{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.9999863865008543,
  "eval_steps": 500,
  "global_step": 36728,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 5.4453996583011714e-05,
      "grad_norm": 0.6514295021273755,
      "learning_rate": 1.8148820326678765e-07,
      "loss": 12.2835,
      "step": 1
    },
    {
      "epoch": 0.00010890799316602343,
      "grad_norm": 0.6445339767412759,
      "learning_rate": 3.629764065335753e-07,
      "loss": 12.0613,
      "step": 2
    },
    {
      "epoch": 0.00016336198974903514,
      "grad_norm": 0.7426420695460909,
      "learning_rate": 5.44464609800363e-07,
      "loss": 12.4091,
      "step": 3
    },
    {
      "epoch": 0.00021781598633204685,
      "grad_norm": 0.7481904256847658,
      "learning_rate": 7.259528130671506e-07,
      "loss": 12.3499,
      "step": 4
    },
    {
      "epoch": 0.0002722699829150586,
      "grad_norm": 0.6345118387930209,
      "learning_rate": 9.074410163339384e-07,
      "loss": 12.1005,
      "step": 5
    },
    {
      "epoch": 0.0003267239794980703,
      "grad_norm": 0.7830858567681372,
      "learning_rate": 1.088929219600726e-06,
      "loss": 12.3444,
      "step": 6
    },
    {
      "epoch": 0.000381177976081082,
      "grad_norm": 0.6552438665550809,
      "learning_rate": 1.2704174228675138e-06,
      "loss": 12.0518,
      "step": 7
    },
    {
      "epoch": 0.0004356319726640937,
      "grad_norm": 1.0230629177664952,
      "learning_rate": 1.4519056261343012e-06,
      "loss": 12.2645,
      "step": 8
    },
    {
      "epoch": 0.0004900859692471054,
      "grad_norm": 0.736508733398066,
      "learning_rate": 1.6333938294010888e-06,
      "loss": 12.4657,
      "step": 9
    },
    {
      "epoch": 0.0005445399658301172,
      "grad_norm": 0.7832204882876573,
      "learning_rate": 1.8148820326678768e-06,
      "loss": 12.2456,
      "step": 10
    },
    {
      "epoch": 0.0005989939624131288,
      "grad_norm": 0.6406598213024922,
      "learning_rate": 1.996370235934664e-06,
      "loss": 12.2987,
      "step": 11
    },
    {
      "epoch": 0.0006534479589961406,
      "grad_norm": 0.6616531858585198,
      "learning_rate": 2.177858439201452e-06,
      "loss": 12.2704,
      "step": 12
    },
    {
      "epoch": 0.0007079019555791523,
      "grad_norm": 0.9041079613798669,
      "learning_rate": 2.35934664246824e-06,
      "loss": 12.219,
      "step": 13
    },
    {
      "epoch": 0.000762355952162164,
      "grad_norm": 0.7350143336611246,
      "learning_rate": 2.5408348457350276e-06,
      "loss": 12.3781,
      "step": 14
    },
    {
      "epoch": 0.0008168099487451757,
      "grad_norm": 0.650775455014501,
      "learning_rate": 2.722323049001815e-06,
      "loss": 12.3572,
      "step": 15
    },
    {
      "epoch": 0.0008712639453281874,
      "grad_norm": 0.7993019309118847,
      "learning_rate": 2.9038112522686024e-06,
      "loss": 12.2421,
      "step": 16
    },
    {
      "epoch": 0.0009257179419111992,
      "grad_norm": 0.7507798789960347,
      "learning_rate": 3.0852994555353906e-06,
      "loss": 12.3405,
      "step": 17
    },
    {
      "epoch": 0.000980171938494211,
      "grad_norm": 0.7232033569227868,
      "learning_rate": 3.2667876588021776e-06,
      "loss": 12.2115,
      "step": 18
    },
    {
      "epoch": 0.0010346259350772226,
      "grad_norm": 0.6730149594636531,
      "learning_rate": 3.448275862068966e-06,
      "loss": 12.3185,
      "step": 19
    },
    {
      "epoch": 0.0010890799316602344,
      "grad_norm": 0.7695953679018918,
      "learning_rate": 3.6297640653357536e-06,
      "loss": 12.3159,
      "step": 20
    },
    {
      "epoch": 0.001143533928243246,
      "grad_norm": 0.711349662965617,
      "learning_rate": 3.8112522686025406e-06,
      "loss": 12.2963,
      "step": 21
    },
    {
      "epoch": 0.0011979879248262576,
      "grad_norm": 0.8016325530794184,
      "learning_rate": 3.992740471869328e-06,
      "loss": 12.3665,
      "step": 22
    },
    {
      "epoch": 0.0012524419214092694,
      "grad_norm": 0.5886320970744596,
      "learning_rate": 4.174228675136116e-06,
      "loss": 12.1546,
      "step": 23
    },
    {
      "epoch": 0.0013068959179922811,
      "grad_norm": 0.6537757775197935,
      "learning_rate": 4.355716878402904e-06,
      "loss": 12.4958,
      "step": 24
    },
    {
      "epoch": 0.0013613499145752929,
      "grad_norm": 0.6751351224388435,
      "learning_rate": 4.537205081669692e-06,
      "loss": 12.3862,
      "step": 25
    },
    {
      "epoch": 0.0014158039111583046,
      "grad_norm": 0.6901807715714254,
      "learning_rate": 4.71869328493648e-06,
      "loss": 12.1915,
      "step": 26
    },
    {
      "epoch": 0.0014702579077413163,
      "grad_norm": 0.7439712686389042,
      "learning_rate": 4.900181488203267e-06,
      "loss": 12.3018,
      "step": 27
    },
    {
      "epoch": 0.001524711904324328,
      "grad_norm": 0.6966653361776312,
      "learning_rate": 5.081669691470055e-06,
      "loss": 12.2465,
      "step": 28
    },
    {
      "epoch": 0.0015791659009073396,
      "grad_norm": 0.6787479289536181,
      "learning_rate": 5.263157894736842e-06,
      "loss": 12.3024,
      "step": 29
    },
    {
      "epoch": 0.0016336198974903514,
      "grad_norm": 0.7046649743230384,
      "learning_rate": 5.44464609800363e-06,
      "loss": 12.3373,
      "step": 30
    },
    {
      "epoch": 0.001688073894073363,
      "grad_norm": 0.6384618515156677,
      "learning_rate": 5.626134301270418e-06,
      "loss": 12.2291,
      "step": 31
    },
    {
      "epoch": 0.0017425278906563748,
      "grad_norm": 0.7312209897280082,
      "learning_rate": 5.807622504537205e-06,
      "loss": 12.2218,
      "step": 32
    },
    {
      "epoch": 0.0017969818872393866,
      "grad_norm": 0.6929587260508587,
      "learning_rate": 5.9891107078039935e-06,
      "loss": 12.2801,
      "step": 33
    },
    {
      "epoch": 0.0018514358838223983,
      "grad_norm": 0.6811907398958092,
      "learning_rate": 6.170598911070781e-06,
      "loss": 12.2846,
      "step": 34
    },
    {
      "epoch": 0.00190588988040541,
      "grad_norm": 0.7114030300924737,
      "learning_rate": 6.352087114337568e-06,
      "loss": 12.2761,
      "step": 35
    },
    {
      "epoch": 0.001960343876988422,
      "grad_norm": 0.7208253804183478,
      "learning_rate": 6.533575317604355e-06,
      "loss": 12.2223,
      "step": 36
    },
    {
      "epoch": 0.0020147978735714335,
      "grad_norm": 0.7173103462219833,
      "learning_rate": 6.715063520871144e-06,
      "loss": 12.2833,
      "step": 37
    },
    {
      "epoch": 0.0020692518701544453,
      "grad_norm": 0.7199801679873419,
      "learning_rate": 6.896551724137932e-06,
      "loss": 12.4004,
      "step": 38
    },
    {
      "epoch": 0.002123705866737457,
      "grad_norm": 0.6283121742102948,
      "learning_rate": 7.078039927404719e-06,
      "loss": 12.2422,
      "step": 39
    },
    {
      "epoch": 0.0021781598633204688,
      "grad_norm": 0.6438610301627259,
      "learning_rate": 7.259528130671507e-06,
      "loss": 12.2721,
      "step": 40
    },
    {
      "epoch": 0.0022326138599034805,
      "grad_norm": 0.673710537418366,
      "learning_rate": 7.441016333938294e-06,
      "loss": 12.2338,
      "step": 41
    },
    {
      "epoch": 0.002287067856486492,
      "grad_norm": 0.7046804607495716,
      "learning_rate": 7.622504537205081e-06,
      "loss": 12.3124,
      "step": 42
    },
    {
      "epoch": 0.0023415218530695035,
      "grad_norm": 0.7040526020482769,
      "learning_rate": 7.80399274047187e-06,
      "loss": 12.2752,
      "step": 43
    },
    {
      "epoch": 0.0023959758496525153,
      "grad_norm": 0.7306043247419787,
      "learning_rate": 7.985480943738657e-06,
      "loss": 12.4015,
      "step": 44
    },
    {
      "epoch": 0.002450429846235527,
      "grad_norm": 0.664490461781453,
      "learning_rate": 8.166969147005445e-06,
      "loss": 12.3169,
      "step": 45
    },
    {
      "epoch": 0.0025048838428185388,
      "grad_norm": 0.6548692389195887,
      "learning_rate": 8.348457350272232e-06,
      "loss": 12.2979,
      "step": 46
    },
    {
      "epoch": 0.0025593378394015505,
      "grad_norm": 0.7790840513869893,
      "learning_rate": 8.52994555353902e-06,
      "loss": 12.3873,
      "step": 47
    },
    {
      "epoch": 0.0026137918359845622,
      "grad_norm": 0.700770109715519,
      "learning_rate": 8.711433756805808e-06,
      "loss": 12.2902,
      "step": 48
    },
    {
      "epoch": 0.002668245832567574,
      "grad_norm": 0.7284900184817878,
      "learning_rate": 8.892921960072596e-06,
      "loss": 12.2719,
      "step": 49
    },
    {
      "epoch": 0.0027226998291505857,
      "grad_norm": 0.8332734018054926,
      "learning_rate": 9.074410163339384e-06,
      "loss": 12.2799,
      "step": 50
    },
    {
      "epoch": 0.0027771538257335975,
      "grad_norm": 0.6965636980670914,
      "learning_rate": 9.255898366606171e-06,
      "loss": 12.2629,
      "step": 51
    },
    {
      "epoch": 0.002831607822316609,
      "grad_norm": 0.6783391272444385,
      "learning_rate": 9.43738656987296e-06,
      "loss": 12.1928,
      "step": 52
    },
    {
      "epoch": 0.002886061818899621,
      "grad_norm": 0.6768160560169009,
      "learning_rate": 9.618874773139747e-06,
      "loss": 12.2738,
      "step": 53
    },
    {
      "epoch": 0.0029405158154826327,
      "grad_norm": 0.7038244532929006,
      "learning_rate": 9.800362976406533e-06,
      "loss": 12.2378,
      "step": 54
    },
    {
      "epoch": 0.0029949698120656444,
      "grad_norm": 0.6932385380100631,
      "learning_rate": 9.981851179673321e-06,
      "loss": 12.3725,
      "step": 55
    },
    {
      "epoch": 0.003049423808648656,
      "grad_norm": 0.6915084120243854,
      "learning_rate": 1.016333938294011e-05,
      "loss": 12.2021,
      "step": 56
    },
    {
      "epoch": 0.003103877805231668,
      "grad_norm": 0.7148909867057305,
      "learning_rate": 1.0344827586206897e-05,
      "loss": 12.3215,
      "step": 57
    },
    {
      "epoch": 0.0031583318018146792,
      "grad_norm": 0.6991958091345991,
      "learning_rate": 1.0526315789473684e-05,
      "loss": 12.2382,
      "step": 58
    },
    {
      "epoch": 0.003212785798397691,
      "grad_norm": 0.6919246914121425,
      "learning_rate": 1.0707803992740472e-05,
      "loss": 12.183,
      "step": 59
    },
    {
      "epoch": 0.0032672397949807027,
      "grad_norm": 0.8025321181544712,
      "learning_rate": 1.088929219600726e-05,
      "loss": 12.4441,
      "step": 60
    },
    {
      "epoch": 0.0033216937915637144,
      "grad_norm": 0.6698039631860974,
      "learning_rate": 1.1070780399274048e-05,
      "loss": 12.2611,
      "step": 61
    },
    {
      "epoch": 0.003376147788146726,
      "grad_norm": 0.7407394943261039,
      "learning_rate": 1.1252268602540836e-05,
      "loss": 12.2053,
      "step": 62
    },
    {
      "epoch": 0.003430601784729738,
      "grad_norm": 0.6865613288316597,
      "learning_rate": 1.1433756805807623e-05,
      "loss": 12.3175,
      "step": 63
    },
    {
      "epoch": 0.0034850557813127497,
      "grad_norm": 0.7041064521893132,
      "learning_rate": 1.161524500907441e-05,
      "loss": 12.2321,
      "step": 64
    },
    {
      "epoch": 0.0035395097778957614,
      "grad_norm": 0.6723425810734814,
      "learning_rate": 1.1796733212341199e-05,
      "loss": 12.2558,
      "step": 65
    },
    {
      "epoch": 0.003593963774478773,
      "grad_norm": 0.661956982839324,
      "learning_rate": 1.1978221415607987e-05,
      "loss": 12.2064,
      "step": 66
    },
    {
      "epoch": 0.003648417771061785,
      "grad_norm": 0.7483218714408623,
      "learning_rate": 1.2159709618874773e-05,
      "loss": 12.3637,
      "step": 67
    },
    {
      "epoch": 0.0037028717676447966,
      "grad_norm": 0.6719289161767829,
      "learning_rate": 1.2341197822141563e-05,
      "loss": 12.2866,
      "step": 68
    },
    {
      "epoch": 0.0037573257642278084,
      "grad_norm": 0.7353482196298523,
      "learning_rate": 1.2522686025408347e-05,
      "loss": 12.3733,
      "step": 69
    },
    {
      "epoch": 0.00381177976081082,
      "grad_norm": 0.7121839832467598,
      "learning_rate": 1.2704174228675136e-05,
      "loss": 12.3107,
      "step": 70
    },
    {
      "epoch": 0.003866233757393832,
      "grad_norm": 0.7298666561899594,
      "learning_rate": 1.2885662431941924e-05,
      "loss": 12.2409,
      "step": 71
    },
    {
      "epoch": 0.003920687753976844,
      "grad_norm": 0.673187183618948,
      "learning_rate": 1.306715063520871e-05,
      "loss": 12.2624,
      "step": 72
    },
    {
      "epoch": 0.003975141750559855,
      "grad_norm": 0.6647579362296864,
      "learning_rate": 1.32486388384755e-05,
      "loss": 12.2228,
      "step": 73
    },
    {
      "epoch": 0.004029595747142867,
      "grad_norm": 0.7126085308286554,
      "learning_rate": 1.3430127041742288e-05,
      "loss": 12.1887,
      "step": 74
    },
    {
      "epoch": 0.004084049743725879,
      "grad_norm": 0.6508629608031615,
      "learning_rate": 1.3611615245009074e-05,
      "loss": 12.2574,
      "step": 75
    },
    {
      "epoch": 0.0041385037403088906,
      "grad_norm": 0.7217793347228008,
      "learning_rate": 1.3793103448275863e-05,
      "loss": 12.3596,
      "step": 76
    },
    {
      "epoch": 0.004192957736891902,
      "grad_norm": 0.7277886621255288,
      "learning_rate": 1.3974591651542651e-05,
      "loss": 12.1593,
      "step": 77
    },
    {
      "epoch": 0.004247411733474914,
      "grad_norm": 0.7476740623784802,
      "learning_rate": 1.4156079854809437e-05,
      "loss": 12.2851,
      "step": 78
    },
    {
      "epoch": 0.004301865730057926,
      "grad_norm": 0.670232339117414,
      "learning_rate": 1.4337568058076225e-05,
      "loss": 12.2627,
      "step": 79
    },
    {
      "epoch": 0.0043563197266409375,
      "grad_norm": 0.6467570917998483,
      "learning_rate": 1.4519056261343015e-05,
      "loss": 12.2336,
      "step": 80
    },
    {
      "epoch": 0.004410773723223949,
      "grad_norm": 0.7120208830515039,
      "learning_rate": 1.47005444646098e-05,
      "loss": 12.357,
      "step": 81
    },
    {
      "epoch": 0.004465227719806961,
      "grad_norm": 0.7432640467149172,
      "learning_rate": 1.4882032667876588e-05,
      "loss": 12.3145,
      "step": 82
    },
    {
      "epoch": 0.004519681716389972,
      "grad_norm": 0.8293675536060945,
      "learning_rate": 1.5063520871143378e-05,
      "loss": 12.6006,
      "step": 83
    },
    {
      "epoch": 0.004574135712972984,
      "grad_norm": 0.7120304671169275,
      "learning_rate": 1.5245009074410162e-05,
      "loss": 12.3627,
      "step": 84
    },
    {
      "epoch": 0.004628589709555995,
      "grad_norm": 0.6673914704270645,
      "learning_rate": 1.542649727767695e-05,
      "loss": 12.2417,
      "step": 85
    },
    {
      "epoch": 0.004683043706139007,
      "grad_norm": 0.6868912483040882,
      "learning_rate": 1.560798548094374e-05,
      "loss": 12.2832,
      "step": 86
    },
    {
      "epoch": 0.004737497702722019,
      "grad_norm": 0.6950953167328401,
      "learning_rate": 1.5789473684210526e-05,
      "loss": 12.2372,
      "step": 87
    },
    {
      "epoch": 0.004791951699305031,
      "grad_norm": 0.7383575421215864,
      "learning_rate": 1.5970961887477314e-05,
      "loss": 12.2887,
      "step": 88
    },
    {
      "epoch": 0.004846405695888042,
      "grad_norm": 0.703029378935624,
      "learning_rate": 1.6152450090744105e-05,
      "loss": 12.3149,
      "step": 89
    },
    {
      "epoch": 0.004900859692471054,
      "grad_norm": 0.7021279833401868,
      "learning_rate": 1.633393829401089e-05,
      "loss": 12.2678,
      "step": 90
    },
    {
      "epoch": 0.004955313689054066,
      "grad_norm": 0.7251509526578306,
      "learning_rate": 1.6515426497277677e-05,
      "loss": 12.3562,
      "step": 91
    },
    {
      "epoch": 0.0050097676856370775,
      "grad_norm": 0.7797560038561253,
      "learning_rate": 1.6696914700544465e-05,
      "loss": 12.276,
      "step": 92
    },
    {
      "epoch": 0.005064221682220089,
      "grad_norm": 0.7064592922388186,
      "learning_rate": 1.6878402903811253e-05,
      "loss": 12.3852,
      "step": 93
    },
    {
      "epoch": 0.005118675678803101,
      "grad_norm": 0.7053365174441039,
      "learning_rate": 1.705989110707804e-05,
      "loss": 12.3483,
      "step": 94
    },
    {
      "epoch": 0.005173129675386113,
      "grad_norm": 0.6738625360373299,
      "learning_rate": 1.7241379310344828e-05,
      "loss": 12.1678,
      "step": 95
    },
    {
      "epoch": 0.0052275836719691245,
      "grad_norm": 0.5972838096238351,
      "learning_rate": 1.7422867513611616e-05,
      "loss": 12.1698,
      "step": 96
    },
    {
      "epoch": 0.005282037668552136,
      "grad_norm": 0.694605357226162,
      "learning_rate": 1.7604355716878404e-05,
      "loss": 12.2028,
      "step": 97
    },
    {
      "epoch": 0.005336491665135148,
      "grad_norm": 0.7235709308856128,
      "learning_rate": 1.7785843920145192e-05,
      "loss": 12.3249,
      "step": 98
    },
    {
      "epoch": 0.00539094566171816,
      "grad_norm": 0.7116272658253867,
      "learning_rate": 1.796733212341198e-05,
      "loss": 12.1328,
      "step": 99
    },
    {
      "epoch": 0.0054453996583011715,
      "grad_norm": 0.7960311740307418,
      "learning_rate": 1.8148820326678767e-05,
      "loss": 12.4241,
      "step": 100
    },
    {
      "epoch": 0.005499853654884183,
      "grad_norm": 0.7607101112707406,
      "learning_rate": 1.8330308529945555e-05,
      "loss": 12.1386,
      "step": 101
    },
    {
      "epoch": 0.005554307651467195,
      "grad_norm": 0.6989823266122064,
      "learning_rate": 1.8511796733212343e-05,
      "loss": 12.3272,
      "step": 102
    },
    {
      "epoch": 0.005608761648050207,
      "grad_norm": 0.6771018070968361,
      "learning_rate": 1.869328493647913e-05,
      "loss": 12.2864,
      "step": 103
    },
    {
      "epoch": 0.005663215644633218,
      "grad_norm": 0.6776164869457959,
      "learning_rate": 1.887477313974592e-05,
      "loss": 12.2483,
      "step": 104
    },
    {
      "epoch": 0.00571766964121623,
      "grad_norm": 0.7161127268232864,
      "learning_rate": 1.9056261343012703e-05,
      "loss": 12.1731,
      "step": 105
    },
    {
      "epoch": 0.005772123637799242,
      "grad_norm": 0.7248124927863123,
      "learning_rate": 1.9237749546279494e-05,
      "loss": 12.4591,
      "step": 106
    },
    {
      "epoch": 0.005826577634382254,
      "grad_norm": 0.7591292383102465,
      "learning_rate": 1.941923774954628e-05,
      "loss": 12.1536,
      "step": 107
    },
    {
      "epoch": 0.005881031630965265,
      "grad_norm": 0.6567533407138826,
      "learning_rate": 1.9600725952813066e-05,
      "loss": 12.2191,
      "step": 108
    },
    {
      "epoch": 0.005935485627548277,
      "grad_norm": 0.7190643868028403,
      "learning_rate": 1.9782214156079858e-05,
      "loss": 12.2527,
      "step": 109
    },
    {
      "epoch": 0.005989939624131289,
      "grad_norm": 0.6898991311631406,
      "learning_rate": 1.9963702359346642e-05,
      "loss": 12.2642,
      "step": 110
    },
    {
      "epoch": 0.006044393620714301,
      "grad_norm": 0.7495187164171233,
      "learning_rate": 2.014519056261343e-05,
      "loss": 12.2646,
      "step": 111
    },
    {
      "epoch": 0.006098847617297312,
      "grad_norm": 0.7170076277469752,
      "learning_rate": 2.032667876588022e-05,
      "loss": 12.3398,
      "step": 112
    },
    {
      "epoch": 0.006153301613880324,
      "grad_norm": 0.731108164640672,
      "learning_rate": 2.0508166969147005e-05,
      "loss": 12.3531,
      "step": 113
    },
    {
      "epoch": 0.006207755610463336,
      "grad_norm": 0.7825773788282819,
      "learning_rate": 2.0689655172413793e-05,
      "loss": 12.2599,
      "step": 114
    },
    {
      "epoch": 0.006262209607046348,
      "grad_norm": 0.6717004811664388,
      "learning_rate": 2.087114337568058e-05,
      "loss": 12.3298,
      "step": 115
    },
    {
      "epoch": 0.0063166636036293584,
      "grad_norm": 0.6969969828201115,
      "learning_rate": 2.105263157894737e-05,
      "loss": 12.2737,
      "step": 116
    },
    {
      "epoch": 0.00637111760021237,
      "grad_norm": 0.7242188258016712,
      "learning_rate": 2.1234119782214157e-05,
      "loss": 12.3695,
      "step": 117
    },
    {
      "epoch": 0.006425571596795382,
      "grad_norm": 0.8811248563175815,
      "learning_rate": 2.1415607985480945e-05,
      "loss": 12.3923,
      "step": 118
    },
    {
      "epoch": 0.006480025593378394,
      "grad_norm": 0.6943426432171022,
      "learning_rate": 2.1597096188747732e-05,
      "loss": 12.2724,
      "step": 119
    },
    {
      "epoch": 0.006534479589961405,
      "grad_norm": 0.8367672147901168,
      "learning_rate": 2.177858439201452e-05,
      "loss": 12.2846,
      "step": 120
    },
    {
      "epoch": 0.006588933586544417,
      "grad_norm": 0.7169323963392524,
      "learning_rate": 2.1960072595281308e-05,
      "loss": 12.3054,
      "step": 121
    },
    {
      "epoch": 0.006643387583127429,
      "grad_norm": 0.7328339351641384,
      "learning_rate": 2.2141560798548096e-05,
      "loss": 12.3401,
      "step": 122
    },
    {
      "epoch": 0.006697841579710441,
      "grad_norm": 0.6950351401076931,
      "learning_rate": 2.2323049001814884e-05,
      "loss": 12.2651,
      "step": 123
    },
    {
      "epoch": 0.006752295576293452,
      "grad_norm": 0.7135461876331347,
      "learning_rate": 2.250453720508167e-05,
      "loss": 12.3066,
      "step": 124
    },
    {
      "epoch": 0.006806749572876464,
      "grad_norm": 0.6740158526405283,
      "learning_rate": 2.2686025408348456e-05,
      "loss": 12.2421,
      "step": 125
    },
    {
      "epoch": 0.006861203569459476,
      "grad_norm": 0.7511554883846614,
      "learning_rate": 2.2867513611615247e-05,
      "loss": 12.2741,
      "step": 126
    },
    {
      "epoch": 0.006915657566042488,
      "grad_norm": 0.6990531199482589,
      "learning_rate": 2.3049001814882035e-05,
      "loss": 12.384,
      "step": 127
    },
    {
      "epoch": 0.006970111562625499,
      "grad_norm": 0.7135501305238853,
      "learning_rate": 2.323049001814882e-05,
      "loss": 12.3623,
      "step": 128
    },
    {
      "epoch": 0.007024565559208511,
      "grad_norm": 0.7041623355500041,
      "learning_rate": 2.341197822141561e-05,
      "loss": 12.3004,
      "step": 129
    },
    {
      "epoch": 0.007079019555791523,
      "grad_norm": 0.6661526549812437,
      "learning_rate": 2.3593466424682398e-05,
      "loss": 12.1402,
      "step": 130
    },
    {
      "epoch": 0.0071334735523745346,
      "grad_norm": 0.7361476840420841,
      "learning_rate": 2.3774954627949183e-05,
      "loss": 12.1733,
      "step": 131
    },
    {
      "epoch": 0.007187927548957546,
      "grad_norm": 0.7967543207030372,
      "learning_rate": 2.3956442831215974e-05,
      "loss": 12.4345,
      "step": 132
    },
    {
      "epoch": 0.007242381545540558,
      "grad_norm": 0.77469629112749,
      "learning_rate": 2.413793103448276e-05,
      "loss": 12.3164,
      "step": 133
    },
    {
      "epoch": 0.00729683554212357,
      "grad_norm": 0.8176680285748424,
      "learning_rate": 2.4319419237749546e-05,
      "loss": 12.3651,
      "step": 134
    },
    {
      "epoch": 0.0073512895387065815,
      "grad_norm": 0.6969513054321705,
      "learning_rate": 2.4500907441016334e-05,
      "loss": 12.3023,
      "step": 135
    },
    {
      "epoch": 0.007405743535289593,
      "grad_norm": 0.6874259890342583,
      "learning_rate": 2.4682395644283125e-05,
      "loss": 12.3488,
      "step": 136
    },
    {
      "epoch": 0.007460197531872605,
      "grad_norm": 0.7848881195123376,
      "learning_rate": 2.486388384754991e-05,
      "loss": 12.3248,
      "step": 137
    },
    {
      "epoch": 0.007514651528455617,
      "grad_norm": 0.7536572549155603,
      "learning_rate": 2.5045372050816694e-05,
      "loss": 12.3101,
      "step": 138
    },
    {
      "epoch": 0.0075691055250386285,
      "grad_norm": 0.6779931950396582,
      "learning_rate": 2.5226860254083485e-05,
      "loss": 12.21,
      "step": 139
    },
    {
      "epoch": 0.00762355952162164,
      "grad_norm": 0.7551152796581928,
      "learning_rate": 2.5408348457350273e-05,
      "loss": 12.4042,
      "step": 140
    },
    {
      "epoch": 0.007678013518204652,
      "grad_norm": 0.9069979314667508,
      "learning_rate": 2.558983666061706e-05,
      "loss": 12.2847,
      "step": 141
    },
    {
      "epoch": 0.007732467514787664,
      "grad_norm": 0.7202987526526302,
      "learning_rate": 2.577132486388385e-05,
      "loss": 12.363,
      "step": 142
    },
    {
      "epoch": 0.0077869215113706754,
      "grad_norm": 0.7524333057239485,
      "learning_rate": 2.595281306715064e-05,
      "loss": 12.3406,
      "step": 143
    },
    {
      "epoch": 0.007841375507953687,
      "grad_norm": 0.7164888079373964,
      "learning_rate": 2.613430127041742e-05,
      "loss": 12.3726,
      "step": 144
    },
    {
      "epoch": 0.007895829504536699,
      "grad_norm": 0.7311551787417221,
      "learning_rate": 2.6315789473684212e-05,
      "loss": 12.2858,
      "step": 145
    },
    {
      "epoch": 0.00795028350111971,
      "grad_norm": 0.6773134174743295,
      "learning_rate": 2.6497277676951e-05,
      "loss": 12.3468,
      "step": 146
    },
    {
      "epoch": 0.008004737497702722,
      "grad_norm": 0.7174401575961286,
      "learning_rate": 2.6678765880217788e-05,
      "loss": 12.2457,
      "step": 147
    },
    {
      "epoch": 0.008059191494285734,
      "grad_norm": 0.6863666889751157,
      "learning_rate": 2.6860254083484575e-05,
      "loss": 12.2749,
      "step": 148
    },
    {
      "epoch": 0.008113645490868746,
      "grad_norm": 0.7262950182241283,
      "learning_rate": 2.7041742286751363e-05,
      "loss": 12.2609,
      "step": 149
    },
    {
      "epoch": 0.008168099487451758,
      "grad_norm": 0.8015975463544986,
      "learning_rate": 2.7223230490018148e-05,
      "loss": 12.2737,
      "step": 150
    },
    {
      "epoch": 0.00822255348403477,
      "grad_norm": 0.6974688735570929,
      "learning_rate": 2.7404718693284935e-05,
      "loss": 12.2764,
      "step": 151
    },
    {
      "epoch": 0.008277007480617781,
      "grad_norm": 0.7612859811519558,
      "learning_rate": 2.7586206896551727e-05,
      "loss": 12.3173,
      "step": 152
    },
    {
      "epoch": 0.008331461477200793,
      "grad_norm": 0.7535272542092525,
      "learning_rate": 2.7767695099818514e-05,
      "loss": 12.2599,
      "step": 153
    },
    {
      "epoch": 0.008385915473783805,
      "grad_norm": 0.8024979560171636,
      "learning_rate": 2.7949183303085302e-05,
      "loss": 12.3722,
      "step": 154
    },
    {
      "epoch": 0.008440369470366816,
      "grad_norm": 0.7531576898161162,
      "learning_rate": 2.813067150635209e-05,
      "loss": 12.3065,
      "step": 155
    },
    {
      "epoch": 0.008494823466949828,
      "grad_norm": 0.7575320428987756,
      "learning_rate": 2.8312159709618874e-05,
      "loss": 12.356,
      "step": 156
    },
    {
      "epoch": 0.00854927746353284,
      "grad_norm": 0.6921716231479544,
      "learning_rate": 2.8493647912885662e-05,
      "loss": 12.2568,
      "step": 157
    },
    {
      "epoch": 0.008603731460115852,
      "grad_norm": 0.6867771987616564,
      "learning_rate": 2.867513611615245e-05,
      "loss": 12.2139,
      "step": 158
    },
    {
      "epoch": 0.008658185456698863,
      "grad_norm": 0.7022464961805333,
      "learning_rate": 2.885662431941924e-05,
      "loss": 12.3242,
      "step": 159
    },
    {
      "epoch": 0.008712639453281875,
      "grad_norm": 0.7220153309753276,
      "learning_rate": 2.903811252268603e-05,
      "loss": 12.3137,
      "step": 160
    },
    {
      "epoch": 0.008767093449864887,
      "grad_norm": 0.7116207468224576,
      "learning_rate": 2.9219600725952817e-05,
      "loss": 12.4212,
      "step": 161
    },
    {
      "epoch": 0.008821547446447899,
      "grad_norm": 0.7606595881201347,
      "learning_rate": 2.94010889292196e-05,
      "loss": 12.3533,
      "step": 162
    },
    {
      "epoch": 0.00887600144303091,
      "grad_norm": 0.8010136363377759,
      "learning_rate": 2.958257713248639e-05,
      "loss": 12.384,
      "step": 163
    },
    {
      "epoch": 0.008930455439613922,
      "grad_norm": 0.7641948223788549,
      "learning_rate": 2.9764065335753177e-05,
      "loss": 12.2878,
      "step": 164
    },
    {
      "epoch": 0.008984909436196934,
      "grad_norm": 0.7313238653805536,
      "learning_rate": 2.9945553539019965e-05,
      "loss": 12.2797,
      "step": 165
    },
    {
      "epoch": 0.009039363432779944,
      "grad_norm": 0.7280246174606982,
      "learning_rate": 3.0127041742286756e-05,
      "loss": 12.2048,
      "step": 166
    },
    {
      "epoch": 0.009093817429362955,
      "grad_norm": 0.6735111329183364,
      "learning_rate": 3.0308529945553544e-05,
      "loss": 12.1298,
      "step": 167
    },
    {
      "epoch": 0.009148271425945967,
      "grad_norm": 0.662515268105397,
      "learning_rate": 3.0490018148820325e-05,
      "loss": 12.2694,
      "step": 168
    },
    {
      "epoch": 0.009202725422528979,
      "grad_norm": 0.6814067150138639,
      "learning_rate": 3.0671506352087116e-05,
      "loss": 12.3329,
      "step": 169
    },
    {
      "epoch": 0.00925717941911199,
      "grad_norm": 0.6299645547657693,
      "learning_rate": 3.08529945553539e-05,
      "loss": 12.2323,
      "step": 170
    },
    {
      "epoch": 0.009311633415695002,
      "grad_norm": 0.7046415195233939,
      "learning_rate": 3.103448275862069e-05,
      "loss": 12.1888,
      "step": 171
    },
    {
      "epoch": 0.009366087412278014,
      "grad_norm": 0.6635786550238101,
      "learning_rate": 3.121597096188748e-05,
      "loss": 12.3145,
      "step": 172
    },
    {
      "epoch": 0.009420541408861026,
      "grad_norm": 0.8405662525571916,
      "learning_rate": 3.139745916515426e-05,
      "loss": 12.3302,
      "step": 173
    },
    {
      "epoch": 0.009474995405444038,
      "grad_norm": 0.7053223200658312,
      "learning_rate": 3.157894736842105e-05,
      "loss": 12.2905,
      "step": 174
    },
    {
      "epoch": 0.00952944940202705,
      "grad_norm": 0.7179732734291909,
      "learning_rate": 3.176043557168784e-05,
      "loss": 12.3143,
      "step": 175
    },
    {
      "epoch": 0.009583903398610061,
      "grad_norm": 0.6713806463041468,
      "learning_rate": 3.194192377495463e-05,
      "loss": 12.2889,
      "step": 176
    },
    {
      "epoch": 0.009638357395193073,
      "grad_norm": 0.8537429522165395,
      "learning_rate": 3.212341197822142e-05,
      "loss": 12.3003,
      "step": 177
    },
    {
      "epoch": 0.009692811391776085,
      "grad_norm": 0.7936643802651663,
      "learning_rate": 3.230490018148821e-05,
      "loss": 12.4024,
      "step": 178
    },
    {
      "epoch": 0.009747265388359096,
      "grad_norm": 0.6673211953219038,
      "learning_rate": 3.248638838475499e-05,
      "loss": 12.3591,
      "step": 179
    },
    {
      "epoch": 0.009801719384942108,
      "grad_norm": 0.7664690630438143,
      "learning_rate": 3.266787658802178e-05,
      "loss": 12.2,
      "step": 180
    },
    {
      "epoch": 0.00985617338152512,
      "grad_norm": 0.7602742645809355,
      "learning_rate": 3.284936479128857e-05,
      "loss": 12.2861,
      "step": 181
    },
    {
      "epoch": 0.009910627378108132,
      "grad_norm": 0.7223425709471121,
      "learning_rate": 3.3030852994555354e-05,
      "loss": 12.3773,
      "step": 182
    },
    {
      "epoch": 0.009965081374691143,
      "grad_norm": 0.7272094124639393,
      "learning_rate": 3.3212341197822145e-05,
      "loss": 12.2859,
      "step": 183
    },
    {
      "epoch": 0.010019535371274155,
      "grad_norm": 0.7226902937341132,
      "learning_rate": 3.339382940108893e-05,
      "loss": 12.3335,
      "step": 184
    },
    {
      "epoch": 0.010073989367857167,
      "grad_norm": 0.7069144564518154,
      "learning_rate": 3.3575317604355714e-05,
      "loss": 12.3361,
      "step": 185
    },
    {
      "epoch": 0.010128443364440179,
      "grad_norm": 0.7351369937399913,
      "learning_rate": 3.3756805807622505e-05,
      "loss": 12.3304,
      "step": 186
    },
    {
      "epoch": 0.01018289736102319,
      "grad_norm": 0.7562449223296611,
      "learning_rate": 3.3938294010889297e-05,
      "loss": 12.4463,
      "step": 187
    },
    {
      "epoch": 0.010237351357606202,
      "grad_norm": 0.7380133543278365,
      "learning_rate": 3.411978221415608e-05,
      "loss": 12.3487,
      "step": 188
    },
    {
      "epoch": 0.010291805354189214,
      "grad_norm": 0.8018559952427972,
      "learning_rate": 3.430127041742287e-05,
      "loss": 12.4735,
      "step": 189
    },
    {
      "epoch": 0.010346259350772226,
      "grad_norm": 0.6992645273661477,
      "learning_rate": 3.4482758620689657e-05,
      "loss": 12.2485,
      "step": 190
    },
    {
      "epoch": 0.010400713347355237,
      "grad_norm": 0.7190265436442772,
      "learning_rate": 3.466424682395644e-05,
      "loss": 12.2231,
      "step": 191
    },
    {
      "epoch": 0.010455167343938249,
      "grad_norm": 0.7728396372356257,
      "learning_rate": 3.484573502722323e-05,
      "loss": 12.4133,
      "step": 192
    },
    {
      "epoch": 0.01050962134052126,
      "grad_norm": 0.8039774543145355,
      "learning_rate": 3.502722323049002e-05,
      "loss": 12.4807,
      "step": 193
    },
    {
      "epoch": 0.010564075337104272,
      "grad_norm": 0.7275544168870578,
      "learning_rate": 3.520871143375681e-05,
      "loss": 12.3294,
      "step": 194
    },
    {
      "epoch": 0.010618529333687284,
      "grad_norm": 0.8566875819319059,
      "learning_rate": 3.53901996370236e-05,
      "loss": 12.4185,
      "step": 195
    },
    {
      "epoch": 0.010672983330270296,
      "grad_norm": 0.8055650218978196,
      "learning_rate": 3.5571687840290383e-05,
      "loss": 12.5118,
      "step": 196
    },
    {
      "epoch": 0.010727437326853308,
      "grad_norm": 0.7349187170227696,
      "learning_rate": 3.575317604355717e-05,
      "loss": 12.2468,
      "step": 197
    },
    {
      "epoch": 0.01078189132343632,
      "grad_norm": 0.7134840505631237,
      "learning_rate": 3.593466424682396e-05,
      "loss": 12.388,
      "step": 198
    },
    {
      "epoch": 0.010836345320019331,
      "grad_norm": 0.6774397050532335,
      "learning_rate": 3.6116152450090743e-05,
      "loss": 12.2235,
      "step": 199
    },
    {
      "epoch": 0.010890799316602343,
      "grad_norm": 0.7744176715025664,
      "learning_rate": 3.6297640653357535e-05,
      "loss": 12.293,
      "step": 200
    },
    {
      "epoch": 0.010945253313185355,
      "grad_norm": 0.7348144637670879,
      "learning_rate": 3.6479128856624326e-05,
      "loss": 12.3172,
      "step": 201
    },
    {
      "epoch": 0.010999707309768366,
      "grad_norm": 0.6968931217183741,
      "learning_rate": 3.666061705989111e-05,
      "loss": 12.1999,
      "step": 202
    },
    {
      "epoch": 0.011054161306351378,
      "grad_norm": 0.847067144995256,
      "learning_rate": 3.6842105263157895e-05,
      "loss": 12.2706,
      "step": 203
    },
    {
      "epoch": 0.01110861530293439,
      "grad_norm": 0.7530363546527266,
      "learning_rate": 3.7023593466424686e-05,
      "loss": 12.2301,
      "step": 204
    },
    {
      "epoch": 0.011163069299517402,
      "grad_norm": 0.7564652633064927,
      "learning_rate": 3.720508166969147e-05,
      "loss": 12.3639,
      "step": 205
    },
    {
      "epoch": 0.011217523296100413,
      "grad_norm": 0.7135173235708773,
      "learning_rate": 3.738656987295826e-05,
      "loss": 12.2338,
      "step": 206
    },
    {
      "epoch": 0.011271977292683425,
      "grad_norm": 0.6858025792749569,
      "learning_rate": 3.7568058076225046e-05,
      "loss": 12.3793,
      "step": 207
    },
    {
      "epoch": 0.011326431289266437,
      "grad_norm": 0.7824021027165805,
      "learning_rate": 3.774954627949184e-05,
      "loss": 12.339,
      "step": 208
    },
    {
      "epoch": 0.011380885285849449,
      "grad_norm": 0.7250247555230146,
      "learning_rate": 3.793103448275862e-05,
      "loss": 12.2351,
      "step": 209
    },
    {
      "epoch": 0.01143533928243246,
      "grad_norm": 0.6809803088998201,
      "learning_rate": 3.8112522686025406e-05,
      "loss": 12.4248,
      "step": 210
    },
    {
      "epoch": 0.011489793279015472,
      "grad_norm": 0.7032058028279533,
      "learning_rate": 3.82940108892922e-05,
      "loss": 12.3732,
      "step": 211
    },
    {
      "epoch": 0.011544247275598484,
      "grad_norm": 0.754428880383836,
      "learning_rate": 3.847549909255899e-05,
      "loss": 12.3197,
      "step": 212
    },
    {
      "epoch": 0.011598701272181496,
      "grad_norm": 0.7504424399849499,
      "learning_rate": 3.865698729582577e-05,
      "loss": 12.3846,
      "step": 213
    },
    {
      "epoch": 0.011653155268764507,
      "grad_norm": 0.8098844516869852,
      "learning_rate": 3.883847549909256e-05,
      "loss": 12.4009,
      "step": 214
    },
    {
      "epoch": 0.011707609265347519,
      "grad_norm": 0.7434306698270551,
      "learning_rate": 3.901996370235935e-05,
      "loss": 12.3121,
      "step": 215
    },
    {
      "epoch": 0.01176206326193053,
      "grad_norm": 0.835756221038369,
      "learning_rate": 3.920145190562613e-05,
      "loss": 12.3561,
      "step": 216
    },
    {
      "epoch": 0.011816517258513543,
      "grad_norm": 0.7441859041127364,
      "learning_rate": 3.9382940108892924e-05,
      "loss": 12.3837,
      "step": 217
    },
    {
      "epoch": 0.011870971255096554,
      "grad_norm": 0.6774272997258979,
      "learning_rate": 3.9564428312159715e-05,
      "loss": 12.321,
      "step": 218
    },
    {
      "epoch": 0.011925425251679566,
      "grad_norm": 0.740858626909941,
      "learning_rate": 3.97459165154265e-05,
      "loss": 12.3377,
      "step": 219
    },
    {
      "epoch": 0.011979879248262578,
      "grad_norm": 0.8029304956373081,
      "learning_rate": 3.9927404718693284e-05,
      "loss": 12.2668,
      "step": 220
    },
    {
      "epoch": 0.01203433324484559,
      "grad_norm": 0.7255710443809265,
      "learning_rate": 4.0108892921960075e-05,
      "loss": 12.2052,
      "step": 221
    },
    {
      "epoch": 0.012088787241428601,
      "grad_norm": 0.7467564299377789,
      "learning_rate": 4.029038112522686e-05,
      "loss": 12.3052,
      "step": 222
    },
    {
      "epoch": 0.012143241238011613,
      "grad_norm": 0.7757543921998024,
      "learning_rate": 4.047186932849365e-05,
      "loss": 12.3313,
      "step": 223
    },
    {
      "epoch": 0.012197695234594625,
      "grad_norm": 0.7663879408559824,
      "learning_rate": 4.065335753176044e-05,
      "loss": 12.3567,
      "step": 224
    },
    {
      "epoch": 0.012252149231177636,
      "grad_norm": 0.830893424480459,
      "learning_rate": 4.0834845735027227e-05,
      "loss": 12.2888,
      "step": 225
    },
    {
      "epoch": 0.012306603227760648,
      "grad_norm": 0.7659576414170449,
      "learning_rate": 4.101633393829401e-05,
      "loss": 12.2534,
      "step": 226
    },
    {
      "epoch": 0.01236105722434366,
      "grad_norm": 0.8240503874061232,
      "learning_rate": 4.11978221415608e-05,
      "loss": 12.2796,
      "step": 227
    },
    {
      "epoch": 0.012415511220926672,
      "grad_norm": 0.8055779272980373,
      "learning_rate": 4.1379310344827587e-05,
      "loss": 12.3734,
      "step": 228
    },
    {
      "epoch": 0.012469965217509683,
      "grad_norm": 0.6990309940422604,
      "learning_rate": 4.156079854809438e-05,
      "loss": 12.2448,
      "step": 229
    },
    {
      "epoch": 0.012524419214092695,
      "grad_norm": 0.7814896272818583,
      "learning_rate": 4.174228675136116e-05,
      "loss": 12.3718,
      "step": 230
    },
    {
      "epoch": 0.012578873210675707,
      "grad_norm": 0.6456843025285693,
      "learning_rate": 4.192377495462795e-05,
      "loss": 12.2105,
      "step": 231
    },
    {
      "epoch": 0.012633327207258717,
      "grad_norm": 0.7374802126161342,
      "learning_rate": 4.210526315789474e-05,
      "loss": 12.3398,
      "step": 232
    },
    {
      "epoch": 0.012687781203841729,
      "grad_norm": 0.720484469540933,
      "learning_rate": 4.228675136116152e-05,
      "loss": 12.3939,
      "step": 233
    },
    {
      "epoch": 0.01274223520042474,
      "grad_norm": 0.7966943666753264,
      "learning_rate": 4.2468239564428313e-05,
      "loss": 12.41,
      "step": 234
    },
    {
      "epoch": 0.012796689197007752,
      "grad_norm": 0.6898793732291435,
      "learning_rate": 4.2649727767695105e-05,
      "loss": 12.1404,
      "step": 235
    },
    {
      "epoch": 0.012851143193590764,
      "grad_norm": 0.7640340826855201,
      "learning_rate": 4.283121597096189e-05,
      "loss": 12.2975,
      "step": 236
    },
    {
      "epoch": 0.012905597190173776,
      "grad_norm": 0.7512000166482308,
      "learning_rate": 4.301270417422868e-05,
      "loss": 12.483,
      "step": 237
    },
    {
      "epoch": 0.012960051186756787,
      "grad_norm": 0.7378596174200351,
      "learning_rate": 4.3194192377495465e-05,
      "loss": 12.3202,
      "step": 238
    },
    {
      "epoch": 0.013014505183339799,
      "grad_norm": 0.813840093552659,
      "learning_rate": 4.337568058076225e-05,
      "loss": 12.3155,
      "step": 239
    },
    {
      "epoch": 0.01306895917992281,
      "grad_norm": 0.7077202129525905,
      "learning_rate": 4.355716878402904e-05,
      "loss": 12.331,
      "step": 240
    },
    {
      "epoch": 0.013123413176505823,
      "grad_norm": 0.7432924329815834,
      "learning_rate": 4.373865698729583e-05,
      "loss": 12.4413,
      "step": 241
    },
    {
      "epoch": 0.013177867173088834,
      "grad_norm": 0.8087370274823864,
      "learning_rate": 4.3920145190562616e-05,
      "loss": 12.352,
      "step": 242
    },
    {
      "epoch": 0.013232321169671846,
      "grad_norm": 0.787315390919198,
      "learning_rate": 4.410163339382941e-05,
      "loss": 12.3767,
      "step": 243
    },
    {
      "epoch": 0.013286775166254858,
      "grad_norm": 0.7186308136487785,
      "learning_rate": 4.428312159709619e-05,
      "loss": 12.3972,
      "step": 244
    },
    {
      "epoch": 0.01334122916283787,
      "grad_norm": 0.6962881361865607,
      "learning_rate": 4.4464609800362976e-05,
      "loss": 12.2565,
      "step": 245
    },
    {
      "epoch": 0.013395683159420881,
      "grad_norm": 0.6962482325758493,
      "learning_rate": 4.464609800362977e-05,
      "loss": 12.2827,
      "step": 246
    },
    {
      "epoch": 0.013450137156003893,
      "grad_norm": 0.8134645016161036,
      "learning_rate": 4.482758620689655e-05,
      "loss": 12.3889,
      "step": 247
    },
    {
      "epoch": 0.013504591152586905,
      "grad_norm": 0.7305915380958484,
      "learning_rate": 4.500907441016334e-05,
      "loss": 12.3915,
      "step": 248
    },
    {
      "epoch": 0.013559045149169916,
      "grad_norm": 0.7618719998357277,
      "learning_rate": 4.5190562613430134e-05,
      "loss": 12.3237,
      "step": 249
    },
    {
      "epoch": 0.013613499145752928,
      "grad_norm": 0.7662374563616653,
      "learning_rate": 4.537205081669691e-05,
      "loss": 12.3536,
      "step": 250
    },
    {
      "epoch": 0.01366795314233594,
      "grad_norm": 0.7878655053271634,
      "learning_rate": 4.55535390199637e-05,
      "loss": 12.2811,
      "step": 251
    },
    {
      "epoch": 0.013722407138918952,
      "grad_norm": 0.7351407959136845,
      "learning_rate": 4.5735027223230494e-05,
      "loss": 12.3271,
      "step": 252
    },
    {
      "epoch": 0.013776861135501963,
      "grad_norm": 0.738532069890918,
      "learning_rate": 4.591651542649728e-05,
      "loss": 12.3281,
      "step": 253
    },
    {
      "epoch": 0.013831315132084975,
      "grad_norm": 0.7684920721548087,
      "learning_rate": 4.609800362976407e-05,
      "loss": 12.174,
      "step": 254
    },
    {
      "epoch": 0.013885769128667987,
      "grad_norm": 0.85709477932742,
      "learning_rate": 4.6279491833030854e-05,
      "loss": 12.3123,
      "step": 255
    },
    {
      "epoch": 0.013940223125250999,
      "grad_norm": 0.7699276923675464,
      "learning_rate": 4.646098003629764e-05,
      "loss": 12.3149,
      "step": 256
    },
    {
      "epoch": 0.01399467712183401,
      "grad_norm": 0.7252363551142195,
      "learning_rate": 4.664246823956443e-05,
      "loss": 12.2997,
      "step": 257
    },
    {
      "epoch": 0.014049131118417022,
      "grad_norm": 0.7307344287538694,
      "learning_rate": 4.682395644283122e-05,
      "loss": 12.3743,
      "step": 258
    },
    {
      "epoch": 0.014103585115000034,
      "grad_norm": 0.7688745278021739,
      "learning_rate": 4.7005444646098005e-05,
      "loss": 12.2888,
      "step": 259
    },
    {
      "epoch": 0.014158039111583046,
      "grad_norm": 0.7519190643126366,
      "learning_rate": 4.7186932849364796e-05,
      "loss": 12.2947,
      "step": 260
    },
    {
      "epoch": 0.014212493108166057,
      "grad_norm": 0.7811339132345416,
      "learning_rate": 4.736842105263158e-05,
      "loss": 12.462,
      "step": 261
    },
    {
      "epoch": 0.014266947104749069,
      "grad_norm": 0.7115737251169268,
      "learning_rate": 4.7549909255898365e-05,
      "loss": 12.3242,
      "step": 262
    },
    {
      "epoch": 0.01432140110133208,
      "grad_norm": 0.7507844491490189,
      "learning_rate": 4.7731397459165156e-05,
      "loss": 12.2234,
      "step": 263
    },
    {
      "epoch": 0.014375855097915093,
      "grad_norm": 0.7610821598373387,
      "learning_rate": 4.791288566243195e-05,
      "loss": 12.4286,
      "step": 264
    },
    {
      "epoch": 0.014430309094498104,
      "grad_norm": 0.8770692142481441,
      "learning_rate": 4.809437386569873e-05,
      "loss": 12.5179,
      "step": 265
    },
    {
      "epoch": 0.014484763091081116,
      "grad_norm": 0.7994207367776012,
      "learning_rate": 4.827586206896552e-05,
      "loss": 12.3988,
      "step": 266
    },
    {
      "epoch": 0.014539217087664128,
      "grad_norm": 0.7317523430466678,
      "learning_rate": 4.845735027223231e-05,
      "loss": 12.3898,
      "step": 267
    },
    {
      "epoch": 0.01459367108424714,
      "grad_norm": 0.7656122484229141,
      "learning_rate": 4.863883847549909e-05,
      "loss": 12.0056,
      "step": 268
    },
    {
      "epoch": 0.014648125080830151,
      "grad_norm": 0.7986586703788588,
      "learning_rate": 4.882032667876588e-05,
      "loss": 12.3948,
      "step": 269
    },
    {
      "epoch": 0.014702579077413163,
      "grad_norm": 0.7657932505480529,
      "learning_rate": 4.900181488203267e-05,
      "loss": 12.3687,
      "step": 270
    },
    {
      "epoch": 0.014757033073996175,
      "grad_norm": 0.7099680506665171,
      "learning_rate": 4.918330308529946e-05,
      "loss": 12.2355,
      "step": 271
    },
    {
      "epoch": 0.014811487070579187,
      "grad_norm": 0.6931982785581737,
      "learning_rate": 4.936479128856625e-05,
      "loss": 12.3633,
      "step": 272
    },
    {
      "epoch": 0.014865941067162198,
      "grad_norm": 0.7474303399881781,
      "learning_rate": 4.954627949183303e-05,
      "loss": 12.3746,
      "step": 273
    },
    {
      "epoch": 0.01492039506374521,
      "grad_norm": 0.862481964480327,
      "learning_rate": 4.972776769509982e-05,
      "loss": 12.4141,
      "step": 274
    },
    {
      "epoch": 0.014974849060328222,
      "grad_norm": 0.7318364097265114,
      "learning_rate": 4.990925589836661e-05,
      "loss": 12.2247,
      "step": 275
    },
    {
      "epoch": 0.015029303056911233,
      "grad_norm": 0.734011719238498,
      "learning_rate": 5.009074410163339e-05,
      "loss": 12.3864,
      "step": 276
    },
    {
      "epoch": 0.015083757053494245,
      "grad_norm": 0.7736991802041899,
      "learning_rate": 5.027223230490018e-05,
      "loss": 12.3802,
      "step": 277
    },
    {
      "epoch": 0.015138211050077257,
      "grad_norm": 0.7148902208446675,
      "learning_rate": 5.045372050816697e-05,
      "loss": 12.3205,
      "step": 278
    },
    {
      "epoch": 0.015192665046660269,
      "grad_norm": 0.7319645221512004,
      "learning_rate": 5.0635208711433755e-05,
      "loss": 12.3342,
      "step": 279
    },
    {
      "epoch": 0.01524711904324328,
      "grad_norm": 0.7299722315856292,
      "learning_rate": 5.0816696914700546e-05,
      "loss": 12.3347,
      "step": 280
    },
    {
      "epoch": 0.015301573039826292,
      "grad_norm": 0.7296940630372005,
      "learning_rate": 5.099818511796734e-05,
      "loss": 12.3668,
      "step": 281
    },
    {
      "epoch": 0.015356027036409304,
      "grad_norm": 0.7073166461717706,
      "learning_rate": 5.117967332123412e-05,
      "loss": 12.3703,
      "step": 282
    },
    {
      "epoch": 0.015410481032992316,
      "grad_norm": 0.7546793817999312,
      "learning_rate": 5.136116152450091e-05,
      "loss": 12.3491,
      "step": 283
    },
    {
      "epoch": 0.015464935029575327,
      "grad_norm": 0.685017972082848,
      "learning_rate": 5.15426497277677e-05,
      "loss": 12.2707,
      "step": 284
    },
    {
      "epoch": 0.01551938902615834,
      "grad_norm": 0.7562946426920689,
      "learning_rate": 5.172413793103449e-05,
      "loss": 12.3115,
      "step": 285
    },
    {
      "epoch": 0.015573843022741351,
      "grad_norm": 0.7216084563358899,
      "learning_rate": 5.190562613430128e-05,
      "loss": 12.3921,
      "step": 286
    },
    {
      "epoch": 0.01562829701932436,
      "grad_norm": 0.6939190861763674,
      "learning_rate": 5.2087114337568064e-05,
      "loss": 12.2032,
      "step": 287
    },
    {
      "epoch": 0.015682751015907374,
      "grad_norm": 0.7239626701676075,
      "learning_rate": 5.226860254083484e-05,
      "loss": 12.2799,
      "step": 288
    },
    {
      "epoch": 0.015737205012490384,
      "grad_norm": 0.7753868516243451,
      "learning_rate": 5.245009074410163e-05,
      "loss": 12.4619,
      "step": 289
    },
    {
      "epoch": 0.015791659009073398,
      "grad_norm": 0.7514557376174034,
      "learning_rate": 5.2631578947368424e-05,
      "loss": 12.4112,
      "step": 290
    },
    {
      "epoch": 0.015846113005656408,
      "grad_norm": 0.7377806576725301,
      "learning_rate": 5.281306715063521e-05,
      "loss": 12.3133,
      "step": 291
    },
    {
      "epoch": 0.01590056700223942,
      "grad_norm": 0.7454703909220026,
      "learning_rate": 5.2994555353902e-05,
      "loss": 12.2604,
      "step": 292
    },
    {
      "epoch": 0.01595502099882243,
      "grad_norm": 0.7425978328340237,
      "learning_rate": 5.3176043557168784e-05,
      "loss": 12.3337,
      "step": 293
    },
    {
      "epoch": 0.016009474995405445,
      "grad_norm": 0.8121816159096134,
      "learning_rate": 5.3357531760435575e-05,
      "loss": 12.3189,
      "step": 294
    },
    {
      "epoch": 0.016063928991988455,
      "grad_norm": 0.874090569506292,
      "learning_rate": 5.3539019963702366e-05,
      "loss": 12.4024,
      "step": 295
    },
    {
      "epoch": 0.01611838298857147,
      "grad_norm": 0.766723150191294,
      "learning_rate": 5.372050816696915e-05,
      "loss": 12.3427,
      "step": 296
    },
    {
      "epoch": 0.01617283698515448,
      "grad_norm": 0.7533549696517355,
      "learning_rate": 5.390199637023594e-05,
      "loss": 12.1945,
      "step": 297
    },
    {
      "epoch": 0.016227290981737492,
      "grad_norm": 0.82295747879306,
      "learning_rate": 5.4083484573502726e-05,
      "loss": 12.3395,
      "step": 298
    },
    {
      "epoch": 0.016281744978320502,
      "grad_norm": 0.7763144627979807,
      "learning_rate": 5.4264972776769504e-05,
      "loss": 12.4658,
      "step": 299
    },
    {
      "epoch": 0.016336198974903515,
      "grad_norm": 0.7803622453108773,
      "learning_rate": 5.4446460980036295e-05,
      "loss": 12.1774,
      "step": 300
    },
    {
      "epoch": 0.016390652971486525,
      "grad_norm": 0.7711036795355979,
      "learning_rate": 5.4627949183303086e-05,
      "loss": 12.3421,
      "step": 301
    },
    {
      "epoch": 0.01644510696806954,
      "grad_norm": 0.7862864570197541,
      "learning_rate": 5.480943738656987e-05,
      "loss": 12.44,
      "step": 302
    },
    {
      "epoch": 0.01649956096465255,
      "grad_norm": 0.7174141772095207,
      "learning_rate": 5.499092558983666e-05,
      "loss": 12.3887,
      "step": 303
    },
    {
      "epoch": 0.016554014961235562,
      "grad_norm": 0.8079655442331966,
      "learning_rate": 5.517241379310345e-05,
      "loss": 12.3736,
      "step": 304
    },
    {
      "epoch": 0.016608468957818572,
      "grad_norm": 0.7882892914456249,
      "learning_rate": 5.535390199637024e-05,
      "loss": 12.3937,
      "step": 305
    },
    {
      "epoch": 0.016662922954401586,
      "grad_norm": 0.7739803428465745,
      "learning_rate": 5.553539019963703e-05,
      "loss": 12.3119,
      "step": 306
    },
    {
      "epoch": 0.016717376950984596,
      "grad_norm": 0.8202440006814529,
      "learning_rate": 5.571687840290381e-05,
      "loss": 12.469,
      "step": 307
    },
    {
      "epoch": 0.01677183094756761,
      "grad_norm": 0.6948421754884823,
      "learning_rate": 5.5898366606170604e-05,
      "loss": 12.38,
      "step": 308
    },
    {
      "epoch": 0.01682628494415062,
      "grad_norm": 0.7278638745473617,
      "learning_rate": 5.6079854809437396e-05,
      "loss": 12.3829,
      "step": 309
    },
    {
      "epoch": 0.016880738940733633,
      "grad_norm": 0.8078662521778953,
      "learning_rate": 5.626134301270418e-05,
      "loss": 12.4609,
      "step": 310
    },
    {
      "epoch": 0.016935192937316643,
      "grad_norm": 0.8163381416660692,
      "learning_rate": 5.644283121597096e-05,
      "loss": 12.296,
      "step": 311
    },
    {
      "epoch": 0.016989646933899656,
      "grad_norm": 0.7499116355114528,
      "learning_rate": 5.662431941923775e-05,
      "loss": 12.2917,
      "step": 312
    },
    {
      "epoch": 0.017044100930482666,
      "grad_norm": 0.8559513637122415,
      "learning_rate": 5.680580762250453e-05,
      "loss": 12.3777,
      "step": 313
    },
    {
      "epoch": 0.01709855492706568,
      "grad_norm": 0.8186258240775369,
      "learning_rate": 5.6987295825771325e-05,
      "loss": 12.449,
      "step": 314
    },
    {
      "epoch": 0.01715300892364869,
      "grad_norm": 0.7314385570608495,
      "learning_rate": 5.7168784029038116e-05,
      "loss": 12.2357,
      "step": 315
    },
    {
      "epoch": 0.017207462920231703,
      "grad_norm": 0.8568191729204323,
      "learning_rate": 5.73502722323049e-05,
      "loss": 12.5074,
      "step": 316
    },
    {
      "epoch": 0.017261916916814713,
      "grad_norm": 0.8062065360577912,
      "learning_rate": 5.753176043557169e-05,
      "loss": 12.357,
      "step": 317
    },
    {
      "epoch": 0.017316370913397727,
      "grad_norm": 0.7803665817022063,
      "learning_rate": 5.771324863883848e-05,
      "loss": 12.4644,
      "step": 318
    },
    {
      "epoch": 0.017370824909980737,
      "grad_norm": 0.742566393860498,
      "learning_rate": 5.789473684210527e-05,
      "loss": 12.3906,
      "step": 319
    },
    {
      "epoch": 0.01742527890656375,
      "grad_norm": 0.7289345098896403,
      "learning_rate": 5.807622504537206e-05,
      "loss": 12.3417,
      "step": 320
    },
    {
      "epoch": 0.01747973290314676,
      "grad_norm": 0.7808326632555322,
      "learning_rate": 5.825771324863884e-05,
      "loss": 12.3817,
      "step": 321
    },
    {
      "epoch": 0.017534186899729774,
      "grad_norm": 0.8249973000669343,
      "learning_rate": 5.8439201451905634e-05,
      "loss": 12.2332,
      "step": 322
    },
    {
      "epoch": 0.017588640896312784,
      "grad_norm": 0.7574058668425114,
      "learning_rate": 5.862068965517241e-05,
      "loss": 12.2956,
      "step": 323
    },
    {
      "epoch": 0.017643094892895797,
      "grad_norm": 0.8298624360645479,
      "learning_rate": 5.88021778584392e-05,
      "loss": 12.36,
      "step": 324
    },
    {
      "epoch": 0.017697548889478807,
      "grad_norm": 0.8547904402189499,
      "learning_rate": 5.898366606170599e-05,
      "loss": 12.3879,
      "step": 325
    },
    {
      "epoch": 0.01775200288606182,
      "grad_norm": 0.7968314498228707,
      "learning_rate": 5.916515426497278e-05,
      "loss": 12.46,
      "step": 326
    },
    {
      "epoch": 0.01780645688264483,
      "grad_norm": 0.8736437654956223,
      "learning_rate": 5.934664246823956e-05,
      "loss": 12.4407,
      "step": 327
    },
    {
      "epoch": 0.017860910879227844,
      "grad_norm": 0.8668907429370268,
      "learning_rate": 5.9528130671506354e-05,
      "loss": 12.4397,
      "step": 328
    },
    {
      "epoch": 0.017915364875810854,
      "grad_norm": 0.7818621779531549,
      "learning_rate": 5.9709618874773145e-05,
      "loss": 12.4018,
      "step": 329
    },
    {
      "epoch": 0.017969818872393867,
      "grad_norm": 0.9103192243585209,
      "learning_rate": 5.989110707803993e-05,
      "loss": 12.5196,
      "step": 330
    },
    {
      "epoch": 0.018024272868976877,
      "grad_norm": 0.7862412367698574,
      "learning_rate": 6.007259528130672e-05,
      "loss": 12.3264,
      "step": 331
    },
    {
      "epoch": 0.018078726865559887,
      "grad_norm": 0.7143113953678407,
      "learning_rate": 6.025408348457351e-05,
      "loss": 12.4047,
      "step": 332
    },
    {
      "epoch": 0.0181331808621429,
      "grad_norm": 0.9052912116449027,
      "learning_rate": 6.0435571687840296e-05,
      "loss": 12.3172,
      "step": 333
    },
    {
      "epoch": 0.01818763485872591,
      "grad_norm": 0.7947826092807085,
      "learning_rate": 6.061705989110709e-05,
      "loss": 12.3744,
      "step": 334
    },
    {
      "epoch": 0.018242088855308924,
      "grad_norm": 0.7545460389607728,
      "learning_rate": 6.0798548094373865e-05,
      "loss": 12.1713,
      "step": 335
    },
    {
      "epoch": 0.018296542851891934,
      "grad_norm": 0.7152244757749424,
      "learning_rate": 6.098003629764065e-05,
      "loss": 12.3317,
      "step": 336
    },
    {
      "epoch": 0.018350996848474948,
      "grad_norm": 0.7996572062771121,
      "learning_rate": 6.116152450090745e-05,
      "loss": 12.4122,
      "step": 337
    },
    {
      "epoch": 0.018405450845057958,
      "grad_norm": 0.8411178606002937,
      "learning_rate": 6.134301270417423e-05,
      "loss": 12.3052,
      "step": 338
    },
    {
      "epoch": 0.01845990484164097,
      "grad_norm": 0.7975036289012045,
      "learning_rate": 6.152450090744102e-05,
      "loss": 12.4601,
      "step": 339
    },
    {
      "epoch": 0.01851435883822398,
      "grad_norm": 0.9137429350223257,
      "learning_rate": 6.17059891107078e-05,
      "loss": 12.3111,
      "step": 340
    },
    {
      "epoch": 0.018568812834806995,
      "grad_norm": 0.7148462961473729,
      "learning_rate": 6.18874773139746e-05,
      "loss": 12.4212,
      "step": 341
    },
    {
      "epoch": 0.018623266831390005,
      "grad_norm": 0.6913217865933418,
      "learning_rate": 6.206896551724138e-05,
      "loss": 12.3248,
      "step": 342
    },
    {
      "epoch": 0.01867772082797302,
      "grad_norm": 0.8838781205805315,
      "learning_rate": 6.225045372050817e-05,
      "loss": 12.3943,
      "step": 343
    },
    {
      "epoch": 0.01873217482455603,
      "grad_norm": 0.7873429833167643,
      "learning_rate": 6.243194192377497e-05,
      "loss": 12.3052,
      "step": 344
    },
    {
      "epoch": 0.018786628821139042,
      "grad_norm": 0.6852179867320145,
      "learning_rate": 6.261343012704175e-05,
      "loss": 12.3556,
      "step": 345
    },
    {
      "epoch": 0.018841082817722052,
      "grad_norm": 0.8621452861175046,
      "learning_rate": 6.279491833030852e-05,
      "loss": 12.5934,
      "step": 346
    },
    {
      "epoch": 0.018895536814305065,
      "grad_norm": 0.9120176242511238,
      "learning_rate": 6.297640653357532e-05,
      "loss": 12.4076,
      "step": 347
    },
    {
      "epoch": 0.018949990810888075,
      "grad_norm": 0.7843427231761305,
      "learning_rate": 6.31578947368421e-05,
      "loss": 12.3899,
      "step": 348
    },
    {
      "epoch": 0.01900444480747109,
      "grad_norm": 0.7893170140012902,
      "learning_rate": 6.333938294010889e-05,
      "loss": 12.2412,
      "step": 349
    },
    {
      "epoch": 0.0190588988040541,
      "grad_norm": 0.8916646848287791,
      "learning_rate": 6.352087114337569e-05,
      "loss": 12.4694,
      "step": 350
    },
    {
      "epoch": 0.019113352800637112,
      "grad_norm": 0.7258088717745435,
      "learning_rate": 6.370235934664247e-05,
      "loss": 12.2489,
      "step": 351
    },
    {
      "epoch": 0.019167806797220122,
      "grad_norm": 0.8650617630444332,
      "learning_rate": 6.388384754990925e-05,
      "loss": 12.2905,
      "step": 352
    },
    {
      "epoch": 0.019222260793803136,
      "grad_norm": 0.7667753049211509,
      "learning_rate": 6.406533575317605e-05,
      "loss": 12.3042,
      "step": 353
    },
    {
      "epoch": 0.019276714790386146,
      "grad_norm": 0.7511282035657841,
      "learning_rate": 6.424682395644284e-05,
      "loss": 12.2579,
      "step": 354
    },
    {
      "epoch": 0.01933116878696916,
      "grad_norm": 0.7735404530594452,
      "learning_rate": 6.442831215970962e-05,
      "loss": 12.3486,
      "step": 355
    },
    {
      "epoch": 0.01938562278355217,
      "grad_norm": 0.8412619305074472,
      "learning_rate": 6.460980036297642e-05,
      "loss": 12.4623,
      "step": 356
    },
    {
      "epoch": 0.019440076780135183,
      "grad_norm": 0.7301192343330496,
      "learning_rate": 6.47912885662432e-05,
      "loss": 12.3276,
      "step": 357
    },
    {
      "epoch": 0.019494530776718193,
      "grad_norm": 0.7804652434665313,
      "learning_rate": 6.497277676950997e-05,
      "loss": 12.2476,
      "step": 358
    },
    {
      "epoch": 0.019548984773301206,
      "grad_norm": 0.8556088812883522,
      "learning_rate": 6.515426497277677e-05,
      "loss": 12.529,
      "step": 359
    },
    {
      "epoch": 0.019603438769884216,
      "grad_norm": 0.8768877684878126,
      "learning_rate": 6.533575317604356e-05,
      "loss": 12.4842,
      "step": 360
    },
    {
      "epoch": 0.01965789276646723,
      "grad_norm": 0.7343479317743452,
      "learning_rate": 6.551724137931034e-05,
      "loss": 12.3482,
      "step": 361
    },
    {
      "epoch": 0.01971234676305024,
      "grad_norm": 0.8436152095043282,
      "learning_rate": 6.569872958257714e-05,
      "loss": 12.422,
      "step": 362
    },
    {
      "epoch": 0.019766800759633253,
      "grad_norm": 0.8293501645016546,
      "learning_rate": 6.588021778584392e-05,
      "loss": 12.3398,
      "step": 363
    },
    {
      "epoch": 0.019821254756216263,
      "grad_norm": 0.757977291896263,
      "learning_rate": 6.606170598911071e-05,
      "loss": 12.3762,
      "step": 364
    },
    {
      "epoch": 0.019875708752799277,
      "grad_norm": 0.9108534009229488,
      "learning_rate": 6.62431941923775e-05,
      "loss": 12.5203,
      "step": 365
    },
    {
      "epoch": 0.019930162749382287,
      "grad_norm": 0.861404684253492,
      "learning_rate": 6.642468239564429e-05,
      "loss": 12.3693,
      "step": 366
    },
    {
      "epoch": 0.0199846167459653,
      "grad_norm": 0.7442435951270349,
      "learning_rate": 6.660617059891108e-05,
      "loss": 12.3365,
      "step": 367
    },
    {
      "epoch": 0.02003907074254831,
      "grad_norm": 0.7481686167086928,
      "learning_rate": 6.678765880217786e-05,
      "loss": 12.3823,
      "step": 368
    },
    {
      "epoch": 0.020093524739131324,
      "grad_norm": 0.8866860856512322,
      "learning_rate": 6.696914700544466e-05,
      "loss": 12.3893,
      "step": 369
    },
    {
      "epoch": 0.020147978735714334,
      "grad_norm": 0.7480340530756074,
      "learning_rate": 6.715063520871143e-05,
      "loss": 12.3733,
      "step": 370
    },
    {
      "epoch": 0.020202432732297347,
      "grad_norm": 0.8429189686136943,
      "learning_rate": 6.733212341197823e-05,
      "loss": 12.3659,
      "step": 371
    },
    {
      "epoch": 0.020256886728880357,
      "grad_norm": 0.927639442740603,
      "learning_rate": 6.751361161524501e-05,
      "loss": 12.287,
      "step": 372
    },
    {
      "epoch": 0.02031134072546337,
      "grad_norm": 0.7857422926853126,
      "learning_rate": 6.76950998185118e-05,
      "loss": 12.412,
      "step": 373
    },
    {
      "epoch": 0.02036579472204638,
      "grad_norm": 0.806848985078179,
      "learning_rate": 6.787658802177859e-05,
      "loss": 12.3987,
      "step": 374
    },
    {
      "epoch": 0.020420248718629394,
      "grad_norm": 0.9004884915515702,
      "learning_rate": 6.805807622504538e-05,
      "loss": 12.3248,
      "step": 375
    },
    {
      "epoch": 0.020474702715212404,
      "grad_norm": 0.8136907332825543,
      "learning_rate": 6.823956442831216e-05,
      "loss": 12.5514,
      "step": 376
    },
    {
      "epoch": 0.020529156711795418,
      "grad_norm": 0.7947680138594297,
      "learning_rate": 6.842105263157895e-05,
      "loss": 12.3048,
      "step": 377
    },
    {
      "epoch": 0.020583610708378428,
      "grad_norm": 0.787939095178154,
      "learning_rate": 6.860254083484574e-05,
      "loss": 12.5616,
      "step": 378
    },
    {
      "epoch": 0.02063806470496144,
      "grad_norm": 0.8053644554897705,
      "learning_rate": 6.878402903811253e-05,
      "loss": 12.342,
      "step": 379
    },
    {
      "epoch": 0.02069251870154445,
      "grad_norm": 0.7333377537038525,
      "learning_rate": 6.896551724137931e-05,
      "loss": 12.346,
      "step": 380
    },
    {
      "epoch": 0.020746972698127465,
      "grad_norm": 0.8115592545530859,
      "learning_rate": 6.914700544464611e-05,
      "loss": 12.5671,
      "step": 381
    },
    {
      "epoch": 0.020801426694710475,
      "grad_norm": 0.9025245777577485,
      "learning_rate": 6.932849364791288e-05,
      "loss": 12.2945,
      "step": 382
    },
    {
      "epoch": 0.020855880691293488,
      "grad_norm": 0.7272441731724826,
      "learning_rate": 6.950998185117967e-05,
      "loss": 12.2068,
      "step": 383
    },
    {
      "epoch": 0.020910334687876498,
      "grad_norm": 0.8508505713825312,
      "learning_rate": 6.969147005444646e-05,
      "loss": 12.4984,
      "step": 384
    },
    {
      "epoch": 0.02096478868445951,
      "grad_norm": 0.7415051803154171,
      "learning_rate": 6.987295825771325e-05,
      "loss": 12.3752,
      "step": 385
    },
    {
      "epoch": 0.02101924268104252,
      "grad_norm": 0.759163767114583,
      "learning_rate": 7.005444646098003e-05,
      "loss": 12.3624,
      "step": 386
    },
    {
      "epoch": 0.021073696677625535,
      "grad_norm": 0.8489341011156437,
      "learning_rate": 7.023593466424683e-05,
      "loss": 12.3969,
      "step": 387
    },
    {
      "epoch": 0.021128150674208545,
      "grad_norm": 0.8462665677785168,
      "learning_rate": 7.041742286751362e-05,
      "loss": 12.2932,
      "step": 388
    },
    {
      "epoch": 0.02118260467079156,
      "grad_norm": 0.75552639338092,
      "learning_rate": 7.05989110707804e-05,
      "loss": 12.3267,
      "step": 389
    },
    {
      "epoch": 0.02123705866737457,
      "grad_norm": 0.8161265444898805,
      "learning_rate": 7.07803992740472e-05,
      "loss": 12.5891,
      "step": 390
    },
    {
      "epoch": 0.021291512663957582,
      "grad_norm": 0.8856078253587745,
      "learning_rate": 7.096188747731398e-05,
      "loss": 12.2956,
      "step": 391
    },
    {
      "epoch": 0.021345966660540592,
      "grad_norm": 0.8152626351928155,
      "learning_rate": 7.114337568058077e-05,
      "loss": 12.3987,
      "step": 392
    },
    {
      "epoch": 0.021400420657123605,
      "grad_norm": 0.8211638500815782,
      "learning_rate": 7.132486388384755e-05,
      "loss": 12.3112,
      "step": 393
    },
    {
      "epoch": 0.021454874653706615,
      "grad_norm": 0.8404912916628079,
      "learning_rate": 7.150635208711434e-05,
      "loss": 12.2929,
      "step": 394
    },
    {
      "epoch": 0.02150932865028963,
      "grad_norm": 0.7378950463318673,
      "learning_rate": 7.168784029038112e-05,
      "loss": 12.4709,
      "step": 395
    },
    {
      "epoch": 0.02156378264687264,
      "grad_norm": 0.7621638941069262,
      "learning_rate": 7.186932849364792e-05,
      "loss": 12.4729,
      "step": 396
    },
    {
      "epoch": 0.021618236643455652,
      "grad_norm": 0.8226698918779496,
      "learning_rate": 7.20508166969147e-05,
      "loss": 12.4705,
      "step": 397
    },
    {
      "epoch": 0.021672690640038662,
      "grad_norm": 0.7052288911047748,
      "learning_rate": 7.223230490018149e-05,
      "loss": 12.4461,
      "step": 398
    },
    {
      "epoch": 0.021727144636621672,
      "grad_norm": 0.7609734401203072,
      "learning_rate": 7.241379310344828e-05,
      "loss": 12.3506,
      "step": 399
    },
    {
      "epoch": 0.021781598633204686,
      "grad_norm": 0.770636318894737,
      "learning_rate": 7.259528130671507e-05,
      "loss": 12.248,
      "step": 400
    },
    {
      "epoch": 0.021836052629787696,
      "grad_norm": 0.7161390931014325,
      "learning_rate": 7.277676950998185e-05,
      "loss": 12.2974,
      "step": 401
    },
    {
      "epoch": 0.02189050662637071,
      "grad_norm": 0.7093602343926348,
      "learning_rate": 7.295825771324865e-05,
      "loss": 12.3475,
      "step": 402
    },
    {
      "epoch": 0.02194496062295372,
      "grad_norm": 0.7846838901586377,
      "learning_rate": 7.313974591651544e-05,
      "loss": 12.4812,
      "step": 403
    },
    {
      "epoch": 0.021999414619536733,
      "grad_norm": 0.7610570252567146,
      "learning_rate": 7.332123411978222e-05,
      "loss": 12.3179,
      "step": 404
    },
    {
      "epoch": 0.022053868616119743,
      "grad_norm": 0.7678646935604143,
      "learning_rate": 7.3502722323049e-05,
      "loss": 12.4395,
      "step": 405
    },
    {
      "epoch": 0.022108322612702756,
      "grad_norm": 0.788227147067854,
      "learning_rate": 7.368421052631579e-05,
      "loss": 12.3643,
      "step": 406
    },
    {
      "epoch": 0.022162776609285766,
      "grad_norm": 0.8555008491336252,
      "learning_rate": 7.386569872958257e-05,
      "loss": 12.5442,
      "step": 407
    },
    {
      "epoch": 0.02221723060586878,
      "grad_norm": 0.7848374239784748,
      "learning_rate": 7.404718693284937e-05,
      "loss": 12.4897,
      "step": 408
    },
    {
      "epoch": 0.02227168460245179,
      "grad_norm": 0.7986911623537468,
      "learning_rate": 7.422867513611616e-05,
      "loss": 12.4207,
      "step": 409
    },
    {
      "epoch": 0.022326138599034803,
      "grad_norm": 0.8578821767689867,
      "learning_rate": 7.441016333938294e-05,
      "loss": 12.406,
      "step": 410
    },
    {
      "epoch": 0.022380592595617813,
      "grad_norm": 0.8579401771922557,
      "learning_rate": 7.459165154264974e-05,
      "loss": 12.4753,
      "step": 411
    },
    {
      "epoch": 0.022435046592200827,
      "grad_norm": 0.7942776463738789,
      "learning_rate": 7.477313974591652e-05,
      "loss": 12.3652,
      "step": 412
    },
    {
      "epoch": 0.022489500588783837,
      "grad_norm": 0.7606530813783086,
      "learning_rate": 7.495462794918331e-05,
      "loss": 12.3301,
      "step": 413
    },
    {
      "epoch": 0.02254395458536685,
      "grad_norm": 0.7820187282722689,
      "learning_rate": 7.513611615245009e-05,
      "loss": 12.2943,
      "step": 414
    },
    {
      "epoch": 0.02259840858194986,
      "grad_norm": 0.8133279352459926,
      "learning_rate": 7.531760435571689e-05,
      "loss": 12.3255,
      "step": 415
    },
    {
      "epoch": 0.022652862578532874,
      "grad_norm": 0.7202675473587253,
      "learning_rate": 7.549909255898367e-05,
      "loss": 12.4437,
      "step": 416
    },
    {
      "epoch": 0.022707316575115884,
      "grad_norm": 0.7250526735890024,
      "learning_rate": 7.568058076225046e-05,
      "loss": 12.4672,
      "step": 417
    },
    {
      "epoch": 0.022761770571698897,
      "grad_norm": 0.7829881797724314,
      "learning_rate": 7.586206896551724e-05,
      "loss": 12.4441,
      "step": 418
    },
    {
      "epoch": 0.022816224568281907,
      "grad_norm": 0.8157466754769012,
      "learning_rate": 7.604355716878403e-05,
      "loss": 12.3637,
      "step": 419
    },
    {
      "epoch": 0.02287067856486492,
      "grad_norm": 0.6878509475121516,
      "learning_rate": 7.622504537205081e-05,
      "loss": 12.265,
      "step": 420
    },
    {
      "epoch": 0.02292513256144793,
      "grad_norm": 0.8133681369767054,
      "learning_rate": 7.640653357531761e-05,
      "loss": 12.4073,
      "step": 421
    },
    {
      "epoch": 0.022979586558030944,
      "grad_norm": 0.7970282881764229,
      "learning_rate": 7.65880217785844e-05,
      "loss": 12.4782,
      "step": 422
    },
    {
      "epoch": 0.023034040554613954,
      "grad_norm": 0.7306816228265234,
      "learning_rate": 7.676950998185118e-05,
      "loss": 12.2968,
      "step": 423
    },
    {
      "epoch": 0.023088494551196968,
      "grad_norm": 0.6634878489767484,
      "learning_rate": 7.695099818511798e-05,
      "loss": 12.43,
      "step": 424
    },
    {
      "epoch": 0.023142948547779978,
      "grad_norm": 0.7106521845447317,
      "learning_rate": 7.713248638838476e-05,
      "loss": 12.2943,
      "step": 425
    },
    {
      "epoch": 0.02319740254436299,
      "grad_norm": 0.6994237071240959,
      "learning_rate": 7.731397459165155e-05,
      "loss": 12.3155,
      "step": 426
    },
    {
      "epoch": 0.023251856540946,
      "grad_norm": 0.8721037875029398,
      "learning_rate": 7.749546279491834e-05,
      "loss": 12.3339,
      "step": 427
    },
    {
      "epoch": 0.023306310537529015,
      "grad_norm": 0.7274699367575759,
      "learning_rate": 7.767695099818511e-05,
      "loss": 12.4143,
      "step": 428
    },
    {
      "epoch": 0.023360764534112025,
      "grad_norm": 0.7122374265862785,
      "learning_rate": 7.78584392014519e-05,
      "loss": 12.1451,
      "step": 429
    },
    {
      "epoch": 0.023415218530695038,
      "grad_norm": 0.8254803830668477,
      "learning_rate": 7.80399274047187e-05,
      "loss": 12.5343,
      "step": 430
    },
    {
      "epoch": 0.023469672527278048,
      "grad_norm": 0.7387526625998517,
      "learning_rate": 7.822141560798548e-05,
      "loss": 12.351,
      "step": 431
    },
    {
      "epoch": 0.02352412652386106,
      "grad_norm": 0.7816935354004675,
      "learning_rate": 7.840290381125227e-05,
      "loss": 12.3752,
      "step": 432
    },
    {
      "epoch": 0.02357858052044407,
      "grad_norm": 0.8477449845456101,
      "learning_rate": 7.858439201451906e-05,
      "loss": 12.4905,
      "step": 433
    },
    {
      "epoch": 0.023633034517027085,
      "grad_norm": 0.7892505674089381,
      "learning_rate": 7.876588021778585e-05,
      "loss": 12.4304,
      "step": 434
    },
    {
      "epoch": 0.023687488513610095,
      "grad_norm": 0.7935738185972664,
      "learning_rate": 7.894736842105263e-05,
      "loss": 12.295,
      "step": 435
    },
    {
      "epoch": 0.02374194251019311,
      "grad_norm": 0.7129426510545228,
      "learning_rate": 7.912885662431943e-05,
      "loss": 12.4046,
      "step": 436
    },
    {
      "epoch": 0.02379639650677612,
      "grad_norm": 0.836938154929551,
      "learning_rate": 7.931034482758621e-05,
      "loss": 12.3877,
      "step": 437
    },
    {
      "epoch": 0.023850850503359132,
      "grad_norm": 1.0221663146021225,
      "learning_rate": 7.9491833030853e-05,
      "loss": 12.3748,
      "step": 438
    },
    {
      "epoch": 0.023905304499942142,
      "grad_norm": 0.9267503989787906,
      "learning_rate": 7.96733212341198e-05,
      "loss": 12.4954,
      "step": 439
    },
    {
      "epoch": 0.023959758496525155,
      "grad_norm": 0.8434026048076346,
      "learning_rate": 7.985480943738657e-05,
      "loss": 12.4728,
      "step": 440
    },
    {
      "epoch": 0.024014212493108165,
      "grad_norm": 0.7878062858231188,
      "learning_rate": 8.003629764065335e-05,
      "loss": 12.3983,
      "step": 441
    },
    {
      "epoch": 0.02406866648969118,
      "grad_norm": 0.710973960203588,
      "learning_rate": 8.021778584392015e-05,
      "loss": 12.3385,
      "step": 442
    },
    {
      "epoch": 0.02412312048627419,
      "grad_norm": 0.7938168455755067,
      "learning_rate": 8.039927404718693e-05,
      "loss": 12.3286,
      "step": 443
    },
    {
      "epoch": 0.024177574482857202,
      "grad_norm": 0.7947193309206884,
      "learning_rate": 8.058076225045372e-05,
      "loss": 12.5106,
      "step": 444
    },
    {
      "epoch": 0.024232028479440212,
      "grad_norm": 0.7584161983962521,
      "learning_rate": 8.076225045372052e-05,
      "loss": 12.3826,
      "step": 445
    },
    {
      "epoch": 0.024286482476023226,
      "grad_norm": 0.8202312029520822,
      "learning_rate": 8.09437386569873e-05,
      "loss": 12.345,
      "step": 446
    },
    {
      "epoch": 0.024340936472606236,
      "grad_norm": 0.7645543641740715,
      "learning_rate": 8.112522686025409e-05,
      "loss": 12.4588,
      "step": 447
    },
    {
      "epoch": 0.02439539046918925,
      "grad_norm": 0.7730517169532368,
      "learning_rate": 8.130671506352088e-05,
      "loss": 12.421,
      "step": 448
    },
    {
      "epoch": 0.02444984446577226,
      "grad_norm": 0.814708665964133,
      "learning_rate": 8.148820326678767e-05,
      "loss": 12.3183,
      "step": 449
    },
    {
      "epoch": 0.024504298462355273,
      "grad_norm": 0.7711615040975552,
      "learning_rate": 8.166969147005445e-05,
      "loss": 12.4889,
      "step": 450
    },
    {
      "epoch": 0.024558752458938283,
      "grad_norm": 0.7528070171512011,
      "learning_rate": 8.185117967332124e-05,
      "loss": 12.4343,
      "step": 451
    },
    {
      "epoch": 0.024613206455521296,
      "grad_norm": 0.8249958258020976,
      "learning_rate": 8.203266787658802e-05,
      "loss": 12.4099,
      "step": 452
    },
    {
      "epoch": 0.024667660452104306,
      "grad_norm": 0.8813431530918817,
      "learning_rate": 8.22141560798548e-05,
      "loss": 12.4739,
      "step": 453
    },
    {
      "epoch": 0.02472211444868732,
      "grad_norm": 0.9494001566035871,
      "learning_rate": 8.23956442831216e-05,
      "loss": 12.4154,
      "step": 454
    },
    {
      "epoch": 0.02477656844527033,
      "grad_norm": 0.7154527838559613,
      "learning_rate": 8.257713248638839e-05,
      "loss": 12.3046,
      "step": 455
    },
    {
      "epoch": 0.024831022441853343,
      "grad_norm": 0.8426756172886806,
      "learning_rate": 8.275862068965517e-05,
      "loss": 12.3878,
      "step": 456
    },
    {
      "epoch": 0.024885476438436353,
      "grad_norm": 0.7868340051489234,
      "learning_rate": 8.294010889292196e-05,
      "loss": 12.4081,
      "step": 457
    },
    {
      "epoch": 0.024939930435019367,
      "grad_norm": 0.7614062874800365,
      "learning_rate": 8.312159709618876e-05,
      "loss": 12.3421,
      "step": 458
    },
    {
      "epoch": 0.024994384431602377,
      "grad_norm": 0.7831978878652606,
      "learning_rate": 8.330308529945554e-05,
      "loss": 12.4271,
      "step": 459
    },
    {
      "epoch": 0.02504883842818539,
      "grad_norm": 0.7742389342308593,
      "learning_rate": 8.348457350272232e-05,
      "loss": 12.6071,
      "step": 460
    },
    {
      "epoch": 0.0251032924247684,
      "grad_norm": 0.7667575505932676,
      "learning_rate": 8.366606170598912e-05,
      "loss": 12.4484,
      "step": 461
    },
    {
      "epoch": 0.025157746421351414,
      "grad_norm": 0.7949954459613233,
      "learning_rate": 8.38475499092559e-05,
      "loss": 12.3982,
      "step": 462
    },
    {
      "epoch": 0.025212200417934424,
      "grad_norm": 0.8839761700709547,
      "learning_rate": 8.402903811252269e-05,
      "loss": 12.3902,
      "step": 463
    },
    {
      "epoch": 0.025266654414517434,
      "grad_norm": 0.7436720063586588,
      "learning_rate": 8.421052631578948e-05,
      "loss": 12.3808,
      "step": 464
    },
    {
      "epoch": 0.025321108411100447,
      "grad_norm": 0.8086165049759212,
      "learning_rate": 8.439201451905626e-05,
      "loss": 12.3742,
      "step": 465
    },
    {
      "epoch": 0.025375562407683457,
      "grad_norm": 0.8217729326448585,
      "learning_rate": 8.457350272232304e-05,
      "loss": 12.3675,
      "step": 466
    },
    {
      "epoch": 0.02543001640426647,
      "grad_norm": 0.8804919476658869,
      "learning_rate": 8.475499092558984e-05,
      "loss": 12.4588,
      "step": 467
    },
    {
      "epoch": 0.02548447040084948,
      "grad_norm": 0.8021891332568722,
      "learning_rate": 8.493647912885663e-05,
      "loss": 12.443,
      "step": 468
    },
    {
      "epoch": 0.025538924397432494,
      "grad_norm": 0.7325722424197364,
      "learning_rate": 8.511796733212341e-05,
      "loss": 12.4545,
      "step": 469
    },
    {
      "epoch": 0.025593378394015504,
      "grad_norm": 0.8306668524452858,
      "learning_rate": 8.529945553539021e-05,
      "loss": 12.5469,
      "step": 470
    },
    {
      "epoch": 0.025647832390598518,
      "grad_norm": 0.7943932754142018,
      "learning_rate": 8.5480943738657e-05,
      "loss": 12.2246,
      "step": 471
    },
    {
      "epoch": 0.025702286387181528,
      "grad_norm": 0.7407462425158048,
      "learning_rate": 8.566243194192378e-05,
      "loss": 12.3846,
      "step": 472
    },
    {
      "epoch": 0.02575674038376454,
      "grad_norm": 0.8711005010892444,
      "learning_rate": 8.584392014519058e-05,
      "loss": 12.5484,
      "step": 473
    },
    {
      "epoch": 0.02581119438034755,
      "grad_norm": 0.7937737358003336,
      "learning_rate": 8.602540834845736e-05,
      "loss": 12.3675,
      "step": 474
    },
    {
      "epoch": 0.025865648376930565,
      "grad_norm": 0.739494251610042,
      "learning_rate": 8.620689655172413e-05,
      "loss": 12.4181,
      "step": 475
    },
    {
      "epoch": 0.025920102373513575,
      "grad_norm": 0.7494105764676788,
      "learning_rate": 8.638838475499093e-05,
      "loss": 12.4294,
      "step": 476
    },
    {
      "epoch": 0.025974556370096588,
      "grad_norm": 0.8112658612116864,
      "learning_rate": 8.656987295825771e-05,
      "loss": 12.3863,
      "step": 477
    },
    {
      "epoch": 0.026029010366679598,
      "grad_norm": 0.7951356603711836,
      "learning_rate": 8.67513611615245e-05,
      "loss": 12.509,
      "step": 478
    },
    {
      "epoch": 0.02608346436326261,
      "grad_norm": 0.8050809097921263,
      "learning_rate": 8.69328493647913e-05,
      "loss": 12.3642,
      "step": 479
    },
    {
      "epoch": 0.02613791835984562,
      "grad_norm": 0.8245382505558528,
      "learning_rate": 8.711433756805808e-05,
      "loss": 12.3835,
      "step": 480
    },
    {
      "epoch": 0.026192372356428635,
      "grad_norm": 0.742494597149955,
      "learning_rate": 8.729582577132486e-05,
      "loss": 12.4518,
      "step": 481
    },
    {
      "epoch": 0.026246826353011645,
      "grad_norm": 0.9027867580588818,
      "learning_rate": 8.747731397459166e-05,
      "loss": 12.3726,
      "step": 482
    },
    {
      "epoch": 0.02630128034959466,
      "grad_norm": 0.7595129443596266,
      "learning_rate": 8.765880217785845e-05,
      "loss": 12.6085,
      "step": 483
    },
    {
      "epoch": 0.02635573434617767,
      "grad_norm": 0.8600852902167964,
      "learning_rate": 8.784029038112523e-05,
      "loss": 12.5357,
      "step": 484
    },
    {
      "epoch": 0.026410188342760682,
      "grad_norm": 0.9430561851745869,
      "learning_rate": 8.802177858439202e-05,
      "loss": 12.5052,
      "step": 485
    },
    {
      "epoch": 0.026464642339343692,
      "grad_norm": 0.7608697999488,
      "learning_rate": 8.820326678765881e-05,
      "loss": 12.3988,
      "step": 486
    },
    {
      "epoch": 0.026519096335926706,
      "grad_norm": 0.9102688809761854,
      "learning_rate": 8.838475499092559e-05,
      "loss": 12.4724,
      "step": 487
    },
    {
      "epoch": 0.026573550332509716,
      "grad_norm": 0.7791000410634494,
      "learning_rate": 8.856624319419238e-05,
      "loss": 12.5351,
      "step": 488
    },
    {
      "epoch": 0.02662800432909273,
      "grad_norm": 0.7489844670021688,
      "learning_rate": 8.874773139745917e-05,
      "loss": 12.2126,
      "step": 489
    },
    {
      "epoch": 0.02668245832567574,
      "grad_norm": 0.8085753500287401,
      "learning_rate": 8.892921960072595e-05,
      "loss": 12.5421,
      "step": 490
    },
    {
      "epoch": 0.026736912322258753,
      "grad_norm": 0.9563889035257984,
      "learning_rate": 8.911070780399275e-05,
      "loss": 12.6741,
      "step": 491
    },
    {
      "epoch": 0.026791366318841763,
      "grad_norm": 0.7063448574074058,
      "learning_rate": 8.929219600725953e-05,
      "loss": 12.2971,
      "step": 492
    },
    {
      "epoch": 0.026845820315424776,
      "grad_norm": 0.7813369753167644,
      "learning_rate": 8.947368421052632e-05,
      "loss": 12.2513,
      "step": 493
    },
    {
      "epoch": 0.026900274312007786,
      "grad_norm": 0.8629014027216114,
      "learning_rate": 8.96551724137931e-05,
      "loss": 12.4148,
      "step": 494
    },
    {
      "epoch": 0.0269547283085908,
      "grad_norm": 0.7481663513315681,
      "learning_rate": 8.98366606170599e-05,
      "loss": 12.5463,
      "step": 495
    },
    {
      "epoch": 0.02700918230517381,
      "grad_norm": 0.8758771305014487,
      "learning_rate": 9.001814882032669e-05,
      "loss": 12.4045,
      "step": 496
    },
    {
      "epoch": 0.027063636301756823,
      "grad_norm": 0.7480406054883233,
      "learning_rate": 9.019963702359347e-05,
      "loss": 12.4557,
      "step": 497
    },
    {
      "epoch": 0.027118090298339833,
      "grad_norm": 0.8152346713462907,
      "learning_rate": 9.038112522686027e-05,
      "loss": 12.3973,
      "step": 498
    },
    {
      "epoch": 0.027172544294922846,
      "grad_norm": 0.8415389004370808,
      "learning_rate": 9.056261343012704e-05,
      "loss": 12.6063,
      "step": 499
    },
    {
      "epoch": 0.027226998291505856,
      "grad_norm": 0.8011366657793219,
      "learning_rate": 9.074410163339382e-05,
      "loss": 12.4955,
      "step": 500
    },
    {
      "epoch": 0.02728145228808887,
      "grad_norm": 0.7380805437115405,
      "learning_rate": 9.092558983666062e-05,
      "loss": 12.3469,
      "step": 501
    },
    {
      "epoch": 0.02733590628467188,
      "grad_norm": 0.7133968415768672,
      "learning_rate": 9.11070780399274e-05,
      "loss": 12.4881,
      "step": 502
    },
    {
      "epoch": 0.027390360281254893,
      "grad_norm": 0.8125970947519109,
      "learning_rate": 9.128856624319419e-05,
      "loss": 12.3151,
      "step": 503
    },
    {
      "epoch": 0.027444814277837903,
      "grad_norm": 0.7077249733189649,
      "learning_rate": 9.147005444646099e-05,
      "loss": 12.3738,
      "step": 504
    },
    {
      "epoch": 0.027499268274420917,
      "grad_norm": 0.7809924557333555,
      "learning_rate": 9.165154264972777e-05,
      "loss": 12.4415,
      "step": 505
    },
    {
      "epoch": 0.027553722271003927,
      "grad_norm": 0.7543662314141218,
      "learning_rate": 9.183303085299456e-05,
      "loss": 12.425,
      "step": 506
    },
    {
      "epoch": 0.02760817626758694,
      "grad_norm": 0.740968620199914,
      "learning_rate": 9.201451905626135e-05,
      "loss": 12.4001,
      "step": 507
    },
    {
      "epoch": 0.02766263026416995,
      "grad_norm": 0.715948782602618,
      "learning_rate": 9.219600725952814e-05,
      "loss": 12.4193,
      "step": 508
    },
    {
      "epoch": 0.027717084260752964,
      "grad_norm": 0.7555285381975604,
      "learning_rate": 9.237749546279492e-05,
      "loss": 12.3583,
      "step": 509
    },
    {
      "epoch": 0.027771538257335974,
      "grad_norm": 0.7926894903352032,
      "learning_rate": 9.255898366606171e-05,
      "loss": 12.5034,
      "step": 510
    },
    {
      "epoch": 0.027825992253918987,
      "grad_norm": 0.722277348946715,
      "learning_rate": 9.274047186932849e-05,
      "loss": 12.4026,
      "step": 511
    },
    {
      "epoch": 0.027880446250501997,
      "grad_norm": 0.8485310235578939,
      "learning_rate": 9.292196007259528e-05,
      "loss": 12.4843,
      "step": 512
    },
    {
      "epoch": 0.02793490024708501,
      "grad_norm": 0.8246964012075222,
      "learning_rate": 9.310344827586207e-05,
      "loss": 12.3967,
      "step": 513
    },
    {
      "epoch": 0.02798935424366802,
      "grad_norm": 0.8152543979538756,
      "learning_rate": 9.328493647912886e-05,
      "loss": 12.26,
      "step": 514
    },
    {
      "epoch": 0.028043808240251034,
      "grad_norm": 0.8555264952532589,
      "learning_rate": 9.346642468239564e-05,
      "loss": 12.3481,
      "step": 515
    },
    {
      "epoch": 0.028098262236834044,
      "grad_norm": 0.7301430730163417,
      "learning_rate": 9.364791288566244e-05,
      "loss": 12.3859,
      "step": 516
    },
    {
      "epoch": 0.028152716233417058,
      "grad_norm": 0.8502343380693529,
      "learning_rate": 9.382940108892923e-05,
      "loss": 12.3946,
      "step": 517
    },
    {
      "epoch": 0.028207170230000068,
      "grad_norm": 0.7616908996355757,
      "learning_rate": 9.401088929219601e-05,
      "loss": 12.4984,
      "step": 518
    },
    {
      "epoch": 0.02826162422658308,
      "grad_norm": 0.848187178061048,
      "learning_rate": 9.419237749546281e-05,
      "loss": 12.5009,
      "step": 519
    },
    {
      "epoch": 0.02831607822316609,
      "grad_norm": 0.7605162192554722,
      "learning_rate": 9.437386569872959e-05,
      "loss": 12.3757,
      "step": 520
    },
    {
      "epoch": 0.028370532219749105,
      "grad_norm": 0.6941160676897854,
      "learning_rate": 9.455535390199638e-05,
      "loss": 12.497,
      "step": 521
    },
    {
      "epoch": 0.028424986216332115,
      "grad_norm": 0.7856867296190392,
      "learning_rate": 9.473684210526316e-05,
      "loss": 12.4029,
      "step": 522
    },
    {
      "epoch": 0.028479440212915128,
      "grad_norm": 0.7048400049802264,
      "learning_rate": 9.491833030852995e-05,
      "loss": 12.4511,
      "step": 523
    },
    {
      "epoch": 0.028533894209498138,
      "grad_norm": 0.8013294724513513,
      "learning_rate": 9.509981851179673e-05,
      "loss": 12.4689,
      "step": 524
    },
    {
      "epoch": 0.02858834820608115,
      "grad_norm": 0.7878969559755142,
      "learning_rate": 9.528130671506353e-05,
      "loss": 12.5229,
      "step": 525
    },
    {
      "epoch": 0.02864280220266416,
      "grad_norm": 0.7987625743494264,
      "learning_rate": 9.546279491833031e-05,
      "loss": 12.3626,
      "step": 526
    },
    {
      "epoch": 0.028697256199247175,
      "grad_norm": 0.8079525903340861,
      "learning_rate": 9.56442831215971e-05,
      "loss": 12.4762,
      "step": 527
    },
    {
      "epoch": 0.028751710195830185,
      "grad_norm": 0.6640448779297942,
      "learning_rate": 9.58257713248639e-05,
      "loss": 12.3718,
      "step": 528
    },
    {
      "epoch": 0.0288061641924132,
      "grad_norm": 0.8399913617521838,
      "learning_rate": 9.600725952813068e-05,
      "loss": 12.342,
      "step": 529
    },
    {
      "epoch": 0.02886061818899621,
      "grad_norm": 0.8352947312719899,
      "learning_rate": 9.618874773139746e-05,
      "loss": 12.4117,
      "step": 530
    },
    {
      "epoch": 0.02891507218557922,
      "grad_norm": 0.7357612574343283,
      "learning_rate": 9.637023593466425e-05,
      "loss": 12.3374,
      "step": 531
    },
    {
      "epoch": 0.028969526182162232,
      "grad_norm": 0.8621291940802033,
      "learning_rate": 9.655172413793105e-05,
      "loss": 12.4102,
      "step": 532
    },
    {
      "epoch": 0.029023980178745242,
      "grad_norm": 0.7496563648684155,
      "learning_rate": 9.673321234119783e-05,
      "loss": 12.553,
      "step": 533
    },
    {
      "epoch": 0.029078434175328256,
      "grad_norm": 0.7178040502697846,
      "learning_rate": 9.691470054446462e-05,
      "loss": 12.3036,
      "step": 534
    },
    {
      "epoch": 0.029132888171911266,
      "grad_norm": 0.7748952398348568,
      "learning_rate": 9.70961887477314e-05,
      "loss": 12.493,
      "step": 535
    },
    {
      "epoch": 0.02918734216849428,
      "grad_norm": 0.7713180455385662,
      "learning_rate": 9.727767695099818e-05,
      "loss": 12.4102,
      "step": 536
    },
    {
      "epoch": 0.02924179616507729,
      "grad_norm": 0.7740687521356002,
      "learning_rate": 9.745916515426497e-05,
      "loss": 12.5748,
      "step": 537
    },
    {
      "epoch": 0.029296250161660303,
      "grad_norm": 0.7865595720640453,
      "learning_rate": 9.764065335753177e-05,
      "loss": 12.4367,
      "step": 538
    },
    {
      "epoch": 0.029350704158243313,
      "grad_norm": 0.7560847556377283,
      "learning_rate": 9.782214156079855e-05,
      "loss": 12.3583,
      "step": 539
    },
    {
      "epoch": 0.029405158154826326,
      "grad_norm": 0.7978359099601086,
      "learning_rate": 9.800362976406534e-05,
      "loss": 12.3999,
      "step": 540
    },
    {
      "epoch": 0.029459612151409336,
      "grad_norm": 0.7535509446044629,
      "learning_rate": 9.818511796733213e-05,
      "loss": 12.3052,
      "step": 541
    },
    {
      "epoch": 0.02951406614799235,
      "grad_norm": 0.7775521738235894,
      "learning_rate": 9.836660617059892e-05,
      "loss": 12.4467,
      "step": 542
    },
    {
      "epoch": 0.02956852014457536,
      "grad_norm": 0.7469699238521446,
      "learning_rate": 9.85480943738657e-05,
      "loss": 12.456,
      "step": 543
    },
    {
      "epoch": 0.029622974141158373,
      "grad_norm": 1.0563298749968926,
      "learning_rate": 9.87295825771325e-05,
      "loss": 12.4247,
      "step": 544
    },
    {
      "epoch": 0.029677428137741383,
      "grad_norm": 0.8361142060905258,
      "learning_rate": 9.891107078039928e-05,
      "loss": 12.5043,
      "step": 545
    },
    {
      "epoch": 0.029731882134324396,
      "grad_norm": 1.0096705276772675,
      "learning_rate": 9.909255898366606e-05,
      "loss": 12.3808,
      "step": 546
    },
    {
      "epoch": 0.029786336130907407,
      "grad_norm": 0.7864328089464631,
      "learning_rate": 9.927404718693285e-05,
      "loss": 12.538,
      "step": 547
    },
    {
      "epoch": 0.02984079012749042,
      "grad_norm": 0.8451585533599,
      "learning_rate": 9.945553539019964e-05,
      "loss": 12.5222,
      "step": 548
    },
    {
      "epoch": 0.02989524412407343,
      "grad_norm": 0.8653338116417337,
      "learning_rate": 9.963702359346642e-05,
      "loss": 12.4985,
      "step": 549
    },
    {
      "epoch": 0.029949698120656443,
      "grad_norm": 0.7662298878370504,
      "learning_rate": 9.981851179673322e-05,
      "loss": 12.4266,
      "step": 550
    },
    {
      "epoch": 0.030004152117239453,
      "grad_norm": 0.7970444594455892,
      "learning_rate": 0.0001,
      "loss": 12.3258,
      "step": 551
    },
    {
      "epoch": 0.030058606113822467,
      "grad_norm": 0.7349918099412,
      "learning_rate": 0.00010018148820326678,
      "loss": 12.4242,
      "step": 552
    },
    {
      "epoch": 0.030113060110405477,
      "grad_norm": 0.854292237492196,
      "learning_rate": 0.00010036297640653359,
      "loss": 12.3636,
      "step": 553
    },
    {
      "epoch": 0.03016751410698849,
      "grad_norm": 0.7395022781358609,
      "learning_rate": 0.00010054446460980036,
      "loss": 12.5779,
      "step": 554
    },
    {
      "epoch": 0.0302219681035715,
      "grad_norm": 0.768148807115022,
      "learning_rate": 0.00010072595281306716,
      "loss": 12.5132,
      "step": 555
    },
    {
      "epoch": 0.030276422100154514,
      "grad_norm": 0.8159993812941795,
      "learning_rate": 0.00010090744101633394,
      "loss": 12.4995,
      "step": 556
    },
    {
      "epoch": 0.030330876096737524,
      "grad_norm": 0.8351323851458299,
      "learning_rate": 0.00010108892921960074,
      "loss": 12.5193,
      "step": 557
    },
    {
      "epoch": 0.030385330093320537,
      "grad_norm": 0.8416946553162289,
      "learning_rate": 0.00010127041742286751,
      "loss": 12.5428,
      "step": 558
    },
    {
      "epoch": 0.030439784089903547,
      "grad_norm": 0.9752655054331537,
      "learning_rate": 0.00010145190562613431,
      "loss": 12.6036,
      "step": 559
    },
    {
      "epoch": 0.03049423808648656,
      "grad_norm": 0.7197724811301942,
      "learning_rate": 0.00010163339382940109,
      "loss": 12.4491,
      "step": 560
    },
    {
      "epoch": 0.03054869208306957,
      "grad_norm": 0.7503723774413532,
      "learning_rate": 0.00010181488203266789,
      "loss": 12.3547,
      "step": 561
    },
    {
      "epoch": 0.030603146079652584,
      "grad_norm": 0.8452417124749888,
      "learning_rate": 0.00010199637023593467,
      "loss": 12.5108,
      "step": 562
    },
    {
      "epoch": 0.030657600076235594,
      "grad_norm": 1.0895902134284718,
      "learning_rate": 0.00010217785843920144,
      "loss": 12.5596,
      "step": 563
    },
    {
      "epoch": 0.030712054072818608,
      "grad_norm": 0.7045960377212881,
      "learning_rate": 0.00010235934664246824,
      "loss": 12.5114,
      "step": 564
    },
    {
      "epoch": 0.030766508069401618,
      "grad_norm": 0.7817959879503318,
      "learning_rate": 0.00010254083484573503,
      "loss": 12.4439,
      "step": 565
    },
    {
      "epoch": 0.03082096206598463,
      "grad_norm": 0.7924652130755838,
      "learning_rate": 0.00010272232304900183,
      "loss": 12.5138,
      "step": 566
    },
    {
      "epoch": 0.03087541606256764,
      "grad_norm": 0.8559510622634047,
      "learning_rate": 0.0001029038112522686,
      "loss": 12.4195,
      "step": 567
    },
    {
      "epoch": 0.030929870059150655,
      "grad_norm": 0.7675767381139006,
      "learning_rate": 0.0001030852994555354,
      "loss": 12.4978,
      "step": 568
    },
    {
      "epoch": 0.030984324055733665,
      "grad_norm": 0.7741050473466831,
      "learning_rate": 0.00010326678765880218,
      "loss": 12.2737,
      "step": 569
    },
    {
      "epoch": 0.03103877805231668,
      "grad_norm": 0.7415387084328533,
      "learning_rate": 0.00010344827586206898,
      "loss": 12.4745,
      "step": 570
    },
    {
      "epoch": 0.03109323204889969,
      "grad_norm": 0.7841681248843312,
      "learning_rate": 0.00010362976406533576,
      "loss": 12.4871,
      "step": 571
    },
    {
      "epoch": 0.031147686045482702,
      "grad_norm": 0.7936691292245915,
      "learning_rate": 0.00010381125226860256,
      "loss": 12.652,
      "step": 572
    },
    {
      "epoch": 0.031202140042065712,
      "grad_norm": 0.8113102552857533,
      "learning_rate": 0.00010399274047186933,
      "loss": 12.5519,
      "step": 573
    },
    {
      "epoch": 0.03125659403864872,
      "grad_norm": 0.7347337505529872,
      "learning_rate": 0.00010417422867513613,
      "loss": 12.3806,
      "step": 574
    },
    {
      "epoch": 0.031311048035231735,
      "grad_norm": 0.746487006322028,
      "learning_rate": 0.00010435571687840291,
      "loss": 12.4832,
      "step": 575
    },
    {
      "epoch": 0.03136550203181475,
      "grad_norm": 0.8489077584446184,
      "learning_rate": 0.00010453720508166968,
      "loss": 12.2913,
      "step": 576
    },
    {
      "epoch": 0.03141995602839776,
      "grad_norm": 0.7185282981457465,
      "learning_rate": 0.00010471869328493648,
      "loss": 12.3301,
      "step": 577
    },
    {
      "epoch": 0.03147441002498077,
      "grad_norm": 0.8659215057338194,
      "learning_rate": 0.00010490018148820327,
      "loss": 12.601,
      "step": 578
    },
    {
      "epoch": 0.03152886402156378,
      "grad_norm": 0.76054791141889,
      "learning_rate": 0.00010508166969147006,
      "loss": 12.3515,
      "step": 579
    },
    {
      "epoch": 0.031583318018146796,
      "grad_norm": 0.8109048061455886,
      "learning_rate": 0.00010526315789473685,
      "loss": 12.5157,
      "step": 580
    },
    {
      "epoch": 0.03163777201472981,
      "grad_norm": 0.7397461059924253,
      "learning_rate": 0.00010544464609800365,
      "loss": 12.4316,
      "step": 581
    },
    {
      "epoch": 0.031692226011312816,
      "grad_norm": 0.7946491784910253,
      "learning_rate": 0.00010562613430127042,
      "loss": 12.3995,
      "step": 582
    },
    {
      "epoch": 0.03174668000789583,
      "grad_norm": 0.8084263112882792,
      "learning_rate": 0.00010580762250453721,
      "loss": 12.4758,
      "step": 583
    },
    {
      "epoch": 0.03180113400447884,
      "grad_norm": 0.7806289943327588,
      "learning_rate": 0.000105989110707804,
      "loss": 12.3552,
      "step": 584
    },
    {
      "epoch": 0.031855588001061856,
      "grad_norm": 0.782911514302089,
      "learning_rate": 0.0001061705989110708,
      "loss": 12.5056,
      "step": 585
    },
    {
      "epoch": 0.03191004199764486,
      "grad_norm": 0.8014304453370548,
      "learning_rate": 0.00010635208711433757,
      "loss": 12.5809,
      "step": 586
    },
    {
      "epoch": 0.031964495994227876,
      "grad_norm": 0.8775607468840907,
      "learning_rate": 0.00010653357531760435,
      "loss": 12.5721,
      "step": 587
    },
    {
      "epoch": 0.03201894999081089,
      "grad_norm": 0.7172825755544577,
      "learning_rate": 0.00010671506352087115,
      "loss": 12.4076,
      "step": 588
    },
    {
      "epoch": 0.0320734039873939,
      "grad_norm": 0.7180909578238679,
      "learning_rate": 0.00010689655172413792,
      "loss": 12.4168,
      "step": 589
    },
    {
      "epoch": 0.03212785798397691,
      "grad_norm": 0.8139633699732863,
      "learning_rate": 0.00010707803992740473,
      "loss": 12.4554,
      "step": 590
    },
    {
      "epoch": 0.03218231198055992,
      "grad_norm": 0.8201333378367557,
      "learning_rate": 0.0001072595281306715,
      "loss": 12.5802,
      "step": 591
    },
    {
      "epoch": 0.03223676597714294,
      "grad_norm": 0.7468982443082229,
      "learning_rate": 0.0001074410163339383,
      "loss": 12.4596,
      "step": 592
    },
    {
      "epoch": 0.03229121997372595,
      "grad_norm": 0.8776024210025977,
      "learning_rate": 0.00010762250453720509,
      "loss": 12.4242,
      "step": 593
    },
    {
      "epoch": 0.03234567397030896,
      "grad_norm": 0.7475469170008233,
      "learning_rate": 0.00010780399274047188,
      "loss": 12.2832,
      "step": 594
    },
    {
      "epoch": 0.03240012796689197,
      "grad_norm": 0.7498393845005602,
      "learning_rate": 0.00010798548094373865,
      "loss": 12.4668,
      "step": 595
    },
    {
      "epoch": 0.032454581963474984,
      "grad_norm": 0.8085189966961024,
      "learning_rate": 0.00010816696914700545,
      "loss": 12.5477,
      "step": 596
    },
    {
      "epoch": 0.03250903596005799,
      "grad_norm": 0.781876240427016,
      "learning_rate": 0.00010834845735027224,
      "loss": 12.3383,
      "step": 597
    },
    {
      "epoch": 0.032563489956641004,
      "grad_norm": 0.7894170574953352,
      "learning_rate": 0.00010852994555353901,
      "loss": 12.505,
      "step": 598
    },
    {
      "epoch": 0.03261794395322402,
      "grad_norm": 0.7576096050152189,
      "learning_rate": 0.00010871143375680582,
      "loss": 12.3167,
      "step": 599
    },
    {
      "epoch": 0.03267239794980703,
      "grad_norm": 0.8250448715677351,
      "learning_rate": 0.00010889292196007259,
      "loss": 12.4931,
      "step": 600
    },
    {
      "epoch": 0.03272685194639004,
      "grad_norm": 0.7191331124925939,
      "learning_rate": 0.00010907441016333939,
      "loss": 12.343,
      "step": 601
    },
    {
      "epoch": 0.03278130594297305,
      "grad_norm": 0.7096756844673696,
      "learning_rate": 0.00010925589836660617,
      "loss": 12.3706,
      "step": 602
    },
    {
      "epoch": 0.032835759939556064,
      "grad_norm": 0.8855413287507164,
      "learning_rate": 0.00010943738656987297,
      "loss": 12.5008,
      "step": 603
    },
    {
      "epoch": 0.03289021393613908,
      "grad_norm": 0.702816878198603,
      "learning_rate": 0.00010961887477313974,
      "loss": 12.4181,
      "step": 604
    },
    {
      "epoch": 0.032944667932722084,
      "grad_norm": 0.7511863541614481,
      "learning_rate": 0.00010980036297640654,
      "loss": 12.4619,
      "step": 605
    },
    {
      "epoch": 0.0329991219293051,
      "grad_norm": 0.7440166154853498,
      "learning_rate": 0.00010998185117967332,
      "loss": 12.5371,
      "step": 606
    },
    {
      "epoch": 0.03305357592588811,
      "grad_norm": 0.7329152168316372,
      "learning_rate": 0.00011016333938294012,
      "loss": 12.4742,
      "step": 607
    },
    {
      "epoch": 0.033108029922471124,
      "grad_norm": 0.736280532598944,
      "learning_rate": 0.0001103448275862069,
      "loss": 12.3711,
      "step": 608
    },
    {
      "epoch": 0.03316248391905413,
      "grad_norm": 0.7413393159584099,
      "learning_rate": 0.0001105263157894737,
      "loss": 12.4609,
      "step": 609
    },
    {
      "epoch": 0.033216937915637144,
      "grad_norm": 0.789987821650849,
      "learning_rate": 0.00011070780399274048,
      "loss": 12.5343,
      "step": 610
    },
    {
      "epoch": 0.03327139191222016,
      "grad_norm": 0.7557116687034697,
      "learning_rate": 0.00011088929219600726,
      "loss": 12.5283,
      "step": 611
    },
    {
      "epoch": 0.03332584590880317,
      "grad_norm": 0.7854327140332928,
      "learning_rate": 0.00011107078039927406,
      "loss": 12.4899,
      "step": 612
    },
    {
      "epoch": 0.03338029990538618,
      "grad_norm": 0.8594935168696665,
      "learning_rate": 0.00011125226860254083,
      "loss": 12.4901,
      "step": 613
    },
    {
      "epoch": 0.03343475390196919,
      "grad_norm": 0.880417761612294,
      "learning_rate": 0.00011143375680580763,
      "loss": 12.5552,
      "step": 614
    },
    {
      "epoch": 0.033489207898552205,
      "grad_norm": 0.7806942770954527,
      "learning_rate": 0.00011161524500907441,
      "loss": 12.6158,
      "step": 615
    },
    {
      "epoch": 0.03354366189513522,
      "grad_norm": 0.8841970711057281,
      "learning_rate": 0.00011179673321234121,
      "loss": 12.4989,
      "step": 616
    },
    {
      "epoch": 0.033598115891718225,
      "grad_norm": 0.7713157169726781,
      "learning_rate": 0.000111978221415608,
      "loss": 12.4892,
      "step": 617
    },
    {
      "epoch": 0.03365256988830124,
      "grad_norm": 0.8324243289561584,
      "learning_rate": 0.00011215970961887479,
      "loss": 12.4635,
      "step": 618
    },
    {
      "epoch": 0.03370702388488425,
      "grad_norm": 0.7807991754630468,
      "learning_rate": 0.00011234119782214156,
      "loss": 12.5345,
      "step": 619
    },
    {
      "epoch": 0.033761477881467265,
      "grad_norm": 0.7865945001071623,
      "learning_rate": 0.00011252268602540836,
      "loss": 12.3791,
      "step": 620
    },
    {
      "epoch": 0.03381593187805027,
      "grad_norm": 0.7872398487857712,
      "learning_rate": 0.00011270417422867514,
      "loss": 12.5145,
      "step": 621
    },
    {
      "epoch": 0.033870385874633285,
      "grad_norm": 0.7885667646810096,
      "learning_rate": 0.00011288566243194192,
      "loss": 12.5641,
      "step": 622
    },
    {
      "epoch": 0.0339248398712163,
      "grad_norm": 0.7741959357450423,
      "learning_rate": 0.00011306715063520871,
      "loss": 12.4703,
      "step": 623
    },
    {
      "epoch": 0.03397929386779931,
      "grad_norm": 0.7564638066325946,
      "learning_rate": 0.0001132486388384755,
      "loss": 12.3965,
      "step": 624
    },
    {
      "epoch": 0.03403374786438232,
      "grad_norm": 0.9519171311859516,
      "learning_rate": 0.0001134301270417423,
      "loss": 12.4289,
      "step": 625
    },
    {
      "epoch": 0.03408820186096533,
      "grad_norm": 0.8107340278897369,
      "learning_rate": 0.00011361161524500907,
      "loss": 12.3782,
      "step": 626
    },
    {
      "epoch": 0.034142655857548346,
      "grad_norm": 0.8481212877491462,
      "learning_rate": 0.00011379310344827588,
      "loss": 12.4214,
      "step": 627
    },
    {
      "epoch": 0.03419710985413136,
      "grad_norm": 0.8650098885796367,
      "learning_rate": 0.00011397459165154265,
      "loss": 12.4615,
      "step": 628
    },
    {
      "epoch": 0.034251563850714366,
      "grad_norm": 0.7457653007753933,
      "learning_rate": 0.00011415607985480945,
      "loss": 12.596,
      "step": 629
    },
    {
      "epoch": 0.03430601784729738,
      "grad_norm": 0.9104614202638347,
      "learning_rate": 0.00011433756805807623,
      "loss": 12.4861,
      "step": 630
    },
    {
      "epoch": 0.03436047184388039,
      "grad_norm": 0.7999052962726083,
      "learning_rate": 0.00011451905626134303,
      "loss": 12.4831,
      "step": 631
    },
    {
      "epoch": 0.034414925840463406,
      "grad_norm": 0.8776766817589621,
      "learning_rate": 0.0001147005444646098,
      "loss": 12.4767,
      "step": 632
    },
    {
      "epoch": 0.03446937983704641,
      "grad_norm": 0.757055231958654,
      "learning_rate": 0.0001148820326678766,
      "loss": 12.4682,
      "step": 633
    },
    {
      "epoch": 0.034523833833629426,
      "grad_norm": 0.7452271311596549,
      "learning_rate": 0.00011506352087114338,
      "loss": 12.4019,
      "step": 634
    },
    {
      "epoch": 0.03457828783021244,
      "grad_norm": 0.8914212821035306,
      "learning_rate": 0.00011524500907441015,
      "loss": 12.5484,
      "step": 635
    },
    {
      "epoch": 0.03463274182679545,
      "grad_norm": 0.8426837477394973,
      "learning_rate": 0.00011542649727767697,
      "loss": 12.6615,
      "step": 636
    },
    {
      "epoch": 0.03468719582337846,
      "grad_norm": 0.8084210678468613,
      "learning_rate": 0.00011560798548094374,
      "loss": 12.3847,
      "step": 637
    },
    {
      "epoch": 0.03474164981996147,
      "grad_norm": 0.8653995572984597,
      "learning_rate": 0.00011578947368421053,
      "loss": 12.5286,
      "step": 638
    },
    {
      "epoch": 0.03479610381654449,
      "grad_norm": 0.7890053357866484,
      "learning_rate": 0.00011597096188747732,
      "loss": 12.4752,
      "step": 639
    },
    {
      "epoch": 0.0348505578131275,
      "grad_norm": 0.8857773461034695,
      "learning_rate": 0.00011615245009074412,
      "loss": 12.537,
      "step": 640
    },
    {
      "epoch": 0.03490501180971051,
      "grad_norm": 0.9416668002877222,
      "learning_rate": 0.00011633393829401089,
      "loss": 12.5413,
      "step": 641
    },
    {
      "epoch": 0.03495946580629352,
      "grad_norm": 0.779793688730529,
      "learning_rate": 0.00011651542649727769,
      "loss": 12.5147,
      "step": 642
    },
    {
      "epoch": 0.035013919802876534,
      "grad_norm": 0.951732854095737,
      "learning_rate": 0.00011669691470054447,
      "loss": 12.684,
      "step": 643
    },
    {
      "epoch": 0.03506837379945955,
      "grad_norm": 0.7967477855943389,
      "learning_rate": 0.00011687840290381127,
      "loss": 12.5166,
      "step": 644
    },
    {
      "epoch": 0.035122827796042554,
      "grad_norm": 0.9520128117212568,
      "learning_rate": 0.00011705989110707805,
      "loss": 12.4273,
      "step": 645
    },
    {
      "epoch": 0.03517728179262557,
      "grad_norm": 0.7204273522718219,
      "learning_rate": 0.00011724137931034482,
      "loss": 12.452,
      "step": 646
    },
    {
      "epoch": 0.03523173578920858,
      "grad_norm": 0.8316091332522132,
      "learning_rate": 0.00011742286751361162,
      "loss": 12.5175,
      "step": 647
    },
    {
      "epoch": 0.035286189785791594,
      "grad_norm": 0.8106043550269486,
      "learning_rate": 0.0001176043557168784,
      "loss": 12.6276,
      "step": 648
    },
    {
      "epoch": 0.0353406437823746,
      "grad_norm": 0.907645004050805,
      "learning_rate": 0.0001177858439201452,
      "loss": 12.5116,
      "step": 649
    },
    {
      "epoch": 0.035395097778957614,
      "grad_norm": 0.8421133812031315,
      "learning_rate": 0.00011796733212341197,
      "loss": 12.5068,
      "step": 650
    },
    {
      "epoch": 0.03544955177554063,
      "grad_norm": 0.8263559879788932,
      "learning_rate": 0.00011814882032667877,
      "loss": 12.5064,
      "step": 651
    },
    {
      "epoch": 0.03550400577212364,
      "grad_norm": 0.8368348477203709,
      "learning_rate": 0.00011833030852994556,
      "loss": 12.6344,
      "step": 652
    },
    {
      "epoch": 0.03555845976870665,
      "grad_norm": 0.8435898015434021,
      "learning_rate": 0.00011851179673321235,
      "loss": 12.5091,
      "step": 653
    },
    {
      "epoch": 0.03561291376528966,
      "grad_norm": 0.8127838512211614,
      "learning_rate": 0.00011869328493647913,
      "loss": 12.5358,
      "step": 654
    },
    {
      "epoch": 0.035667367761872674,
      "grad_norm": 0.844221855096733,
      "learning_rate": 0.00011887477313974594,
      "loss": 12.5023,
      "step": 655
    },
    {
      "epoch": 0.03572182175845569,
      "grad_norm": 0.830609883523681,
      "learning_rate": 0.00011905626134301271,
      "loss": 12.5483,
      "step": 656
    },
    {
      "epoch": 0.035776275755038695,
      "grad_norm": 0.7751682709968774,
      "learning_rate": 0.00011923774954627949,
      "loss": 12.5475,
      "step": 657
    },
    {
      "epoch": 0.03583072975162171,
      "grad_norm": 0.7594651681846976,
      "learning_rate": 0.00011941923774954629,
      "loss": 12.4376,
      "step": 658
    },
    {
      "epoch": 0.03588518374820472,
      "grad_norm": 0.8105747332842143,
      "learning_rate": 0.00011960072595281306,
      "loss": 12.5585,
      "step": 659
    },
    {
      "epoch": 0.035939637744787735,
      "grad_norm": 0.7793088239529253,
      "learning_rate": 0.00011978221415607986,
      "loss": 12.5071,
      "step": 660
    },
    {
      "epoch": 0.03599409174137074,
      "grad_norm": 0.8091456420082649,
      "learning_rate": 0.00011996370235934664,
      "loss": 12.4204,
      "step": 661
    },
    {
      "epoch": 0.036048545737953755,
      "grad_norm": 0.7951004347492463,
      "learning_rate": 0.00012014519056261344,
      "loss": 12.4689,
      "step": 662
    },
    {
      "epoch": 0.03610299973453677,
      "grad_norm": 0.8203582718986129,
      "learning_rate": 0.00012032667876588021,
      "loss": 12.4161,
      "step": 663
    },
    {
      "epoch": 0.036157453731119775,
      "grad_norm": 0.9104629783176926,
      "learning_rate": 0.00012050816696914702,
      "loss": 12.4031,
      "step": 664
    },
    {
      "epoch": 0.03621190772770279,
      "grad_norm": 0.8353893787785487,
      "learning_rate": 0.0001206896551724138,
      "loss": 12.4805,
      "step": 665
    },
    {
      "epoch": 0.0362663617242858,
      "grad_norm": 0.8025972229636333,
      "learning_rate": 0.00012087114337568059,
      "loss": 12.5601,
      "step": 666
    },
    {
      "epoch": 0.036320815720868815,
      "grad_norm": 0.7563712264723594,
      "learning_rate": 0.00012105263157894738,
      "loss": 12.4652,
      "step": 667
    },
    {
      "epoch": 0.03637526971745182,
      "grad_norm": 0.8122372235020665,
      "learning_rate": 0.00012123411978221418,
      "loss": 12.4633,
      "step": 668
    },
    {
      "epoch": 0.036429723714034835,
      "grad_norm": 0.8880903012742153,
      "learning_rate": 0.00012141560798548095,
      "loss": 12.4917,
      "step": 669
    },
    {
      "epoch": 0.03648417771061785,
      "grad_norm": 0.7902079118875632,
      "learning_rate": 0.00012159709618874773,
      "loss": 12.5041,
      "step": 670
    },
    {
      "epoch": 0.03653863170720086,
      "grad_norm": 0.7919532434646256,
      "learning_rate": 0.00012177858439201453,
      "loss": 12.4096,
      "step": 671
    },
    {
      "epoch": 0.03659308570378387,
      "grad_norm": 0.786591506755876,
      "learning_rate": 0.0001219600725952813,
      "loss": 12.5606,
      "step": 672
    },
    {
      "epoch": 0.03664753970036688,
      "grad_norm": 0.8038365184195617,
      "learning_rate": 0.0001221415607985481,
      "loss": 12.4756,
      "step": 673
    },
    {
      "epoch": 0.036701993696949896,
      "grad_norm": 0.770757422573286,
      "learning_rate": 0.0001223230490018149,
      "loss": 12.4893,
      "step": 674
    },
    {
      "epoch": 0.03675644769353291,
      "grad_norm": 0.761612989544494,
      "learning_rate": 0.0001225045372050817,
      "loss": 12.3772,
      "step": 675
    },
    {
      "epoch": 0.036810901690115916,
      "grad_norm": 0.8670862349575074,
      "learning_rate": 0.00012268602540834846,
      "loss": 12.3902,
      "step": 676
    },
    {
      "epoch": 0.03686535568669893,
      "grad_norm": 0.7716718708781327,
      "learning_rate": 0.00012286751361161526,
      "loss": 12.5174,
      "step": 677
    },
    {
      "epoch": 0.03691980968328194,
      "grad_norm": 0.7626179805058609,
      "learning_rate": 0.00012304900181488203,
      "loss": 12.5441,
      "step": 678
    },
    {
      "epoch": 0.036974263679864956,
      "grad_norm": 0.8312158992798632,
      "learning_rate": 0.00012323049001814883,
      "loss": 12.3531,
      "step": 679
    },
    {
      "epoch": 0.03702871767644796,
      "grad_norm": 0.8564858249915114,
      "learning_rate": 0.0001234119782214156,
      "loss": 12.5319,
      "step": 680
    },
    {
      "epoch": 0.037083171673030976,
      "grad_norm": 0.8648959668269681,
      "learning_rate": 0.0001235934664246824,
      "loss": 12.5591,
      "step": 681
    },
    {
      "epoch": 0.03713762566961399,
      "grad_norm": 0.8916475653141226,
      "learning_rate": 0.0001237749546279492,
      "loss": 12.6703,
      "step": 682
    },
    {
      "epoch": 0.037192079666197,
      "grad_norm": 0.9084031257077833,
      "learning_rate": 0.00012395644283121597,
      "loss": 12.5349,
      "step": 683
    },
    {
      "epoch": 0.03724653366278001,
      "grad_norm": 0.8797221532380368,
      "learning_rate": 0.00012413793103448277,
      "loss": 12.4662,
      "step": 684
    },
    {
      "epoch": 0.03730098765936302,
      "grad_norm": 0.8043936491667408,
      "learning_rate": 0.00012431941923774954,
      "loss": 12.5472,
      "step": 685
    },
    {
      "epoch": 0.03735544165594604,
      "grad_norm": 0.817146472456262,
      "learning_rate": 0.00012450090744101634,
      "loss": 12.4383,
      "step": 686
    },
    {
      "epoch": 0.03740989565252905,
      "grad_norm": 0.8114469583343945,
      "learning_rate": 0.00012468239564428313,
      "loss": 12.5172,
      "step": 687
    },
    {
      "epoch": 0.03746434964911206,
      "grad_norm": 0.8326000918570917,
      "learning_rate": 0.00012486388384754993,
      "loss": 12.5178,
      "step": 688
    },
    {
      "epoch": 0.03751880364569507,
      "grad_norm": 0.7723478809577488,
      "learning_rate": 0.0001250453720508167,
      "loss": 12.5362,
      "step": 689
    },
    {
      "epoch": 0.037573257642278084,
      "grad_norm": 0.9410693950400154,
      "learning_rate": 0.0001252268602540835,
      "loss": 12.5571,
      "step": 690
    },
    {
      "epoch": 0.0376277116388611,
      "grad_norm": 0.7561947345119391,
      "learning_rate": 0.00012540834845735027,
      "loss": 12.4258,
      "step": 691
    },
    {
      "epoch": 0.037682165635444104,
      "grad_norm": 0.8418495240004382,
      "learning_rate": 0.00012558983666061704,
      "loss": 12.5777,
      "step": 692
    },
    {
      "epoch": 0.03773661963202712,
      "grad_norm": 0.8746311873777267,
      "learning_rate": 0.00012577132486388387,
      "loss": 12.5951,
      "step": 693
    },
    {
      "epoch": 0.03779107362861013,
      "grad_norm": 0.7625921236694155,
      "learning_rate": 0.00012595281306715064,
      "loss": 12.4046,
      "step": 694
    },
    {
      "epoch": 0.037845527625193144,
      "grad_norm": 0.734823565054105,
      "learning_rate": 0.00012613430127041744,
      "loss": 12.5139,
      "step": 695
    },
    {
      "epoch": 0.03789998162177615,
      "grad_norm": 0.7621770331854553,
      "learning_rate": 0.0001263157894736842,
      "loss": 12.6277,
      "step": 696
    },
    {
      "epoch": 0.037954435618359164,
      "grad_norm": 0.7166086175672672,
      "learning_rate": 0.000126497277676951,
      "loss": 12.4117,
      "step": 697
    },
    {
      "epoch": 0.03800888961494218,
      "grad_norm": 0.7971843612908569,
      "learning_rate": 0.00012667876588021778,
      "loss": 12.4814,
      "step": 698
    },
    {
      "epoch": 0.03806334361152519,
      "grad_norm": 0.7481586297499027,
      "learning_rate": 0.00012686025408348457,
      "loss": 12.4657,
      "step": 699
    },
    {
      "epoch": 0.0381177976081082,
      "grad_norm": 0.7143053642727076,
      "learning_rate": 0.00012704174228675137,
      "loss": 12.3673,
      "step": 700
    },
    {
      "epoch": 0.03817225160469121,
      "grad_norm": 0.8593540767983657,
      "learning_rate": 0.00012722323049001817,
      "loss": 12.5246,
      "step": 701
    },
    {
      "epoch": 0.038226705601274225,
      "grad_norm": 0.7833857321998503,
      "learning_rate": 0.00012740471869328494,
      "loss": 12.5754,
      "step": 702
    },
    {
      "epoch": 0.03828115959785724,
      "grad_norm": 0.7740278489930515,
      "learning_rate": 0.00012758620689655174,
      "loss": 12.6015,
      "step": 703
    },
    {
      "epoch": 0.038335613594440245,
      "grad_norm": 0.8410629541952023,
      "learning_rate": 0.0001277676950998185,
      "loss": 12.6456,
      "step": 704
    },
    {
      "epoch": 0.03839006759102326,
      "grad_norm": 0.8093859296985529,
      "learning_rate": 0.0001279491833030853,
      "loss": 12.4783,
      "step": 705
    },
    {
      "epoch": 0.03844452158760627,
      "grad_norm": 0.7619677944718481,
      "learning_rate": 0.0001281306715063521,
      "loss": 12.4919,
      "step": 706
    },
    {
      "epoch": 0.038498975584189285,
      "grad_norm": 0.7627275451844259,
      "learning_rate": 0.00012831215970961888,
      "loss": 12.4902,
      "step": 707
    },
    {
      "epoch": 0.03855342958077229,
      "grad_norm": 0.7874103603037563,
      "learning_rate": 0.00012849364791288567,
      "loss": 12.4671,
      "step": 708
    },
    {
      "epoch": 0.038607883577355305,
      "grad_norm": 0.8572838287744005,
      "learning_rate": 0.00012867513611615244,
      "loss": 12.5543,
      "step": 709
    },
    {
      "epoch": 0.03866233757393832,
      "grad_norm": 0.8810561414897864,
      "learning_rate": 0.00012885662431941924,
      "loss": 12.51,
      "step": 710
    },
    {
      "epoch": 0.03871679157052133,
      "grad_norm": 0.7541640787422855,
      "learning_rate": 0.00012903811252268604,
      "loss": 12.5667,
      "step": 711
    },
    {
      "epoch": 0.03877124556710434,
      "grad_norm": 0.773591066392057,
      "learning_rate": 0.00012921960072595284,
      "loss": 12.3739,
      "step": 712
    },
    {
      "epoch": 0.03882569956368735,
      "grad_norm": 0.7308375378339267,
      "learning_rate": 0.0001294010889292196,
      "loss": 12.5431,
      "step": 713
    },
    {
      "epoch": 0.038880153560270365,
      "grad_norm": 0.8362238065065555,
      "learning_rate": 0.0001295825771324864,
      "loss": 12.5286,
      "step": 714
    },
    {
      "epoch": 0.03893460755685338,
      "grad_norm": 0.8560896444701315,
      "learning_rate": 0.00012976406533575318,
      "loss": 12.6039,
      "step": 715
    },
    {
      "epoch": 0.038989061553436385,
      "grad_norm": 0.7718918195967964,
      "learning_rate": 0.00012994555353901995,
      "loss": 12.6776,
      "step": 716
    },
    {
      "epoch": 0.0390435155500194,
      "grad_norm": 0.8479818318382927,
      "learning_rate": 0.00013012704174228675,
      "loss": 12.4291,
      "step": 717
    },
    {
      "epoch": 0.03909796954660241,
      "grad_norm": 0.8189985279402068,
      "learning_rate": 0.00013030852994555355,
      "loss": 12.4984,
      "step": 718
    },
    {
      "epoch": 0.039152423543185426,
      "grad_norm": 0.8252447601960348,
      "learning_rate": 0.00013049001814882034,
      "loss": 12.5111,
      "step": 719
    },
    {
      "epoch": 0.03920687753976843,
      "grad_norm": 0.8136491572382893,
      "learning_rate": 0.00013067150635208711,
      "loss": 12.5477,
      "step": 720
    },
    {
      "epoch": 0.039261331536351446,
      "grad_norm": 0.741260225544626,
      "learning_rate": 0.0001308529945553539,
      "loss": 12.5397,
      "step": 721
    },
    {
      "epoch": 0.03931578553293446,
      "grad_norm": 1.0188705229204873,
      "learning_rate": 0.00013103448275862068,
      "loss": 12.5885,
      "step": 722
    },
    {
      "epoch": 0.03937023952951747,
      "grad_norm": 0.8344084817037877,
      "learning_rate": 0.00013121597096188748,
      "loss": 12.6934,
      "step": 723
    },
    {
      "epoch": 0.03942469352610048,
      "grad_norm": 0.8182160497160557,
      "learning_rate": 0.00013139745916515428,
      "loss": 12.5394,
      "step": 724
    },
    {
      "epoch": 0.03947914752268349,
      "grad_norm": 0.754364334548919,
      "learning_rate": 0.00013157894736842108,
      "loss": 12.5191,
      "step": 725
    },
    {
      "epoch": 0.039533601519266506,
      "grad_norm": 0.7740832117265269,
      "learning_rate": 0.00013176043557168785,
      "loss": 12.5467,
      "step": 726
    },
    {
      "epoch": 0.03958805551584952,
      "grad_norm": 0.8335032571621467,
      "learning_rate": 0.00013194192377495462,
      "loss": 12.4449,
      "step": 727
    },
    {
      "epoch": 0.039642509512432526,
      "grad_norm": 0.8051614632134781,
      "learning_rate": 0.00013212341197822142,
      "loss": 12.6346,
      "step": 728
    },
    {
      "epoch": 0.03969696350901554,
      "grad_norm": 0.7995481068509528,
      "learning_rate": 0.0001323049001814882,
      "loss": 12.6264,
      "step": 729
    },
    {
      "epoch": 0.03975141750559855,
      "grad_norm": 0.8623273622651532,
      "learning_rate": 0.000132486388384755,
      "loss": 12.5711,
      "step": 730
    },
    {
      "epoch": 0.03980587150218156,
      "grad_norm": 0.8437033455684732,
      "learning_rate": 0.00013266787658802178,
      "loss": 12.5684,
      "step": 731
    },
    {
      "epoch": 0.03986032549876457,
      "grad_norm": 0.9133141357437732,
      "learning_rate": 0.00013284936479128858,
      "loss": 12.602,
      "step": 732
    },
    {
      "epoch": 0.03991477949534759,
      "grad_norm": 0.8189575723503095,
      "learning_rate": 0.00013303085299455535,
      "loss": 12.4223,
      "step": 733
    },
    {
      "epoch": 0.0399692334919306,
      "grad_norm": 0.7768915209961624,
      "learning_rate": 0.00013321234119782215,
      "loss": 12.5371,
      "step": 734
    },
    {
      "epoch": 0.04002368748851361,
      "grad_norm": 0.8949770926241271,
      "learning_rate": 0.00013339382940108892,
      "loss": 12.5584,
      "step": 735
    },
    {
      "epoch": 0.04007814148509662,
      "grad_norm": 0.8251457768920586,
      "learning_rate": 0.00013357531760435572,
      "loss": 12.5211,
      "step": 736
    },
    {
      "epoch": 0.040132595481679634,
      "grad_norm": 1.057308901725991,
      "learning_rate": 0.00013375680580762252,
      "loss": 12.7114,
      "step": 737
    },
    {
      "epoch": 0.04018704947826265,
      "grad_norm": 0.7555355053742839,
      "learning_rate": 0.00013393829401088931,
      "loss": 12.4727,
      "step": 738
    },
    {
      "epoch": 0.040241503474845654,
      "grad_norm": 0.9506351004654765,
      "learning_rate": 0.00013411978221415609,
      "loss": 12.5368,
      "step": 739
    },
    {
      "epoch": 0.04029595747142867,
      "grad_norm": 0.8585509422496327,
      "learning_rate": 0.00013430127041742286,
      "loss": 12.59,
      "step": 740
    },
    {
      "epoch": 0.04035041146801168,
      "grad_norm": 0.8650805842588604,
      "learning_rate": 0.00013448275862068965,
      "loss": 12.4794,
      "step": 741
    },
    {
      "epoch": 0.040404865464594694,
      "grad_norm": 0.8002843337946893,
      "learning_rate": 0.00013466424682395645,
      "loss": 12.3748,
      "step": 742
    },
    {
      "epoch": 0.0404593194611777,
      "grad_norm": 0.7811881804163763,
      "learning_rate": 0.00013484573502722325,
      "loss": 12.5016,
      "step": 743
    },
    {
      "epoch": 0.040513773457760714,
      "grad_norm": 0.8033795374596253,
      "learning_rate": 0.00013502722323049002,
      "loss": 12.5511,
      "step": 744
    },
    {
      "epoch": 0.04056822745434373,
      "grad_norm": 0.7797716186956701,
      "learning_rate": 0.00013520871143375682,
      "loss": 12.6652,
      "step": 745
    },
    {
      "epoch": 0.04062268145092674,
      "grad_norm": 0.8900245517652725,
      "learning_rate": 0.0001353901996370236,
      "loss": 12.5599,
      "step": 746
    },
    {
      "epoch": 0.04067713544750975,
      "grad_norm": 0.942216054700293,
      "learning_rate": 0.0001355716878402904,
      "loss": 12.5546,
      "step": 747
    },
    {
      "epoch": 0.04073158944409276,
      "grad_norm": 1.0359805567146756,
      "learning_rate": 0.00013575317604355719,
      "loss": 12.5748,
      "step": 748
    },
    {
      "epoch": 0.040786043440675775,
      "grad_norm": 1.0075629890341031,
      "learning_rate": 0.00013593466424682398,
      "loss": 12.5473,
      "step": 749
    },
    {
      "epoch": 0.04084049743725879,
      "grad_norm": 0.7558468233619073,
      "learning_rate": 0.00013611615245009076,
      "loss": 12.4132,
      "step": 750
    },
    {
      "epoch": 0.040894951433841795,
      "grad_norm": 0.778620606134502,
      "learning_rate": 0.00013629764065335753,
      "loss": 12.5153,
      "step": 751
    },
    {
      "epoch": 0.04094940543042481,
      "grad_norm": 0.8510205117821609,
      "learning_rate": 0.00013647912885662432,
      "loss": 12.6411,
      "step": 752
    },
    {
      "epoch": 0.04100385942700782,
      "grad_norm": 0.9177016995670307,
      "learning_rate": 0.0001366606170598911,
      "loss": 12.5692,
      "step": 753
    },
    {
      "epoch": 0.041058313423590835,
      "grad_norm": 0.7745160379530536,
      "learning_rate": 0.0001368421052631579,
      "loss": 12.5341,
      "step": 754
    },
    {
      "epoch": 0.04111276742017384,
      "grad_norm": 0.8140311412247214,
      "learning_rate": 0.0001370235934664247,
      "loss": 12.5958,
      "step": 755
    },
    {
      "epoch": 0.041167221416756855,
      "grad_norm": 0.8551770393253671,
      "learning_rate": 0.0001372050816696915,
      "loss": 12.5419,
      "step": 756
    },
    {
      "epoch": 0.04122167541333987,
      "grad_norm": 0.7937937119864299,
      "learning_rate": 0.00013738656987295826,
      "loss": 12.5787,
      "step": 757
    },
    {
      "epoch": 0.04127612940992288,
      "grad_norm": 0.8020893372298973,
      "learning_rate": 0.00013756805807622506,
      "loss": 12.5617,
      "step": 758
    },
    {
      "epoch": 0.04133058340650589,
      "grad_norm": 0.903708509231395,
      "learning_rate": 0.00013774954627949183,
      "loss": 12.5472,
      "step": 759
    },
    {
      "epoch": 0.0413850374030889,
      "grad_norm": 0.8577396403644533,
      "learning_rate": 0.00013793103448275863,
      "loss": 12.5458,
      "step": 760
    },
    {
      "epoch": 0.041439491399671916,
      "grad_norm": 1.091639804694237,
      "learning_rate": 0.00013811252268602542,
      "loss": 12.5178,
      "step": 761
    },
    {
      "epoch": 0.04149394539625493,
      "grad_norm": 0.842781245898835,
      "learning_rate": 0.00013829401088929222,
      "loss": 12.662,
      "step": 762
    },
    {
      "epoch": 0.041548399392837936,
      "grad_norm": 0.9767804515104263,
      "learning_rate": 0.000138475499092559,
      "loss": 12.5239,
      "step": 763
    },
    {
      "epoch": 0.04160285338942095,
      "grad_norm": 0.8484900134788992,
      "learning_rate": 0.00013865698729582576,
      "loss": 12.5559,
      "step": 764
    },
    {
      "epoch": 0.04165730738600396,
      "grad_norm": 0.8415112219861786,
      "learning_rate": 0.00013883847549909256,
      "loss": 12.527,
      "step": 765
    },
    {
      "epoch": 0.041711761382586976,
      "grad_norm": 1.0104772032943936,
      "learning_rate": 0.00013901996370235933,
      "loss": 12.5081,
      "step": 766
    },
    {
      "epoch": 0.04176621537916998,
      "grad_norm": 0.7789779394160212,
      "learning_rate": 0.00013920145190562616,
      "loss": 12.5802,
      "step": 767
    },
    {
      "epoch": 0.041820669375752996,
      "grad_norm": 0.8828806284149939,
      "learning_rate": 0.00013938294010889293,
      "loss": 12.5638,
      "step": 768
    },
    {
      "epoch": 0.04187512337233601,
      "grad_norm": 0.9095358751899403,
      "learning_rate": 0.00013956442831215973,
      "loss": 12.6153,
      "step": 769
    },
    {
      "epoch": 0.04192957736891902,
      "grad_norm": 0.7915235702796843,
      "learning_rate": 0.0001397459165154265,
      "loss": 12.575,
      "step": 770
    },
    {
      "epoch": 0.04198403136550203,
      "grad_norm": 0.9294936050905224,
      "learning_rate": 0.0001399274047186933,
      "loss": 12.4846,
      "step": 771
    },
    {
      "epoch": 0.04203848536208504,
      "grad_norm": 0.749715657221197,
      "learning_rate": 0.00014010889292196007,
      "loss": 12.5267,
      "step": 772
    },
    {
      "epoch": 0.042092939358668056,
      "grad_norm": 0.8036323955706655,
      "learning_rate": 0.00014029038112522686,
      "loss": 12.602,
      "step": 773
    },
    {
      "epoch": 0.04214739335525107,
      "grad_norm": 0.8151381024538288,
      "learning_rate": 0.00014047186932849366,
      "loss": 12.5653,
      "step": 774
    },
    {
      "epoch": 0.042201847351834076,
      "grad_norm": 0.9971274493801566,
      "learning_rate": 0.00014065335753176043,
      "loss": 12.6751,
      "step": 775
    },
    {
      "epoch": 0.04225630134841709,
      "grad_norm": 0.8513572095537125,
      "learning_rate": 0.00014083484573502723,
      "loss": 12.3767,
      "step": 776
    },
    {
      "epoch": 0.0423107553450001,
      "grad_norm": 1.0023920639584978,
      "learning_rate": 0.000141016333938294,
      "loss": 12.7133,
      "step": 777
    },
    {
      "epoch": 0.04236520934158312,
      "grad_norm": 0.8628820897885917,
      "learning_rate": 0.0001411978221415608,
      "loss": 12.5328,
      "step": 778
    },
    {
      "epoch": 0.04241966333816612,
      "grad_norm": 0.7938920881523879,
      "learning_rate": 0.0001413793103448276,
      "loss": 12.3853,
      "step": 779
    },
    {
      "epoch": 0.04247411733474914,
      "grad_norm": 0.8646632555026025,
      "learning_rate": 0.0001415607985480944,
      "loss": 12.6662,
      "step": 780
    },
    {
      "epoch": 0.04252857133133215,
      "grad_norm": 0.8439048939595939,
      "learning_rate": 0.00014174228675136117,
      "loss": 12.3081,
      "step": 781
    },
    {
      "epoch": 0.042583025327915164,
      "grad_norm": 0.7599687354406947,
      "learning_rate": 0.00014192377495462796,
      "loss": 12.4467,
      "step": 782
    },
    {
      "epoch": 0.04263747932449817,
      "grad_norm": 0.8070920519030709,
      "learning_rate": 0.00014210526315789474,
      "loss": 12.531,
      "step": 783
    },
    {
      "epoch": 0.042691933321081184,
      "grad_norm": 0.873705703606423,
      "learning_rate": 0.00014228675136116153,
      "loss": 12.4829,
      "step": 784
    },
    {
      "epoch": 0.0427463873176642,
      "grad_norm": 0.8746134110818439,
      "learning_rate": 0.00014246823956442833,
      "loss": 12.5633,
      "step": 785
    },
    {
      "epoch": 0.04280084131424721,
      "grad_norm": 0.8003175609249429,
      "learning_rate": 0.0001426497277676951,
      "loss": 12.6093,
      "step": 786
    },
    {
      "epoch": 0.04285529531083022,
      "grad_norm": 0.9226227234251844,
      "learning_rate": 0.0001428312159709619,
      "loss": 12.7582,
      "step": 787
    },
    {
      "epoch": 0.04290974930741323,
      "grad_norm": 0.774081515279731,
      "learning_rate": 0.00014301270417422867,
      "loss": 12.4773,
      "step": 788
    },
    {
      "epoch": 0.042964203303996244,
      "grad_norm": 0.7770717492469534,
      "learning_rate": 0.00014319419237749547,
      "loss": 12.5703,
      "step": 789
    },
    {
      "epoch": 0.04301865730057926,
      "grad_norm": 0.8342543205146827,
      "learning_rate": 0.00014337568058076224,
      "loss": 12.596,
      "step": 790
    },
    {
      "epoch": 0.043073111297162264,
      "grad_norm": 0.8516389169196312,
      "learning_rate": 0.00014355716878402904,
      "loss": 12.5668,
      "step": 791
    },
    {
      "epoch": 0.04312756529374528,
      "grad_norm": 0.7896009308409137,
      "learning_rate": 0.00014373865698729584,
      "loss": 12.519,
      "step": 792
    },
    {
      "epoch": 0.04318201929032829,
      "grad_norm": 0.8052339672016775,
      "learning_rate": 0.00014392014519056263,
      "loss": 12.5517,
      "step": 793
    },
    {
      "epoch": 0.043236473286911305,
      "grad_norm": 0.7530572458315676,
      "learning_rate": 0.0001441016333938294,
      "loss": 12.541,
      "step": 794
    },
    {
      "epoch": 0.04329092728349431,
      "grad_norm": 0.8662017245842865,
      "learning_rate": 0.0001442831215970962,
      "loss": 12.5754,
      "step": 795
    },
    {
      "epoch": 0.043345381280077325,
      "grad_norm": 0.7645406815099872,
      "learning_rate": 0.00014446460980036297,
      "loss": 12.4066,
      "step": 796
    },
    {
      "epoch": 0.04339983527666034,
      "grad_norm": 0.8330667873972226,
      "learning_rate": 0.00014464609800362977,
      "loss": 12.4053,
      "step": 797
    },
    {
      "epoch": 0.043454289273243345,
      "grad_norm": 0.8205108353659064,
      "learning_rate": 0.00014482758620689657,
      "loss": 12.5986,
      "step": 798
    },
    {
      "epoch": 0.04350874326982636,
      "grad_norm": 0.8520375890357685,
      "learning_rate": 0.00014500907441016334,
      "loss": 12.5274,
      "step": 799
    },
    {
      "epoch": 0.04356319726640937,
      "grad_norm": 0.8080865183543499,
      "learning_rate": 0.00014519056261343014,
      "loss": 12.703,
      "step": 800
    },
    {
      "epoch": 0.043617651262992385,
      "grad_norm": 0.8028752755990547,
      "learning_rate": 0.0001453720508166969,
      "loss": 12.4813,
      "step": 801
    },
    {
      "epoch": 0.04367210525957539,
      "grad_norm": 0.7991778930135419,
      "learning_rate": 0.0001455535390199637,
      "loss": 12.4828,
      "step": 802
    },
    {
      "epoch": 0.043726559256158405,
      "grad_norm": 0.7638276877678696,
      "learning_rate": 0.00014573502722323048,
      "loss": 12.4735,
      "step": 803
    },
    {
      "epoch": 0.04378101325274142,
      "grad_norm": 0.818424687275988,
      "learning_rate": 0.0001459165154264973,
      "loss": 12.6986,
      "step": 804
    },
    {
      "epoch": 0.04383546724932443,
      "grad_norm": 1.082495586362405,
      "learning_rate": 0.00014609800362976407,
      "loss": 12.5908,
      "step": 805
    },
    {
      "epoch": 0.04388992124590744,
      "grad_norm": 1.325984013203311,
      "learning_rate": 0.00014627949183303087,
      "loss": 12.6857,
      "step": 806
    },
    {
      "epoch": 0.04394437524249045,
      "grad_norm": 0.8686683284879365,
      "learning_rate": 0.00014646098003629764,
      "loss": 12.4566,
      "step": 807
    },
    {
      "epoch": 0.043998829239073466,
      "grad_norm": 0.8177369162606946,
      "learning_rate": 0.00014664246823956444,
      "loss": 12.5558,
      "step": 808
    },
    {
      "epoch": 0.04405328323565648,
      "grad_norm": 0.7760195423893618,
      "learning_rate": 0.0001468239564428312,
      "loss": 12.4437,
      "step": 809
    },
    {
      "epoch": 0.044107737232239486,
      "grad_norm": 0.8298412976489291,
      "learning_rate": 0.000147005444646098,
      "loss": 12.4522,
      "step": 810
    },
    {
      "epoch": 0.0441621912288225,
      "grad_norm": 0.7607087893488296,
      "learning_rate": 0.0001471869328493648,
      "loss": 12.5662,
      "step": 811
    },
    {
      "epoch": 0.04421664522540551,
      "grad_norm": 0.8810991381778492,
      "learning_rate": 0.00014736842105263158,
      "loss": 12.6585,
      "step": 812
    },
    {
      "epoch": 0.044271099221988526,
      "grad_norm": 0.8024073570168333,
      "learning_rate": 0.00014754990925589838,
      "loss": 12.5763,
      "step": 813
    },
    {
      "epoch": 0.04432555321857153,
      "grad_norm": 0.7715357348992065,
      "learning_rate": 0.00014773139745916515,
      "loss": 12.6189,
      "step": 814
    },
    {
      "epoch": 0.044380007215154546,
      "grad_norm": 0.8995056229545437,
      "learning_rate": 0.00014791288566243195,
      "loss": 12.6572,
      "step": 815
    },
    {
      "epoch": 0.04443446121173756,
      "grad_norm": 0.8248284303171265,
      "learning_rate": 0.00014809437386569874,
      "loss": 12.5962,
      "step": 816
    },
    {
      "epoch": 0.04448891520832057,
      "grad_norm": 0.866524307728336,
      "learning_rate": 0.00014827586206896554,
      "loss": 12.6484,
      "step": 817
    },
    {
      "epoch": 0.04454336920490358,
      "grad_norm": 0.8879113215826534,
      "learning_rate": 0.0001484573502722323,
      "loss": 12.6412,
      "step": 818
    },
    {
      "epoch": 0.04459782320148659,
      "grad_norm": 0.8758200396877941,
      "learning_rate": 0.0001486388384754991,
      "loss": 12.4537,
      "step": 819
    },
    {
      "epoch": 0.044652277198069606,
      "grad_norm": 0.8436526664202744,
      "learning_rate": 0.00014882032667876588,
      "loss": 12.5899,
      "step": 820
    },
    {
      "epoch": 0.04470673119465262,
      "grad_norm": 0.8545141478541453,
      "learning_rate": 0.00014900181488203265,
      "loss": 12.3574,
      "step": 821
    },
    {
      "epoch": 0.044761185191235627,
      "grad_norm": 1.0267167465528357,
      "learning_rate": 0.00014918330308529948,
      "loss": 12.5645,
      "step": 822
    },
    {
      "epoch": 0.04481563918781864,
      "grad_norm": 0.784780075284951,
      "learning_rate": 0.00014936479128856625,
      "loss": 12.4824,
      "step": 823
    },
    {
      "epoch": 0.04487009318440165,
      "grad_norm": 0.9789745719784654,
      "learning_rate": 0.00014954627949183305,
      "loss": 12.8387,
      "step": 824
    },
    {
      "epoch": 0.04492454718098467,
      "grad_norm": 0.8341195518544883,
      "learning_rate": 0.00014972776769509982,
      "loss": 12.5543,
      "step": 825
    },
    {
      "epoch": 0.04497900117756767,
      "grad_norm": 0.869629747247344,
      "learning_rate": 0.00014990925589836661,
      "loss": 12.6488,
      "step": 826
    },
    {
      "epoch": 0.04503345517415069,
      "grad_norm": 0.826578908620228,
      "learning_rate": 0.00015009074410163339,
      "loss": 12.7158,
      "step": 827
    },
    {
      "epoch": 0.0450879091707337,
      "grad_norm": 0.8756522771620907,
      "learning_rate": 0.00015027223230490018,
      "loss": 12.6569,
      "step": 828
    },
    {
      "epoch": 0.045142363167316714,
      "grad_norm": 0.9101434033316181,
      "learning_rate": 0.00015045372050816698,
      "loss": 12.6332,
      "step": 829
    },
    {
      "epoch": 0.04519681716389972,
      "grad_norm": 0.8952132643866115,
      "learning_rate": 0.00015063520871143378,
      "loss": 12.5787,
      "step": 830
    },
    {
      "epoch": 0.045251271160482734,
      "grad_norm": 0.8011479463268877,
      "learning_rate": 0.00015081669691470055,
      "loss": 12.608,
      "step": 831
    },
    {
      "epoch": 0.04530572515706575,
      "grad_norm": 0.8206162307714958,
      "learning_rate": 0.00015099818511796735,
      "loss": 12.4505,
      "step": 832
    },
    {
      "epoch": 0.04536017915364876,
      "grad_norm": 0.832128042031696,
      "learning_rate": 0.00015117967332123412,
      "loss": 12.6066,
      "step": 833
    },
    {
      "epoch": 0.04541463315023177,
      "grad_norm": 0.8683483549303054,
      "learning_rate": 0.00015136116152450092,
      "loss": 12.606,
      "step": 834
    },
    {
      "epoch": 0.04546908714681478,
      "grad_norm": 0.8436085523266798,
      "learning_rate": 0.00015154264972776772,
      "loss": 12.6469,
      "step": 835
    },
    {
      "epoch": 0.045523541143397794,
      "grad_norm": 0.8528405540241153,
      "learning_rate": 0.00015172413793103449,
      "loss": 12.6791,
      "step": 836
    },
    {
      "epoch": 0.04557799513998081,
      "grad_norm": 0.8141272474203417,
      "learning_rate": 0.00015190562613430128,
      "loss": 12.6382,
      "step": 837
    },
    {
      "epoch": 0.045632449136563814,
      "grad_norm": 0.74720840520126,
      "learning_rate": 0.00015208711433756806,
      "loss": 12.5676,
      "step": 838
    },
    {
      "epoch": 0.04568690313314683,
      "grad_norm": 0.7927722643407407,
      "learning_rate": 0.00015226860254083485,
      "loss": 12.5094,
      "step": 839
    },
    {
      "epoch": 0.04574135712972984,
      "grad_norm": 0.8636739467475668,
      "learning_rate": 0.00015245009074410162,
      "loss": 12.6322,
      "step": 840
    },
    {
      "epoch": 0.045795811126312855,
      "grad_norm": 0.7951696899986485,
      "learning_rate": 0.00015263157894736845,
      "loss": 12.5994,
      "step": 841
    },
    {
      "epoch": 0.04585026512289586,
      "grad_norm": 0.8781451899122102,
      "learning_rate": 0.00015281306715063522,
      "loss": 12.7485,
      "step": 842
    },
    {
      "epoch": 0.045904719119478875,
      "grad_norm": 0.7856930125310769,
      "learning_rate": 0.00015299455535390202,
      "loss": 12.6924,
      "step": 843
    },
    {
      "epoch": 0.04595917311606189,
      "grad_norm": 0.8669106113156154,
      "learning_rate": 0.0001531760435571688,
      "loss": 12.756,
      "step": 844
    },
    {
      "epoch": 0.0460136271126449,
      "grad_norm": 0.7645949228985675,
      "learning_rate": 0.00015335753176043556,
      "loss": 12.5389,
      "step": 845
    },
    {
      "epoch": 0.04606808110922791,
      "grad_norm": 0.8798600529095105,
      "learning_rate": 0.00015353901996370236,
      "loss": 12.728,
      "step": 846
    },
    {
      "epoch": 0.04612253510581092,
      "grad_norm": 0.741923863402535,
      "learning_rate": 0.00015372050816696916,
      "loss": 12.5727,
      "step": 847
    },
    {
      "epoch": 0.046176989102393935,
      "grad_norm": 0.8260769631339705,
      "learning_rate": 0.00015390199637023595,
      "loss": 12.6141,
      "step": 848
    },
    {
      "epoch": 0.04623144309897695,
      "grad_norm": 0.8194036460957305,
      "learning_rate": 0.00015408348457350272,
      "loss": 12.5255,
      "step": 849
    },
    {
      "epoch": 0.046285897095559955,
      "grad_norm": 0.8691842016409564,
      "learning_rate": 0.00015426497277676952,
      "loss": 12.665,
      "step": 850
    },
    {
      "epoch": 0.04634035109214297,
      "grad_norm": 0.8265824746983291,
      "learning_rate": 0.0001544464609800363,
      "loss": 12.6078,
      "step": 851
    },
    {
      "epoch": 0.04639480508872598,
      "grad_norm": 0.7402054058609698,
      "learning_rate": 0.0001546279491833031,
      "loss": 12.6317,
      "step": 852
    },
    {
      "epoch": 0.046449259085308996,
      "grad_norm": 0.8385811912548654,
      "learning_rate": 0.0001548094373865699,
      "loss": 12.5836,
      "step": 853
    },
    {
      "epoch": 0.046503713081892,
      "grad_norm": 0.8237170640137944,
      "learning_rate": 0.0001549909255898367,
      "loss": 12.6291,
      "step": 854
    },
    {
      "epoch": 0.046558167078475016,
      "grad_norm": 0.823899899114155,
      "learning_rate": 0.00015517241379310346,
      "loss": 12.4743,
      "step": 855
    },
    {
      "epoch": 0.04661262107505803,
      "grad_norm": 0.7988258901543164,
      "learning_rate": 0.00015535390199637023,
      "loss": 12.4996,
      "step": 856
    },
    {
      "epoch": 0.04666707507164104,
      "grad_norm": 0.809244770017492,
      "learning_rate": 0.00015553539019963703,
      "loss": 12.575,
      "step": 857
    },
    {
      "epoch": 0.04672152906822405,
      "grad_norm": 0.7959350249409557,
      "learning_rate": 0.0001557168784029038,
      "loss": 12.5737,
      "step": 858
    },
    {
      "epoch": 0.04677598306480706,
      "grad_norm": 1.0155065239576782,
      "learning_rate": 0.00015589836660617062,
      "loss": 12.55,
      "step": 859
    },
    {
      "epoch": 0.046830437061390076,
      "grad_norm": 0.9194464079420726,
      "learning_rate": 0.0001560798548094374,
      "loss": 12.551,
      "step": 860
    },
    {
      "epoch": 0.04688489105797308,
      "grad_norm": 0.9893867712993335,
      "learning_rate": 0.0001562613430127042,
      "loss": 12.5736,
      "step": 861
    },
    {
      "epoch": 0.046939345054556096,
      "grad_norm": 0.862259978923311,
      "learning_rate": 0.00015644283121597096,
      "loss": 12.5943,
      "step": 862
    },
    {
      "epoch": 0.04699379905113911,
      "grad_norm": 0.7750438451575176,
      "learning_rate": 0.00015662431941923776,
      "loss": 12.5714,
      "step": 863
    },
    {
      "epoch": 0.04704825304772212,
      "grad_norm": 0.8220468097414089,
      "learning_rate": 0.00015680580762250453,
      "loss": 12.6296,
      "step": 864
    },
    {
      "epoch": 0.04710270704430513,
      "grad_norm": 0.8713481011755143,
      "learning_rate": 0.00015698729582577133,
      "loss": 12.5893,
      "step": 865
    },
    {
      "epoch": 0.04715716104088814,
      "grad_norm": 0.9801443084095905,
      "learning_rate": 0.00015716878402903813,
      "loss": 12.7519,
      "step": 866
    },
    {
      "epoch": 0.04721161503747116,
      "grad_norm": 0.9026000187524962,
      "learning_rate": 0.00015735027223230493,
      "loss": 12.7312,
      "step": 867
    },
    {
      "epoch": 0.04726606903405417,
      "grad_norm": 0.895500696590104,
      "learning_rate": 0.0001575317604355717,
      "loss": 12.5189,
      "step": 868
    },
    {
      "epoch": 0.04732052303063718,
      "grad_norm": 0.8867758320775264,
      "learning_rate": 0.00015771324863883847,
      "loss": 12.5417,
      "step": 869
    },
    {
      "epoch": 0.04737497702722019,
      "grad_norm": 0.8124655980552189,
      "learning_rate": 0.00015789473684210527,
      "loss": 12.6201,
      "step": 870
    },
    {
      "epoch": 0.047429431023803204,
      "grad_norm": 0.9722011505333534,
      "learning_rate": 0.00015807622504537206,
      "loss": 12.5834,
      "step": 871
    },
    {
      "epoch": 0.04748388502038622,
      "grad_norm": 0.982984567178049,
      "learning_rate": 0.00015825771324863886,
      "loss": 12.6461,
      "step": 872
    },
    {
      "epoch": 0.047538339016969224,
      "grad_norm": 0.8381062655513668,
      "learning_rate": 0.00015843920145190563,
      "loss": 12.5809,
      "step": 873
    },
    {
      "epoch": 0.04759279301355224,
      "grad_norm": 0.9650603897304773,
      "learning_rate": 0.00015862068965517243,
      "loss": 12.7033,
      "step": 874
    },
    {
      "epoch": 0.04764724701013525,
      "grad_norm": 0.8854713870148178,
      "learning_rate": 0.0001588021778584392,
      "loss": 12.5962,
      "step": 875
    },
    {
      "epoch": 0.047701701006718264,
      "grad_norm": 0.7761059432684236,
      "learning_rate": 0.000158983666061706,
      "loss": 12.6441,
      "step": 876
    },
    {
      "epoch": 0.04775615500330127,
      "grad_norm": 0.8718350296220381,
      "learning_rate": 0.00015916515426497277,
      "loss": 12.6047,
      "step": 877
    },
    {
      "epoch": 0.047810608999884284,
      "grad_norm": 0.82164764356862,
      "learning_rate": 0.0001593466424682396,
      "loss": 12.5409,
      "step": 878
    },
    {
      "epoch": 0.0478650629964673,
      "grad_norm": 0.8196215669312907,
      "learning_rate": 0.00015952813067150637,
      "loss": 12.5776,
      "step": 879
    },
    {
      "epoch": 0.04791951699305031,
      "grad_norm": 1.014476253091219,
      "learning_rate": 0.00015970961887477314,
      "loss": 12.6802,
      "step": 880
    },
    {
      "epoch": 0.04797397098963332,
      "grad_norm": 0.8195727306565402,
      "learning_rate": 0.00015989110707803993,
      "loss": 12.5736,
      "step": 881
    },
    {
      "epoch": 0.04802842498621633,
      "grad_norm": 0.862536103470985,
      "learning_rate": 0.0001600725952813067,
      "loss": 12.4775,
      "step": 882
    },
    {
      "epoch": 0.048082878982799344,
      "grad_norm": 0.7498353482743876,
      "learning_rate": 0.0001602540834845735,
      "loss": 12.6313,
      "step": 883
    },
    {
      "epoch": 0.04813733297938236,
      "grad_norm": 0.8062614637177298,
      "learning_rate": 0.0001604355716878403,
      "loss": 12.4909,
      "step": 884
    },
    {
      "epoch": 0.048191786975965364,
      "grad_norm": 0.8415146193366887,
      "learning_rate": 0.0001606170598911071,
      "loss": 12.6701,
      "step": 885
    },
    {
      "epoch": 0.04824624097254838,
      "grad_norm": 0.8429914823102769,
      "learning_rate": 0.00016079854809437387,
      "loss": 12.6284,
      "step": 886
    },
    {
      "epoch": 0.04830069496913139,
      "grad_norm": 0.8222857362939741,
      "learning_rate": 0.00016098003629764067,
      "loss": 12.5522,
      "step": 887
    },
    {
      "epoch": 0.048355148965714405,
      "grad_norm": 0.7890962523234009,
      "learning_rate": 0.00016116152450090744,
      "loss": 12.5168,
      "step": 888
    },
    {
      "epoch": 0.04840960296229741,
      "grad_norm": 0.8912584208255049,
      "learning_rate": 0.00016134301270417424,
      "loss": 12.6498,
      "step": 889
    },
    {
      "epoch": 0.048464056958880425,
      "grad_norm": 0.9530476347867869,
      "learning_rate": 0.00016152450090744103,
      "loss": 12.6872,
      "step": 890
    },
    {
      "epoch": 0.04851851095546344,
      "grad_norm": 0.7724850359717667,
      "learning_rate": 0.0001617059891107078,
      "loss": 12.5858,
      "step": 891
    },
    {
      "epoch": 0.04857296495204645,
      "grad_norm": 1.0184330473902963,
      "learning_rate": 0.0001618874773139746,
      "loss": 12.6251,
      "step": 892
    },
    {
      "epoch": 0.04862741894862946,
      "grad_norm": 0.9254790514667174,
      "learning_rate": 0.00016206896551724137,
      "loss": 12.6039,
      "step": 893
    },
    {
      "epoch": 0.04868187294521247,
      "grad_norm": 0.758029348988625,
      "learning_rate": 0.00016225045372050817,
      "loss": 12.4594,
      "step": 894
    },
    {
      "epoch": 0.048736326941795485,
      "grad_norm": 0.7274927678606498,
      "learning_rate": 0.00016243194192377494,
      "loss": 12.5977,
      "step": 895
    },
    {
      "epoch": 0.0487907809383785,
      "grad_norm": 0.8384075646167376,
      "learning_rate": 0.00016261343012704177,
      "loss": 12.6449,
      "step": 896
    },
    {
      "epoch": 0.048845234934961505,
      "grad_norm": 0.7486479322661992,
      "learning_rate": 0.00016279491833030854,
      "loss": 12.6152,
      "step": 897
    },
    {
      "epoch": 0.04889968893154452,
      "grad_norm": 0.9106636506573582,
      "learning_rate": 0.00016297640653357534,
      "loss": 12.6102,
      "step": 898
    },
    {
      "epoch": 0.04895414292812753,
      "grad_norm": 0.8113137306900519,
      "learning_rate": 0.0001631578947368421,
      "loss": 12.5951,
      "step": 899
    },
    {
      "epoch": 0.049008596924710546,
      "grad_norm": 0.8959399175330679,
      "learning_rate": 0.0001633393829401089,
      "loss": 12.7228,
      "step": 900
    },
    {
      "epoch": 0.04906305092129355,
      "grad_norm": 1.0275683355568288,
      "learning_rate": 0.00016352087114337568,
      "loss": 12.7082,
      "step": 901
    },
    {
      "epoch": 0.049117504917876566,
      "grad_norm": 0.889496269768351,
      "learning_rate": 0.00016370235934664247,
      "loss": 12.6843,
      "step": 902
    },
    {
      "epoch": 0.04917195891445958,
      "grad_norm": 0.7581300503636256,
      "learning_rate": 0.00016388384754990927,
      "loss": 12.5467,
      "step": 903
    },
    {
      "epoch": 0.04922641291104259,
      "grad_norm": 0.8974381502103214,
      "learning_rate": 0.00016406533575317604,
      "loss": 12.6412,
      "step": 904
    },
    {
      "epoch": 0.0492808669076256,
      "grad_norm": 0.9315736025199278,
      "learning_rate": 0.00016424682395644284,
      "loss": 12.6383,
      "step": 905
    },
    {
      "epoch": 0.04933532090420861,
      "grad_norm": 0.8746410644890605,
      "learning_rate": 0.0001644283121597096,
      "loss": 12.689,
      "step": 906
    },
    {
      "epoch": 0.049389774900791626,
      "grad_norm": 0.8032178195748642,
      "learning_rate": 0.0001646098003629764,
      "loss": 12.5981,
      "step": 907
    },
    {
      "epoch": 0.04944422889737464,
      "grad_norm": 1.055609092234511,
      "learning_rate": 0.0001647912885662432,
      "loss": 12.6674,
      "step": 908
    },
    {
      "epoch": 0.049498682893957646,
      "grad_norm": 0.9329468352906664,
      "learning_rate": 0.00016497277676951,
      "loss": 12.6298,
      "step": 909
    },
    {
      "epoch": 0.04955313689054066,
      "grad_norm": 0.9100919688447834,
      "learning_rate": 0.00016515426497277678,
      "loss": 12.6856,
      "step": 910
    },
    {
      "epoch": 0.04960759088712367,
      "grad_norm": 0.8157553859700658,
      "learning_rate": 0.00016533575317604358,
      "loss": 12.6711,
      "step": 911
    },
    {
      "epoch": 0.04966204488370669,
      "grad_norm": 0.8407602222280705,
      "learning_rate": 0.00016551724137931035,
      "loss": 12.5952,
      "step": 912
    },
    {
      "epoch": 0.04971649888028969,
      "grad_norm": 0.8536838879477414,
      "learning_rate": 0.00016569872958257714,
      "loss": 12.4985,
      "step": 913
    },
    {
      "epoch": 0.04977095287687271,
      "grad_norm": 0.818610133507912,
      "learning_rate": 0.00016588021778584392,
      "loss": 12.5662,
      "step": 914
    },
    {
      "epoch": 0.04982540687345572,
      "grad_norm": 0.8852001994986728,
      "learning_rate": 0.0001660617059891107,
      "loss": 12.5936,
      "step": 915
    },
    {
      "epoch": 0.049879860870038734,
      "grad_norm": 0.9487098832759733,
      "learning_rate": 0.0001662431941923775,
      "loss": 12.7971,
      "step": 916
    },
    {
      "epoch": 0.04993431486662174,
      "grad_norm": 0.872066612504273,
      "learning_rate": 0.00016642468239564428,
      "loss": 12.7543,
      "step": 917
    },
    {
      "epoch": 0.049988768863204754,
      "grad_norm": 0.7908053266666842,
      "learning_rate": 0.00016660617059891108,
      "loss": 12.6731,
      "step": 918
    },
    {
      "epoch": 0.05004322285978777,
      "grad_norm": 0.8909361958279155,
      "learning_rate": 0.00016678765880217785,
      "loss": 12.6195,
      "step": 919
    },
    {
      "epoch": 0.05009767685637078,
      "grad_norm": 0.8839930655370404,
      "learning_rate": 0.00016696914700544465,
      "loss": 12.6823,
      "step": 920
    },
    {
      "epoch": 0.05015213085295379,
      "grad_norm": 0.8593282266531492,
      "learning_rate": 0.00016715063520871145,
      "loss": 12.5395,
      "step": 921
    },
    {
      "epoch": 0.0502065848495368,
      "grad_norm": 0.7522025735405317,
      "learning_rate": 0.00016733212341197824,
      "loss": 12.5567,
      "step": 922
    },
    {
      "epoch": 0.050261038846119814,
      "grad_norm": 1.0190661738060314,
      "learning_rate": 0.00016751361161524502,
      "loss": 12.6318,
      "step": 923
    },
    {
      "epoch": 0.05031549284270283,
      "grad_norm": 0.7614223532031728,
      "learning_rate": 0.0001676950998185118,
      "loss": 12.6078,
      "step": 924
    },
    {
      "epoch": 0.050369946839285834,
      "grad_norm": 0.9321891712221156,
      "learning_rate": 0.00016787658802177858,
      "loss": 12.6678,
      "step": 925
    },
    {
      "epoch": 0.05042440083586885,
      "grad_norm": 0.8467437632932115,
      "learning_rate": 0.00016805807622504538,
      "loss": 12.6101,
      "step": 926
    },
    {
      "epoch": 0.05047885483245186,
      "grad_norm": 1.0609254909084949,
      "learning_rate": 0.00016823956442831218,
      "loss": 12.7275,
      "step": 927
    },
    {
      "epoch": 0.05053330882903487,
      "grad_norm": 0.9175270724116192,
      "learning_rate": 0.00016842105263157895,
      "loss": 12.7194,
      "step": 928
    },
    {
      "epoch": 0.05058776282561788,
      "grad_norm": 0.9166307896802638,
      "learning_rate": 0.00016860254083484575,
      "loss": 12.6248,
      "step": 929
    },
    {
      "epoch": 0.050642216822200894,
      "grad_norm": 0.8315736647865203,
      "learning_rate": 0.00016878402903811252,
      "loss": 12.6652,
      "step": 930
    },
    {
      "epoch": 0.05069667081878391,
      "grad_norm": 0.779456315691902,
      "learning_rate": 0.00016896551724137932,
      "loss": 12.6193,
      "step": 931
    },
    {
      "epoch": 0.050751124815366915,
      "grad_norm": 0.9332276655363869,
      "learning_rate": 0.0001691470054446461,
      "loss": 12.4871,
      "step": 932
    },
    {
      "epoch": 0.05080557881194993,
      "grad_norm": 0.7483055003595003,
      "learning_rate": 0.0001693284936479129,
      "loss": 12.3438,
      "step": 933
    },
    {
      "epoch": 0.05086003280853294,
      "grad_norm": 1.043133034199254,
      "learning_rate": 0.00016950998185117968,
      "loss": 12.5724,
      "step": 934
    },
    {
      "epoch": 0.050914486805115955,
      "grad_norm": 0.9885656883730269,
      "learning_rate": 0.00016969147005444648,
      "loss": 12.5337,
      "step": 935
    },
    {
      "epoch": 0.05096894080169896,
      "grad_norm": 0.8284948943642783,
      "learning_rate": 0.00016987295825771325,
      "loss": 12.7829,
      "step": 936
    },
    {
      "epoch": 0.051023394798281975,
      "grad_norm": 0.8429753478015432,
      "learning_rate": 0.00017005444646098005,
      "loss": 12.671,
      "step": 937
    },
    {
      "epoch": 0.05107784879486499,
      "grad_norm": 0.8150872100250276,
      "learning_rate": 0.00017023593466424682,
      "loss": 12.5752,
      "step": 938
    },
    {
      "epoch": 0.051132302791448,
      "grad_norm": 0.975085573256272,
      "learning_rate": 0.00017041742286751362,
      "loss": 12.7308,
      "step": 939
    },
    {
      "epoch": 0.05118675678803101,
      "grad_norm": 0.8226356283220444,
      "learning_rate": 0.00017059891107078042,
      "loss": 12.7762,
      "step": 940
    },
    {
      "epoch": 0.05124121078461402,
      "grad_norm": 0.9192672341554673,
      "learning_rate": 0.0001707803992740472,
      "loss": 12.6354,
      "step": 941
    },
    {
      "epoch": 0.051295664781197035,
      "grad_norm": 0.8457327315713598,
      "learning_rate": 0.000170961887477314,
      "loss": 12.665,
      "step": 942
    },
    {
      "epoch": 0.05135011877778005,
      "grad_norm": 0.8549173394249183,
      "learning_rate": 0.00017114337568058076,
      "loss": 12.546,
      "step": 943
    },
    {
      "epoch": 0.051404572774363055,
      "grad_norm": 0.9118763020226056,
      "learning_rate": 0.00017132486388384756,
      "loss": 12.6458,
      "step": 944
    },
    {
      "epoch": 0.05145902677094607,
      "grad_norm": 0.8463997651394616,
      "learning_rate": 0.00017150635208711435,
      "loss": 12.7347,
      "step": 945
    },
    {
      "epoch": 0.05151348076752908,
      "grad_norm": 0.8627001066806589,
      "learning_rate": 0.00017168784029038115,
      "loss": 12.5186,
      "step": 946
    },
    {
      "epoch": 0.051567934764112096,
      "grad_norm": 0.8447051949506521,
      "learning_rate": 0.00017186932849364792,
      "loss": 12.6107,
      "step": 947
    },
    {
      "epoch": 0.0516223887606951,
      "grad_norm": 0.8246470203738206,
      "learning_rate": 0.00017205081669691472,
      "loss": 12.6495,
      "step": 948
    },
    {
      "epoch": 0.051676842757278116,
      "grad_norm": 0.8653752959166282,
      "learning_rate": 0.0001722323049001815,
      "loss": 12.5561,
      "step": 949
    },
    {
      "epoch": 0.05173129675386113,
      "grad_norm": 0.8007262251384583,
      "learning_rate": 0.00017241379310344826,
      "loss": 12.6482,
      "step": 950
    },
    {
      "epoch": 0.05178575075044414,
      "grad_norm": 0.8102077511779113,
      "learning_rate": 0.00017259528130671506,
      "loss": 12.6646,
      "step": 951
    },
    {
      "epoch": 0.05184020474702715,
      "grad_norm": 0.904787409745801,
      "learning_rate": 0.00017277676950998186,
      "loss": 12.613,
      "step": 952
    },
    {
      "epoch": 0.05189465874361016,
      "grad_norm": 0.8622064273999556,
      "learning_rate": 0.00017295825771324866,
      "loss": 12.6379,
      "step": 953
    },
    {
      "epoch": 0.051949112740193176,
      "grad_norm": 0.9227253225257225,
      "learning_rate": 0.00017313974591651543,
      "loss": 12.4684,
      "step": 954
    },
    {
      "epoch": 0.05200356673677619,
      "grad_norm": 0.9007687103506802,
      "learning_rate": 0.00017332123411978223,
      "loss": 12.6496,
      "step": 955
    },
    {
      "epoch": 0.052058020733359196,
      "grad_norm": 0.9285057014615055,
      "learning_rate": 0.000173502722323049,
      "loss": 12.5537,
      "step": 956
    },
    {
      "epoch": 0.05211247472994221,
      "grad_norm": 0.789205719752802,
      "learning_rate": 0.0001736842105263158,
      "loss": 12.5936,
      "step": 957
    },
    {
      "epoch": 0.05216692872652522,
      "grad_norm": 0.8941984741380373,
      "learning_rate": 0.0001738656987295826,
      "loss": 12.5961,
      "step": 958
    },
    {
      "epoch": 0.05222138272310824,
      "grad_norm": 0.9388036268594567,
      "learning_rate": 0.0001740471869328494,
      "loss": 12.609,
      "step": 959
    },
    {
      "epoch": 0.05227583671969124,
      "grad_norm": 0.9588542625267311,
      "learning_rate": 0.00017422867513611616,
      "loss": 12.662,
      "step": 960
    },
    {
      "epoch": 0.05233029071627426,
      "grad_norm": 0.9761067195333801,
      "learning_rate": 0.00017441016333938296,
      "loss": 12.7561,
      "step": 961
    },
    {
      "epoch": 0.05238474471285727,
      "grad_norm": 0.7943628439855379,
      "learning_rate": 0.00017459165154264973,
      "loss": 12.661,
      "step": 962
    },
    {
      "epoch": 0.052439198709440284,
      "grad_norm": 0.7609648264074853,
      "learning_rate": 0.0001747731397459165,
      "loss": 12.6289,
      "step": 963
    },
    {
      "epoch": 0.05249365270602329,
      "grad_norm": 0.9609887685299381,
      "learning_rate": 0.00017495462794918333,
      "loss": 12.7005,
      "step": 964
    },
    {
      "epoch": 0.052548106702606304,
      "grad_norm": 0.7698262991659199,
      "learning_rate": 0.0001751361161524501,
      "loss": 12.5711,
      "step": 965
    },
    {
      "epoch": 0.05260256069918932,
      "grad_norm": 0.8387736314013369,
      "learning_rate": 0.0001753176043557169,
      "loss": 12.6618,
      "step": 966
    },
    {
      "epoch": 0.05265701469577233,
      "grad_norm": 0.8919145099367959,
      "learning_rate": 0.00017549909255898367,
      "loss": 12.6663,
      "step": 967
    },
    {
      "epoch": 0.05271146869235534,
      "grad_norm": 0.9217171184647402,
      "learning_rate": 0.00017568058076225046,
      "loss": 12.6593,
      "step": 968
    },
    {
      "epoch": 0.05276592268893835,
      "grad_norm": 0.8225240714439287,
      "learning_rate": 0.00017586206896551723,
      "loss": 12.7085,
      "step": 969
    },
    {
      "epoch": 0.052820376685521364,
      "grad_norm": 0.8413886223855485,
      "learning_rate": 0.00017604355716878403,
      "loss": 12.5573,
      "step": 970
    },
    {
      "epoch": 0.05287483068210438,
      "grad_norm": 0.7944122346053981,
      "learning_rate": 0.00017622504537205083,
      "loss": 12.649,
      "step": 971
    },
    {
      "epoch": 0.052929284678687384,
      "grad_norm": 0.9027497161747975,
      "learning_rate": 0.00017640653357531763,
      "loss": 12.6321,
      "step": 972
    },
    {
      "epoch": 0.0529837386752704,
      "grad_norm": 0.7864322574182031,
      "learning_rate": 0.0001765880217785844,
      "loss": 12.7023,
      "step": 973
    },
    {
      "epoch": 0.05303819267185341,
      "grad_norm": 0.8749044448504391,
      "learning_rate": 0.00017676950998185117,
      "loss": 12.6465,
      "step": 974
    },
    {
      "epoch": 0.053092646668436425,
      "grad_norm": 0.733046163574283,
      "learning_rate": 0.00017695099818511797,
      "loss": 12.5625,
      "step": 975
    },
    {
      "epoch": 0.05314710066501943,
      "grad_norm": 0.850086651338118,
      "learning_rate": 0.00017713248638838477,
      "loss": 12.6172,
      "step": 976
    },
    {
      "epoch": 0.053201554661602445,
      "grad_norm": 0.9315575438714128,
      "learning_rate": 0.00017731397459165156,
      "loss": 12.7023,
      "step": 977
    },
    {
      "epoch": 0.05325600865818546,
      "grad_norm": 0.8522769717057842,
      "learning_rate": 0.00017749546279491833,
      "loss": 12.7316,
      "step": 978
    },
    {
      "epoch": 0.05331046265476847,
      "grad_norm": 0.7952996316417338,
      "learning_rate": 0.00017767695099818513,
      "loss": 12.5978,
      "step": 979
    },
    {
      "epoch": 0.05336491665135148,
      "grad_norm": 0.8405135630987488,
      "learning_rate": 0.0001778584392014519,
      "loss": 12.8137,
      "step": 980
    },
    {
      "epoch": 0.05341937064793449,
      "grad_norm": 0.7651645706970682,
      "learning_rate": 0.0001780399274047187,
      "loss": 12.5507,
      "step": 981
    },
    {
      "epoch": 0.053473824644517505,
      "grad_norm": 0.8459505872252346,
      "learning_rate": 0.0001782214156079855,
      "loss": 12.5907,
      "step": 982
    },
    {
      "epoch": 0.05352827864110052,
      "grad_norm": 0.7574463269859852,
      "learning_rate": 0.0001784029038112523,
      "loss": 12.4691,
      "step": 983
    },
    {
      "epoch": 0.053582732637683525,
      "grad_norm": 0.8922042819317508,
      "learning_rate": 0.00017858439201451907,
      "loss": 12.7275,
      "step": 984
    },
    {
      "epoch": 0.05363718663426654,
      "grad_norm": 0.7755357628949647,
      "learning_rate": 0.00017876588021778584,
      "loss": 12.6564,
      "step": 985
    },
    {
      "epoch": 0.05369164063084955,
      "grad_norm": 0.888720018295191,
      "learning_rate": 0.00017894736842105264,
      "loss": 12.4645,
      "step": 986
    },
    {
      "epoch": 0.053746094627432565,
      "grad_norm": 0.8258444684364085,
      "learning_rate": 0.0001791288566243194,
      "loss": 12.7308,
      "step": 987
    },
    {
      "epoch": 0.05380054862401557,
      "grad_norm": 0.8517525708787187,
      "learning_rate": 0.0001793103448275862,
      "loss": 12.6665,
      "step": 988
    },
    {
      "epoch": 0.053855002620598585,
      "grad_norm": 0.9308572420303274,
      "learning_rate": 0.000179491833030853,
      "loss": 12.6683,
      "step": 989
    },
    {
      "epoch": 0.0539094566171816,
      "grad_norm": 0.947701272403253,
      "learning_rate": 0.0001796733212341198,
      "loss": 12.7302,
      "step": 990
    },
    {
      "epoch": 0.05396391061376461,
      "grad_norm": 0.8412669138813667,
      "learning_rate": 0.00017985480943738657,
      "loss": 12.4787,
      "step": 991
    },
    {
      "epoch": 0.05401836461034762,
      "grad_norm": 0.9849026278127962,
      "learning_rate": 0.00018003629764065337,
      "loss": 12.5891,
      "step": 992
    },
    {
      "epoch": 0.05407281860693063,
      "grad_norm": 0.9591250834751449,
      "learning_rate": 0.00018021778584392014,
      "loss": 12.6623,
      "step": 993
    },
    {
      "epoch": 0.054127272603513646,
      "grad_norm": 0.9657335153684073,
      "learning_rate": 0.00018039927404718694,
      "loss": 12.6658,
      "step": 994
    },
    {
      "epoch": 0.05418172660009665,
      "grad_norm": 0.7862727136981313,
      "learning_rate": 0.00018058076225045374,
      "loss": 12.6252,
      "step": 995
    },
    {
      "epoch": 0.054236180596679666,
      "grad_norm": 0.8994392012305145,
      "learning_rate": 0.00018076225045372054,
      "loss": 12.6117,
      "step": 996
    },
    {
      "epoch": 0.05429063459326268,
      "grad_norm": 0.9888859609422734,
      "learning_rate": 0.0001809437386569873,
      "loss": 12.542,
      "step": 997
    },
    {
      "epoch": 0.05434508858984569,
      "grad_norm": 0.8085407049387496,
      "learning_rate": 0.00018112522686025408,
      "loss": 12.511,
      "step": 998
    },
    {
      "epoch": 0.0543995425864287,
      "grad_norm": 0.8706037750909749,
      "learning_rate": 0.00018130671506352088,
      "loss": 12.6018,
      "step": 999
    },
    {
      "epoch": 0.05445399658301171,
      "grad_norm": 0.9259418524451194,
      "learning_rate": 0.00018148820326678765,
      "loss": 12.8483,
      "step": 1000
    },
    {
      "epoch": 0.054508450579594726,
      "grad_norm": 0.7746764072383874,
      "learning_rate": 0.00018166969147005447,
      "loss": 12.5888,
      "step": 1001
    },
    {
      "epoch": 0.05456290457617774,
      "grad_norm": 0.9766058753944424,
      "learning_rate": 0.00018185117967332124,
      "loss": 12.5827,
      "step": 1002
    },
    {
      "epoch": 0.054617358572760746,
      "grad_norm": 0.8856003767869254,
      "learning_rate": 0.00018203266787658804,
      "loss": 12.4993,
      "step": 1003
    },
    {
      "epoch": 0.05467181256934376,
      "grad_norm": 0.9485411740557078,
      "learning_rate": 0.0001822141560798548,
      "loss": 12.677,
      "step": 1004
    },
    {
      "epoch": 0.05472626656592677,
      "grad_norm": 0.7672312949692742,
      "learning_rate": 0.0001823956442831216,
      "loss": 12.6134,
      "step": 1005
    },
    {
      "epoch": 0.05478072056250979,
      "grad_norm": 0.9172643751252387,
      "learning_rate": 0.00018257713248638838,
      "loss": 12.7224,
      "step": 1006
    },
    {
      "epoch": 0.05483517455909279,
      "grad_norm": 0.8420022646409505,
      "learning_rate": 0.00018275862068965518,
      "loss": 12.6461,
      "step": 1007
    },
    {
      "epoch": 0.05488962855567581,
      "grad_norm": 0.8333960383700091,
      "learning_rate": 0.00018294010889292198,
      "loss": 12.6246,
      "step": 1008
    },
    {
      "epoch": 0.05494408255225882,
      "grad_norm": 1.0096244098506704,
      "learning_rate": 0.00018312159709618875,
      "loss": 12.7259,
      "step": 1009
    },
    {
      "epoch": 0.054998536548841834,
      "grad_norm": 0.7871373395387745,
      "learning_rate": 0.00018330308529945554,
      "loss": 12.7275,
      "step": 1010
    },
    {
      "epoch": 0.05505299054542484,
      "grad_norm": 1.032343747821382,
      "learning_rate": 0.00018348457350272232,
      "loss": 12.6628,
      "step": 1011
    },
    {
      "epoch": 0.055107444542007854,
      "grad_norm": 0.7083079315534552,
      "learning_rate": 0.00018366606170598911,
      "loss": 12.4748,
      "step": 1012
    },
    {
      "epoch": 0.05516189853859087,
      "grad_norm": 1.1907392275444992,
      "learning_rate": 0.0001838475499092559,
      "loss": 12.5132,
      "step": 1013
    },
    {
      "epoch": 0.05521635253517388,
      "grad_norm": 0.8442442447866033,
      "learning_rate": 0.0001840290381125227,
      "loss": 12.6859,
      "step": 1014
    },
    {
      "epoch": 0.05527080653175689,
      "grad_norm": 0.9375890211793874,
      "learning_rate": 0.00018421052631578948,
      "loss": 12.6139,
      "step": 1015
    },
    {
      "epoch": 0.0553252605283399,
      "grad_norm": 0.7957705162002837,
      "learning_rate": 0.00018439201451905628,
      "loss": 12.6227,
      "step": 1016
    },
    {
      "epoch": 0.055379714524922914,
      "grad_norm": 0.8572364025943078,
      "learning_rate": 0.00018457350272232305,
      "loss": 12.4793,
      "step": 1017
    },
    {
      "epoch": 0.05543416852150593,
      "grad_norm": 0.8594165745763462,
      "learning_rate": 0.00018475499092558985,
      "loss": 12.7353,
      "step": 1018
    },
    {
      "epoch": 0.055488622518088934,
      "grad_norm": 0.8947901476734842,
      "learning_rate": 0.00018493647912885665,
      "loss": 12.5524,
      "step": 1019
    },
    {
      "epoch": 0.05554307651467195,
      "grad_norm": 0.8317346016577231,
      "learning_rate": 0.00018511796733212342,
      "loss": 12.5814,
      "step": 1020
    },
    {
      "epoch": 0.05559753051125496,
      "grad_norm": 0.7741287564610906,
      "learning_rate": 0.00018529945553539021,
      "loss": 12.7118,
      "step": 1021
    },
    {
      "epoch": 0.055651984507837975,
      "grad_norm": 0.9229904453397703,
      "learning_rate": 0.00018548094373865698,
      "loss": 12.7203,
      "step": 1022
    },
    {
      "epoch": 0.05570643850442098,
      "grad_norm": 0.7189120710008146,
      "learning_rate": 0.00018566243194192378,
      "loss": 12.7316,
      "step": 1023
    },
    {
      "epoch": 0.055760892501003995,
      "grad_norm": 0.7996729763717355,
      "learning_rate": 0.00018584392014519055,
      "loss": 12.6907,
      "step": 1024
    },
    {
      "epoch": 0.05581534649758701,
      "grad_norm": 0.8282088203606525,
      "learning_rate": 0.00018602540834845735,
      "loss": 12.6942,
      "step": 1025
    },
    {
      "epoch": 0.05586980049417002,
      "grad_norm": 0.8101927238311741,
      "learning_rate": 0.00018620689655172415,
      "loss": 12.6429,
      "step": 1026
    },
    {
      "epoch": 0.05592425449075303,
      "grad_norm": 0.8723302771571373,
      "learning_rate": 0.00018638838475499095,
      "loss": 12.715,
      "step": 1027
    },
    {
      "epoch": 0.05597870848733604,
      "grad_norm": 0.8993457263793324,
      "learning_rate": 0.00018656987295825772,
      "loss": 12.7494,
      "step": 1028
    },
    {
      "epoch": 0.056033162483919055,
      "grad_norm": 0.8956846970014478,
      "learning_rate": 0.00018675136116152452,
      "loss": 12.7523,
      "step": 1029
    },
    {
      "epoch": 0.05608761648050207,
      "grad_norm": 0.848471231997586,
      "learning_rate": 0.0001869328493647913,
      "loss": 12.7121,
      "step": 1030
    },
    {
      "epoch": 0.056142070477085075,
      "grad_norm": 0.7681001667107862,
      "learning_rate": 0.00018711433756805809,
      "loss": 12.6024,
      "step": 1031
    },
    {
      "epoch": 0.05619652447366809,
      "grad_norm": 0.7924794070425718,
      "learning_rate": 0.00018729582577132488,
      "loss": 12.7012,
      "step": 1032
    },
    {
      "epoch": 0.0562509784702511,
      "grad_norm": 0.8349770749508899,
      "learning_rate": 0.00018747731397459165,
      "loss": 12.6386,
      "step": 1033
    },
    {
      "epoch": 0.056305432466834116,
      "grad_norm": 0.8191879420778659,
      "learning_rate": 0.00018765880217785845,
      "loss": 12.6633,
      "step": 1034
    },
    {
      "epoch": 0.05635988646341712,
      "grad_norm": 0.9108979371089774,
      "learning_rate": 0.00018784029038112522,
      "loss": 12.7074,
      "step": 1035
    },
    {
      "epoch": 0.056414340460000136,
      "grad_norm": 0.9995188221538889,
      "learning_rate": 0.00018802177858439202,
      "loss": 12.7757,
      "step": 1036
    },
    {
      "epoch": 0.05646879445658315,
      "grad_norm": 0.8331304603254887,
      "learning_rate": 0.0001882032667876588,
      "loss": 12.7268,
      "step": 1037
    },
    {
      "epoch": 0.05652324845316616,
      "grad_norm": 0.8421314134400244,
      "learning_rate": 0.00018838475499092562,
      "loss": 12.6786,
      "step": 1038
    },
    {
      "epoch": 0.05657770244974917,
      "grad_norm": 0.9249062144746272,
      "learning_rate": 0.0001885662431941924,
      "loss": 12.6439,
      "step": 1039
    },
    {
      "epoch": 0.05663215644633218,
      "grad_norm": 1.0740033540400613,
      "learning_rate": 0.00018874773139745919,
      "loss": 12.6613,
      "step": 1040
    },
    {
      "epoch": 0.056686610442915196,
      "grad_norm": 0.9515879897654931,
      "learning_rate": 0.00018892921960072596,
      "loss": 12.7961,
      "step": 1041
    },
    {
      "epoch": 0.05674106443949821,
      "grad_norm": 1.0687652941035266,
      "learning_rate": 0.00018911070780399275,
      "loss": 12.7123,
      "step": 1042
    },
    {
      "epoch": 0.056795518436081216,
      "grad_norm": 0.9446694097624536,
      "learning_rate": 0.00018929219600725953,
      "loss": 12.7562,
      "step": 1043
    },
    {
      "epoch": 0.05684997243266423,
      "grad_norm": 0.9237974447352099,
      "learning_rate": 0.00018947368421052632,
      "loss": 12.8357,
      "step": 1044
    },
    {
      "epoch": 0.05690442642924724,
      "grad_norm": 0.8936942172764373,
      "learning_rate": 0.00018965517241379312,
      "loss": 12.7792,
      "step": 1045
    },
    {
      "epoch": 0.056958880425830256,
      "grad_norm": 0.9961818489725659,
      "learning_rate": 0.0001898366606170599,
      "loss": 12.6665,
      "step": 1046
    },
    {
      "epoch": 0.05701333442241326,
      "grad_norm": 0.8390675260109299,
      "learning_rate": 0.0001900181488203267,
      "loss": 12.6463,
      "step": 1047
    },
    {
      "epoch": 0.057067788418996276,
      "grad_norm": 0.9127513873491976,
      "learning_rate": 0.00019019963702359346,
      "loss": 12.6908,
      "step": 1048
    },
    {
      "epoch": 0.05712224241557929,
      "grad_norm": 0.7822521272987107,
      "learning_rate": 0.00019038112522686026,
      "loss": 12.6916,
      "step": 1049
    },
    {
      "epoch": 0.0571766964121623,
      "grad_norm": 0.9316417618579463,
      "learning_rate": 0.00019056261343012706,
      "loss": 12.6148,
      "step": 1050
    },
    {
      "epoch": 0.05723115040874531,
      "grad_norm": 0.8267893855283287,
      "learning_rate": 0.00019074410163339386,
      "loss": 12.6444,
      "step": 1051
    },
    {
      "epoch": 0.05728560440532832,
      "grad_norm": 0.8756184417618663,
      "learning_rate": 0.00019092558983666063,
      "loss": 12.6479,
      "step": 1052
    },
    {
      "epoch": 0.05734005840191134,
      "grad_norm": 0.8537715754578956,
      "learning_rate": 0.00019110707803992742,
      "loss": 12.8705,
      "step": 1053
    },
    {
      "epoch": 0.05739451239849435,
      "grad_norm": 0.8793465083226172,
      "learning_rate": 0.0001912885662431942,
      "loss": 12.6592,
      "step": 1054
    },
    {
      "epoch": 0.05744896639507736,
      "grad_norm": 0.914347190314965,
      "learning_rate": 0.000191470054446461,
      "loss": 12.6625,
      "step": 1055
    },
    {
      "epoch": 0.05750342039166037,
      "grad_norm": 0.8856692561805453,
      "learning_rate": 0.0001916515426497278,
      "loss": 12.692,
      "step": 1056
    },
    {
      "epoch": 0.057557874388243384,
      "grad_norm": 0.8458952197724661,
      "learning_rate": 0.00019183303085299456,
      "loss": 12.7096,
      "step": 1057
    },
    {
      "epoch": 0.0576123283848264,
      "grad_norm": 0.7757824745642351,
      "learning_rate": 0.00019201451905626136,
      "loss": 12.7721,
      "step": 1058
    },
    {
      "epoch": 0.057666782381409404,
      "grad_norm": 0.975994260166484,
      "learning_rate": 0.00019219600725952813,
      "loss": 12.719,
      "step": 1059
    },
    {
      "epoch": 0.05772123637799242,
      "grad_norm": 0.8585148684375852,
      "learning_rate": 0.00019237749546279493,
      "loss": 12.7498,
      "step": 1060
    },
    {
      "epoch": 0.05777569037457543,
      "grad_norm": 0.8545141793726162,
      "learning_rate": 0.0001925589836660617,
      "loss": 12.7521,
      "step": 1061
    },
    {
      "epoch": 0.05783014437115844,
      "grad_norm": 1.0351696982338359,
      "learning_rate": 0.0001927404718693285,
      "loss": 12.6858,
      "step": 1062
    },
    {
      "epoch": 0.05788459836774145,
      "grad_norm": 0.8932909538412702,
      "learning_rate": 0.0001929219600725953,
      "loss": 12.6172,
      "step": 1063
    },
    {
      "epoch": 0.057939052364324464,
      "grad_norm": 0.8033540231795703,
      "learning_rate": 0.0001931034482758621,
      "loss": 12.7224,
      "step": 1064
    },
    {
      "epoch": 0.05799350636090748,
      "grad_norm": 0.9647008699499642,
      "learning_rate": 0.00019328493647912886,
      "loss": 12.8202,
      "step": 1065
    },
    {
      "epoch": 0.058047960357490484,
      "grad_norm": 0.8107858754859842,
      "learning_rate": 0.00019346642468239566,
      "loss": 12.6884,
      "step": 1066
    },
    {
      "epoch": 0.0581024143540735,
      "grad_norm": 0.8763919745645723,
      "learning_rate": 0.00019364791288566243,
      "loss": 12.5539,
      "step": 1067
    },
    {
      "epoch": 0.05815686835065651,
      "grad_norm": 1.0992738520488685,
      "learning_rate": 0.00019382940108892923,
      "loss": 12.6674,
      "step": 1068
    },
    {
      "epoch": 0.058211322347239525,
      "grad_norm": 0.7850194207297764,
      "learning_rate": 0.00019401088929219603,
      "loss": 12.597,
      "step": 1069
    },
    {
      "epoch": 0.05826577634382253,
      "grad_norm": 0.9852227560575982,
      "learning_rate": 0.0001941923774954628,
      "loss": 12.6947,
      "step": 1070
    },
    {
      "epoch": 0.058320230340405545,
      "grad_norm": 1.0687907738165234,
      "learning_rate": 0.0001943738656987296,
      "loss": 12.8078,
      "step": 1071
    },
    {
      "epoch": 0.05837468433698856,
      "grad_norm": 0.8874665780436435,
      "learning_rate": 0.00019455535390199637,
      "loss": 12.6525,
      "step": 1072
    },
    {
      "epoch": 0.05842913833357157,
      "grad_norm": 0.8938001131329727,
      "learning_rate": 0.00019473684210526317,
      "loss": 12.7372,
      "step": 1073
    },
    {
      "epoch": 0.05848359233015458,
      "grad_norm": 0.8296677243268067,
      "learning_rate": 0.00019491833030852994,
      "loss": 12.6129,
      "step": 1074
    },
    {
      "epoch": 0.05853804632673759,
      "grad_norm": 0.9888068104357823,
      "learning_rate": 0.00019509981851179676,
      "loss": 12.8547,
      "step": 1075
    },
    {
      "epoch": 0.058592500323320605,
      "grad_norm": 1.028019721476723,
      "learning_rate": 0.00019528130671506353,
      "loss": 12.7056,
      "step": 1076
    },
    {
      "epoch": 0.05864695431990362,
      "grad_norm": 0.8361357407910752,
      "learning_rate": 0.00019546279491833033,
      "loss": 12.6857,
      "step": 1077
    },
    {
      "epoch": 0.058701408316486625,
      "grad_norm": 0.8973490125437626,
      "learning_rate": 0.0001956442831215971,
      "loss": 12.6222,
      "step": 1078
    },
    {
      "epoch": 0.05875586231306964,
      "grad_norm": 0.7877561882710619,
      "learning_rate": 0.00019582577132486387,
      "loss": 12.7484,
      "step": 1079
    },
    {
      "epoch": 0.05881031630965265,
      "grad_norm": 1.346493036428632,
      "learning_rate": 0.00019600725952813067,
      "loss": 12.7032,
      "step": 1080
    },
    {
      "epoch": 0.058864770306235666,
      "grad_norm": 0.9387631520562696,
      "learning_rate": 0.00019618874773139747,
      "loss": 12.6168,
      "step": 1081
    },
    {
      "epoch": 0.05891922430281867,
      "grad_norm": 0.8426429211134219,
      "learning_rate": 0.00019637023593466427,
      "loss": 12.7582,
      "step": 1082
    },
    {
      "epoch": 0.058973678299401686,
      "grad_norm": 0.8821282555993558,
      "learning_rate": 0.00019655172413793104,
      "loss": 12.6952,
      "step": 1083
    },
    {
      "epoch": 0.0590281322959847,
      "grad_norm": 0.9413833624250499,
      "learning_rate": 0.00019673321234119784,
      "loss": 12.8895,
      "step": 1084
    },
    {
      "epoch": 0.05908258629256771,
      "grad_norm": 1.0551890250213953,
      "learning_rate": 0.0001969147005444646,
      "loss": 12.7802,
      "step": 1085
    },
    {
      "epoch": 0.05913704028915072,
      "grad_norm": 0.826483127495897,
      "learning_rate": 0.0001970961887477314,
      "loss": 12.7205,
      "step": 1086
    },
    {
      "epoch": 0.05919149428573373,
      "grad_norm": 0.953767015147213,
      "learning_rate": 0.0001972776769509982,
      "loss": 12.6614,
      "step": 1087
    },
    {
      "epoch": 0.059245948282316746,
      "grad_norm": 0.8755683812872677,
      "learning_rate": 0.000197459165154265,
      "loss": 12.7046,
      "step": 1088
    },
    {
      "epoch": 0.05930040227889976,
      "grad_norm": 0.8242367546212285,
      "learning_rate": 0.00019764065335753177,
      "loss": 12.5865,
      "step": 1089
    },
    {
      "epoch": 0.059354856275482766,
      "grad_norm": 0.8118383627116624,
      "learning_rate": 0.00019782214156079857,
      "loss": 12.7108,
      "step": 1090
    },
    {
      "epoch": 0.05940931027206578,
      "grad_norm": 0.8309785056550677,
      "learning_rate": 0.00019800362976406534,
      "loss": 12.7555,
      "step": 1091
    },
    {
      "epoch": 0.05946376426864879,
      "grad_norm": 0.8702493286280903,
      "learning_rate": 0.0001981851179673321,
      "loss": 12.7094,
      "step": 1092
    },
    {
      "epoch": 0.059518218265231806,
      "grad_norm": 0.8935609387918308,
      "learning_rate": 0.00019836660617059894,
      "loss": 12.8853,
      "step": 1093
    },
    {
      "epoch": 0.05957267226181481,
      "grad_norm": 0.8176063965684935,
      "learning_rate": 0.0001985480943738657,
      "loss": 12.7473,
      "step": 1094
    },
    {
      "epoch": 0.059627126258397826,
      "grad_norm": 0.9382886609351457,
      "learning_rate": 0.0001987295825771325,
      "loss": 12.6793,
      "step": 1095
    },
    {
      "epoch": 0.05968158025498084,
      "grad_norm": 0.8780758861923207,
      "learning_rate": 0.00019891107078039928,
      "loss": 12.6795,
      "step": 1096
    },
    {
      "epoch": 0.05973603425156385,
      "grad_norm": 0.9060522534957417,
      "learning_rate": 0.00019909255898366607,
      "loss": 12.6617,
      "step": 1097
    },
    {
      "epoch": 0.05979048824814686,
      "grad_norm": 0.8515080058550354,
      "learning_rate": 0.00019927404718693284,
      "loss": 12.6893,
      "step": 1098
    },
    {
      "epoch": 0.05984494224472987,
      "grad_norm": 0.8872822863676731,
      "learning_rate": 0.00019945553539019964,
      "loss": 12.7036,
      "step": 1099
    },
    {
      "epoch": 0.05989939624131289,
      "grad_norm": 0.838829157388531,
      "learning_rate": 0.00019963702359346644,
      "loss": 12.6507,
      "step": 1100
    },
    {
      "epoch": 0.0599538502378959,
      "grad_norm": 0.8884415432987369,
      "learning_rate": 0.00019981851179673324,
      "loss": 12.8487,
      "step": 1101
    },
    {
      "epoch": 0.06000830423447891,
      "grad_norm": 0.9740412298051592,
      "learning_rate": 0.0002,
      "loss": 12.6881,
      "step": 1102
    },
    {
      "epoch": 0.06006275823106192,
      "grad_norm": 0.9241866593303006,
      "learning_rate": 0.0001999999996111916,
      "loss": 12.7269,
      "step": 1103
    },
    {
      "epoch": 0.060117212227644934,
      "grad_norm": 0.9278070088616288,
      "learning_rate": 0.00019999999844476647,
      "loss": 12.7711,
      "step": 1104
    },
    {
      "epoch": 0.06017166622422795,
      "grad_norm": 0.8756109875155865,
      "learning_rate": 0.00019999999650072457,
      "loss": 12.6504,
      "step": 1105
    },
    {
      "epoch": 0.060226120220810954,
      "grad_norm": 0.9555653657587291,
      "learning_rate": 0.0001999999937790659,
      "loss": 12.5121,
      "step": 1106
    },
    {
      "epoch": 0.06028057421739397,
      "grad_norm": 0.8096454821951842,
      "learning_rate": 0.00019999999027979054,
      "loss": 12.6582,
      "step": 1107
    },
    {
      "epoch": 0.06033502821397698,
      "grad_norm": 0.9296910383062582,
      "learning_rate": 0.00019999998600289846,
      "loss": 12.6657,
      "step": 1108
    },
    {
      "epoch": 0.060389482210559994,
      "grad_norm": 0.8466965321383215,
      "learning_rate": 0.00019999998094838973,
      "loss": 12.6374,
      "step": 1109
    },
    {
      "epoch": 0.060443936207143,
      "grad_norm": 0.8264775366654458,
      "learning_rate": 0.0001999999751162644,
      "loss": 12.7052,
      "step": 1110
    },
    {
      "epoch": 0.060498390203726014,
      "grad_norm": 0.9400289147748783,
      "learning_rate": 0.00019999996850652245,
      "loss": 12.7985,
      "step": 1111
    },
    {
      "epoch": 0.06055284420030903,
      "grad_norm": 0.940019638160756,
      "learning_rate": 0.00019999996111916399,
      "loss": 12.7924,
      "step": 1112
    },
    {
      "epoch": 0.06060729819689204,
      "grad_norm": 0.7509979058254492,
      "learning_rate": 0.00019999995295418908,
      "loss": 12.6273,
      "step": 1113
    },
    {
      "epoch": 0.06066175219347505,
      "grad_norm": 0.8144515839177705,
      "learning_rate": 0.00019999994401159775,
      "loss": 12.7125,
      "step": 1114
    },
    {
      "epoch": 0.06071620619005806,
      "grad_norm": 0.7871352320666705,
      "learning_rate": 0.0001999999342913901,
      "loss": 12.6371,
      "step": 1115
    },
    {
      "epoch": 0.060770660186641075,
      "grad_norm": 0.8128165695381447,
      "learning_rate": 0.0001999999237935662,
      "loss": 12.7798,
      "step": 1116
    },
    {
      "epoch": 0.06082511418322409,
      "grad_norm": 0.9145599160586337,
      "learning_rate": 0.00019999991251812608,
      "loss": 12.7207,
      "step": 1117
    },
    {
      "epoch": 0.060879568179807095,
      "grad_norm": 0.8061894716041041,
      "learning_rate": 0.00019999990046506988,
      "loss": 12.6761,
      "step": 1118
    },
    {
      "epoch": 0.06093402217639011,
      "grad_norm": 0.84955895633201,
      "learning_rate": 0.00019999988763439773,
      "loss": 12.6494,
      "step": 1119
    },
    {
      "epoch": 0.06098847617297312,
      "grad_norm": 0.8119447540792254,
      "learning_rate": 0.00019999987402610962,
      "loss": 12.7347,
      "step": 1120
    },
    {
      "epoch": 0.061042930169556135,
      "grad_norm": 0.8194282221432447,
      "learning_rate": 0.00019999985964020577,
      "loss": 12.7614,
      "step": 1121
    },
    {
      "epoch": 0.06109738416613914,
      "grad_norm": 0.8014853175758311,
      "learning_rate": 0.00019999984447668622,
      "loss": 12.7345,
      "step": 1122
    },
    {
      "epoch": 0.061151838162722155,
      "grad_norm": 0.9225218876342778,
      "learning_rate": 0.00019999982853555111,
      "loss": 12.751,
      "step": 1123
    },
    {
      "epoch": 0.06120629215930517,
      "grad_norm": 0.8304127287116028,
      "learning_rate": 0.00019999981181680057,
      "loss": 12.7568,
      "step": 1124
    },
    {
      "epoch": 0.06126074615588818,
      "grad_norm": 0.8598278997579997,
      "learning_rate": 0.00019999979432043472,
      "loss": 12.5819,
      "step": 1125
    },
    {
      "epoch": 0.06131520015247119,
      "grad_norm": 0.8826576027775591,
      "learning_rate": 0.00019999977604645368,
      "loss": 12.7199,
      "step": 1126
    },
    {
      "epoch": 0.0613696541490542,
      "grad_norm": 0.804636809658474,
      "learning_rate": 0.00019999975699485763,
      "loss": 12.5365,
      "step": 1127
    },
    {
      "epoch": 0.061424108145637216,
      "grad_norm": 0.8346707911072948,
      "learning_rate": 0.00019999973716564672,
      "loss": 12.7627,
      "step": 1128
    },
    {
      "epoch": 0.06147856214222022,
      "grad_norm": 0.8549089778900794,
      "learning_rate": 0.00019999971655882106,
      "loss": 12.6977,
      "step": 1129
    },
    {
      "epoch": 0.061533016138803236,
      "grad_norm": 0.8472158811477527,
      "learning_rate": 0.0001999996951743808,
      "loss": 12.785,
      "step": 1130
    },
    {
      "epoch": 0.06158747013538625,
      "grad_norm": 0.7968932169209947,
      "learning_rate": 0.00019999967301232623,
      "loss": 12.6148,
      "step": 1131
    },
    {
      "epoch": 0.06164192413196926,
      "grad_norm": 0.8026579849549752,
      "learning_rate": 0.00019999965007265735,
      "loss": 12.6353,
      "step": 1132
    },
    {
      "epoch": 0.06169637812855227,
      "grad_norm": 0.793885187224609,
      "learning_rate": 0.00019999962635537446,
      "loss": 12.6965,
      "step": 1133
    },
    {
      "epoch": 0.06175083212513528,
      "grad_norm": 0.8813812746509808,
      "learning_rate": 0.0001999996018604777,
      "loss": 12.6435,
      "step": 1134
    },
    {
      "epoch": 0.061805286121718296,
      "grad_norm": 0.7731513844126909,
      "learning_rate": 0.00019999957658796725,
      "loss": 12.6346,
      "step": 1135
    },
    {
      "epoch": 0.06185974011830131,
      "grad_norm": 0.8927790695605196,
      "learning_rate": 0.00019999955053784336,
      "loss": 12.8121,
      "step": 1136
    },
    {
      "epoch": 0.061914194114884316,
      "grad_norm": 0.8686464281487228,
      "learning_rate": 0.00019999952371010617,
      "loss": 12.7914,
      "step": 1137
    },
    {
      "epoch": 0.06196864811146733,
      "grad_norm": 0.8148901326895414,
      "learning_rate": 0.0001999994961047559,
      "loss": 12.7512,
      "step": 1138
    },
    {
      "epoch": 0.06202310210805034,
      "grad_norm": 0.8212055968907241,
      "learning_rate": 0.00019999946772179282,
      "loss": 12.7898,
      "step": 1139
    },
    {
      "epoch": 0.06207755610463336,
      "grad_norm": 0.859935862765019,
      "learning_rate": 0.00019999943856121707,
      "loss": 12.7415,
      "step": 1140
    },
    {
      "epoch": 0.06213201010121636,
      "grad_norm": 0.8271274292210274,
      "learning_rate": 0.00019999940862302893,
      "loss": 12.5209,
      "step": 1141
    },
    {
      "epoch": 0.06218646409779938,
      "grad_norm": 0.9194182466076581,
      "learning_rate": 0.0001999993779072286,
      "loss": 12.8127,
      "step": 1142
    },
    {
      "epoch": 0.06224091809438239,
      "grad_norm": 0.7947629670968179,
      "learning_rate": 0.00019999934641381635,
      "loss": 12.5794,
      "step": 1143
    },
    {
      "epoch": 0.062295372090965404,
      "grad_norm": 0.8589934847949596,
      "learning_rate": 0.0001999993141427924,
      "loss": 12.862,
      "step": 1144
    },
    {
      "epoch": 0.06234982608754841,
      "grad_norm": 0.9370285835777666,
      "learning_rate": 0.00019999928109415706,
      "loss": 12.6531,
      "step": 1145
    },
    {
      "epoch": 0.062404280084131424,
      "grad_norm": 0.8650870724910286,
      "learning_rate": 0.00019999924726791051,
      "loss": 12.7084,
      "step": 1146
    },
    {
      "epoch": 0.06245873408071444,
      "grad_norm": 0.8105626060240043,
      "learning_rate": 0.00019999921266405303,
      "loss": 12.6377,
      "step": 1147
    },
    {
      "epoch": 0.06251318807729744,
      "grad_norm": 0.8396580432219618,
      "learning_rate": 0.00019999917728258493,
      "loss": 12.626,
      "step": 1148
    },
    {
      "epoch": 0.06256764207388046,
      "grad_norm": 0.8616677580171016,
      "learning_rate": 0.00019999914112350643,
      "loss": 12.7841,
      "step": 1149
    },
    {
      "epoch": 0.06262209607046347,
      "grad_norm": 0.8225606805167363,
      "learning_rate": 0.00019999910418681783,
      "loss": 12.6528,
      "step": 1150
    },
    {
      "epoch": 0.06267655006704648,
      "grad_norm": 0.8853599634712473,
      "learning_rate": 0.00019999906647251946,
      "loss": 12.9739,
      "step": 1151
    },
    {
      "epoch": 0.0627310040636295,
      "grad_norm": 0.8499949719325622,
      "learning_rate": 0.00019999902798061156,
      "loss": 12.7154,
      "step": 1152
    },
    {
      "epoch": 0.0627854580602125,
      "grad_norm": 0.7984338515402579,
      "learning_rate": 0.00019999898871109445,
      "loss": 12.7525,
      "step": 1153
    },
    {
      "epoch": 0.06283991205679552,
      "grad_norm": 0.8270221917898859,
      "learning_rate": 0.00019999894866396846,
      "loss": 12.6632,
      "step": 1154
    },
    {
      "epoch": 0.06289436605337853,
      "grad_norm": 0.8417958776114451,
      "learning_rate": 0.00019999890783923386,
      "loss": 12.6455,
      "step": 1155
    },
    {
      "epoch": 0.06294882004996154,
      "grad_norm": 0.9328871763734922,
      "learning_rate": 0.00019999886623689098,
      "loss": 12.8024,
      "step": 1156
    },
    {
      "epoch": 0.06300327404654456,
      "grad_norm": 1.0856841110950344,
      "learning_rate": 0.00019999882385694014,
      "loss": 12.6466,
      "step": 1157
    },
    {
      "epoch": 0.06305772804312756,
      "grad_norm": 0.7988820949419523,
      "learning_rate": 0.00019999878069938167,
      "loss": 12.8665,
      "step": 1158
    },
    {
      "epoch": 0.06311218203971057,
      "grad_norm": 0.906338199416811,
      "learning_rate": 0.00019999873676421594,
      "loss": 12.826,
      "step": 1159
    },
    {
      "epoch": 0.06316663603629359,
      "grad_norm": 0.9190725587084829,
      "learning_rate": 0.00019999869205144323,
      "loss": 12.8269,
      "step": 1160
    },
    {
      "epoch": 0.0632210900328766,
      "grad_norm": 0.8270321207992188,
      "learning_rate": 0.00019999864656106392,
      "loss": 12.7298,
      "step": 1161
    },
    {
      "epoch": 0.06327554402945962,
      "grad_norm": 0.7746905689590335,
      "learning_rate": 0.0001999986002930784,
      "loss": 12.559,
      "step": 1162
    },
    {
      "epoch": 0.06332999802604262,
      "grad_norm": 0.8378946907958874,
      "learning_rate": 0.00019999855324748697,
      "loss": 12.8177,
      "step": 1163
    },
    {
      "epoch": 0.06338445202262563,
      "grad_norm": 0.7859625331062614,
      "learning_rate": 0.00019999850542429002,
      "loss": 12.7153,
      "step": 1164
    },
    {
      "epoch": 0.06343890601920865,
      "grad_norm": 0.8331880794666584,
      "learning_rate": 0.00019999845682348792,
      "loss": 12.6974,
      "step": 1165
    },
    {
      "epoch": 0.06349336001579166,
      "grad_norm": 0.962658613381509,
      "learning_rate": 0.00019999840744508107,
      "loss": 12.6871,
      "step": 1166
    },
    {
      "epoch": 0.06354781401237466,
      "grad_norm": 0.819743012172273,
      "learning_rate": 0.00019999835728906984,
      "loss": 12.5648,
      "step": 1167
    },
    {
      "epoch": 0.06360226800895769,
      "grad_norm": 0.8145608023687774,
      "learning_rate": 0.00019999830635545457,
      "loss": 12.7575,
      "step": 1168
    },
    {
      "epoch": 0.06365672200554069,
      "grad_norm": 0.9500062161145167,
      "learning_rate": 0.00019999825464423574,
      "loss": 12.7535,
      "step": 1169
    },
    {
      "epoch": 0.06371117600212371,
      "grad_norm": 0.8339180428418321,
      "learning_rate": 0.0001999982021554137,
      "loss": 12.7517,
      "step": 1170
    },
    {
      "epoch": 0.06376562999870672,
      "grad_norm": 0.8859042989078137,
      "learning_rate": 0.00019999814888898887,
      "loss": 12.7061,
      "step": 1171
    },
    {
      "epoch": 0.06382008399528973,
      "grad_norm": 0.9593859826719203,
      "learning_rate": 0.00019999809484496167,
      "loss": 12.769,
      "step": 1172
    },
    {
      "epoch": 0.06387453799187275,
      "grad_norm": 0.9254926975424156,
      "learning_rate": 0.0001999980400233325,
      "loss": 12.7988,
      "step": 1173
    },
    {
      "epoch": 0.06392899198845575,
      "grad_norm": 0.9218656660623168,
      "learning_rate": 0.00019999798442410177,
      "loss": 12.6074,
      "step": 1174
    },
    {
      "epoch": 0.06398344598503876,
      "grad_norm": 0.9195323497831199,
      "learning_rate": 0.00019999792804727,
      "loss": 12.7693,
      "step": 1175
    },
    {
      "epoch": 0.06403789998162178,
      "grad_norm": 0.8757789350337974,
      "learning_rate": 0.00019999787089283757,
      "loss": 12.666,
      "step": 1176
    },
    {
      "epoch": 0.06409235397820479,
      "grad_norm": 0.896203375036291,
      "learning_rate": 0.0001999978129608049,
      "loss": 12.8525,
      "step": 1177
    },
    {
      "epoch": 0.0641468079747878,
      "grad_norm": 0.9466387119592451,
      "learning_rate": 0.0001999977542511725,
      "loss": 12.7746,
      "step": 1178
    },
    {
      "epoch": 0.06420126197137081,
      "grad_norm": 0.9107062802083105,
      "learning_rate": 0.00019999769476394076,
      "loss": 12.7753,
      "step": 1179
    },
    {
      "epoch": 0.06425571596795382,
      "grad_norm": 0.8633923015901338,
      "learning_rate": 0.00019999763449911017,
      "loss": 12.7459,
      "step": 1180
    },
    {
      "epoch": 0.06431016996453684,
      "grad_norm": 0.8523060925255115,
      "learning_rate": 0.00019999757345668122,
      "loss": 12.7009,
      "step": 1181
    },
    {
      "epoch": 0.06436462396111985,
      "grad_norm": 0.8732343539755694,
      "learning_rate": 0.00019999751163665437,
      "loss": 12.8505,
      "step": 1182
    },
    {
      "epoch": 0.06441907795770285,
      "grad_norm": 0.8955887659856802,
      "learning_rate": 0.00019999744903903007,
      "loss": 12.6876,
      "step": 1183
    },
    {
      "epoch": 0.06447353195428587,
      "grad_norm": 0.8869982370172741,
      "learning_rate": 0.00019999738566380887,
      "loss": 12.6886,
      "step": 1184
    },
    {
      "epoch": 0.06452798595086888,
      "grad_norm": 0.8861061504796613,
      "learning_rate": 0.00019999732151099124,
      "loss": 12.892,
      "step": 1185
    },
    {
      "epoch": 0.0645824399474519,
      "grad_norm": 0.8650163616204314,
      "learning_rate": 0.00019999725658057766,
      "loss": 12.6152,
      "step": 1186
    },
    {
      "epoch": 0.0646368939440349,
      "grad_norm": 0.7317105713251505,
      "learning_rate": 0.00019999719087256864,
      "loss": 12.6659,
      "step": 1187
    },
    {
      "epoch": 0.06469134794061791,
      "grad_norm": 0.9458481135838617,
      "learning_rate": 0.00019999712438696467,
      "loss": 12.9147,
      "step": 1188
    },
    {
      "epoch": 0.06474580193720093,
      "grad_norm": 0.7615024396322205,
      "learning_rate": 0.00019999705712376632,
      "loss": 12.7225,
      "step": 1189
    },
    {
      "epoch": 0.06480025593378394,
      "grad_norm": 0.8568408849038813,
      "learning_rate": 0.00019999698908297408,
      "loss": 12.5842,
      "step": 1190
    },
    {
      "epoch": 0.06485470993036695,
      "grad_norm": 0.814394862279894,
      "learning_rate": 0.00019999692026458847,
      "loss": 12.8761,
      "step": 1191
    },
    {
      "epoch": 0.06490916392694997,
      "grad_norm": 0.8714145177343299,
      "learning_rate": 0.00019999685066861007,
      "loss": 12.8574,
      "step": 1192
    },
    {
      "epoch": 0.06496361792353297,
      "grad_norm": 0.8264656686816955,
      "learning_rate": 0.00019999678029503936,
      "loss": 12.7122,
      "step": 1193
    },
    {
      "epoch": 0.06501807192011598,
      "grad_norm": 0.8104604521918491,
      "learning_rate": 0.00019999670914387695,
      "loss": 12.6977,
      "step": 1194
    },
    {
      "epoch": 0.065072525916699,
      "grad_norm": 0.8354849449147481,
      "learning_rate": 0.0001999966372151233,
      "loss": 12.7495,
      "step": 1195
    },
    {
      "epoch": 0.06512697991328201,
      "grad_norm": 0.8670770016037759,
      "learning_rate": 0.00019999656450877908,
      "loss": 12.8445,
      "step": 1196
    },
    {
      "epoch": 0.06518143390986503,
      "grad_norm": 0.761947793544306,
      "learning_rate": 0.00019999649102484475,
      "loss": 12.7165,
      "step": 1197
    },
    {
      "epoch": 0.06523588790644803,
      "grad_norm": 0.8575494256497836,
      "learning_rate": 0.00019999641676332098,
      "loss": 12.7475,
      "step": 1198
    },
    {
      "epoch": 0.06529034190303104,
      "grad_norm": 0.9446326318271256,
      "learning_rate": 0.00019999634172420834,
      "loss": 12.7995,
      "step": 1199
    },
    {
      "epoch": 0.06534479589961406,
      "grad_norm": 0.8926293577591496,
      "learning_rate": 0.00019999626590750733,
      "loss": 12.8856,
      "step": 1200
    },
    {
      "epoch": 0.06539924989619707,
      "grad_norm": 0.8785564167937711,
      "learning_rate": 0.00019999618931321859,
      "loss": 12.4816,
      "step": 1201
    },
    {
      "epoch": 0.06545370389278007,
      "grad_norm": 0.8769465556308361,
      "learning_rate": 0.00019999611194134272,
      "loss": 12.7773,
      "step": 1202
    },
    {
      "epoch": 0.0655081578893631,
      "grad_norm": 0.8078239982110514,
      "learning_rate": 0.0001999960337918803,
      "loss": 12.6893,
      "step": 1203
    },
    {
      "epoch": 0.0655626118859461,
      "grad_norm": 0.8764038524011952,
      "learning_rate": 0.000199995954864832,
      "loss": 12.6881,
      "step": 1204
    },
    {
      "epoch": 0.06561706588252912,
      "grad_norm": 1.0046378226101262,
      "learning_rate": 0.00019999587516019834,
      "loss": 12.6412,
      "step": 1205
    },
    {
      "epoch": 0.06567151987911213,
      "grad_norm": 0.8499068641220393,
      "learning_rate": 0.00019999579467797998,
      "loss": 12.8251,
      "step": 1206
    },
    {
      "epoch": 0.06572597387569513,
      "grad_norm": 0.8114198595859603,
      "learning_rate": 0.00019999571341817755,
      "loss": 12.7518,
      "step": 1207
    },
    {
      "epoch": 0.06578042787227815,
      "grad_norm": 0.7826828314807258,
      "learning_rate": 0.0001999956313807917,
      "loss": 12.773,
      "step": 1208
    },
    {
      "epoch": 0.06583488186886116,
      "grad_norm": 0.9134842134989745,
      "learning_rate": 0.00019999554856582304,
      "loss": 12.6953,
      "step": 1209
    },
    {
      "epoch": 0.06588933586544417,
      "grad_norm": 0.8648511371370636,
      "learning_rate": 0.00019999546497327227,
      "loss": 12.6783,
      "step": 1210
    },
    {
      "epoch": 0.06594378986202719,
      "grad_norm": 0.937972103939476,
      "learning_rate": 0.00019999538060313995,
      "loss": 12.7148,
      "step": 1211
    },
    {
      "epoch": 0.0659982438586102,
      "grad_norm": 0.7882262998747774,
      "learning_rate": 0.00019999529545542677,
      "loss": 12.5461,
      "step": 1212
    },
    {
      "epoch": 0.06605269785519322,
      "grad_norm": 0.8935868580453479,
      "learning_rate": 0.00019999520953013344,
      "loss": 12.8204,
      "step": 1213
    },
    {
      "epoch": 0.06610715185177622,
      "grad_norm": 0.8836712629861396,
      "learning_rate": 0.00019999512282726055,
      "loss": 12.6649,
      "step": 1214
    },
    {
      "epoch": 0.06616160584835923,
      "grad_norm": 0.7804057926443057,
      "learning_rate": 0.00019999503534680888,
      "loss": 12.7262,
      "step": 1215
    },
    {
      "epoch": 0.06621605984494225,
      "grad_norm": 0.8700122862674238,
      "learning_rate": 0.000199994947088779,
      "loss": 12.8055,
      "step": 1216
    },
    {
      "epoch": 0.06627051384152526,
      "grad_norm": 0.8834004841790215,
      "learning_rate": 0.00019999485805317164,
      "loss": 12.762,
      "step": 1217
    },
    {
      "epoch": 0.06632496783810826,
      "grad_norm": 0.8343108775202628,
      "learning_rate": 0.00019999476823998752,
      "loss": 12.6515,
      "step": 1218
    },
    {
      "epoch": 0.06637942183469128,
      "grad_norm": 0.8579726093320554,
      "learning_rate": 0.00019999467764922728,
      "loss": 12.7467,
      "step": 1219
    },
    {
      "epoch": 0.06643387583127429,
      "grad_norm": 0.8654133283698366,
      "learning_rate": 0.00019999458628089167,
      "loss": 12.7794,
      "step": 1220
    },
    {
      "epoch": 0.06648832982785731,
      "grad_norm": 0.8325988034952059,
      "learning_rate": 0.00019999449413498138,
      "loss": 12.6766,
      "step": 1221
    },
    {
      "epoch": 0.06654278382444032,
      "grad_norm": 0.7726895332022075,
      "learning_rate": 0.00019999440121149715,
      "loss": 12.7095,
      "step": 1222
    },
    {
      "epoch": 0.06659723782102332,
      "grad_norm": 0.8422111747797787,
      "learning_rate": 0.00019999430751043972,
      "loss": 12.7101,
      "step": 1223
    },
    {
      "epoch": 0.06665169181760634,
      "grad_norm": 0.8713740577866859,
      "learning_rate": 0.00019999421303180972,
      "loss": 12.7866,
      "step": 1224
    },
    {
      "epoch": 0.06670614581418935,
      "grad_norm": 0.7861309399978694,
      "learning_rate": 0.000199994117775608,
      "loss": 12.5477,
      "step": 1225
    },
    {
      "epoch": 0.06676059981077236,
      "grad_norm": 0.7909438744345703,
      "learning_rate": 0.00019999402174183524,
      "loss": 12.6011,
      "step": 1226
    },
    {
      "epoch": 0.06681505380735538,
      "grad_norm": 0.805144263404179,
      "learning_rate": 0.00019999392493049215,
      "loss": 12.6455,
      "step": 1227
    },
    {
      "epoch": 0.06686950780393838,
      "grad_norm": 0.9161247444164494,
      "learning_rate": 0.0001999938273415796,
      "loss": 12.9511,
      "step": 1228
    },
    {
      "epoch": 0.0669239618005214,
      "grad_norm": 0.8007339932273023,
      "learning_rate": 0.00019999372897509826,
      "loss": 12.4325,
      "step": 1229
    },
    {
      "epoch": 0.06697841579710441,
      "grad_norm": 0.8580605244250328,
      "learning_rate": 0.00019999362983104887,
      "loss": 12.7317,
      "step": 1230
    },
    {
      "epoch": 0.06703286979368742,
      "grad_norm": 0.9518365532593344,
      "learning_rate": 0.0001999935299094323,
      "loss": 12.8734,
      "step": 1231
    },
    {
      "epoch": 0.06708732379027044,
      "grad_norm": 0.8239272304844115,
      "learning_rate": 0.00019999342921024927,
      "loss": 12.7499,
      "step": 1232
    },
    {
      "epoch": 0.06714177778685344,
      "grad_norm": 0.8968558651975471,
      "learning_rate": 0.00019999332773350053,
      "loss": 12.8772,
      "step": 1233
    },
    {
      "epoch": 0.06719623178343645,
      "grad_norm": 0.9650739082159776,
      "learning_rate": 0.00019999322547918692,
      "loss": 12.864,
      "step": 1234
    },
    {
      "epoch": 0.06725068578001947,
      "grad_norm": 0.8170253300214879,
      "learning_rate": 0.00019999312244730924,
      "loss": 12.8578,
      "step": 1235
    },
    {
      "epoch": 0.06730513977660248,
      "grad_norm": 0.8107661672269623,
      "learning_rate": 0.00019999301863786825,
      "loss": 12.6231,
      "step": 1236
    },
    {
      "epoch": 0.0673595937731855,
      "grad_norm": 0.8584076187816582,
      "learning_rate": 0.00019999291405086477,
      "loss": 12.8479,
      "step": 1237
    },
    {
      "epoch": 0.0674140477697685,
      "grad_norm": 0.8387218124859667,
      "learning_rate": 0.00019999280868629964,
      "loss": 12.8035,
      "step": 1238
    },
    {
      "epoch": 0.06746850176635151,
      "grad_norm": 0.833310340002492,
      "learning_rate": 0.00019999270254417363,
      "loss": 12.813,
      "step": 1239
    },
    {
      "epoch": 0.06752295576293453,
      "grad_norm": 0.7848080365904714,
      "learning_rate": 0.00019999259562448766,
      "loss": 12.5327,
      "step": 1240
    },
    {
      "epoch": 0.06757740975951754,
      "grad_norm": 0.8079709147478482,
      "learning_rate": 0.00019999248792724244,
      "loss": 12.7529,
      "step": 1241
    },
    {
      "epoch": 0.06763186375610054,
      "grad_norm": 0.797562748653487,
      "learning_rate": 0.0001999923794524389,
      "loss": 12.8353,
      "step": 1242
    },
    {
      "epoch": 0.06768631775268356,
      "grad_norm": 0.7568940240260569,
      "learning_rate": 0.00019999227020007783,
      "loss": 12.5846,
      "step": 1243
    },
    {
      "epoch": 0.06774077174926657,
      "grad_norm": 0.9021563734495636,
      "learning_rate": 0.00019999216017016006,
      "loss": 12.7845,
      "step": 1244
    },
    {
      "epoch": 0.06779522574584959,
      "grad_norm": 0.7172017086809768,
      "learning_rate": 0.00019999204936268656,
      "loss": 12.3948,
      "step": 1245
    },
    {
      "epoch": 0.0678496797424326,
      "grad_norm": 0.899680702607057,
      "learning_rate": 0.00019999193777765805,
      "loss": 12.6401,
      "step": 1246
    },
    {
      "epoch": 0.0679041337390156,
      "grad_norm": 0.8631946648104156,
      "learning_rate": 0.0001999918254150755,
      "loss": 12.7803,
      "step": 1247
    },
    {
      "epoch": 0.06795858773559862,
      "grad_norm": 0.8901750091644829,
      "learning_rate": 0.00019999171227493974,
      "loss": 12.7478,
      "step": 1248
    },
    {
      "epoch": 0.06801304173218163,
      "grad_norm": 0.8298302381975877,
      "learning_rate": 0.00019999159835725166,
      "loss": 12.6882,
      "step": 1249
    },
    {
      "epoch": 0.06806749572876464,
      "grad_norm": 0.8391719325869385,
      "learning_rate": 0.00019999148366201214,
      "loss": 12.7615,
      "step": 1250
    },
    {
      "epoch": 0.06812194972534766,
      "grad_norm": 0.9542640810854296,
      "learning_rate": 0.0001999913681892221,
      "loss": 12.8767,
      "step": 1251
    },
    {
      "epoch": 0.06817640372193066,
      "grad_norm": 0.7886309533998018,
      "learning_rate": 0.00019999125193888238,
      "loss": 12.6535,
      "step": 1252
    },
    {
      "epoch": 0.06823085771851368,
      "grad_norm": 0.8938502848779079,
      "learning_rate": 0.0001999911349109939,
      "loss": 12.7479,
      "step": 1253
    },
    {
      "epoch": 0.06828531171509669,
      "grad_norm": 1.01894846212262,
      "learning_rate": 0.00019999101710555762,
      "loss": 12.7659,
      "step": 1254
    },
    {
      "epoch": 0.0683397657116797,
      "grad_norm": 0.8200912021011945,
      "learning_rate": 0.0001999908985225744,
      "loss": 12.7416,
      "step": 1255
    },
    {
      "epoch": 0.06839421970826272,
      "grad_norm": 0.7572550774775388,
      "learning_rate": 0.00019999077916204517,
      "loss": 12.7203,
      "step": 1256
    },
    {
      "epoch": 0.06844867370484572,
      "grad_norm": 0.9595690781436814,
      "learning_rate": 0.00019999065902397093,
      "loss": 12.9055,
      "step": 1257
    },
    {
      "epoch": 0.06850312770142873,
      "grad_norm": 0.9216699050913904,
      "learning_rate": 0.00019999053810835254,
      "loss": 12.9216,
      "step": 1258
    },
    {
      "epoch": 0.06855758169801175,
      "grad_norm": 0.7783220350551281,
      "learning_rate": 0.00019999041641519095,
      "loss": 12.7104,
      "step": 1259
    },
    {
      "epoch": 0.06861203569459476,
      "grad_norm": 0.7964697170367746,
      "learning_rate": 0.0001999902939444871,
      "loss": 12.5829,
      "step": 1260
    },
    {
      "epoch": 0.06866648969117777,
      "grad_norm": 0.947797767672639,
      "learning_rate": 0.00019999017069624193,
      "loss": 12.8261,
      "step": 1261
    },
    {
      "epoch": 0.06872094368776079,
      "grad_norm": 0.9990216177536839,
      "learning_rate": 0.00019999004667045647,
      "loss": 12.6765,
      "step": 1262
    },
    {
      "epoch": 0.06877539768434379,
      "grad_norm": 0.9003760583623563,
      "learning_rate": 0.00019998992186713165,
      "loss": 12.7849,
      "step": 1263
    },
    {
      "epoch": 0.06882985168092681,
      "grad_norm": 0.8353313742840284,
      "learning_rate": 0.00019998979628626837,
      "loss": 12.6698,
      "step": 1264
    },
    {
      "epoch": 0.06888430567750982,
      "grad_norm": 0.8651149020580153,
      "learning_rate": 0.00019998966992786768,
      "loss": 12.6586,
      "step": 1265
    },
    {
      "epoch": 0.06893875967409283,
      "grad_norm": 0.9382816369907496,
      "learning_rate": 0.0001999895427919306,
      "loss": 12.9221,
      "step": 1266
    },
    {
      "epoch": 0.06899321367067585,
      "grad_norm": 0.7998177833076818,
      "learning_rate": 0.00019998941487845803,
      "loss": 12.6578,
      "step": 1267
    },
    {
      "epoch": 0.06904766766725885,
      "grad_norm": 0.8184372156049693,
      "learning_rate": 0.00019998928618745102,
      "loss": 12.7793,
      "step": 1268
    },
    {
      "epoch": 0.06910212166384186,
      "grad_norm": 0.8071861459286777,
      "learning_rate": 0.00019998915671891055,
      "loss": 12.3847,
      "step": 1269
    },
    {
      "epoch": 0.06915657566042488,
      "grad_norm": 0.8132926016895157,
      "learning_rate": 0.0001999890264728376,
      "loss": 12.8209,
      "step": 1270
    },
    {
      "epoch": 0.06921102965700789,
      "grad_norm": 0.8625809229455872,
      "learning_rate": 0.00019998889544923322,
      "loss": 12.6077,
      "step": 1271
    },
    {
      "epoch": 0.0692654836535909,
      "grad_norm": 0.877250661583724,
      "learning_rate": 0.00019998876364809843,
      "loss": 12.888,
      "step": 1272
    },
    {
      "epoch": 0.06931993765017391,
      "grad_norm": 0.9665899385261744,
      "learning_rate": 0.00019998863106943427,
      "loss": 12.8492,
      "step": 1273
    },
    {
      "epoch": 0.06937439164675692,
      "grad_norm": 0.8873863859176445,
      "learning_rate": 0.0001999884977132417,
      "loss": 12.7165,
      "step": 1274
    },
    {
      "epoch": 0.06942884564333994,
      "grad_norm": 0.8998207159482029,
      "learning_rate": 0.00019998836357952183,
      "loss": 12.7232,
      "step": 1275
    },
    {
      "epoch": 0.06948329963992295,
      "grad_norm": 0.9357481029675431,
      "learning_rate": 0.00019998822866827568,
      "loss": 12.7706,
      "step": 1276
    },
    {
      "epoch": 0.06953775363650595,
      "grad_norm": 0.9390340910364658,
      "learning_rate": 0.0001999880929795043,
      "loss": 12.8162,
      "step": 1277
    },
    {
      "epoch": 0.06959220763308897,
      "grad_norm": 0.8788392373945151,
      "learning_rate": 0.00019998795651320875,
      "loss": 12.7985,
      "step": 1278
    },
    {
      "epoch": 0.06964666162967198,
      "grad_norm": 0.9910757288059426,
      "learning_rate": 0.00019998781926939004,
      "loss": 12.7973,
      "step": 1279
    },
    {
      "epoch": 0.069701115626255,
      "grad_norm": 0.7825317381332015,
      "learning_rate": 0.00019998768124804931,
      "loss": 12.7743,
      "step": 1280
    },
    {
      "epoch": 0.069755569622838,
      "grad_norm": 0.9466682022844718,
      "learning_rate": 0.0001999875424491876,
      "loss": 12.8456,
      "step": 1281
    },
    {
      "epoch": 0.06981002361942101,
      "grad_norm": 0.8392171551646117,
      "learning_rate": 0.00019998740287280597,
      "loss": 12.7803,
      "step": 1282
    },
    {
      "epoch": 0.06986447761600403,
      "grad_norm": 0.9682561669853998,
      "learning_rate": 0.00019998726251890556,
      "loss": 12.7737,
      "step": 1283
    },
    {
      "epoch": 0.06991893161258704,
      "grad_norm": 1.0730646542580322,
      "learning_rate": 0.0001999871213874874,
      "loss": 12.6476,
      "step": 1284
    },
    {
      "epoch": 0.06997338560917005,
      "grad_norm": 0.8438483865520613,
      "learning_rate": 0.00019998697947855263,
      "loss": 12.8169,
      "step": 1285
    },
    {
      "epoch": 0.07002783960575307,
      "grad_norm": 0.883713442906129,
      "learning_rate": 0.00019998683679210236,
      "loss": 12.7374,
      "step": 1286
    },
    {
      "epoch": 0.07008229360233607,
      "grad_norm": 0.940886253494083,
      "learning_rate": 0.00019998669332813764,
      "loss": 12.7748,
      "step": 1287
    },
    {
      "epoch": 0.0701367475989191,
      "grad_norm": 0.7718253307822253,
      "learning_rate": 0.00019998654908665966,
      "loss": 12.665,
      "step": 1288
    },
    {
      "epoch": 0.0701912015955021,
      "grad_norm": 0.9335549581515296,
      "learning_rate": 0.00019998640406766947,
      "loss": 12.728,
      "step": 1289
    },
    {
      "epoch": 0.07024565559208511,
      "grad_norm": 0.7689425613584284,
      "learning_rate": 0.00019998625827116827,
      "loss": 12.7429,
      "step": 1290
    },
    {
      "epoch": 0.07030010958866813,
      "grad_norm": 0.8506528158595922,
      "learning_rate": 0.00019998611169715712,
      "loss": 12.8497,
      "step": 1291
    },
    {
      "epoch": 0.07035456358525113,
      "grad_norm": 0.7830197778249123,
      "learning_rate": 0.00019998596434563724,
      "loss": 12.8308,
      "step": 1292
    },
    {
      "epoch": 0.07040901758183414,
      "grad_norm": 0.7674751880153996,
      "learning_rate": 0.00019998581621660973,
      "loss": 12.6265,
      "step": 1293
    },
    {
      "epoch": 0.07046347157841716,
      "grad_norm": 0.8792718237656185,
      "learning_rate": 0.0001999856673100757,
      "loss": 12.6221,
      "step": 1294
    },
    {
      "epoch": 0.07051792557500017,
      "grad_norm": 0.888901523985518,
      "learning_rate": 0.0001999855176260364,
      "loss": 12.6924,
      "step": 1295
    },
    {
      "epoch": 0.07057237957158319,
      "grad_norm": 0.8764202643621799,
      "learning_rate": 0.00019998536716449292,
      "loss": 12.7861,
      "step": 1296
    },
    {
      "epoch": 0.0706268335681662,
      "grad_norm": 1.0514054139949702,
      "learning_rate": 0.00019998521592544646,
      "loss": 12.738,
      "step": 1297
    },
    {
      "epoch": 0.0706812875647492,
      "grad_norm": 0.9078005384364782,
      "learning_rate": 0.0001999850639088982,
      "loss": 12.6706,
      "step": 1298
    },
    {
      "epoch": 0.07073574156133222,
      "grad_norm": 1.0033928702575605,
      "learning_rate": 0.00019998491111484934,
      "loss": 12.7786,
      "step": 1299
    },
    {
      "epoch": 0.07079019555791523,
      "grad_norm": 0.9070619957929708,
      "learning_rate": 0.000199984757543301,
      "loss": 12.7996,
      "step": 1300
    },
    {
      "epoch": 0.07084464955449823,
      "grad_norm": 0.8637899760457025,
      "learning_rate": 0.00019998460319425445,
      "loss": 12.6295,
      "step": 1301
    },
    {
      "epoch": 0.07089910355108126,
      "grad_norm": 0.9270227663127247,
      "learning_rate": 0.00019998444806771084,
      "loss": 12.693,
      "step": 1302
    },
    {
      "epoch": 0.07095355754766426,
      "grad_norm": 0.8678327197234903,
      "learning_rate": 0.0001999842921636714,
      "loss": 12.6845,
      "step": 1303
    },
    {
      "epoch": 0.07100801154424728,
      "grad_norm": 0.9123627237275738,
      "learning_rate": 0.00019998413548213734,
      "loss": 12.738,
      "step": 1304
    },
    {
      "epoch": 0.07106246554083029,
      "grad_norm": 0.7763699733385914,
      "learning_rate": 0.00019998397802310986,
      "loss": 12.6883,
      "step": 1305
    },
    {
      "epoch": 0.0711169195374133,
      "grad_norm": 1.0035685910073238,
      "learning_rate": 0.00019998381978659024,
      "loss": 12.889,
      "step": 1306
    },
    {
      "epoch": 0.07117137353399632,
      "grad_norm": 0.890221905170563,
      "learning_rate": 0.00019998366077257962,
      "loss": 12.8601,
      "step": 1307
    },
    {
      "epoch": 0.07122582753057932,
      "grad_norm": 0.8082291689304159,
      "learning_rate": 0.0001999835009810793,
      "loss": 12.7015,
      "step": 1308
    },
    {
      "epoch": 0.07128028152716233,
      "grad_norm": 0.8360757752561405,
      "learning_rate": 0.00019998334041209054,
      "loss": 12.7729,
      "step": 1309
    },
    {
      "epoch": 0.07133473552374535,
      "grad_norm": 0.8218382034287375,
      "learning_rate": 0.00019998317906561454,
      "loss": 12.8634,
      "step": 1310
    },
    {
      "epoch": 0.07138918952032836,
      "grad_norm": 0.924703496644343,
      "learning_rate": 0.00019998301694165255,
      "loss": 12.7538,
      "step": 1311
    },
    {
      "epoch": 0.07144364351691138,
      "grad_norm": 0.8849010458774731,
      "learning_rate": 0.00019998285404020588,
      "loss": 12.7549,
      "step": 1312
    },
    {
      "epoch": 0.07149809751349438,
      "grad_norm": 0.7943815027792276,
      "learning_rate": 0.00019998269036127577,
      "loss": 12.6778,
      "step": 1313
    },
    {
      "epoch": 0.07155255151007739,
      "grad_norm": 0.7558448698202527,
      "learning_rate": 0.00019998252590486346,
      "loss": 12.7548,
      "step": 1314
    },
    {
      "epoch": 0.07160700550666041,
      "grad_norm": 0.8911125068611316,
      "learning_rate": 0.00019998236067097033,
      "loss": 12.7131,
      "step": 1315
    },
    {
      "epoch": 0.07166145950324342,
      "grad_norm": 0.8561803781699435,
      "learning_rate": 0.00019998219465959752,
      "loss": 12.8283,
      "step": 1316
    },
    {
      "epoch": 0.07171591349982642,
      "grad_norm": 0.8473732850401972,
      "learning_rate": 0.00019998202787074645,
      "loss": 12.8933,
      "step": 1317
    },
    {
      "epoch": 0.07177036749640944,
      "grad_norm": 0.727468497024035,
      "learning_rate": 0.00019998186030441832,
      "loss": 12.6992,
      "step": 1318
    },
    {
      "epoch": 0.07182482149299245,
      "grad_norm": 0.8180382041248009,
      "learning_rate": 0.00019998169196061452,
      "loss": 12.6889,
      "step": 1319
    },
    {
      "epoch": 0.07187927548957547,
      "grad_norm": 0.7852274604537989,
      "learning_rate": 0.0001999815228393363,
      "loss": 12.7212,
      "step": 1320
    },
    {
      "epoch": 0.07193372948615848,
      "grad_norm": 0.9769106293559628,
      "learning_rate": 0.00019998135294058497,
      "loss": 12.8875,
      "step": 1321
    },
    {
      "epoch": 0.07198818348274148,
      "grad_norm": 0.8503420912576485,
      "learning_rate": 0.0001999811822643619,
      "loss": 12.7531,
      "step": 1322
    },
    {
      "epoch": 0.0720426374793245,
      "grad_norm": 0.7948220930979146,
      "learning_rate": 0.00019998101081066837,
      "loss": 12.5656,
      "step": 1323
    },
    {
      "epoch": 0.07209709147590751,
      "grad_norm": 0.8488679894413691,
      "learning_rate": 0.00019998083857950577,
      "loss": 12.8234,
      "step": 1324
    },
    {
      "epoch": 0.07215154547249052,
      "grad_norm": 0.9413227053637365,
      "learning_rate": 0.00019998066557087537,
      "loss": 12.7785,
      "step": 1325
    },
    {
      "epoch": 0.07220599946907354,
      "grad_norm": 0.7645388464604446,
      "learning_rate": 0.00019998049178477853,
      "loss": 12.6192,
      "step": 1326
    },
    {
      "epoch": 0.07226045346565654,
      "grad_norm": 0.9032030787263513,
      "learning_rate": 0.00019998031722121663,
      "loss": 12.8004,
      "step": 1327
    },
    {
      "epoch": 0.07231490746223955,
      "grad_norm": 0.9060112901526959,
      "learning_rate": 0.00019998014188019105,
      "loss": 12.8844,
      "step": 1328
    },
    {
      "epoch": 0.07236936145882257,
      "grad_norm": 0.9041037289993008,
      "learning_rate": 0.00019997996576170312,
      "loss": 12.6895,
      "step": 1329
    },
    {
      "epoch": 0.07242381545540558,
      "grad_norm": 0.8757460620636227,
      "learning_rate": 0.00019997978886575416,
      "loss": 12.7804,
      "step": 1330
    },
    {
      "epoch": 0.0724782694519886,
      "grad_norm": 0.8629108827418805,
      "learning_rate": 0.00019997961119234563,
      "loss": 12.6409,
      "step": 1331
    },
    {
      "epoch": 0.0725327234485716,
      "grad_norm": 0.8312445852962548,
      "learning_rate": 0.00019997943274147889,
      "loss": 12.7261,
      "step": 1332
    },
    {
      "epoch": 0.07258717744515461,
      "grad_norm": 0.8520995556695088,
      "learning_rate": 0.00019997925351315527,
      "loss": 12.6756,
      "step": 1333
    },
    {
      "epoch": 0.07264163144173763,
      "grad_norm": 0.9686838517002854,
      "learning_rate": 0.00019997907350737624,
      "loss": 12.8004,
      "step": 1334
    },
    {
      "epoch": 0.07269608543832064,
      "grad_norm": 0.9052860715073252,
      "learning_rate": 0.0001999788927241432,
      "loss": 12.8299,
      "step": 1335
    },
    {
      "epoch": 0.07275053943490364,
      "grad_norm": 0.7694932364124429,
      "learning_rate": 0.00019997871116345746,
      "loss": 12.5894,
      "step": 1336
    },
    {
      "epoch": 0.07280499343148666,
      "grad_norm": 0.8697208727797224,
      "learning_rate": 0.00019997852882532052,
      "loss": 12.7327,
      "step": 1337
    },
    {
      "epoch": 0.07285944742806967,
      "grad_norm": 0.7950392651172854,
      "learning_rate": 0.00019997834570973378,
      "loss": 12.6369,
      "step": 1338
    },
    {
      "epoch": 0.07291390142465269,
      "grad_norm": 0.9149058956676394,
      "learning_rate": 0.00019997816181669865,
      "loss": 12.6927,
      "step": 1339
    },
    {
      "epoch": 0.0729683554212357,
      "grad_norm": 0.7917477900076075,
      "learning_rate": 0.00019997797714621656,
      "loss": 12.6357,
      "step": 1340
    },
    {
      "epoch": 0.0730228094178187,
      "grad_norm": 0.8974376761470795,
      "learning_rate": 0.00019997779169828896,
      "loss": 12.5907,
      "step": 1341
    },
    {
      "epoch": 0.07307726341440172,
      "grad_norm": 0.9046560938278421,
      "learning_rate": 0.0001999776054729173,
      "loss": 12.63,
      "step": 1342
    },
    {
      "epoch": 0.07313171741098473,
      "grad_norm": 0.8687122090517582,
      "learning_rate": 0.000199977418470103,
      "loss": 12.6174,
      "step": 1343
    },
    {
      "epoch": 0.07318617140756774,
      "grad_norm": 0.8219971025694719,
      "learning_rate": 0.00019997723068984754,
      "loss": 12.7169,
      "step": 1344
    },
    {
      "epoch": 0.07324062540415076,
      "grad_norm": 0.9499603674891286,
      "learning_rate": 0.00019997704213215234,
      "loss": 12.7951,
      "step": 1345
    },
    {
      "epoch": 0.07329507940073376,
      "grad_norm": 0.8403131431498408,
      "learning_rate": 0.00019997685279701889,
      "loss": 12.6047,
      "step": 1346
    },
    {
      "epoch": 0.07334953339731679,
      "grad_norm": 1.0116847450199355,
      "learning_rate": 0.00019997666268444872,
      "loss": 12.7429,
      "step": 1347
    },
    {
      "epoch": 0.07340398739389979,
      "grad_norm": 0.8687286331334413,
      "learning_rate": 0.00019997647179444323,
      "loss": 12.7573,
      "step": 1348
    },
    {
      "epoch": 0.0734584413904828,
      "grad_norm": 0.9707217570791576,
      "learning_rate": 0.0001999762801270039,
      "loss": 12.6194,
      "step": 1349
    },
    {
      "epoch": 0.07351289538706582,
      "grad_norm": 0.913308078894722,
      "learning_rate": 0.0001999760876821323,
      "loss": 12.791,
      "step": 1350
    },
    {
      "epoch": 0.07356734938364883,
      "grad_norm": 0.870495234403029,
      "learning_rate": 0.00019997589445982982,
      "loss": 12.6617,
      "step": 1351
    },
    {
      "epoch": 0.07362180338023183,
      "grad_norm": 0.9077511397743955,
      "learning_rate": 0.00019997570046009807,
      "loss": 12.7826,
      "step": 1352
    },
    {
      "epoch": 0.07367625737681485,
      "grad_norm": 0.865216146586963,
      "learning_rate": 0.00019997550568293847,
      "loss": 12.6991,
      "step": 1353
    },
    {
      "epoch": 0.07373071137339786,
      "grad_norm": 0.8039294697903542,
      "learning_rate": 0.00019997531012835257,
      "loss": 12.718,
      "step": 1354
    },
    {
      "epoch": 0.07378516536998088,
      "grad_norm": 0.8036977410464696,
      "learning_rate": 0.00019997511379634192,
      "loss": 12.6545,
      "step": 1355
    },
    {
      "epoch": 0.07383961936656389,
      "grad_norm": 0.9204505253050527,
      "learning_rate": 0.00019997491668690803,
      "loss": 12.6631,
      "step": 1356
    },
    {
      "epoch": 0.07389407336314689,
      "grad_norm": 0.8244381847188533,
      "learning_rate": 0.0001999747188000524,
      "loss": 12.7989,
      "step": 1357
    },
    {
      "epoch": 0.07394852735972991,
      "grad_norm": 0.794070214572969,
      "learning_rate": 0.00019997452013577658,
      "loss": 12.6709,
      "step": 1358
    },
    {
      "epoch": 0.07400298135631292,
      "grad_norm": 0.8115791022284864,
      "learning_rate": 0.00019997432069408214,
      "loss": 12.7089,
      "step": 1359
    },
    {
      "epoch": 0.07405743535289593,
      "grad_norm": 0.8152891881701941,
      "learning_rate": 0.00019997412047497058,
      "loss": 12.6603,
      "step": 1360
    },
    {
      "epoch": 0.07411188934947895,
      "grad_norm": 0.8932045405545638,
      "learning_rate": 0.00019997391947844354,
      "loss": 12.6571,
      "step": 1361
    },
    {
      "epoch": 0.07416634334606195,
      "grad_norm": 0.8189990385705692,
      "learning_rate": 0.00019997371770450256,
      "loss": 12.7136,
      "step": 1362
    },
    {
      "epoch": 0.07422079734264497,
      "grad_norm": 0.8729171047097313,
      "learning_rate": 0.00019997351515314913,
      "loss": 12.744,
      "step": 1363
    },
    {
      "epoch": 0.07427525133922798,
      "grad_norm": 0.9197472633047581,
      "learning_rate": 0.0001999733118243849,
      "loss": 12.6353,
      "step": 1364
    },
    {
      "epoch": 0.07432970533581099,
      "grad_norm": 0.8243862529909959,
      "learning_rate": 0.00019997310771821143,
      "loss": 12.8734,
      "step": 1365
    },
    {
      "epoch": 0.074384159332394,
      "grad_norm": 0.7922404837557487,
      "learning_rate": 0.0001999729028346303,
      "loss": 12.7866,
      "step": 1366
    },
    {
      "epoch": 0.07443861332897701,
      "grad_norm": 0.8674908828154206,
      "learning_rate": 0.00019997269717364312,
      "loss": 12.632,
      "step": 1367
    },
    {
      "epoch": 0.07449306732556002,
      "grad_norm": 0.8979652665766763,
      "learning_rate": 0.0001999724907352515,
      "loss": 12.6652,
      "step": 1368
    },
    {
      "epoch": 0.07454752132214304,
      "grad_norm": 0.8543331532248469,
      "learning_rate": 0.000199972283519457,
      "loss": 12.7693,
      "step": 1369
    },
    {
      "epoch": 0.07460197531872605,
      "grad_norm": 0.8961519185426239,
      "learning_rate": 0.00019997207552626127,
      "loss": 12.7032,
      "step": 1370
    },
    {
      "epoch": 0.07465642931530907,
      "grad_norm": 0.8094646158907232,
      "learning_rate": 0.0001999718667556659,
      "loss": 12.7456,
      "step": 1371
    },
    {
      "epoch": 0.07471088331189207,
      "grad_norm": 1.108898840885292,
      "learning_rate": 0.00019997165720767255,
      "loss": 12.7149,
      "step": 1372
    },
    {
      "epoch": 0.07476533730847508,
      "grad_norm": 0.8660662469532117,
      "learning_rate": 0.00019997144688228282,
      "loss": 12.765,
      "step": 1373
    },
    {
      "epoch": 0.0748197913050581,
      "grad_norm": 0.8121742729568635,
      "learning_rate": 0.00019997123577949837,
      "loss": 12.7631,
      "step": 1374
    },
    {
      "epoch": 0.0748742453016411,
      "grad_norm": 0.9143360210967595,
      "learning_rate": 0.0001999710238993208,
      "loss": 12.7528,
      "step": 1375
    },
    {
      "epoch": 0.07492869929822411,
      "grad_norm": 0.8538749301886916,
      "learning_rate": 0.0001999708112417518,
      "loss": 12.7464,
      "step": 1376
    },
    {
      "epoch": 0.07498315329480713,
      "grad_norm": 0.8345594107229769,
      "learning_rate": 0.000199970597806793,
      "loss": 12.7875,
      "step": 1377
    },
    {
      "epoch": 0.07503760729139014,
      "grad_norm": 0.799878598540185,
      "learning_rate": 0.00019997038359444605,
      "loss": 12.6608,
      "step": 1378
    },
    {
      "epoch": 0.07509206128797316,
      "grad_norm": 0.7981199107212789,
      "learning_rate": 0.00019997016860471268,
      "loss": 12.7673,
      "step": 1379
    },
    {
      "epoch": 0.07514651528455617,
      "grad_norm": 0.8155962023244945,
      "learning_rate": 0.00019996995283759445,
      "loss": 12.6999,
      "step": 1380
    },
    {
      "epoch": 0.07520096928113917,
      "grad_norm": 0.8283187517960794,
      "learning_rate": 0.00019996973629309316,
      "loss": 12.7195,
      "step": 1381
    },
    {
      "epoch": 0.0752554232777222,
      "grad_norm": 0.8160714565733528,
      "learning_rate": 0.0001999695189712104,
      "loss": 12.5985,
      "step": 1382
    },
    {
      "epoch": 0.0753098772743052,
      "grad_norm": 0.7522090236557449,
      "learning_rate": 0.0001999693008719479,
      "loss": 12.6562,
      "step": 1383
    },
    {
      "epoch": 0.07536433127088821,
      "grad_norm": 0.8246447349504396,
      "learning_rate": 0.00019996908199530736,
      "loss": 12.7008,
      "step": 1384
    },
    {
      "epoch": 0.07541878526747123,
      "grad_norm": 0.7906471474347675,
      "learning_rate": 0.00019996886234129046,
      "loss": 12.6111,
      "step": 1385
    },
    {
      "epoch": 0.07547323926405423,
      "grad_norm": 0.8896552341485157,
      "learning_rate": 0.00019996864190989895,
      "loss": 12.6309,
      "step": 1386
    },
    {
      "epoch": 0.07552769326063725,
      "grad_norm": 0.8374448428066728,
      "learning_rate": 0.00019996842070113449,
      "loss": 12.8523,
      "step": 1387
    },
    {
      "epoch": 0.07558214725722026,
      "grad_norm": 0.8911290392829221,
      "learning_rate": 0.00019996819871499882,
      "loss": 12.7662,
      "step": 1388
    },
    {
      "epoch": 0.07563660125380327,
      "grad_norm": 0.8776686874771998,
      "learning_rate": 0.00019996797595149367,
      "loss": 12.6586,
      "step": 1389
    },
    {
      "epoch": 0.07569105525038629,
      "grad_norm": 0.7757694249013276,
      "learning_rate": 0.0001999677524106208,
      "loss": 12.7784,
      "step": 1390
    },
    {
      "epoch": 0.0757455092469693,
      "grad_norm": 0.8463963270041872,
      "learning_rate": 0.00019996752809238192,
      "loss": 12.7496,
      "step": 1391
    },
    {
      "epoch": 0.0757999632435523,
      "grad_norm": 0.9666247978951433,
      "learning_rate": 0.0001999673029967788,
      "loss": 12.6967,
      "step": 1392
    },
    {
      "epoch": 0.07585441724013532,
      "grad_norm": 0.911715433375072,
      "learning_rate": 0.00019996707712381312,
      "loss": 12.7605,
      "step": 1393
    },
    {
      "epoch": 0.07590887123671833,
      "grad_norm": 0.896211013201681,
      "learning_rate": 0.0001999668504734867,
      "loss": 12.7567,
      "step": 1394
    },
    {
      "epoch": 0.07596332523330133,
      "grad_norm": 0.7875639968187476,
      "learning_rate": 0.00019996662304580127,
      "loss": 12.7262,
      "step": 1395
    },
    {
      "epoch": 0.07601777922988436,
      "grad_norm": 0.8028651920579518,
      "learning_rate": 0.00019996639484075863,
      "loss": 12.6632,
      "step": 1396
    },
    {
      "epoch": 0.07607223322646736,
      "grad_norm": 0.8290344892272751,
      "learning_rate": 0.00019996616585836056,
      "loss": 12.7777,
      "step": 1397
    },
    {
      "epoch": 0.07612668722305038,
      "grad_norm": 0.808369979827406,
      "learning_rate": 0.0001999659360986088,
      "loss": 12.6464,
      "step": 1398
    },
    {
      "epoch": 0.07618114121963339,
      "grad_norm": 0.8593156464362707,
      "learning_rate": 0.00019996570556150516,
      "loss": 12.8318,
      "step": 1399
    },
    {
      "epoch": 0.0762355952162164,
      "grad_norm": 0.8476622102536071,
      "learning_rate": 0.0001999654742470514,
      "loss": 12.5885,
      "step": 1400
    },
    {
      "epoch": 0.07629004921279942,
      "grad_norm": 1.0197744142411682,
      "learning_rate": 0.0001999652421552494,
      "loss": 12.8219,
      "step": 1401
    },
    {
      "epoch": 0.07634450320938242,
      "grad_norm": 0.8796520538054026,
      "learning_rate": 0.0001999650092861009,
      "loss": 12.7627,
      "step": 1402
    },
    {
      "epoch": 0.07639895720596543,
      "grad_norm": 0.8767347301844162,
      "learning_rate": 0.0001999647756396077,
      "loss": 12.6461,
      "step": 1403
    },
    {
      "epoch": 0.07645341120254845,
      "grad_norm": 0.8968474287614663,
      "learning_rate": 0.00019996454121577167,
      "loss": 12.7267,
      "step": 1404
    },
    {
      "epoch": 0.07650786519913146,
      "grad_norm": 0.8363621234719137,
      "learning_rate": 0.00019996430601459454,
      "loss": 12.627,
      "step": 1405
    },
    {
      "epoch": 0.07656231919571448,
      "grad_norm": 0.8442098672535624,
      "learning_rate": 0.00019996407003607827,
      "loss": 12.7895,
      "step": 1406
    },
    {
      "epoch": 0.07661677319229748,
      "grad_norm": 0.895294258933168,
      "learning_rate": 0.0001999638332802246,
      "loss": 12.8037,
      "step": 1407
    },
    {
      "epoch": 0.07667122718888049,
      "grad_norm": 0.8150125735904853,
      "learning_rate": 0.0001999635957470354,
      "loss": 12.8318,
      "step": 1408
    },
    {
      "epoch": 0.07672568118546351,
      "grad_norm": 0.8107636017840302,
      "learning_rate": 0.00019996335743651254,
      "loss": 12.7149,
      "step": 1409
    },
    {
      "epoch": 0.07678013518204652,
      "grad_norm": 0.8164062918816573,
      "learning_rate": 0.00019996311834865783,
      "loss": 12.7795,
      "step": 1410
    },
    {
      "epoch": 0.07683458917862952,
      "grad_norm": 0.7278267164603288,
      "learning_rate": 0.00019996287848347315,
      "loss": 12.7514,
      "step": 1411
    },
    {
      "epoch": 0.07688904317521254,
      "grad_norm": 0.8448076956301601,
      "learning_rate": 0.00019996263784096034,
      "loss": 12.8286,
      "step": 1412
    },
    {
      "epoch": 0.07694349717179555,
      "grad_norm": 0.8097564652529006,
      "learning_rate": 0.00019996239642112133,
      "loss": 12.5989,
      "step": 1413
    },
    {
      "epoch": 0.07699795116837857,
      "grad_norm": 0.8107001306743004,
      "learning_rate": 0.00019996215422395794,
      "loss": 12.7783,
      "step": 1414
    },
    {
      "epoch": 0.07705240516496158,
      "grad_norm": 0.8086448265638508,
      "learning_rate": 0.00019996191124947208,
      "loss": 12.6879,
      "step": 1415
    },
    {
      "epoch": 0.07710685916154458,
      "grad_norm": 0.7665013421105268,
      "learning_rate": 0.00019996166749766564,
      "loss": 12.6297,
      "step": 1416
    },
    {
      "epoch": 0.0771613131581276,
      "grad_norm": 0.7943762132967175,
      "learning_rate": 0.00019996142296854047,
      "loss": 12.7468,
      "step": 1417
    },
    {
      "epoch": 0.07721576715471061,
      "grad_norm": 0.8981478243989491,
      "learning_rate": 0.00019996117766209857,
      "loss": 12.8986,
      "step": 1418
    },
    {
      "epoch": 0.07727022115129362,
      "grad_norm": 0.8143901857096051,
      "learning_rate": 0.00019996093157834176,
      "loss": 12.78,
      "step": 1419
    },
    {
      "epoch": 0.07732467514787664,
      "grad_norm": 0.8756714785828305,
      "learning_rate": 0.000199960684717272,
      "loss": 12.785,
      "step": 1420
    },
    {
      "epoch": 0.07737912914445964,
      "grad_norm": 0.8496567372331549,
      "learning_rate": 0.00019996043707889118,
      "loss": 12.7625,
      "step": 1421
    },
    {
      "epoch": 0.07743358314104266,
      "grad_norm": 0.8563016596814594,
      "learning_rate": 0.00019996018866320122,
      "loss": 12.818,
      "step": 1422
    },
    {
      "epoch": 0.07748803713762567,
      "grad_norm": 0.7546198210158892,
      "learning_rate": 0.0001999599394702041,
      "loss": 12.6427,
      "step": 1423
    },
    {
      "epoch": 0.07754249113420868,
      "grad_norm": 1.0243034781564102,
      "learning_rate": 0.00019995968949990171,
      "loss": 12.7954,
      "step": 1424
    },
    {
      "epoch": 0.0775969451307917,
      "grad_norm": 0.8101715088607566,
      "learning_rate": 0.000199959438752296,
      "loss": 12.7001,
      "step": 1425
    },
    {
      "epoch": 0.0776513991273747,
      "grad_norm": 0.8761233776508695,
      "learning_rate": 0.000199959187227389,
      "loss": 12.8675,
      "step": 1426
    },
    {
      "epoch": 0.07770585312395771,
      "grad_norm": 0.9545124675812537,
      "learning_rate": 0.00019995893492518252,
      "loss": 12.7311,
      "step": 1427
    },
    {
      "epoch": 0.07776030712054073,
      "grad_norm": 1.0653521779212813,
      "learning_rate": 0.00019995868184567863,
      "loss": 12.837,
      "step": 1428
    },
    {
      "epoch": 0.07781476111712374,
      "grad_norm": 0.7823863372608583,
      "learning_rate": 0.00019995842798887925,
      "loss": 12.646,
      "step": 1429
    },
    {
      "epoch": 0.07786921511370676,
      "grad_norm": 0.9468592289955438,
      "learning_rate": 0.0001999581733547864,
      "loss": 12.5773,
      "step": 1430
    },
    {
      "epoch": 0.07792366911028976,
      "grad_norm": 0.8859844301333014,
      "learning_rate": 0.000199957917943402,
      "loss": 12.6881,
      "step": 1431
    },
    {
      "epoch": 0.07797812310687277,
      "grad_norm": 0.7958834591179493,
      "learning_rate": 0.00019995766175472807,
      "loss": 12.7938,
      "step": 1432
    },
    {
      "epoch": 0.07803257710345579,
      "grad_norm": 0.9045806271425504,
      "learning_rate": 0.00019995740478876662,
      "loss": 12.8938,
      "step": 1433
    },
    {
      "epoch": 0.0780870311000388,
      "grad_norm": 0.9066374152311184,
      "learning_rate": 0.0001999571470455196,
      "loss": 12.8734,
      "step": 1434
    },
    {
      "epoch": 0.0781414850966218,
      "grad_norm": 1.105244040197418,
      "learning_rate": 0.00019995688852498907,
      "loss": 12.8091,
      "step": 1435
    },
    {
      "epoch": 0.07819593909320482,
      "grad_norm": 0.9163991440480068,
      "learning_rate": 0.000199956629227177,
      "loss": 12.8646,
      "step": 1436
    },
    {
      "epoch": 0.07825039308978783,
      "grad_norm": 0.9264755380848951,
      "learning_rate": 0.0001999563691520854,
      "loss": 12.6942,
      "step": 1437
    },
    {
      "epoch": 0.07830484708637085,
      "grad_norm": 0.8982000129129306,
      "learning_rate": 0.00019995610829971633,
      "loss": 12.7016,
      "step": 1438
    },
    {
      "epoch": 0.07835930108295386,
      "grad_norm": 0.8429653568543607,
      "learning_rate": 0.0001999558466700718,
      "loss": 12.7855,
      "step": 1439
    },
    {
      "epoch": 0.07841375507953686,
      "grad_norm": 1.02653089378891,
      "learning_rate": 0.00019995558426315384,
      "loss": 12.699,
      "step": 1440
    },
    {
      "epoch": 0.07846820907611989,
      "grad_norm": 0.8931709995895065,
      "learning_rate": 0.0001999553210789645,
      "loss": 12.8778,
      "step": 1441
    },
    {
      "epoch": 0.07852266307270289,
      "grad_norm": 1.077265824329578,
      "learning_rate": 0.00019995505711750583,
      "loss": 12.7602,
      "step": 1442
    },
    {
      "epoch": 0.0785771170692859,
      "grad_norm": 0.8471373686576927,
      "learning_rate": 0.00019995479237877985,
      "loss": 12.7904,
      "step": 1443
    },
    {
      "epoch": 0.07863157106586892,
      "grad_norm": 1.0498061251024426,
      "learning_rate": 0.00019995452686278866,
      "loss": 12.9102,
      "step": 1444
    },
    {
      "epoch": 0.07868602506245193,
      "grad_norm": 0.9447132721604138,
      "learning_rate": 0.00019995426056953428,
      "loss": 12.7107,
      "step": 1445
    },
    {
      "epoch": 0.07874047905903495,
      "grad_norm": 0.963176706304791,
      "learning_rate": 0.00019995399349901884,
      "loss": 12.9,
      "step": 1446
    },
    {
      "epoch": 0.07879493305561795,
      "grad_norm": 0.9043190032368775,
      "learning_rate": 0.00019995372565124436,
      "loss": 12.8733,
      "step": 1447
    },
    {
      "epoch": 0.07884938705220096,
      "grad_norm": 0.895109525929898,
      "learning_rate": 0.00019995345702621296,
      "loss": 12.7444,
      "step": 1448
    },
    {
      "epoch": 0.07890384104878398,
      "grad_norm": 1.0157615390283048,
      "learning_rate": 0.00019995318762392673,
      "loss": 12.7979,
      "step": 1449
    },
    {
      "epoch": 0.07895829504536699,
      "grad_norm": 0.9177562381958061,
      "learning_rate": 0.0001999529174443877,
      "loss": 12.7878,
      "step": 1450
    },
    {
      "epoch": 0.07901274904194999,
      "grad_norm": 0.8671856971809547,
      "learning_rate": 0.0001999526464875981,
      "loss": 12.8859,
      "step": 1451
    },
    {
      "epoch": 0.07906720303853301,
      "grad_norm": 0.7667322554377042,
      "learning_rate": 0.0001999523747535599,
      "loss": 12.7204,
      "step": 1452
    },
    {
      "epoch": 0.07912165703511602,
      "grad_norm": 0.8422632651938327,
      "learning_rate": 0.0001999521022422753,
      "loss": 12.7557,
      "step": 1453
    },
    {
      "epoch": 0.07917611103169904,
      "grad_norm": 0.9579270469578615,
      "learning_rate": 0.00019995182895374635,
      "loss": 12.6818,
      "step": 1454
    },
    {
      "epoch": 0.07923056502828205,
      "grad_norm": 0.8821151364436076,
      "learning_rate": 0.00019995155488797525,
      "loss": 12.6819,
      "step": 1455
    },
    {
      "epoch": 0.07928501902486505,
      "grad_norm": 0.8051216332043438,
      "learning_rate": 0.0001999512800449641,
      "loss": 12.7483,
      "step": 1456
    },
    {
      "epoch": 0.07933947302144807,
      "grad_norm": 0.8706203100578653,
      "learning_rate": 0.00019995100442471504,
      "loss": 12.7462,
      "step": 1457
    },
    {
      "epoch": 0.07939392701803108,
      "grad_norm": 0.8392511128264428,
      "learning_rate": 0.00019995072802723017,
      "loss": 12.7625,
      "step": 1458
    },
    {
      "epoch": 0.07944838101461409,
      "grad_norm": 0.7159408022254062,
      "learning_rate": 0.00019995045085251172,
      "loss": 12.6584,
      "step": 1459
    },
    {
      "epoch": 0.0795028350111971,
      "grad_norm": 0.8182234292300353,
      "learning_rate": 0.00019995017290056177,
      "loss": 12.7386,
      "step": 1460
    },
    {
      "epoch": 0.07955728900778011,
      "grad_norm": 0.8776394827159079,
      "learning_rate": 0.00019994989417138252,
      "loss": 12.84,
      "step": 1461
    },
    {
      "epoch": 0.07961174300436312,
      "grad_norm": 0.7774015631160571,
      "learning_rate": 0.00019994961466497614,
      "loss": 12.6897,
      "step": 1462
    },
    {
      "epoch": 0.07966619700094614,
      "grad_norm": 0.7153588530373372,
      "learning_rate": 0.0001999493343813448,
      "loss": 12.696,
      "step": 1463
    },
    {
      "epoch": 0.07972065099752915,
      "grad_norm": 0.8526001778470991,
      "learning_rate": 0.00019994905332049067,
      "loss": 12.6799,
      "step": 1464
    },
    {
      "epoch": 0.07977510499411217,
      "grad_norm": 0.8395065430273043,
      "learning_rate": 0.00019994877148241593,
      "loss": 12.7107,
      "step": 1465
    },
    {
      "epoch": 0.07982955899069517,
      "grad_norm": 0.7910115963722115,
      "learning_rate": 0.0001999484888671228,
      "loss": 12.7074,
      "step": 1466
    },
    {
      "epoch": 0.07988401298727818,
      "grad_norm": 0.8736867046566489,
      "learning_rate": 0.00019994820547461343,
      "loss": 12.7821,
      "step": 1467
    },
    {
      "epoch": 0.0799384669838612,
      "grad_norm": 0.7909169429327706,
      "learning_rate": 0.0001999479213048901,
      "loss": 12.7732,
      "step": 1468
    },
    {
      "epoch": 0.07999292098044421,
      "grad_norm": 0.712764262632445,
      "learning_rate": 0.00019994763635795493,
      "loss": 12.5218,
      "step": 1469
    },
    {
      "epoch": 0.08004737497702721,
      "grad_norm": 0.7489196228051317,
      "learning_rate": 0.00019994735063381017,
      "loss": 12.7551,
      "step": 1470
    },
    {
      "epoch": 0.08010182897361023,
      "grad_norm": 0.8125103235341982,
      "learning_rate": 0.0001999470641324581,
      "loss": 12.7572,
      "step": 1471
    },
    {
      "epoch": 0.08015628297019324,
      "grad_norm": 0.8649553566871733,
      "learning_rate": 0.00019994677685390087,
      "loss": 12.7905,
      "step": 1472
    },
    {
      "epoch": 0.08021073696677626,
      "grad_norm": 0.8974811724427579,
      "learning_rate": 0.00019994648879814074,
      "loss": 12.7806,
      "step": 1473
    },
    {
      "epoch": 0.08026519096335927,
      "grad_norm": 0.7657442868661224,
      "learning_rate": 0.00019994619996517997,
      "loss": 12.7352,
      "step": 1474
    },
    {
      "epoch": 0.08031964495994227,
      "grad_norm": 0.8360366489275451,
      "learning_rate": 0.00019994591035502076,
      "loss": 12.7474,
      "step": 1475
    },
    {
      "epoch": 0.0803740989565253,
      "grad_norm": 0.7867683365999412,
      "learning_rate": 0.0001999456199676654,
      "loss": 12.5452,
      "step": 1476
    },
    {
      "epoch": 0.0804285529531083,
      "grad_norm": 0.8163665133622418,
      "learning_rate": 0.00019994532880311617,
      "loss": 12.6865,
      "step": 1477
    },
    {
      "epoch": 0.08048300694969131,
      "grad_norm": 0.8767810719540375,
      "learning_rate": 0.00019994503686137524,
      "loss": 12.7188,
      "step": 1478
    },
    {
      "epoch": 0.08053746094627433,
      "grad_norm": 0.7954817734841462,
      "learning_rate": 0.000199944744142445,
      "loss": 12.7262,
      "step": 1479
    },
    {
      "epoch": 0.08059191494285733,
      "grad_norm": 0.7956972360870302,
      "learning_rate": 0.00019994445064632762,
      "loss": 12.81,
      "step": 1480
    },
    {
      "epoch": 0.08064636893944035,
      "grad_norm": 0.9637155793432769,
      "learning_rate": 0.00019994415637302547,
      "loss": 12.6966,
      "step": 1481
    },
    {
      "epoch": 0.08070082293602336,
      "grad_norm": 0.8875279185921682,
      "learning_rate": 0.0001999438613225408,
      "loss": 12.6457,
      "step": 1482
    },
    {
      "epoch": 0.08075527693260637,
      "grad_norm": 0.8889887759230662,
      "learning_rate": 0.00019994356549487587,
      "loss": 12.7212,
      "step": 1483
    },
    {
      "epoch": 0.08080973092918939,
      "grad_norm": 0.8712764841724392,
      "learning_rate": 0.00019994326889003302,
      "loss": 12.709,
      "step": 1484
    },
    {
      "epoch": 0.0808641849257724,
      "grad_norm": 0.8618471723921115,
      "learning_rate": 0.0001999429715080146,
      "loss": 12.6181,
      "step": 1485
    },
    {
      "epoch": 0.0809186389223554,
      "grad_norm": 0.9185534941177032,
      "learning_rate": 0.00019994267334882282,
      "loss": 12.7084,
      "step": 1486
    },
    {
      "epoch": 0.08097309291893842,
      "grad_norm": 0.8517384145727733,
      "learning_rate": 0.0001999423744124601,
      "loss": 12.7456,
      "step": 1487
    },
    {
      "epoch": 0.08102754691552143,
      "grad_norm": 0.7427396610603257,
      "learning_rate": 0.00019994207469892867,
      "loss": 12.6107,
      "step": 1488
    },
    {
      "epoch": 0.08108200091210445,
      "grad_norm": 0.80926578166257,
      "learning_rate": 0.00019994177420823092,
      "loss": 12.7398,
      "step": 1489
    },
    {
      "epoch": 0.08113645490868746,
      "grad_norm": 0.9206143365338525,
      "learning_rate": 0.00019994147294036916,
      "loss": 12.7456,
      "step": 1490
    },
    {
      "epoch": 0.08119090890527046,
      "grad_norm": 0.8103141575787716,
      "learning_rate": 0.00019994117089534576,
      "loss": 12.7612,
      "step": 1491
    },
    {
      "epoch": 0.08124536290185348,
      "grad_norm": 0.8039408920478894,
      "learning_rate": 0.00019994086807316306,
      "loss": 12.6768,
      "step": 1492
    },
    {
      "epoch": 0.08129981689843649,
      "grad_norm": 0.9364347767261306,
      "learning_rate": 0.0001999405644738234,
      "loss": 12.7028,
      "step": 1493
    },
    {
      "epoch": 0.0813542708950195,
      "grad_norm": 0.7758284076712086,
      "learning_rate": 0.00019994026009732916,
      "loss": 12.7189,
      "step": 1494
    },
    {
      "epoch": 0.08140872489160252,
      "grad_norm": 0.7628202869652245,
      "learning_rate": 0.00019993995494368272,
      "loss": 12.7589,
      "step": 1495
    },
    {
      "epoch": 0.08146317888818552,
      "grad_norm": 0.8970131980378315,
      "learning_rate": 0.00019993964901288637,
      "loss": 12.7318,
      "step": 1496
    },
    {
      "epoch": 0.08151763288476854,
      "grad_norm": 0.9069018572400294,
      "learning_rate": 0.0001999393423049426,
      "loss": 12.8936,
      "step": 1497
    },
    {
      "epoch": 0.08157208688135155,
      "grad_norm": 0.876220223123507,
      "learning_rate": 0.00019993903481985373,
      "loss": 12.812,
      "step": 1498
    },
    {
      "epoch": 0.08162654087793456,
      "grad_norm": 0.8535583445255587,
      "learning_rate": 0.00019993872655762215,
      "loss": 12.698,
      "step": 1499
    },
    {
      "epoch": 0.08168099487451758,
      "grad_norm": 0.8631114608236895,
      "learning_rate": 0.00019993841751825032,
      "loss": 12.8378,
      "step": 1500
    },
    {
      "epoch": 0.08173544887110058,
      "grad_norm": 0.9638776193483425,
      "learning_rate": 0.00019993810770174055,
      "loss": 12.7563,
      "step": 1501
    },
    {
      "epoch": 0.08178990286768359,
      "grad_norm": 1.1874397730440935,
      "learning_rate": 0.00019993779710809532,
      "loss": 12.4875,
      "step": 1502
    },
    {
      "epoch": 0.08184435686426661,
      "grad_norm": 0.7936581239575987,
      "learning_rate": 0.00019993748573731698,
      "loss": 12.714,
      "step": 1503
    },
    {
      "epoch": 0.08189881086084962,
      "grad_norm": 0.7976386685488491,
      "learning_rate": 0.00019993717358940803,
      "loss": 12.7046,
      "step": 1504
    },
    {
      "epoch": 0.08195326485743264,
      "grad_norm": 0.9118200058009364,
      "learning_rate": 0.00019993686066437086,
      "loss": 12.7126,
      "step": 1505
    },
    {
      "epoch": 0.08200771885401564,
      "grad_norm": 0.8982301897740976,
      "learning_rate": 0.00019993654696220787,
      "loss": 12.7474,
      "step": 1506
    },
    {
      "epoch": 0.08206217285059865,
      "grad_norm": 0.8340940733070505,
      "learning_rate": 0.00019993623248292156,
      "loss": 12.6143,
      "step": 1507
    },
    {
      "epoch": 0.08211662684718167,
      "grad_norm": 0.7856628911786119,
      "learning_rate": 0.00019993591722651432,
      "loss": 12.6957,
      "step": 1508
    },
    {
      "epoch": 0.08217108084376468,
      "grad_norm": 0.8563602762901024,
      "learning_rate": 0.00019993560119298866,
      "loss": 12.7298,
      "step": 1509
    },
    {
      "epoch": 0.08222553484034768,
      "grad_norm": 0.9064468196397422,
      "learning_rate": 0.00019993528438234698,
      "loss": 12.7308,
      "step": 1510
    },
    {
      "epoch": 0.0822799888369307,
      "grad_norm": 0.7929295665041554,
      "learning_rate": 0.0001999349667945918,
      "loss": 12.6453,
      "step": 1511
    },
    {
      "epoch": 0.08233444283351371,
      "grad_norm": 0.7409195103316495,
      "learning_rate": 0.00019993464842972552,
      "loss": 12.7021,
      "step": 1512
    },
    {
      "epoch": 0.08238889683009673,
      "grad_norm": 1.0174626901426223,
      "learning_rate": 0.00019993432928775069,
      "loss": 12.654,
      "step": 1513
    },
    {
      "epoch": 0.08244335082667974,
      "grad_norm": 0.8385420276571517,
      "learning_rate": 0.00019993400936866974,
      "loss": 12.7606,
      "step": 1514
    },
    {
      "epoch": 0.08249780482326274,
      "grad_norm": 0.7773434141608896,
      "learning_rate": 0.00019993368867248518,
      "loss": 12.6926,
      "step": 1515
    },
    {
      "epoch": 0.08255225881984576,
      "grad_norm": 0.948935487487568,
      "learning_rate": 0.0001999333671991995,
      "loss": 12.7262,
      "step": 1516
    },
    {
      "epoch": 0.08260671281642877,
      "grad_norm": 0.7845942917925295,
      "learning_rate": 0.0001999330449488152,
      "loss": 12.7695,
      "step": 1517
    },
    {
      "epoch": 0.08266116681301178,
      "grad_norm": 0.796696184682058,
      "learning_rate": 0.00019993272192133477,
      "loss": 12.7048,
      "step": 1518
    },
    {
      "epoch": 0.0827156208095948,
      "grad_norm": 0.8115038277668638,
      "learning_rate": 0.00019993239811676075,
      "loss": 12.7077,
      "step": 1519
    },
    {
      "epoch": 0.0827700748061778,
      "grad_norm": 0.8598970799174874,
      "learning_rate": 0.00019993207353509562,
      "loss": 12.7663,
      "step": 1520
    },
    {
      "epoch": 0.08282452880276082,
      "grad_norm": 0.8329175862926992,
      "learning_rate": 0.00019993174817634196,
      "loss": 12.6002,
      "step": 1521
    },
    {
      "epoch": 0.08287898279934383,
      "grad_norm": 0.9493548901124595,
      "learning_rate": 0.00019993142204050224,
      "loss": 12.8797,
      "step": 1522
    },
    {
      "epoch": 0.08293343679592684,
      "grad_norm": 0.7844916289661439,
      "learning_rate": 0.00019993109512757903,
      "loss": 12.7075,
      "step": 1523
    },
    {
      "epoch": 0.08298789079250986,
      "grad_norm": 0.7485570224627376,
      "learning_rate": 0.00019993076743757485,
      "loss": 12.7142,
      "step": 1524
    },
    {
      "epoch": 0.08304234478909286,
      "grad_norm": 0.8230611251453327,
      "learning_rate": 0.0001999304389704923,
      "loss": 12.6775,
      "step": 1525
    },
    {
      "epoch": 0.08309679878567587,
      "grad_norm": 0.7802141642317684,
      "learning_rate": 0.00019993010972633389,
      "loss": 12.7862,
      "step": 1526
    },
    {
      "epoch": 0.08315125278225889,
      "grad_norm": 0.804109574654945,
      "learning_rate": 0.0001999297797051022,
      "loss": 12.7384,
      "step": 1527
    },
    {
      "epoch": 0.0832057067788419,
      "grad_norm": 0.7549845829476725,
      "learning_rate": 0.00019992944890679976,
      "loss": 12.6077,
      "step": 1528
    },
    {
      "epoch": 0.0832601607754249,
      "grad_norm": 0.8158889763260669,
      "learning_rate": 0.00019992911733142916,
      "loss": 12.7293,
      "step": 1529
    },
    {
      "epoch": 0.08331461477200792,
      "grad_norm": 0.8322818323951467,
      "learning_rate": 0.00019992878497899298,
      "loss": 12.7941,
      "step": 1530
    },
    {
      "epoch": 0.08336906876859093,
      "grad_norm": 0.7974712778159135,
      "learning_rate": 0.00019992845184949384,
      "loss": 12.6936,
      "step": 1531
    },
    {
      "epoch": 0.08342352276517395,
      "grad_norm": 0.789866577373669,
      "learning_rate": 0.0001999281179429343,
      "loss": 12.684,
      "step": 1532
    },
    {
      "epoch": 0.08347797676175696,
      "grad_norm": 0.8534707399357608,
      "learning_rate": 0.00019992778325931694,
      "loss": 12.6751,
      "step": 1533
    },
    {
      "epoch": 0.08353243075833997,
      "grad_norm": 0.9367458502912297,
      "learning_rate": 0.00019992744779864438,
      "loss": 12.7332,
      "step": 1534
    },
    {
      "epoch": 0.08358688475492299,
      "grad_norm": 0.9117832539518956,
      "learning_rate": 0.00019992711156091925,
      "loss": 12.6615,
      "step": 1535
    },
    {
      "epoch": 0.08364133875150599,
      "grad_norm": 0.8034947883057498,
      "learning_rate": 0.00019992677454614414,
      "loss": 12.6091,
      "step": 1536
    },
    {
      "epoch": 0.083695792748089,
      "grad_norm": 0.8610815668935233,
      "learning_rate": 0.00019992643675432163,
      "loss": 12.6785,
      "step": 1537
    },
    {
      "epoch": 0.08375024674467202,
      "grad_norm": 0.7634725773013094,
      "learning_rate": 0.00019992609818545443,
      "loss": 12.6252,
      "step": 1538
    },
    {
      "epoch": 0.08380470074125503,
      "grad_norm": 0.773853435934736,
      "learning_rate": 0.00019992575883954512,
      "loss": 12.6299,
      "step": 1539
    },
    {
      "epoch": 0.08385915473783805,
      "grad_norm": 0.83449596965377,
      "learning_rate": 0.00019992541871659636,
      "loss": 12.7319,
      "step": 1540
    },
    {
      "epoch": 0.08391360873442105,
      "grad_norm": 0.7817968068976223,
      "learning_rate": 0.00019992507781661076,
      "loss": 12.6707,
      "step": 1541
    },
    {
      "epoch": 0.08396806273100406,
      "grad_norm": 0.8437153661982023,
      "learning_rate": 0.00019992473613959102,
      "loss": 12.7417,
      "step": 1542
    },
    {
      "epoch": 0.08402251672758708,
      "grad_norm": 0.7850470107731357,
      "learning_rate": 0.00019992439368553977,
      "loss": 12.7926,
      "step": 1543
    },
    {
      "epoch": 0.08407697072417009,
      "grad_norm": 0.8146054061971123,
      "learning_rate": 0.0001999240504544597,
      "loss": 12.8017,
      "step": 1544
    },
    {
      "epoch": 0.08413142472075309,
      "grad_norm": 0.871613332041884,
      "learning_rate": 0.00019992370644635342,
      "loss": 12.8044,
      "step": 1545
    },
    {
      "epoch": 0.08418587871733611,
      "grad_norm": 0.8555636063343115,
      "learning_rate": 0.00019992336166122366,
      "loss": 12.7711,
      "step": 1546
    },
    {
      "epoch": 0.08424033271391912,
      "grad_norm": 0.9492028973039249,
      "learning_rate": 0.0001999230160990731,
      "loss": 12.8954,
      "step": 1547
    },
    {
      "epoch": 0.08429478671050214,
      "grad_norm": 0.8221100770793226,
      "learning_rate": 0.00019992266975990436,
      "loss": 12.8115,
      "step": 1548
    },
    {
      "epoch": 0.08434924070708515,
      "grad_norm": 0.8717001137503292,
      "learning_rate": 0.00019992232264372023,
      "loss": 12.7496,
      "step": 1549
    },
    {
      "epoch": 0.08440369470366815,
      "grad_norm": 0.8772885860945219,
      "learning_rate": 0.00019992197475052334,
      "loss": 12.6613,
      "step": 1550
    },
    {
      "epoch": 0.08445814870025117,
      "grad_norm": 0.7647801603416208,
      "learning_rate": 0.00019992162608031643,
      "loss": 12.605,
      "step": 1551
    },
    {
      "epoch": 0.08451260269683418,
      "grad_norm": 0.8249989839894225,
      "learning_rate": 0.00019992127663310218,
      "loss": 12.713,
      "step": 1552
    },
    {
      "epoch": 0.08456705669341719,
      "grad_norm": 0.8004163246461077,
      "learning_rate": 0.00019992092640888336,
      "loss": 12.8538,
      "step": 1553
    },
    {
      "epoch": 0.0846215106900002,
      "grad_norm": 0.7652912812630724,
      "learning_rate": 0.00019992057540766262,
      "loss": 12.6518,
      "step": 1554
    },
    {
      "epoch": 0.08467596468658321,
      "grad_norm": 0.7636164105422508,
      "learning_rate": 0.00019992022362944276,
      "loss": 12.7801,
      "step": 1555
    },
    {
      "epoch": 0.08473041868316623,
      "grad_norm": 0.7770954475005418,
      "learning_rate": 0.00019991987107422646,
      "loss": 12.7198,
      "step": 1556
    },
    {
      "epoch": 0.08478487267974924,
      "grad_norm": 0.7876480663521178,
      "learning_rate": 0.0001999195177420165,
      "loss": 12.7493,
      "step": 1557
    },
    {
      "epoch": 0.08483932667633225,
      "grad_norm": 0.7950431925010902,
      "learning_rate": 0.0001999191636328156,
      "loss": 12.7189,
      "step": 1558
    },
    {
      "epoch": 0.08489378067291527,
      "grad_norm": 0.8317608535674442,
      "learning_rate": 0.00019991880874662655,
      "loss": 12.8068,
      "step": 1559
    },
    {
      "epoch": 0.08494823466949827,
      "grad_norm": 0.9077379398287838,
      "learning_rate": 0.00019991845308345204,
      "loss": 12.9445,
      "step": 1560
    },
    {
      "epoch": 0.08500268866608128,
      "grad_norm": 1.2586595675046723,
      "learning_rate": 0.00019991809664329492,
      "loss": 12.7941,
      "step": 1561
    },
    {
      "epoch": 0.0850571426626643,
      "grad_norm": 0.934969280480126,
      "learning_rate": 0.00019991773942615795,
      "loss": 12.7453,
      "step": 1562
    },
    {
      "epoch": 0.08511159665924731,
      "grad_norm": 0.8689759469261831,
      "learning_rate": 0.00019991738143204383,
      "loss": 12.716,
      "step": 1563
    },
    {
      "epoch": 0.08516605065583033,
      "grad_norm": 0.9048636278651105,
      "learning_rate": 0.00019991702266095542,
      "loss": 12.6813,
      "step": 1564
    },
    {
      "epoch": 0.08522050465241333,
      "grad_norm": 0.9309895881664149,
      "learning_rate": 0.0001999166631128955,
      "loss": 12.8226,
      "step": 1565
    },
    {
      "epoch": 0.08527495864899634,
      "grad_norm": 1.085437499269038,
      "learning_rate": 0.00019991630278786682,
      "loss": 12.7647,
      "step": 1566
    },
    {
      "epoch": 0.08532941264557936,
      "grad_norm": 0.9087661866473474,
      "learning_rate": 0.00019991594168587224,
      "loss": 12.8076,
      "step": 1567
    },
    {
      "epoch": 0.08538386664216237,
      "grad_norm": 0.9896982322788401,
      "learning_rate": 0.00019991557980691453,
      "loss": 12.7972,
      "step": 1568
    },
    {
      "epoch": 0.08543832063874537,
      "grad_norm": 0.832692895973604,
      "learning_rate": 0.0001999152171509965,
      "loss": 12.7677,
      "step": 1569
    },
    {
      "epoch": 0.0854927746353284,
      "grad_norm": 0.9248339271879307,
      "learning_rate": 0.00019991485371812103,
      "loss": 12.789,
      "step": 1570
    },
    {
      "epoch": 0.0855472286319114,
      "grad_norm": 0.767994607022334,
      "learning_rate": 0.00019991448950829085,
      "loss": 12.6611,
      "step": 1571
    },
    {
      "epoch": 0.08560168262849442,
      "grad_norm": 0.8421196864236498,
      "learning_rate": 0.0001999141245215089,
      "loss": 12.7529,
      "step": 1572
    },
    {
      "epoch": 0.08565613662507743,
      "grad_norm": 0.9014060216445708,
      "learning_rate": 0.0001999137587577779,
      "loss": 12.7877,
      "step": 1573
    },
    {
      "epoch": 0.08571059062166043,
      "grad_norm": 0.866376171224473,
      "learning_rate": 0.00019991339221710078,
      "loss": 12.7103,
      "step": 1574
    },
    {
      "epoch": 0.08576504461824346,
      "grad_norm": 0.721762025104229,
      "learning_rate": 0.0001999130248994804,
      "loss": 12.7141,
      "step": 1575
    },
    {
      "epoch": 0.08581949861482646,
      "grad_norm": 0.7766649078647998,
      "learning_rate": 0.00019991265680491954,
      "loss": 12.6779,
      "step": 1576
    },
    {
      "epoch": 0.08587395261140947,
      "grad_norm": 0.809840598959927,
      "learning_rate": 0.00019991228793342112,
      "loss": 12.8861,
      "step": 1577
    },
    {
      "epoch": 0.08592840660799249,
      "grad_norm": 0.8275739487902802,
      "learning_rate": 0.000199911918284988,
      "loss": 12.5711,
      "step": 1578
    },
    {
      "epoch": 0.0859828606045755,
      "grad_norm": 0.8132028689679895,
      "learning_rate": 0.00019991154785962306,
      "loss": 12.8249,
      "step": 1579
    },
    {
      "epoch": 0.08603731460115852,
      "grad_norm": 1.3463942717961974,
      "learning_rate": 0.00019991117665732914,
      "loss": 12.8725,
      "step": 1580
    },
    {
      "epoch": 0.08609176859774152,
      "grad_norm": 0.7744003920615815,
      "learning_rate": 0.00019991080467810917,
      "loss": 12.733,
      "step": 1581
    },
    {
      "epoch": 0.08614622259432453,
      "grad_norm": 0.8797671865198825,
      "learning_rate": 0.00019991043192196602,
      "loss": 12.7362,
      "step": 1582
    },
    {
      "epoch": 0.08620067659090755,
      "grad_norm": 0.8685316217744672,
      "learning_rate": 0.0001999100583889026,
      "loss": 12.8417,
      "step": 1583
    },
    {
      "epoch": 0.08625513058749056,
      "grad_norm": 0.8634938681153937,
      "learning_rate": 0.0001999096840789218,
      "loss": 12.8137,
      "step": 1584
    },
    {
      "epoch": 0.08630958458407356,
      "grad_norm": 0.9845889721614345,
      "learning_rate": 0.00019990930899202656,
      "loss": 12.8088,
      "step": 1585
    },
    {
      "epoch": 0.08636403858065658,
      "grad_norm": 0.9203403556282245,
      "learning_rate": 0.00019990893312821976,
      "loss": 12.8132,
      "step": 1586
    },
    {
      "epoch": 0.08641849257723959,
      "grad_norm": 0.8059258371430177,
      "learning_rate": 0.00019990855648750438,
      "loss": 12.6877,
      "step": 1587
    },
    {
      "epoch": 0.08647294657382261,
      "grad_norm": 0.9359421414283161,
      "learning_rate": 0.00019990817906988327,
      "loss": 12.7852,
      "step": 1588
    },
    {
      "epoch": 0.08652740057040562,
      "grad_norm": 0.828140212206367,
      "learning_rate": 0.00019990780087535942,
      "loss": 12.7481,
      "step": 1589
    },
    {
      "epoch": 0.08658185456698862,
      "grad_norm": 0.8451668669812522,
      "learning_rate": 0.00019990742190393573,
      "loss": 12.7679,
      "step": 1590
    },
    {
      "epoch": 0.08663630856357164,
      "grad_norm": 0.9171546488032268,
      "learning_rate": 0.0001999070421556152,
      "loss": 12.7237,
      "step": 1591
    },
    {
      "epoch": 0.08669076256015465,
      "grad_norm": 0.7928322850405436,
      "learning_rate": 0.00019990666163040077,
      "loss": 12.831,
      "step": 1592
    },
    {
      "epoch": 0.08674521655673766,
      "grad_norm": 0.7971408312985833,
      "learning_rate": 0.00019990628032829537,
      "loss": 12.6751,
      "step": 1593
    },
    {
      "epoch": 0.08679967055332068,
      "grad_norm": 0.7932785961720318,
      "learning_rate": 0.00019990589824930198,
      "loss": 12.732,
      "step": 1594
    },
    {
      "epoch": 0.08685412454990368,
      "grad_norm": 0.8749998597118366,
      "learning_rate": 0.00019990551539342355,
      "loss": 12.7074,
      "step": 1595
    },
    {
      "epoch": 0.08690857854648669,
      "grad_norm": 0.81543285620202,
      "learning_rate": 0.0001999051317606631,
      "loss": 12.7376,
      "step": 1596
    },
    {
      "epoch": 0.08696303254306971,
      "grad_norm": 0.9261825740964793,
      "learning_rate": 0.0001999047473510236,
      "loss": 12.8214,
      "step": 1597
    },
    {
      "epoch": 0.08701748653965272,
      "grad_norm": 0.8622490108458055,
      "learning_rate": 0.00019990436216450803,
      "loss": 12.7227,
      "step": 1598
    },
    {
      "epoch": 0.08707194053623574,
      "grad_norm": 0.8550542644779822,
      "learning_rate": 0.00019990397620111937,
      "loss": 12.7084,
      "step": 1599
    },
    {
      "epoch": 0.08712639453281874,
      "grad_norm": 0.8381618824955894,
      "learning_rate": 0.00019990358946086063,
      "loss": 12.6713,
      "step": 1600
    },
    {
      "epoch": 0.08718084852940175,
      "grad_norm": 0.9796927090666376,
      "learning_rate": 0.00019990320194373485,
      "loss": 12.7857,
      "step": 1601
    },
    {
      "epoch": 0.08723530252598477,
      "grad_norm": 0.9539240134811797,
      "learning_rate": 0.000199902813649745,
      "loss": 12.7795,
      "step": 1602
    },
    {
      "epoch": 0.08728975652256778,
      "grad_norm": 0.8105784092854924,
      "learning_rate": 0.0001999024245788941,
      "loss": 12.6445,
      "step": 1603
    },
    {
      "epoch": 0.08734421051915078,
      "grad_norm": 0.7883626144517288,
      "learning_rate": 0.00019990203473118522,
      "loss": 12.6751,
      "step": 1604
    },
    {
      "epoch": 0.0873986645157338,
      "grad_norm": 0.8767347617569384,
      "learning_rate": 0.00019990164410662136,
      "loss": 12.6106,
      "step": 1605
    },
    {
      "epoch": 0.08745311851231681,
      "grad_norm": 0.9084313776833122,
      "learning_rate": 0.00019990125270520558,
      "loss": 12.7858,
      "step": 1606
    },
    {
      "epoch": 0.08750757250889983,
      "grad_norm": 0.8747776366189766,
      "learning_rate": 0.0001999008605269409,
      "loss": 12.9065,
      "step": 1607
    },
    {
      "epoch": 0.08756202650548284,
      "grad_norm": 0.8489601291243307,
      "learning_rate": 0.00019990046757183033,
      "loss": 12.8027,
      "step": 1608
    },
    {
      "epoch": 0.08761648050206584,
      "grad_norm": 0.9162027477048691,
      "learning_rate": 0.00019990007383987698,
      "loss": 12.7236,
      "step": 1609
    },
    {
      "epoch": 0.08767093449864886,
      "grad_norm": 0.8386437850394985,
      "learning_rate": 0.00019989967933108394,
      "loss": 12.7247,
      "step": 1610
    },
    {
      "epoch": 0.08772538849523187,
      "grad_norm": 0.8829733589907728,
      "learning_rate": 0.00019989928404545425,
      "loss": 12.7478,
      "step": 1611
    },
    {
      "epoch": 0.08777984249181488,
      "grad_norm": 0.8150772135220602,
      "learning_rate": 0.00019989888798299093,
      "loss": 12.6919,
      "step": 1612
    },
    {
      "epoch": 0.0878342964883979,
      "grad_norm": 0.7878375599715177,
      "learning_rate": 0.0001998984911436971,
      "loss": 12.6992,
      "step": 1613
    },
    {
      "epoch": 0.0878887504849809,
      "grad_norm": 0.8416899312712404,
      "learning_rate": 0.0001998980935275759,
      "loss": 12.7049,
      "step": 1614
    },
    {
      "epoch": 0.08794320448156392,
      "grad_norm": 0.9133401631169028,
      "learning_rate": 0.00019989769513463035,
      "loss": 12.8479,
      "step": 1615
    },
    {
      "epoch": 0.08799765847814693,
      "grad_norm": 0.8952845893190265,
      "learning_rate": 0.00019989729596486355,
      "loss": 12.7254,
      "step": 1616
    },
    {
      "epoch": 0.08805211247472994,
      "grad_norm": 0.8407850339653142,
      "learning_rate": 0.00019989689601827864,
      "loss": 12.7772,
      "step": 1617
    },
    {
      "epoch": 0.08810656647131296,
      "grad_norm": 0.8027908080124432,
      "learning_rate": 0.0001998964952948787,
      "loss": 12.793,
      "step": 1618
    },
    {
      "epoch": 0.08816102046789596,
      "grad_norm": 0.8720305737368109,
      "learning_rate": 0.00019989609379466688,
      "loss": 12.8101,
      "step": 1619
    },
    {
      "epoch": 0.08821547446447897,
      "grad_norm": 0.8423254972283053,
      "learning_rate": 0.0001998956915176463,
      "loss": 12.7649,
      "step": 1620
    },
    {
      "epoch": 0.08826992846106199,
      "grad_norm": 0.8906736340443412,
      "learning_rate": 0.00019989528846382,
      "loss": 12.6544,
      "step": 1621
    },
    {
      "epoch": 0.088324382457645,
      "grad_norm": 0.8685979598804452,
      "learning_rate": 0.00019989488463319127,
      "loss": 12.7049,
      "step": 1622
    },
    {
      "epoch": 0.08837883645422802,
      "grad_norm": 0.9201523573659224,
      "learning_rate": 0.0001998944800257631,
      "loss": 12.7724,
      "step": 1623
    },
    {
      "epoch": 0.08843329045081103,
      "grad_norm": 0.7950617489837553,
      "learning_rate": 0.00019989407464153874,
      "loss": 12.6128,
      "step": 1624
    },
    {
      "epoch": 0.08848774444739403,
      "grad_norm": 0.8431903484712208,
      "learning_rate": 0.00019989366848052127,
      "loss": 12.6943,
      "step": 1625
    },
    {
      "epoch": 0.08854219844397705,
      "grad_norm": 0.792854903121601,
      "learning_rate": 0.0001998932615427139,
      "loss": 12.7372,
      "step": 1626
    },
    {
      "epoch": 0.08859665244056006,
      "grad_norm": 0.919581588175274,
      "learning_rate": 0.00019989285382811977,
      "loss": 12.8211,
      "step": 1627
    },
    {
      "epoch": 0.08865110643714307,
      "grad_norm": 1.0225144687152612,
      "learning_rate": 0.00019989244533674208,
      "loss": 12.9169,
      "step": 1628
    },
    {
      "epoch": 0.08870556043372609,
      "grad_norm": 0.9628652507615445,
      "learning_rate": 0.00019989203606858395,
      "loss": 12.7378,
      "step": 1629
    },
    {
      "epoch": 0.08876001443030909,
      "grad_norm": 0.9482872431610143,
      "learning_rate": 0.0001998916260236486,
      "loss": 12.7291,
      "step": 1630
    },
    {
      "epoch": 0.08881446842689211,
      "grad_norm": 0.8991062958487321,
      "learning_rate": 0.00019989121520193925,
      "loss": 12.817,
      "step": 1631
    },
    {
      "epoch": 0.08886892242347512,
      "grad_norm": 0.9023899125767259,
      "learning_rate": 0.00019989080360345902,
      "loss": 12.7704,
      "step": 1632
    },
    {
      "epoch": 0.08892337642005813,
      "grad_norm": 0.7729477313833728,
      "learning_rate": 0.00019989039122821116,
      "loss": 12.6245,
      "step": 1633
    },
    {
      "epoch": 0.08897783041664115,
      "grad_norm": 1.0417855932195912,
      "learning_rate": 0.00019988997807619886,
      "loss": 12.7979,
      "step": 1634
    },
    {
      "epoch": 0.08903228441322415,
      "grad_norm": 0.7937034858967675,
      "learning_rate": 0.00019988956414742536,
      "loss": 12.7046,
      "step": 1635
    },
    {
      "epoch": 0.08908673840980716,
      "grad_norm": 0.8001933910250998,
      "learning_rate": 0.00019988914944189386,
      "loss": 12.7312,
      "step": 1636
    },
    {
      "epoch": 0.08914119240639018,
      "grad_norm": 0.8558583761877688,
      "learning_rate": 0.00019988873395960756,
      "loss": 12.7965,
      "step": 1637
    },
    {
      "epoch": 0.08919564640297319,
      "grad_norm": 0.7599770895603871,
      "learning_rate": 0.00019988831770056972,
      "loss": 12.7082,
      "step": 1638
    },
    {
      "epoch": 0.0892501003995562,
      "grad_norm": 0.7960868786708084,
      "learning_rate": 0.00019988790066478358,
      "loss": 12.6888,
      "step": 1639
    },
    {
      "epoch": 0.08930455439613921,
      "grad_norm": 0.7261475247206844,
      "learning_rate": 0.00019988748285225237,
      "loss": 12.5913,
      "step": 1640
    },
    {
      "epoch": 0.08935900839272222,
      "grad_norm": 0.7993605836884233,
      "learning_rate": 0.00019988706426297932,
      "loss": 12.717,
      "step": 1641
    },
    {
      "epoch": 0.08941346238930524,
      "grad_norm": 0.748685997004754,
      "learning_rate": 0.00019988664489696773,
      "loss": 12.794,
      "step": 1642
    },
    {
      "epoch": 0.08946791638588825,
      "grad_norm": 0.8745510045470801,
      "learning_rate": 0.00019988622475422085,
      "loss": 12.7614,
      "step": 1643
    },
    {
      "epoch": 0.08952237038247125,
      "grad_norm": 0.8887894206010839,
      "learning_rate": 0.00019988580383474192,
      "loss": 12.8313,
      "step": 1644
    },
    {
      "epoch": 0.08957682437905427,
      "grad_norm": 0.7790629439842283,
      "learning_rate": 0.0001998853821385342,
      "loss": 12.6754,
      "step": 1645
    },
    {
      "epoch": 0.08963127837563728,
      "grad_norm": 0.7548331983129816,
      "learning_rate": 0.00019988495966560103,
      "loss": 12.6716,
      "step": 1646
    },
    {
      "epoch": 0.0896857323722203,
      "grad_norm": 0.8683759239019688,
      "learning_rate": 0.00019988453641594568,
      "loss": 12.7828,
      "step": 1647
    },
    {
      "epoch": 0.0897401863688033,
      "grad_norm": 0.8333977638249082,
      "learning_rate": 0.0001998841123895714,
      "loss": 12.7488,
      "step": 1648
    },
    {
      "epoch": 0.08979464036538631,
      "grad_norm": 0.7821755083844757,
      "learning_rate": 0.0001998836875864815,
      "loss": 12.803,
      "step": 1649
    },
    {
      "epoch": 0.08984909436196933,
      "grad_norm": 0.8495799064371253,
      "learning_rate": 0.00019988326200667933,
      "loss": 12.614,
      "step": 1650
    },
    {
      "epoch": 0.08990354835855234,
      "grad_norm": 0.8922029842895346,
      "learning_rate": 0.00019988283565016812,
      "loss": 12.7251,
      "step": 1651
    },
    {
      "epoch": 0.08995800235513535,
      "grad_norm": 0.8249423440125538,
      "learning_rate": 0.00019988240851695125,
      "loss": 12.7249,
      "step": 1652
    },
    {
      "epoch": 0.09001245635171837,
      "grad_norm": 0.7715976194103749,
      "learning_rate": 0.00019988198060703205,
      "loss": 12.7746,
      "step": 1653
    },
    {
      "epoch": 0.09006691034830137,
      "grad_norm": 0.8493084075161741,
      "learning_rate": 0.00019988155192041378,
      "loss": 12.6983,
      "step": 1654
    },
    {
      "epoch": 0.0901213643448844,
      "grad_norm": 0.8833809799334171,
      "learning_rate": 0.00019988112245709983,
      "loss": 12.6754,
      "step": 1655
    },
    {
      "epoch": 0.0901758183414674,
      "grad_norm": 0.813960208688227,
      "learning_rate": 0.00019988069221709348,
      "loss": 12.7352,
      "step": 1656
    },
    {
      "epoch": 0.09023027233805041,
      "grad_norm": 0.7960213057660804,
      "learning_rate": 0.0001998802612003982,
      "loss": 12.8401,
      "step": 1657
    },
    {
      "epoch": 0.09028472633463343,
      "grad_norm": 0.7846256038589693,
      "learning_rate": 0.0001998798294070172,
      "loss": 12.8505,
      "step": 1658
    },
    {
      "epoch": 0.09033918033121643,
      "grad_norm": 0.911526256365673,
      "learning_rate": 0.0001998793968369539,
      "loss": 12.8778,
      "step": 1659
    },
    {
      "epoch": 0.09039363432779944,
      "grad_norm": 0.8647590123578177,
      "learning_rate": 0.00019987896349021167,
      "loss": 12.6736,
      "step": 1660
    },
    {
      "epoch": 0.09044808832438246,
      "grad_norm": 0.9390747596810028,
      "learning_rate": 0.00019987852936679388,
      "loss": 12.7272,
      "step": 1661
    },
    {
      "epoch": 0.09050254232096547,
      "grad_norm": 0.9456846322227296,
      "learning_rate": 0.00019987809446670387,
      "loss": 12.7014,
      "step": 1662
    },
    {
      "epoch": 0.09055699631754847,
      "grad_norm": 0.9041210520828457,
      "learning_rate": 0.00019987765878994507,
      "loss": 12.6473,
      "step": 1663
    },
    {
      "epoch": 0.0906114503141315,
      "grad_norm": 0.7426745800172005,
      "learning_rate": 0.00019987722233652086,
      "loss": 12.7287,
      "step": 1664
    },
    {
      "epoch": 0.0906659043107145,
      "grad_norm": 1.0350350328989921,
      "learning_rate": 0.00019987678510643457,
      "loss": 12.7622,
      "step": 1665
    },
    {
      "epoch": 0.09072035830729752,
      "grad_norm": 0.7795116162583914,
      "learning_rate": 0.0001998763470996897,
      "loss": 12.644,
      "step": 1666
    },
    {
      "epoch": 0.09077481230388053,
      "grad_norm": 1.0945356128978212,
      "learning_rate": 0.00019987590831628955,
      "loss": 12.7748,
      "step": 1667
    },
    {
      "epoch": 0.09082926630046353,
      "grad_norm": 0.9341997157336556,
      "learning_rate": 0.00019987546875623765,
      "loss": 12.6985,
      "step": 1668
    },
    {
      "epoch": 0.09088372029704656,
      "grad_norm": 1.0754647338306391,
      "learning_rate": 0.0001998750284195373,
      "loss": 12.743,
      "step": 1669
    },
    {
      "epoch": 0.09093817429362956,
      "grad_norm": 0.8777951178194842,
      "learning_rate": 0.00019987458730619202,
      "loss": 12.6949,
      "step": 1670
    },
    {
      "epoch": 0.09099262829021257,
      "grad_norm": 0.7755947654045108,
      "learning_rate": 0.0001998741454162052,
      "loss": 12.7449,
      "step": 1671
    },
    {
      "epoch": 0.09104708228679559,
      "grad_norm": 0.9579356976820738,
      "learning_rate": 0.00019987370274958025,
      "loss": 12.6444,
      "step": 1672
    },
    {
      "epoch": 0.0911015362833786,
      "grad_norm": 0.8891914081952949,
      "learning_rate": 0.00019987325930632065,
      "loss": 12.878,
      "step": 1673
    },
    {
      "epoch": 0.09115599027996162,
      "grad_norm": 0.9225125348962756,
      "learning_rate": 0.00019987281508642983,
      "loss": 12.5637,
      "step": 1674
    },
    {
      "epoch": 0.09121044427654462,
      "grad_norm": 0.7929658741370371,
      "learning_rate": 0.00019987237008991127,
      "loss": 12.5481,
      "step": 1675
    },
    {
      "epoch": 0.09126489827312763,
      "grad_norm": 0.8286918511585734,
      "learning_rate": 0.00019987192431676843,
      "loss": 12.7087,
      "step": 1676
    },
    {
      "epoch": 0.09131935226971065,
      "grad_norm": 0.8387876060507591,
      "learning_rate": 0.00019987147776700473,
      "loss": 12.6977,
      "step": 1677
    },
    {
      "epoch": 0.09137380626629366,
      "grad_norm": 0.8037519330448577,
      "learning_rate": 0.00019987103044062364,
      "loss": 12.7798,
      "step": 1678
    },
    {
      "epoch": 0.09142826026287666,
      "grad_norm": 0.8990989280319842,
      "learning_rate": 0.0001998705823376287,
      "loss": 12.7042,
      "step": 1679
    },
    {
      "epoch": 0.09148271425945968,
      "grad_norm": 0.801847772818663,
      "learning_rate": 0.00019987013345802336,
      "loss": 12.8197,
      "step": 1680
    },
    {
      "epoch": 0.09153716825604269,
      "grad_norm": 0.8304429597862878,
      "learning_rate": 0.00019986968380181113,
      "loss": 12.682,
      "step": 1681
    },
    {
      "epoch": 0.09159162225262571,
      "grad_norm": 0.886711334594854,
      "learning_rate": 0.00019986923336899547,
      "loss": 12.5952,
      "step": 1682
    },
    {
      "epoch": 0.09164607624920872,
      "grad_norm": 0.7732843373918673,
      "learning_rate": 0.0001998687821595799,
      "loss": 12.6595,
      "step": 1683
    },
    {
      "epoch": 0.09170053024579172,
      "grad_norm": 0.9707310684742881,
      "learning_rate": 0.00019986833017356797,
      "loss": 12.7915,
      "step": 1684
    },
    {
      "epoch": 0.09175498424237474,
      "grad_norm": 0.7450978514787958,
      "learning_rate": 0.00019986787741096311,
      "loss": 12.6324,
      "step": 1685
    },
    {
      "epoch": 0.09180943823895775,
      "grad_norm": 0.8637568515593369,
      "learning_rate": 0.0001998674238717689,
      "loss": 12.729,
      "step": 1686
    },
    {
      "epoch": 0.09186389223554076,
      "grad_norm": 0.7476349343697685,
      "learning_rate": 0.0001998669695559889,
      "loss": 12.6907,
      "step": 1687
    },
    {
      "epoch": 0.09191834623212378,
      "grad_norm": 0.7931163580976857,
      "learning_rate": 0.00019986651446362653,
      "loss": 12.8222,
      "step": 1688
    },
    {
      "epoch": 0.09197280022870678,
      "grad_norm": 0.7635941349745159,
      "learning_rate": 0.00019986605859468543,
      "loss": 12.6869,
      "step": 1689
    },
    {
      "epoch": 0.0920272542252898,
      "grad_norm": 0.7402233735864691,
      "learning_rate": 0.0001998656019491691,
      "loss": 12.6336,
      "step": 1690
    },
    {
      "epoch": 0.09208170822187281,
      "grad_norm": 0.7765142940286349,
      "learning_rate": 0.0001998651445270811,
      "loss": 12.6721,
      "step": 1691
    },
    {
      "epoch": 0.09213616221845582,
      "grad_norm": 0.8834448067628252,
      "learning_rate": 0.000199864686328425,
      "loss": 12.763,
      "step": 1692
    },
    {
      "epoch": 0.09219061621503884,
      "grad_norm": 0.7297738087025569,
      "learning_rate": 0.00019986422735320436,
      "loss": 12.7451,
      "step": 1693
    },
    {
      "epoch": 0.09224507021162184,
      "grad_norm": 0.7648076695333648,
      "learning_rate": 0.00019986376760142274,
      "loss": 12.8448,
      "step": 1694
    },
    {
      "epoch": 0.09229952420820485,
      "grad_norm": 0.7793752591738599,
      "learning_rate": 0.00019986330707308367,
      "loss": 12.9426,
      "step": 1695
    },
    {
      "epoch": 0.09235397820478787,
      "grad_norm": 0.7778053571595327,
      "learning_rate": 0.00019986284576819084,
      "loss": 12.6704,
      "step": 1696
    },
    {
      "epoch": 0.09240843220137088,
      "grad_norm": 0.7202035826372535,
      "learning_rate": 0.00019986238368674774,
      "loss": 12.7045,
      "step": 1697
    },
    {
      "epoch": 0.0924628861979539,
      "grad_norm": 0.7528090876710528,
      "learning_rate": 0.000199861920828758,
      "loss": 12.7146,
      "step": 1698
    },
    {
      "epoch": 0.0925173401945369,
      "grad_norm": 0.7248727230127,
      "learning_rate": 0.00019986145719422523,
      "loss": 12.6796,
      "step": 1699
    },
    {
      "epoch": 0.09257179419111991,
      "grad_norm": 0.7771517513852315,
      "learning_rate": 0.000199860992783153,
      "loss": 12.7017,
      "step": 1700
    },
    {
      "epoch": 0.09262624818770293,
      "grad_norm": 0.7915660755684254,
      "learning_rate": 0.00019986052759554497,
      "loss": 12.6844,
      "step": 1701
    },
    {
      "epoch": 0.09268070218428594,
      "grad_norm": 0.8352445974040995,
      "learning_rate": 0.0001998600616314047,
      "loss": 12.7781,
      "step": 1702
    },
    {
      "epoch": 0.09273515618086894,
      "grad_norm": 0.7784112929413094,
      "learning_rate": 0.00019985959489073586,
      "loss": 12.6843,
      "step": 1703
    },
    {
      "epoch": 0.09278961017745196,
      "grad_norm": 0.9289410263110156,
      "learning_rate": 0.00019985912737354206,
      "loss": 12.7159,
      "step": 1704
    },
    {
      "epoch": 0.09284406417403497,
      "grad_norm": 0.7633534931473822,
      "learning_rate": 0.00019985865907982695,
      "loss": 12.6632,
      "step": 1705
    },
    {
      "epoch": 0.09289851817061799,
      "grad_norm": 0.7276428028327624,
      "learning_rate": 0.00019985819000959416,
      "loss": 12.7251,
      "step": 1706
    },
    {
      "epoch": 0.092952972167201,
      "grad_norm": 0.819345051368717,
      "learning_rate": 0.0001998577201628473,
      "loss": 12.667,
      "step": 1707
    },
    {
      "epoch": 0.093007426163784,
      "grad_norm": 0.7978315345932591,
      "learning_rate": 0.00019985724953959012,
      "loss": 12.7466,
      "step": 1708
    },
    {
      "epoch": 0.09306188016036702,
      "grad_norm": 0.7802810303182017,
      "learning_rate": 0.0001998567781398262,
      "loss": 12.6264,
      "step": 1709
    },
    {
      "epoch": 0.09311633415695003,
      "grad_norm": 0.8275515696845572,
      "learning_rate": 0.0001998563059635592,
      "loss": 12.6937,
      "step": 1710
    },
    {
      "epoch": 0.09317078815353304,
      "grad_norm": 0.8539871682918575,
      "learning_rate": 0.00019985583301079286,
      "loss": 12.7544,
      "step": 1711
    },
    {
      "epoch": 0.09322524215011606,
      "grad_norm": 0.7965901249303493,
      "learning_rate": 0.00019985535928153077,
      "loss": 12.743,
      "step": 1712
    },
    {
      "epoch": 0.09327969614669906,
      "grad_norm": 1.0469523307408424,
      "learning_rate": 0.00019985488477577672,
      "loss": 12.835,
      "step": 1713
    },
    {
      "epoch": 0.09333415014328209,
      "grad_norm": 1.0200281444576251,
      "learning_rate": 0.0001998544094935343,
      "loss": 12.7051,
      "step": 1714
    },
    {
      "epoch": 0.09338860413986509,
      "grad_norm": 0.9224592123697221,
      "learning_rate": 0.00019985393343480726,
      "loss": 12.7287,
      "step": 1715
    },
    {
      "epoch": 0.0934430581364481,
      "grad_norm": 0.8094401652320615,
      "learning_rate": 0.00019985345659959927,
      "loss": 12.7504,
      "step": 1716
    },
    {
      "epoch": 0.09349751213303112,
      "grad_norm": 0.8806549796355795,
      "learning_rate": 0.00019985297898791407,
      "loss": 12.7822,
      "step": 1717
    },
    {
      "epoch": 0.09355196612961413,
      "grad_norm": 0.8017319253189918,
      "learning_rate": 0.00019985250059975534,
      "loss": 12.7851,
      "step": 1718
    },
    {
      "epoch": 0.09360642012619713,
      "grad_norm": 0.7970534334621997,
      "learning_rate": 0.00019985202143512688,
      "loss": 12.727,
      "step": 1719
    },
    {
      "epoch": 0.09366087412278015,
      "grad_norm": 0.8444063092172269,
      "learning_rate": 0.00019985154149403228,
      "loss": 12.6435,
      "step": 1720
    },
    {
      "epoch": 0.09371532811936316,
      "grad_norm": 0.7854418054238925,
      "learning_rate": 0.00019985106077647543,
      "loss": 12.575,
      "step": 1721
    },
    {
      "epoch": 0.09376978211594617,
      "grad_norm": 0.9488296508732591,
      "learning_rate": 0.00019985057928245992,
      "loss": 12.689,
      "step": 1722
    },
    {
      "epoch": 0.09382423611252919,
      "grad_norm": 0.9478793448789539,
      "learning_rate": 0.00019985009701198957,
      "loss": 12.8562,
      "step": 1723
    },
    {
      "epoch": 0.09387869010911219,
      "grad_norm": 1.3421603271758058,
      "learning_rate": 0.00019984961396506815,
      "loss": 12.7497,
      "step": 1724
    },
    {
      "epoch": 0.09393314410569521,
      "grad_norm": 1.0297410530163407,
      "learning_rate": 0.00019984913014169938,
      "loss": 12.7798,
      "step": 1725
    },
    {
      "epoch": 0.09398759810227822,
      "grad_norm": 1.1291195574382247,
      "learning_rate": 0.000199848645541887,
      "loss": 12.648,
      "step": 1726
    },
    {
      "epoch": 0.09404205209886123,
      "grad_norm": 0.9849925329312679,
      "learning_rate": 0.00019984816016563483,
      "loss": 12.5487,
      "step": 1727
    },
    {
      "epoch": 0.09409650609544425,
      "grad_norm": 0.8181122519141973,
      "learning_rate": 0.0001998476740129466,
      "loss": 12.6672,
      "step": 1728
    },
    {
      "epoch": 0.09415096009202725,
      "grad_norm": 0.9015974890306336,
      "learning_rate": 0.00019984718708382615,
      "loss": 12.771,
      "step": 1729
    },
    {
      "epoch": 0.09420541408861026,
      "grad_norm": 0.8528008739132253,
      "learning_rate": 0.00019984669937827719,
      "loss": 12.7735,
      "step": 1730
    },
    {
      "epoch": 0.09425986808519328,
      "grad_norm": 0.8831817396018113,
      "learning_rate": 0.00019984621089630356,
      "loss": 12.696,
      "step": 1731
    },
    {
      "epoch": 0.09431432208177629,
      "grad_norm": 0.834736105594587,
      "learning_rate": 0.00019984572163790908,
      "loss": 12.7276,
      "step": 1732
    },
    {
      "epoch": 0.0943687760783593,
      "grad_norm": 0.9257863305824491,
      "learning_rate": 0.00019984523160309752,
      "loss": 12.8682,
      "step": 1733
    },
    {
      "epoch": 0.09442323007494231,
      "grad_norm": 0.8519996446656369,
      "learning_rate": 0.00019984474079187266,
      "loss": 12.7941,
      "step": 1734
    },
    {
      "epoch": 0.09447768407152532,
      "grad_norm": 0.8464983573387057,
      "learning_rate": 0.00019984424920423837,
      "loss": 12.7706,
      "step": 1735
    },
    {
      "epoch": 0.09453213806810834,
      "grad_norm": 0.8616173053640237,
      "learning_rate": 0.00019984375684019848,
      "loss": 12.7734,
      "step": 1736
    },
    {
      "epoch": 0.09458659206469135,
      "grad_norm": 0.8584513670065499,
      "learning_rate": 0.00019984326369975675,
      "loss": 12.8588,
      "step": 1737
    },
    {
      "epoch": 0.09464104606127435,
      "grad_norm": 0.7275419976228897,
      "learning_rate": 0.00019984276978291709,
      "loss": 12.6037,
      "step": 1738
    },
    {
      "epoch": 0.09469550005785737,
      "grad_norm": 0.90704252581912,
      "learning_rate": 0.00019984227508968328,
      "loss": 12.8174,
      "step": 1739
    },
    {
      "epoch": 0.09474995405444038,
      "grad_norm": 0.7890718965669796,
      "learning_rate": 0.0001998417796200592,
      "loss": 12.7361,
      "step": 1740
    },
    {
      "epoch": 0.0948044080510234,
      "grad_norm": 0.8700908686746459,
      "learning_rate": 0.0001998412833740487,
      "loss": 12.7187,
      "step": 1741
    },
    {
      "epoch": 0.09485886204760641,
      "grad_norm": 0.8820330592465839,
      "learning_rate": 0.00019984078635165565,
      "loss": 12.7653,
      "step": 1742
    },
    {
      "epoch": 0.09491331604418941,
      "grad_norm": 0.8480096022522114,
      "learning_rate": 0.0001998402885528839,
      "loss": 12.7305,
      "step": 1743
    },
    {
      "epoch": 0.09496777004077243,
      "grad_norm": 0.844964622166045,
      "learning_rate": 0.00019983978997773733,
      "loss": 12.6504,
      "step": 1744
    },
    {
      "epoch": 0.09502222403735544,
      "grad_norm": 0.7337818119185138,
      "learning_rate": 0.0001998392906262198,
      "loss": 12.6787,
      "step": 1745
    },
    {
      "epoch": 0.09507667803393845,
      "grad_norm": 0.8951156428654814,
      "learning_rate": 0.0001998387904983352,
      "loss": 12.629,
      "step": 1746
    },
    {
      "epoch": 0.09513113203052147,
      "grad_norm": 0.7417498529985256,
      "learning_rate": 0.00019983828959408743,
      "loss": 12.6818,
      "step": 1747
    },
    {
      "epoch": 0.09518558602710447,
      "grad_norm": 0.8106578184852027,
      "learning_rate": 0.00019983778791348038,
      "loss": 12.6571,
      "step": 1748
    },
    {
      "epoch": 0.0952400400236875,
      "grad_norm": 0.8339705961484438,
      "learning_rate": 0.00019983728545651795,
      "loss": 12.7501,
      "step": 1749
    },
    {
      "epoch": 0.0952944940202705,
      "grad_norm": 0.7656032487484683,
      "learning_rate": 0.00019983678222320402,
      "loss": 12.6431,
      "step": 1750
    },
    {
      "epoch": 0.09534894801685351,
      "grad_norm": 0.7678729913466037,
      "learning_rate": 0.00019983627821354254,
      "loss": 12.7806,
      "step": 1751
    },
    {
      "epoch": 0.09540340201343653,
      "grad_norm": 0.7412100270850847,
      "learning_rate": 0.00019983577342753744,
      "loss": 12.8495,
      "step": 1752
    },
    {
      "epoch": 0.09545785601001953,
      "grad_norm": 0.8329923403455379,
      "learning_rate": 0.0001998352678651926,
      "loss": 12.6991,
      "step": 1753
    },
    {
      "epoch": 0.09551231000660254,
      "grad_norm": 0.6852399910270895,
      "learning_rate": 0.00019983476152651196,
      "loss": 12.6196,
      "step": 1754
    },
    {
      "epoch": 0.09556676400318556,
      "grad_norm": 0.7473720505363164,
      "learning_rate": 0.0001998342544114995,
      "loss": 12.641,
      "step": 1755
    },
    {
      "epoch": 0.09562121799976857,
      "grad_norm": 0.8072951108918488,
      "learning_rate": 0.00019983374652015915,
      "loss": 12.6861,
      "step": 1756
    },
    {
      "epoch": 0.09567567199635159,
      "grad_norm": 0.8901914678671833,
      "learning_rate": 0.0001998332378524948,
      "loss": 12.7401,
      "step": 1757
    },
    {
      "epoch": 0.0957301259929346,
      "grad_norm": 0.7381858416182421,
      "learning_rate": 0.00019983272840851048,
      "loss": 12.6173,
      "step": 1758
    },
    {
      "epoch": 0.0957845799895176,
      "grad_norm": 0.7780328757663166,
      "learning_rate": 0.00019983221818821011,
      "loss": 12.6119,
      "step": 1759
    },
    {
      "epoch": 0.09583903398610062,
      "grad_norm": 0.8312609985976824,
      "learning_rate": 0.00019983170719159769,
      "loss": 12.7331,
      "step": 1760
    },
    {
      "epoch": 0.09589348798268363,
      "grad_norm": 0.8830377568322108,
      "learning_rate": 0.00019983119541867718,
      "loss": 12.9335,
      "step": 1761
    },
    {
      "epoch": 0.09594794197926663,
      "grad_norm": 1.0031691115579495,
      "learning_rate": 0.0001998306828694525,
      "loss": 12.6282,
      "step": 1762
    },
    {
      "epoch": 0.09600239597584966,
      "grad_norm": 0.8776848123359052,
      "learning_rate": 0.00019983016954392771,
      "loss": 12.5339,
      "step": 1763
    },
    {
      "epoch": 0.09605684997243266,
      "grad_norm": 0.9444493402268572,
      "learning_rate": 0.0001998296554421068,
      "loss": 12.797,
      "step": 1764
    },
    {
      "epoch": 0.09611130396901568,
      "grad_norm": 0.8453358988930049,
      "learning_rate": 0.00019982914056399374,
      "loss": 12.7444,
      "step": 1765
    },
    {
      "epoch": 0.09616575796559869,
      "grad_norm": 1.0057800902831977,
      "learning_rate": 0.00019982862490959256,
      "loss": 12.615,
      "step": 1766
    },
    {
      "epoch": 0.0962202119621817,
      "grad_norm": 0.8227321761404245,
      "learning_rate": 0.0001998281084789072,
      "loss": 12.7163,
      "step": 1767
    },
    {
      "epoch": 0.09627466595876472,
      "grad_norm": 0.9078437574448277,
      "learning_rate": 0.00019982759127194178,
      "loss": 12.8353,
      "step": 1768
    },
    {
      "epoch": 0.09632911995534772,
      "grad_norm": 0.8607841260012521,
      "learning_rate": 0.00019982707328870025,
      "loss": 12.7217,
      "step": 1769
    },
    {
      "epoch": 0.09638357395193073,
      "grad_norm": 0.806857734594643,
      "learning_rate": 0.00019982655452918663,
      "loss": 12.8509,
      "step": 1770
    },
    {
      "epoch": 0.09643802794851375,
      "grad_norm": 0.9016798428985953,
      "learning_rate": 0.00019982603499340502,
      "loss": 12.7054,
      "step": 1771
    },
    {
      "epoch": 0.09649248194509676,
      "grad_norm": 0.8270136349810251,
      "learning_rate": 0.00019982551468135943,
      "loss": 12.7688,
      "step": 1772
    },
    {
      "epoch": 0.09654693594167978,
      "grad_norm": 0.7586339597744727,
      "learning_rate": 0.00019982499359305384,
      "loss": 12.6027,
      "step": 1773
    },
    {
      "epoch": 0.09660138993826278,
      "grad_norm": 0.8039689280170982,
      "learning_rate": 0.00019982447172849243,
      "loss": 12.6849,
      "step": 1774
    },
    {
      "epoch": 0.09665584393484579,
      "grad_norm": 0.864875110019144,
      "learning_rate": 0.00019982394908767912,
      "loss": 12.8059,
      "step": 1775
    },
    {
      "epoch": 0.09671029793142881,
      "grad_norm": 0.7660481506729493,
      "learning_rate": 0.00019982342567061807,
      "loss": 12.6182,
      "step": 1776
    },
    {
      "epoch": 0.09676475192801182,
      "grad_norm": 0.8262935586777914,
      "learning_rate": 0.00019982290147731334,
      "loss": 12.6749,
      "step": 1777
    },
    {
      "epoch": 0.09681920592459482,
      "grad_norm": 0.8597501934936786,
      "learning_rate": 0.00019982237650776897,
      "loss": 12.8307,
      "step": 1778
    },
    {
      "epoch": 0.09687365992117784,
      "grad_norm": 0.9670302026180937,
      "learning_rate": 0.00019982185076198905,
      "loss": 12.6684,
      "step": 1779
    },
    {
      "epoch": 0.09692811391776085,
      "grad_norm": 1.0795139067045474,
      "learning_rate": 0.0001998213242399777,
      "loss": 12.7065,
      "step": 1780
    },
    {
      "epoch": 0.09698256791434387,
      "grad_norm": 0.8816692193509686,
      "learning_rate": 0.00019982079694173897,
      "loss": 12.6734,
      "step": 1781
    },
    {
      "epoch": 0.09703702191092688,
      "grad_norm": 0.7591381912463186,
      "learning_rate": 0.00019982026886727702,
      "loss": 12.5892,
      "step": 1782
    },
    {
      "epoch": 0.09709147590750988,
      "grad_norm": 0.831267387648959,
      "learning_rate": 0.00019981974001659586,
      "loss": 12.687,
      "step": 1783
    },
    {
      "epoch": 0.0971459299040929,
      "grad_norm": 0.8723663496857355,
      "learning_rate": 0.0001998192103896997,
      "loss": 12.7374,
      "step": 1784
    },
    {
      "epoch": 0.09720038390067591,
      "grad_norm": 0.7613084972321603,
      "learning_rate": 0.0001998186799865926,
      "loss": 12.8568,
      "step": 1785
    },
    {
      "epoch": 0.09725483789725892,
      "grad_norm": 0.9693531135750679,
      "learning_rate": 0.00019981814880727875,
      "loss": 12.7736,
      "step": 1786
    },
    {
      "epoch": 0.09730929189384194,
      "grad_norm": 0.8070502979590324,
      "learning_rate": 0.00019981761685176222,
      "loss": 12.7267,
      "step": 1787
    },
    {
      "epoch": 0.09736374589042494,
      "grad_norm": 0.7585035282744843,
      "learning_rate": 0.0001998170841200471,
      "loss": 12.6695,
      "step": 1788
    },
    {
      "epoch": 0.09741819988700795,
      "grad_norm": 0.8562878599847409,
      "learning_rate": 0.00019981655061213766,
      "loss": 12.7517,
      "step": 1789
    },
    {
      "epoch": 0.09747265388359097,
      "grad_norm": 0.8076468198038834,
      "learning_rate": 0.000199816016328038,
      "loss": 12.6599,
      "step": 1790
    },
    {
      "epoch": 0.09752710788017398,
      "grad_norm": 0.8624308020640535,
      "learning_rate": 0.0001998154812677522,
      "loss": 12.8433,
      "step": 1791
    },
    {
      "epoch": 0.097581561876757,
      "grad_norm": 0.8124216890531085,
      "learning_rate": 0.00019981494543128448,
      "loss": 12.7618,
      "step": 1792
    },
    {
      "epoch": 0.09763601587334,
      "grad_norm": 0.9145054786820154,
      "learning_rate": 0.00019981440881863905,
      "loss": 12.6682,
      "step": 1793
    },
    {
      "epoch": 0.09769046986992301,
      "grad_norm": 0.8483562639490578,
      "learning_rate": 0.00019981387142982003,
      "loss": 12.8717,
      "step": 1794
    },
    {
      "epoch": 0.09774492386650603,
      "grad_norm": 0.8409069976588321,
      "learning_rate": 0.00019981333326483158,
      "loss": 12.7535,
      "step": 1795
    },
    {
      "epoch": 0.09779937786308904,
      "grad_norm": 0.8193328017494412,
      "learning_rate": 0.0001998127943236779,
      "loss": 12.8387,
      "step": 1796
    },
    {
      "epoch": 0.09785383185967204,
      "grad_norm": 0.7893282197935113,
      "learning_rate": 0.00019981225460636326,
      "loss": 12.6631,
      "step": 1797
    },
    {
      "epoch": 0.09790828585625506,
      "grad_norm": 0.9814262516729859,
      "learning_rate": 0.00019981171411289172,
      "loss": 12.7136,
      "step": 1798
    },
    {
      "epoch": 0.09796273985283807,
      "grad_norm": 0.8465352965252302,
      "learning_rate": 0.00019981117284326757,
      "loss": 12.7594,
      "step": 1799
    },
    {
      "epoch": 0.09801719384942109,
      "grad_norm": 0.8189406451797356,
      "learning_rate": 0.00019981063079749505,
      "loss": 12.6765,
      "step": 1800
    },
    {
      "epoch": 0.0980716478460041,
      "grad_norm": 0.8344960848336206,
      "learning_rate": 0.00019981008797557827,
      "loss": 12.7143,
      "step": 1801
    },
    {
      "epoch": 0.0981261018425871,
      "grad_norm": 0.793799228357131,
      "learning_rate": 0.00019980954437752153,
      "loss": 12.6669,
      "step": 1802
    },
    {
      "epoch": 0.09818055583917012,
      "grad_norm": 0.8231053066247628,
      "learning_rate": 0.00019980900000332903,
      "loss": 12.7411,
      "step": 1803
    },
    {
      "epoch": 0.09823500983575313,
      "grad_norm": 0.8060131383467297,
      "learning_rate": 0.000199808454853005,
      "loss": 12.7213,
      "step": 1804
    },
    {
      "epoch": 0.09828946383233614,
      "grad_norm": 0.7542976640094788,
      "learning_rate": 0.0001998079089265537,
      "loss": 12.7119,
      "step": 1805
    },
    {
      "epoch": 0.09834391782891916,
      "grad_norm": 0.8811074779603901,
      "learning_rate": 0.0001998073622239794,
      "loss": 12.7143,
      "step": 1806
    },
    {
      "epoch": 0.09839837182550216,
      "grad_norm": 0.7443022336030499,
      "learning_rate": 0.00019980681474528623,
      "loss": 12.5875,
      "step": 1807
    },
    {
      "epoch": 0.09845282582208519,
      "grad_norm": 0.8255299157055661,
      "learning_rate": 0.0001998062664904786,
      "loss": 12.777,
      "step": 1808
    },
    {
      "epoch": 0.09850727981866819,
      "grad_norm": 0.7914682742250475,
      "learning_rate": 0.00019980571745956068,
      "loss": 12.6854,
      "step": 1809
    },
    {
      "epoch": 0.0985617338152512,
      "grad_norm": 0.9831540550159134,
      "learning_rate": 0.00019980516765253674,
      "loss": 12.8013,
      "step": 1810
    },
    {
      "epoch": 0.09861618781183422,
      "grad_norm": 0.7181653532335807,
      "learning_rate": 0.0001998046170694111,
      "loss": 12.6832,
      "step": 1811
    },
    {
      "epoch": 0.09867064180841723,
      "grad_norm": 0.8613942793337783,
      "learning_rate": 0.000199804065710188,
      "loss": 12.7333,
      "step": 1812
    },
    {
      "epoch": 0.09872509580500023,
      "grad_norm": 0.7765773016663301,
      "learning_rate": 0.00019980351357487178,
      "loss": 12.7315,
      "step": 1813
    },
    {
      "epoch": 0.09877954980158325,
      "grad_norm": 0.6808776096578668,
      "learning_rate": 0.0001998029606634667,
      "loss": 12.7699,
      "step": 1814
    },
    {
      "epoch": 0.09883400379816626,
      "grad_norm": 0.8799736462156778,
      "learning_rate": 0.00019980240697597704,
      "loss": 12.7738,
      "step": 1815
    },
    {
      "epoch": 0.09888845779474928,
      "grad_norm": 0.8627772359823945,
      "learning_rate": 0.00019980185251240715,
      "loss": 12.7422,
      "step": 1816
    },
    {
      "epoch": 0.09894291179133229,
      "grad_norm": 0.9523158797483452,
      "learning_rate": 0.00019980129727276128,
      "loss": 12.6161,
      "step": 1817
    },
    {
      "epoch": 0.09899736578791529,
      "grad_norm": 0.8316339425431188,
      "learning_rate": 0.00019980074125704381,
      "loss": 12.7533,
      "step": 1818
    },
    {
      "epoch": 0.09905181978449831,
      "grad_norm": 0.7555606839824176,
      "learning_rate": 0.00019980018446525904,
      "loss": 12.8045,
      "step": 1819
    },
    {
      "epoch": 0.09910627378108132,
      "grad_norm": 0.8736839244670254,
      "learning_rate": 0.00019979962689741133,
      "loss": 12.8051,
      "step": 1820
    },
    {
      "epoch": 0.09916072777766433,
      "grad_norm": 0.8298289371961125,
      "learning_rate": 0.00019979906855350493,
      "loss": 12.7442,
      "step": 1821
    },
    {
      "epoch": 0.09921518177424735,
      "grad_norm": 0.7359470330154858,
      "learning_rate": 0.00019979850943354429,
      "loss": 12.6789,
      "step": 1822
    },
    {
      "epoch": 0.09926963577083035,
      "grad_norm": 0.7984255387074274,
      "learning_rate": 0.00019979794953753368,
      "loss": 12.6387,
      "step": 1823
    },
    {
      "epoch": 0.09932408976741337,
      "grad_norm": 0.7948787479597286,
      "learning_rate": 0.00019979738886547748,
      "loss": 12.7372,
      "step": 1824
    },
    {
      "epoch": 0.09937854376399638,
      "grad_norm": 0.7525955093049125,
      "learning_rate": 0.00019979682741738005,
      "loss": 12.5815,
      "step": 1825
    },
    {
      "epoch": 0.09943299776057939,
      "grad_norm": 0.8045976161147929,
      "learning_rate": 0.00019979626519324572,
      "loss": 12.6558,
      "step": 1826
    },
    {
      "epoch": 0.0994874517571624,
      "grad_norm": 0.868763583150912,
      "learning_rate": 0.00019979570219307892,
      "loss": 12.9435,
      "step": 1827
    },
    {
      "epoch": 0.09954190575374541,
      "grad_norm": 0.8254280429266189,
      "learning_rate": 0.000199795138416884,
      "loss": 12.7005,
      "step": 1828
    },
    {
      "epoch": 0.09959635975032842,
      "grad_norm": 0.8215715474865103,
      "learning_rate": 0.00019979457386466536,
      "loss": 12.6894,
      "step": 1829
    },
    {
      "epoch": 0.09965081374691144,
      "grad_norm": 0.8617949069115154,
      "learning_rate": 0.0001997940085364274,
      "loss": 12.8104,
      "step": 1830
    },
    {
      "epoch": 0.09970526774349445,
      "grad_norm": 0.7084497799870297,
      "learning_rate": 0.00019979344243217445,
      "loss": 12.5418,
      "step": 1831
    },
    {
      "epoch": 0.09975972174007747,
      "grad_norm": 0.7914428213251324,
      "learning_rate": 0.00019979287555191096,
      "loss": 12.7525,
      "step": 1832
    },
    {
      "epoch": 0.09981417573666047,
      "grad_norm": 0.8036534649424923,
      "learning_rate": 0.00019979230789564137,
      "loss": 12.8148,
      "step": 1833
    },
    {
      "epoch": 0.09986862973324348,
      "grad_norm": 0.783771502898549,
      "learning_rate": 0.00019979173946337,
      "loss": 12.6985,
      "step": 1834
    },
    {
      "epoch": 0.0999230837298265,
      "grad_norm": 0.8433229115010147,
      "learning_rate": 0.00019979117025510136,
      "loss": 12.8037,
      "step": 1835
    },
    {
      "epoch": 0.09997753772640951,
      "grad_norm": 0.7781085291475797,
      "learning_rate": 0.00019979060027083988,
      "loss": 12.6502,
      "step": 1836
    },
    {
      "epoch": 0.10003199172299251,
      "grad_norm": 0.7396520761371812,
      "learning_rate": 0.00019979002951058992,
      "loss": 12.6156,
      "step": 1837
    },
    {
      "epoch": 0.10008644571957553,
      "grad_norm": 0.7502516601071137,
      "learning_rate": 0.00019978945797435594,
      "loss": 12.6214,
      "step": 1838
    },
    {
      "epoch": 0.10014089971615854,
      "grad_norm": 0.7945760298938975,
      "learning_rate": 0.00019978888566214245,
      "loss": 12.7648,
      "step": 1839
    },
    {
      "epoch": 0.10019535371274156,
      "grad_norm": 0.805994217809907,
      "learning_rate": 0.00019978831257395384,
      "loss": 12.6749,
      "step": 1840
    },
    {
      "epoch": 0.10024980770932457,
      "grad_norm": 0.7585193198751154,
      "learning_rate": 0.00019978773870979452,
      "loss": 12.6268,
      "step": 1841
    },
    {
      "epoch": 0.10030426170590757,
      "grad_norm": 0.8305633847794065,
      "learning_rate": 0.00019978716406966905,
      "loss": 12.7269,
      "step": 1842
    },
    {
      "epoch": 0.1003587157024906,
      "grad_norm": 0.7422976190993655,
      "learning_rate": 0.00019978658865358185,
      "loss": 12.7719,
      "step": 1843
    },
    {
      "epoch": 0.1004131696990736,
      "grad_norm": 0.740572357468796,
      "learning_rate": 0.00019978601246153742,
      "loss": 12.7352,
      "step": 1844
    },
    {
      "epoch": 0.10046762369565661,
      "grad_norm": 0.7707640168766531,
      "learning_rate": 0.00019978543549354022,
      "loss": 12.4927,
      "step": 1845
    },
    {
      "epoch": 0.10052207769223963,
      "grad_norm": 0.815737711358069,
      "learning_rate": 0.00019978485774959474,
      "loss": 12.6872,
      "step": 1846
    },
    {
      "epoch": 0.10057653168882263,
      "grad_norm": 0.7948585836889104,
      "learning_rate": 0.00019978427922970546,
      "loss": 12.6934,
      "step": 1847
    },
    {
      "epoch": 0.10063098568540566,
      "grad_norm": 0.7752338166617883,
      "learning_rate": 0.0001997836999338769,
      "loss": 12.7227,
      "step": 1848
    },
    {
      "epoch": 0.10068543968198866,
      "grad_norm": 0.7549758786488137,
      "learning_rate": 0.00019978311986211354,
      "loss": 12.7107,
      "step": 1849
    },
    {
      "epoch": 0.10073989367857167,
      "grad_norm": 0.7228918115206341,
      "learning_rate": 0.00019978253901441992,
      "loss": 12.7228,
      "step": 1850
    },
    {
      "epoch": 0.10079434767515469,
      "grad_norm": 0.9869084616467323,
      "learning_rate": 0.00019978195739080054,
      "loss": 12.691,
      "step": 1851
    },
    {
      "epoch": 0.1008488016717377,
      "grad_norm": 0.7541298740075157,
      "learning_rate": 0.00019978137499125994,
      "loss": 12.6926,
      "step": 1852
    },
    {
      "epoch": 0.1009032556683207,
      "grad_norm": 0.8353392921949634,
      "learning_rate": 0.0001997807918158026,
      "loss": 12.693,
      "step": 1853
    },
    {
      "epoch": 0.10095770966490372,
      "grad_norm": 0.9022504957175957,
      "learning_rate": 0.00019978020786443312,
      "loss": 12.8224,
      "step": 1854
    },
    {
      "epoch": 0.10101216366148673,
      "grad_norm": 0.7754694247261438,
      "learning_rate": 0.00019977962313715602,
      "loss": 12.6533,
      "step": 1855
    },
    {
      "epoch": 0.10106661765806974,
      "grad_norm": 0.9496107528055869,
      "learning_rate": 0.0001997790376339758,
      "loss": 12.6253,
      "step": 1856
    },
    {
      "epoch": 0.10112107165465276,
      "grad_norm": 0.8295457108745374,
      "learning_rate": 0.00019977845135489707,
      "loss": 12.759,
      "step": 1857
    },
    {
      "epoch": 0.10117552565123576,
      "grad_norm": 0.9132320161804102,
      "learning_rate": 0.00019977786429992438,
      "loss": 12.7149,
      "step": 1858
    },
    {
      "epoch": 0.10122997964781878,
      "grad_norm": 0.9099955640502901,
      "learning_rate": 0.0001997772764690623,
      "loss": 12.7591,
      "step": 1859
    },
    {
      "epoch": 0.10128443364440179,
      "grad_norm": 0.9136455396020913,
      "learning_rate": 0.00019977668786231534,
      "loss": 12.6479,
      "step": 1860
    },
    {
      "epoch": 0.1013388876409848,
      "grad_norm": 1.0443368465604608,
      "learning_rate": 0.00019977609847968812,
      "loss": 12.7831,
      "step": 1861
    },
    {
      "epoch": 0.10139334163756782,
      "grad_norm": 0.8544793584331826,
      "learning_rate": 0.00019977550832118526,
      "loss": 12.8603,
      "step": 1862
    },
    {
      "epoch": 0.10144779563415082,
      "grad_norm": 0.9758123126638216,
      "learning_rate": 0.00019977491738681132,
      "loss": 12.8229,
      "step": 1863
    },
    {
      "epoch": 0.10150224963073383,
      "grad_norm": 1.0450927875618519,
      "learning_rate": 0.00019977432567657086,
      "loss": 12.785,
      "step": 1864
    },
    {
      "epoch": 0.10155670362731685,
      "grad_norm": 0.8321896909389046,
      "learning_rate": 0.0001997737331904685,
      "loss": 12.7621,
      "step": 1865
    },
    {
      "epoch": 0.10161115762389986,
      "grad_norm": 1.086697442673203,
      "learning_rate": 0.0001997731399285089,
      "loss": 12.7405,
      "step": 1866
    },
    {
      "epoch": 0.10166561162048288,
      "grad_norm": 0.8795583094614052,
      "learning_rate": 0.0001997725458906966,
      "loss": 12.6583,
      "step": 1867
    },
    {
      "epoch": 0.10172006561706588,
      "grad_norm": 0.9093661938885378,
      "learning_rate": 0.00019977195107703625,
      "loss": 12.6382,
      "step": 1868
    },
    {
      "epoch": 0.10177451961364889,
      "grad_norm": 0.8206637853045706,
      "learning_rate": 0.0001997713554875325,
      "loss": 12.7288,
      "step": 1869
    },
    {
      "epoch": 0.10182897361023191,
      "grad_norm": 0.7959548053380612,
      "learning_rate": 0.00019977075912218996,
      "loss": 12.5389,
      "step": 1870
    },
    {
      "epoch": 0.10188342760681492,
      "grad_norm": 0.8940054567632745,
      "learning_rate": 0.00019977016198101326,
      "loss": 12.7477,
      "step": 1871
    },
    {
      "epoch": 0.10193788160339792,
      "grad_norm": 0.7983708631762738,
      "learning_rate": 0.00019976956406400704,
      "loss": 12.6809,
      "step": 1872
    },
    {
      "epoch": 0.10199233559998094,
      "grad_norm": 0.8947777592247796,
      "learning_rate": 0.00019976896537117597,
      "loss": 12.7298,
      "step": 1873
    },
    {
      "epoch": 0.10204678959656395,
      "grad_norm": 0.7639695051807167,
      "learning_rate": 0.00019976836590252469,
      "loss": 12.8679,
      "step": 1874
    },
    {
      "epoch": 0.10210124359314697,
      "grad_norm": 0.8503436782329791,
      "learning_rate": 0.00019976776565805787,
      "loss": 12.7481,
      "step": 1875
    },
    {
      "epoch": 0.10215569758972998,
      "grad_norm": 0.7299504612088192,
      "learning_rate": 0.00019976716463778016,
      "loss": 12.6675,
      "step": 1876
    },
    {
      "epoch": 0.10221015158631298,
      "grad_norm": 0.8034290287601704,
      "learning_rate": 0.00019976656284169625,
      "loss": 12.7463,
      "step": 1877
    },
    {
      "epoch": 0.102264605582896,
      "grad_norm": 0.9378384856057707,
      "learning_rate": 0.0001997659602698108,
      "loss": 12.7667,
      "step": 1878
    },
    {
      "epoch": 0.10231905957947901,
      "grad_norm": 0.8112566177222267,
      "learning_rate": 0.00019976535692212854,
      "loss": 12.6914,
      "step": 1879
    },
    {
      "epoch": 0.10237351357606202,
      "grad_norm": 0.8836096715216687,
      "learning_rate": 0.00019976475279865415,
      "loss": 12.5431,
      "step": 1880
    },
    {
      "epoch": 0.10242796757264504,
      "grad_norm": 0.7678332184149458,
      "learning_rate": 0.00019976414789939226,
      "loss": 12.68,
      "step": 1881
    },
    {
      "epoch": 0.10248242156922804,
      "grad_norm": 0.9820205706063538,
      "learning_rate": 0.00019976354222434766,
      "loss": 12.5327,
      "step": 1882
    },
    {
      "epoch": 0.10253687556581106,
      "grad_norm": 0.839689545120463,
      "learning_rate": 0.00019976293577352502,
      "loss": 12.7281,
      "step": 1883
    },
    {
      "epoch": 0.10259132956239407,
      "grad_norm": 0.7763360763117715,
      "learning_rate": 0.00019976232854692903,
      "loss": 12.7423,
      "step": 1884
    },
    {
      "epoch": 0.10264578355897708,
      "grad_norm": 0.8263614999513452,
      "learning_rate": 0.0001997617205445645,
      "loss": 12.7349,
      "step": 1885
    },
    {
      "epoch": 0.1027002375555601,
      "grad_norm": 0.8544881902889455,
      "learning_rate": 0.00019976111176643607,
      "loss": 12.8311,
      "step": 1886
    },
    {
      "epoch": 0.1027546915521431,
      "grad_norm": 0.803396672866949,
      "learning_rate": 0.0001997605022125485,
      "loss": 12.75,
      "step": 1887
    },
    {
      "epoch": 0.10280914554872611,
      "grad_norm": 0.7346963496646222,
      "learning_rate": 0.00019975989188290654,
      "loss": 12.6567,
      "step": 1888
    },
    {
      "epoch": 0.10286359954530913,
      "grad_norm": 0.8449617676754794,
      "learning_rate": 0.00019975928077751496,
      "loss": 12.6853,
      "step": 1889
    },
    {
      "epoch": 0.10291805354189214,
      "grad_norm": 0.7721976538792544,
      "learning_rate": 0.00019975866889637844,
      "loss": 12.657,
      "step": 1890
    },
    {
      "epoch": 0.10297250753847516,
      "grad_norm": 0.8370477843159757,
      "learning_rate": 0.0001997580562395018,
      "loss": 12.7145,
      "step": 1891
    },
    {
      "epoch": 0.10302696153505816,
      "grad_norm": 0.786273334397099,
      "learning_rate": 0.0001997574428068898,
      "loss": 12.7155,
      "step": 1892
    },
    {
      "epoch": 0.10308141553164117,
      "grad_norm": 0.7773834981716732,
      "learning_rate": 0.00019975682859854716,
      "loss": 12.6723,
      "step": 1893
    },
    {
      "epoch": 0.10313586952822419,
      "grad_norm": 0.7131770678235348,
      "learning_rate": 0.0001997562136144787,
      "loss": 12.7067,
      "step": 1894
    },
    {
      "epoch": 0.1031903235248072,
      "grad_norm": 0.751203222047551,
      "learning_rate": 0.00019975559785468923,
      "loss": 12.7865,
      "step": 1895
    },
    {
      "epoch": 0.1032447775213902,
      "grad_norm": 0.8029767491515284,
      "learning_rate": 0.00019975498131918348,
      "loss": 12.8447,
      "step": 1896
    },
    {
      "epoch": 0.10329923151797323,
      "grad_norm": 0.8498280896911178,
      "learning_rate": 0.00019975436400796625,
      "loss": 12.6351,
      "step": 1897
    },
    {
      "epoch": 0.10335368551455623,
      "grad_norm": 0.8046931953621171,
      "learning_rate": 0.00019975374592104235,
      "loss": 12.814,
      "step": 1898
    },
    {
      "epoch": 0.10340813951113925,
      "grad_norm": 0.8911829559360334,
      "learning_rate": 0.00019975312705841663,
      "loss": 12.833,
      "step": 1899
    },
    {
      "epoch": 0.10346259350772226,
      "grad_norm": 0.9755131176201265,
      "learning_rate": 0.00019975250742009382,
      "loss": 12.7744,
      "step": 1900
    },
    {
      "epoch": 0.10351704750430527,
      "grad_norm": 0.7652603388028444,
      "learning_rate": 0.00019975188700607882,
      "loss": 12.6047,
      "step": 1901
    },
    {
      "epoch": 0.10357150150088829,
      "grad_norm": 0.7308159389641743,
      "learning_rate": 0.00019975126581637642,
      "loss": 12.54,
      "step": 1902
    },
    {
      "epoch": 0.10362595549747129,
      "grad_norm": 0.8841537311014933,
      "learning_rate": 0.00019975064385099143,
      "loss": 12.8176,
      "step": 1903
    },
    {
      "epoch": 0.1036804094940543,
      "grad_norm": 0.7959266490780836,
      "learning_rate": 0.0001997500211099287,
      "loss": 12.8648,
      "step": 1904
    },
    {
      "epoch": 0.10373486349063732,
      "grad_norm": 0.7932715396835438,
      "learning_rate": 0.0001997493975931931,
      "loss": 12.7503,
      "step": 1905
    },
    {
      "epoch": 0.10378931748722033,
      "grad_norm": 0.7059892133233987,
      "learning_rate": 0.00019974877330078945,
      "loss": 12.6694,
      "step": 1906
    },
    {
      "epoch": 0.10384377148380335,
      "grad_norm": 0.7647967486749511,
      "learning_rate": 0.00019974814823272265,
      "loss": 12.7022,
      "step": 1907
    },
    {
      "epoch": 0.10389822548038635,
      "grad_norm": 0.7149325329573871,
      "learning_rate": 0.00019974752238899744,
      "loss": 12.7661,
      "step": 1908
    },
    {
      "epoch": 0.10395267947696936,
      "grad_norm": 0.8302278048238623,
      "learning_rate": 0.00019974689576961882,
      "loss": 12.7787,
      "step": 1909
    },
    {
      "epoch": 0.10400713347355238,
      "grad_norm": 0.7650845662765747,
      "learning_rate": 0.00019974626837459161,
      "loss": 12.7433,
      "step": 1910
    },
    {
      "epoch": 0.10406158747013539,
      "grad_norm": 0.7263652929619784,
      "learning_rate": 0.00019974564020392067,
      "loss": 12.8098,
      "step": 1911
    },
    {
      "epoch": 0.10411604146671839,
      "grad_norm": 0.75783854526348,
      "learning_rate": 0.00019974501125761092,
      "loss": 12.6675,
      "step": 1912
    },
    {
      "epoch": 0.10417049546330141,
      "grad_norm": 0.9010729595464053,
      "learning_rate": 0.00019974438153566723,
      "loss": 12.7539,
      "step": 1913
    },
    {
      "epoch": 0.10422494945988442,
      "grad_norm": 0.8126529024951368,
      "learning_rate": 0.00019974375103809448,
      "loss": 12.8125,
      "step": 1914
    },
    {
      "epoch": 0.10427940345646744,
      "grad_norm": 0.941997554515698,
      "learning_rate": 0.0001997431197648976,
      "loss": 12.7994,
      "step": 1915
    },
    {
      "epoch": 0.10433385745305045,
      "grad_norm": 0.9094933655259666,
      "learning_rate": 0.00019974248771608154,
      "loss": 12.6271,
      "step": 1916
    },
    {
      "epoch": 0.10438831144963345,
      "grad_norm": 0.8626037953851565,
      "learning_rate": 0.00019974185489165112,
      "loss": 12.6484,
      "step": 1917
    },
    {
      "epoch": 0.10444276544621647,
      "grad_norm": 0.8310424783833855,
      "learning_rate": 0.00019974122129161133,
      "loss": 12.8424,
      "step": 1918
    },
    {
      "epoch": 0.10449721944279948,
      "grad_norm": 0.8262241343181942,
      "learning_rate": 0.00019974058691596706,
      "loss": 12.7997,
      "step": 1919
    },
    {
      "epoch": 0.10455167343938249,
      "grad_norm": 0.8746503937477267,
      "learning_rate": 0.0001997399517647233,
      "loss": 12.8625,
      "step": 1920
    },
    {
      "epoch": 0.1046061274359655,
      "grad_norm": 0.7630310464515302,
      "learning_rate": 0.0001997393158378849,
      "loss": 12.7733,
      "step": 1921
    },
    {
      "epoch": 0.10466058143254851,
      "grad_norm": 0.9309510895312205,
      "learning_rate": 0.0001997386791354569,
      "loss": 12.6717,
      "step": 1922
    },
    {
      "epoch": 0.10471503542913152,
      "grad_norm": 0.7652281673953589,
      "learning_rate": 0.00019973804165744418,
      "loss": 12.6441,
      "step": 1923
    },
    {
      "epoch": 0.10476948942571454,
      "grad_norm": 0.9094132289341906,
      "learning_rate": 0.0001997374034038517,
      "loss": 12.8794,
      "step": 1924
    },
    {
      "epoch": 0.10482394342229755,
      "grad_norm": 0.9693633912024958,
      "learning_rate": 0.0001997367643746845,
      "loss": 12.6089,
      "step": 1925
    },
    {
      "epoch": 0.10487839741888057,
      "grad_norm": 0.7918707360267294,
      "learning_rate": 0.00019973612456994743,
      "loss": 12.7389,
      "step": 1926
    },
    {
      "epoch": 0.10493285141546357,
      "grad_norm": 0.8084363062606581,
      "learning_rate": 0.00019973548398964557,
      "loss": 12.7046,
      "step": 1927
    },
    {
      "epoch": 0.10498730541204658,
      "grad_norm": 0.9597969610620458,
      "learning_rate": 0.00019973484263378387,
      "loss": 12.865,
      "step": 1928
    },
    {
      "epoch": 0.1050417594086296,
      "grad_norm": 0.8646966847427582,
      "learning_rate": 0.00019973420050236728,
      "loss": 12.7711,
      "step": 1929
    },
    {
      "epoch": 0.10509621340521261,
      "grad_norm": 0.8524972206331253,
      "learning_rate": 0.00019973355759540082,
      "loss": 12.7456,
      "step": 1930
    },
    {
      "epoch": 0.10515066740179561,
      "grad_norm": 0.7654825536901725,
      "learning_rate": 0.00019973291391288953,
      "loss": 12.7896,
      "step": 1931
    },
    {
      "epoch": 0.10520512139837863,
      "grad_norm": 0.8152675935793777,
      "learning_rate": 0.00019973226945483834,
      "loss": 12.7541,
      "step": 1932
    },
    {
      "epoch": 0.10525957539496164,
      "grad_norm": 0.8117049500801464,
      "learning_rate": 0.0001997316242212523,
      "loss": 12.8113,
      "step": 1933
    },
    {
      "epoch": 0.10531402939154466,
      "grad_norm": 0.8842097075775519,
      "learning_rate": 0.00019973097821213642,
      "loss": 12.7517,
      "step": 1934
    },
    {
      "epoch": 0.10536848338812767,
      "grad_norm": 0.7566468144569976,
      "learning_rate": 0.00019973033142749576,
      "loss": 12.7313,
      "step": 1935
    },
    {
      "epoch": 0.10542293738471067,
      "grad_norm": 0.7788525110569708,
      "learning_rate": 0.00019972968386733532,
      "loss": 12.8033,
      "step": 1936
    },
    {
      "epoch": 0.1054773913812937,
      "grad_norm": 0.8621093718166254,
      "learning_rate": 0.0001997290355316601,
      "loss": 12.9192,
      "step": 1937
    },
    {
      "epoch": 0.1055318453778767,
      "grad_norm": 0.732165906619715,
      "learning_rate": 0.0001997283864204752,
      "loss": 12.6428,
      "step": 1938
    },
    {
      "epoch": 0.10558629937445971,
      "grad_norm": 0.7521490290685846,
      "learning_rate": 0.00019972773653378562,
      "loss": 12.6545,
      "step": 1939
    },
    {
      "epoch": 0.10564075337104273,
      "grad_norm": 0.9478641791415195,
      "learning_rate": 0.00019972708587159642,
      "loss": 12.7364,
      "step": 1940
    },
    {
      "epoch": 0.10569520736762573,
      "grad_norm": 0.7784312137825782,
      "learning_rate": 0.0001997264344339127,
      "loss": 12.664,
      "step": 1941
    },
    {
      "epoch": 0.10574966136420876,
      "grad_norm": 0.7947087795298028,
      "learning_rate": 0.00019972578222073953,
      "loss": 12.7374,
      "step": 1942
    },
    {
      "epoch": 0.10580411536079176,
      "grad_norm": 0.7660446958449624,
      "learning_rate": 0.00019972512923208192,
      "loss": 12.6068,
      "step": 1943
    },
    {
      "epoch": 0.10585856935737477,
      "grad_norm": 0.9478451868901601,
      "learning_rate": 0.000199724475467945,
      "loss": 12.7313,
      "step": 1944
    },
    {
      "epoch": 0.10591302335395779,
      "grad_norm": 0.8443286080664616,
      "learning_rate": 0.00019972382092833381,
      "loss": 12.7797,
      "step": 1945
    },
    {
      "epoch": 0.1059674773505408,
      "grad_norm": 0.7757027982670757,
      "learning_rate": 0.00019972316561325348,
      "loss": 12.7802,
      "step": 1946
    },
    {
      "epoch": 0.1060219313471238,
      "grad_norm": 0.7323537505609665,
      "learning_rate": 0.0001997225095227091,
      "loss": 12.7715,
      "step": 1947
    },
    {
      "epoch": 0.10607638534370682,
      "grad_norm": 0.7258173389061959,
      "learning_rate": 0.00019972185265670572,
      "loss": 12.7192,
      "step": 1948
    },
    {
      "epoch": 0.10613083934028983,
      "grad_norm": 0.9886967263255692,
      "learning_rate": 0.00019972119501524853,
      "loss": 12.948,
      "step": 1949
    },
    {
      "epoch": 0.10618529333687285,
      "grad_norm": 0.779434641578507,
      "learning_rate": 0.0001997205365983426,
      "loss": 12.6265,
      "step": 1950
    },
    {
      "epoch": 0.10623974733345586,
      "grad_norm": 0.7378281448418083,
      "learning_rate": 0.00019971987740599305,
      "loss": 12.6904,
      "step": 1951
    },
    {
      "epoch": 0.10629420133003886,
      "grad_norm": 0.8763514796537106,
      "learning_rate": 0.00019971921743820503,
      "loss": 12.7581,
      "step": 1952
    },
    {
      "epoch": 0.10634865532662188,
      "grad_norm": 0.7598026179643976,
      "learning_rate": 0.00019971855669498364,
      "loss": 12.6264,
      "step": 1953
    },
    {
      "epoch": 0.10640310932320489,
      "grad_norm": 0.7677533412606428,
      "learning_rate": 0.00019971789517633402,
      "loss": 12.5165,
      "step": 1954
    },
    {
      "epoch": 0.1064575633197879,
      "grad_norm": 0.8014143428907597,
      "learning_rate": 0.00019971723288226133,
      "loss": 12.8213,
      "step": 1955
    },
    {
      "epoch": 0.10651201731637092,
      "grad_norm": 1.0166532682991876,
      "learning_rate": 0.0001997165698127707,
      "loss": 12.6291,
      "step": 1956
    },
    {
      "epoch": 0.10656647131295392,
      "grad_norm": 0.7397763869887779,
      "learning_rate": 0.00019971590596786732,
      "loss": 12.7036,
      "step": 1957
    },
    {
      "epoch": 0.10662092530953694,
      "grad_norm": 0.8243836863860697,
      "learning_rate": 0.0001997152413475563,
      "loss": 12.7095,
      "step": 1958
    },
    {
      "epoch": 0.10667537930611995,
      "grad_norm": 0.7607529958960203,
      "learning_rate": 0.0001997145759518429,
      "loss": 12.7681,
      "step": 1959
    },
    {
      "epoch": 0.10672983330270296,
      "grad_norm": 0.8500255738770411,
      "learning_rate": 0.00019971390978073219,
      "loss": 12.8752,
      "step": 1960
    },
    {
      "epoch": 0.10678428729928598,
      "grad_norm": 0.8429022005007072,
      "learning_rate": 0.0001997132428342294,
      "loss": 12.7818,
      "step": 1961
    },
    {
      "epoch": 0.10683874129586898,
      "grad_norm": 0.8973653852038352,
      "learning_rate": 0.00019971257511233975,
      "loss": 12.829,
      "step": 1962
    },
    {
      "epoch": 0.10689319529245199,
      "grad_norm": 0.9450106747699376,
      "learning_rate": 0.00019971190661506832,
      "loss": 12.7089,
      "step": 1963
    },
    {
      "epoch": 0.10694764928903501,
      "grad_norm": 0.8972410673730398,
      "learning_rate": 0.00019971123734242044,
      "loss": 12.469,
      "step": 1964
    },
    {
      "epoch": 0.10700210328561802,
      "grad_norm": 0.781966142046788,
      "learning_rate": 0.00019971056729440126,
      "loss": 12.5841,
      "step": 1965
    },
    {
      "epoch": 0.10705655728220104,
      "grad_norm": 0.8277637906824876,
      "learning_rate": 0.00019970989647101597,
      "loss": 12.6589,
      "step": 1966
    },
    {
      "epoch": 0.10711101127878404,
      "grad_norm": 0.8526458345610508,
      "learning_rate": 0.0001997092248722698,
      "loss": 12.7865,
      "step": 1967
    },
    {
      "epoch": 0.10716546527536705,
      "grad_norm": 0.7697247304481947,
      "learning_rate": 0.00019970855249816798,
      "loss": 12.7212,
      "step": 1968
    },
    {
      "epoch": 0.10721991927195007,
      "grad_norm": 0.8853362692299411,
      "learning_rate": 0.00019970787934871573,
      "loss": 12.7994,
      "step": 1969
    },
    {
      "epoch": 0.10727437326853308,
      "grad_norm": 0.7867428966469842,
      "learning_rate": 0.0001997072054239183,
      "loss": 12.5694,
      "step": 1970
    },
    {
      "epoch": 0.10732882726511608,
      "grad_norm": 0.7358861967218404,
      "learning_rate": 0.0001997065307237809,
      "loss": 12.5993,
      "step": 1971
    },
    {
      "epoch": 0.1073832812616991,
      "grad_norm": 0.7950033939733909,
      "learning_rate": 0.00019970585524830883,
      "loss": 12.6651,
      "step": 1972
    },
    {
      "epoch": 0.10743773525828211,
      "grad_norm": 0.8758062462747864,
      "learning_rate": 0.0001997051789975073,
      "loss": 12.7358,
      "step": 1973
    },
    {
      "epoch": 0.10749218925486513,
      "grad_norm": 0.8597195443725842,
      "learning_rate": 0.00019970450197138155,
      "loss": 12.7349,
      "step": 1974
    },
    {
      "epoch": 0.10754664325144814,
      "grad_norm": 0.790691546135162,
      "learning_rate": 0.00019970382416993688,
      "loss": 12.8114,
      "step": 1975
    },
    {
      "epoch": 0.10760109724803114,
      "grad_norm": 0.9024575737104155,
      "learning_rate": 0.00019970314559317854,
      "loss": 12.8282,
      "step": 1976
    },
    {
      "epoch": 0.10765555124461416,
      "grad_norm": 0.7525079128703486,
      "learning_rate": 0.00019970246624111186,
      "loss": 12.7042,
      "step": 1977
    },
    {
      "epoch": 0.10771000524119717,
      "grad_norm": 0.7830885538690692,
      "learning_rate": 0.00019970178611374207,
      "loss": 12.7363,
      "step": 1978
    },
    {
      "epoch": 0.10776445923778018,
      "grad_norm": 0.7758410027452511,
      "learning_rate": 0.00019970110521107446,
      "loss": 12.711,
      "step": 1979
    },
    {
      "epoch": 0.1078189132343632,
      "grad_norm": 0.786509695907984,
      "learning_rate": 0.00019970042353311434,
      "loss": 12.6893,
      "step": 1980
    },
    {
      "epoch": 0.1078733672309462,
      "grad_norm": 0.7910151871874543,
      "learning_rate": 0.00019969974107986703,
      "loss": 12.6859,
      "step": 1981
    },
    {
      "epoch": 0.10792782122752922,
      "grad_norm": 0.7356011892291389,
      "learning_rate": 0.00019969905785133775,
      "loss": 12.6665,
      "step": 1982
    },
    {
      "epoch": 0.10798227522411223,
      "grad_norm": 0.791857100119298,
      "learning_rate": 0.00019969837384753195,
      "loss": 12.6574,
      "step": 1983
    },
    {
      "epoch": 0.10803672922069524,
      "grad_norm": 0.9453268512645497,
      "learning_rate": 0.00019969768906845484,
      "loss": 12.9026,
      "step": 1984
    },
    {
      "epoch": 0.10809118321727826,
      "grad_norm": 0.7369807898916599,
      "learning_rate": 0.00019969700351411178,
      "loss": 12.4894,
      "step": 1985
    },
    {
      "epoch": 0.10814563721386126,
      "grad_norm": 0.7844522057126512,
      "learning_rate": 0.0001996963171845081,
      "loss": 12.6189,
      "step": 1986
    },
    {
      "epoch": 0.10820009121044427,
      "grad_norm": 0.8845416386732657,
      "learning_rate": 0.00019969563007964913,
      "loss": 12.8518,
      "step": 1987
    },
    {
      "epoch": 0.10825454520702729,
      "grad_norm": 0.8037598314151181,
      "learning_rate": 0.00019969494219954025,
      "loss": 12.7211,
      "step": 1988
    },
    {
      "epoch": 0.1083089992036103,
      "grad_norm": 0.9389458047150615,
      "learning_rate": 0.00019969425354418675,
      "loss": 12.7243,
      "step": 1989
    },
    {
      "epoch": 0.1083634532001933,
      "grad_norm": 0.7320127631185521,
      "learning_rate": 0.00019969356411359405,
      "loss": 12.6032,
      "step": 1990
    },
    {
      "epoch": 0.10841790719677633,
      "grad_norm": 0.8657140221983711,
      "learning_rate": 0.00019969287390776748,
      "loss": 12.7469,
      "step": 1991
    },
    {
      "epoch": 0.10847236119335933,
      "grad_norm": 0.7649625558778893,
      "learning_rate": 0.00019969218292671234,
      "loss": 12.7505,
      "step": 1992
    },
    {
      "epoch": 0.10852681518994235,
      "grad_norm": 0.8960064777973007,
      "learning_rate": 0.00019969149117043413,
      "loss": 12.6676,
      "step": 1993
    },
    {
      "epoch": 0.10858126918652536,
      "grad_norm": 1.017277538626505,
      "learning_rate": 0.00019969079863893817,
      "loss": 12.7488,
      "step": 1994
    },
    {
      "epoch": 0.10863572318310837,
      "grad_norm": 0.8590189883083232,
      "learning_rate": 0.00019969010533222982,
      "loss": 12.742,
      "step": 1995
    },
    {
      "epoch": 0.10869017717969139,
      "grad_norm": 0.8725955485441322,
      "learning_rate": 0.00019968941125031447,
      "loss": 12.7131,
      "step": 1996
    },
    {
      "epoch": 0.10874463117627439,
      "grad_norm": 0.9533012684001186,
      "learning_rate": 0.00019968871639319756,
      "loss": 12.8064,
      "step": 1997
    },
    {
      "epoch": 0.1087990851728574,
      "grad_norm": 0.8515093684993654,
      "learning_rate": 0.0001996880207608845,
      "loss": 12.7675,
      "step": 1998
    },
    {
      "epoch": 0.10885353916944042,
      "grad_norm": 0.9449399025658869,
      "learning_rate": 0.00019968732435338062,
      "loss": 12.7264,
      "step": 1999
    },
    {
      "epoch": 0.10890799316602343,
      "grad_norm": 0.8716562621301317,
      "learning_rate": 0.0001996866271706914,
      "loss": 12.7379,
      "step": 2000
    },
    {
      "epoch": 0.10896244716260645,
      "grad_norm": 0.7605747741023537,
      "learning_rate": 0.00019968592921282228,
      "loss": 12.6486,
      "step": 2001
    },
    {
      "epoch": 0.10901690115918945,
      "grad_norm": 0.8690279429757708,
      "learning_rate": 0.00019968523047977864,
      "loss": 12.727,
      "step": 2002
    },
    {
      "epoch": 0.10907135515577246,
      "grad_norm": 0.7740230628936833,
      "learning_rate": 0.00019968453097156594,
      "loss": 12.7432,
      "step": 2003
    },
    {
      "epoch": 0.10912580915235548,
      "grad_norm": 0.8301607370104841,
      "learning_rate": 0.0001996838306881896,
      "loss": 12.7203,
      "step": 2004
    },
    {
      "epoch": 0.10918026314893849,
      "grad_norm": 0.7939352424523266,
      "learning_rate": 0.00019968312962965508,
      "loss": 12.695,
      "step": 2005
    },
    {
      "epoch": 0.10923471714552149,
      "grad_norm": 0.880346087195917,
      "learning_rate": 0.00019968242779596783,
      "loss": 12.8456,
      "step": 2006
    },
    {
      "epoch": 0.10928917114210451,
      "grad_norm": 0.7813320479992395,
      "learning_rate": 0.00019968172518713327,
      "loss": 12.7121,
      "step": 2007
    },
    {
      "epoch": 0.10934362513868752,
      "grad_norm": 0.8484785526716306,
      "learning_rate": 0.00019968102180315696,
      "loss": 12.7572,
      "step": 2008
    },
    {
      "epoch": 0.10939807913527054,
      "grad_norm": 0.898153371419266,
      "learning_rate": 0.00019968031764404427,
      "loss": 12.7857,
      "step": 2009
    },
    {
      "epoch": 0.10945253313185355,
      "grad_norm": 0.741570262515584,
      "learning_rate": 0.0001996796127098007,
      "loss": 12.8034,
      "step": 2010
    },
    {
      "epoch": 0.10950698712843655,
      "grad_norm": 0.9657599315611324,
      "learning_rate": 0.00019967890700043177,
      "loss": 12.5358,
      "step": 2011
    },
    {
      "epoch": 0.10956144112501957,
      "grad_norm": 0.7940396053085513,
      "learning_rate": 0.00019967820051594294,
      "loss": 12.7363,
      "step": 2012
    },
    {
      "epoch": 0.10961589512160258,
      "grad_norm": 1.0561796034970556,
      "learning_rate": 0.0001996774932563397,
      "loss": 12.7496,
      "step": 2013
    },
    {
      "epoch": 0.10967034911818559,
      "grad_norm": 0.8155241579906948,
      "learning_rate": 0.00019967678522162758,
      "loss": 12.8103,
      "step": 2014
    },
    {
      "epoch": 0.10972480311476861,
      "grad_norm": 0.7664867207293443,
      "learning_rate": 0.00019967607641181205,
      "loss": 12.7256,
      "step": 2015
    },
    {
      "epoch": 0.10977925711135161,
      "grad_norm": 0.7413458849382605,
      "learning_rate": 0.00019967536682689862,
      "loss": 12.6486,
      "step": 2016
    },
    {
      "epoch": 0.10983371110793463,
      "grad_norm": 0.7641870409209854,
      "learning_rate": 0.00019967465646689284,
      "loss": 12.6495,
      "step": 2017
    },
    {
      "epoch": 0.10988816510451764,
      "grad_norm": 0.7952154969764736,
      "learning_rate": 0.0001996739453318002,
      "loss": 12.7041,
      "step": 2018
    },
    {
      "epoch": 0.10994261910110065,
      "grad_norm": 0.8621410230361722,
      "learning_rate": 0.00019967323342162625,
      "loss": 12.7359,
      "step": 2019
    },
    {
      "epoch": 0.10999707309768367,
      "grad_norm": 0.7941226482770086,
      "learning_rate": 0.0001996725207363765,
      "loss": 12.7605,
      "step": 2020
    },
    {
      "epoch": 0.11005152709426667,
      "grad_norm": 0.8875273838164063,
      "learning_rate": 0.00019967180727605656,
      "loss": 12.8796,
      "step": 2021
    },
    {
      "epoch": 0.11010598109084968,
      "grad_norm": 0.9317449387044903,
      "learning_rate": 0.0001996710930406719,
      "loss": 12.7256,
      "step": 2022
    },
    {
      "epoch": 0.1101604350874327,
      "grad_norm": 0.7733791956195418,
      "learning_rate": 0.00019967037803022812,
      "loss": 12.7287,
      "step": 2023
    },
    {
      "epoch": 0.11021488908401571,
      "grad_norm": 0.839253425039756,
      "learning_rate": 0.00019966966224473076,
      "loss": 12.8589,
      "step": 2024
    },
    {
      "epoch": 0.11026934308059873,
      "grad_norm": 0.8383895898304654,
      "learning_rate": 0.0001996689456841854,
      "loss": 12.6426,
      "step": 2025
    },
    {
      "epoch": 0.11032379707718173,
      "grad_norm": 0.80800769816932,
      "learning_rate": 0.00019966822834859759,
      "loss": 12.7299,
      "step": 2026
    },
    {
      "epoch": 0.11037825107376474,
      "grad_norm": 0.9158105841488882,
      "learning_rate": 0.00019966751023797294,
      "loss": 12.8265,
      "step": 2027
    },
    {
      "epoch": 0.11043270507034776,
      "grad_norm": 0.869986832787026,
      "learning_rate": 0.00019966679135231702,
      "loss": 12.7817,
      "step": 2028
    },
    {
      "epoch": 0.11048715906693077,
      "grad_norm": 0.8128647516680227,
      "learning_rate": 0.00019966607169163538,
      "loss": 12.7022,
      "step": 2029
    },
    {
      "epoch": 0.11054161306351377,
      "grad_norm": 0.8051905540730242,
      "learning_rate": 0.00019966535125593368,
      "loss": 12.8388,
      "step": 2030
    },
    {
      "epoch": 0.1105960670600968,
      "grad_norm": 0.7719518429212509,
      "learning_rate": 0.0001996646300452175,
      "loss": 12.7637,
      "step": 2031
    },
    {
      "epoch": 0.1106505210566798,
      "grad_norm": 0.8971123617842752,
      "learning_rate": 0.00019966390805949242,
      "loss": 12.8074,
      "step": 2032
    },
    {
      "epoch": 0.11070497505326282,
      "grad_norm": 0.8091233612763397,
      "learning_rate": 0.0001996631852987641,
      "loss": 12.6037,
      "step": 2033
    },
    {
      "epoch": 0.11075942904984583,
      "grad_norm": 0.8335833189649199,
      "learning_rate": 0.0001996624617630381,
      "loss": 12.5976,
      "step": 2034
    },
    {
      "epoch": 0.11081388304642883,
      "grad_norm": 0.7532461021786356,
      "learning_rate": 0.00019966173745232011,
      "loss": 12.7016,
      "step": 2035
    },
    {
      "epoch": 0.11086833704301186,
      "grad_norm": 0.9038619999472759,
      "learning_rate": 0.00019966101236661575,
      "loss": 12.5796,
      "step": 2036
    },
    {
      "epoch": 0.11092279103959486,
      "grad_norm": 0.9020584325889739,
      "learning_rate": 0.00019966028650593063,
      "loss": 12.8904,
      "step": 2037
    },
    {
      "epoch": 0.11097724503617787,
      "grad_norm": 0.8719156748425685,
      "learning_rate": 0.0001996595598702704,
      "loss": 12.8904,
      "step": 2038
    },
    {
      "epoch": 0.11103169903276089,
      "grad_norm": 0.8031254838524301,
      "learning_rate": 0.0001996588324596407,
      "loss": 12.7358,
      "step": 2039
    },
    {
      "epoch": 0.1110861530293439,
      "grad_norm": 0.8312455018258538,
      "learning_rate": 0.00019965810427404726,
      "loss": 12.7067,
      "step": 2040
    },
    {
      "epoch": 0.11114060702592692,
      "grad_norm": 0.6682438162781495,
      "learning_rate": 0.00019965737531349567,
      "loss": 12.5912,
      "step": 2041
    },
    {
      "epoch": 0.11119506102250992,
      "grad_norm": 0.9103969344133255,
      "learning_rate": 0.00019965664557799163,
      "loss": 12.9311,
      "step": 2042
    },
    {
      "epoch": 0.11124951501909293,
      "grad_norm": 0.7586090916444304,
      "learning_rate": 0.00019965591506754076,
      "loss": 12.7437,
      "step": 2043
    },
    {
      "epoch": 0.11130396901567595,
      "grad_norm": 0.7919965882278415,
      "learning_rate": 0.0001996551837821488,
      "loss": 12.7225,
      "step": 2044
    },
    {
      "epoch": 0.11135842301225896,
      "grad_norm": 0.7519166731231169,
      "learning_rate": 0.00019965445172182142,
      "loss": 12.729,
      "step": 2045
    },
    {
      "epoch": 0.11141287700884196,
      "grad_norm": 0.8124237368088452,
      "learning_rate": 0.0001996537188865643,
      "loss": 12.6507,
      "step": 2046
    },
    {
      "epoch": 0.11146733100542498,
      "grad_norm": 0.8276721559395761,
      "learning_rate": 0.0001996529852763832,
      "loss": 12.849,
      "step": 2047
    },
    {
      "epoch": 0.11152178500200799,
      "grad_norm": 0.7338358319650686,
      "learning_rate": 0.00019965225089128372,
      "loss": 12.7346,
      "step": 2048
    },
    {
      "epoch": 0.11157623899859101,
      "grad_norm": 0.7879128849080722,
      "learning_rate": 0.00019965151573127164,
      "loss": 12.7471,
      "step": 2049
    },
    {
      "epoch": 0.11163069299517402,
      "grad_norm": 0.7036905561446233,
      "learning_rate": 0.00019965077979635268,
      "loss": 12.691,
      "step": 2050
    },
    {
      "epoch": 0.11168514699175702,
      "grad_norm": 0.7659057759800266,
      "learning_rate": 0.00019965004308653253,
      "loss": 12.7796,
      "step": 2051
    },
    {
      "epoch": 0.11173960098834004,
      "grad_norm": 0.9112024972953926,
      "learning_rate": 0.00019964930560181695,
      "loss": 12.6461,
      "step": 2052
    },
    {
      "epoch": 0.11179405498492305,
      "grad_norm": 0.815708476482424,
      "learning_rate": 0.00019964856734221162,
      "loss": 12.662,
      "step": 2053
    },
    {
      "epoch": 0.11184850898150606,
      "grad_norm": 0.7861656973503325,
      "learning_rate": 0.00019964782830772236,
      "loss": 12.6978,
      "step": 2054
    },
    {
      "epoch": 0.11190296297808908,
      "grad_norm": 0.7568981781640535,
      "learning_rate": 0.00019964708849835484,
      "loss": 12.8035,
      "step": 2055
    },
    {
      "epoch": 0.11195741697467208,
      "grad_norm": 0.7885340729467678,
      "learning_rate": 0.00019964634791411488,
      "loss": 12.8172,
      "step": 2056
    },
    {
      "epoch": 0.11201187097125509,
      "grad_norm": 0.7049502508081198,
      "learning_rate": 0.00019964560655500818,
      "loss": 12.6953,
      "step": 2057
    },
    {
      "epoch": 0.11206632496783811,
      "grad_norm": 0.7515432089172787,
      "learning_rate": 0.00019964486442104057,
      "loss": 12.6345,
      "step": 2058
    },
    {
      "epoch": 0.11212077896442112,
      "grad_norm": 0.810079398805017,
      "learning_rate": 0.00019964412151221773,
      "loss": 12.8406,
      "step": 2059
    },
    {
      "epoch": 0.11217523296100414,
      "grad_norm": 0.8140262106072949,
      "learning_rate": 0.00019964337782854555,
      "loss": 12.5698,
      "step": 2060
    },
    {
      "epoch": 0.11222968695758714,
      "grad_norm": 0.7161724218226323,
      "learning_rate": 0.00019964263337002972,
      "loss": 12.6271,
      "step": 2061
    },
    {
      "epoch": 0.11228414095417015,
      "grad_norm": 0.753009588216702,
      "learning_rate": 0.00019964188813667607,
      "loss": 12.6349,
      "step": 2062
    },
    {
      "epoch": 0.11233859495075317,
      "grad_norm": 0.7505899607318048,
      "learning_rate": 0.0001996411421284904,
      "loss": 12.5799,
      "step": 2063
    },
    {
      "epoch": 0.11239304894733618,
      "grad_norm": 0.8278026206975396,
      "learning_rate": 0.00019964039534547847,
      "loss": 12.7407,
      "step": 2064
    },
    {
      "epoch": 0.11244750294391918,
      "grad_norm": 0.8331622863627176,
      "learning_rate": 0.00019963964778764613,
      "loss": 12.6814,
      "step": 2065
    },
    {
      "epoch": 0.1125019569405022,
      "grad_norm": 0.7699158187386606,
      "learning_rate": 0.00019963889945499917,
      "loss": 12.6718,
      "step": 2066
    },
    {
      "epoch": 0.11255641093708521,
      "grad_norm": 0.7747270710988973,
      "learning_rate": 0.00019963815034754344,
      "loss": 12.7746,
      "step": 2067
    },
    {
      "epoch": 0.11261086493366823,
      "grad_norm": 0.7494007434705581,
      "learning_rate": 0.00019963740046528475,
      "loss": 12.6735,
      "step": 2068
    },
    {
      "epoch": 0.11266531893025124,
      "grad_norm": 0.746165912951711,
      "learning_rate": 0.0001996366498082289,
      "loss": 12.766,
      "step": 2069
    },
    {
      "epoch": 0.11271977292683424,
      "grad_norm": 0.8449008773597039,
      "learning_rate": 0.00019963589837638176,
      "loss": 12.7827,
      "step": 2070
    },
    {
      "epoch": 0.11277422692341726,
      "grad_norm": 0.7303376725361176,
      "learning_rate": 0.00019963514616974916,
      "loss": 12.738,
      "step": 2071
    },
    {
      "epoch": 0.11282868092000027,
      "grad_norm": 0.7473221651838649,
      "learning_rate": 0.00019963439318833697,
      "loss": 12.7155,
      "step": 2072
    },
    {
      "epoch": 0.11288313491658328,
      "grad_norm": 0.825100614677872,
      "learning_rate": 0.000199633639432151,
      "loss": 12.7331,
      "step": 2073
    },
    {
      "epoch": 0.1129375889131663,
      "grad_norm": 0.7861781030471263,
      "learning_rate": 0.00019963288490119717,
      "loss": 12.7108,
      "step": 2074
    },
    {
      "epoch": 0.1129920429097493,
      "grad_norm": 0.7526941015138744,
      "learning_rate": 0.0001996321295954813,
      "loss": 12.7691,
      "step": 2075
    },
    {
      "epoch": 0.11304649690633232,
      "grad_norm": 0.8979029163830556,
      "learning_rate": 0.00019963137351500931,
      "loss": 12.6463,
      "step": 2076
    },
    {
      "epoch": 0.11310095090291533,
      "grad_norm": 0.905851466359616,
      "learning_rate": 0.00019963061665978705,
      "loss": 12.8106,
      "step": 2077
    },
    {
      "epoch": 0.11315540489949834,
      "grad_norm": 0.8874248846001883,
      "learning_rate": 0.00019962985902982036,
      "loss": 12.8702,
      "step": 2078
    },
    {
      "epoch": 0.11320985889608136,
      "grad_norm": 0.7718457677154574,
      "learning_rate": 0.00019962910062511525,
      "loss": 12.7201,
      "step": 2079
    },
    {
      "epoch": 0.11326431289266436,
      "grad_norm": 0.7637357437072458,
      "learning_rate": 0.00019962834144567748,
      "loss": 12.7155,
      "step": 2080
    },
    {
      "epoch": 0.11331876688924737,
      "grad_norm": 0.8570310071456682,
      "learning_rate": 0.00019962758149151303,
      "loss": 12.8059,
      "step": 2081
    },
    {
      "epoch": 0.11337322088583039,
      "grad_norm": 0.8012256975464873,
      "learning_rate": 0.00019962682076262781,
      "loss": 12.7262,
      "step": 2082
    },
    {
      "epoch": 0.1134276748824134,
      "grad_norm": 0.7009570353759675,
      "learning_rate": 0.00019962605925902775,
      "loss": 12.6513,
      "step": 2083
    },
    {
      "epoch": 0.11348212887899642,
      "grad_norm": 0.8049734064740511,
      "learning_rate": 0.00019962529698071873,
      "loss": 12.8019,
      "step": 2084
    },
    {
      "epoch": 0.11353658287557943,
      "grad_norm": 0.7857397204885804,
      "learning_rate": 0.00019962453392770668,
      "loss": 12.7982,
      "step": 2085
    },
    {
      "epoch": 0.11359103687216243,
      "grad_norm": 0.7731529643017948,
      "learning_rate": 0.00019962377009999756,
      "loss": 12.7888,
      "step": 2086
    },
    {
      "epoch": 0.11364549086874545,
      "grad_norm": 0.741120737210802,
      "learning_rate": 0.0001996230054975973,
      "loss": 12.5751,
      "step": 2087
    },
    {
      "epoch": 0.11369994486532846,
      "grad_norm": 0.7620596420969351,
      "learning_rate": 0.0001996222401205118,
      "loss": 12.7663,
      "step": 2088
    },
    {
      "epoch": 0.11375439886191147,
      "grad_norm": 0.7492434504656282,
      "learning_rate": 0.0001996214739687471,
      "loss": 12.7384,
      "step": 2089
    },
    {
      "epoch": 0.11380885285849449,
      "grad_norm": 0.7840620448864547,
      "learning_rate": 0.0001996207070423091,
      "loss": 12.6857,
      "step": 2090
    },
    {
      "epoch": 0.11386330685507749,
      "grad_norm": 0.8200841690305539,
      "learning_rate": 0.00019961993934120378,
      "loss": 12.7391,
      "step": 2091
    },
    {
      "epoch": 0.11391776085166051,
      "grad_norm": 0.8040405029865603,
      "learning_rate": 0.0001996191708654371,
      "loss": 12.7379,
      "step": 2092
    },
    {
      "epoch": 0.11397221484824352,
      "grad_norm": 0.7587519143034004,
      "learning_rate": 0.00019961840161501505,
      "loss": 12.6308,
      "step": 2093
    },
    {
      "epoch": 0.11402666884482653,
      "grad_norm": 0.784489810754187,
      "learning_rate": 0.0001996176315899436,
      "loss": 12.7887,
      "step": 2094
    },
    {
      "epoch": 0.11408112284140955,
      "grad_norm": 0.8164944082645404,
      "learning_rate": 0.0001996168607902287,
      "loss": 12.7018,
      "step": 2095
    },
    {
      "epoch": 0.11413557683799255,
      "grad_norm": 0.835692331915501,
      "learning_rate": 0.00019961608921587645,
      "loss": 12.7757,
      "step": 2096
    },
    {
      "epoch": 0.11419003083457556,
      "grad_norm": 0.7558616368210048,
      "learning_rate": 0.00019961531686689274,
      "loss": 12.5786,
      "step": 2097
    },
    {
      "epoch": 0.11424448483115858,
      "grad_norm": 0.7079213163165838,
      "learning_rate": 0.00019961454374328364,
      "loss": 12.6923,
      "step": 2098
    },
    {
      "epoch": 0.11429893882774159,
      "grad_norm": 0.7888926831701923,
      "learning_rate": 0.00019961376984505512,
      "loss": 12.825,
      "step": 2099
    },
    {
      "epoch": 0.1143533928243246,
      "grad_norm": 0.774072981035808,
      "learning_rate": 0.00019961299517221324,
      "loss": 12.7196,
      "step": 2100
    },
    {
      "epoch": 0.11440784682090761,
      "grad_norm": 0.6690168651460398,
      "learning_rate": 0.00019961221972476402,
      "loss": 12.5567,
      "step": 2101
    },
    {
      "epoch": 0.11446230081749062,
      "grad_norm": 0.8531511807742205,
      "learning_rate": 0.00019961144350271343,
      "loss": 12.7074,
      "step": 2102
    },
    {
      "epoch": 0.11451675481407364,
      "grad_norm": 0.7090196312756721,
      "learning_rate": 0.0001996106665060676,
      "loss": 12.6886,
      "step": 2103
    },
    {
      "epoch": 0.11457120881065665,
      "grad_norm": 0.7685376179371318,
      "learning_rate": 0.0001996098887348325,
      "loss": 12.7699,
      "step": 2104
    },
    {
      "epoch": 0.11462566280723965,
      "grad_norm": 0.740224080424988,
      "learning_rate": 0.0001996091101890142,
      "loss": 12.6349,
      "step": 2105
    },
    {
      "epoch": 0.11468011680382267,
      "grad_norm": 0.9454100900329399,
      "learning_rate": 0.00019960833086861873,
      "loss": 12.8987,
      "step": 2106
    },
    {
      "epoch": 0.11473457080040568,
      "grad_norm": 0.7763313314121196,
      "learning_rate": 0.00019960755077365222,
      "loss": 12.6243,
      "step": 2107
    },
    {
      "epoch": 0.1147890247969887,
      "grad_norm": 0.7103990252977854,
      "learning_rate": 0.0001996067699041207,
      "loss": 12.6774,
      "step": 2108
    },
    {
      "epoch": 0.11484347879357171,
      "grad_norm": 0.7405021657321451,
      "learning_rate": 0.00019960598826003018,
      "loss": 12.6477,
      "step": 2109
    },
    {
      "epoch": 0.11489793279015471,
      "grad_norm": 0.8219760829652516,
      "learning_rate": 0.00019960520584138682,
      "loss": 12.7941,
      "step": 2110
    },
    {
      "epoch": 0.11495238678673773,
      "grad_norm": 0.9959298108093985,
      "learning_rate": 0.00019960442264819664,
      "loss": 12.7902,
      "step": 2111
    },
    {
      "epoch": 0.11500684078332074,
      "grad_norm": 0.6913064619941739,
      "learning_rate": 0.00019960363868046582,
      "loss": 12.5739,
      "step": 2112
    },
    {
      "epoch": 0.11506129477990375,
      "grad_norm": 0.7620375931662848,
      "learning_rate": 0.00019960285393820037,
      "loss": 12.653,
      "step": 2113
    },
    {
      "epoch": 0.11511574877648677,
      "grad_norm": 0.8280792459478163,
      "learning_rate": 0.00019960206842140643,
      "loss": 12.6963,
      "step": 2114
    },
    {
      "epoch": 0.11517020277306977,
      "grad_norm": 0.7494683299858886,
      "learning_rate": 0.0001996012821300901,
      "loss": 12.6244,
      "step": 2115
    },
    {
      "epoch": 0.1152246567696528,
      "grad_norm": 0.706202371763292,
      "learning_rate": 0.00019960049506425752,
      "loss": 12.6273,
      "step": 2116
    },
    {
      "epoch": 0.1152791107662358,
      "grad_norm": 0.8221177533629022,
      "learning_rate": 0.00019959970722391478,
      "loss": 12.8283,
      "step": 2117
    },
    {
      "epoch": 0.11533356476281881,
      "grad_norm": 0.831064960973766,
      "learning_rate": 0.00019959891860906799,
      "loss": 12.6878,
      "step": 2118
    },
    {
      "epoch": 0.11538801875940183,
      "grad_norm": 0.881850146847089,
      "learning_rate": 0.00019959812921972332,
      "loss": 12.8005,
      "step": 2119
    },
    {
      "epoch": 0.11544247275598483,
      "grad_norm": 0.8511223229165487,
      "learning_rate": 0.0001995973390558869,
      "loss": 12.7423,
      "step": 2120
    },
    {
      "epoch": 0.11549692675256784,
      "grad_norm": 0.8376433573079606,
      "learning_rate": 0.00019959654811756486,
      "loss": 12.8614,
      "step": 2121
    },
    {
      "epoch": 0.11555138074915086,
      "grad_norm": 0.8510792016650545,
      "learning_rate": 0.00019959575640476334,
      "loss": 12.6251,
      "step": 2122
    },
    {
      "epoch": 0.11560583474573387,
      "grad_norm": 0.8033614865041665,
      "learning_rate": 0.00019959496391748857,
      "loss": 12.7687,
      "step": 2123
    },
    {
      "epoch": 0.11566028874231687,
      "grad_norm": 0.8177322004397564,
      "learning_rate": 0.00019959417065574663,
      "loss": 12.5712,
      "step": 2124
    },
    {
      "epoch": 0.1157147427388999,
      "grad_norm": 0.8300869860612424,
      "learning_rate": 0.00019959337661954368,
      "loss": 12.8096,
      "step": 2125
    },
    {
      "epoch": 0.1157691967354829,
      "grad_norm": 0.7413698447803599,
      "learning_rate": 0.000199592581808886,
      "loss": 12.6082,
      "step": 2126
    },
    {
      "epoch": 0.11582365073206592,
      "grad_norm": 0.7823550575389198,
      "learning_rate": 0.00019959178622377965,
      "loss": 12.6772,
      "step": 2127
    },
    {
      "epoch": 0.11587810472864893,
      "grad_norm": 0.8882803396640111,
      "learning_rate": 0.00019959098986423087,
      "loss": 12.8569,
      "step": 2128
    },
    {
      "epoch": 0.11593255872523194,
      "grad_norm": 0.826671747324549,
      "learning_rate": 0.00019959019273024588,
      "loss": 12.7348,
      "step": 2129
    },
    {
      "epoch": 0.11598701272181496,
      "grad_norm": 0.7841105201498944,
      "learning_rate": 0.00019958939482183084,
      "loss": 12.7031,
      "step": 2130
    },
    {
      "epoch": 0.11604146671839796,
      "grad_norm": 0.7900489427215688,
      "learning_rate": 0.00019958859613899196,
      "loss": 12.5509,
      "step": 2131
    },
    {
      "epoch": 0.11609592071498097,
      "grad_norm": 0.723579613398403,
      "learning_rate": 0.0001995877966817355,
      "loss": 12.6316,
      "step": 2132
    },
    {
      "epoch": 0.11615037471156399,
      "grad_norm": 1.030053378973563,
      "learning_rate": 0.00019958699645006758,
      "loss": 12.9159,
      "step": 2133
    },
    {
      "epoch": 0.116204828708147,
      "grad_norm": 0.857899856222025,
      "learning_rate": 0.00019958619544399445,
      "loss": 12.744,
      "step": 2134
    },
    {
      "epoch": 0.11625928270473002,
      "grad_norm": 0.7480106693179064,
      "learning_rate": 0.00019958539366352241,
      "loss": 12.8448,
      "step": 2135
    },
    {
      "epoch": 0.11631373670131302,
      "grad_norm": 0.7751551904107694,
      "learning_rate": 0.00019958459110865765,
      "loss": 12.6368,
      "step": 2136
    },
    {
      "epoch": 0.11636819069789603,
      "grad_norm": 0.7603853151602749,
      "learning_rate": 0.0001995837877794064,
      "loss": 12.7677,
      "step": 2137
    },
    {
      "epoch": 0.11642264469447905,
      "grad_norm": 0.881158263097917,
      "learning_rate": 0.0001995829836757749,
      "loss": 12.7606,
      "step": 2138
    },
    {
      "epoch": 0.11647709869106206,
      "grad_norm": 0.7802242944553751,
      "learning_rate": 0.00019958217879776944,
      "loss": 12.6844,
      "step": 2139
    },
    {
      "epoch": 0.11653155268764506,
      "grad_norm": 0.9848652595670689,
      "learning_rate": 0.00019958137314539625,
      "loss": 12.5426,
      "step": 2140
    },
    {
      "epoch": 0.11658600668422808,
      "grad_norm": 0.7941566177687074,
      "learning_rate": 0.00019958056671866162,
      "loss": 12.8575,
      "step": 2141
    },
    {
      "epoch": 0.11664046068081109,
      "grad_norm": 0.7449880770493477,
      "learning_rate": 0.00019957975951757177,
      "loss": 12.6769,
      "step": 2142
    },
    {
      "epoch": 0.11669491467739411,
      "grad_norm": 0.7723677676886868,
      "learning_rate": 0.00019957895154213302,
      "loss": 12.6147,
      "step": 2143
    },
    {
      "epoch": 0.11674936867397712,
      "grad_norm": 0.7912735436279764,
      "learning_rate": 0.00019957814279235165,
      "loss": 12.6037,
      "step": 2144
    },
    {
      "epoch": 0.11680382267056012,
      "grad_norm": 0.7972391102056795,
      "learning_rate": 0.00019957733326823394,
      "loss": 12.7607,
      "step": 2145
    },
    {
      "epoch": 0.11685827666714314,
      "grad_norm": 0.8313684183370367,
      "learning_rate": 0.00019957652296978618,
      "loss": 12.6734,
      "step": 2146
    },
    {
      "epoch": 0.11691273066372615,
      "grad_norm": 0.792697041691428,
      "learning_rate": 0.00019957571189701469,
      "loss": 12.5648,
      "step": 2147
    },
    {
      "epoch": 0.11696718466030916,
      "grad_norm": 0.9063581429753217,
      "learning_rate": 0.00019957490004992575,
      "loss": 12.7155,
      "step": 2148
    },
    {
      "epoch": 0.11702163865689218,
      "grad_norm": 0.8434538546620092,
      "learning_rate": 0.00019957408742852573,
      "loss": 12.6305,
      "step": 2149
    },
    {
      "epoch": 0.11707609265347518,
      "grad_norm": 0.87158769555442,
      "learning_rate": 0.00019957327403282088,
      "loss": 12.6689,
      "step": 2150
    },
    {
      "epoch": 0.1171305466500582,
      "grad_norm": 0.8126919819323308,
      "learning_rate": 0.00019957245986281755,
      "loss": 12.6945,
      "step": 2151
    },
    {
      "epoch": 0.11718500064664121,
      "grad_norm": 0.8233288581078374,
      "learning_rate": 0.00019957164491852207,
      "loss": 12.7854,
      "step": 2152
    },
    {
      "epoch": 0.11723945464322422,
      "grad_norm": 0.7885447903071905,
      "learning_rate": 0.0001995708291999408,
      "loss": 12.5977,
      "step": 2153
    },
    {
      "epoch": 0.11729390863980724,
      "grad_norm": 0.8239947426294945,
      "learning_rate": 0.00019957001270708003,
      "loss": 12.6609,
      "step": 2154
    },
    {
      "epoch": 0.11734836263639024,
      "grad_norm": 0.7073707699017527,
      "learning_rate": 0.00019956919543994615,
      "loss": 12.6269,
      "step": 2155
    },
    {
      "epoch": 0.11740281663297325,
      "grad_norm": 0.8425301365131279,
      "learning_rate": 0.00019956837739854556,
      "loss": 12.8125,
      "step": 2156
    },
    {
      "epoch": 0.11745727062955627,
      "grad_norm": 0.7271759383651856,
      "learning_rate": 0.0001995675585828845,
      "loss": 12.6695,
      "step": 2157
    },
    {
      "epoch": 0.11751172462613928,
      "grad_norm": 0.8336817554060225,
      "learning_rate": 0.00019956673899296944,
      "loss": 12.7262,
      "step": 2158
    },
    {
      "epoch": 0.1175661786227223,
      "grad_norm": 0.8076930572191388,
      "learning_rate": 0.00019956591862880675,
      "loss": 12.7619,
      "step": 2159
    },
    {
      "epoch": 0.1176206326193053,
      "grad_norm": 0.7989577147925232,
      "learning_rate": 0.00019956509749040273,
      "loss": 12.6479,
      "step": 2160
    },
    {
      "epoch": 0.11767508661588831,
      "grad_norm": 0.8037586122289786,
      "learning_rate": 0.00019956427557776384,
      "loss": 12.6592,
      "step": 2161
    },
    {
      "epoch": 0.11772954061247133,
      "grad_norm": 0.7611174663838151,
      "learning_rate": 0.00019956345289089643,
      "loss": 12.6262,
      "step": 2162
    },
    {
      "epoch": 0.11778399460905434,
      "grad_norm": 0.8435221901079686,
      "learning_rate": 0.0001995626294298069,
      "loss": 12.72,
      "step": 2163
    },
    {
      "epoch": 0.11783844860563734,
      "grad_norm": 0.8431094578143535,
      "learning_rate": 0.00019956180519450171,
      "loss": 12.7281,
      "step": 2164
    },
    {
      "epoch": 0.11789290260222036,
      "grad_norm": 1.0147865888423293,
      "learning_rate": 0.00019956098018498723,
      "loss": 12.7111,
      "step": 2165
    },
    {
      "epoch": 0.11794735659880337,
      "grad_norm": 0.7481049960190732,
      "learning_rate": 0.00019956015440126984,
      "loss": 12.751,
      "step": 2166
    },
    {
      "epoch": 0.11800181059538639,
      "grad_norm": 1.0240643900963289,
      "learning_rate": 0.000199559327843356,
      "loss": 12.8251,
      "step": 2167
    },
    {
      "epoch": 0.1180562645919694,
      "grad_norm": 0.8048856773837297,
      "learning_rate": 0.00019955850051125214,
      "loss": 12.7436,
      "step": 2168
    },
    {
      "epoch": 0.1181107185885524,
      "grad_norm": 0.7799355673998167,
      "learning_rate": 0.0001995576724049647,
      "loss": 12.7746,
      "step": 2169
    },
    {
      "epoch": 0.11816517258513543,
      "grad_norm": 0.898073428935087,
      "learning_rate": 0.00019955684352450007,
      "loss": 12.6445,
      "step": 2170
    },
    {
      "epoch": 0.11821962658171843,
      "grad_norm": 0.8455278039776454,
      "learning_rate": 0.00019955601386986471,
      "loss": 12.9123,
      "step": 2171
    },
    {
      "epoch": 0.11827408057830144,
      "grad_norm": 0.8917322796775545,
      "learning_rate": 0.00019955518344106512,
      "loss": 12.63,
      "step": 2172
    },
    {
      "epoch": 0.11832853457488446,
      "grad_norm": 0.8617661704418199,
      "learning_rate": 0.00019955435223810772,
      "loss": 12.8358,
      "step": 2173
    },
    {
      "epoch": 0.11838298857146747,
      "grad_norm": 0.915808254151994,
      "learning_rate": 0.00019955352026099901,
      "loss": 12.5354,
      "step": 2174
    },
    {
      "epoch": 0.11843744256805049,
      "grad_norm": 1.055264887510652,
      "learning_rate": 0.0001995526875097454,
      "loss": 12.6227,
      "step": 2175
    },
    {
      "epoch": 0.11849189656463349,
      "grad_norm": 0.9186264289083459,
      "learning_rate": 0.0001995518539843534,
      "loss": 12.617,
      "step": 2176
    },
    {
      "epoch": 0.1185463505612165,
      "grad_norm": 1.0774450319609858,
      "learning_rate": 0.0001995510196848295,
      "loss": 12.8788,
      "step": 2177
    },
    {
      "epoch": 0.11860080455779952,
      "grad_norm": 0.7694483552357013,
      "learning_rate": 0.00019955018461118018,
      "loss": 12.7367,
      "step": 2178
    },
    {
      "epoch": 0.11865525855438253,
      "grad_norm": 1.2952253320732638,
      "learning_rate": 0.0001995493487634119,
      "loss": 12.7683,
      "step": 2179
    },
    {
      "epoch": 0.11870971255096553,
      "grad_norm": 0.8391668405024848,
      "learning_rate": 0.00019954851214153124,
      "loss": 12.5749,
      "step": 2180
    },
    {
      "epoch": 0.11876416654754855,
      "grad_norm": 0.9121798107531427,
      "learning_rate": 0.0001995476747455446,
      "loss": 12.7191,
      "step": 2181
    },
    {
      "epoch": 0.11881862054413156,
      "grad_norm": 0.8153326608698361,
      "learning_rate": 0.0001995468365754586,
      "loss": 12.6872,
      "step": 2182
    },
    {
      "epoch": 0.11887307454071458,
      "grad_norm": 0.7900303177250834,
      "learning_rate": 0.00019954599763127967,
      "loss": 12.6046,
      "step": 2183
    },
    {
      "epoch": 0.11892752853729759,
      "grad_norm": 0.7569958139616827,
      "learning_rate": 0.00019954515791301437,
      "loss": 12.6619,
      "step": 2184
    },
    {
      "epoch": 0.11898198253388059,
      "grad_norm": 0.8371377500237001,
      "learning_rate": 0.00019954431742066927,
      "loss": 12.6206,
      "step": 2185
    },
    {
      "epoch": 0.11903643653046361,
      "grad_norm": 0.853149233833002,
      "learning_rate": 0.00019954347615425082,
      "loss": 12.6786,
      "step": 2186
    },
    {
      "epoch": 0.11909089052704662,
      "grad_norm": 0.7817696062905213,
      "learning_rate": 0.0001995426341137656,
      "loss": 12.7055,
      "step": 2187
    },
    {
      "epoch": 0.11914534452362963,
      "grad_norm": 0.7871792577118936,
      "learning_rate": 0.00019954179129922018,
      "loss": 12.6681,
      "step": 2188
    },
    {
      "epoch": 0.11919979852021265,
      "grad_norm": 0.8416226746173138,
      "learning_rate": 0.0001995409477106211,
      "loss": 12.8002,
      "step": 2189
    },
    {
      "epoch": 0.11925425251679565,
      "grad_norm": 0.7951692811202323,
      "learning_rate": 0.0001995401033479749,
      "loss": 12.8236,
      "step": 2190
    },
    {
      "epoch": 0.11930870651337866,
      "grad_norm": 0.8553946427982531,
      "learning_rate": 0.0001995392582112882,
      "loss": 12.6695,
      "step": 2191
    },
    {
      "epoch": 0.11936316050996168,
      "grad_norm": 0.7860369946268659,
      "learning_rate": 0.00019953841230056752,
      "loss": 12.5614,
      "step": 2192
    },
    {
      "epoch": 0.11941761450654469,
      "grad_norm": 0.7248201454741638,
      "learning_rate": 0.0001995375656158194,
      "loss": 12.5566,
      "step": 2193
    },
    {
      "epoch": 0.1194720685031277,
      "grad_norm": 0.7820765623167737,
      "learning_rate": 0.00019953671815705056,
      "loss": 12.7546,
      "step": 2194
    },
    {
      "epoch": 0.11952652249971071,
      "grad_norm": 0.7850126988810134,
      "learning_rate": 0.00019953586992426747,
      "loss": 12.589,
      "step": 2195
    },
    {
      "epoch": 0.11958097649629372,
      "grad_norm": 0.7990527946487643,
      "learning_rate": 0.00019953502091747677,
      "loss": 12.6696,
      "step": 2196
    },
    {
      "epoch": 0.11963543049287674,
      "grad_norm": 0.8866117370502644,
      "learning_rate": 0.00019953417113668505,
      "loss": 12.8036,
      "step": 2197
    },
    {
      "epoch": 0.11968988448945975,
      "grad_norm": 0.7082177203292337,
      "learning_rate": 0.00019953332058189892,
      "loss": 12.6838,
      "step": 2198
    },
    {
      "epoch": 0.11974433848604275,
      "grad_norm": 0.8070197457949011,
      "learning_rate": 0.000199532469253125,
      "loss": 12.7203,
      "step": 2199
    },
    {
      "epoch": 0.11979879248262577,
      "grad_norm": 0.8369868055876668,
      "learning_rate": 0.00019953161715036992,
      "loss": 12.6899,
      "step": 2200
    },
    {
      "epoch": 0.11985324647920878,
      "grad_norm": 0.8999597921541851,
      "learning_rate": 0.0001995307642736403,
      "loss": 12.6945,
      "step": 2201
    },
    {
      "epoch": 0.1199077004757918,
      "grad_norm": 0.7692650720140853,
      "learning_rate": 0.00019952991062294274,
      "loss": 12.7636,
      "step": 2202
    },
    {
      "epoch": 0.11996215447237481,
      "grad_norm": 0.8510153607320751,
      "learning_rate": 0.0001995290561982839,
      "loss": 12.7078,
      "step": 2203
    },
    {
      "epoch": 0.12001660846895781,
      "grad_norm": 0.783743892081945,
      "learning_rate": 0.00019952820099967044,
      "loss": 12.7044,
      "step": 2204
    },
    {
      "epoch": 0.12007106246554083,
      "grad_norm": 0.7653316210747961,
      "learning_rate": 0.000199527345027109,
      "loss": 12.7126,
      "step": 2205
    },
    {
      "epoch": 0.12012551646212384,
      "grad_norm": 0.8536028075609401,
      "learning_rate": 0.00019952648828060622,
      "loss": 12.7369,
      "step": 2206
    },
    {
      "epoch": 0.12017997045870685,
      "grad_norm": 0.7582973421226827,
      "learning_rate": 0.0001995256307601688,
      "loss": 12.7263,
      "step": 2207
    },
    {
      "epoch": 0.12023442445528987,
      "grad_norm": 0.9344165946344044,
      "learning_rate": 0.00019952477246580337,
      "loss": 12.6734,
      "step": 2208
    },
    {
      "epoch": 0.12028887845187287,
      "grad_norm": 0.7814808828459261,
      "learning_rate": 0.00019952391339751665,
      "loss": 12.7027,
      "step": 2209
    },
    {
      "epoch": 0.1203433324484559,
      "grad_norm": 0.8403333966762496,
      "learning_rate": 0.00019952305355531525,
      "loss": 12.7408,
      "step": 2210
    },
    {
      "epoch": 0.1203977864450389,
      "grad_norm": 0.8319704410053225,
      "learning_rate": 0.0001995221929392059,
      "loss": 12.7889,
      "step": 2211
    },
    {
      "epoch": 0.12045224044162191,
      "grad_norm": 0.7947473187321004,
      "learning_rate": 0.00019952133154919527,
      "loss": 12.7497,
      "step": 2212
    },
    {
      "epoch": 0.12050669443820493,
      "grad_norm": 0.7839426718385665,
      "learning_rate": 0.00019952046938529012,
      "loss": 12.7165,
      "step": 2213
    },
    {
      "epoch": 0.12056114843478793,
      "grad_norm": 0.7638925777090524,
      "learning_rate": 0.00019951960644749706,
      "loss": 12.6444,
      "step": 2214
    },
    {
      "epoch": 0.12061560243137094,
      "grad_norm": 0.7897163234635378,
      "learning_rate": 0.00019951874273582291,
      "loss": 12.6192,
      "step": 2215
    },
    {
      "epoch": 0.12067005642795396,
      "grad_norm": 0.7563778420238759,
      "learning_rate": 0.0001995178782502743,
      "loss": 12.6769,
      "step": 2216
    },
    {
      "epoch": 0.12072451042453697,
      "grad_norm": 0.9390704289738171,
      "learning_rate": 0.00019951701299085798,
      "loss": 12.8826,
      "step": 2217
    },
    {
      "epoch": 0.12077896442111999,
      "grad_norm": 0.8678848428558147,
      "learning_rate": 0.00019951614695758069,
      "loss": 12.8051,
      "step": 2218
    },
    {
      "epoch": 0.120833418417703,
      "grad_norm": 0.9110099563247891,
      "learning_rate": 0.00019951528015044913,
      "loss": 12.7489,
      "step": 2219
    },
    {
      "epoch": 0.120887872414286,
      "grad_norm": 0.8804716108031068,
      "learning_rate": 0.00019951441256947007,
      "loss": 12.6187,
      "step": 2220
    },
    {
      "epoch": 0.12094232641086902,
      "grad_norm": 0.7909037529558258,
      "learning_rate": 0.00019951354421465024,
      "loss": 12.688,
      "step": 2221
    },
    {
      "epoch": 0.12099678040745203,
      "grad_norm": 0.8543656619867279,
      "learning_rate": 0.00019951267508599643,
      "loss": 12.7867,
      "step": 2222
    },
    {
      "epoch": 0.12105123440403504,
      "grad_norm": 0.6982517176985668,
      "learning_rate": 0.00019951180518351536,
      "loss": 12.6393,
      "step": 2223
    },
    {
      "epoch": 0.12110568840061806,
      "grad_norm": 0.9337327935082366,
      "learning_rate": 0.0001995109345072138,
      "loss": 12.8072,
      "step": 2224
    },
    {
      "epoch": 0.12116014239720106,
      "grad_norm": 0.9208777629968543,
      "learning_rate": 0.00019951006305709852,
      "loss": 12.6814,
      "step": 2225
    },
    {
      "epoch": 0.12121459639378408,
      "grad_norm": 0.7661250813181665,
      "learning_rate": 0.0001995091908331763,
      "loss": 12.7313,
      "step": 2226
    },
    {
      "epoch": 0.12126905039036709,
      "grad_norm": 0.8286799047391251,
      "learning_rate": 0.00019950831783545393,
      "loss": 12.6088,
      "step": 2227
    },
    {
      "epoch": 0.1213235043869501,
      "grad_norm": 0.9007022814790493,
      "learning_rate": 0.00019950744406393818,
      "loss": 12.8683,
      "step": 2228
    },
    {
      "epoch": 0.12137795838353312,
      "grad_norm": 0.7584611212657733,
      "learning_rate": 0.00019950656951863588,
      "loss": 12.6202,
      "step": 2229
    },
    {
      "epoch": 0.12143241238011612,
      "grad_norm": 0.852208382943685,
      "learning_rate": 0.0001995056941995538,
      "loss": 12.8656,
      "step": 2230
    },
    {
      "epoch": 0.12148686637669913,
      "grad_norm": 0.6967002768684686,
      "learning_rate": 0.00019950481810669874,
      "loss": 12.6892,
      "step": 2231
    },
    {
      "epoch": 0.12154132037328215,
      "grad_norm": 0.8661162169371703,
      "learning_rate": 0.00019950394124007757,
      "loss": 12.5622,
      "step": 2232
    },
    {
      "epoch": 0.12159577436986516,
      "grad_norm": 0.7339221495289832,
      "learning_rate": 0.000199503063599697,
      "loss": 12.6426,
      "step": 2233
    },
    {
      "epoch": 0.12165022836644818,
      "grad_norm": 0.8185295105254345,
      "learning_rate": 0.000199502185185564,
      "loss": 12.76,
      "step": 2234
    },
    {
      "epoch": 0.12170468236303118,
      "grad_norm": 0.7841512755708057,
      "learning_rate": 0.00019950130599768527,
      "loss": 12.8623,
      "step": 2235
    },
    {
      "epoch": 0.12175913635961419,
      "grad_norm": 0.8584145810028976,
      "learning_rate": 0.0001995004260360677,
      "loss": 12.7982,
      "step": 2236
    },
    {
      "epoch": 0.12181359035619721,
      "grad_norm": 0.8331097929496447,
      "learning_rate": 0.00019949954530071814,
      "loss": 12.7162,
      "step": 2237
    },
    {
      "epoch": 0.12186804435278022,
      "grad_norm": 0.7656528692633603,
      "learning_rate": 0.00019949866379164344,
      "loss": 12.6292,
      "step": 2238
    },
    {
      "epoch": 0.12192249834936322,
      "grad_norm": 0.8247940047932069,
      "learning_rate": 0.00019949778150885042,
      "loss": 12.5262,
      "step": 2239
    },
    {
      "epoch": 0.12197695234594624,
      "grad_norm": 0.7842828299270601,
      "learning_rate": 0.00019949689845234598,
      "loss": 12.7852,
      "step": 2240
    },
    {
      "epoch": 0.12203140634252925,
      "grad_norm": 0.8133244862462937,
      "learning_rate": 0.00019949601462213696,
      "loss": 12.7031,
      "step": 2241
    },
    {
      "epoch": 0.12208586033911227,
      "grad_norm": 0.713675822257346,
      "learning_rate": 0.00019949513001823026,
      "loss": 12.592,
      "step": 2242
    },
    {
      "epoch": 0.12214031433569528,
      "grad_norm": 0.8027291483722292,
      "learning_rate": 0.00019949424464063274,
      "loss": 12.7108,
      "step": 2243
    },
    {
      "epoch": 0.12219476833227828,
      "grad_norm": 0.8099594121442285,
      "learning_rate": 0.0001994933584893513,
      "loss": 12.5928,
      "step": 2244
    },
    {
      "epoch": 0.1222492223288613,
      "grad_norm": 0.772771167685069,
      "learning_rate": 0.0001994924715643928,
      "loss": 12.5073,
      "step": 2245
    },
    {
      "epoch": 0.12230367632544431,
      "grad_norm": 0.7813372535261173,
      "learning_rate": 0.00019949158386576413,
      "loss": 12.4561,
      "step": 2246
    },
    {
      "epoch": 0.12235813032202732,
      "grad_norm": 0.789388708629161,
      "learning_rate": 0.00019949069539347227,
      "loss": 12.7137,
      "step": 2247
    },
    {
      "epoch": 0.12241258431861034,
      "grad_norm": 0.7318944286969588,
      "learning_rate": 0.00019948980614752405,
      "loss": 12.5061,
      "step": 2248
    },
    {
      "epoch": 0.12246703831519334,
      "grad_norm": 0.8059878366303711,
      "learning_rate": 0.00019948891612792645,
      "loss": 12.6272,
      "step": 2249
    },
    {
      "epoch": 0.12252149231177636,
      "grad_norm": 0.8177049921666123,
      "learning_rate": 0.0001994880253346863,
      "loss": 12.6948,
      "step": 2250
    },
    {
      "epoch": 0.12257594630835937,
      "grad_norm": 0.760296658253682,
      "learning_rate": 0.00019948713376781064,
      "loss": 12.6752,
      "step": 2251
    },
    {
      "epoch": 0.12263040030494238,
      "grad_norm": 0.781897864488266,
      "learning_rate": 0.0001994862414273063,
      "loss": 12.6803,
      "step": 2252
    },
    {
      "epoch": 0.1226848543015254,
      "grad_norm": 0.7677497443346096,
      "learning_rate": 0.00019948534831318025,
      "loss": 12.7446,
      "step": 2253
    },
    {
      "epoch": 0.1227393082981084,
      "grad_norm": 0.7855649193744076,
      "learning_rate": 0.0001994844544254395,
      "loss": 12.7643,
      "step": 2254
    },
    {
      "epoch": 0.12279376229469141,
      "grad_norm": 0.8211053950537586,
      "learning_rate": 0.0001994835597640909,
      "loss": 12.7967,
      "step": 2255
    },
    {
      "epoch": 0.12284821629127443,
      "grad_norm": 0.8089345545600491,
      "learning_rate": 0.00019948266432914146,
      "loss": 12.8287,
      "step": 2256
    },
    {
      "epoch": 0.12290267028785744,
      "grad_norm": 0.9045745536594514,
      "learning_rate": 0.00019948176812059818,
      "loss": 12.7589,
      "step": 2257
    },
    {
      "epoch": 0.12295712428444044,
      "grad_norm": 0.8362285356153368,
      "learning_rate": 0.00019948087113846796,
      "loss": 12.9202,
      "step": 2258
    },
    {
      "epoch": 0.12301157828102346,
      "grad_norm": 0.7167029837106066,
      "learning_rate": 0.0001994799733827578,
      "loss": 12.6848,
      "step": 2259
    },
    {
      "epoch": 0.12306603227760647,
      "grad_norm": 0.9069713997609858,
      "learning_rate": 0.0001994790748534747,
      "loss": 12.6965,
      "step": 2260
    },
    {
      "epoch": 0.12312048627418949,
      "grad_norm": 0.8544585569223123,
      "learning_rate": 0.00019947817555062563,
      "loss": 12.774,
      "step": 2261
    },
    {
      "epoch": 0.1231749402707725,
      "grad_norm": 0.7448066164304611,
      "learning_rate": 0.00019947727547421756,
      "loss": 12.6796,
      "step": 2262
    },
    {
      "epoch": 0.1232293942673555,
      "grad_norm": 0.7350628811540768,
      "learning_rate": 0.00019947637462425753,
      "loss": 12.7487,
      "step": 2263
    },
    {
      "epoch": 0.12328384826393853,
      "grad_norm": 0.8324612782436918,
      "learning_rate": 0.00019947547300075254,
      "loss": 12.6417,
      "step": 2264
    },
    {
      "epoch": 0.12333830226052153,
      "grad_norm": 0.8118064750379741,
      "learning_rate": 0.00019947457060370957,
      "loss": 12.6985,
      "step": 2265
    },
    {
      "epoch": 0.12339275625710454,
      "grad_norm": 0.7497361623227577,
      "learning_rate": 0.00019947366743313568,
      "loss": 12.6324,
      "step": 2266
    },
    {
      "epoch": 0.12344721025368756,
      "grad_norm": 0.8414180110506619,
      "learning_rate": 0.00019947276348903782,
      "loss": 12.6279,
      "step": 2267
    },
    {
      "epoch": 0.12350166425027057,
      "grad_norm": 0.7324293159778035,
      "learning_rate": 0.00019947185877142314,
      "loss": 12.7224,
      "step": 2268
    },
    {
      "epoch": 0.12355611824685359,
      "grad_norm": 0.8854429184534627,
      "learning_rate": 0.00019947095328029857,
      "loss": 12.7163,
      "step": 2269
    },
    {
      "epoch": 0.12361057224343659,
      "grad_norm": 0.9119324234099492,
      "learning_rate": 0.00019947004701567118,
      "loss": 12.8091,
      "step": 2270
    },
    {
      "epoch": 0.1236650262400196,
      "grad_norm": 0.8905496427432454,
      "learning_rate": 0.00019946913997754802,
      "loss": 12.7258,
      "step": 2271
    },
    {
      "epoch": 0.12371948023660262,
      "grad_norm": 0.9118610556427518,
      "learning_rate": 0.00019946823216593614,
      "loss": 12.7396,
      "step": 2272
    },
    {
      "epoch": 0.12377393423318563,
      "grad_norm": 0.7735477988456516,
      "learning_rate": 0.00019946732358084264,
      "loss": 12.8204,
      "step": 2273
    },
    {
      "epoch": 0.12382838822976863,
      "grad_norm": 0.9140441735532921,
      "learning_rate": 0.00019946641422227454,
      "loss": 12.6388,
      "step": 2274
    },
    {
      "epoch": 0.12388284222635165,
      "grad_norm": 0.7906245809149918,
      "learning_rate": 0.00019946550409023889,
      "loss": 12.6582,
      "step": 2275
    },
    {
      "epoch": 0.12393729622293466,
      "grad_norm": 0.9227199554541721,
      "learning_rate": 0.00019946459318474283,
      "loss": 12.7565,
      "step": 2276
    },
    {
      "epoch": 0.12399175021951768,
      "grad_norm": 0.8634134768192193,
      "learning_rate": 0.00019946368150579343,
      "loss": 12.7553,
      "step": 2277
    },
    {
      "epoch": 0.12404620421610069,
      "grad_norm": 0.8388803922129331,
      "learning_rate": 0.00019946276905339773,
      "loss": 12.872,
      "step": 2278
    },
    {
      "epoch": 0.12410065821268369,
      "grad_norm": 0.8528527510163847,
      "learning_rate": 0.00019946185582756288,
      "loss": 12.8716,
      "step": 2279
    },
    {
      "epoch": 0.12415511220926671,
      "grad_norm": 0.9367156268725985,
      "learning_rate": 0.00019946094182829595,
      "loss": 12.7734,
      "step": 2280
    },
    {
      "epoch": 0.12420956620584972,
      "grad_norm": 0.7778224462800217,
      "learning_rate": 0.00019946002705560406,
      "loss": 12.6889,
      "step": 2281
    },
    {
      "epoch": 0.12426402020243273,
      "grad_norm": 0.7491322077833296,
      "learning_rate": 0.0001994591115094943,
      "loss": 12.6262,
      "step": 2282
    },
    {
      "epoch": 0.12431847419901575,
      "grad_norm": 0.8402926697141994,
      "learning_rate": 0.00019945819518997384,
      "loss": 12.7234,
      "step": 2283
    },
    {
      "epoch": 0.12437292819559875,
      "grad_norm": 0.8636335512953711,
      "learning_rate": 0.00019945727809704975,
      "loss": 12.608,
      "step": 2284
    },
    {
      "epoch": 0.12442738219218177,
      "grad_norm": 0.7371257107997498,
      "learning_rate": 0.0001994563602307292,
      "loss": 12.5674,
      "step": 2285
    },
    {
      "epoch": 0.12448183618876478,
      "grad_norm": 0.7708748365298516,
      "learning_rate": 0.0001994554415910193,
      "loss": 12.6949,
      "step": 2286
    },
    {
      "epoch": 0.12453629018534779,
      "grad_norm": 0.8447973867023547,
      "learning_rate": 0.0001994545221779272,
      "loss": 12.688,
      "step": 2287
    },
    {
      "epoch": 0.12459074418193081,
      "grad_norm": 1.108122665821199,
      "learning_rate": 0.0001994536019914601,
      "loss": 12.9816,
      "step": 2288
    },
    {
      "epoch": 0.12464519817851381,
      "grad_norm": 0.8541583693169076,
      "learning_rate": 0.00019945268103162506,
      "loss": 12.6994,
      "step": 2289
    },
    {
      "epoch": 0.12469965217509682,
      "grad_norm": 0.8429082152082202,
      "learning_rate": 0.00019945175929842935,
      "loss": 12.7596,
      "step": 2290
    },
    {
      "epoch": 0.12475410617167984,
      "grad_norm": 1.0515128515445635,
      "learning_rate": 0.00019945083679188006,
      "loss": 12.8312,
      "step": 2291
    },
    {
      "epoch": 0.12480856016826285,
      "grad_norm": 0.8531126268343717,
      "learning_rate": 0.00019944991351198435,
      "loss": 12.7377,
      "step": 2292
    },
    {
      "epoch": 0.12486301416484587,
      "grad_norm": 0.7856715624294648,
      "learning_rate": 0.0001994489894587495,
      "loss": 12.6877,
      "step": 2293
    },
    {
      "epoch": 0.12491746816142887,
      "grad_norm": 1.0596739665009962,
      "learning_rate": 0.00019944806463218257,
      "loss": 12.6832,
      "step": 2294
    },
    {
      "epoch": 0.12497192215801188,
      "grad_norm": 0.8027737513539224,
      "learning_rate": 0.0001994471390322908,
      "loss": 12.7159,
      "step": 2295
    },
    {
      "epoch": 0.1250263761545949,
      "grad_norm": 0.9906726771062633,
      "learning_rate": 0.00019944621265908147,
      "loss": 12.6231,
      "step": 2296
    },
    {
      "epoch": 0.1250808301511779,
      "grad_norm": 0.8841377858178393,
      "learning_rate": 0.00019944528551256167,
      "loss": 12.8105,
      "step": 2297
    },
    {
      "epoch": 0.12513528414776093,
      "grad_norm": 0.9171916869214579,
      "learning_rate": 0.00019944435759273866,
      "loss": 12.7818,
      "step": 2298
    },
    {
      "epoch": 0.12518973814434392,
      "grad_norm": 0.8517112317085067,
      "learning_rate": 0.00019944342889961964,
      "loss": 12.6482,
      "step": 2299
    },
    {
      "epoch": 0.12524419214092694,
      "grad_norm": 0.8005258077612063,
      "learning_rate": 0.00019944249943321186,
      "loss": 12.7086,
      "step": 2300
    },
    {
      "epoch": 0.12529864613750996,
      "grad_norm": 0.7865738100260907,
      "learning_rate": 0.00019944156919352252,
      "loss": 12.6759,
      "step": 2301
    },
    {
      "epoch": 0.12535310013409295,
      "grad_norm": 0.8039482381347336,
      "learning_rate": 0.00019944063818055888,
      "loss": 12.6076,
      "step": 2302
    },
    {
      "epoch": 0.12540755413067597,
      "grad_norm": 0.7764854255852931,
      "learning_rate": 0.0001994397063943281,
      "loss": 12.7396,
      "step": 2303
    },
    {
      "epoch": 0.125462008127259,
      "grad_norm": 0.837158284917924,
      "learning_rate": 0.00019943877383483757,
      "loss": 12.5878,
      "step": 2304
    },
    {
      "epoch": 0.12551646212384202,
      "grad_norm": 0.8371200249719656,
      "learning_rate": 0.0001994378405020944,
      "loss": 12.7985,
      "step": 2305
    },
    {
      "epoch": 0.125570916120425,
      "grad_norm": 0.8137839369614454,
      "learning_rate": 0.00019943690639610592,
      "loss": 12.8366,
      "step": 2306
    },
    {
      "epoch": 0.12562537011700803,
      "grad_norm": 0.8556346327405181,
      "learning_rate": 0.00019943597151687937,
      "loss": 12.7881,
      "step": 2307
    },
    {
      "epoch": 0.12567982411359105,
      "grad_norm": 0.7709814149035997,
      "learning_rate": 0.00019943503586442206,
      "loss": 12.7591,
      "step": 2308
    },
    {
      "epoch": 0.12573427811017404,
      "grad_norm": 0.7898640756861948,
      "learning_rate": 0.00019943409943874122,
      "loss": 12.6926,
      "step": 2309
    },
    {
      "epoch": 0.12578873210675706,
      "grad_norm": 0.7367569366625164,
      "learning_rate": 0.00019943316223984414,
      "loss": 12.7746,
      "step": 2310
    },
    {
      "epoch": 0.12584318610334008,
      "grad_norm": 0.7454982746348482,
      "learning_rate": 0.00019943222426773813,
      "loss": 12.7185,
      "step": 2311
    },
    {
      "epoch": 0.12589764009992307,
      "grad_norm": 0.7967554594677826,
      "learning_rate": 0.00019943128552243046,
      "loss": 12.6746,
      "step": 2312
    },
    {
      "epoch": 0.1259520940965061,
      "grad_norm": 0.7758457456666698,
      "learning_rate": 0.00019943034600392845,
      "loss": 12.783,
      "step": 2313
    },
    {
      "epoch": 0.12600654809308912,
      "grad_norm": 0.8175081193982631,
      "learning_rate": 0.00019942940571223935,
      "loss": 12.7546,
      "step": 2314
    },
    {
      "epoch": 0.1260610020896721,
      "grad_norm": 0.7223717731557533,
      "learning_rate": 0.00019942846464737058,
      "loss": 12.6316,
      "step": 2315
    },
    {
      "epoch": 0.12611545608625513,
      "grad_norm": 0.7201639049114413,
      "learning_rate": 0.00019942752280932937,
      "loss": 12.6579,
      "step": 2316
    },
    {
      "epoch": 0.12616991008283815,
      "grad_norm": 0.6723939651791366,
      "learning_rate": 0.00019942658019812305,
      "loss": 12.7038,
      "step": 2317
    },
    {
      "epoch": 0.12622436407942114,
      "grad_norm": 0.756418238434827,
      "learning_rate": 0.000199425636813759,
      "loss": 12.7699,
      "step": 2318
    },
    {
      "epoch": 0.12627881807600416,
      "grad_norm": 0.7184296523805659,
      "learning_rate": 0.0001994246926562445,
      "loss": 12.6331,
      "step": 2319
    },
    {
      "epoch": 0.12633327207258718,
      "grad_norm": 0.8591570290913682,
      "learning_rate": 0.0001994237477255869,
      "loss": 12.7931,
      "step": 2320
    },
    {
      "epoch": 0.12638772606917018,
      "grad_norm": 0.7893364107713335,
      "learning_rate": 0.0001994228020217936,
      "loss": 12.7506,
      "step": 2321
    },
    {
      "epoch": 0.1264421800657532,
      "grad_norm": 0.8951456925670717,
      "learning_rate": 0.00019942185554487193,
      "loss": 12.7711,
      "step": 2322
    },
    {
      "epoch": 0.12649663406233622,
      "grad_norm": 0.9598649738838202,
      "learning_rate": 0.0001994209082948292,
      "loss": 12.616,
      "step": 2323
    },
    {
      "epoch": 0.12655108805891924,
      "grad_norm": 0.7527902561752199,
      "learning_rate": 0.00019941996027167286,
      "loss": 12.5971,
      "step": 2324
    },
    {
      "epoch": 0.12660554205550223,
      "grad_norm": 0.768501031065377,
      "learning_rate": 0.0001994190114754102,
      "loss": 12.6549,
      "step": 2325
    },
    {
      "epoch": 0.12665999605208525,
      "grad_norm": 0.817273213089173,
      "learning_rate": 0.00019941806190604863,
      "loss": 12.5479,
      "step": 2326
    },
    {
      "epoch": 0.12671445004866827,
      "grad_norm": 0.8645238416141345,
      "learning_rate": 0.00019941711156359554,
      "loss": 12.8375,
      "step": 2327
    },
    {
      "epoch": 0.12676890404525126,
      "grad_norm": 1.278825422993769,
      "learning_rate": 0.00019941616044805833,
      "loss": 12.6886,
      "step": 2328
    },
    {
      "epoch": 0.12682335804183428,
      "grad_norm": 0.7926814342834742,
      "learning_rate": 0.0001994152085594444,
      "loss": 12.8379,
      "step": 2329
    },
    {
      "epoch": 0.1268778120384173,
      "grad_norm": 0.8287504474355063,
      "learning_rate": 0.0001994142558977611,
      "loss": 12.6508,
      "step": 2330
    },
    {
      "epoch": 0.1269322660350003,
      "grad_norm": 1.0534286243495066,
      "learning_rate": 0.0001994133024630159,
      "loss": 12.7352,
      "step": 2331
    },
    {
      "epoch": 0.12698672003158332,
      "grad_norm": 0.7493620548408583,
      "learning_rate": 0.00019941234825521616,
      "loss": 12.7401,
      "step": 2332
    },
    {
      "epoch": 0.12704117402816634,
      "grad_norm": 0.8318357837689145,
      "learning_rate": 0.00019941139327436935,
      "loss": 12.652,
      "step": 2333
    },
    {
      "epoch": 0.12709562802474933,
      "grad_norm": 0.7715913785784512,
      "learning_rate": 0.0001994104375204829,
      "loss": 12.7926,
      "step": 2334
    },
    {
      "epoch": 0.12715008202133235,
      "grad_norm": 0.7542855191671053,
      "learning_rate": 0.00019940948099356418,
      "loss": 12.6063,
      "step": 2335
    },
    {
      "epoch": 0.12720453601791537,
      "grad_norm": 0.9658985909567381,
      "learning_rate": 0.00019940852369362068,
      "loss": 12.6083,
      "step": 2336
    },
    {
      "epoch": 0.12725899001449836,
      "grad_norm": 0.6851974060541999,
      "learning_rate": 0.00019940756562065982,
      "loss": 12.5249,
      "step": 2337
    },
    {
      "epoch": 0.12731344401108138,
      "grad_norm": 0.7785530678240814,
      "learning_rate": 0.00019940660677468904,
      "loss": 12.7989,
      "step": 2338
    },
    {
      "epoch": 0.1273678980076644,
      "grad_norm": 0.7379846325689866,
      "learning_rate": 0.00019940564715571586,
      "loss": 12.6929,
      "step": 2339
    },
    {
      "epoch": 0.12742235200424742,
      "grad_norm": 0.792901212371482,
      "learning_rate": 0.00019940468676374765,
      "loss": 12.779,
      "step": 2340
    },
    {
      "epoch": 0.12747680600083042,
      "grad_norm": 0.7337004708678737,
      "learning_rate": 0.000199403725598792,
      "loss": 12.6837,
      "step": 2341
    },
    {
      "epoch": 0.12753125999741344,
      "grad_norm": 0.7368460634198486,
      "learning_rate": 0.00019940276366085623,
      "loss": 12.6762,
      "step": 2342
    },
    {
      "epoch": 0.12758571399399646,
      "grad_norm": 0.7485113299926779,
      "learning_rate": 0.00019940180094994792,
      "loss": 12.6945,
      "step": 2343
    },
    {
      "epoch": 0.12764016799057945,
      "grad_norm": 0.7121957433105658,
      "learning_rate": 0.00019940083746607456,
      "loss": 12.6726,
      "step": 2344
    },
    {
      "epoch": 0.12769462198716247,
      "grad_norm": 1.5233087180492444,
      "learning_rate": 0.0001993998732092436,
      "loss": 12.6615,
      "step": 2345
    },
    {
      "epoch": 0.1277490759837455,
      "grad_norm": 0.7962138637784169,
      "learning_rate": 0.00019939890817946259,
      "loss": 12.6639,
      "step": 2346
    },
    {
      "epoch": 0.12780352998032848,
      "grad_norm": 0.7149436169520963,
      "learning_rate": 0.00019939794237673896,
      "loss": 12.7055,
      "step": 2347
    },
    {
      "epoch": 0.1278579839769115,
      "grad_norm": 0.7602346657839104,
      "learning_rate": 0.00019939697580108025,
      "loss": 12.7092,
      "step": 2348
    },
    {
      "epoch": 0.12791243797349452,
      "grad_norm": 0.751533657680629,
      "learning_rate": 0.00019939600845249403,
      "loss": 12.6796,
      "step": 2349
    },
    {
      "epoch": 0.12796689197007752,
      "grad_norm": 0.7980077399207712,
      "learning_rate": 0.00019939504033098776,
      "loss": 12.6877,
      "step": 2350
    },
    {
      "epoch": 0.12802134596666054,
      "grad_norm": 0.7532150415596107,
      "learning_rate": 0.000199394071436569,
      "loss": 12.7105,
      "step": 2351
    },
    {
      "epoch": 0.12807579996324356,
      "grad_norm": 0.7315030759058961,
      "learning_rate": 0.00019939310176924523,
      "loss": 12.646,
      "step": 2352
    },
    {
      "epoch": 0.12813025395982655,
      "grad_norm": 0.7864302503318793,
      "learning_rate": 0.00019939213132902408,
      "loss": 12.6704,
      "step": 2353
    },
    {
      "epoch": 0.12818470795640957,
      "grad_norm": 0.7913586628644312,
      "learning_rate": 0.00019939116011591303,
      "loss": 12.7471,
      "step": 2354
    },
    {
      "epoch": 0.1282391619529926,
      "grad_norm": 0.8119181526158408,
      "learning_rate": 0.00019939018812991966,
      "loss": 12.7335,
      "step": 2355
    },
    {
      "epoch": 0.1282936159495756,
      "grad_norm": 0.7735205615622864,
      "learning_rate": 0.00019938921537105152,
      "loss": 12.7511,
      "step": 2356
    },
    {
      "epoch": 0.1283480699461586,
      "grad_norm": 0.7968665962275414,
      "learning_rate": 0.00019938824183931617,
      "loss": 12.7516,
      "step": 2357
    },
    {
      "epoch": 0.12840252394274163,
      "grad_norm": 0.7338392610553128,
      "learning_rate": 0.00019938726753472116,
      "loss": 12.6123,
      "step": 2358
    },
    {
      "epoch": 0.12845697793932465,
      "grad_norm": 0.7827847883010737,
      "learning_rate": 0.00019938629245727413,
      "loss": 12.7633,
      "step": 2359
    },
    {
      "epoch": 0.12851143193590764,
      "grad_norm": 0.7450380140451616,
      "learning_rate": 0.00019938531660698258,
      "loss": 12.773,
      "step": 2360
    },
    {
      "epoch": 0.12856588593249066,
      "grad_norm": 0.7989051885239961,
      "learning_rate": 0.00019938433998385418,
      "loss": 12.6847,
      "step": 2361
    },
    {
      "epoch": 0.12862033992907368,
      "grad_norm": 0.7356763210194389,
      "learning_rate": 0.00019938336258789647,
      "loss": 12.655,
      "step": 2362
    },
    {
      "epoch": 0.12867479392565667,
      "grad_norm": 0.8986725123346193,
      "learning_rate": 0.00019938238441911705,
      "loss": 12.9576,
      "step": 2363
    },
    {
      "epoch": 0.1287292479222397,
      "grad_norm": 0.7481316704368125,
      "learning_rate": 0.00019938140547752354,
      "loss": 12.737,
      "step": 2364
    },
    {
      "epoch": 0.1287837019188227,
      "grad_norm": 0.7894228636204875,
      "learning_rate": 0.0001993804257631236,
      "loss": 12.7876,
      "step": 2365
    },
    {
      "epoch": 0.1288381559154057,
      "grad_norm": 0.8122936643074704,
      "learning_rate": 0.0001993794452759248,
      "loss": 12.7443,
      "step": 2366
    },
    {
      "epoch": 0.12889260991198873,
      "grad_norm": 0.7765834068020498,
      "learning_rate": 0.00019937846401593473,
      "loss": 12.8403,
      "step": 2367
    },
    {
      "epoch": 0.12894706390857175,
      "grad_norm": 0.6921187212435841,
      "learning_rate": 0.00019937748198316105,
      "loss": 12.6311,
      "step": 2368
    },
    {
      "epoch": 0.12900151790515474,
      "grad_norm": 0.7502571830149722,
      "learning_rate": 0.00019937649917761143,
      "loss": 12.6613,
      "step": 2369
    },
    {
      "epoch": 0.12905597190173776,
      "grad_norm": 0.7821551845777908,
      "learning_rate": 0.0001993755155992935,
      "loss": 12.8988,
      "step": 2370
    },
    {
      "epoch": 0.12911042589832078,
      "grad_norm": 0.7418220682205031,
      "learning_rate": 0.00019937453124821487,
      "loss": 12.6488,
      "step": 2371
    },
    {
      "epoch": 0.1291648798949038,
      "grad_norm": 0.8117136365721119,
      "learning_rate": 0.00019937354612438321,
      "loss": 12.6928,
      "step": 2372
    },
    {
      "epoch": 0.1292193338914868,
      "grad_norm": 0.7981778766751755,
      "learning_rate": 0.00019937256022780622,
      "loss": 12.8224,
      "step": 2373
    },
    {
      "epoch": 0.1292737878880698,
      "grad_norm": 0.7479712328362667,
      "learning_rate": 0.00019937157355849153,
      "loss": 12.6721,
      "step": 2374
    },
    {
      "epoch": 0.12932824188465283,
      "grad_norm": 0.7884826639891351,
      "learning_rate": 0.0001993705861164468,
      "loss": 12.8831,
      "step": 2375
    },
    {
      "epoch": 0.12938269588123583,
      "grad_norm": 0.7255531485760002,
      "learning_rate": 0.00019936959790167974,
      "loss": 12.6682,
      "step": 2376
    },
    {
      "epoch": 0.12943714987781885,
      "grad_norm": 0.7100455346830856,
      "learning_rate": 0.00019936860891419804,
      "loss": 12.669,
      "step": 2377
    },
    {
      "epoch": 0.12949160387440187,
      "grad_norm": 0.7806212789623882,
      "learning_rate": 0.00019936761915400936,
      "loss": 12.532,
      "step": 2378
    },
    {
      "epoch": 0.12954605787098486,
      "grad_norm": 0.7431760599900052,
      "learning_rate": 0.0001993666286211214,
      "loss": 12.6516,
      "step": 2379
    },
    {
      "epoch": 0.12960051186756788,
      "grad_norm": 0.7213877200770814,
      "learning_rate": 0.00019936563731554188,
      "loss": 12.6298,
      "step": 2380
    },
    {
      "epoch": 0.1296549658641509,
      "grad_norm": 0.8736262624845132,
      "learning_rate": 0.0001993646452372785,
      "loss": 12.8446,
      "step": 2381
    },
    {
      "epoch": 0.1297094198607339,
      "grad_norm": 0.7855509988563504,
      "learning_rate": 0.000199363652386339,
      "loss": 12.7075,
      "step": 2382
    },
    {
      "epoch": 0.1297638738573169,
      "grad_norm": 0.7299958322594098,
      "learning_rate": 0.00019936265876273104,
      "loss": 12.6899,
      "step": 2383
    },
    {
      "epoch": 0.12981832785389993,
      "grad_norm": 0.8168217259134454,
      "learning_rate": 0.00019936166436646237,
      "loss": 12.7358,
      "step": 2384
    },
    {
      "epoch": 0.12987278185048293,
      "grad_norm": 0.7768691061368659,
      "learning_rate": 0.00019936066919754077,
      "loss": 12.705,
      "step": 2385
    },
    {
      "epoch": 0.12992723584706595,
      "grad_norm": 0.8847332828074436,
      "learning_rate": 0.00019935967325597392,
      "loss": 12.7719,
      "step": 2386
    },
    {
      "epoch": 0.12998168984364897,
      "grad_norm": 0.7083908366634801,
      "learning_rate": 0.0001993586765417696,
      "loss": 12.7458,
      "step": 2387
    },
    {
      "epoch": 0.13003614384023196,
      "grad_norm": 0.7875102036153997,
      "learning_rate": 0.00019935767905493557,
      "loss": 12.6281,
      "step": 2388
    },
    {
      "epoch": 0.13009059783681498,
      "grad_norm": 0.7395792011876929,
      "learning_rate": 0.00019935668079547957,
      "loss": 12.6863,
      "step": 2389
    },
    {
      "epoch": 0.130145051833398,
      "grad_norm": 0.7621670186428621,
      "learning_rate": 0.00019935568176340928,
      "loss": 12.7617,
      "step": 2390
    },
    {
      "epoch": 0.13019950582998102,
      "grad_norm": 0.7847645774595151,
      "learning_rate": 0.00019935468195873262,
      "loss": 12.7119,
      "step": 2391
    },
    {
      "epoch": 0.13025395982656401,
      "grad_norm": 0.7375661269533667,
      "learning_rate": 0.00019935368138145727,
      "loss": 12.7158,
      "step": 2392
    },
    {
      "epoch": 0.13030841382314703,
      "grad_norm": 0.8126269955672756,
      "learning_rate": 0.00019935268003159101,
      "loss": 12.6302,
      "step": 2393
    },
    {
      "epoch": 0.13036286781973005,
      "grad_norm": 0.8268543789843175,
      "learning_rate": 0.0001993516779091417,
      "loss": 12.7678,
      "step": 2394
    },
    {
      "epoch": 0.13041732181631305,
      "grad_norm": 0.8205187707857627,
      "learning_rate": 0.00019935067501411705,
      "loss": 12.6628,
      "step": 2395
    },
    {
      "epoch": 0.13047177581289607,
      "grad_norm": 0.7378866568993926,
      "learning_rate": 0.0001993496713465249,
      "loss": 12.7025,
      "step": 2396
    },
    {
      "epoch": 0.1305262298094791,
      "grad_norm": 0.877940758716135,
      "learning_rate": 0.00019934866690637302,
      "loss": 12.72,
      "step": 2397
    },
    {
      "epoch": 0.13058068380606208,
      "grad_norm": 0.7267513693465012,
      "learning_rate": 0.00019934766169366929,
      "loss": 12.6185,
      "step": 2398
    },
    {
      "epoch": 0.1306351378026451,
      "grad_norm": 0.6846617254702555,
      "learning_rate": 0.0001993466557084214,
      "loss": 12.6491,
      "step": 2399
    },
    {
      "epoch": 0.13068959179922812,
      "grad_norm": 0.7491981620181035,
      "learning_rate": 0.00019934564895063734,
      "loss": 12.6363,
      "step": 2400
    },
    {
      "epoch": 0.13074404579581111,
      "grad_norm": 0.7751118812572666,
      "learning_rate": 0.00019934464142032482,
      "loss": 12.7001,
      "step": 2401
    },
    {
      "epoch": 0.13079849979239414,
      "grad_norm": 0.8264496185916214,
      "learning_rate": 0.00019934363311749172,
      "loss": 12.73,
      "step": 2402
    },
    {
      "epoch": 0.13085295378897716,
      "grad_norm": 0.8321637617252079,
      "learning_rate": 0.00019934262404214584,
      "loss": 12.6028,
      "step": 2403
    },
    {
      "epoch": 0.13090740778556015,
      "grad_norm": 0.7848027006827053,
      "learning_rate": 0.0001993416141942951,
      "loss": 12.9119,
      "step": 2404
    },
    {
      "epoch": 0.13096186178214317,
      "grad_norm": 0.857679708710815,
      "learning_rate": 0.00019934060357394725,
      "loss": 12.6421,
      "step": 2405
    },
    {
      "epoch": 0.1310163157787262,
      "grad_norm": 0.8055289289746956,
      "learning_rate": 0.00019933959218111026,
      "loss": 12.6846,
      "step": 2406
    },
    {
      "epoch": 0.1310707697753092,
      "grad_norm": 0.7747666351779556,
      "learning_rate": 0.0001993385800157919,
      "loss": 12.7859,
      "step": 2407
    },
    {
      "epoch": 0.1311252237718922,
      "grad_norm": 0.756568540020912,
      "learning_rate": 0.00019933756707800012,
      "loss": 12.8728,
      "step": 2408
    },
    {
      "epoch": 0.13117967776847522,
      "grad_norm": 0.7997659508246298,
      "learning_rate": 0.00019933655336774276,
      "loss": 12.7844,
      "step": 2409
    },
    {
      "epoch": 0.13123413176505824,
      "grad_norm": 0.8394640526431785,
      "learning_rate": 0.00019933553888502767,
      "loss": 12.7839,
      "step": 2410
    },
    {
      "epoch": 0.13128858576164124,
      "grad_norm": 0.8828481353034261,
      "learning_rate": 0.0001993345236298628,
      "loss": 12.6463,
      "step": 2411
    },
    {
      "epoch": 0.13134303975822426,
      "grad_norm": 0.7343617366920848,
      "learning_rate": 0.000199333507602256,
      "loss": 12.7517,
      "step": 2412
    },
    {
      "epoch": 0.13139749375480728,
      "grad_norm": 0.8459936383316501,
      "learning_rate": 0.0001993324908022152,
      "loss": 12.7808,
      "step": 2413
    },
    {
      "epoch": 0.13145194775139027,
      "grad_norm": 0.7608363034537843,
      "learning_rate": 0.00019933147322974827,
      "loss": 12.7196,
      "step": 2414
    },
    {
      "epoch": 0.1315064017479733,
      "grad_norm": 0.6668086521669773,
      "learning_rate": 0.00019933045488486318,
      "loss": 12.5531,
      "step": 2415
    },
    {
      "epoch": 0.1315608557445563,
      "grad_norm": 0.7002068468454027,
      "learning_rate": 0.00019932943576756777,
      "loss": 12.7032,
      "step": 2416
    },
    {
      "epoch": 0.1316153097411393,
      "grad_norm": 0.9000736480641097,
      "learning_rate": 0.00019932841587787006,
      "loss": 12.6522,
      "step": 2417
    },
    {
      "epoch": 0.13166976373772232,
      "grad_norm": 0.7721331574493356,
      "learning_rate": 0.0001993273952157779,
      "loss": 12.6671,
      "step": 2418
    },
    {
      "epoch": 0.13172421773430534,
      "grad_norm": 0.8308437806311205,
      "learning_rate": 0.00019932637378129926,
      "loss": 12.5842,
      "step": 2419
    },
    {
      "epoch": 0.13177867173088834,
      "grad_norm": 0.8219447869064825,
      "learning_rate": 0.00019932535157444206,
      "loss": 12.9223,
      "step": 2420
    },
    {
      "epoch": 0.13183312572747136,
      "grad_norm": 0.8560251642378786,
      "learning_rate": 0.00019932432859521432,
      "loss": 12.7095,
      "step": 2421
    },
    {
      "epoch": 0.13188757972405438,
      "grad_norm": 0.7308984604514989,
      "learning_rate": 0.00019932330484362392,
      "loss": 12.7873,
      "step": 2422
    },
    {
      "epoch": 0.1319420337206374,
      "grad_norm": 0.810039047049506,
      "learning_rate": 0.00019932228031967886,
      "loss": 12.6626,
      "step": 2423
    },
    {
      "epoch": 0.1319964877172204,
      "grad_norm": 0.7508769945880449,
      "learning_rate": 0.00019932125502338706,
      "loss": 12.7308,
      "step": 2424
    },
    {
      "epoch": 0.1320509417138034,
      "grad_norm": 0.756496840443002,
      "learning_rate": 0.00019932022895475653,
      "loss": 12.7622,
      "step": 2425
    },
    {
      "epoch": 0.13210539571038643,
      "grad_norm": 0.8579954679459876,
      "learning_rate": 0.00019931920211379526,
      "loss": 12.6203,
      "step": 2426
    },
    {
      "epoch": 0.13215984970696942,
      "grad_norm": 0.8188439639621717,
      "learning_rate": 0.00019931817450051121,
      "loss": 12.777,
      "step": 2427
    },
    {
      "epoch": 0.13221430370355244,
      "grad_norm": 0.7532868906855189,
      "learning_rate": 0.00019931714611491237,
      "loss": 12.8348,
      "step": 2428
    },
    {
      "epoch": 0.13226875770013546,
      "grad_norm": 0.7958028147926751,
      "learning_rate": 0.00019931611695700677,
      "loss": 12.694,
      "step": 2429
    },
    {
      "epoch": 0.13232321169671846,
      "grad_norm": 0.7814894757386551,
      "learning_rate": 0.00019931508702680236,
      "loss": 12.6925,
      "step": 2430
    },
    {
      "epoch": 0.13237766569330148,
      "grad_norm": 0.8324479778200059,
      "learning_rate": 0.00019931405632430722,
      "loss": 12.6425,
      "step": 2431
    },
    {
      "epoch": 0.1324321196898845,
      "grad_norm": 0.8092206373047455,
      "learning_rate": 0.00019931302484952926,
      "loss": 12.8163,
      "step": 2432
    },
    {
      "epoch": 0.1324865736864675,
      "grad_norm": 0.7377978092615135,
      "learning_rate": 0.00019931199260247664,
      "loss": 12.5553,
      "step": 2433
    },
    {
      "epoch": 0.1325410276830505,
      "grad_norm": 0.755796514648496,
      "learning_rate": 0.00019931095958315725,
      "loss": 12.7638,
      "step": 2434
    },
    {
      "epoch": 0.13259548167963353,
      "grad_norm": 0.8040857355178929,
      "learning_rate": 0.00019930992579157922,
      "loss": 12.7484,
      "step": 2435
    },
    {
      "epoch": 0.13264993567621652,
      "grad_norm": 0.7939845278937419,
      "learning_rate": 0.00019930889122775054,
      "loss": 12.5594,
      "step": 2436
    },
    {
      "epoch": 0.13270438967279954,
      "grad_norm": 0.7699514168539291,
      "learning_rate": 0.00019930785589167927,
      "loss": 12.7626,
      "step": 2437
    },
    {
      "epoch": 0.13275884366938256,
      "grad_norm": 1.033416481849656,
      "learning_rate": 0.00019930681978337343,
      "loss": 12.8065,
      "step": 2438
    },
    {
      "epoch": 0.13281329766596559,
      "grad_norm": 0.7417687328317938,
      "learning_rate": 0.00019930578290284115,
      "loss": 12.6928,
      "step": 2439
    },
    {
      "epoch": 0.13286775166254858,
      "grad_norm": 0.8562815513817471,
      "learning_rate": 0.00019930474525009043,
      "loss": 12.8132,
      "step": 2440
    },
    {
      "epoch": 0.1329222056591316,
      "grad_norm": 0.8832737109439536,
      "learning_rate": 0.00019930370682512936,
      "loss": 12.7486,
      "step": 2441
    },
    {
      "epoch": 0.13297665965571462,
      "grad_norm": 0.8655967771917551,
      "learning_rate": 0.000199302667627966,
      "loss": 12.7677,
      "step": 2442
    },
    {
      "epoch": 0.1330311136522976,
      "grad_norm": 0.8352025276927129,
      "learning_rate": 0.00019930162765860847,
      "loss": 12.6883,
      "step": 2443
    },
    {
      "epoch": 0.13308556764888063,
      "grad_norm": 0.7657269519474376,
      "learning_rate": 0.0001993005869170648,
      "loss": 12.639,
      "step": 2444
    },
    {
      "epoch": 0.13314002164546365,
      "grad_norm": 1.0200413938057031,
      "learning_rate": 0.00019929954540334315,
      "loss": 12.6656,
      "step": 2445
    },
    {
      "epoch": 0.13319447564204664,
      "grad_norm": 0.7208134883027584,
      "learning_rate": 0.00019929850311745155,
      "loss": 12.7271,
      "step": 2446
    },
    {
      "epoch": 0.13324892963862967,
      "grad_norm": 0.8273881026402172,
      "learning_rate": 0.00019929746005939813,
      "loss": 12.7771,
      "step": 2447
    },
    {
      "epoch": 0.13330338363521269,
      "grad_norm": 0.7493445587390899,
      "learning_rate": 0.00019929641622919104,
      "loss": 12.6896,
      "step": 2448
    },
    {
      "epoch": 0.13335783763179568,
      "grad_norm": 0.742375521459718,
      "learning_rate": 0.00019929537162683835,
      "loss": 12.6528,
      "step": 2449
    },
    {
      "epoch": 0.1334122916283787,
      "grad_norm": 0.9208561020185162,
      "learning_rate": 0.0001992943262523482,
      "loss": 12.7464,
      "step": 2450
    },
    {
      "epoch": 0.13346674562496172,
      "grad_norm": 0.7293788209577327,
      "learning_rate": 0.0001992932801057287,
      "loss": 12.6756,
      "step": 2451
    },
    {
      "epoch": 0.1335211996215447,
      "grad_norm": 0.9920197065465052,
      "learning_rate": 0.00019929223318698803,
      "loss": 12.7579,
      "step": 2452
    },
    {
      "epoch": 0.13357565361812773,
      "grad_norm": 0.9922516429306217,
      "learning_rate": 0.0001992911854961343,
      "loss": 12.6291,
      "step": 2453
    },
    {
      "epoch": 0.13363010761471075,
      "grad_norm": 0.8123075917723392,
      "learning_rate": 0.00019929013703317563,
      "loss": 12.7533,
      "step": 2454
    },
    {
      "epoch": 0.13368456161129375,
      "grad_norm": 0.8599656764223896,
      "learning_rate": 0.00019928908779812026,
      "loss": 12.6901,
      "step": 2455
    },
    {
      "epoch": 0.13373901560787677,
      "grad_norm": 0.8907052736792533,
      "learning_rate": 0.00019928803779097623,
      "loss": 12.7852,
      "step": 2456
    },
    {
      "epoch": 0.13379346960445979,
      "grad_norm": 0.8338265561480035,
      "learning_rate": 0.00019928698701175178,
      "loss": 12.6099,
      "step": 2457
    },
    {
      "epoch": 0.1338479236010428,
      "grad_norm": 0.9079405685140904,
      "learning_rate": 0.0001992859354604551,
      "loss": 12.6389,
      "step": 2458
    },
    {
      "epoch": 0.1339023775976258,
      "grad_norm": 0.8278803604932522,
      "learning_rate": 0.00019928488313709434,
      "loss": 12.7932,
      "step": 2459
    },
    {
      "epoch": 0.13395683159420882,
      "grad_norm": 0.8003952204297358,
      "learning_rate": 0.00019928383004167764,
      "loss": 12.7772,
      "step": 2460
    },
    {
      "epoch": 0.13401128559079184,
      "grad_norm": 0.672183505197408,
      "learning_rate": 0.00019928277617421326,
      "loss": 12.5482,
      "step": 2461
    },
    {
      "epoch": 0.13406573958737483,
      "grad_norm": 0.7396340985238629,
      "learning_rate": 0.00019928172153470933,
      "loss": 12.7081,
      "step": 2462
    },
    {
      "epoch": 0.13412019358395785,
      "grad_norm": 0.735938888362427,
      "learning_rate": 0.00019928066612317412,
      "loss": 12.5721,
      "step": 2463
    },
    {
      "epoch": 0.13417464758054087,
      "grad_norm": 0.7331259242018381,
      "learning_rate": 0.00019927960993961578,
      "loss": 12.5909,
      "step": 2464
    },
    {
      "epoch": 0.13422910157712387,
      "grad_norm": 0.8850945952602003,
      "learning_rate": 0.00019927855298404253,
      "loss": 12.7318,
      "step": 2465
    },
    {
      "epoch": 0.1342835555737069,
      "grad_norm": 0.7471688438849723,
      "learning_rate": 0.00019927749525646264,
      "loss": 12.6139,
      "step": 2466
    },
    {
      "epoch": 0.1343380095702899,
      "grad_norm": 0.8094440813264298,
      "learning_rate": 0.0001992764367568843,
      "loss": 12.8425,
      "step": 2467
    },
    {
      "epoch": 0.1343924635668729,
      "grad_norm": 0.8310365902118073,
      "learning_rate": 0.0001992753774853157,
      "loss": 12.652,
      "step": 2468
    },
    {
      "epoch": 0.13444691756345592,
      "grad_norm": 0.7309480519652607,
      "learning_rate": 0.00019927431744176514,
      "loss": 12.6415,
      "step": 2469
    },
    {
      "epoch": 0.13450137156003894,
      "grad_norm": 0.8196147699360571,
      "learning_rate": 0.00019927325662624082,
      "loss": 12.7008,
      "step": 2470
    },
    {
      "epoch": 0.13455582555662193,
      "grad_norm": 0.8298741408117133,
      "learning_rate": 0.00019927219503875103,
      "loss": 12.7277,
      "step": 2471
    },
    {
      "epoch": 0.13461027955320495,
      "grad_norm": 0.6554252989687059,
      "learning_rate": 0.00019927113267930398,
      "loss": 12.6143,
      "step": 2472
    },
    {
      "epoch": 0.13466473354978797,
      "grad_norm": 0.7473824587961554,
      "learning_rate": 0.000199270069547908,
      "loss": 12.6095,
      "step": 2473
    },
    {
      "epoch": 0.134719187546371,
      "grad_norm": 0.8160647428862018,
      "learning_rate": 0.00019926900564457128,
      "loss": 12.692,
      "step": 2474
    },
    {
      "epoch": 0.134773641542954,
      "grad_norm": 0.7492014235360235,
      "learning_rate": 0.0001992679409693021,
      "loss": 12.7103,
      "step": 2475
    },
    {
      "epoch": 0.134828095539537,
      "grad_norm": 0.839607875095279,
      "learning_rate": 0.0001992668755221088,
      "loss": 12.8071,
      "step": 2476
    },
    {
      "epoch": 0.13488254953612003,
      "grad_norm": 0.801927344699619,
      "learning_rate": 0.00019926580930299963,
      "loss": 12.7339,
      "step": 2477
    },
    {
      "epoch": 0.13493700353270302,
      "grad_norm": 0.8960165647196202,
      "learning_rate": 0.00019926474231198285,
      "loss": 12.6735,
      "step": 2478
    },
    {
      "epoch": 0.13499145752928604,
      "grad_norm": 0.7933722921821766,
      "learning_rate": 0.0001992636745490668,
      "loss": 12.7974,
      "step": 2479
    },
    {
      "epoch": 0.13504591152586906,
      "grad_norm": 0.9451088853717208,
      "learning_rate": 0.0001992626060142598,
      "loss": 12.5382,
      "step": 2480
    },
    {
      "epoch": 0.13510036552245205,
      "grad_norm": 0.8345545262101434,
      "learning_rate": 0.00019926153670757007,
      "loss": 12.7088,
      "step": 2481
    },
    {
      "epoch": 0.13515481951903507,
      "grad_norm": 0.7211833903776345,
      "learning_rate": 0.00019926046662900606,
      "loss": 12.6001,
      "step": 2482
    },
    {
      "epoch": 0.1352092735156181,
      "grad_norm": 0.8170867246642093,
      "learning_rate": 0.00019925939577857595,
      "loss": 12.7586,
      "step": 2483
    },
    {
      "epoch": 0.1352637275122011,
      "grad_norm": 0.8766455026502107,
      "learning_rate": 0.00019925832415628817,
      "loss": 12.8397,
      "step": 2484
    },
    {
      "epoch": 0.1353181815087841,
      "grad_norm": 0.8304795223103192,
      "learning_rate": 0.000199257251762151,
      "loss": 12.8194,
      "step": 2485
    },
    {
      "epoch": 0.13537263550536713,
      "grad_norm": 0.8258332528487458,
      "learning_rate": 0.0001992561785961728,
      "loss": 12.8221,
      "step": 2486
    },
    {
      "epoch": 0.13542708950195012,
      "grad_norm": 0.7722041721631009,
      "learning_rate": 0.0001992551046583619,
      "loss": 12.6855,
      "step": 2487
    },
    {
      "epoch": 0.13548154349853314,
      "grad_norm": 0.8196913899421888,
      "learning_rate": 0.00019925402994872666,
      "loss": 12.6272,
      "step": 2488
    },
    {
      "epoch": 0.13553599749511616,
      "grad_norm": 0.8220844770061915,
      "learning_rate": 0.00019925295446727548,
      "loss": 12.7254,
      "step": 2489
    },
    {
      "epoch": 0.13559045149169918,
      "grad_norm": 0.7441266743639062,
      "learning_rate": 0.00019925187821401663,
      "loss": 12.6433,
      "step": 2490
    },
    {
      "epoch": 0.13564490548828217,
      "grad_norm": 0.7922384453675508,
      "learning_rate": 0.00019925080118895857,
      "loss": 12.7242,
      "step": 2491
    },
    {
      "epoch": 0.1356993594848652,
      "grad_norm": 0.7615577137990498,
      "learning_rate": 0.00019924972339210962,
      "loss": 12.6124,
      "step": 2492
    },
    {
      "epoch": 0.13575381348144822,
      "grad_norm": 0.8763841639212409,
      "learning_rate": 0.00019924864482347813,
      "loss": 12.7149,
      "step": 2493
    },
    {
      "epoch": 0.1358082674780312,
      "grad_norm": 0.755861381321944,
      "learning_rate": 0.00019924756548307258,
      "loss": 12.7621,
      "step": 2494
    },
    {
      "epoch": 0.13586272147461423,
      "grad_norm": 0.7151226118739447,
      "learning_rate": 0.00019924648537090133,
      "loss": 12.5913,
      "step": 2495
    },
    {
      "epoch": 0.13591717547119725,
      "grad_norm": 0.7134477879040643,
      "learning_rate": 0.00019924540448697275,
      "loss": 12.7356,
      "step": 2496
    },
    {
      "epoch": 0.13597162946778024,
      "grad_norm": 0.7416597457191845,
      "learning_rate": 0.00019924432283129526,
      "loss": 12.6654,
      "step": 2497
    },
    {
      "epoch": 0.13602608346436326,
      "grad_norm": 0.8960485887635284,
      "learning_rate": 0.00019924324040387727,
      "loss": 12.6783,
      "step": 2498
    },
    {
      "epoch": 0.13608053746094628,
      "grad_norm": 0.8043344559108634,
      "learning_rate": 0.0001992421572047272,
      "loss": 12.6638,
      "step": 2499
    },
    {
      "epoch": 0.13613499145752928,
      "grad_norm": 0.8477672467383391,
      "learning_rate": 0.00019924107323385348,
      "loss": 12.6786,
      "step": 2500
    },
    {
      "epoch": 0.1361894454541123,
      "grad_norm": 0.8350506299141978,
      "learning_rate": 0.00019923998849126452,
      "loss": 12.7616,
      "step": 2501
    },
    {
      "epoch": 0.13624389945069532,
      "grad_norm": 0.755115387664609,
      "learning_rate": 0.0001992389029769688,
      "loss": 12.5811,
      "step": 2502
    },
    {
      "epoch": 0.1362983534472783,
      "grad_norm": 0.8296141248216201,
      "learning_rate": 0.00019923781669097467,
      "loss": 12.697,
      "step": 2503
    },
    {
      "epoch": 0.13635280744386133,
      "grad_norm": 0.777487444662502,
      "learning_rate": 0.00019923672963329068,
      "loss": 12.9226,
      "step": 2504
    },
    {
      "epoch": 0.13640726144044435,
      "grad_norm": 0.8009633102281982,
      "learning_rate": 0.00019923564180392522,
      "loss": 12.7052,
      "step": 2505
    },
    {
      "epoch": 0.13646171543702737,
      "grad_norm": 0.8660245138597285,
      "learning_rate": 0.00019923455320288678,
      "loss": 12.6371,
      "step": 2506
    },
    {
      "epoch": 0.13651616943361036,
      "grad_norm": 0.799333483782979,
      "learning_rate": 0.0001992334638301838,
      "loss": 12.5216,
      "step": 2507
    },
    {
      "epoch": 0.13657062343019338,
      "grad_norm": 0.7890530468567644,
      "learning_rate": 0.00019923237368582478,
      "loss": 12.7124,
      "step": 2508
    },
    {
      "epoch": 0.1366250774267764,
      "grad_norm": 0.8051322273979854,
      "learning_rate": 0.00019923128276981816,
      "loss": 12.7529,
      "step": 2509
    },
    {
      "epoch": 0.1366795314233594,
      "grad_norm": 0.7613188323047255,
      "learning_rate": 0.00019923019108217244,
      "loss": 12.6688,
      "step": 2510
    },
    {
      "epoch": 0.13673398541994242,
      "grad_norm": 0.7175554055582198,
      "learning_rate": 0.00019922909862289613,
      "loss": 12.6997,
      "step": 2511
    },
    {
      "epoch": 0.13678843941652544,
      "grad_norm": 0.7617772690976709,
      "learning_rate": 0.00019922800539199772,
      "loss": 12.7001,
      "step": 2512
    },
    {
      "epoch": 0.13684289341310843,
      "grad_norm": 0.8815313899645151,
      "learning_rate": 0.00019922691138948566,
      "loss": 12.7137,
      "step": 2513
    },
    {
      "epoch": 0.13689734740969145,
      "grad_norm": 0.6953988947363619,
      "learning_rate": 0.0001992258166153685,
      "loss": 12.6242,
      "step": 2514
    },
    {
      "epoch": 0.13695180140627447,
      "grad_norm": 0.7211808446872298,
      "learning_rate": 0.00019922472106965482,
      "loss": 12.6297,
      "step": 2515
    },
    {
      "epoch": 0.13700625540285746,
      "grad_norm": 0.7805289346970223,
      "learning_rate": 0.00019922362475235301,
      "loss": 12.7185,
      "step": 2516
    },
    {
      "epoch": 0.13706070939944048,
      "grad_norm": 0.7435082825109165,
      "learning_rate": 0.00019922252766347166,
      "loss": 12.6931,
      "step": 2517
    },
    {
      "epoch": 0.1371151633960235,
      "grad_norm": 0.7593964045770272,
      "learning_rate": 0.0001992214298030193,
      "loss": 12.7587,
      "step": 2518
    },
    {
      "epoch": 0.1371696173926065,
      "grad_norm": 1.0711067900247304,
      "learning_rate": 0.00019922033117100447,
      "loss": 12.7403,
      "step": 2519
    },
    {
      "epoch": 0.13722407138918952,
      "grad_norm": 0.8442713964578787,
      "learning_rate": 0.0001992192317674357,
      "loss": 12.6896,
      "step": 2520
    },
    {
      "epoch": 0.13727852538577254,
      "grad_norm": 0.7379985680047709,
      "learning_rate": 0.00019921813159232151,
      "loss": 12.5902,
      "step": 2521
    },
    {
      "epoch": 0.13733297938235553,
      "grad_norm": 0.641792785472628,
      "learning_rate": 0.00019921703064567056,
      "loss": 12.5831,
      "step": 2522
    },
    {
      "epoch": 0.13738743337893855,
      "grad_norm": 0.7085982767459995,
      "learning_rate": 0.0001992159289274913,
      "loss": 12.5308,
      "step": 2523
    },
    {
      "epoch": 0.13744188737552157,
      "grad_norm": 0.8689263472913269,
      "learning_rate": 0.00019921482643779235,
      "loss": 12.6573,
      "step": 2524
    },
    {
      "epoch": 0.1374963413721046,
      "grad_norm": 0.8102736269735316,
      "learning_rate": 0.00019921372317658224,
      "loss": 12.7661,
      "step": 2525
    },
    {
      "epoch": 0.13755079536868758,
      "grad_norm": 0.9204667558381068,
      "learning_rate": 0.00019921261914386963,
      "loss": 12.6912,
      "step": 2526
    },
    {
      "epoch": 0.1376052493652706,
      "grad_norm": 0.7654727932975922,
      "learning_rate": 0.000199211514339663,
      "loss": 12.7644,
      "step": 2527
    },
    {
      "epoch": 0.13765970336185362,
      "grad_norm": 1.0129557668524476,
      "learning_rate": 0.00019921040876397103,
      "loss": 12.7797,
      "step": 2528
    },
    {
      "epoch": 0.13771415735843662,
      "grad_norm": 0.7456783582541698,
      "learning_rate": 0.0001992093024168023,
      "loss": 12.6864,
      "step": 2529
    },
    {
      "epoch": 0.13776861135501964,
      "grad_norm": 0.7895332665858877,
      "learning_rate": 0.00019920819529816534,
      "loss": 12.686,
      "step": 2530
    },
    {
      "epoch": 0.13782306535160266,
      "grad_norm": 0.7372087027065335,
      "learning_rate": 0.0001992070874080689,
      "loss": 12.7634,
      "step": 2531
    },
    {
      "epoch": 0.13787751934818565,
      "grad_norm": 0.8003564961950486,
      "learning_rate": 0.00019920597874652143,
      "loss": 12.6631,
      "step": 2532
    },
    {
      "epoch": 0.13793197334476867,
      "grad_norm": 0.7813288190540522,
      "learning_rate": 0.0001992048693135317,
      "loss": 12.6787,
      "step": 2533
    },
    {
      "epoch": 0.1379864273413517,
      "grad_norm": 0.7680793879946148,
      "learning_rate": 0.0001992037591091082,
      "loss": 12.772,
      "step": 2534
    },
    {
      "epoch": 0.13804088133793468,
      "grad_norm": 0.7710086521748657,
      "learning_rate": 0.00019920264813325965,
      "loss": 12.6736,
      "step": 2535
    },
    {
      "epoch": 0.1380953353345177,
      "grad_norm": 0.7414826352875737,
      "learning_rate": 0.0001992015363859947,
      "loss": 12.7004,
      "step": 2536
    },
    {
      "epoch": 0.13814978933110073,
      "grad_norm": 0.8210043929211484,
      "learning_rate": 0.00019920042386732196,
      "loss": 12.6852,
      "step": 2537
    },
    {
      "epoch": 0.13820424332768372,
      "grad_norm": 0.7534293641843519,
      "learning_rate": 0.0001991993105772501,
      "loss": 12.6567,
      "step": 2538
    },
    {
      "epoch": 0.13825869732426674,
      "grad_norm": 0.8809000111534994,
      "learning_rate": 0.00019919819651578776,
      "loss": 12.7414,
      "step": 2539
    },
    {
      "epoch": 0.13831315132084976,
      "grad_norm": 0.7685448480984839,
      "learning_rate": 0.0001991970816829436,
      "loss": 12.686,
      "step": 2540
    },
    {
      "epoch": 0.13836760531743278,
      "grad_norm": 0.7667664978318831,
      "learning_rate": 0.0001991959660787263,
      "loss": 12.6296,
      "step": 2541
    },
    {
      "epoch": 0.13842205931401577,
      "grad_norm": 0.810234524969098,
      "learning_rate": 0.00019919484970314452,
      "loss": 12.7798,
      "step": 2542
    },
    {
      "epoch": 0.1384765133105988,
      "grad_norm": 0.7663707870937774,
      "learning_rate": 0.00019919373255620695,
      "loss": 12.6499,
      "step": 2543
    },
    {
      "epoch": 0.1385309673071818,
      "grad_norm": 0.7614050036552538,
      "learning_rate": 0.0001991926146379223,
      "loss": 12.7253,
      "step": 2544
    },
    {
      "epoch": 0.1385854213037648,
      "grad_norm": 0.7614062286590654,
      "learning_rate": 0.00019919149594829926,
      "loss": 12.6519,
      "step": 2545
    },
    {
      "epoch": 0.13863987530034783,
      "grad_norm": 0.7781884061851773,
      "learning_rate": 0.00019919037648734647,
      "loss": 12.7314,
      "step": 2546
    },
    {
      "epoch": 0.13869432929693085,
      "grad_norm": 0.7678996328527208,
      "learning_rate": 0.0001991892562550727,
      "loss": 12.628,
      "step": 2547
    },
    {
      "epoch": 0.13874878329351384,
      "grad_norm": 0.8359242351768541,
      "learning_rate": 0.00019918813525148665,
      "loss": 12.6445,
      "step": 2548
    },
    {
      "epoch": 0.13880323729009686,
      "grad_norm": 0.772133255022356,
      "learning_rate": 0.000199187013476597,
      "loss": 12.7278,
      "step": 2549
    },
    {
      "epoch": 0.13885769128667988,
      "grad_norm": 0.9285555529661113,
      "learning_rate": 0.0001991858909304125,
      "loss": 12.6475,
      "step": 2550
    },
    {
      "epoch": 0.13891214528326287,
      "grad_norm": 0.748614685498673,
      "learning_rate": 0.00019918476761294191,
      "loss": 12.6584,
      "step": 2551
    },
    {
      "epoch": 0.1389665992798459,
      "grad_norm": 0.9228432707306571,
      "learning_rate": 0.00019918364352419391,
      "loss": 12.6811,
      "step": 2552
    },
    {
      "epoch": 0.1390210532764289,
      "grad_norm": 0.8046804415097579,
      "learning_rate": 0.00019918251866417729,
      "loss": 12.5896,
      "step": 2553
    },
    {
      "epoch": 0.1390755072730119,
      "grad_norm": 0.841468945990471,
      "learning_rate": 0.00019918139303290073,
      "loss": 12.5972,
      "step": 2554
    },
    {
      "epoch": 0.13912996126959493,
      "grad_norm": 1.0219679946081544,
      "learning_rate": 0.00019918026663037305,
      "loss": 12.4939,
      "step": 2555
    },
    {
      "epoch": 0.13918441526617795,
      "grad_norm": 0.7614948596107192,
      "learning_rate": 0.000199179139456603,
      "loss": 12.7957,
      "step": 2556
    },
    {
      "epoch": 0.13923886926276097,
      "grad_norm": 1.125763541075716,
      "learning_rate": 0.00019917801151159931,
      "loss": 12.6848,
      "step": 2557
    },
    {
      "epoch": 0.13929332325934396,
      "grad_norm": 0.7167558919471144,
      "learning_rate": 0.00019917688279537076,
      "loss": 12.6508,
      "step": 2558
    },
    {
      "epoch": 0.13934777725592698,
      "grad_norm": 1.0221540450131401,
      "learning_rate": 0.00019917575330792616,
      "loss": 12.6141,
      "step": 2559
    },
    {
      "epoch": 0.13940223125251,
      "grad_norm": 0.863351781333219,
      "learning_rate": 0.00019917462304927424,
      "loss": 12.7706,
      "step": 2560
    },
    {
      "epoch": 0.139456685249093,
      "grad_norm": 0.845744534563339,
      "learning_rate": 0.00019917349201942385,
      "loss": 12.8127,
      "step": 2561
    },
    {
      "epoch": 0.139511139245676,
      "grad_norm": 0.8801687590672354,
      "learning_rate": 0.00019917236021838375,
      "loss": 12.7274,
      "step": 2562
    },
    {
      "epoch": 0.13956559324225903,
      "grad_norm": 0.7298449140222769,
      "learning_rate": 0.00019917122764616274,
      "loss": 12.779,
      "step": 2563
    },
    {
      "epoch": 0.13962004723884203,
      "grad_norm": 0.7832122595156031,
      "learning_rate": 0.00019917009430276962,
      "loss": 12.5244,
      "step": 2564
    },
    {
      "epoch": 0.13967450123542505,
      "grad_norm": 0.7527806598249291,
      "learning_rate": 0.00019916896018821323,
      "loss": 12.5743,
      "step": 2565
    },
    {
      "epoch": 0.13972895523200807,
      "grad_norm": 0.7631874231541864,
      "learning_rate": 0.0001991678253025024,
      "loss": 12.6855,
      "step": 2566
    },
    {
      "epoch": 0.13978340922859106,
      "grad_norm": 0.7998953837645839,
      "learning_rate": 0.0001991666896456459,
      "loss": 12.6305,
      "step": 2567
    },
    {
      "epoch": 0.13983786322517408,
      "grad_norm": 0.7844095551718671,
      "learning_rate": 0.00019916555321765258,
      "loss": 12.7029,
      "step": 2568
    },
    {
      "epoch": 0.1398923172217571,
      "grad_norm": 0.761532138143132,
      "learning_rate": 0.0001991644160185313,
      "loss": 12.6633,
      "step": 2569
    },
    {
      "epoch": 0.1399467712183401,
      "grad_norm": 0.7509558404118452,
      "learning_rate": 0.0001991632780482909,
      "loss": 12.7437,
      "step": 2570
    },
    {
      "epoch": 0.14000122521492311,
      "grad_norm": 0.7299977982300054,
      "learning_rate": 0.0001991621393069402,
      "loss": 12.5211,
      "step": 2571
    },
    {
      "epoch": 0.14005567921150613,
      "grad_norm": 0.8211312042072629,
      "learning_rate": 0.0001991609997944881,
      "loss": 12.7196,
      "step": 2572
    },
    {
      "epoch": 0.14011013320808915,
      "grad_norm": 0.735678027919518,
      "learning_rate": 0.00019915985951094342,
      "loss": 12.6854,
      "step": 2573
    },
    {
      "epoch": 0.14016458720467215,
      "grad_norm": 0.7765696874719373,
      "learning_rate": 0.00019915871845631506,
      "loss": 12.6538,
      "step": 2574
    },
    {
      "epoch": 0.14021904120125517,
      "grad_norm": 0.7117218844333242,
      "learning_rate": 0.00019915757663061188,
      "loss": 12.6685,
      "step": 2575
    },
    {
      "epoch": 0.1402734951978382,
      "grad_norm": 0.8627660601894691,
      "learning_rate": 0.00019915643403384272,
      "loss": 12.7439,
      "step": 2576
    },
    {
      "epoch": 0.14032794919442118,
      "grad_norm": 0.7356120518935028,
      "learning_rate": 0.00019915529066601652,
      "loss": 12.8082,
      "step": 2577
    },
    {
      "epoch": 0.1403824031910042,
      "grad_norm": 0.7130187369102607,
      "learning_rate": 0.00019915414652714217,
      "loss": 12.7221,
      "step": 2578
    },
    {
      "epoch": 0.14043685718758722,
      "grad_norm": 0.7332688257439224,
      "learning_rate": 0.00019915300161722852,
      "loss": 12.7258,
      "step": 2579
    },
    {
      "epoch": 0.14049131118417021,
      "grad_norm": 0.7630906052367811,
      "learning_rate": 0.00019915185593628453,
      "loss": 12.6905,
      "step": 2580
    },
    {
      "epoch": 0.14054576518075323,
      "grad_norm": 0.9138139283846225,
      "learning_rate": 0.00019915070948431905,
      "loss": 12.6803,
      "step": 2581
    },
    {
      "epoch": 0.14060021917733626,
      "grad_norm": 0.7234398462032869,
      "learning_rate": 0.00019914956226134103,
      "loss": 12.669,
      "step": 2582
    },
    {
      "epoch": 0.14065467317391925,
      "grad_norm": 0.7252327764689702,
      "learning_rate": 0.0001991484142673594,
      "loss": 12.6976,
      "step": 2583
    },
    {
      "epoch": 0.14070912717050227,
      "grad_norm": 0.7898203683406123,
      "learning_rate": 0.00019914726550238307,
      "loss": 12.8086,
      "step": 2584
    },
    {
      "epoch": 0.1407635811670853,
      "grad_norm": 0.752966658793422,
      "learning_rate": 0.00019914611596642096,
      "loss": 12.6919,
      "step": 2585
    },
    {
      "epoch": 0.14081803516366828,
      "grad_norm": 0.9067939979456567,
      "learning_rate": 0.00019914496565948207,
      "loss": 12.6859,
      "step": 2586
    },
    {
      "epoch": 0.1408724891602513,
      "grad_norm": 0.715271796852438,
      "learning_rate": 0.00019914381458157525,
      "loss": 12.6533,
      "step": 2587
    },
    {
      "epoch": 0.14092694315683432,
      "grad_norm": 0.763482898726477,
      "learning_rate": 0.00019914266273270953,
      "loss": 12.5664,
      "step": 2588
    },
    {
      "epoch": 0.14098139715341731,
      "grad_norm": 0.7376861678712301,
      "learning_rate": 0.0001991415101128938,
      "loss": 12.6123,
      "step": 2589
    },
    {
      "epoch": 0.14103585115000034,
      "grad_norm": 0.8084686482048893,
      "learning_rate": 0.00019914035672213712,
      "loss": 12.6054,
      "step": 2590
    },
    {
      "epoch": 0.14109030514658336,
      "grad_norm": 0.8438341417341487,
      "learning_rate": 0.00019913920256044837,
      "loss": 12.7978,
      "step": 2591
    },
    {
      "epoch": 0.14114475914316638,
      "grad_norm": 0.7656913542268167,
      "learning_rate": 0.00019913804762783656,
      "loss": 12.6378,
      "step": 2592
    },
    {
      "epoch": 0.14119921313974937,
      "grad_norm": 0.7226525503659902,
      "learning_rate": 0.00019913689192431065,
      "loss": 12.6956,
      "step": 2593
    },
    {
      "epoch": 0.1412536671363324,
      "grad_norm": 0.7566159616987096,
      "learning_rate": 0.00019913573544987968,
      "loss": 12.7507,
      "step": 2594
    },
    {
      "epoch": 0.1413081211329154,
      "grad_norm": 0.9359513791113954,
      "learning_rate": 0.00019913457820455258,
      "loss": 12.7758,
      "step": 2595
    },
    {
      "epoch": 0.1413625751294984,
      "grad_norm": 0.7716699789726437,
      "learning_rate": 0.00019913342018833835,
      "loss": 12.5928,
      "step": 2596
    },
    {
      "epoch": 0.14141702912608142,
      "grad_norm": 0.855705424725422,
      "learning_rate": 0.00019913226140124608,
      "loss": 12.6545,
      "step": 2597
    },
    {
      "epoch": 0.14147148312266444,
      "grad_norm": 0.7477286757754348,
      "learning_rate": 0.00019913110184328466,
      "loss": 12.6477,
      "step": 2598
    },
    {
      "epoch": 0.14152593711924744,
      "grad_norm": 0.7656128177871533,
      "learning_rate": 0.0001991299415144632,
      "loss": 12.6694,
      "step": 2599
    },
    {
      "epoch": 0.14158039111583046,
      "grad_norm": 0.8201103526357508,
      "learning_rate": 0.00019912878041479067,
      "loss": 12.7311,
      "step": 2600
    },
    {
      "epoch": 0.14163484511241348,
      "grad_norm": 0.7757724450627341,
      "learning_rate": 0.00019912761854427616,
      "loss": 12.7812,
      "step": 2601
    },
    {
      "epoch": 0.14168929910899647,
      "grad_norm": 0.7652702562645756,
      "learning_rate": 0.00019912645590292865,
      "loss": 12.8233,
      "step": 2602
    },
    {
      "epoch": 0.1417437531055795,
      "grad_norm": 0.8294807319246501,
      "learning_rate": 0.00019912529249075718,
      "loss": 12.606,
      "step": 2603
    },
    {
      "epoch": 0.1417982071021625,
      "grad_norm": 0.7085809011050793,
      "learning_rate": 0.0001991241283077708,
      "loss": 12.6265,
      "step": 2604
    },
    {
      "epoch": 0.1418526610987455,
      "grad_norm": 0.7706834778729315,
      "learning_rate": 0.00019912296335397863,
      "loss": 12.7192,
      "step": 2605
    },
    {
      "epoch": 0.14190711509532852,
      "grad_norm": 0.6831567962305096,
      "learning_rate": 0.00019912179762938964,
      "loss": 12.5742,
      "step": 2606
    },
    {
      "epoch": 0.14196156909191154,
      "grad_norm": 0.7245976056716881,
      "learning_rate": 0.0001991206311340129,
      "loss": 12.7203,
      "step": 2607
    },
    {
      "epoch": 0.14201602308849456,
      "grad_norm": 0.7092967526585179,
      "learning_rate": 0.00019911946386785755,
      "loss": 12.7227,
      "step": 2608
    },
    {
      "epoch": 0.14207047708507756,
      "grad_norm": 0.8299215978940133,
      "learning_rate": 0.0001991182958309326,
      "loss": 12.6424,
      "step": 2609
    },
    {
      "epoch": 0.14212493108166058,
      "grad_norm": 0.7427920778837738,
      "learning_rate": 0.00019911712702324716,
      "loss": 12.7091,
      "step": 2610
    },
    {
      "epoch": 0.1421793850782436,
      "grad_norm": 0.8184591293143213,
      "learning_rate": 0.00019911595744481034,
      "loss": 12.7721,
      "step": 2611
    },
    {
      "epoch": 0.1422338390748266,
      "grad_norm": 0.8594294714337313,
      "learning_rate": 0.00019911478709563123,
      "loss": 12.7675,
      "step": 2612
    },
    {
      "epoch": 0.1422882930714096,
      "grad_norm": 0.7905177187484469,
      "learning_rate": 0.00019911361597571887,
      "loss": 12.7313,
      "step": 2613
    },
    {
      "epoch": 0.14234274706799263,
      "grad_norm": 0.6542339698773333,
      "learning_rate": 0.00019911244408508241,
      "loss": 12.6825,
      "step": 2614
    },
    {
      "epoch": 0.14239720106457562,
      "grad_norm": 0.6864872742719939,
      "learning_rate": 0.000199111271423731,
      "loss": 12.7059,
      "step": 2615
    },
    {
      "epoch": 0.14245165506115864,
      "grad_norm": 0.8509600880961752,
      "learning_rate": 0.0001991100979916737,
      "loss": 12.825,
      "step": 2616
    },
    {
      "epoch": 0.14250610905774166,
      "grad_norm": 0.8027983418211375,
      "learning_rate": 0.00019910892378891966,
      "loss": 12.7105,
      "step": 2617
    },
    {
      "epoch": 0.14256056305432466,
      "grad_norm": 0.7234317331840034,
      "learning_rate": 0.000199107748815478,
      "loss": 12.6505,
      "step": 2618
    },
    {
      "epoch": 0.14261501705090768,
      "grad_norm": 0.7664867974063245,
      "learning_rate": 0.0001991065730713579,
      "loss": 12.7388,
      "step": 2619
    },
    {
      "epoch": 0.1426694710474907,
      "grad_norm": 0.7636742822664422,
      "learning_rate": 0.00019910539655656844,
      "loss": 12.6586,
      "step": 2620
    },
    {
      "epoch": 0.1427239250440737,
      "grad_norm": 0.785599042442663,
      "learning_rate": 0.0001991042192711188,
      "loss": 12.6612,
      "step": 2621
    },
    {
      "epoch": 0.1427783790406567,
      "grad_norm": 0.6841416305877701,
      "learning_rate": 0.00019910304121501811,
      "loss": 12.5579,
      "step": 2622
    },
    {
      "epoch": 0.14283283303723973,
      "grad_norm": 0.7405822791815138,
      "learning_rate": 0.00019910186238827557,
      "loss": 12.7123,
      "step": 2623
    },
    {
      "epoch": 0.14288728703382275,
      "grad_norm": 0.7209279711717004,
      "learning_rate": 0.00019910068279090036,
      "loss": 12.6044,
      "step": 2624
    },
    {
      "epoch": 0.14294174103040574,
      "grad_norm": 0.8261997294453759,
      "learning_rate": 0.0001990995024229016,
      "loss": 12.6075,
      "step": 2625
    },
    {
      "epoch": 0.14299619502698876,
      "grad_norm": 0.7613534265460365,
      "learning_rate": 0.00019909832128428846,
      "loss": 12.6447,
      "step": 2626
    },
    {
      "epoch": 0.14305064902357179,
      "grad_norm": 0.7563704893104031,
      "learning_rate": 0.0001990971393750702,
      "loss": 12.5855,
      "step": 2627
    },
    {
      "epoch": 0.14310510302015478,
      "grad_norm": 0.7399588497910308,
      "learning_rate": 0.00019909595669525594,
      "loss": 12.6363,
      "step": 2628
    },
    {
      "epoch": 0.1431595570167378,
      "grad_norm": 0.7423730889180716,
      "learning_rate": 0.0001990947732448549,
      "loss": 12.623,
      "step": 2629
    },
    {
      "epoch": 0.14321401101332082,
      "grad_norm": 0.7780993727686799,
      "learning_rate": 0.00019909358902387626,
      "loss": 12.6195,
      "step": 2630
    },
    {
      "epoch": 0.1432684650099038,
      "grad_norm": 0.695759998723745,
      "learning_rate": 0.0001990924040323293,
      "loss": 12.6837,
      "step": 2631
    },
    {
      "epoch": 0.14332291900648683,
      "grad_norm": 0.7818291616707426,
      "learning_rate": 0.0001990912182702232,
      "loss": 12.7589,
      "step": 2632
    },
    {
      "epoch": 0.14337737300306985,
      "grad_norm": 0.728177780364699,
      "learning_rate": 0.0001990900317375671,
      "loss": 12.6438,
      "step": 2633
    },
    {
      "epoch": 0.14343182699965284,
      "grad_norm": 0.7192789143055213,
      "learning_rate": 0.00019908884443437037,
      "loss": 12.7923,
      "step": 2634
    },
    {
      "epoch": 0.14348628099623587,
      "grad_norm": 0.7560552245070992,
      "learning_rate": 0.00019908765636064213,
      "loss": 12.5752,
      "step": 2635
    },
    {
      "epoch": 0.14354073499281889,
      "grad_norm": 0.7917742140949682,
      "learning_rate": 0.00019908646751639166,
      "loss": 12.7673,
      "step": 2636
    },
    {
      "epoch": 0.14359518898940188,
      "grad_norm": 0.7889192817451162,
      "learning_rate": 0.00019908527790162822,
      "loss": 12.7232,
      "step": 2637
    },
    {
      "epoch": 0.1436496429859849,
      "grad_norm": 0.7590576115205909,
      "learning_rate": 0.000199084087516361,
      "loss": 12.629,
      "step": 2638
    },
    {
      "epoch": 0.14370409698256792,
      "grad_norm": 0.7968942636474549,
      "learning_rate": 0.00019908289636059933,
      "loss": 12.7596,
      "step": 2639
    },
    {
      "epoch": 0.14375855097915094,
      "grad_norm": 0.7118388249750108,
      "learning_rate": 0.00019908170443435244,
      "loss": 12.6103,
      "step": 2640
    },
    {
      "epoch": 0.14381300497573393,
      "grad_norm": 0.7676927176741307,
      "learning_rate": 0.0001990805117376296,
      "loss": 12.659,
      "step": 2641
    },
    {
      "epoch": 0.14386745897231695,
      "grad_norm": 0.74628843718605,
      "learning_rate": 0.0001990793182704401,
      "loss": 12.6684,
      "step": 2642
    },
    {
      "epoch": 0.14392191296889997,
      "grad_norm": 0.765142973884477,
      "learning_rate": 0.00019907812403279314,
      "loss": 12.6766,
      "step": 2643
    },
    {
      "epoch": 0.14397636696548297,
      "grad_norm": 1.120038400913752,
      "learning_rate": 0.00019907692902469814,
      "loss": 12.7644,
      "step": 2644
    },
    {
      "epoch": 0.144030820962066,
      "grad_norm": 0.8613421582516949,
      "learning_rate": 0.0001990757332461643,
      "loss": 12.7654,
      "step": 2645
    },
    {
      "epoch": 0.144085274958649,
      "grad_norm": 0.7597908331512231,
      "learning_rate": 0.00019907453669720096,
      "loss": 12.732,
      "step": 2646
    },
    {
      "epoch": 0.144139728955232,
      "grad_norm": 0.9086438951971259,
      "learning_rate": 0.0001990733393778174,
      "loss": 12.6952,
      "step": 2647
    },
    {
      "epoch": 0.14419418295181502,
      "grad_norm": 0.735107309446148,
      "learning_rate": 0.00019907214128802293,
      "loss": 12.6305,
      "step": 2648
    },
    {
      "epoch": 0.14424863694839804,
      "grad_norm": 0.8510749009659634,
      "learning_rate": 0.0001990709424278269,
      "loss": 12.6497,
      "step": 2649
    },
    {
      "epoch": 0.14430309094498103,
      "grad_norm": 2.553487336315287,
      "learning_rate": 0.00019906974279723858,
      "loss": 12.859,
      "step": 2650
    },
    {
      "epoch": 0.14435754494156405,
      "grad_norm": 0.8497281216272348,
      "learning_rate": 0.00019906854239626733,
      "loss": 12.6034,
      "step": 2651
    },
    {
      "epoch": 0.14441199893814707,
      "grad_norm": 0.7741178450896294,
      "learning_rate": 0.0001990673412249225,
      "loss": 12.7448,
      "step": 2652
    },
    {
      "epoch": 0.14446645293473007,
      "grad_norm": 0.8050047338057631,
      "learning_rate": 0.00019906613928321338,
      "loss": 12.5269,
      "step": 2653
    },
    {
      "epoch": 0.1445209069313131,
      "grad_norm": 0.764721188085812,
      "learning_rate": 0.00019906493657114935,
      "loss": 12.8101,
      "step": 2654
    },
    {
      "epoch": 0.1445753609278961,
      "grad_norm": 0.9205425051456191,
      "learning_rate": 0.0001990637330887398,
      "loss": 12.6569,
      "step": 2655
    },
    {
      "epoch": 0.1446298149244791,
      "grad_norm": 0.7721467979457826,
      "learning_rate": 0.00019906252883599402,
      "loss": 12.7271,
      "step": 2656
    },
    {
      "epoch": 0.14468426892106212,
      "grad_norm": 0.7858920669134861,
      "learning_rate": 0.0001990613238129214,
      "loss": 12.8252,
      "step": 2657
    },
    {
      "epoch": 0.14473872291764514,
      "grad_norm": 0.765720003106791,
      "learning_rate": 0.0001990601180195313,
      "loss": 12.7792,
      "step": 2658
    },
    {
      "epoch": 0.14479317691422816,
      "grad_norm": 0.7591642681701061,
      "learning_rate": 0.00019905891145583312,
      "loss": 12.6342,
      "step": 2659
    },
    {
      "epoch": 0.14484763091081115,
      "grad_norm": 0.7738866144071866,
      "learning_rate": 0.00019905770412183626,
      "loss": 12.7038,
      "step": 2660
    },
    {
      "epoch": 0.14490208490739417,
      "grad_norm": 0.7751819170403272,
      "learning_rate": 0.00019905649601755006,
      "loss": 12.7029,
      "step": 2661
    },
    {
      "epoch": 0.1449565389039772,
      "grad_norm": 1.037253868229172,
      "learning_rate": 0.00019905528714298393,
      "loss": 12.6876,
      "step": 2662
    },
    {
      "epoch": 0.1450109929005602,
      "grad_norm": 0.8228831140879636,
      "learning_rate": 0.00019905407749814727,
      "loss": 12.6693,
      "step": 2663
    },
    {
      "epoch": 0.1450654468971432,
      "grad_norm": 0.7472181487251265,
      "learning_rate": 0.0001990528670830495,
      "loss": 12.5575,
      "step": 2664
    },
    {
      "epoch": 0.14511990089372623,
      "grad_norm": 0.8676633248694123,
      "learning_rate": 0.00019905165589770003,
      "loss": 12.6504,
      "step": 2665
    },
    {
      "epoch": 0.14517435489030922,
      "grad_norm": 0.833166831577227,
      "learning_rate": 0.00019905044394210827,
      "loss": 12.8559,
      "step": 2666
    },
    {
      "epoch": 0.14522880888689224,
      "grad_norm": 0.7431818143738881,
      "learning_rate": 0.00019904923121628367,
      "loss": 12.599,
      "step": 2667
    },
    {
      "epoch": 0.14528326288347526,
      "grad_norm": 0.9159990154597104,
      "learning_rate": 0.0001990480177202356,
      "loss": 12.7133,
      "step": 2668
    },
    {
      "epoch": 0.14533771688005825,
      "grad_norm": 0.7690112766314912,
      "learning_rate": 0.00019904680345397355,
      "loss": 12.7107,
      "step": 2669
    },
    {
      "epoch": 0.14539217087664127,
      "grad_norm": 0.7763570974348,
      "learning_rate": 0.00019904558841750696,
      "loss": 12.797,
      "step": 2670
    },
    {
      "epoch": 0.1454466248732243,
      "grad_norm": 0.885475603482203,
      "learning_rate": 0.00019904437261084526,
      "loss": 12.648,
      "step": 2671
    },
    {
      "epoch": 0.1455010788698073,
      "grad_norm": 0.7985814589081149,
      "learning_rate": 0.00019904315603399788,
      "loss": 12.9738,
      "step": 2672
    },
    {
      "epoch": 0.1455555328663903,
      "grad_norm": 0.937349046219652,
      "learning_rate": 0.00019904193868697432,
      "loss": 12.7084,
      "step": 2673
    },
    {
      "epoch": 0.14560998686297333,
      "grad_norm": 0.7784287617162705,
      "learning_rate": 0.00019904072056978404,
      "loss": 12.6306,
      "step": 2674
    },
    {
      "epoch": 0.14566444085955635,
      "grad_norm": 0.6986676816712994,
      "learning_rate": 0.00019903950168243654,
      "loss": 12.6791,
      "step": 2675
    },
    {
      "epoch": 0.14571889485613934,
      "grad_norm": 0.7982549576873136,
      "learning_rate": 0.0001990382820249412,
      "loss": 12.5907,
      "step": 2676
    },
    {
      "epoch": 0.14577334885272236,
      "grad_norm": 0.7865765018957946,
      "learning_rate": 0.00019903706159730763,
      "loss": 12.683,
      "step": 2677
    },
    {
      "epoch": 0.14582780284930538,
      "grad_norm": 0.7727965652112779,
      "learning_rate": 0.00019903584039954525,
      "loss": 12.73,
      "step": 2678
    },
    {
      "epoch": 0.14588225684588838,
      "grad_norm": 0.7201995605130986,
      "learning_rate": 0.00019903461843166352,
      "loss": 12.5938,
      "step": 2679
    },
    {
      "epoch": 0.1459367108424714,
      "grad_norm": 0.776229206051119,
      "learning_rate": 0.00019903339569367202,
      "loss": 12.6367,
      "step": 2680
    },
    {
      "epoch": 0.14599116483905442,
      "grad_norm": 0.7445473575011761,
      "learning_rate": 0.00019903217218558025,
      "loss": 12.7577,
      "step": 2681
    },
    {
      "epoch": 0.1460456188356374,
      "grad_norm": 0.7552656467499095,
      "learning_rate": 0.0001990309479073977,
      "loss": 12.7018,
      "step": 2682
    },
    {
      "epoch": 0.14610007283222043,
      "grad_norm": 0.7021799060374931,
      "learning_rate": 0.00019902972285913386,
      "loss": 12.5492,
      "step": 2683
    },
    {
      "epoch": 0.14615452682880345,
      "grad_norm": 0.7101796742359953,
      "learning_rate": 0.0001990284970407983,
      "loss": 12.6654,
      "step": 2684
    },
    {
      "epoch": 0.14620898082538644,
      "grad_norm": 0.7770953225119109,
      "learning_rate": 0.00019902727045240055,
      "loss": 12.7807,
      "step": 2685
    },
    {
      "epoch": 0.14626343482196946,
      "grad_norm": 0.8205011594444994,
      "learning_rate": 0.0001990260430939501,
      "loss": 12.7381,
      "step": 2686
    },
    {
      "epoch": 0.14631788881855248,
      "grad_norm": 0.8011777697339707,
      "learning_rate": 0.00019902481496545657,
      "loss": 12.6545,
      "step": 2687
    },
    {
      "epoch": 0.14637234281513548,
      "grad_norm": 0.7561686468407363,
      "learning_rate": 0.0001990235860669295,
      "loss": 12.7167,
      "step": 2688
    },
    {
      "epoch": 0.1464267968117185,
      "grad_norm": 0.7239528910960851,
      "learning_rate": 0.00019902235639837837,
      "loss": 12.7289,
      "step": 2689
    },
    {
      "epoch": 0.14648125080830152,
      "grad_norm": 0.7739801417470364,
      "learning_rate": 0.0001990211259598128,
      "loss": 12.7053,
      "step": 2690
    },
    {
      "epoch": 0.14653570480488454,
      "grad_norm": 0.8893441290926933,
      "learning_rate": 0.00019901989475124235,
      "loss": 12.7693,
      "step": 2691
    },
    {
      "epoch": 0.14659015880146753,
      "grad_norm": 0.7881993519858137,
      "learning_rate": 0.00019901866277267658,
      "loss": 12.7999,
      "step": 2692
    },
    {
      "epoch": 0.14664461279805055,
      "grad_norm": 0.7960471770020839,
      "learning_rate": 0.0001990174300241251,
      "loss": 12.7491,
      "step": 2693
    },
    {
      "epoch": 0.14669906679463357,
      "grad_norm": 0.7753083056275317,
      "learning_rate": 0.00019901619650559749,
      "loss": 12.7001,
      "step": 2694
    },
    {
      "epoch": 0.14675352079121656,
      "grad_norm": 0.7696722499154163,
      "learning_rate": 0.00019901496221710332,
      "loss": 12.6874,
      "step": 2695
    },
    {
      "epoch": 0.14680797478779958,
      "grad_norm": 0.767633402938913,
      "learning_rate": 0.00019901372715865217,
      "loss": 12.6353,
      "step": 2696
    },
    {
      "epoch": 0.1468624287843826,
      "grad_norm": 0.7438423585527448,
      "learning_rate": 0.0001990124913302537,
      "loss": 12.7425,
      "step": 2697
    },
    {
      "epoch": 0.1469168827809656,
      "grad_norm": 0.7757341838499866,
      "learning_rate": 0.0001990112547319175,
      "loss": 12.8279,
      "step": 2698
    },
    {
      "epoch": 0.14697133677754862,
      "grad_norm": 0.8258565883469542,
      "learning_rate": 0.00019901001736365317,
      "loss": 12.6837,
      "step": 2699
    },
    {
      "epoch": 0.14702579077413164,
      "grad_norm": 0.8004203175040449,
      "learning_rate": 0.00019900877922547034,
      "loss": 12.7775,
      "step": 2700
    },
    {
      "epoch": 0.14708024477071463,
      "grad_norm": 0.7947067824549463,
      "learning_rate": 0.00019900754031737866,
      "loss": 12.7518,
      "step": 2701
    },
    {
      "epoch": 0.14713469876729765,
      "grad_norm": 0.9069636423116467,
      "learning_rate": 0.00019900630063938773,
      "loss": 12.7515,
      "step": 2702
    },
    {
      "epoch": 0.14718915276388067,
      "grad_norm": 0.7338969251876433,
      "learning_rate": 0.00019900506019150717,
      "loss": 12.6569,
      "step": 2703
    },
    {
      "epoch": 0.14724360676046366,
      "grad_norm": 0.747056261107894,
      "learning_rate": 0.00019900381897374668,
      "loss": 12.6114,
      "step": 2704
    },
    {
      "epoch": 0.14729806075704668,
      "grad_norm": 0.792429734575613,
      "learning_rate": 0.0001990025769861159,
      "loss": 12.7098,
      "step": 2705
    },
    {
      "epoch": 0.1473525147536297,
      "grad_norm": 0.7958552148528896,
      "learning_rate": 0.0001990013342286245,
      "loss": 12.6546,
      "step": 2706
    },
    {
      "epoch": 0.14740696875021272,
      "grad_norm": 0.7635541867716035,
      "learning_rate": 0.00019900009070128208,
      "loss": 12.6584,
      "step": 2707
    },
    {
      "epoch": 0.14746142274679572,
      "grad_norm": 0.820738536195993,
      "learning_rate": 0.00019899884640409837,
      "loss": 12.8891,
      "step": 2708
    },
    {
      "epoch": 0.14751587674337874,
      "grad_norm": 0.678436148866974,
      "learning_rate": 0.00019899760133708304,
      "loss": 12.7063,
      "step": 2709
    },
    {
      "epoch": 0.14757033073996176,
      "grad_norm": 0.7338891818564695,
      "learning_rate": 0.00019899635550024573,
      "loss": 12.6804,
      "step": 2710
    },
    {
      "epoch": 0.14762478473654475,
      "grad_norm": 0.7677957679961459,
      "learning_rate": 0.0001989951088935962,
      "loss": 12.5545,
      "step": 2711
    },
    {
      "epoch": 0.14767923873312777,
      "grad_norm": 0.7455124550308035,
      "learning_rate": 0.00019899386151714407,
      "loss": 12.7571,
      "step": 2712
    },
    {
      "epoch": 0.1477336927297108,
      "grad_norm": 0.7554064266083648,
      "learning_rate": 0.00019899261337089907,
      "loss": 12.7199,
      "step": 2713
    },
    {
      "epoch": 0.14778814672629378,
      "grad_norm": 0.7000666033089813,
      "learning_rate": 0.0001989913644548709,
      "loss": 12.7881,
      "step": 2714
    },
    {
      "epoch": 0.1478426007228768,
      "grad_norm": 0.7409220000487262,
      "learning_rate": 0.00019899011476906932,
      "loss": 12.6374,
      "step": 2715
    },
    {
      "epoch": 0.14789705471945983,
      "grad_norm": 0.7127590442514861,
      "learning_rate": 0.00019898886431350397,
      "loss": 12.6687,
      "step": 2716
    },
    {
      "epoch": 0.14795150871604282,
      "grad_norm": 0.8103966832252935,
      "learning_rate": 0.0001989876130881846,
      "loss": 12.6737,
      "step": 2717
    },
    {
      "epoch": 0.14800596271262584,
      "grad_norm": 0.7294411546682413,
      "learning_rate": 0.000198986361093121,
      "loss": 12.6861,
      "step": 2718
    },
    {
      "epoch": 0.14806041670920886,
      "grad_norm": 0.7964709794829142,
      "learning_rate": 0.0001989851083283228,
      "loss": 12.5385,
      "step": 2719
    },
    {
      "epoch": 0.14811487070579185,
      "grad_norm": 0.7113450423677704,
      "learning_rate": 0.00019898385479379986,
      "loss": 12.601,
      "step": 2720
    },
    {
      "epoch": 0.14816932470237487,
      "grad_norm": 0.7822473870313964,
      "learning_rate": 0.00019898260048956183,
      "loss": 12.603,
      "step": 2721
    },
    {
      "epoch": 0.1482237786989579,
      "grad_norm": 0.8135540599153726,
      "learning_rate": 0.0001989813454156185,
      "loss": 12.6545,
      "step": 2722
    },
    {
      "epoch": 0.14827823269554088,
      "grad_norm": 0.8233043530388209,
      "learning_rate": 0.00019898008957197966,
      "loss": 12.7226,
      "step": 2723
    },
    {
      "epoch": 0.1483326866921239,
      "grad_norm": 0.9156143678477184,
      "learning_rate": 0.000198978832958655,
      "loss": 12.7056,
      "step": 2724
    },
    {
      "epoch": 0.14838714068870693,
      "grad_norm": 0.810162685105311,
      "learning_rate": 0.00019897757557565436,
      "loss": 12.682,
      "step": 2725
    },
    {
      "epoch": 0.14844159468528995,
      "grad_norm": 0.8342449403585397,
      "learning_rate": 0.00019897631742298746,
      "loss": 12.7605,
      "step": 2726
    },
    {
      "epoch": 0.14849604868187294,
      "grad_norm": 0.7489492357339741,
      "learning_rate": 0.00019897505850066414,
      "loss": 12.6675,
      "step": 2727
    },
    {
      "epoch": 0.14855050267845596,
      "grad_norm": 0.8945862280358247,
      "learning_rate": 0.00019897379880869418,
      "loss": 12.7221,
      "step": 2728
    },
    {
      "epoch": 0.14860495667503898,
      "grad_norm": 0.8578394375204317,
      "learning_rate": 0.00019897253834708735,
      "loss": 12.6227,
      "step": 2729
    },
    {
      "epoch": 0.14865941067162197,
      "grad_norm": 0.8320179117170514,
      "learning_rate": 0.00019897127711585347,
      "loss": 12.7534,
      "step": 2730
    },
    {
      "epoch": 0.148713864668205,
      "grad_norm": 0.913571833439365,
      "learning_rate": 0.00019897001511500232,
      "loss": 12.6558,
      "step": 2731
    },
    {
      "epoch": 0.148768318664788,
      "grad_norm": 0.7032303064815117,
      "learning_rate": 0.00019896875234454378,
      "loss": 12.6742,
      "step": 2732
    },
    {
      "epoch": 0.148822772661371,
      "grad_norm": 0.863979360450013,
      "learning_rate": 0.0001989674888044876,
      "loss": 12.6833,
      "step": 2733
    },
    {
      "epoch": 0.14887722665795403,
      "grad_norm": 0.8477605359647903,
      "learning_rate": 0.00019896622449484363,
      "loss": 12.724,
      "step": 2734
    },
    {
      "epoch": 0.14893168065453705,
      "grad_norm": 0.7009812417053154,
      "learning_rate": 0.00019896495941562167,
      "loss": 12.477,
      "step": 2735
    },
    {
      "epoch": 0.14898613465112004,
      "grad_norm": 1.1028258025088578,
      "learning_rate": 0.00019896369356683165,
      "loss": 12.7799,
      "step": 2736
    },
    {
      "epoch": 0.14904058864770306,
      "grad_norm": 0.823652454643868,
      "learning_rate": 0.00019896242694848333,
      "loss": 12.7377,
      "step": 2737
    },
    {
      "epoch": 0.14909504264428608,
      "grad_norm": 0.9733281948709586,
      "learning_rate": 0.00019896115956058655,
      "loss": 12.7747,
      "step": 2738
    },
    {
      "epoch": 0.14914949664086907,
      "grad_norm": 0.7976004474814257,
      "learning_rate": 0.00019895989140315123,
      "loss": 12.8456,
      "step": 2739
    },
    {
      "epoch": 0.1492039506374521,
      "grad_norm": 0.9644150577357924,
      "learning_rate": 0.0001989586224761872,
      "loss": 12.7945,
      "step": 2740
    },
    {
      "epoch": 0.1492584046340351,
      "grad_norm": 0.8759892203051738,
      "learning_rate": 0.0001989573527797043,
      "loss": 12.7189,
      "step": 2741
    },
    {
      "epoch": 0.14931285863061813,
      "grad_norm": 0.8336760294714186,
      "learning_rate": 0.00019895608231371246,
      "loss": 12.6849,
      "step": 2742
    },
    {
      "epoch": 0.14936731262720113,
      "grad_norm": 0.8626112894874401,
      "learning_rate": 0.0001989548110782215,
      "loss": 12.733,
      "step": 2743
    },
    {
      "epoch": 0.14942176662378415,
      "grad_norm": 0.7318897808235327,
      "learning_rate": 0.00019895353907324137,
      "loss": 12.6482,
      "step": 2744
    },
    {
      "epoch": 0.14947622062036717,
      "grad_norm": 0.9435418862621813,
      "learning_rate": 0.0001989522662987819,
      "loss": 12.788,
      "step": 2745
    },
    {
      "epoch": 0.14953067461695016,
      "grad_norm": 0.8022696319211658,
      "learning_rate": 0.000198950992754853,
      "loss": 12.438,
      "step": 2746
    },
    {
      "epoch": 0.14958512861353318,
      "grad_norm": 0.7575661421408524,
      "learning_rate": 0.00019894971844146463,
      "loss": 12.7393,
      "step": 2747
    },
    {
      "epoch": 0.1496395826101162,
      "grad_norm": 0.7805792001603048,
      "learning_rate": 0.00019894844335862662,
      "loss": 12.657,
      "step": 2748
    },
    {
      "epoch": 0.1496940366066992,
      "grad_norm": 0.8126920101810677,
      "learning_rate": 0.00019894716750634892,
      "loss": 12.8064,
      "step": 2749
    },
    {
      "epoch": 0.1497484906032822,
      "grad_norm": 0.8474201711084006,
      "learning_rate": 0.00019894589088464146,
      "loss": 12.754,
      "step": 2750
    },
    {
      "epoch": 0.14980294459986523,
      "grad_norm": 0.7755472736544625,
      "learning_rate": 0.00019894461349351415,
      "loss": 12.822,
      "step": 2751
    },
    {
      "epoch": 0.14985739859644823,
      "grad_norm": 0.7593805950832129,
      "learning_rate": 0.00019894333533297694,
      "loss": 12.6108,
      "step": 2752
    },
    {
      "epoch": 0.14991185259303125,
      "grad_norm": 0.7520301260416404,
      "learning_rate": 0.0001989420564030398,
      "loss": 12.6183,
      "step": 2753
    },
    {
      "epoch": 0.14996630658961427,
      "grad_norm": 0.749234270126392,
      "learning_rate": 0.0001989407767037126,
      "loss": 12.6935,
      "step": 2754
    },
    {
      "epoch": 0.15002076058619726,
      "grad_norm": 0.8756663060744522,
      "learning_rate": 0.00019893949623500534,
      "loss": 12.7225,
      "step": 2755
    },
    {
      "epoch": 0.15007521458278028,
      "grad_norm": 0.7676286733455003,
      "learning_rate": 0.00019893821499692793,
      "loss": 12.7744,
      "step": 2756
    },
    {
      "epoch": 0.1501296685793633,
      "grad_norm": 0.7605119624902635,
      "learning_rate": 0.0001989369329894904,
      "loss": 12.6357,
      "step": 2757
    },
    {
      "epoch": 0.15018412257594632,
      "grad_norm": 0.8580930249417181,
      "learning_rate": 0.00019893565021270268,
      "loss": 12.6745,
      "step": 2758
    },
    {
      "epoch": 0.15023857657252931,
      "grad_norm": 0.7583813012738259,
      "learning_rate": 0.00019893436666657474,
      "loss": 12.7093,
      "step": 2759
    },
    {
      "epoch": 0.15029303056911233,
      "grad_norm": 0.8563488809263123,
      "learning_rate": 0.0001989330823511166,
      "loss": 12.7741,
      "step": 2760
    },
    {
      "epoch": 0.15034748456569536,
      "grad_norm": 0.7543344960375925,
      "learning_rate": 0.00019893179726633822,
      "loss": 12.6322,
      "step": 2761
    },
    {
      "epoch": 0.15040193856227835,
      "grad_norm": 0.8365295252292132,
      "learning_rate": 0.0001989305114122496,
      "loss": 12.6755,
      "step": 2762
    },
    {
      "epoch": 0.15045639255886137,
      "grad_norm": 0.717356381989756,
      "learning_rate": 0.00019892922478886068,
      "loss": 12.6843,
      "step": 2763
    },
    {
      "epoch": 0.1505108465554444,
      "grad_norm": 0.7970472395126367,
      "learning_rate": 0.00019892793739618157,
      "loss": 12.7954,
      "step": 2764
    },
    {
      "epoch": 0.15056530055202738,
      "grad_norm": 0.8229387020294725,
      "learning_rate": 0.0001989266492342222,
      "loss": 12.6806,
      "step": 2765
    },
    {
      "epoch": 0.1506197545486104,
      "grad_norm": 0.7299889919023793,
      "learning_rate": 0.00019892536030299262,
      "loss": 12.7584,
      "step": 2766
    },
    {
      "epoch": 0.15067420854519342,
      "grad_norm": 0.8202907454446939,
      "learning_rate": 0.00019892407060250286,
      "loss": 12.8171,
      "step": 2767
    },
    {
      "epoch": 0.15072866254177641,
      "grad_norm": 0.7692958776460064,
      "learning_rate": 0.00019892278013276292,
      "loss": 12.7063,
      "step": 2768
    },
    {
      "epoch": 0.15078311653835944,
      "grad_norm": 0.7407039841520368,
      "learning_rate": 0.0001989214888937829,
      "loss": 12.6678,
      "step": 2769
    },
    {
      "epoch": 0.15083757053494246,
      "grad_norm": 0.6944323470913601,
      "learning_rate": 0.00019892019688557273,
      "loss": 12.6697,
      "step": 2770
    },
    {
      "epoch": 0.15089202453152545,
      "grad_norm": 0.9503595753704032,
      "learning_rate": 0.00019891890410814257,
      "loss": 12.6659,
      "step": 2771
    },
    {
      "epoch": 0.15094647852810847,
      "grad_norm": 0.7002113714765726,
      "learning_rate": 0.0001989176105615024,
      "loss": 12.5938,
      "step": 2772
    },
    {
      "epoch": 0.1510009325246915,
      "grad_norm": 0.7207990649669148,
      "learning_rate": 0.0001989163162456623,
      "loss": 12.6238,
      "step": 2773
    },
    {
      "epoch": 0.1510553865212745,
      "grad_norm": 0.7609110684132062,
      "learning_rate": 0.00019891502116063233,
      "loss": 12.5131,
      "step": 2774
    },
    {
      "epoch": 0.1511098405178575,
      "grad_norm": 0.7841808488718762,
      "learning_rate": 0.00019891372530642256,
      "loss": 12.7415,
      "step": 2775
    },
    {
      "epoch": 0.15116429451444052,
      "grad_norm": 0.803554084916481,
      "learning_rate": 0.0001989124286830431,
      "loss": 12.848,
      "step": 2776
    },
    {
      "epoch": 0.15121874851102354,
      "grad_norm": 0.7090519416935699,
      "learning_rate": 0.00019891113129050402,
      "loss": 12.6934,
      "step": 2777
    },
    {
      "epoch": 0.15127320250760654,
      "grad_norm": 0.9183418134038173,
      "learning_rate": 0.00019890983312881538,
      "loss": 12.6013,
      "step": 2778
    },
    {
      "epoch": 0.15132765650418956,
      "grad_norm": 0.6772283912015566,
      "learning_rate": 0.00019890853419798728,
      "loss": 12.6931,
      "step": 2779
    },
    {
      "epoch": 0.15138211050077258,
      "grad_norm": 0.7876816429288338,
      "learning_rate": 0.0001989072344980298,
      "loss": 12.753,
      "step": 2780
    },
    {
      "epoch": 0.15143656449735557,
      "grad_norm": 0.6960728512524404,
      "learning_rate": 0.00019890593402895312,
      "loss": 12.606,
      "step": 2781
    },
    {
      "epoch": 0.1514910184939386,
      "grad_norm": 0.7555357427749649,
      "learning_rate": 0.00019890463279076731,
      "loss": 12.5668,
      "step": 2782
    },
    {
      "epoch": 0.1515454724905216,
      "grad_norm": 0.7010977415108224,
      "learning_rate": 0.00019890333078348248,
      "loss": 12.6256,
      "step": 2783
    },
    {
      "epoch": 0.1515999264871046,
      "grad_norm": 0.7232246106573289,
      "learning_rate": 0.00019890202800710877,
      "loss": 12.7359,
      "step": 2784
    },
    {
      "epoch": 0.15165438048368762,
      "grad_norm": 0.809163964335571,
      "learning_rate": 0.0001989007244616563,
      "loss": 12.8173,
      "step": 2785
    },
    {
      "epoch": 0.15170883448027064,
      "grad_norm": 0.790540688648389,
      "learning_rate": 0.0001988994201471352,
      "loss": 12.5647,
      "step": 2786
    },
    {
      "epoch": 0.15176328847685364,
      "grad_norm": 0.6886477145769496,
      "learning_rate": 0.00019889811506355564,
      "loss": 12.5209,
      "step": 2787
    },
    {
      "epoch": 0.15181774247343666,
      "grad_norm": 0.6892518048460101,
      "learning_rate": 0.00019889680921092776,
      "loss": 12.5624,
      "step": 2788
    },
    {
      "epoch": 0.15187219647001968,
      "grad_norm": 0.7063700559267442,
      "learning_rate": 0.0001988955025892617,
      "loss": 12.6218,
      "step": 2789
    },
    {
      "epoch": 0.15192665046660267,
      "grad_norm": 0.7893300858011191,
      "learning_rate": 0.0001988941951985676,
      "loss": 12.641,
      "step": 2790
    },
    {
      "epoch": 0.1519811044631857,
      "grad_norm": 0.7314835383366215,
      "learning_rate": 0.0001988928870388557,
      "loss": 12.5973,
      "step": 2791
    },
    {
      "epoch": 0.1520355584597687,
      "grad_norm": 0.7727702514239536,
      "learning_rate": 0.00019889157811013607,
      "loss": 12.6172,
      "step": 2792
    },
    {
      "epoch": 0.15209001245635173,
      "grad_norm": 0.7357199420721806,
      "learning_rate": 0.000198890268412419,
      "loss": 12.6283,
      "step": 2793
    },
    {
      "epoch": 0.15214446645293472,
      "grad_norm": 0.7029293740093456,
      "learning_rate": 0.00019888895794571457,
      "loss": 12.7138,
      "step": 2794
    },
    {
      "epoch": 0.15219892044951774,
      "grad_norm": 0.8140800206572754,
      "learning_rate": 0.00019888764671003304,
      "loss": 12.7915,
      "step": 2795
    },
    {
      "epoch": 0.15225337444610076,
      "grad_norm": 0.7189172652761978,
      "learning_rate": 0.0001988863347053846,
      "loss": 12.6233,
      "step": 2796
    },
    {
      "epoch": 0.15230782844268376,
      "grad_norm": 0.6867836990626499,
      "learning_rate": 0.00019888502193177944,
      "loss": 12.671,
      "step": 2797
    },
    {
      "epoch": 0.15236228243926678,
      "grad_norm": 0.7207526699098658,
      "learning_rate": 0.00019888370838922774,
      "loss": 12.6913,
      "step": 2798
    },
    {
      "epoch": 0.1524167364358498,
      "grad_norm": 0.834167090031831,
      "learning_rate": 0.00019888239407773973,
      "loss": 12.5075,
      "step": 2799
    },
    {
      "epoch": 0.1524711904324328,
      "grad_norm": 0.7527501201976892,
      "learning_rate": 0.00019888107899732567,
      "loss": 12.5776,
      "step": 2800
    },
    {
      "epoch": 0.1525256444290158,
      "grad_norm": 0.705797037985442,
      "learning_rate": 0.00019887976314799576,
      "loss": 12.5589,
      "step": 2801
    },
    {
      "epoch": 0.15258009842559883,
      "grad_norm": 0.8131126845625408,
      "learning_rate": 0.00019887844652976023,
      "loss": 12.9504,
      "step": 2802
    },
    {
      "epoch": 0.15263455242218182,
      "grad_norm": 0.7197483465561019,
      "learning_rate": 0.00019887712914262932,
      "loss": 12.6358,
      "step": 2803
    },
    {
      "epoch": 0.15268900641876484,
      "grad_norm": 0.7211535406910872,
      "learning_rate": 0.00019887581098661326,
      "loss": 12.6082,
      "step": 2804
    },
    {
      "epoch": 0.15274346041534786,
      "grad_norm": 0.7153783931890983,
      "learning_rate": 0.0001988744920617223,
      "loss": 12.6324,
      "step": 2805
    },
    {
      "epoch": 0.15279791441193086,
      "grad_norm": 0.8625198065469974,
      "learning_rate": 0.00019887317236796673,
      "loss": 12.6345,
      "step": 2806
    },
    {
      "epoch": 0.15285236840851388,
      "grad_norm": 0.7268672211214469,
      "learning_rate": 0.00019887185190535676,
      "loss": 12.5933,
      "step": 2807
    },
    {
      "epoch": 0.1529068224050969,
      "grad_norm": 0.7191003964132267,
      "learning_rate": 0.00019887053067390271,
      "loss": 12.718,
      "step": 2808
    },
    {
      "epoch": 0.15296127640167992,
      "grad_norm": 0.9156558752399968,
      "learning_rate": 0.00019886920867361486,
      "loss": 12.5851,
      "step": 2809
    },
    {
      "epoch": 0.1530157303982629,
      "grad_norm": 0.7778164323577543,
      "learning_rate": 0.00019886788590450343,
      "loss": 12.5485,
      "step": 2810
    },
    {
      "epoch": 0.15307018439484593,
      "grad_norm": 0.7567328985770456,
      "learning_rate": 0.00019886656236657875,
      "loss": 12.7041,
      "step": 2811
    },
    {
      "epoch": 0.15312463839142895,
      "grad_norm": 0.7678411182063787,
      "learning_rate": 0.00019886523805985108,
      "loss": 12.6737,
      "step": 2812
    },
    {
      "epoch": 0.15317909238801194,
      "grad_norm": 0.8627798690144588,
      "learning_rate": 0.0001988639129843308,
      "loss": 12.6411,
      "step": 2813
    },
    {
      "epoch": 0.15323354638459497,
      "grad_norm": 0.735304833237844,
      "learning_rate": 0.00019886258714002807,
      "loss": 12.6742,
      "step": 2814
    },
    {
      "epoch": 0.15328800038117799,
      "grad_norm": 0.7593330629316116,
      "learning_rate": 0.00019886126052695333,
      "loss": 12.6313,
      "step": 2815
    },
    {
      "epoch": 0.15334245437776098,
      "grad_norm": 0.6571157408790251,
      "learning_rate": 0.00019885993314511686,
      "loss": 12.6673,
      "step": 2816
    },
    {
      "epoch": 0.153396908374344,
      "grad_norm": 0.7836367975743452,
      "learning_rate": 0.00019885860499452895,
      "loss": 12.6894,
      "step": 2817
    },
    {
      "epoch": 0.15345136237092702,
      "grad_norm": 0.7174222449230672,
      "learning_rate": 0.00019885727607519993,
      "loss": 12.5857,
      "step": 2818
    },
    {
      "epoch": 0.15350581636751,
      "grad_norm": 0.8054844096099334,
      "learning_rate": 0.00019885594638714018,
      "loss": 12.7942,
      "step": 2819
    },
    {
      "epoch": 0.15356027036409303,
      "grad_norm": 0.7531539053210494,
      "learning_rate": 0.00019885461593036,
      "loss": 12.7152,
      "step": 2820
    },
    {
      "epoch": 0.15361472436067605,
      "grad_norm": 0.8023736412652699,
      "learning_rate": 0.00019885328470486976,
      "loss": 12.8768,
      "step": 2821
    },
    {
      "epoch": 0.15366917835725905,
      "grad_norm": 0.7543875214548945,
      "learning_rate": 0.0001988519527106798,
      "loss": 12.5622,
      "step": 2822
    },
    {
      "epoch": 0.15372363235384207,
      "grad_norm": 0.7248015081840921,
      "learning_rate": 0.0001988506199478005,
      "loss": 12.6281,
      "step": 2823
    },
    {
      "epoch": 0.15377808635042509,
      "grad_norm": 0.7369565113377408,
      "learning_rate": 0.00019884928641624217,
      "loss": 12.5889,
      "step": 2824
    },
    {
      "epoch": 0.1538325403470081,
      "grad_norm": 0.755269816391,
      "learning_rate": 0.00019884795211601522,
      "loss": 12.659,
      "step": 2825
    },
    {
      "epoch": 0.1538869943435911,
      "grad_norm": 0.8702353886072459,
      "learning_rate": 0.00019884661704713003,
      "loss": 12.6518,
      "step": 2826
    },
    {
      "epoch": 0.15394144834017412,
      "grad_norm": 0.8471764604812616,
      "learning_rate": 0.00019884528120959693,
      "loss": 12.6379,
      "step": 2827
    },
    {
      "epoch": 0.15399590233675714,
      "grad_norm": 0.7872512739321531,
      "learning_rate": 0.00019884394460342636,
      "loss": 12.509,
      "step": 2828
    },
    {
      "epoch": 0.15405035633334013,
      "grad_norm": 0.8235514382992309,
      "learning_rate": 0.00019884260722862873,
      "loss": 12.6816,
      "step": 2829
    },
    {
      "epoch": 0.15410481032992315,
      "grad_norm": 1.414682065801912,
      "learning_rate": 0.0001988412690852144,
      "loss": 12.7264,
      "step": 2830
    },
    {
      "epoch": 0.15415926432650617,
      "grad_norm": 0.7508374862892704,
      "learning_rate": 0.00019883993017319376,
      "loss": 12.7554,
      "step": 2831
    },
    {
      "epoch": 0.15421371832308917,
      "grad_norm": 0.7707106426936741,
      "learning_rate": 0.00019883859049257726,
      "loss": 12.6894,
      "step": 2832
    },
    {
      "epoch": 0.1542681723196722,
      "grad_norm": 0.7810345921420442,
      "learning_rate": 0.00019883725004337532,
      "loss": 12.6557,
      "step": 2833
    },
    {
      "epoch": 0.1543226263162552,
      "grad_norm": 0.6986731637448959,
      "learning_rate": 0.00019883590882559834,
      "loss": 12.7629,
      "step": 2834
    },
    {
      "epoch": 0.1543770803128382,
      "grad_norm": 0.8182701064880522,
      "learning_rate": 0.0001988345668392568,
      "loss": 12.7564,
      "step": 2835
    },
    {
      "epoch": 0.15443153430942122,
      "grad_norm": 0.7583269846678126,
      "learning_rate": 0.00019883322408436102,
      "loss": 12.5635,
      "step": 2836
    },
    {
      "epoch": 0.15448598830600424,
      "grad_norm": 0.721085359058006,
      "learning_rate": 0.00019883188056092155,
      "loss": 12.5803,
      "step": 2837
    },
    {
      "epoch": 0.15454044230258723,
      "grad_norm": 0.8972990116057311,
      "learning_rate": 0.00019883053626894878,
      "loss": 12.721,
      "step": 2838
    },
    {
      "epoch": 0.15459489629917025,
      "grad_norm": 0.7696583704912467,
      "learning_rate": 0.00019882919120845324,
      "loss": 12.6906,
      "step": 2839
    },
    {
      "epoch": 0.15464935029575327,
      "grad_norm": 0.8683192989979782,
      "learning_rate": 0.0001988278453794453,
      "loss": 12.7031,
      "step": 2840
    },
    {
      "epoch": 0.1547038042923363,
      "grad_norm": 0.7580129590742768,
      "learning_rate": 0.00019882649878193544,
      "loss": 12.66,
      "step": 2841
    },
    {
      "epoch": 0.1547582582889193,
      "grad_norm": 0.8619060801915377,
      "learning_rate": 0.00019882515141593417,
      "loss": 12.7066,
      "step": 2842
    },
    {
      "epoch": 0.1548127122855023,
      "grad_norm": 0.8092195760188854,
      "learning_rate": 0.00019882380328145195,
      "loss": 12.7587,
      "step": 2843
    },
    {
      "epoch": 0.15486716628208533,
      "grad_norm": 0.7464848175934536,
      "learning_rate": 0.0001988224543784993,
      "loss": 12.6538,
      "step": 2844
    },
    {
      "epoch": 0.15492162027866832,
      "grad_norm": 0.8128263278093952,
      "learning_rate": 0.0001988211047070866,
      "loss": 12.6386,
      "step": 2845
    },
    {
      "epoch": 0.15497607427525134,
      "grad_norm": 0.7961419281399771,
      "learning_rate": 0.0001988197542672245,
      "loss": 12.6517,
      "step": 2846
    },
    {
      "epoch": 0.15503052827183436,
      "grad_norm": 0.8514430955903212,
      "learning_rate": 0.00019881840305892336,
      "loss": 12.6183,
      "step": 2847
    },
    {
      "epoch": 0.15508498226841735,
      "grad_norm": 0.7891271901770917,
      "learning_rate": 0.00019881705108219376,
      "loss": 12.6508,
      "step": 2848
    },
    {
      "epoch": 0.15513943626500037,
      "grad_norm": 0.8303345806312178,
      "learning_rate": 0.0001988156983370462,
      "loss": 12.604,
      "step": 2849
    },
    {
      "epoch": 0.1551938902615834,
      "grad_norm": 0.8482866449361063,
      "learning_rate": 0.0001988143448234912,
      "loss": 12.6734,
      "step": 2850
    },
    {
      "epoch": 0.1552483442581664,
      "grad_norm": 0.7796472708220716,
      "learning_rate": 0.0001988129905415393,
      "loss": 12.4922,
      "step": 2851
    },
    {
      "epoch": 0.1553027982547494,
      "grad_norm": 0.9465182394861125,
      "learning_rate": 0.000198811635491201,
      "loss": 12.7001,
      "step": 2852
    },
    {
      "epoch": 0.15535725225133243,
      "grad_norm": 0.7281216127951547,
      "learning_rate": 0.00019881027967248683,
      "loss": 12.6122,
      "step": 2853
    },
    {
      "epoch": 0.15541170624791542,
      "grad_norm": 0.9286753088079683,
      "learning_rate": 0.00019880892308540737,
      "loss": 12.8227,
      "step": 2854
    },
    {
      "epoch": 0.15546616024449844,
      "grad_norm": 0.7763893618414947,
      "learning_rate": 0.00019880756572997316,
      "loss": 12.7216,
      "step": 2855
    },
    {
      "epoch": 0.15552061424108146,
      "grad_norm": 0.7916498937728372,
      "learning_rate": 0.00019880620760619476,
      "loss": 12.6203,
      "step": 2856
    },
    {
      "epoch": 0.15557506823766445,
      "grad_norm": 0.767571823535825,
      "learning_rate": 0.0001988048487140827,
      "loss": 12.5859,
      "step": 2857
    },
    {
      "epoch": 0.15562952223424747,
      "grad_norm": 0.7645270956833853,
      "learning_rate": 0.00019880348905364757,
      "loss": 12.7418,
      "step": 2858
    },
    {
      "epoch": 0.1556839762308305,
      "grad_norm": 0.9638854534597893,
      "learning_rate": 0.00019880212862489994,
      "loss": 12.6037,
      "step": 2859
    },
    {
      "epoch": 0.15573843022741352,
      "grad_norm": 0.7436913641362939,
      "learning_rate": 0.0001988007674278504,
      "loss": 12.6679,
      "step": 2860
    },
    {
      "epoch": 0.1557928842239965,
      "grad_norm": 0.8370044615080934,
      "learning_rate": 0.00019879940546250953,
      "loss": 12.623,
      "step": 2861
    },
    {
      "epoch": 0.15584733822057953,
      "grad_norm": 0.6821081371810797,
      "learning_rate": 0.0001987980427288879,
      "loss": 12.5769,
      "step": 2862
    },
    {
      "epoch": 0.15590179221716255,
      "grad_norm": 0.7644536575584995,
      "learning_rate": 0.0001987966792269961,
      "loss": 12.503,
      "step": 2863
    },
    {
      "epoch": 0.15595624621374554,
      "grad_norm": 0.7734674709924113,
      "learning_rate": 0.00019879531495684477,
      "loss": 12.6612,
      "step": 2864
    },
    {
      "epoch": 0.15601070021032856,
      "grad_norm": 0.7339254537144414,
      "learning_rate": 0.00019879394991844453,
      "loss": 12.5314,
      "step": 2865
    },
    {
      "epoch": 0.15606515420691158,
      "grad_norm": 0.754806928649199,
      "learning_rate": 0.00019879258411180595,
      "loss": 12.6488,
      "step": 2866
    },
    {
      "epoch": 0.15611960820349458,
      "grad_norm": 0.7838025302628356,
      "learning_rate": 0.00019879121753693966,
      "loss": 12.624,
      "step": 2867
    },
    {
      "epoch": 0.1561740622000776,
      "grad_norm": 0.9476835270509917,
      "learning_rate": 0.00019878985019385629,
      "loss": 12.7108,
      "step": 2868
    },
    {
      "epoch": 0.15622851619666062,
      "grad_norm": 0.7778603460412541,
      "learning_rate": 0.0001987884820825665,
      "loss": 12.7608,
      "step": 2869
    },
    {
      "epoch": 0.1562829701932436,
      "grad_norm": 0.8146019433611156,
      "learning_rate": 0.00019878711320308088,
      "loss": 12.5931,
      "step": 2870
    },
    {
      "epoch": 0.15633742418982663,
      "grad_norm": 0.8538501969506916,
      "learning_rate": 0.00019878574355541013,
      "loss": 12.7687,
      "step": 2871
    },
    {
      "epoch": 0.15639187818640965,
      "grad_norm": 0.7748395850487926,
      "learning_rate": 0.00019878437313956485,
      "loss": 12.5725,
      "step": 2872
    },
    {
      "epoch": 0.15644633218299264,
      "grad_norm": 0.9468791007489141,
      "learning_rate": 0.00019878300195555574,
      "loss": 12.7611,
      "step": 2873
    },
    {
      "epoch": 0.15650078617957566,
      "grad_norm": 0.8541710954520909,
      "learning_rate": 0.0001987816300033934,
      "loss": 12.757,
      "step": 2874
    },
    {
      "epoch": 0.15655524017615868,
      "grad_norm": 0.8001891904172195,
      "learning_rate": 0.00019878025728308857,
      "loss": 12.5571,
      "step": 2875
    },
    {
      "epoch": 0.1566096941727417,
      "grad_norm": 0.7420531715125143,
      "learning_rate": 0.0001987788837946519,
      "loss": 12.8099,
      "step": 2876
    },
    {
      "epoch": 0.1566641481693247,
      "grad_norm": 0.8668620926535715,
      "learning_rate": 0.00019877750953809403,
      "loss": 12.8152,
      "step": 2877
    },
    {
      "epoch": 0.15671860216590772,
      "grad_norm": 0.7066024735778503,
      "learning_rate": 0.00019877613451342572,
      "loss": 12.6544,
      "step": 2878
    },
    {
      "epoch": 0.15677305616249074,
      "grad_norm": 0.7301587483127819,
      "learning_rate": 0.0001987747587206576,
      "loss": 12.6587,
      "step": 2879
    },
    {
      "epoch": 0.15682751015907373,
      "grad_norm": 0.8221851904995162,
      "learning_rate": 0.0001987733821598004,
      "loss": 12.614,
      "step": 2880
    },
    {
      "epoch": 0.15688196415565675,
      "grad_norm": 0.7800050428325034,
      "learning_rate": 0.00019877200483086482,
      "loss": 12.7346,
      "step": 2881
    },
    {
      "epoch": 0.15693641815223977,
      "grad_norm": 0.7538371294006235,
      "learning_rate": 0.00019877062673386155,
      "loss": 12.7771,
      "step": 2882
    },
    {
      "epoch": 0.15699087214882276,
      "grad_norm": 0.7737338005678903,
      "learning_rate": 0.00019876924786880136,
      "loss": 12.6477,
      "step": 2883
    },
    {
      "epoch": 0.15704532614540578,
      "grad_norm": 0.7794075495170949,
      "learning_rate": 0.0001987678682356949,
      "loss": 12.8022,
      "step": 2884
    },
    {
      "epoch": 0.1570997801419888,
      "grad_norm": 0.8710909802578073,
      "learning_rate": 0.00019876648783455293,
      "loss": 12.659,
      "step": 2885
    },
    {
      "epoch": 0.1571542341385718,
      "grad_norm": 0.668760916103844,
      "learning_rate": 0.0001987651066653862,
      "loss": 12.686,
      "step": 2886
    },
    {
      "epoch": 0.15720868813515482,
      "grad_norm": 0.7562285864065703,
      "learning_rate": 0.00019876372472820545,
      "loss": 12.6519,
      "step": 2887
    },
    {
      "epoch": 0.15726314213173784,
      "grad_norm": 0.7185663162783882,
      "learning_rate": 0.0001987623420230214,
      "loss": 12.6415,
      "step": 2888
    },
    {
      "epoch": 0.15731759612832083,
      "grad_norm": 0.6729839607124669,
      "learning_rate": 0.00019876095854984483,
      "loss": 12.6349,
      "step": 2889
    },
    {
      "epoch": 0.15737205012490385,
      "grad_norm": 0.7545275225102389,
      "learning_rate": 0.00019875957430868645,
      "loss": 12.7718,
      "step": 2890
    },
    {
      "epoch": 0.15742650412148687,
      "grad_norm": 0.7892422343726105,
      "learning_rate": 0.0001987581892995571,
      "loss": 12.566,
      "step": 2891
    },
    {
      "epoch": 0.1574809581180699,
      "grad_norm": 0.800924414122386,
      "learning_rate": 0.0001987568035224675,
      "loss": 12.7672,
      "step": 2892
    },
    {
      "epoch": 0.15753541211465288,
      "grad_norm": 0.7405628682209897,
      "learning_rate": 0.0001987554169774284,
      "loss": 12.6177,
      "step": 2893
    },
    {
      "epoch": 0.1575898661112359,
      "grad_norm": 0.8188446538091821,
      "learning_rate": 0.00019875402966445065,
      "loss": 12.8018,
      "step": 2894
    },
    {
      "epoch": 0.15764432010781892,
      "grad_norm": 0.6775815545337175,
      "learning_rate": 0.00019875264158354498,
      "loss": 12.9058,
      "step": 2895
    },
    {
      "epoch": 0.15769877410440192,
      "grad_norm": 0.7745018906486485,
      "learning_rate": 0.00019875125273472222,
      "loss": 12.6926,
      "step": 2896
    },
    {
      "epoch": 0.15775322810098494,
      "grad_norm": 0.6856967013924328,
      "learning_rate": 0.00019874986311799316,
      "loss": 12.6651,
      "step": 2897
    },
    {
      "epoch": 0.15780768209756796,
      "grad_norm": 0.8311969803551889,
      "learning_rate": 0.00019874847273336862,
      "loss": 12.8152,
      "step": 2898
    },
    {
      "epoch": 0.15786213609415095,
      "grad_norm": 0.7484518446515533,
      "learning_rate": 0.00019874708158085938,
      "loss": 12.7422,
      "step": 2899
    },
    {
      "epoch": 0.15791659009073397,
      "grad_norm": 0.9253746140131502,
      "learning_rate": 0.00019874568966047625,
      "loss": 12.8505,
      "step": 2900
    },
    {
      "epoch": 0.157971044087317,
      "grad_norm": 0.7642243120959947,
      "learning_rate": 0.0001987442969722301,
      "loss": 12.6175,
      "step": 2901
    },
    {
      "epoch": 0.15802549808389998,
      "grad_norm": 0.7379359930600351,
      "learning_rate": 0.00019874290351613177,
      "loss": 12.7439,
      "step": 2902
    },
    {
      "epoch": 0.158079952080483,
      "grad_norm": 0.7361247160464598,
      "learning_rate": 0.000198741509292192,
      "loss": 12.5702,
      "step": 2903
    },
    {
      "epoch": 0.15813440607706603,
      "grad_norm": 0.7632560096569945,
      "learning_rate": 0.00019874011430042173,
      "loss": 12.5885,
      "step": 2904
    },
    {
      "epoch": 0.15818886007364902,
      "grad_norm": 0.7121434411200048,
      "learning_rate": 0.00019873871854083177,
      "loss": 12.6405,
      "step": 2905
    },
    {
      "epoch": 0.15824331407023204,
      "grad_norm": 0.6974010357137779,
      "learning_rate": 0.00019873732201343297,
      "loss": 12.6232,
      "step": 2906
    },
    {
      "epoch": 0.15829776806681506,
      "grad_norm": 0.6934937196933434,
      "learning_rate": 0.00019873592471823622,
      "loss": 12.5748,
      "step": 2907
    },
    {
      "epoch": 0.15835222206339808,
      "grad_norm": 0.8136084193231012,
      "learning_rate": 0.0001987345266552523,
      "loss": 12.705,
      "step": 2908
    },
    {
      "epoch": 0.15840667605998107,
      "grad_norm": 0.8153925561329763,
      "learning_rate": 0.0001987331278244922,
      "loss": 12.6548,
      "step": 2909
    },
    {
      "epoch": 0.1584611300565641,
      "grad_norm": 0.7461307062300498,
      "learning_rate": 0.00019873172822596673,
      "loss": 12.6724,
      "step": 2910
    },
    {
      "epoch": 0.1585155840531471,
      "grad_norm": 0.7591005159059452,
      "learning_rate": 0.0001987303278596868,
      "loss": 12.7604,
      "step": 2911
    },
    {
      "epoch": 0.1585700380497301,
      "grad_norm": 0.7719825542635549,
      "learning_rate": 0.00019872892672566326,
      "loss": 12.755,
      "step": 2912
    },
    {
      "epoch": 0.15862449204631313,
      "grad_norm": 0.7409777070497231,
      "learning_rate": 0.000198727524823907,
      "loss": 12.7466,
      "step": 2913
    },
    {
      "epoch": 0.15867894604289615,
      "grad_norm": 0.8675143707913091,
      "learning_rate": 0.000198726122154429,
      "loss": 12.642,
      "step": 2914
    },
    {
      "epoch": 0.15873340003947914,
      "grad_norm": 0.8154491006459972,
      "learning_rate": 0.0001987247187172401,
      "loss": 12.749,
      "step": 2915
    },
    {
      "epoch": 0.15878785403606216,
      "grad_norm": 0.8774062749892383,
      "learning_rate": 0.0001987233145123512,
      "loss": 12.744,
      "step": 2916
    },
    {
      "epoch": 0.15884230803264518,
      "grad_norm": 0.7171864769206286,
      "learning_rate": 0.00019872190953977331,
      "loss": 12.6279,
      "step": 2917
    },
    {
      "epoch": 0.15889676202922817,
      "grad_norm": 0.8983152653112244,
      "learning_rate": 0.0001987205037995173,
      "loss": 12.6355,
      "step": 2918
    },
    {
      "epoch": 0.1589512160258112,
      "grad_norm": 0.8848065508819202,
      "learning_rate": 0.00019871909729159403,
      "loss": 12.8295,
      "step": 2919
    },
    {
      "epoch": 0.1590056700223942,
      "grad_norm": 0.8467832737921628,
      "learning_rate": 0.00019871769001601454,
      "loss": 12.5814,
      "step": 2920
    },
    {
      "epoch": 0.1590601240189772,
      "grad_norm": 0.84887487191556,
      "learning_rate": 0.00019871628197278972,
      "loss": 12.8861,
      "step": 2921
    },
    {
      "epoch": 0.15911457801556023,
      "grad_norm": 0.8061164552631324,
      "learning_rate": 0.00019871487316193057,
      "loss": 12.7834,
      "step": 2922
    },
    {
      "epoch": 0.15916903201214325,
      "grad_norm": 0.7107138305932041,
      "learning_rate": 0.000198713463583448,
      "loss": 12.5469,
      "step": 2923
    },
    {
      "epoch": 0.15922348600872624,
      "grad_norm": 0.6743243016861461,
      "learning_rate": 0.00019871205323735298,
      "loss": 12.6979,
      "step": 2924
    },
    {
      "epoch": 0.15927794000530926,
      "grad_norm": 0.7720761548238595,
      "learning_rate": 0.00019871064212365647,
      "loss": 12.5449,
      "step": 2925
    },
    {
      "epoch": 0.15933239400189228,
      "grad_norm": 0.7922355008399666,
      "learning_rate": 0.00019870923024236948,
      "loss": 12.6402,
      "step": 2926
    },
    {
      "epoch": 0.1593868479984753,
      "grad_norm": 0.8070662326895096,
      "learning_rate": 0.00019870781759350292,
      "loss": 12.6296,
      "step": 2927
    },
    {
      "epoch": 0.1594413019950583,
      "grad_norm": 0.6803895819196606,
      "learning_rate": 0.00019870640417706784,
      "loss": 12.5325,
      "step": 2928
    },
    {
      "epoch": 0.1594957559916413,
      "grad_norm": 0.7126599815491631,
      "learning_rate": 0.00019870498999307522,
      "loss": 12.6576,
      "step": 2929
    },
    {
      "epoch": 0.15955020998822433,
      "grad_norm": 0.8448286288702788,
      "learning_rate": 0.00019870357504153603,
      "loss": 12.7188,
      "step": 2930
    },
    {
      "epoch": 0.15960466398480733,
      "grad_norm": 0.7214711269702817,
      "learning_rate": 0.00019870215932246127,
      "loss": 12.6394,
      "step": 2931
    },
    {
      "epoch": 0.15965911798139035,
      "grad_norm": 0.829531978726736,
      "learning_rate": 0.000198700742835862,
      "loss": 12.7261,
      "step": 2932
    },
    {
      "epoch": 0.15971357197797337,
      "grad_norm": 0.7058088612808299,
      "learning_rate": 0.00019869932558174919,
      "loss": 12.678,
      "step": 2933
    },
    {
      "epoch": 0.15976802597455636,
      "grad_norm": 0.7317555508306628,
      "learning_rate": 0.00019869790756013385,
      "loss": 12.5755,
      "step": 2934
    },
    {
      "epoch": 0.15982247997113938,
      "grad_norm": 0.6647603557904068,
      "learning_rate": 0.00019869648877102707,
      "loss": 12.6979,
      "step": 2935
    },
    {
      "epoch": 0.1598769339677224,
      "grad_norm": 0.8407777414091054,
      "learning_rate": 0.00019869506921443982,
      "loss": 12.7168,
      "step": 2936
    },
    {
      "epoch": 0.1599313879643054,
      "grad_norm": 0.743608301092564,
      "learning_rate": 0.0001986936488903832,
      "loss": 12.756,
      "step": 2937
    },
    {
      "epoch": 0.15998584196088841,
      "grad_norm": 0.8049838269368632,
      "learning_rate": 0.00019869222779886814,
      "loss": 12.6823,
      "step": 2938
    },
    {
      "epoch": 0.16004029595747143,
      "grad_norm": 0.7606150998743767,
      "learning_rate": 0.00019869080593990578,
      "loss": 12.6691,
      "step": 2939
    },
    {
      "epoch": 0.16009474995405443,
      "grad_norm": 0.6970909985961943,
      "learning_rate": 0.0001986893833135072,
      "loss": 12.7718,
      "step": 2940
    },
    {
      "epoch": 0.16014920395063745,
      "grad_norm": 0.7769916732202724,
      "learning_rate": 0.00019868795991968342,
      "loss": 12.6826,
      "step": 2941
    },
    {
      "epoch": 0.16020365794722047,
      "grad_norm": 0.7659872792765248,
      "learning_rate": 0.0001986865357584455,
      "loss": 12.7583,
      "step": 2942
    },
    {
      "epoch": 0.1602581119438035,
      "grad_norm": 0.8445126991147179,
      "learning_rate": 0.00019868511082980455,
      "loss": 12.8178,
      "step": 2943
    },
    {
      "epoch": 0.16031256594038648,
      "grad_norm": 0.763277351074957,
      "learning_rate": 0.0001986836851337716,
      "loss": 12.6646,
      "step": 2944
    },
    {
      "epoch": 0.1603670199369695,
      "grad_norm": 1.0276730264139622,
      "learning_rate": 0.00019868225867035778,
      "loss": 12.837,
      "step": 2945
    },
    {
      "epoch": 0.16042147393355252,
      "grad_norm": 1.0233540700530248,
      "learning_rate": 0.00019868083143957416,
      "loss": 12.7812,
      "step": 2946
    },
    {
      "epoch": 0.16047592793013551,
      "grad_norm": 0.7844462560812835,
      "learning_rate": 0.00019867940344143185,
      "loss": 12.6027,
      "step": 2947
    },
    {
      "epoch": 0.16053038192671854,
      "grad_norm": 0.8081790004592988,
      "learning_rate": 0.00019867797467594195,
      "loss": 12.5724,
      "step": 2948
    },
    {
      "epoch": 0.16058483592330156,
      "grad_norm": 0.7615334540876233,
      "learning_rate": 0.00019867654514311558,
      "loss": 12.6042,
      "step": 2949
    },
    {
      "epoch": 0.16063928991988455,
      "grad_norm": 0.7693417927056728,
      "learning_rate": 0.00019867511484296385,
      "loss": 12.7132,
      "step": 2950
    },
    {
      "epoch": 0.16069374391646757,
      "grad_norm": 0.7432777721279483,
      "learning_rate": 0.00019867368377549785,
      "loss": 12.6545,
      "step": 2951
    },
    {
      "epoch": 0.1607481979130506,
      "grad_norm": 0.7256121338548468,
      "learning_rate": 0.00019867225194072875,
      "loss": 12.6685,
      "step": 2952
    },
    {
      "epoch": 0.16080265190963358,
      "grad_norm": 0.7448417158991302,
      "learning_rate": 0.00019867081933866768,
      "loss": 12.711,
      "step": 2953
    },
    {
      "epoch": 0.1608571059062166,
      "grad_norm": 0.8333907442536371,
      "learning_rate": 0.0001986693859693258,
      "loss": 12.6638,
      "step": 2954
    },
    {
      "epoch": 0.16091155990279962,
      "grad_norm": 0.813839754003139,
      "learning_rate": 0.00019866795183271418,
      "loss": 12.7326,
      "step": 2955
    },
    {
      "epoch": 0.16096601389938262,
      "grad_norm": 0.8254595639514176,
      "learning_rate": 0.000198666516928844,
      "loss": 12.8003,
      "step": 2956
    },
    {
      "epoch": 0.16102046789596564,
      "grad_norm": 0.7082473755766474,
      "learning_rate": 0.0001986650812577265,
      "loss": 12.7089,
      "step": 2957
    },
    {
      "epoch": 0.16107492189254866,
      "grad_norm": 0.8686973365777078,
      "learning_rate": 0.00019866364481937275,
      "loss": 12.8251,
      "step": 2958
    },
    {
      "epoch": 0.16112937588913168,
      "grad_norm": 0.8270284101134693,
      "learning_rate": 0.00019866220761379397,
      "loss": 12.8409,
      "step": 2959
    },
    {
      "epoch": 0.16118382988571467,
      "grad_norm": 0.7209478339761264,
      "learning_rate": 0.00019866076964100132,
      "loss": 12.7463,
      "step": 2960
    },
    {
      "epoch": 0.1612382838822977,
      "grad_norm": 0.9011747411779443,
      "learning_rate": 0.00019865933090100593,
      "loss": 12.7234,
      "step": 2961
    },
    {
      "epoch": 0.1612927378788807,
      "grad_norm": 0.7302090317140539,
      "learning_rate": 0.00019865789139381906,
      "loss": 12.6734,
      "step": 2962
    },
    {
      "epoch": 0.1613471918754637,
      "grad_norm": 0.8439839869720785,
      "learning_rate": 0.00019865645111945192,
      "loss": 12.7653,
      "step": 2963
    },
    {
      "epoch": 0.16140164587204672,
      "grad_norm": 0.7831387332858022,
      "learning_rate": 0.00019865501007791564,
      "loss": 12.6617,
      "step": 2964
    },
    {
      "epoch": 0.16145609986862974,
      "grad_norm": 0.751759533220739,
      "learning_rate": 0.00019865356826922147,
      "loss": 12.6347,
      "step": 2965
    },
    {
      "epoch": 0.16151055386521274,
      "grad_norm": 0.784713141210689,
      "learning_rate": 0.0001986521256933806,
      "loss": 12.6752,
      "step": 2966
    },
    {
      "epoch": 0.16156500786179576,
      "grad_norm": 0.9349312628518618,
      "learning_rate": 0.00019865068235040427,
      "loss": 12.8224,
      "step": 2967
    },
    {
      "epoch": 0.16161946185837878,
      "grad_norm": 0.796775119153196,
      "learning_rate": 0.00019864923824030367,
      "loss": 12.6793,
      "step": 2968
    },
    {
      "epoch": 0.16167391585496177,
      "grad_norm": 0.9158652239242536,
      "learning_rate": 0.00019864779336309005,
      "loss": 12.752,
      "step": 2969
    },
    {
      "epoch": 0.1617283698515448,
      "grad_norm": 0.7597183903240505,
      "learning_rate": 0.0001986463477187747,
      "loss": 12.7695,
      "step": 2970
    },
    {
      "epoch": 0.1617828238481278,
      "grad_norm": 0.9777933769595428,
      "learning_rate": 0.00019864490130736874,
      "loss": 12.7579,
      "step": 2971
    },
    {
      "epoch": 0.1618372778447108,
      "grad_norm": 0.6985247110538119,
      "learning_rate": 0.0001986434541288835,
      "loss": 12.6177,
      "step": 2972
    },
    {
      "epoch": 0.16189173184129382,
      "grad_norm": 0.7980997965207286,
      "learning_rate": 0.00019864200618333023,
      "loss": 12.7239,
      "step": 2973
    },
    {
      "epoch": 0.16194618583787684,
      "grad_norm": 0.8623726372473786,
      "learning_rate": 0.00019864055747072018,
      "loss": 12.7306,
      "step": 2974
    },
    {
      "epoch": 0.16200063983445986,
      "grad_norm": 0.7770705515852193,
      "learning_rate": 0.0001986391079910646,
      "loss": 12.8426,
      "step": 2975
    },
    {
      "epoch": 0.16205509383104286,
      "grad_norm": 0.7478972825442104,
      "learning_rate": 0.00019863765774437477,
      "loss": 12.6541,
      "step": 2976
    },
    {
      "epoch": 0.16210954782762588,
      "grad_norm": 0.7143585535799926,
      "learning_rate": 0.000198636206730662,
      "loss": 12.7268,
      "step": 2977
    },
    {
      "epoch": 0.1621640018242089,
      "grad_norm": 0.8776883368027423,
      "learning_rate": 0.0001986347549499375,
      "loss": 12.831,
      "step": 2978
    },
    {
      "epoch": 0.1622184558207919,
      "grad_norm": 0.7508476394175179,
      "learning_rate": 0.00019863330240221263,
      "loss": 12.833,
      "step": 2979
    },
    {
      "epoch": 0.1622729098173749,
      "grad_norm": 0.8346959291607994,
      "learning_rate": 0.00019863184908749866,
      "loss": 12.6451,
      "step": 2980
    },
    {
      "epoch": 0.16232736381395793,
      "grad_norm": 0.7329729295070289,
      "learning_rate": 0.0001986303950058069,
      "loss": 12.743,
      "step": 2981
    },
    {
      "epoch": 0.16238181781054092,
      "grad_norm": 0.7357999171068264,
      "learning_rate": 0.00019862894015714865,
      "loss": 12.6297,
      "step": 2982
    },
    {
      "epoch": 0.16243627180712394,
      "grad_norm": 0.7464982470254528,
      "learning_rate": 0.00019862748454153523,
      "loss": 12.7155,
      "step": 2983
    },
    {
      "epoch": 0.16249072580370696,
      "grad_norm": 0.7276796561806051,
      "learning_rate": 0.00019862602815897792,
      "loss": 12.7921,
      "step": 2984
    },
    {
      "epoch": 0.16254517980028996,
      "grad_norm": 0.7468609325498567,
      "learning_rate": 0.0001986245710094881,
      "loss": 12.7216,
      "step": 2985
    },
    {
      "epoch": 0.16259963379687298,
      "grad_norm": 0.7581697225369602,
      "learning_rate": 0.00019862311309307704,
      "loss": 12.6114,
      "step": 2986
    },
    {
      "epoch": 0.162654087793456,
      "grad_norm": 0.7915931346131438,
      "learning_rate": 0.00019862165440975616,
      "loss": 12.745,
      "step": 2987
    },
    {
      "epoch": 0.162708541790039,
      "grad_norm": 0.775253020930781,
      "learning_rate": 0.00019862019495953673,
      "loss": 12.6253,
      "step": 2988
    },
    {
      "epoch": 0.162762995786622,
      "grad_norm": 0.7540019789596962,
      "learning_rate": 0.0001986187347424301,
      "loss": 12.6637,
      "step": 2989
    },
    {
      "epoch": 0.16281744978320503,
      "grad_norm": 0.7955802651375394,
      "learning_rate": 0.00019861727375844769,
      "loss": 12.6314,
      "step": 2990
    },
    {
      "epoch": 0.16287190377978802,
      "grad_norm": 0.704076718526564,
      "learning_rate": 0.00019861581200760083,
      "loss": 12.654,
      "step": 2991
    },
    {
      "epoch": 0.16292635777637104,
      "grad_norm": 0.752020939332643,
      "learning_rate": 0.00019861434948990084,
      "loss": 12.7554,
      "step": 2992
    },
    {
      "epoch": 0.16298081177295407,
      "grad_norm": 0.8500661663949628,
      "learning_rate": 0.00019861288620535915,
      "loss": 12.7574,
      "step": 2993
    },
    {
      "epoch": 0.16303526576953709,
      "grad_norm": 0.7151514384071872,
      "learning_rate": 0.00019861142215398713,
      "loss": 12.6612,
      "step": 2994
    },
    {
      "epoch": 0.16308971976612008,
      "grad_norm": 0.8014697789286036,
      "learning_rate": 0.00019860995733579615,
      "loss": 12.6775,
      "step": 2995
    },
    {
      "epoch": 0.1631441737627031,
      "grad_norm": 0.7902169978055278,
      "learning_rate": 0.0001986084917507976,
      "loss": 12.8139,
      "step": 2996
    },
    {
      "epoch": 0.16319862775928612,
      "grad_norm": 0.7667894415875761,
      "learning_rate": 0.00019860702539900287,
      "loss": 12.5831,
      "step": 2997
    },
    {
      "epoch": 0.1632530817558691,
      "grad_norm": 0.7606419870046596,
      "learning_rate": 0.00019860555828042338,
      "loss": 12.5801,
      "step": 2998
    },
    {
      "epoch": 0.16330753575245213,
      "grad_norm": 0.7268333122039216,
      "learning_rate": 0.00019860409039507054,
      "loss": 12.6711,
      "step": 2999
    },
    {
      "epoch": 0.16336198974903515,
      "grad_norm": 0.8293781927525129,
      "learning_rate": 0.00019860262174295574,
      "loss": 12.667,
      "step": 3000
    },
    {
      "epoch": 0.16341644374561815,
      "grad_norm": 0.6821855115896598,
      "learning_rate": 0.00019860115232409045,
      "loss": 12.646,
      "step": 3001
    },
    {
      "epoch": 0.16347089774220117,
      "grad_norm": 0.8003391933674313,
      "learning_rate": 0.00019859968213848604,
      "loss": 12.5705,
      "step": 3002
    },
    {
      "epoch": 0.16352535173878419,
      "grad_norm": 0.8315845013677727,
      "learning_rate": 0.00019859821118615396,
      "loss": 12.7818,
      "step": 3003
    },
    {
      "epoch": 0.16357980573536718,
      "grad_norm": 0.8041144709489726,
      "learning_rate": 0.00019859673946710568,
      "loss": 12.6929,
      "step": 3004
    },
    {
      "epoch": 0.1636342597319502,
      "grad_norm": 0.7491663856126463,
      "learning_rate": 0.0001985952669813526,
      "loss": 12.6536,
      "step": 3005
    },
    {
      "epoch": 0.16368871372853322,
      "grad_norm": 0.7213642535047672,
      "learning_rate": 0.0001985937937289062,
      "loss": 12.6771,
      "step": 3006
    },
    {
      "epoch": 0.1637431677251162,
      "grad_norm": 0.7064321339419524,
      "learning_rate": 0.00019859231970977792,
      "loss": 12.7313,
      "step": 3007
    },
    {
      "epoch": 0.16379762172169923,
      "grad_norm": 0.7118779395124208,
      "learning_rate": 0.00019859084492397923,
      "loss": 12.6198,
      "step": 3008
    },
    {
      "epoch": 0.16385207571828225,
      "grad_norm": 0.776431535884499,
      "learning_rate": 0.0001985893693715216,
      "loss": 12.6581,
      "step": 3009
    },
    {
      "epoch": 0.16390652971486527,
      "grad_norm": 0.7060943938035943,
      "learning_rate": 0.00019858789305241648,
      "loss": 12.6482,
      "step": 3010
    },
    {
      "epoch": 0.16396098371144827,
      "grad_norm": 0.7462608617503164,
      "learning_rate": 0.0001985864159666754,
      "loss": 12.7181,
      "step": 3011
    },
    {
      "epoch": 0.1640154377080313,
      "grad_norm": 0.7949141644882773,
      "learning_rate": 0.0001985849381143098,
      "loss": 12.7098,
      "step": 3012
    },
    {
      "epoch": 0.1640698917046143,
      "grad_norm": 0.7700097507545192,
      "learning_rate": 0.00019858345949533117,
      "loss": 12.6542,
      "step": 3013
    },
    {
      "epoch": 0.1641243457011973,
      "grad_norm": 0.7223619675646541,
      "learning_rate": 0.00019858198010975106,
      "loss": 12.7549,
      "step": 3014
    },
    {
      "epoch": 0.16417879969778032,
      "grad_norm": 0.6751448073396646,
      "learning_rate": 0.00019858049995758094,
      "loss": 12.6315,
      "step": 3015
    },
    {
      "epoch": 0.16423325369436334,
      "grad_norm": 0.7771226973080849,
      "learning_rate": 0.0001985790190388323,
      "loss": 12.6497,
      "step": 3016
    },
    {
      "epoch": 0.16428770769094633,
      "grad_norm": 0.6711637065491515,
      "learning_rate": 0.00019857753735351668,
      "loss": 12.5573,
      "step": 3017
    },
    {
      "epoch": 0.16434216168752935,
      "grad_norm": 0.712214129858097,
      "learning_rate": 0.0001985760549016456,
      "loss": 12.7238,
      "step": 3018
    },
    {
      "epoch": 0.16439661568411237,
      "grad_norm": 0.7331697045163766,
      "learning_rate": 0.0001985745716832306,
      "loss": 12.5742,
      "step": 3019
    },
    {
      "epoch": 0.16445106968069537,
      "grad_norm": 0.7119826674527502,
      "learning_rate": 0.0001985730876982832,
      "loss": 12.6483,
      "step": 3020
    },
    {
      "epoch": 0.1645055236772784,
      "grad_norm": 0.6974600839247409,
      "learning_rate": 0.0001985716029468149,
      "loss": 12.6442,
      "step": 3021
    },
    {
      "epoch": 0.1645599776738614,
      "grad_norm": 0.8670722829202118,
      "learning_rate": 0.00019857011742883734,
      "loss": 12.8303,
      "step": 3022
    },
    {
      "epoch": 0.1646144316704444,
      "grad_norm": 0.7635478127989597,
      "learning_rate": 0.00019856863114436197,
      "loss": 12.5336,
      "step": 3023
    },
    {
      "epoch": 0.16466888566702742,
      "grad_norm": 0.7542722269262729,
      "learning_rate": 0.0001985671440934004,
      "loss": 12.6527,
      "step": 3024
    },
    {
      "epoch": 0.16472333966361044,
      "grad_norm": 0.8416074307555329,
      "learning_rate": 0.00019856565627596423,
      "loss": 12.7418,
      "step": 3025
    },
    {
      "epoch": 0.16477779366019346,
      "grad_norm": 0.7287356370803173,
      "learning_rate": 0.00019856416769206496,
      "loss": 12.7479,
      "step": 3026
    },
    {
      "epoch": 0.16483224765677645,
      "grad_norm": 0.8005722489828518,
      "learning_rate": 0.00019856267834171418,
      "loss": 12.6369,
      "step": 3027
    },
    {
      "epoch": 0.16488670165335947,
      "grad_norm": 0.8305008235817432,
      "learning_rate": 0.00019856118822492348,
      "loss": 12.6217,
      "step": 3028
    },
    {
      "epoch": 0.1649411556499425,
      "grad_norm": 0.7033320894025007,
      "learning_rate": 0.0001985596973417045,
      "loss": 12.6356,
      "step": 3029
    },
    {
      "epoch": 0.1649956096465255,
      "grad_norm": 0.7793475423768594,
      "learning_rate": 0.00019855820569206873,
      "loss": 12.5976,
      "step": 3030
    },
    {
      "epoch": 0.1650500636431085,
      "grad_norm": 0.7107485974300436,
      "learning_rate": 0.00019855671327602786,
      "loss": 12.6629,
      "step": 3031
    },
    {
      "epoch": 0.16510451763969153,
      "grad_norm": 0.8246162928773088,
      "learning_rate": 0.00019855522009359344,
      "loss": 12.7947,
      "step": 3032
    },
    {
      "epoch": 0.16515897163627452,
      "grad_norm": 0.8131775899827965,
      "learning_rate": 0.0001985537261447771,
      "loss": 12.7613,
      "step": 3033
    },
    {
      "epoch": 0.16521342563285754,
      "grad_norm": 0.7587550524533624,
      "learning_rate": 0.00019855223142959045,
      "loss": 12.6487,
      "step": 3034
    },
    {
      "epoch": 0.16526787962944056,
      "grad_norm": 0.9463315526005036,
      "learning_rate": 0.00019855073594804513,
      "loss": 12.8261,
      "step": 3035
    },
    {
      "epoch": 0.16532233362602355,
      "grad_norm": 0.6858694256651935,
      "learning_rate": 0.0001985492397001528,
      "loss": 12.6521,
      "step": 3036
    },
    {
      "epoch": 0.16537678762260657,
      "grad_norm": 0.9290574372859011,
      "learning_rate": 0.00019854774268592503,
      "loss": 12.7764,
      "step": 3037
    },
    {
      "epoch": 0.1654312416191896,
      "grad_norm": 0.7226513185149467,
      "learning_rate": 0.00019854624490537345,
      "loss": 12.5919,
      "step": 3038
    },
    {
      "epoch": 0.1654856956157726,
      "grad_norm": 0.8268556528120298,
      "learning_rate": 0.00019854474635850976,
      "loss": 12.5752,
      "step": 3039
    },
    {
      "epoch": 0.1655401496123556,
      "grad_norm": 0.6935520582092968,
      "learning_rate": 0.0001985432470453456,
      "loss": 12.624,
      "step": 3040
    },
    {
      "epoch": 0.16559460360893863,
      "grad_norm": 0.7905359262554176,
      "learning_rate": 0.00019854174696589265,
      "loss": 12.6458,
      "step": 3041
    },
    {
      "epoch": 0.16564905760552165,
      "grad_norm": 0.8584511171853852,
      "learning_rate": 0.00019854024612016256,
      "loss": 12.787,
      "step": 3042
    },
    {
      "epoch": 0.16570351160210464,
      "grad_norm": 0.7234888567257534,
      "learning_rate": 0.00019853874450816695,
      "loss": 12.7128,
      "step": 3043
    },
    {
      "epoch": 0.16575796559868766,
      "grad_norm": 0.8364271048465408,
      "learning_rate": 0.00019853724212991756,
      "loss": 12.7128,
      "step": 3044
    },
    {
      "epoch": 0.16581241959527068,
      "grad_norm": 0.7495999449297123,
      "learning_rate": 0.00019853573898542603,
      "loss": 12.7308,
      "step": 3045
    },
    {
      "epoch": 0.16586687359185368,
      "grad_norm": 0.7474271252052067,
      "learning_rate": 0.0001985342350747041,
      "loss": 12.5932,
      "step": 3046
    },
    {
      "epoch": 0.1659213275884367,
      "grad_norm": 0.7774008075239626,
      "learning_rate": 0.00019853273039776342,
      "loss": 12.7495,
      "step": 3047
    },
    {
      "epoch": 0.16597578158501972,
      "grad_norm": 0.7127847304079973,
      "learning_rate": 0.0001985312249546157,
      "loss": 12.8242,
      "step": 3048
    },
    {
      "epoch": 0.1660302355816027,
      "grad_norm": 0.8142909144845354,
      "learning_rate": 0.00019852971874527263,
      "loss": 12.7933,
      "step": 3049
    },
    {
      "epoch": 0.16608468957818573,
      "grad_norm": 0.8571399956290684,
      "learning_rate": 0.00019852821176974595,
      "loss": 12.7599,
      "step": 3050
    },
    {
      "epoch": 0.16613914357476875,
      "grad_norm": 0.7255126471391168,
      "learning_rate": 0.0001985267040280474,
      "loss": 12.6355,
      "step": 3051
    },
    {
      "epoch": 0.16619359757135174,
      "grad_norm": 0.8548735300335262,
      "learning_rate": 0.00019852519552018868,
      "loss": 12.7333,
      "step": 3052
    },
    {
      "epoch": 0.16624805156793476,
      "grad_norm": 0.9585000229029373,
      "learning_rate": 0.0001985236862461815,
      "loss": 12.6209,
      "step": 3053
    },
    {
      "epoch": 0.16630250556451778,
      "grad_norm": 0.8346856967389,
      "learning_rate": 0.0001985221762060376,
      "loss": 12.8314,
      "step": 3054
    },
    {
      "epoch": 0.16635695956110078,
      "grad_norm": 0.781151732816868,
      "learning_rate": 0.00019852066539976875,
      "loss": 12.6076,
      "step": 3055
    },
    {
      "epoch": 0.1664114135576838,
      "grad_norm": 0.8753576736840428,
      "learning_rate": 0.00019851915382738668,
      "loss": 12.7246,
      "step": 3056
    },
    {
      "epoch": 0.16646586755426682,
      "grad_norm": 0.7551967445969222,
      "learning_rate": 0.00019851764148890317,
      "loss": 12.672,
      "step": 3057
    },
    {
      "epoch": 0.1665203215508498,
      "grad_norm": 0.7104804983466242,
      "learning_rate": 0.00019851612838432992,
      "loss": 12.5429,
      "step": 3058
    },
    {
      "epoch": 0.16657477554743283,
      "grad_norm": 0.810574169985711,
      "learning_rate": 0.00019851461451367877,
      "loss": 12.6264,
      "step": 3059
    },
    {
      "epoch": 0.16662922954401585,
      "grad_norm": 0.7584177948039127,
      "learning_rate": 0.00019851309987696145,
      "loss": 12.7277,
      "step": 3060
    },
    {
      "epoch": 0.16668368354059887,
      "grad_norm": 0.7381274284204892,
      "learning_rate": 0.00019851158447418973,
      "loss": 12.7544,
      "step": 3061
    },
    {
      "epoch": 0.16673813753718186,
      "grad_norm": 0.7600336760368321,
      "learning_rate": 0.00019851006830537543,
      "loss": 12.6901,
      "step": 3062
    },
    {
      "epoch": 0.16679259153376488,
      "grad_norm": 0.8158717737436204,
      "learning_rate": 0.00019850855137053028,
      "loss": 12.7702,
      "step": 3063
    },
    {
      "epoch": 0.1668470455303479,
      "grad_norm": 0.74919777778861,
      "learning_rate": 0.00019850703366966615,
      "loss": 12.605,
      "step": 3064
    },
    {
      "epoch": 0.1669014995269309,
      "grad_norm": 0.7882382250155103,
      "learning_rate": 0.00019850551520279478,
      "loss": 12.806,
      "step": 3065
    },
    {
      "epoch": 0.16695595352351392,
      "grad_norm": 0.7955354153451767,
      "learning_rate": 0.00019850399596992802,
      "loss": 12.6832,
      "step": 3066
    },
    {
      "epoch": 0.16701040752009694,
      "grad_norm": 0.7322099613336235,
      "learning_rate": 0.00019850247597107766,
      "loss": 12.6294,
      "step": 3067
    },
    {
      "epoch": 0.16706486151667993,
      "grad_norm": 0.7805140231750468,
      "learning_rate": 0.00019850095520625556,
      "loss": 12.7636,
      "step": 3068
    },
    {
      "epoch": 0.16711931551326295,
      "grad_norm": 0.7115740455768171,
      "learning_rate": 0.00019849943367547347,
      "loss": 12.5487,
      "step": 3069
    },
    {
      "epoch": 0.16717376950984597,
      "grad_norm": 0.7621711099837365,
      "learning_rate": 0.0001984979113787433,
      "loss": 12.7246,
      "step": 3070
    },
    {
      "epoch": 0.16722822350642896,
      "grad_norm": 0.7449652831037106,
      "learning_rate": 0.00019849638831607682,
      "loss": 12.6964,
      "step": 3071
    },
    {
      "epoch": 0.16728267750301198,
      "grad_norm": 0.7552484504379975,
      "learning_rate": 0.00019849486448748592,
      "loss": 12.73,
      "step": 3072
    },
    {
      "epoch": 0.167337131499595,
      "grad_norm": 0.7649404312476997,
      "learning_rate": 0.00019849333989298246,
      "loss": 12.6314,
      "step": 3073
    },
    {
      "epoch": 0.167391585496178,
      "grad_norm": 0.8408831313771824,
      "learning_rate": 0.00019849181453257826,
      "loss": 12.7805,
      "step": 3074
    },
    {
      "epoch": 0.16744603949276102,
      "grad_norm": 0.7302841277403656,
      "learning_rate": 0.00019849028840628516,
      "loss": 12.5063,
      "step": 3075
    },
    {
      "epoch": 0.16750049348934404,
      "grad_norm": 0.7544122102561781,
      "learning_rate": 0.00019848876151411511,
      "loss": 12.7941,
      "step": 3076
    },
    {
      "epoch": 0.16755494748592706,
      "grad_norm": 0.7918574476711356,
      "learning_rate": 0.00019848723385607989,
      "loss": 12.6725,
      "step": 3077
    },
    {
      "epoch": 0.16760940148251005,
      "grad_norm": 0.751101721987298,
      "learning_rate": 0.00019848570543219146,
      "loss": 12.6924,
      "step": 3078
    },
    {
      "epoch": 0.16766385547909307,
      "grad_norm": 0.6838336506900237,
      "learning_rate": 0.00019848417624246162,
      "loss": 12.6729,
      "step": 3079
    },
    {
      "epoch": 0.1677183094756761,
      "grad_norm": 0.7408676965449393,
      "learning_rate": 0.00019848264628690237,
      "loss": 12.7387,
      "step": 3080
    },
    {
      "epoch": 0.16777276347225908,
      "grad_norm": 0.7696366140174506,
      "learning_rate": 0.00019848111556552552,
      "loss": 12.7856,
      "step": 3081
    },
    {
      "epoch": 0.1678272174688421,
      "grad_norm": 0.8200316785096359,
      "learning_rate": 0.00019847958407834298,
      "loss": 12.6265,
      "step": 3082
    },
    {
      "epoch": 0.16788167146542513,
      "grad_norm": 0.681234990809023,
      "learning_rate": 0.0001984780518253667,
      "loss": 12.6042,
      "step": 3083
    },
    {
      "epoch": 0.16793612546200812,
      "grad_norm": 0.8458155782816993,
      "learning_rate": 0.00019847651880660857,
      "loss": 12.6334,
      "step": 3084
    },
    {
      "epoch": 0.16799057945859114,
      "grad_norm": 0.8541159959566602,
      "learning_rate": 0.00019847498502208053,
      "loss": 12.6321,
      "step": 3085
    },
    {
      "epoch": 0.16804503345517416,
      "grad_norm": 0.7817962520050384,
      "learning_rate": 0.00019847345047179445,
      "loss": 12.6858,
      "step": 3086
    },
    {
      "epoch": 0.16809948745175715,
      "grad_norm": 0.740149125300525,
      "learning_rate": 0.00019847191515576235,
      "loss": 12.6561,
      "step": 3087
    },
    {
      "epoch": 0.16815394144834017,
      "grad_norm": 0.7569091163765901,
      "learning_rate": 0.0001984703790739961,
      "loss": 12.6237,
      "step": 3088
    },
    {
      "epoch": 0.1682083954449232,
      "grad_norm": 0.8628887769130403,
      "learning_rate": 0.00019846884222650768,
      "loss": 12.7761,
      "step": 3089
    },
    {
      "epoch": 0.16826284944150618,
      "grad_norm": 0.8704712641782385,
      "learning_rate": 0.00019846730461330902,
      "loss": 12.76,
      "step": 3090
    },
    {
      "epoch": 0.1683173034380892,
      "grad_norm": 0.7408081447934876,
      "learning_rate": 0.0001984657662344121,
      "loss": 12.6615,
      "step": 3091
    },
    {
      "epoch": 0.16837175743467223,
      "grad_norm": 0.710881601350767,
      "learning_rate": 0.00019846422708982885,
      "loss": 12.5017,
      "step": 3092
    },
    {
      "epoch": 0.16842621143125525,
      "grad_norm": 0.8098661766532628,
      "learning_rate": 0.00019846268717957127,
      "loss": 12.6211,
      "step": 3093
    },
    {
      "epoch": 0.16848066542783824,
      "grad_norm": 0.775012426784901,
      "learning_rate": 0.00019846114650365134,
      "loss": 12.6198,
      "step": 3094
    },
    {
      "epoch": 0.16853511942442126,
      "grad_norm": 0.80326601496314,
      "learning_rate": 0.000198459605062081,
      "loss": 12.6613,
      "step": 3095
    },
    {
      "epoch": 0.16858957342100428,
      "grad_norm": 0.7354652502997332,
      "learning_rate": 0.00019845806285487228,
      "loss": 12.615,
      "step": 3096
    },
    {
      "epoch": 0.16864402741758727,
      "grad_norm": 0.7450886072732027,
      "learning_rate": 0.00019845651988203712,
      "loss": 12.54,
      "step": 3097
    },
    {
      "epoch": 0.1686984814141703,
      "grad_norm": 0.7933552614013386,
      "learning_rate": 0.00019845497614358757,
      "loss": 12.5981,
      "step": 3098
    },
    {
      "epoch": 0.1687529354107533,
      "grad_norm": 0.7968877726596918,
      "learning_rate": 0.0001984534316395356,
      "loss": 12.6495,
      "step": 3099
    },
    {
      "epoch": 0.1688073894073363,
      "grad_norm": 0.7612335336482233,
      "learning_rate": 0.00019845188636989324,
      "loss": 12.7011,
      "step": 3100
    },
    {
      "epoch": 0.16886184340391933,
      "grad_norm": 0.7379465764326891,
      "learning_rate": 0.00019845034033467253,
      "loss": 12.647,
      "step": 3101
    },
    {
      "epoch": 0.16891629740050235,
      "grad_norm": 0.6342380056369946,
      "learning_rate": 0.0001984487935338854,
      "loss": 12.3162,
      "step": 3102
    },
    {
      "epoch": 0.16897075139708534,
      "grad_norm": 0.7992548336863188,
      "learning_rate": 0.000198447245967544,
      "loss": 12.6207,
      "step": 3103
    },
    {
      "epoch": 0.16902520539366836,
      "grad_norm": 0.7347690982602108,
      "learning_rate": 0.0001984456976356603,
      "loss": 12.6189,
      "step": 3104
    },
    {
      "epoch": 0.16907965939025138,
      "grad_norm": 0.7411609138616736,
      "learning_rate": 0.0001984441485382463,
      "loss": 12.6495,
      "step": 3105
    },
    {
      "epoch": 0.16913411338683437,
      "grad_norm": 0.8311106556317931,
      "learning_rate": 0.00019844259867531414,
      "loss": 12.6338,
      "step": 3106
    },
    {
      "epoch": 0.1691885673834174,
      "grad_norm": 0.766495636804695,
      "learning_rate": 0.00019844104804687582,
      "loss": 12.7428,
      "step": 3107
    },
    {
      "epoch": 0.1692430213800004,
      "grad_norm": 0.8369928669689569,
      "learning_rate": 0.0001984394966529434,
      "loss": 12.7464,
      "step": 3108
    },
    {
      "epoch": 0.16929747537658343,
      "grad_norm": 0.8005274422552616,
      "learning_rate": 0.00019843794449352892,
      "loss": 12.5939,
      "step": 3109
    },
    {
      "epoch": 0.16935192937316643,
      "grad_norm": 0.6739084256004799,
      "learning_rate": 0.0001984363915686445,
      "loss": 12.4955,
      "step": 3110
    },
    {
      "epoch": 0.16940638336974945,
      "grad_norm": 0.8504633667447858,
      "learning_rate": 0.00019843483787830216,
      "loss": 12.6238,
      "step": 3111
    },
    {
      "epoch": 0.16946083736633247,
      "grad_norm": 0.8152522438802968,
      "learning_rate": 0.00019843328342251406,
      "loss": 12.8807,
      "step": 3112
    },
    {
      "epoch": 0.16951529136291546,
      "grad_norm": 0.7834123281067574,
      "learning_rate": 0.0001984317282012922,
      "loss": 12.7353,
      "step": 3113
    },
    {
      "epoch": 0.16956974535949848,
      "grad_norm": 0.7587210169588774,
      "learning_rate": 0.00019843017221464875,
      "loss": 12.6709,
      "step": 3114
    },
    {
      "epoch": 0.1696241993560815,
      "grad_norm": 0.8104141569775231,
      "learning_rate": 0.00019842861546259574,
      "loss": 12.6332,
      "step": 3115
    },
    {
      "epoch": 0.1696786533526645,
      "grad_norm": 0.959464087999072,
      "learning_rate": 0.00019842705794514533,
      "loss": 12.6767,
      "step": 3116
    },
    {
      "epoch": 0.16973310734924751,
      "grad_norm": 0.7914692125288751,
      "learning_rate": 0.00019842549966230961,
      "loss": 12.8013,
      "step": 3117
    },
    {
      "epoch": 0.16978756134583053,
      "grad_norm": 0.6492714647380093,
      "learning_rate": 0.0001984239406141007,
      "loss": 12.5668,
      "step": 3118
    },
    {
      "epoch": 0.16984201534241353,
      "grad_norm": 0.9351858633515623,
      "learning_rate": 0.00019842238080053072,
      "loss": 12.64,
      "step": 3119
    },
    {
      "epoch": 0.16989646933899655,
      "grad_norm": 0.8610702832263224,
      "learning_rate": 0.00019842082022161178,
      "loss": 12.7856,
      "step": 3120
    },
    {
      "epoch": 0.16995092333557957,
      "grad_norm": 0.7780332218014057,
      "learning_rate": 0.00019841925887735607,
      "loss": 12.6922,
      "step": 3121
    },
    {
      "epoch": 0.17000537733216256,
      "grad_norm": 0.8377695759243816,
      "learning_rate": 0.0001984176967677757,
      "loss": 12.7686,
      "step": 3122
    },
    {
      "epoch": 0.17005983132874558,
      "grad_norm": 0.8259157927171062,
      "learning_rate": 0.0001984161338928828,
      "loss": 12.6289,
      "step": 3123
    },
    {
      "epoch": 0.1701142853253286,
      "grad_norm": 0.797915984011378,
      "learning_rate": 0.00019841457025268953,
      "loss": 12.6509,
      "step": 3124
    },
    {
      "epoch": 0.1701687393219116,
      "grad_norm": 0.7578127442759067,
      "learning_rate": 0.00019841300584720807,
      "loss": 12.6557,
      "step": 3125
    },
    {
      "epoch": 0.17022319331849461,
      "grad_norm": 0.8260447831694434,
      "learning_rate": 0.00019841144067645058,
      "loss": 12.6095,
      "step": 3126
    },
    {
      "epoch": 0.17027764731507763,
      "grad_norm": 0.9170401923117737,
      "learning_rate": 0.00019840987474042918,
      "loss": 12.6943,
      "step": 3127
    },
    {
      "epoch": 0.17033210131166066,
      "grad_norm": 0.6838717466562656,
      "learning_rate": 0.0001984083080391561,
      "loss": 12.5913,
      "step": 3128
    },
    {
      "epoch": 0.17038655530824365,
      "grad_norm": 0.6734878857682747,
      "learning_rate": 0.00019840674057264356,
      "loss": 12.5914,
      "step": 3129
    },
    {
      "epoch": 0.17044100930482667,
      "grad_norm": 0.8851280009559688,
      "learning_rate": 0.00019840517234090367,
      "loss": 12.7171,
      "step": 3130
    },
    {
      "epoch": 0.1704954633014097,
      "grad_norm": 0.8336628423928786,
      "learning_rate": 0.00019840360334394869,
      "loss": 12.5769,
      "step": 3131
    },
    {
      "epoch": 0.17054991729799268,
      "grad_norm": 0.8233460415229792,
      "learning_rate": 0.00019840203358179075,
      "loss": 12.6803,
      "step": 3132
    },
    {
      "epoch": 0.1706043712945757,
      "grad_norm": 0.8575100075433871,
      "learning_rate": 0.00019840046305444212,
      "loss": 12.6334,
      "step": 3133
    },
    {
      "epoch": 0.17065882529115872,
      "grad_norm": 0.8402671498996254,
      "learning_rate": 0.00019839889176191498,
      "loss": 12.7051,
      "step": 3134
    },
    {
      "epoch": 0.17071327928774171,
      "grad_norm": 0.8442756476235956,
      "learning_rate": 0.00019839731970422152,
      "loss": 12.598,
      "step": 3135
    },
    {
      "epoch": 0.17076773328432474,
      "grad_norm": 0.7790665163489613,
      "learning_rate": 0.00019839574688137405,
      "loss": 12.673,
      "step": 3136
    },
    {
      "epoch": 0.17082218728090776,
      "grad_norm": 0.6776444996263663,
      "learning_rate": 0.00019839417329338473,
      "loss": 12.5547,
      "step": 3137
    },
    {
      "epoch": 0.17087664127749075,
      "grad_norm": 0.7170712222884784,
      "learning_rate": 0.00019839259894026583,
      "loss": 12.6168,
      "step": 3138
    },
    {
      "epoch": 0.17093109527407377,
      "grad_norm": 0.8250292301702511,
      "learning_rate": 0.00019839102382202957,
      "loss": 12.6966,
      "step": 3139
    },
    {
      "epoch": 0.1709855492706568,
      "grad_norm": 0.776378688447562,
      "learning_rate": 0.00019838944793868822,
      "loss": 12.6753,
      "step": 3140
    },
    {
      "epoch": 0.17104000326723978,
      "grad_norm": 0.7560489984821958,
      "learning_rate": 0.000198387871290254,
      "loss": 12.5989,
      "step": 3141
    },
    {
      "epoch": 0.1710944572638228,
      "grad_norm": 0.7974557424591885,
      "learning_rate": 0.0001983862938767392,
      "loss": 12.5405,
      "step": 3142
    },
    {
      "epoch": 0.17114891126040582,
      "grad_norm": 0.7014349054792098,
      "learning_rate": 0.0001983847156981561,
      "loss": 12.5353,
      "step": 3143
    },
    {
      "epoch": 0.17120336525698884,
      "grad_norm": 0.723315929267737,
      "learning_rate": 0.00019838313675451695,
      "loss": 12.6474,
      "step": 3144
    },
    {
      "epoch": 0.17125781925357184,
      "grad_norm": 0.9779302802477249,
      "learning_rate": 0.00019838155704583402,
      "loss": 12.8977,
      "step": 3145
    },
    {
      "epoch": 0.17131227325015486,
      "grad_norm": 0.7365367954267793,
      "learning_rate": 0.00019837997657211958,
      "loss": 12.68,
      "step": 3146
    },
    {
      "epoch": 0.17136672724673788,
      "grad_norm": 0.7836697915049639,
      "learning_rate": 0.00019837839533338599,
      "loss": 12.5582,
      "step": 3147
    },
    {
      "epoch": 0.17142118124332087,
      "grad_norm": 0.694905046743744,
      "learning_rate": 0.00019837681332964544,
      "loss": 12.6245,
      "step": 3148
    },
    {
      "epoch": 0.1714756352399039,
      "grad_norm": 0.7345872077956427,
      "learning_rate": 0.00019837523056091035,
      "loss": 12.6434,
      "step": 3149
    },
    {
      "epoch": 0.1715300892364869,
      "grad_norm": 0.8676001382683689,
      "learning_rate": 0.00019837364702719295,
      "loss": 12.7774,
      "step": 3150
    },
    {
      "epoch": 0.1715845432330699,
      "grad_norm": 0.749859401437326,
      "learning_rate": 0.00019837206272850554,
      "loss": 12.7502,
      "step": 3151
    },
    {
      "epoch": 0.17163899722965292,
      "grad_norm": 0.7496417092437363,
      "learning_rate": 0.00019837047766486053,
      "loss": 12.7435,
      "step": 3152
    },
    {
      "epoch": 0.17169345122623594,
      "grad_norm": 0.7099237668025427,
      "learning_rate": 0.00019836889183627015,
      "loss": 12.495,
      "step": 3153
    },
    {
      "epoch": 0.17174790522281894,
      "grad_norm": 0.7285641308612487,
      "learning_rate": 0.0001983673052427468,
      "loss": 12.5386,
      "step": 3154
    },
    {
      "epoch": 0.17180235921940196,
      "grad_norm": 0.7009655388375987,
      "learning_rate": 0.00019836571788430275,
      "loss": 12.7553,
      "step": 3155
    },
    {
      "epoch": 0.17185681321598498,
      "grad_norm": 0.7463493574611223,
      "learning_rate": 0.0001983641297609504,
      "loss": 12.636,
      "step": 3156
    },
    {
      "epoch": 0.17191126721256797,
      "grad_norm": 0.787048181874437,
      "learning_rate": 0.00019836254087270206,
      "loss": 12.7262,
      "step": 3157
    },
    {
      "epoch": 0.171965721209151,
      "grad_norm": 0.814045454316862,
      "learning_rate": 0.00019836095121957014,
      "loss": 12.6681,
      "step": 3158
    },
    {
      "epoch": 0.172020175205734,
      "grad_norm": 0.8867722235844003,
      "learning_rate": 0.00019835936080156698,
      "loss": 12.5865,
      "step": 3159
    },
    {
      "epoch": 0.17207462920231703,
      "grad_norm": 0.7674062863483792,
      "learning_rate": 0.00019835776961870492,
      "loss": 12.5926,
      "step": 3160
    },
    {
      "epoch": 0.17212908319890002,
      "grad_norm": 0.877774539505825,
      "learning_rate": 0.00019835617767099633,
      "loss": 12.604,
      "step": 3161
    },
    {
      "epoch": 0.17218353719548304,
      "grad_norm": 0.7515196326795087,
      "learning_rate": 0.00019835458495845362,
      "loss": 12.683,
      "step": 3162
    },
    {
      "epoch": 0.17223799119206606,
      "grad_norm": 0.7810814803217557,
      "learning_rate": 0.00019835299148108918,
      "loss": 12.5438,
      "step": 3163
    },
    {
      "epoch": 0.17229244518864906,
      "grad_norm": 0.8523834949136776,
      "learning_rate": 0.00019835139723891536,
      "loss": 12.7393,
      "step": 3164
    },
    {
      "epoch": 0.17234689918523208,
      "grad_norm": 0.7607407245315153,
      "learning_rate": 0.0001983498022319446,
      "loss": 12.7176,
      "step": 3165
    },
    {
      "epoch": 0.1724013531818151,
      "grad_norm": 0.76191705067529,
      "learning_rate": 0.00019834820646018927,
      "loss": 12.5316,
      "step": 3166
    },
    {
      "epoch": 0.1724558071783981,
      "grad_norm": 0.7260989726107212,
      "learning_rate": 0.0001983466099236618,
      "loss": 12.7281,
      "step": 3167
    },
    {
      "epoch": 0.1725102611749811,
      "grad_norm": 0.7763558291867316,
      "learning_rate": 0.00019834501262237458,
      "loss": 12.6782,
      "step": 3168
    },
    {
      "epoch": 0.17256471517156413,
      "grad_norm": 0.7983010911998738,
      "learning_rate": 0.0001983434145563401,
      "loss": 12.7117,
      "step": 3169
    },
    {
      "epoch": 0.17261916916814712,
      "grad_norm": 0.7605819883968827,
      "learning_rate": 0.00019834181572557066,
      "loss": 12.7948,
      "step": 3170
    },
    {
      "epoch": 0.17267362316473014,
      "grad_norm": 0.8374294116547367,
      "learning_rate": 0.00019834021613007882,
      "loss": 12.7513,
      "step": 3171
    },
    {
      "epoch": 0.17272807716131316,
      "grad_norm": 0.6585356583533877,
      "learning_rate": 0.00019833861576987698,
      "loss": 12.5781,
      "step": 3172
    },
    {
      "epoch": 0.17278253115789616,
      "grad_norm": 0.9425149602084731,
      "learning_rate": 0.00019833701464497754,
      "loss": 12.7727,
      "step": 3173
    },
    {
      "epoch": 0.17283698515447918,
      "grad_norm": 0.7464082223439957,
      "learning_rate": 0.000198335412755393,
      "loss": 12.7494,
      "step": 3174
    },
    {
      "epoch": 0.1728914391510622,
      "grad_norm": 0.8587155233360285,
      "learning_rate": 0.00019833381010113578,
      "loss": 12.5224,
      "step": 3175
    },
    {
      "epoch": 0.17294589314764522,
      "grad_norm": 0.747952703741679,
      "learning_rate": 0.0001983322066822184,
      "loss": 12.4944,
      "step": 3176
    },
    {
      "epoch": 0.1730003471442282,
      "grad_norm": 0.7827107540027283,
      "learning_rate": 0.00019833060249865327,
      "loss": 12.6914,
      "step": 3177
    },
    {
      "epoch": 0.17305480114081123,
      "grad_norm": 0.7433316310571197,
      "learning_rate": 0.0001983289975504529,
      "loss": 12.4551,
      "step": 3178
    },
    {
      "epoch": 0.17310925513739425,
      "grad_norm": 0.8347754411825807,
      "learning_rate": 0.00019832739183762969,
      "loss": 12.7181,
      "step": 3179
    },
    {
      "epoch": 0.17316370913397724,
      "grad_norm": 0.7751754459911796,
      "learning_rate": 0.00019832578536019626,
      "loss": 12.6883,
      "step": 3180
    },
    {
      "epoch": 0.17321816313056027,
      "grad_norm": 0.8741734159288691,
      "learning_rate": 0.00019832417811816501,
      "loss": 12.7347,
      "step": 3181
    },
    {
      "epoch": 0.17327261712714329,
      "grad_norm": 0.8318468746826131,
      "learning_rate": 0.00019832257011154845,
      "loss": 12.6819,
      "step": 3182
    },
    {
      "epoch": 0.17332707112372628,
      "grad_norm": 0.7039626157093785,
      "learning_rate": 0.0001983209613403591,
      "loss": 12.674,
      "step": 3183
    },
    {
      "epoch": 0.1733815251203093,
      "grad_norm": 0.9025010506269673,
      "learning_rate": 0.0001983193518046095,
      "loss": 12.5735,
      "step": 3184
    },
    {
      "epoch": 0.17343597911689232,
      "grad_norm": 0.7546511643836435,
      "learning_rate": 0.00019831774150431213,
      "loss": 12.6793,
      "step": 3185
    },
    {
      "epoch": 0.1734904331134753,
      "grad_norm": 0.8394354751748568,
      "learning_rate": 0.00019831613043947948,
      "loss": 12.6723,
      "step": 3186
    },
    {
      "epoch": 0.17354488711005833,
      "grad_norm": 0.7574481668785278,
      "learning_rate": 0.0001983145186101241,
      "loss": 12.5683,
      "step": 3187
    },
    {
      "epoch": 0.17359934110664135,
      "grad_norm": 0.9026138172851492,
      "learning_rate": 0.0001983129060162586,
      "loss": 12.638,
      "step": 3188
    },
    {
      "epoch": 0.17365379510322435,
      "grad_norm": 0.6765043031700043,
      "learning_rate": 0.00019831129265789537,
      "loss": 12.6092,
      "step": 3189
    },
    {
      "epoch": 0.17370824909980737,
      "grad_norm": 0.7886541026462125,
      "learning_rate": 0.0001983096785350471,
      "loss": 12.7237,
      "step": 3190
    },
    {
      "epoch": 0.1737627030963904,
      "grad_norm": 0.7844971851503274,
      "learning_rate": 0.0001983080636477263,
      "loss": 12.7892,
      "step": 3191
    },
    {
      "epoch": 0.17381715709297338,
      "grad_norm": 0.6930078606339654,
      "learning_rate": 0.00019830644799594547,
      "loss": 12.5887,
      "step": 3192
    },
    {
      "epoch": 0.1738716110895564,
      "grad_norm": 0.7923600954734632,
      "learning_rate": 0.00019830483157971723,
      "loss": 12.6886,
      "step": 3193
    },
    {
      "epoch": 0.17392606508613942,
      "grad_norm": 0.8338987623626151,
      "learning_rate": 0.00019830321439905414,
      "loss": 12.7361,
      "step": 3194
    },
    {
      "epoch": 0.17398051908272244,
      "grad_norm": 0.7749506167348895,
      "learning_rate": 0.00019830159645396875,
      "loss": 12.849,
      "step": 3195
    },
    {
      "epoch": 0.17403497307930543,
      "grad_norm": 0.7390801603555475,
      "learning_rate": 0.00019829997774447366,
      "loss": 12.5025,
      "step": 3196
    },
    {
      "epoch": 0.17408942707588845,
      "grad_norm": 0.6948756460703287,
      "learning_rate": 0.0001982983582705815,
      "loss": 12.6153,
      "step": 3197
    },
    {
      "epoch": 0.17414388107247147,
      "grad_norm": 0.7011689581732615,
      "learning_rate": 0.00019829673803230477,
      "loss": 12.7191,
      "step": 3198
    },
    {
      "epoch": 0.17419833506905447,
      "grad_norm": 0.6850588411098556,
      "learning_rate": 0.00019829511702965612,
      "loss": 12.5695,
      "step": 3199
    },
    {
      "epoch": 0.1742527890656375,
      "grad_norm": 0.7668874363449892,
      "learning_rate": 0.0001982934952626482,
      "loss": 12.7662,
      "step": 3200
    },
    {
      "epoch": 0.1743072430622205,
      "grad_norm": 0.7504905218030471,
      "learning_rate": 0.00019829187273129355,
      "loss": 12.6012,
      "step": 3201
    },
    {
      "epoch": 0.1743616970588035,
      "grad_norm": 0.7171816702004552,
      "learning_rate": 0.00019829024943560485,
      "loss": 12.6831,
      "step": 3202
    },
    {
      "epoch": 0.17441615105538652,
      "grad_norm": 0.698158237637559,
      "learning_rate": 0.00019828862537559462,
      "loss": 12.5953,
      "step": 3203
    },
    {
      "epoch": 0.17447060505196954,
      "grad_norm": 0.8732362510339932,
      "learning_rate": 0.0001982870005512756,
      "loss": 12.6716,
      "step": 3204
    },
    {
      "epoch": 0.17452505904855253,
      "grad_norm": 0.8154029138565516,
      "learning_rate": 0.00019828537496266037,
      "loss": 12.6764,
      "step": 3205
    },
    {
      "epoch": 0.17457951304513555,
      "grad_norm": 0.7699923056388698,
      "learning_rate": 0.0001982837486097616,
      "loss": 12.6628,
      "step": 3206
    },
    {
      "epoch": 0.17463396704171857,
      "grad_norm": 0.7861215643164291,
      "learning_rate": 0.00019828212149259188,
      "loss": 12.6717,
      "step": 3207
    },
    {
      "epoch": 0.17468842103830157,
      "grad_norm": 0.6983338883252923,
      "learning_rate": 0.0001982804936111639,
      "loss": 12.6699,
      "step": 3208
    },
    {
      "epoch": 0.1747428750348846,
      "grad_norm": 0.8298596274818021,
      "learning_rate": 0.00019827886496549037,
      "loss": 12.5887,
      "step": 3209
    },
    {
      "epoch": 0.1747973290314676,
      "grad_norm": 0.7776100334544354,
      "learning_rate": 0.0001982772355555839,
      "loss": 12.6827,
      "step": 3210
    },
    {
      "epoch": 0.17485178302805063,
      "grad_norm": 0.7710869955834241,
      "learning_rate": 0.00019827560538145713,
      "loss": 12.6449,
      "step": 3211
    },
    {
      "epoch": 0.17490623702463362,
      "grad_norm": 0.7444032286300318,
      "learning_rate": 0.0001982739744431228,
      "loss": 12.6241,
      "step": 3212
    },
    {
      "epoch": 0.17496069102121664,
      "grad_norm": 0.8584217284851458,
      "learning_rate": 0.00019827234274059352,
      "loss": 12.6233,
      "step": 3213
    },
    {
      "epoch": 0.17501514501779966,
      "grad_norm": 0.7472171044808118,
      "learning_rate": 0.00019827071027388207,
      "loss": 12.7727,
      "step": 3214
    },
    {
      "epoch": 0.17506959901438265,
      "grad_norm": 0.7241449526372566,
      "learning_rate": 0.00019826907704300107,
      "loss": 12.6059,
      "step": 3215
    },
    {
      "epoch": 0.17512405301096567,
      "grad_norm": 0.6725069804670222,
      "learning_rate": 0.00019826744304796327,
      "loss": 12.5751,
      "step": 3216
    },
    {
      "epoch": 0.1751785070075487,
      "grad_norm": 0.6894594253663173,
      "learning_rate": 0.00019826580828878135,
      "loss": 12.6274,
      "step": 3217
    },
    {
      "epoch": 0.1752329610041317,
      "grad_norm": 0.7304834406583183,
      "learning_rate": 0.000198264172765468,
      "loss": 12.7152,
      "step": 3218
    },
    {
      "epoch": 0.1752874150007147,
      "grad_norm": 0.8386256205995041,
      "learning_rate": 0.00019826253647803598,
      "loss": 12.6445,
      "step": 3219
    },
    {
      "epoch": 0.17534186899729773,
      "grad_norm": 1.270172642023828,
      "learning_rate": 0.000198260899426498,
      "loss": 12.5894,
      "step": 3220
    },
    {
      "epoch": 0.17539632299388072,
      "grad_norm": 0.7344319767853419,
      "learning_rate": 0.00019825926161086679,
      "loss": 12.7958,
      "step": 3221
    },
    {
      "epoch": 0.17545077699046374,
      "grad_norm": 0.734680231052794,
      "learning_rate": 0.00019825762303115506,
      "loss": 12.6942,
      "step": 3222
    },
    {
      "epoch": 0.17550523098704676,
      "grad_norm": 0.8336006775340267,
      "learning_rate": 0.00019825598368737558,
      "loss": 12.8308,
      "step": 3223
    },
    {
      "epoch": 0.17555968498362975,
      "grad_norm": 0.7262882153581018,
      "learning_rate": 0.00019825434357954112,
      "loss": 12.6866,
      "step": 3224
    },
    {
      "epoch": 0.17561413898021278,
      "grad_norm": 0.7392613877216176,
      "learning_rate": 0.0001982527027076644,
      "loss": 12.6137,
      "step": 3225
    },
    {
      "epoch": 0.1756685929767958,
      "grad_norm": 0.6975694053545508,
      "learning_rate": 0.0001982510610717582,
      "loss": 12.6543,
      "step": 3226
    },
    {
      "epoch": 0.17572304697337882,
      "grad_norm": 0.7534491680096834,
      "learning_rate": 0.00019824941867183524,
      "loss": 12.6165,
      "step": 3227
    },
    {
      "epoch": 0.1757775009699618,
      "grad_norm": 0.7788563975747334,
      "learning_rate": 0.00019824777550790835,
      "loss": 12.5636,
      "step": 3228
    },
    {
      "epoch": 0.17583195496654483,
      "grad_norm": 0.7506448121279704,
      "learning_rate": 0.00019824613157999027,
      "loss": 12.5972,
      "step": 3229
    },
    {
      "epoch": 0.17588640896312785,
      "grad_norm": 0.6771530701488644,
      "learning_rate": 0.0001982444868880938,
      "loss": 12.6503,
      "step": 3230
    },
    {
      "epoch": 0.17594086295971084,
      "grad_norm": 0.8799280336028142,
      "learning_rate": 0.0001982428414322317,
      "loss": 12.7639,
      "step": 3231
    },
    {
      "epoch": 0.17599531695629386,
      "grad_norm": 0.7902749117766416,
      "learning_rate": 0.0001982411952124168,
      "loss": 12.768,
      "step": 3232
    },
    {
      "epoch": 0.17604977095287688,
      "grad_norm": 0.6973865579783748,
      "learning_rate": 0.00019823954822866191,
      "loss": 12.5541,
      "step": 3233
    },
    {
      "epoch": 0.17610422494945988,
      "grad_norm": 0.6975249553608402,
      "learning_rate": 0.0001982379004809798,
      "loss": 12.6062,
      "step": 3234
    },
    {
      "epoch": 0.1761586789460429,
      "grad_norm": 0.7168173724680111,
      "learning_rate": 0.0001982362519693833,
      "loss": 12.7842,
      "step": 3235
    },
    {
      "epoch": 0.17621313294262592,
      "grad_norm": 0.7565153526890861,
      "learning_rate": 0.00019823460269388526,
      "loss": 12.7059,
      "step": 3236
    },
    {
      "epoch": 0.1762675869392089,
      "grad_norm": 0.8209036578975398,
      "learning_rate": 0.00019823295265449843,
      "loss": 12.6638,
      "step": 3237
    },
    {
      "epoch": 0.17632204093579193,
      "grad_norm": 0.7981299201729861,
      "learning_rate": 0.00019823130185123573,
      "loss": 12.7187,
      "step": 3238
    },
    {
      "epoch": 0.17637649493237495,
      "grad_norm": 0.7747231650804178,
      "learning_rate": 0.00019822965028410994,
      "loss": 12.6225,
      "step": 3239
    },
    {
      "epoch": 0.17643094892895794,
      "grad_norm": 0.7333857540750321,
      "learning_rate": 0.0001982279979531339,
      "loss": 12.6742,
      "step": 3240
    },
    {
      "epoch": 0.17648540292554096,
      "grad_norm": 0.8397253043859492,
      "learning_rate": 0.0001982263448583205,
      "loss": 12.7436,
      "step": 3241
    },
    {
      "epoch": 0.17653985692212398,
      "grad_norm": 0.7815864103858401,
      "learning_rate": 0.00019822469099968254,
      "loss": 12.6333,
      "step": 3242
    },
    {
      "epoch": 0.176594310918707,
      "grad_norm": 0.8209808262548153,
      "learning_rate": 0.0001982230363772329,
      "loss": 12.7196,
      "step": 3243
    },
    {
      "epoch": 0.17664876491529,
      "grad_norm": 0.7262043789621151,
      "learning_rate": 0.0001982213809909845,
      "loss": 12.6364,
      "step": 3244
    },
    {
      "epoch": 0.17670321891187302,
      "grad_norm": 0.8607138924882243,
      "learning_rate": 0.00019821972484095017,
      "loss": 12.7051,
      "step": 3245
    },
    {
      "epoch": 0.17675767290845604,
      "grad_norm": 0.8057809652765885,
      "learning_rate": 0.00019821806792714276,
      "loss": 12.5776,
      "step": 3246
    },
    {
      "epoch": 0.17681212690503903,
      "grad_norm": 0.7042559054435703,
      "learning_rate": 0.0001982164102495752,
      "loss": 12.4916,
      "step": 3247
    },
    {
      "epoch": 0.17686658090162205,
      "grad_norm": 1.057714095909052,
      "learning_rate": 0.00019821475180826034,
      "loss": 12.668,
      "step": 3248
    },
    {
      "epoch": 0.17692103489820507,
      "grad_norm": 0.7899993707377523,
      "learning_rate": 0.00019821309260321113,
      "loss": 12.5721,
      "step": 3249
    },
    {
      "epoch": 0.17697548889478806,
      "grad_norm": 0.8471101192669516,
      "learning_rate": 0.00019821143263444043,
      "loss": 12.5475,
      "step": 3250
    },
    {
      "epoch": 0.17702994289137108,
      "grad_norm": 0.6735795320454601,
      "learning_rate": 0.00019820977190196116,
      "loss": 12.5648,
      "step": 3251
    },
    {
      "epoch": 0.1770843968879541,
      "grad_norm": 0.9204359196176464,
      "learning_rate": 0.00019820811040578625,
      "loss": 12.9584,
      "step": 3252
    },
    {
      "epoch": 0.1771388508845371,
      "grad_norm": 0.9138396791378408,
      "learning_rate": 0.00019820644814592858,
      "loss": 12.6978,
      "step": 3253
    },
    {
      "epoch": 0.17719330488112012,
      "grad_norm": 0.7301330238149067,
      "learning_rate": 0.0001982047851224011,
      "loss": 12.545,
      "step": 3254
    },
    {
      "epoch": 0.17724775887770314,
      "grad_norm": 0.9013668175356127,
      "learning_rate": 0.00019820312133521674,
      "loss": 12.6496,
      "step": 3255
    },
    {
      "epoch": 0.17730221287428613,
      "grad_norm": 0.9341817253221993,
      "learning_rate": 0.00019820145678438845,
      "loss": 12.6747,
      "step": 3256
    },
    {
      "epoch": 0.17735666687086915,
      "grad_norm": 0.7502328588779547,
      "learning_rate": 0.00019819979146992915,
      "loss": 12.6536,
      "step": 3257
    },
    {
      "epoch": 0.17741112086745217,
      "grad_norm": 0.7681904747639378,
      "learning_rate": 0.0001981981253918518,
      "loss": 12.7236,
      "step": 3258
    },
    {
      "epoch": 0.17746557486403516,
      "grad_norm": 0.8920610418799703,
      "learning_rate": 0.0001981964585501694,
      "loss": 12.6539,
      "step": 3259
    },
    {
      "epoch": 0.17752002886061818,
      "grad_norm": 0.7628111397992703,
      "learning_rate": 0.00019819479094489482,
      "loss": 12.714,
      "step": 3260
    },
    {
      "epoch": 0.1775744828572012,
      "grad_norm": 0.7071311753164595,
      "learning_rate": 0.00019819312257604113,
      "loss": 12.6685,
      "step": 3261
    },
    {
      "epoch": 0.17762893685378423,
      "grad_norm": 0.8942352428065906,
      "learning_rate": 0.0001981914534436212,
      "loss": 12.7463,
      "step": 3262
    },
    {
      "epoch": 0.17768339085036722,
      "grad_norm": 0.8637475451289109,
      "learning_rate": 0.0001981897835476481,
      "loss": 12.5307,
      "step": 3263
    },
    {
      "epoch": 0.17773784484695024,
      "grad_norm": 0.9470761933823619,
      "learning_rate": 0.00019818811288813476,
      "loss": 12.7053,
      "step": 3264
    },
    {
      "epoch": 0.17779229884353326,
      "grad_norm": 0.7384704398906923,
      "learning_rate": 0.0001981864414650942,
      "loss": 12.6304,
      "step": 3265
    },
    {
      "epoch": 0.17784675284011625,
      "grad_norm": 0.7974334541792593,
      "learning_rate": 0.00019818476927853937,
      "loss": 12.7698,
      "step": 3266
    },
    {
      "epoch": 0.17790120683669927,
      "grad_norm": 0.728317539288044,
      "learning_rate": 0.00019818309632848333,
      "loss": 12.6083,
      "step": 3267
    },
    {
      "epoch": 0.1779556608332823,
      "grad_norm": 0.6856666037687628,
      "learning_rate": 0.00019818142261493908,
      "loss": 12.6041,
      "step": 3268
    },
    {
      "epoch": 0.17801011482986528,
      "grad_norm": 0.8080880119981516,
      "learning_rate": 0.00019817974813791963,
      "loss": 12.7661,
      "step": 3269
    },
    {
      "epoch": 0.1780645688264483,
      "grad_norm": 0.7107996562679639,
      "learning_rate": 0.00019817807289743798,
      "loss": 12.6347,
      "step": 3270
    },
    {
      "epoch": 0.17811902282303133,
      "grad_norm": 0.790679300367646,
      "learning_rate": 0.00019817639689350715,
      "loss": 12.6127,
      "step": 3271
    },
    {
      "epoch": 0.17817347681961432,
      "grad_norm": 0.7567804862897068,
      "learning_rate": 0.00019817472012614021,
      "loss": 12.7102,
      "step": 3272
    },
    {
      "epoch": 0.17822793081619734,
      "grad_norm": 0.8069231009329779,
      "learning_rate": 0.00019817304259535018,
      "loss": 12.7045,
      "step": 3273
    },
    {
      "epoch": 0.17828238481278036,
      "grad_norm": 0.7595982136123148,
      "learning_rate": 0.00019817136430115012,
      "loss": 12.6997,
      "step": 3274
    },
    {
      "epoch": 0.17833683880936335,
      "grad_norm": 0.7309321844207629,
      "learning_rate": 0.00019816968524355306,
      "loss": 12.6139,
      "step": 3275
    },
    {
      "epoch": 0.17839129280594637,
      "grad_norm": 0.7192650353233917,
      "learning_rate": 0.0001981680054225721,
      "loss": 12.6906,
      "step": 3276
    },
    {
      "epoch": 0.1784457468025294,
      "grad_norm": 0.7683459462222892,
      "learning_rate": 0.0001981663248382202,
      "loss": 12.758,
      "step": 3277
    },
    {
      "epoch": 0.1785002007991124,
      "grad_norm": 0.7604309449895051,
      "learning_rate": 0.00019816464349051055,
      "loss": 12.708,
      "step": 3278
    },
    {
      "epoch": 0.1785546547956954,
      "grad_norm": 0.6915703638272156,
      "learning_rate": 0.00019816296137945612,
      "loss": 12.638,
      "step": 3279
    },
    {
      "epoch": 0.17860910879227843,
      "grad_norm": 1.1253518154266444,
      "learning_rate": 0.00019816127850507008,
      "loss": 12.7063,
      "step": 3280
    },
    {
      "epoch": 0.17866356278886145,
      "grad_norm": 0.7630225540272567,
      "learning_rate": 0.00019815959486736543,
      "loss": 12.627,
      "step": 3281
    },
    {
      "epoch": 0.17871801678544444,
      "grad_norm": 0.9935145198532191,
      "learning_rate": 0.00019815791046635538,
      "loss": 12.7206,
      "step": 3282
    },
    {
      "epoch": 0.17877247078202746,
      "grad_norm": 0.736334650149649,
      "learning_rate": 0.0001981562253020529,
      "loss": 12.8125,
      "step": 3283
    },
    {
      "epoch": 0.17882692477861048,
      "grad_norm": 0.7808149186506571,
      "learning_rate": 0.00019815453937447117,
      "loss": 12.6405,
      "step": 3284
    },
    {
      "epoch": 0.17888137877519347,
      "grad_norm": 0.8210946819500441,
      "learning_rate": 0.0001981528526836233,
      "loss": 12.6571,
      "step": 3285
    },
    {
      "epoch": 0.1789358327717765,
      "grad_norm": 0.86965004642108,
      "learning_rate": 0.00019815116522952235,
      "loss": 12.7401,
      "step": 3286
    },
    {
      "epoch": 0.1789902867683595,
      "grad_norm": 0.7868117017342933,
      "learning_rate": 0.0001981494770121815,
      "loss": 12.5413,
      "step": 3287
    },
    {
      "epoch": 0.1790447407649425,
      "grad_norm": 0.7415827071978902,
      "learning_rate": 0.00019814778803161387,
      "loss": 12.6923,
      "step": 3288
    },
    {
      "epoch": 0.17909919476152553,
      "grad_norm": 0.6851120650197461,
      "learning_rate": 0.00019814609828783258,
      "loss": 12.56,
      "step": 3289
    },
    {
      "epoch": 0.17915364875810855,
      "grad_norm": 0.8225961915115911,
      "learning_rate": 0.00019814440778085076,
      "loss": 12.6163,
      "step": 3290
    },
    {
      "epoch": 0.17920810275469154,
      "grad_norm": 0.7409751466179318,
      "learning_rate": 0.00019814271651068154,
      "loss": 12.6108,
      "step": 3291
    },
    {
      "epoch": 0.17926255675127456,
      "grad_norm": 0.7492258593142511,
      "learning_rate": 0.0001981410244773381,
      "loss": 12.5378,
      "step": 3292
    },
    {
      "epoch": 0.17931701074785758,
      "grad_norm": 0.7253161388755572,
      "learning_rate": 0.00019813933168083363,
      "loss": 12.7377,
      "step": 3293
    },
    {
      "epoch": 0.1793714647444406,
      "grad_norm": 0.8344633552358383,
      "learning_rate": 0.00019813763812118125,
      "loss": 12.7151,
      "step": 3294
    },
    {
      "epoch": 0.1794259187410236,
      "grad_norm": 0.7765629199089185,
      "learning_rate": 0.00019813594379839415,
      "loss": 12.5645,
      "step": 3295
    },
    {
      "epoch": 0.1794803727376066,
      "grad_norm": 0.8082931219937859,
      "learning_rate": 0.00019813424871248546,
      "loss": 12.7144,
      "step": 3296
    },
    {
      "epoch": 0.17953482673418963,
      "grad_norm": 0.7589009678676194,
      "learning_rate": 0.00019813255286346842,
      "loss": 12.7288,
      "step": 3297
    },
    {
      "epoch": 0.17958928073077263,
      "grad_norm": 0.790735241959672,
      "learning_rate": 0.0001981308562513562,
      "loss": 12.6141,
      "step": 3298
    },
    {
      "epoch": 0.17964373472735565,
      "grad_norm": 0.7073433938535153,
      "learning_rate": 0.00019812915887616194,
      "loss": 12.5315,
      "step": 3299
    },
    {
      "epoch": 0.17969818872393867,
      "grad_norm": 0.7446488567697064,
      "learning_rate": 0.00019812746073789893,
      "loss": 12.6907,
      "step": 3300
    },
    {
      "epoch": 0.17975264272052166,
      "grad_norm": 0.7491534452924706,
      "learning_rate": 0.0001981257618365803,
      "loss": 12.6864,
      "step": 3301
    },
    {
      "epoch": 0.17980709671710468,
      "grad_norm": 0.821475530739584,
      "learning_rate": 0.0001981240621722193,
      "loss": 12.6571,
      "step": 3302
    },
    {
      "epoch": 0.1798615507136877,
      "grad_norm": 0.8260413049475901,
      "learning_rate": 0.00019812236174482917,
      "loss": 12.7932,
      "step": 3303
    },
    {
      "epoch": 0.1799160047102707,
      "grad_norm": 0.8552027146811687,
      "learning_rate": 0.00019812066055442307,
      "loss": 12.7166,
      "step": 3304
    },
    {
      "epoch": 0.17997045870685371,
      "grad_norm": 0.691431305896408,
      "learning_rate": 0.00019811895860101424,
      "loss": 12.7565,
      "step": 3305
    },
    {
      "epoch": 0.18002491270343673,
      "grad_norm": 0.7541206039195266,
      "learning_rate": 0.00019811725588461595,
      "loss": 12.7832,
      "step": 3306
    },
    {
      "epoch": 0.18007936670001973,
      "grad_norm": 0.7329115364844615,
      "learning_rate": 0.00019811555240524143,
      "loss": 12.6535,
      "step": 3307
    },
    {
      "epoch": 0.18013382069660275,
      "grad_norm": 0.7848715954841441,
      "learning_rate": 0.00019811384816290393,
      "loss": 12.7121,
      "step": 3308
    },
    {
      "epoch": 0.18018827469318577,
      "grad_norm": 0.7401571327497739,
      "learning_rate": 0.0001981121431576167,
      "loss": 12.6728,
      "step": 3309
    },
    {
      "epoch": 0.1802427286897688,
      "grad_norm": 0.7385315373003121,
      "learning_rate": 0.00019811043738939295,
      "loss": 12.6642,
      "step": 3310
    },
    {
      "epoch": 0.18029718268635178,
      "grad_norm": 0.795431374670674,
      "learning_rate": 0.00019810873085824603,
      "loss": 12.5564,
      "step": 3311
    },
    {
      "epoch": 0.1803516366829348,
      "grad_norm": 0.7905420016747329,
      "learning_rate": 0.00019810702356418914,
      "loss": 12.7727,
      "step": 3312
    },
    {
      "epoch": 0.18040609067951782,
      "grad_norm": 0.856463722382322,
      "learning_rate": 0.00019810531550723556,
      "loss": 12.7771,
      "step": 3313
    },
    {
      "epoch": 0.18046054467610081,
      "grad_norm": 0.663682842099032,
      "learning_rate": 0.00019810360668739863,
      "loss": 12.5482,
      "step": 3314
    },
    {
      "epoch": 0.18051499867268384,
      "grad_norm": 0.7304808709709537,
      "learning_rate": 0.0001981018971046916,
      "loss": 12.6683,
      "step": 3315
    },
    {
      "epoch": 0.18056945266926686,
      "grad_norm": 0.8716973827013581,
      "learning_rate": 0.00019810018675912774,
      "loss": 12.6943,
      "step": 3316
    },
    {
      "epoch": 0.18062390666584985,
      "grad_norm": 0.8052038652161817,
      "learning_rate": 0.0001980984756507204,
      "loss": 12.7423,
      "step": 3317
    },
    {
      "epoch": 0.18067836066243287,
      "grad_norm": 0.6865640501576407,
      "learning_rate": 0.00019809676377948287,
      "loss": 12.5838,
      "step": 3318
    },
    {
      "epoch": 0.1807328146590159,
      "grad_norm": 0.7787199776120453,
      "learning_rate": 0.00019809505114542843,
      "loss": 12.7606,
      "step": 3319
    },
    {
      "epoch": 0.18078726865559888,
      "grad_norm": 0.6657871497753294,
      "learning_rate": 0.00019809333774857045,
      "loss": 12.619,
      "step": 3320
    },
    {
      "epoch": 0.1808417226521819,
      "grad_norm": 0.7472678299687542,
      "learning_rate": 0.00019809162358892218,
      "loss": 12.778,
      "step": 3321
    },
    {
      "epoch": 0.18089617664876492,
      "grad_norm": 0.7521286002919139,
      "learning_rate": 0.00019808990866649703,
      "loss": 12.5975,
      "step": 3322
    },
    {
      "epoch": 0.18095063064534792,
      "grad_norm": 0.7485712192091213,
      "learning_rate": 0.0001980881929813083,
      "loss": 12.5831,
      "step": 3323
    },
    {
      "epoch": 0.18100508464193094,
      "grad_norm": 0.7729968986430557,
      "learning_rate": 0.00019808647653336934,
      "loss": 12.6564,
      "step": 3324
    },
    {
      "epoch": 0.18105953863851396,
      "grad_norm": 0.7335308303636128,
      "learning_rate": 0.00019808475932269346,
      "loss": 12.628,
      "step": 3325
    },
    {
      "epoch": 0.18111399263509695,
      "grad_norm": 0.8883852418423221,
      "learning_rate": 0.00019808304134929405,
      "loss": 12.7038,
      "step": 3326
    },
    {
      "epoch": 0.18116844663167997,
      "grad_norm": 0.8086481974347323,
      "learning_rate": 0.00019808132261318447,
      "loss": 12.5788,
      "step": 3327
    },
    {
      "epoch": 0.181222900628263,
      "grad_norm": 0.7483757045666115,
      "learning_rate": 0.00019807960311437809,
      "loss": 12.7052,
      "step": 3328
    },
    {
      "epoch": 0.181277354624846,
      "grad_norm": 0.7724274921487599,
      "learning_rate": 0.00019807788285288824,
      "loss": 12.6693,
      "step": 3329
    },
    {
      "epoch": 0.181331808621429,
      "grad_norm": 0.741722193902972,
      "learning_rate": 0.00019807616182872833,
      "loss": 12.5813,
      "step": 3330
    },
    {
      "epoch": 0.18138626261801202,
      "grad_norm": 0.7272278492201305,
      "learning_rate": 0.00019807444004191175,
      "loss": 12.5707,
      "step": 3331
    },
    {
      "epoch": 0.18144071661459504,
      "grad_norm": 0.7854612846382387,
      "learning_rate": 0.00019807271749245188,
      "loss": 12.8055,
      "step": 3332
    },
    {
      "epoch": 0.18149517061117804,
      "grad_norm": 0.823783487606722,
      "learning_rate": 0.00019807099418036212,
      "loss": 12.6623,
      "step": 3333
    },
    {
      "epoch": 0.18154962460776106,
      "grad_norm": 0.7936635697631835,
      "learning_rate": 0.00019806927010565585,
      "loss": 12.5691,
      "step": 3334
    },
    {
      "epoch": 0.18160407860434408,
      "grad_norm": 0.7383972750382136,
      "learning_rate": 0.0001980675452683465,
      "loss": 12.4969,
      "step": 3335
    },
    {
      "epoch": 0.18165853260092707,
      "grad_norm": 0.8732299978656494,
      "learning_rate": 0.00019806581966844746,
      "loss": 12.6761,
      "step": 3336
    },
    {
      "epoch": 0.1817129865975101,
      "grad_norm": 0.8064179318291677,
      "learning_rate": 0.00019806409330597218,
      "loss": 12.7272,
      "step": 3337
    },
    {
      "epoch": 0.1817674405940931,
      "grad_norm": 0.7422343085071372,
      "learning_rate": 0.00019806236618093403,
      "loss": 12.6766,
      "step": 3338
    },
    {
      "epoch": 0.1818218945906761,
      "grad_norm": 0.8175942277868645,
      "learning_rate": 0.0001980606382933465,
      "loss": 12.6871,
      "step": 3339
    },
    {
      "epoch": 0.18187634858725912,
      "grad_norm": 0.7404435885300982,
      "learning_rate": 0.000198058909643223,
      "loss": 12.5923,
      "step": 3340
    },
    {
      "epoch": 0.18193080258384214,
      "grad_norm": 0.8017199061331057,
      "learning_rate": 0.00019805718023057695,
      "loss": 12.6781,
      "step": 3341
    },
    {
      "epoch": 0.18198525658042514,
      "grad_norm": 0.6885014995804388,
      "learning_rate": 0.00019805545005542184,
      "loss": 12.7684,
      "step": 3342
    },
    {
      "epoch": 0.18203971057700816,
      "grad_norm": 0.7907680166328611,
      "learning_rate": 0.00019805371911777112,
      "loss": 12.8546,
      "step": 3343
    },
    {
      "epoch": 0.18209416457359118,
      "grad_norm": 0.7060355670276862,
      "learning_rate": 0.00019805198741763822,
      "loss": 12.4949,
      "step": 3344
    },
    {
      "epoch": 0.1821486185701742,
      "grad_norm": 0.9516032000572783,
      "learning_rate": 0.00019805025495503662,
      "loss": 12.7103,
      "step": 3345
    },
    {
      "epoch": 0.1822030725667572,
      "grad_norm": 0.7449253439223767,
      "learning_rate": 0.00019804852172997981,
      "loss": 12.5891,
      "step": 3346
    },
    {
      "epoch": 0.1822575265633402,
      "grad_norm": 0.729946216620088,
      "learning_rate": 0.00019804678774248125,
      "loss": 12.8393,
      "step": 3347
    },
    {
      "epoch": 0.18231198055992323,
      "grad_norm": 0.9178332662915702,
      "learning_rate": 0.00019804505299255443,
      "loss": 12.6575,
      "step": 3348
    },
    {
      "epoch": 0.18236643455650622,
      "grad_norm": 0.7749283115288399,
      "learning_rate": 0.00019804331748021282,
      "loss": 12.6209,
      "step": 3349
    },
    {
      "epoch": 0.18242088855308924,
      "grad_norm": 0.9139733963471919,
      "learning_rate": 0.00019804158120546994,
      "loss": 12.7139,
      "step": 3350
    },
    {
      "epoch": 0.18247534254967226,
      "grad_norm": 0.7763906942557518,
      "learning_rate": 0.00019803984416833927,
      "loss": 12.5467,
      "step": 3351
    },
    {
      "epoch": 0.18252979654625526,
      "grad_norm": 0.7252198432139424,
      "learning_rate": 0.00019803810636883436,
      "loss": 12.6515,
      "step": 3352
    },
    {
      "epoch": 0.18258425054283828,
      "grad_norm": 0.9978346602920308,
      "learning_rate": 0.00019803636780696865,
      "loss": 12.6464,
      "step": 3353
    },
    {
      "epoch": 0.1826387045394213,
      "grad_norm": 0.7093383174056861,
      "learning_rate": 0.00019803462848275573,
      "loss": 12.7392,
      "step": 3354
    },
    {
      "epoch": 0.1826931585360043,
      "grad_norm": 0.9341940050524234,
      "learning_rate": 0.00019803288839620911,
      "loss": 12.7433,
      "step": 3355
    },
    {
      "epoch": 0.1827476125325873,
      "grad_norm": 0.7787445813299527,
      "learning_rate": 0.00019803114754734228,
      "loss": 12.6275,
      "step": 3356
    },
    {
      "epoch": 0.18280206652917033,
      "grad_norm": 0.7238973947276285,
      "learning_rate": 0.00019802940593616883,
      "loss": 12.5868,
      "step": 3357
    },
    {
      "epoch": 0.18285652052575332,
      "grad_norm": 0.8909256720217067,
      "learning_rate": 0.00019802766356270227,
      "loss": 12.6952,
      "step": 3358
    },
    {
      "epoch": 0.18291097452233634,
      "grad_norm": 0.9060266000302456,
      "learning_rate": 0.00019802592042695614,
      "loss": 12.6288,
      "step": 3359
    },
    {
      "epoch": 0.18296542851891937,
      "grad_norm": 0.8317096102878461,
      "learning_rate": 0.00019802417652894406,
      "loss": 12.8364,
      "step": 3360
    },
    {
      "epoch": 0.18301988251550239,
      "grad_norm": 0.7046624142474951,
      "learning_rate": 0.00019802243186867953,
      "loss": 12.6454,
      "step": 3361
    },
    {
      "epoch": 0.18307433651208538,
      "grad_norm": 0.7724845162821498,
      "learning_rate": 0.0001980206864461761,
      "loss": 12.5423,
      "step": 3362
    },
    {
      "epoch": 0.1831287905086684,
      "grad_norm": 0.7007103141816504,
      "learning_rate": 0.0001980189402614474,
      "loss": 12.5998,
      "step": 3363
    },
    {
      "epoch": 0.18318324450525142,
      "grad_norm": 0.7136727791052012,
      "learning_rate": 0.000198017193314507,
      "loss": 12.5667,
      "step": 3364
    },
    {
      "epoch": 0.1832376985018344,
      "grad_norm": 0.7296237009613211,
      "learning_rate": 0.00019801544560536845,
      "loss": 12.6287,
      "step": 3365
    },
    {
      "epoch": 0.18329215249841743,
      "grad_norm": 0.8388171730170142,
      "learning_rate": 0.00019801369713404535,
      "loss": 12.6726,
      "step": 3366
    },
    {
      "epoch": 0.18334660649500045,
      "grad_norm": 0.7969860000049219,
      "learning_rate": 0.0001980119479005513,
      "loss": 12.6525,
      "step": 3367
    },
    {
      "epoch": 0.18340106049158345,
      "grad_norm": 0.8607464188262216,
      "learning_rate": 0.00019801019790489992,
      "loss": 12.8035,
      "step": 3368
    },
    {
      "epoch": 0.18345551448816647,
      "grad_norm": 0.8417130114551891,
      "learning_rate": 0.00019800844714710478,
      "loss": 12.6438,
      "step": 3369
    },
    {
      "epoch": 0.18350996848474949,
      "grad_norm": 0.7454307447608005,
      "learning_rate": 0.00019800669562717956,
      "loss": 12.6587,
      "step": 3370
    },
    {
      "epoch": 0.18356442248133248,
      "grad_norm": 0.7564182643950726,
      "learning_rate": 0.00019800494334513781,
      "loss": 12.7091,
      "step": 3371
    },
    {
      "epoch": 0.1836188764779155,
      "grad_norm": 0.751046314032216,
      "learning_rate": 0.00019800319030099318,
      "loss": 12.5357,
      "step": 3372
    },
    {
      "epoch": 0.18367333047449852,
      "grad_norm": 0.6986462694900774,
      "learning_rate": 0.00019800143649475934,
      "loss": 12.6685,
      "step": 3373
    },
    {
      "epoch": 0.1837277844710815,
      "grad_norm": 0.7858772949677835,
      "learning_rate": 0.00019799968192644985,
      "loss": 12.7326,
      "step": 3374
    },
    {
      "epoch": 0.18378223846766453,
      "grad_norm": 0.6825472324506604,
      "learning_rate": 0.00019799792659607842,
      "loss": 12.6151,
      "step": 3375
    },
    {
      "epoch": 0.18383669246424755,
      "grad_norm": 0.7411115867200715,
      "learning_rate": 0.0001979961705036587,
      "loss": 12.5658,
      "step": 3376
    },
    {
      "epoch": 0.18389114646083055,
      "grad_norm": 0.8472704769040416,
      "learning_rate": 0.0001979944136492043,
      "loss": 12.7508,
      "step": 3377
    },
    {
      "epoch": 0.18394560045741357,
      "grad_norm": 0.756717561964962,
      "learning_rate": 0.0001979926560327289,
      "loss": 12.6634,
      "step": 3378
    },
    {
      "epoch": 0.1840000544539966,
      "grad_norm": 0.7606214364101355,
      "learning_rate": 0.00019799089765424618,
      "loss": 12.5667,
      "step": 3379
    },
    {
      "epoch": 0.1840545084505796,
      "grad_norm": 0.7270193995130935,
      "learning_rate": 0.00019798913851376982,
      "loss": 12.6351,
      "step": 3380
    },
    {
      "epoch": 0.1841089624471626,
      "grad_norm": 0.6805493559802995,
      "learning_rate": 0.00019798737861131347,
      "loss": 12.6746,
      "step": 3381
    },
    {
      "epoch": 0.18416341644374562,
      "grad_norm": 0.7056558366013929,
      "learning_rate": 0.00019798561794689086,
      "loss": 12.6943,
      "step": 3382
    },
    {
      "epoch": 0.18421787044032864,
      "grad_norm": 0.7657220496359567,
      "learning_rate": 0.0001979838565205156,
      "loss": 12.6128,
      "step": 3383
    },
    {
      "epoch": 0.18427232443691163,
      "grad_norm": 0.9039108815858633,
      "learning_rate": 0.0001979820943322015,
      "loss": 12.6562,
      "step": 3384
    },
    {
      "epoch": 0.18432677843349465,
      "grad_norm": 0.86946493008817,
      "learning_rate": 0.00019798033138196218,
      "loss": 12.5324,
      "step": 3385
    },
    {
      "epoch": 0.18438123243007767,
      "grad_norm": 0.7912167800969766,
      "learning_rate": 0.00019797856766981135,
      "loss": 12.6925,
      "step": 3386
    },
    {
      "epoch": 0.18443568642666067,
      "grad_norm": 0.7999489254168872,
      "learning_rate": 0.00019797680319576276,
      "loss": 12.6894,
      "step": 3387
    },
    {
      "epoch": 0.1844901404232437,
      "grad_norm": 0.7324214346710298,
      "learning_rate": 0.00019797503795983016,
      "loss": 12.6803,
      "step": 3388
    },
    {
      "epoch": 0.1845445944198267,
      "grad_norm": 0.756609872779989,
      "learning_rate": 0.00019797327196202717,
      "loss": 12.6434,
      "step": 3389
    },
    {
      "epoch": 0.1845990484164097,
      "grad_norm": 0.7015324007008875,
      "learning_rate": 0.00019797150520236762,
      "loss": 12.6552,
      "step": 3390
    },
    {
      "epoch": 0.18465350241299272,
      "grad_norm": 0.7122255001024863,
      "learning_rate": 0.0001979697376808652,
      "loss": 12.5259,
      "step": 3391
    },
    {
      "epoch": 0.18470795640957574,
      "grad_norm": 0.9522221558042216,
      "learning_rate": 0.0001979679693975337,
      "loss": 12.673,
      "step": 3392
    },
    {
      "epoch": 0.18476241040615873,
      "grad_norm": 0.7110517403519583,
      "learning_rate": 0.00019796620035238678,
      "loss": 12.6476,
      "step": 3393
    },
    {
      "epoch": 0.18481686440274175,
      "grad_norm": 0.7595909509678056,
      "learning_rate": 0.00019796443054543832,
      "loss": 12.6126,
      "step": 3394
    },
    {
      "epoch": 0.18487131839932477,
      "grad_norm": 0.687617268889062,
      "learning_rate": 0.000197962659976702,
      "loss": 12.5402,
      "step": 3395
    },
    {
      "epoch": 0.1849257723959078,
      "grad_norm": 0.7039148224733114,
      "learning_rate": 0.00019796088864619158,
      "loss": 12.7266,
      "step": 3396
    },
    {
      "epoch": 0.1849802263924908,
      "grad_norm": 0.7788472692865894,
      "learning_rate": 0.00019795911655392089,
      "loss": 12.5487,
      "step": 3397
    },
    {
      "epoch": 0.1850346803890738,
      "grad_norm": 0.7193934435877664,
      "learning_rate": 0.00019795734369990368,
      "loss": 12.6761,
      "step": 3398
    },
    {
      "epoch": 0.18508913438565683,
      "grad_norm": 0.7146683469628885,
      "learning_rate": 0.00019795557008415372,
      "loss": 12.4951,
      "step": 3399
    },
    {
      "epoch": 0.18514358838223982,
      "grad_norm": 0.781946581429394,
      "learning_rate": 0.00019795379570668487,
      "loss": 12.5245,
      "step": 3400
    },
    {
      "epoch": 0.18519804237882284,
      "grad_norm": 0.6925827278733167,
      "learning_rate": 0.00019795202056751082,
      "loss": 12.6528,
      "step": 3401
    },
    {
      "epoch": 0.18525249637540586,
      "grad_norm": 0.7960254736251621,
      "learning_rate": 0.00019795024466664546,
      "loss": 12.8152,
      "step": 3402
    },
    {
      "epoch": 0.18530695037198885,
      "grad_norm": 0.7310673158847482,
      "learning_rate": 0.00019794846800410255,
      "loss": 12.6397,
      "step": 3403
    },
    {
      "epoch": 0.18536140436857187,
      "grad_norm": 0.73383372208105,
      "learning_rate": 0.00019794669057989595,
      "loss": 12.5688,
      "step": 3404
    },
    {
      "epoch": 0.1854158583651549,
      "grad_norm": 0.836279865981678,
      "learning_rate": 0.00019794491239403944,
      "loss": 12.5942,
      "step": 3405
    },
    {
      "epoch": 0.1854703123617379,
      "grad_norm": 0.7712799032342063,
      "learning_rate": 0.0001979431334465469,
      "loss": 12.6425,
      "step": 3406
    },
    {
      "epoch": 0.1855247663583209,
      "grad_norm": 0.6909617201882076,
      "learning_rate": 0.0001979413537374321,
      "loss": 12.588,
      "step": 3407
    },
    {
      "epoch": 0.18557922035490393,
      "grad_norm": 0.7048457518660487,
      "learning_rate": 0.0001979395732667089,
      "loss": 12.6108,
      "step": 3408
    },
    {
      "epoch": 0.18563367435148692,
      "grad_norm": 0.7466368679226278,
      "learning_rate": 0.00019793779203439117,
      "loss": 12.6437,
      "step": 3409
    },
    {
      "epoch": 0.18568812834806994,
      "grad_norm": 0.7581910439112242,
      "learning_rate": 0.00019793601004049273,
      "loss": 12.6338,
      "step": 3410
    },
    {
      "epoch": 0.18574258234465296,
      "grad_norm": 0.6889479667097653,
      "learning_rate": 0.00019793422728502747,
      "loss": 12.6501,
      "step": 3411
    },
    {
      "epoch": 0.18579703634123598,
      "grad_norm": 0.7053095188967593,
      "learning_rate": 0.00019793244376800924,
      "loss": 12.6259,
      "step": 3412
    },
    {
      "epoch": 0.18585149033781898,
      "grad_norm": 0.9028178062423818,
      "learning_rate": 0.0001979306594894519,
      "loss": 12.6088,
      "step": 3413
    },
    {
      "epoch": 0.185905944334402,
      "grad_norm": 0.7099854745460383,
      "learning_rate": 0.00019792887444936933,
      "loss": 12.6394,
      "step": 3414
    },
    {
      "epoch": 0.18596039833098502,
      "grad_norm": 0.6928213414862964,
      "learning_rate": 0.0001979270886477754,
      "loss": 12.4603,
      "step": 3415
    },
    {
      "epoch": 0.186014852327568,
      "grad_norm": 0.6587384490980607,
      "learning_rate": 0.000197925302084684,
      "loss": 12.6143,
      "step": 3416
    },
    {
      "epoch": 0.18606930632415103,
      "grad_norm": 0.7064330914174174,
      "learning_rate": 0.00019792351476010905,
      "loss": 12.5708,
      "step": 3417
    },
    {
      "epoch": 0.18612376032073405,
      "grad_norm": 0.7396116906610933,
      "learning_rate": 0.00019792172667406442,
      "loss": 12.6673,
      "step": 3418
    },
    {
      "epoch": 0.18617821431731704,
      "grad_norm": 0.7228600413696578,
      "learning_rate": 0.000197919937826564,
      "loss": 12.5551,
      "step": 3419
    },
    {
      "epoch": 0.18623266831390006,
      "grad_norm": 0.8579167739390514,
      "learning_rate": 0.00019791814821762174,
      "loss": 12.6024,
      "step": 3420
    },
    {
      "epoch": 0.18628712231048308,
      "grad_norm": 0.8120542124616528,
      "learning_rate": 0.00019791635784725155,
      "loss": 12.8604,
      "step": 3421
    },
    {
      "epoch": 0.18634157630706608,
      "grad_norm": 0.7673698957399567,
      "learning_rate": 0.00019791456671546732,
      "loss": 12.6082,
      "step": 3422
    },
    {
      "epoch": 0.1863960303036491,
      "grad_norm": 0.785400566174054,
      "learning_rate": 0.000197912774822283,
      "loss": 12.6288,
      "step": 3423
    },
    {
      "epoch": 0.18645048430023212,
      "grad_norm": 0.7907865406949788,
      "learning_rate": 0.00019791098216771254,
      "loss": 12.6943,
      "step": 3424
    },
    {
      "epoch": 0.1865049382968151,
      "grad_norm": 0.8620753392475704,
      "learning_rate": 0.00019790918875176985,
      "loss": 12.8335,
      "step": 3425
    },
    {
      "epoch": 0.18655939229339813,
      "grad_norm": 0.7712543156805853,
      "learning_rate": 0.0001979073945744689,
      "loss": 12.6657,
      "step": 3426
    },
    {
      "epoch": 0.18661384628998115,
      "grad_norm": 0.8981380922682508,
      "learning_rate": 0.00019790559963582362,
      "loss": 12.7522,
      "step": 3427
    },
    {
      "epoch": 0.18666830028656417,
      "grad_norm": 0.7264599236766378,
      "learning_rate": 0.00019790380393584797,
      "loss": 12.6306,
      "step": 3428
    },
    {
      "epoch": 0.18672275428314716,
      "grad_norm": 0.7341988030503047,
      "learning_rate": 0.00019790200747455593,
      "loss": 12.6287,
      "step": 3429
    },
    {
      "epoch": 0.18677720827973018,
      "grad_norm": 0.6866438472063808,
      "learning_rate": 0.00019790021025196147,
      "loss": 12.6171,
      "step": 3430
    },
    {
      "epoch": 0.1868316622763132,
      "grad_norm": 0.8609424467053118,
      "learning_rate": 0.00019789841226807857,
      "loss": 12.6022,
      "step": 3431
    },
    {
      "epoch": 0.1868861162728962,
      "grad_norm": 0.735664009505058,
      "learning_rate": 0.00019789661352292116,
      "loss": 12.7115,
      "step": 3432
    },
    {
      "epoch": 0.18694057026947922,
      "grad_norm": 0.8061194136233919,
      "learning_rate": 0.0001978948140165033,
      "loss": 12.6571,
      "step": 3433
    },
    {
      "epoch": 0.18699502426606224,
      "grad_norm": 0.7001891909756394,
      "learning_rate": 0.00019789301374883894,
      "loss": 12.6699,
      "step": 3434
    },
    {
      "epoch": 0.18704947826264523,
      "grad_norm": 0.6471394560553497,
      "learning_rate": 0.00019789121271994206,
      "loss": 12.6167,
      "step": 3435
    },
    {
      "epoch": 0.18710393225922825,
      "grad_norm": 0.8114243930971807,
      "learning_rate": 0.00019788941092982674,
      "loss": 12.5519,
      "step": 3436
    },
    {
      "epoch": 0.18715838625581127,
      "grad_norm": 0.7556975256688439,
      "learning_rate": 0.00019788760837850694,
      "loss": 12.6453,
      "step": 3437
    },
    {
      "epoch": 0.18721284025239426,
      "grad_norm": 0.8111179653669358,
      "learning_rate": 0.00019788580506599664,
      "loss": 12.7268,
      "step": 3438
    },
    {
      "epoch": 0.18726729424897728,
      "grad_norm": 0.7792236041663045,
      "learning_rate": 0.00019788400099230992,
      "loss": 12.471,
      "step": 3439
    },
    {
      "epoch": 0.1873217482455603,
      "grad_norm": 0.7636202013337339,
      "learning_rate": 0.00019788219615746083,
      "loss": 12.7225,
      "step": 3440
    },
    {
      "epoch": 0.1873762022421433,
      "grad_norm": 0.7703390768900522,
      "learning_rate": 0.00019788039056146332,
      "loss": 12.77,
      "step": 3441
    },
    {
      "epoch": 0.18743065623872632,
      "grad_norm": 0.6989497989715391,
      "learning_rate": 0.0001978785842043315,
      "loss": 12.572,
      "step": 3442
    },
    {
      "epoch": 0.18748511023530934,
      "grad_norm": 0.7531763129520263,
      "learning_rate": 0.0001978767770860794,
      "loss": 12.6281,
      "step": 3443
    },
    {
      "epoch": 0.18753956423189233,
      "grad_norm": 0.7656064978206467,
      "learning_rate": 0.00019787496920672107,
      "loss": 12.683,
      "step": 3444
    },
    {
      "epoch": 0.18759401822847535,
      "grad_norm": 0.6959064866014217,
      "learning_rate": 0.00019787316056627053,
      "loss": 12.6926,
      "step": 3445
    },
    {
      "epoch": 0.18764847222505837,
      "grad_norm": 0.7310517195944778,
      "learning_rate": 0.00019787135116474191,
      "loss": 12.6944,
      "step": 3446
    },
    {
      "epoch": 0.1877029262216414,
      "grad_norm": 0.7050888293219245,
      "learning_rate": 0.00019786954100214926,
      "loss": 12.6899,
      "step": 3447
    },
    {
      "epoch": 0.18775738021822438,
      "grad_norm": 0.7331840748186392,
      "learning_rate": 0.00019786773007850664,
      "loss": 12.7043,
      "step": 3448
    },
    {
      "epoch": 0.1878118342148074,
      "grad_norm": 0.8760959354346283,
      "learning_rate": 0.0001978659183938281,
      "loss": 12.5749,
      "step": 3449
    },
    {
      "epoch": 0.18786628821139043,
      "grad_norm": 0.6531939284147806,
      "learning_rate": 0.00019786410594812784,
      "loss": 12.533,
      "step": 3450
    },
    {
      "epoch": 0.18792074220797342,
      "grad_norm": 0.7905262008052769,
      "learning_rate": 0.00019786229274141982,
      "loss": 12.6418,
      "step": 3451
    },
    {
      "epoch": 0.18797519620455644,
      "grad_norm": 0.7837407214029306,
      "learning_rate": 0.00019786047877371821,
      "loss": 12.4968,
      "step": 3452
    },
    {
      "epoch": 0.18802965020113946,
      "grad_norm": 0.6829892747844226,
      "learning_rate": 0.0001978586640450371,
      "loss": 12.5865,
      "step": 3453
    },
    {
      "epoch": 0.18808410419772245,
      "grad_norm": 0.7905919519547869,
      "learning_rate": 0.00019785684855539066,
      "loss": 12.7048,
      "step": 3454
    },
    {
      "epoch": 0.18813855819430547,
      "grad_norm": 0.7202467701468647,
      "learning_rate": 0.0001978550323047929,
      "loss": 12.6159,
      "step": 3455
    },
    {
      "epoch": 0.1881930121908885,
      "grad_norm": 0.7070151712650198,
      "learning_rate": 0.00019785321529325803,
      "loss": 12.654,
      "step": 3456
    },
    {
      "epoch": 0.18824746618747148,
      "grad_norm": 0.7513944704198623,
      "learning_rate": 0.0001978513975208001,
      "loss": 12.6776,
      "step": 3457
    },
    {
      "epoch": 0.1883019201840545,
      "grad_norm": 0.7102718478964493,
      "learning_rate": 0.00019784957898743335,
      "loss": 12.4791,
      "step": 3458
    },
    {
      "epoch": 0.18835637418063753,
      "grad_norm": 0.6849626063754578,
      "learning_rate": 0.00019784775969317183,
      "loss": 12.6471,
      "step": 3459
    },
    {
      "epoch": 0.18841082817722052,
      "grad_norm": 0.7217393791466998,
      "learning_rate": 0.00019784593963802975,
      "loss": 12.6078,
      "step": 3460
    },
    {
      "epoch": 0.18846528217380354,
      "grad_norm": 0.9547534304507546,
      "learning_rate": 0.0001978441188220212,
      "loss": 12.6762,
      "step": 3461
    },
    {
      "epoch": 0.18851973617038656,
      "grad_norm": 0.7400562576011662,
      "learning_rate": 0.0001978422972451604,
      "loss": 12.5686,
      "step": 3462
    },
    {
      "epoch": 0.18857419016696958,
      "grad_norm": 0.7819096937512829,
      "learning_rate": 0.00019784047490746146,
      "loss": 12.7465,
      "step": 3463
    },
    {
      "epoch": 0.18862864416355257,
      "grad_norm": 0.6978935409395687,
      "learning_rate": 0.00019783865180893862,
      "loss": 12.6414,
      "step": 3464
    },
    {
      "epoch": 0.1886830981601356,
      "grad_norm": 0.745264747694646,
      "learning_rate": 0.000197836827949606,
      "loss": 12.6252,
      "step": 3465
    },
    {
      "epoch": 0.1887375521567186,
      "grad_norm": 0.7120841832106395,
      "learning_rate": 0.0001978350033294778,
      "loss": 12.5873,
      "step": 3466
    },
    {
      "epoch": 0.1887920061533016,
      "grad_norm": 0.7915147314262276,
      "learning_rate": 0.00019783317794856817,
      "loss": 12.4969,
      "step": 3467
    },
    {
      "epoch": 0.18884646014988463,
      "grad_norm": 0.7766806085597008,
      "learning_rate": 0.00019783135180689138,
      "loss": 12.5835,
      "step": 3468
    },
    {
      "epoch": 0.18890091414646765,
      "grad_norm": 0.8489177098808214,
      "learning_rate": 0.0001978295249044616,
      "loss": 12.7021,
      "step": 3469
    },
    {
      "epoch": 0.18895536814305064,
      "grad_norm": 0.7755385665913826,
      "learning_rate": 0.000197827697241293,
      "loss": 12.656,
      "step": 3470
    },
    {
      "epoch": 0.18900982213963366,
      "grad_norm": 0.7276129651295314,
      "learning_rate": 0.00019782586881739983,
      "loss": 12.6165,
      "step": 3471
    },
    {
      "epoch": 0.18906427613621668,
      "grad_norm": 0.7669102297638349,
      "learning_rate": 0.0001978240396327963,
      "loss": 12.6595,
      "step": 3472
    },
    {
      "epoch": 0.18911873013279967,
      "grad_norm": 0.7420823944772779,
      "learning_rate": 0.00019782220968749665,
      "loss": 12.5876,
      "step": 3473
    },
    {
      "epoch": 0.1891731841293827,
      "grad_norm": 0.7468557020621868,
      "learning_rate": 0.0001978203789815151,
      "loss": 12.6916,
      "step": 3474
    },
    {
      "epoch": 0.1892276381259657,
      "grad_norm": 0.7106718007187686,
      "learning_rate": 0.00019781854751486582,
      "loss": 12.5696,
      "step": 3475
    },
    {
      "epoch": 0.1892820921225487,
      "grad_norm": 0.7332562200016313,
      "learning_rate": 0.00019781671528756314,
      "loss": 12.7741,
      "step": 3476
    },
    {
      "epoch": 0.18933654611913173,
      "grad_norm": 0.7833748917159779,
      "learning_rate": 0.00019781488229962132,
      "loss": 12.689,
      "step": 3477
    },
    {
      "epoch": 0.18939100011571475,
      "grad_norm": 0.7741583004705924,
      "learning_rate": 0.0001978130485510545,
      "loss": 12.7469,
      "step": 3478
    },
    {
      "epoch": 0.18944545411229777,
      "grad_norm": 0.7177523289604667,
      "learning_rate": 0.00019781121404187707,
      "loss": 12.4962,
      "step": 3479
    },
    {
      "epoch": 0.18949990810888076,
      "grad_norm": 0.809544019401128,
      "learning_rate": 0.0001978093787721032,
      "loss": 12.7359,
      "step": 3480
    },
    {
      "epoch": 0.18955436210546378,
      "grad_norm": 0.8553144004000792,
      "learning_rate": 0.00019780754274174723,
      "loss": 12.818,
      "step": 3481
    },
    {
      "epoch": 0.1896088161020468,
      "grad_norm": 0.7081591841393877,
      "learning_rate": 0.00019780570595082336,
      "loss": 12.5759,
      "step": 3482
    },
    {
      "epoch": 0.1896632700986298,
      "grad_norm": 0.8695116509656418,
      "learning_rate": 0.00019780386839934595,
      "loss": 12.6974,
      "step": 3483
    },
    {
      "epoch": 0.18971772409521281,
      "grad_norm": 0.8223589842834956,
      "learning_rate": 0.00019780203008732924,
      "loss": 12.6353,
      "step": 3484
    },
    {
      "epoch": 0.18977217809179583,
      "grad_norm": 0.7446285543581759,
      "learning_rate": 0.00019780019101478758,
      "loss": 12.6826,
      "step": 3485
    },
    {
      "epoch": 0.18982663208837883,
      "grad_norm": 0.6896485167003834,
      "learning_rate": 0.00019779835118173523,
      "loss": 12.4591,
      "step": 3486
    },
    {
      "epoch": 0.18988108608496185,
      "grad_norm": 0.7866481478000416,
      "learning_rate": 0.00019779651058818645,
      "loss": 12.7327,
      "step": 3487
    },
    {
      "epoch": 0.18993554008154487,
      "grad_norm": 0.7697027786000047,
      "learning_rate": 0.00019779466923415564,
      "loss": 12.5679,
      "step": 3488
    },
    {
      "epoch": 0.18998999407812786,
      "grad_norm": 0.779594290786733,
      "learning_rate": 0.00019779282711965705,
      "loss": 12.643,
      "step": 3489
    },
    {
      "epoch": 0.19004444807471088,
      "grad_norm": 0.8184831223384603,
      "learning_rate": 0.00019779098424470507,
      "loss": 12.6132,
      "step": 3490
    },
    {
      "epoch": 0.1900989020712939,
      "grad_norm": 0.7787519451121007,
      "learning_rate": 0.000197789140609314,
      "loss": 12.7536,
      "step": 3491
    },
    {
      "epoch": 0.1901533560678769,
      "grad_norm": 0.7319492566811413,
      "learning_rate": 0.00019778729621349817,
      "loss": 12.6267,
      "step": 3492
    },
    {
      "epoch": 0.19020781006445991,
      "grad_norm": 0.9438662989100811,
      "learning_rate": 0.0001977854510572719,
      "loss": 12.7885,
      "step": 3493
    },
    {
      "epoch": 0.19026226406104293,
      "grad_norm": 0.7472861900488181,
      "learning_rate": 0.0001977836051406496,
      "loss": 12.5697,
      "step": 3494
    },
    {
      "epoch": 0.19031671805762596,
      "grad_norm": 0.7530749349726638,
      "learning_rate": 0.00019778175846364558,
      "loss": 12.7655,
      "step": 3495
    },
    {
      "epoch": 0.19037117205420895,
      "grad_norm": 0.717289086680717,
      "learning_rate": 0.00019777991102627417,
      "loss": 12.7446,
      "step": 3496
    },
    {
      "epoch": 0.19042562605079197,
      "grad_norm": 0.7927481726756973,
      "learning_rate": 0.0001977780628285498,
      "loss": 12.6446,
      "step": 3497
    },
    {
      "epoch": 0.190480080047375,
      "grad_norm": 0.7975134354075496,
      "learning_rate": 0.00019777621387048684,
      "loss": 12.6087,
      "step": 3498
    },
    {
      "epoch": 0.19053453404395798,
      "grad_norm": 0.7714088806598007,
      "learning_rate": 0.0001977743641520996,
      "loss": 12.6812,
      "step": 3499
    },
    {
      "epoch": 0.190588988040541,
      "grad_norm": 0.7761793014499357,
      "learning_rate": 0.00019777251367340254,
      "loss": 12.4691,
      "step": 3500
    },
    {
      "epoch": 0.19064344203712402,
      "grad_norm": 0.7577645649898556,
      "learning_rate": 0.00019777066243441,
      "loss": 12.7787,
      "step": 3501
    },
    {
      "epoch": 0.19069789603370702,
      "grad_norm": 0.8089804177245533,
      "learning_rate": 0.0001977688104351364,
      "loss": 12.6915,
      "step": 3502
    },
    {
      "epoch": 0.19075235003029004,
      "grad_norm": 0.7639733935034829,
      "learning_rate": 0.00019776695767559615,
      "loss": 12.6868,
      "step": 3503
    },
    {
      "epoch": 0.19080680402687306,
      "grad_norm": 0.8485259089297601,
      "learning_rate": 0.0001977651041558036,
      "loss": 12.6876,
      "step": 3504
    },
    {
      "epoch": 0.19086125802345605,
      "grad_norm": 0.7641466460154173,
      "learning_rate": 0.00019776324987577323,
      "loss": 12.7128,
      "step": 3505
    },
    {
      "epoch": 0.19091571202003907,
      "grad_norm": 1.031112607794016,
      "learning_rate": 0.00019776139483551944,
      "loss": 12.7414,
      "step": 3506
    },
    {
      "epoch": 0.1909701660166221,
      "grad_norm": 0.7165868709934979,
      "learning_rate": 0.00019775953903505665,
      "loss": 12.5893,
      "step": 3507
    },
    {
      "epoch": 0.19102462001320508,
      "grad_norm": 0.6812113548886141,
      "learning_rate": 0.00019775768247439927,
      "loss": 12.5462,
      "step": 3508
    },
    {
      "epoch": 0.1910790740097881,
      "grad_norm": 0.8234924571777671,
      "learning_rate": 0.0001977558251535618,
      "loss": 12.5584,
      "step": 3509
    },
    {
      "epoch": 0.19113352800637112,
      "grad_norm": 0.7551867552780744,
      "learning_rate": 0.0001977539670725586,
      "loss": 12.651,
      "step": 3510
    },
    {
      "epoch": 0.19118798200295412,
      "grad_norm": 0.8335507001913836,
      "learning_rate": 0.00019775210823140416,
      "loss": 12.7434,
      "step": 3511
    },
    {
      "epoch": 0.19124243599953714,
      "grad_norm": 0.9366970779404721,
      "learning_rate": 0.00019775024863011293,
      "loss": 12.7378,
      "step": 3512
    },
    {
      "epoch": 0.19129688999612016,
      "grad_norm": 0.788139924247736,
      "learning_rate": 0.0001977483882686994,
      "loss": 12.6645,
      "step": 3513
    },
    {
      "epoch": 0.19135134399270318,
      "grad_norm": 1.008796680829356,
      "learning_rate": 0.000197746527147178,
      "loss": 12.79,
      "step": 3514
    },
    {
      "epoch": 0.19140579798928617,
      "grad_norm": 0.7957919244741967,
      "learning_rate": 0.0001977446652655632,
      "loss": 12.727,
      "step": 3515
    },
    {
      "epoch": 0.1914602519858692,
      "grad_norm": 0.7216544125020525,
      "learning_rate": 0.0001977428026238695,
      "loss": 12.7369,
      "step": 3516
    },
    {
      "epoch": 0.1915147059824522,
      "grad_norm": 0.7966202844772511,
      "learning_rate": 0.00019774093922211137,
      "loss": 12.6249,
      "step": 3517
    },
    {
      "epoch": 0.1915691599790352,
      "grad_norm": 0.761783076179875,
      "learning_rate": 0.00019773907506030332,
      "loss": 12.6354,
      "step": 3518
    },
    {
      "epoch": 0.19162361397561822,
      "grad_norm": 0.7296691536804387,
      "learning_rate": 0.0001977372101384598,
      "loss": 12.6756,
      "step": 3519
    },
    {
      "epoch": 0.19167806797220124,
      "grad_norm": 0.7538202468078059,
      "learning_rate": 0.00019773534445659537,
      "loss": 12.7087,
      "step": 3520
    },
    {
      "epoch": 0.19173252196878424,
      "grad_norm": 0.8182761392904533,
      "learning_rate": 0.00019773347801472452,
      "loss": 12.6986,
      "step": 3521
    },
    {
      "epoch": 0.19178697596536726,
      "grad_norm": 0.6941479890687625,
      "learning_rate": 0.00019773161081286172,
      "loss": 12.7346,
      "step": 3522
    },
    {
      "epoch": 0.19184142996195028,
      "grad_norm": 0.7328857273974302,
      "learning_rate": 0.00019772974285102156,
      "loss": 12.707,
      "step": 3523
    },
    {
      "epoch": 0.19189588395853327,
      "grad_norm": 0.8779759648629172,
      "learning_rate": 0.00019772787412921853,
      "loss": 12.6799,
      "step": 3524
    },
    {
      "epoch": 0.1919503379551163,
      "grad_norm": 0.8102918945732309,
      "learning_rate": 0.00019772600464746715,
      "loss": 12.7078,
      "step": 3525
    },
    {
      "epoch": 0.1920047919516993,
      "grad_norm": 0.7432385246336675,
      "learning_rate": 0.00019772413440578197,
      "loss": 12.576,
      "step": 3526
    },
    {
      "epoch": 0.1920592459482823,
      "grad_norm": 0.7526038693013306,
      "learning_rate": 0.00019772226340417754,
      "loss": 12.6807,
      "step": 3527
    },
    {
      "epoch": 0.19211369994486532,
      "grad_norm": 0.8225261794530042,
      "learning_rate": 0.00019772039164266838,
      "loss": 12.5654,
      "step": 3528
    },
    {
      "epoch": 0.19216815394144834,
      "grad_norm": 0.725848010653964,
      "learning_rate": 0.00019771851912126908,
      "loss": 12.6766,
      "step": 3529
    },
    {
      "epoch": 0.19222260793803136,
      "grad_norm": 0.7908395411886413,
      "learning_rate": 0.00019771664583999418,
      "loss": 12.5595,
      "step": 3530
    },
    {
      "epoch": 0.19227706193461436,
      "grad_norm": 0.745581408980533,
      "learning_rate": 0.00019771477179885826,
      "loss": 12.7509,
      "step": 3531
    },
    {
      "epoch": 0.19233151593119738,
      "grad_norm": 0.8018799413676838,
      "learning_rate": 0.00019771289699787589,
      "loss": 12.695,
      "step": 3532
    },
    {
      "epoch": 0.1923859699277804,
      "grad_norm": 0.810353107267892,
      "learning_rate": 0.00019771102143706167,
      "loss": 12.6875,
      "step": 3533
    },
    {
      "epoch": 0.1924404239243634,
      "grad_norm": 0.7348926531608867,
      "learning_rate": 0.00019770914511643012,
      "loss": 12.6932,
      "step": 3534
    },
    {
      "epoch": 0.1924948779209464,
      "grad_norm": 0.8973143644014755,
      "learning_rate": 0.0001977072680359959,
      "loss": 12.6792,
      "step": 3535
    },
    {
      "epoch": 0.19254933191752943,
      "grad_norm": 0.7632452197424912,
      "learning_rate": 0.00019770539019577357,
      "loss": 12.7185,
      "step": 3536
    },
    {
      "epoch": 0.19260378591411242,
      "grad_norm": 0.7927514340501035,
      "learning_rate": 0.00019770351159577773,
      "loss": 12.5981,
      "step": 3537
    },
    {
      "epoch": 0.19265823991069544,
      "grad_norm": 0.7068230142827054,
      "learning_rate": 0.000197701632236023,
      "loss": 12.8504,
      "step": 3538
    },
    {
      "epoch": 0.19271269390727847,
      "grad_norm": 0.7134631121518779,
      "learning_rate": 0.000197699752116524,
      "loss": 12.6706,
      "step": 3539
    },
    {
      "epoch": 0.19276714790386146,
      "grad_norm": 0.8083137902876478,
      "learning_rate": 0.00019769787123729535,
      "loss": 12.6535,
      "step": 3540
    },
    {
      "epoch": 0.19282160190044448,
      "grad_norm": 0.7201291945896695,
      "learning_rate": 0.00019769598959835168,
      "loss": 12.6102,
      "step": 3541
    },
    {
      "epoch": 0.1928760558970275,
      "grad_norm": 0.7248858192915315,
      "learning_rate": 0.00019769410719970757,
      "loss": 12.5687,
      "step": 3542
    },
    {
      "epoch": 0.1929305098936105,
      "grad_norm": 0.8362500648989917,
      "learning_rate": 0.00019769222404137773,
      "loss": 12.7465,
      "step": 3543
    },
    {
      "epoch": 0.1929849638901935,
      "grad_norm": 0.705968235830725,
      "learning_rate": 0.00019769034012337677,
      "loss": 12.5646,
      "step": 3544
    },
    {
      "epoch": 0.19303941788677653,
      "grad_norm": 0.7222014227029238,
      "learning_rate": 0.00019768845544571931,
      "loss": 12.6912,
      "step": 3545
    },
    {
      "epoch": 0.19309387188335955,
      "grad_norm": 0.8222074073217709,
      "learning_rate": 0.0001976865700084201,
      "loss": 12.558,
      "step": 3546
    },
    {
      "epoch": 0.19314832587994255,
      "grad_norm": 0.803715808297221,
      "learning_rate": 0.0001976846838114937,
      "loss": 12.765,
      "step": 3547
    },
    {
      "epoch": 0.19320277987652557,
      "grad_norm": 0.9054215650732284,
      "learning_rate": 0.00019768279685495482,
      "loss": 12.6645,
      "step": 3548
    },
    {
      "epoch": 0.19325723387310859,
      "grad_norm": 0.7785454014057418,
      "learning_rate": 0.00019768090913881815,
      "loss": 12.7655,
      "step": 3549
    },
    {
      "epoch": 0.19331168786969158,
      "grad_norm": 0.734109054191284,
      "learning_rate": 0.00019767902066309832,
      "loss": 12.6601,
      "step": 3550
    },
    {
      "epoch": 0.1933661418662746,
      "grad_norm": 0.838383336586159,
      "learning_rate": 0.00019767713142781007,
      "loss": 12.6776,
      "step": 3551
    },
    {
      "epoch": 0.19342059586285762,
      "grad_norm": 0.7860534851895626,
      "learning_rate": 0.00019767524143296804,
      "loss": 12.6996,
      "step": 3552
    },
    {
      "epoch": 0.1934750498594406,
      "grad_norm": 0.7393473966552592,
      "learning_rate": 0.00019767335067858696,
      "loss": 12.7366,
      "step": 3553
    },
    {
      "epoch": 0.19352950385602363,
      "grad_norm": 0.7676063068921294,
      "learning_rate": 0.00019767145916468155,
      "loss": 12.7784,
      "step": 3554
    },
    {
      "epoch": 0.19358395785260665,
      "grad_norm": 0.6940158010728265,
      "learning_rate": 0.00019766956689126647,
      "loss": 12.7075,
      "step": 3555
    },
    {
      "epoch": 0.19363841184918965,
      "grad_norm": 0.8176845403633383,
      "learning_rate": 0.00019766767385835646,
      "loss": 12.6988,
      "step": 3556
    },
    {
      "epoch": 0.19369286584577267,
      "grad_norm": 0.8256601427666457,
      "learning_rate": 0.00019766578006596625,
      "loss": 12.705,
      "step": 3557
    },
    {
      "epoch": 0.1937473198423557,
      "grad_norm": 0.7528657863935839,
      "learning_rate": 0.00019766388551411055,
      "loss": 12.5481,
      "step": 3558
    },
    {
      "epoch": 0.19380177383893868,
      "grad_norm": 0.7907231207947625,
      "learning_rate": 0.00019766199020280407,
      "loss": 12.5655,
      "step": 3559
    },
    {
      "epoch": 0.1938562278355217,
      "grad_norm": 0.8433905921188369,
      "learning_rate": 0.0001976600941320616,
      "loss": 12.6689,
      "step": 3560
    },
    {
      "epoch": 0.19391068183210472,
      "grad_norm": 0.7864133499683497,
      "learning_rate": 0.00019765819730189788,
      "loss": 12.5944,
      "step": 3561
    },
    {
      "epoch": 0.19396513582868774,
      "grad_norm": 0.7921089125035029,
      "learning_rate": 0.00019765629971232762,
      "loss": 12.7421,
      "step": 3562
    },
    {
      "epoch": 0.19401958982527073,
      "grad_norm": 0.769961242712185,
      "learning_rate": 0.00019765440136336563,
      "loss": 12.5976,
      "step": 3563
    },
    {
      "epoch": 0.19407404382185375,
      "grad_norm": 0.7854683420579901,
      "learning_rate": 0.0001976525022550266,
      "loss": 12.7104,
      "step": 3564
    },
    {
      "epoch": 0.19412849781843677,
      "grad_norm": 0.7055786570242942,
      "learning_rate": 0.00019765060238732533,
      "loss": 12.6875,
      "step": 3565
    },
    {
      "epoch": 0.19418295181501977,
      "grad_norm": 0.6676731987699449,
      "learning_rate": 0.0001976487017602766,
      "loss": 12.6477,
      "step": 3566
    },
    {
      "epoch": 0.1942374058116028,
      "grad_norm": 0.7538115788203865,
      "learning_rate": 0.0001976468003738952,
      "loss": 12.7183,
      "step": 3567
    },
    {
      "epoch": 0.1942918598081858,
      "grad_norm": 0.75439765158688,
      "learning_rate": 0.00019764489822819594,
      "loss": 12.6667,
      "step": 3568
    },
    {
      "epoch": 0.1943463138047688,
      "grad_norm": 0.7020514027822288,
      "learning_rate": 0.00019764299532319354,
      "loss": 12.6256,
      "step": 3569
    },
    {
      "epoch": 0.19440076780135182,
      "grad_norm": 0.7834423558108528,
      "learning_rate": 0.00019764109165890283,
      "loss": 12.6853,
      "step": 3570
    },
    {
      "epoch": 0.19445522179793484,
      "grad_norm": 0.8136321623655445,
      "learning_rate": 0.00019763918723533864,
      "loss": 12.6909,
      "step": 3571
    },
    {
      "epoch": 0.19450967579451783,
      "grad_norm": 0.8091687627500386,
      "learning_rate": 0.00019763728205251572,
      "loss": 12.8608,
      "step": 3572
    },
    {
      "epoch": 0.19456412979110085,
      "grad_norm": 0.8657728047541944,
      "learning_rate": 0.00019763537611044892,
      "loss": 12.6988,
      "step": 3573
    },
    {
      "epoch": 0.19461858378768387,
      "grad_norm": 0.7915471591130552,
      "learning_rate": 0.0001976334694091531,
      "loss": 12.491,
      "step": 3574
    },
    {
      "epoch": 0.19467303778426687,
      "grad_norm": 0.7494389684777423,
      "learning_rate": 0.00019763156194864306,
      "loss": 12.721,
      "step": 3575
    },
    {
      "epoch": 0.1947274917808499,
      "grad_norm": 0.73901543888061,
      "learning_rate": 0.0001976296537289336,
      "loss": 12.7059,
      "step": 3576
    },
    {
      "epoch": 0.1947819457774329,
      "grad_norm": 0.7546077616744777,
      "learning_rate": 0.00019762774475003955,
      "loss": 12.6357,
      "step": 3577
    },
    {
      "epoch": 0.1948363997740159,
      "grad_norm": 0.7994713205953073,
      "learning_rate": 0.00019762583501197582,
      "loss": 12.7197,
      "step": 3578
    },
    {
      "epoch": 0.19489085377059892,
      "grad_norm": 0.7869063678294197,
      "learning_rate": 0.00019762392451475722,
      "loss": 12.6343,
      "step": 3579
    },
    {
      "epoch": 0.19494530776718194,
      "grad_norm": 0.7443898321946537,
      "learning_rate": 0.0001976220132583986,
      "loss": 12.5181,
      "step": 3580
    },
    {
      "epoch": 0.19499976176376496,
      "grad_norm": 0.6598295189631749,
      "learning_rate": 0.00019762010124291484,
      "loss": 12.4168,
      "step": 3581
    },
    {
      "epoch": 0.19505421576034795,
      "grad_norm": 0.6901751080579652,
      "learning_rate": 0.0001976181884683208,
      "loss": 12.7246,
      "step": 3582
    },
    {
      "epoch": 0.19510866975693097,
      "grad_norm": 0.8222129196616311,
      "learning_rate": 0.00019761627493463136,
      "loss": 12.6252,
      "step": 3583
    },
    {
      "epoch": 0.195163123753514,
      "grad_norm": 0.694561116471691,
      "learning_rate": 0.00019761436064186138,
      "loss": 12.6051,
      "step": 3584
    },
    {
      "epoch": 0.195217577750097,
      "grad_norm": 0.7677200945317284,
      "learning_rate": 0.0001976124455900258,
      "loss": 12.7399,
      "step": 3585
    },
    {
      "epoch": 0.19527203174668,
      "grad_norm": 0.7603252493444852,
      "learning_rate": 0.00019761052977913942,
      "loss": 12.5219,
      "step": 3586
    },
    {
      "epoch": 0.19532648574326303,
      "grad_norm": 0.7413408765375056,
      "learning_rate": 0.00019760861320921723,
      "loss": 12.7147,
      "step": 3587
    },
    {
      "epoch": 0.19538093973984602,
      "grad_norm": 0.7076795749184328,
      "learning_rate": 0.00019760669588027408,
      "loss": 12.5845,
      "step": 3588
    },
    {
      "epoch": 0.19543539373642904,
      "grad_norm": 0.7097969871713746,
      "learning_rate": 0.0001976047777923249,
      "loss": 12.5254,
      "step": 3589
    },
    {
      "epoch": 0.19548984773301206,
      "grad_norm": 0.6822010258754557,
      "learning_rate": 0.0001976028589453846,
      "loss": 12.5417,
      "step": 3590
    },
    {
      "epoch": 0.19554430172959505,
      "grad_norm": 0.7519241744027907,
      "learning_rate": 0.00019760093933946809,
      "loss": 12.5946,
      "step": 3591
    },
    {
      "epoch": 0.19559875572617808,
      "grad_norm": 0.7397640484556057,
      "learning_rate": 0.00019759901897459033,
      "loss": 12.6998,
      "step": 3592
    },
    {
      "epoch": 0.1956532097227611,
      "grad_norm": 0.7444813318525035,
      "learning_rate": 0.0001975970978507662,
      "loss": 12.6904,
      "step": 3593
    },
    {
      "epoch": 0.1957076637193441,
      "grad_norm": 0.6485345517894925,
      "learning_rate": 0.0001975951759680107,
      "loss": 12.5317,
      "step": 3594
    },
    {
      "epoch": 0.1957621177159271,
      "grad_norm": 0.7359840794178918,
      "learning_rate": 0.00019759325332633872,
      "loss": 12.683,
      "step": 3595
    },
    {
      "epoch": 0.19581657171251013,
      "grad_norm": 0.7701808720342013,
      "learning_rate": 0.00019759132992576528,
      "loss": 12.7741,
      "step": 3596
    },
    {
      "epoch": 0.19587102570909315,
      "grad_norm": 0.7356083817207066,
      "learning_rate": 0.00019758940576630524,
      "loss": 12.445,
      "step": 3597
    },
    {
      "epoch": 0.19592547970567614,
      "grad_norm": 0.7055571693231458,
      "learning_rate": 0.00019758748084797363,
      "loss": 12.6499,
      "step": 3598
    },
    {
      "epoch": 0.19597993370225916,
      "grad_norm": 0.7139018855523392,
      "learning_rate": 0.00019758555517078544,
      "loss": 12.5885,
      "step": 3599
    },
    {
      "epoch": 0.19603438769884218,
      "grad_norm": 0.6686289752594662,
      "learning_rate": 0.00019758362873475557,
      "loss": 12.6985,
      "step": 3600
    },
    {
      "epoch": 0.19608884169542518,
      "grad_norm": 0.7238026367304898,
      "learning_rate": 0.00019758170153989904,
      "loss": 12.6118,
      "step": 3601
    },
    {
      "epoch": 0.1961432956920082,
      "grad_norm": 0.7231655713054286,
      "learning_rate": 0.00019757977358623083,
      "loss": 12.5382,
      "step": 3602
    },
    {
      "epoch": 0.19619774968859122,
      "grad_norm": 0.7240620390875896,
      "learning_rate": 0.00019757784487376597,
      "loss": 12.6104,
      "step": 3603
    },
    {
      "epoch": 0.1962522036851742,
      "grad_norm": 0.8309518403394776,
      "learning_rate": 0.00019757591540251937,
      "loss": 12.786,
      "step": 3604
    },
    {
      "epoch": 0.19630665768175723,
      "grad_norm": 0.7182033651055718,
      "learning_rate": 0.00019757398517250612,
      "loss": 12.6781,
      "step": 3605
    },
    {
      "epoch": 0.19636111167834025,
      "grad_norm": 0.6624124827710981,
      "learning_rate": 0.0001975720541837412,
      "loss": 12.6035,
      "step": 3606
    },
    {
      "epoch": 0.19641556567492324,
      "grad_norm": 0.8103886947473927,
      "learning_rate": 0.00019757012243623963,
      "loss": 12.5608,
      "step": 3607
    },
    {
      "epoch": 0.19647001967150626,
      "grad_norm": 0.6924633612516438,
      "learning_rate": 0.0001975681899300164,
      "loss": 12.6219,
      "step": 3608
    },
    {
      "epoch": 0.19652447366808928,
      "grad_norm": 0.7250157464819681,
      "learning_rate": 0.0001975662566650866,
      "loss": 12.6682,
      "step": 3609
    },
    {
      "epoch": 0.19657892766467228,
      "grad_norm": 0.8019069046132882,
      "learning_rate": 0.0001975643226414652,
      "loss": 12.5714,
      "step": 3610
    },
    {
      "epoch": 0.1966333816612553,
      "grad_norm": 0.8310541251566375,
      "learning_rate": 0.00019756238785916729,
      "loss": 12.6828,
      "step": 3611
    },
    {
      "epoch": 0.19668783565783832,
      "grad_norm": 0.6882459675921699,
      "learning_rate": 0.00019756045231820784,
      "loss": 12.6901,
      "step": 3612
    },
    {
      "epoch": 0.19674228965442134,
      "grad_norm": 0.8262341699163686,
      "learning_rate": 0.000197558516018602,
      "loss": 12.7507,
      "step": 3613
    },
    {
      "epoch": 0.19679674365100433,
      "grad_norm": 0.7657928247465275,
      "learning_rate": 0.00019755657896036475,
      "loss": 12.6165,
      "step": 3614
    },
    {
      "epoch": 0.19685119764758735,
      "grad_norm": 0.7631725744086381,
      "learning_rate": 0.0001975546411435112,
      "loss": 12.5987,
      "step": 3615
    },
    {
      "epoch": 0.19690565164417037,
      "grad_norm": 0.7345946704374849,
      "learning_rate": 0.0001975527025680564,
      "loss": 12.594,
      "step": 3616
    },
    {
      "epoch": 0.19696010564075336,
      "grad_norm": 0.7681731733906019,
      "learning_rate": 0.00019755076323401543,
      "loss": 12.6932,
      "step": 3617
    },
    {
      "epoch": 0.19701455963733638,
      "grad_norm": 0.8080127465641304,
      "learning_rate": 0.00019754882314140335,
      "loss": 12.7142,
      "step": 3618
    },
    {
      "epoch": 0.1970690136339194,
      "grad_norm": 0.8400502525702276,
      "learning_rate": 0.00019754688229023528,
      "loss": 12.6429,
      "step": 3619
    },
    {
      "epoch": 0.1971234676305024,
      "grad_norm": 0.7696932357860664,
      "learning_rate": 0.00019754494068052628,
      "loss": 12.6779,
      "step": 3620
    },
    {
      "epoch": 0.19717792162708542,
      "grad_norm": 0.7362750272307614,
      "learning_rate": 0.00019754299831229146,
      "loss": 12.6583,
      "step": 3621
    },
    {
      "epoch": 0.19723237562366844,
      "grad_norm": 0.9548179295323732,
      "learning_rate": 0.00019754105518554594,
      "loss": 12.7402,
      "step": 3622
    },
    {
      "epoch": 0.19728682962025143,
      "grad_norm": 0.7760525845636607,
      "learning_rate": 0.0001975391113003048,
      "loss": 12.7086,
      "step": 3623
    },
    {
      "epoch": 0.19734128361683445,
      "grad_norm": 0.9345824260359603,
      "learning_rate": 0.0001975371666565832,
      "loss": 12.653,
      "step": 3624
    },
    {
      "epoch": 0.19739573761341747,
      "grad_norm": 0.8025717083886728,
      "learning_rate": 0.00019753522125439622,
      "loss": 12.6398,
      "step": 3625
    },
    {
      "epoch": 0.19745019161000046,
      "grad_norm": 0.7705338143926113,
      "learning_rate": 0.00019753327509375898,
      "loss": 12.5353,
      "step": 3626
    },
    {
      "epoch": 0.19750464560658348,
      "grad_norm": 0.9081350449246679,
      "learning_rate": 0.00019753132817468667,
      "loss": 12.6306,
      "step": 3627
    },
    {
      "epoch": 0.1975590996031665,
      "grad_norm": 0.7150794542815331,
      "learning_rate": 0.00019752938049719438,
      "loss": 12.6129,
      "step": 3628
    },
    {
      "epoch": 0.19761355359974953,
      "grad_norm": 0.7866917662670726,
      "learning_rate": 0.0001975274320612973,
      "loss": 12.7477,
      "step": 3629
    },
    {
      "epoch": 0.19766800759633252,
      "grad_norm": 0.7704851762300791,
      "learning_rate": 0.00019752548286701053,
      "loss": 12.6923,
      "step": 3630
    },
    {
      "epoch": 0.19772246159291554,
      "grad_norm": 0.830753690438923,
      "learning_rate": 0.00019752353291434922,
      "loss": 12.7148,
      "step": 3631
    },
    {
      "epoch": 0.19777691558949856,
      "grad_norm": 0.9358040617724722,
      "learning_rate": 0.00019752158220332858,
      "loss": 12.8388,
      "step": 3632
    },
    {
      "epoch": 0.19783136958608155,
      "grad_norm": 0.7476298552772048,
      "learning_rate": 0.0001975196307339638,
      "loss": 12.6567,
      "step": 3633
    },
    {
      "epoch": 0.19788582358266457,
      "grad_norm": 0.7800103956608793,
      "learning_rate": 0.00019751767850627,
      "loss": 12.6094,
      "step": 3634
    },
    {
      "epoch": 0.1979402775792476,
      "grad_norm": 0.7566188148483881,
      "learning_rate": 0.00019751572552026235,
      "loss": 12.6221,
      "step": 3635
    },
    {
      "epoch": 0.19799473157583058,
      "grad_norm": 0.6787264592319261,
      "learning_rate": 0.0001975137717759561,
      "loss": 12.7796,
      "step": 3636
    },
    {
      "epoch": 0.1980491855724136,
      "grad_norm": 0.7822758510904512,
      "learning_rate": 0.00019751181727336637,
      "loss": 12.6914,
      "step": 3637
    },
    {
      "epoch": 0.19810363956899663,
      "grad_norm": 0.710249806995049,
      "learning_rate": 0.00019750986201250842,
      "loss": 12.7077,
      "step": 3638
    },
    {
      "epoch": 0.19815809356557962,
      "grad_norm": 0.7490688178996289,
      "learning_rate": 0.00019750790599339744,
      "loss": 12.7766,
      "step": 3639
    },
    {
      "epoch": 0.19821254756216264,
      "grad_norm": 0.7043863922849712,
      "learning_rate": 0.00019750594921604862,
      "loss": 12.6596,
      "step": 3640
    },
    {
      "epoch": 0.19826700155874566,
      "grad_norm": 0.8152263313748491,
      "learning_rate": 0.0001975039916804772,
      "loss": 12.7449,
      "step": 3641
    },
    {
      "epoch": 0.19832145555532865,
      "grad_norm": 0.7191860510140848,
      "learning_rate": 0.00019750203338669836,
      "loss": 12.7179,
      "step": 3642
    },
    {
      "epoch": 0.19837590955191167,
      "grad_norm": 0.7064986821688003,
      "learning_rate": 0.00019750007433472737,
      "loss": 12.6284,
      "step": 3643
    },
    {
      "epoch": 0.1984303635484947,
      "grad_norm": 1.002927774669751,
      "learning_rate": 0.00019749811452457946,
      "loss": 12.4528,
      "step": 3644
    },
    {
      "epoch": 0.19848481754507769,
      "grad_norm": 0.7787618250614149,
      "learning_rate": 0.00019749615395626985,
      "loss": 12.6582,
      "step": 3645
    },
    {
      "epoch": 0.1985392715416607,
      "grad_norm": 0.8308895758175826,
      "learning_rate": 0.0001974941926298138,
      "loss": 12.7564,
      "step": 3646
    },
    {
      "epoch": 0.19859372553824373,
      "grad_norm": 0.697413418831933,
      "learning_rate": 0.00019749223054522656,
      "loss": 12.6837,
      "step": 3647
    },
    {
      "epoch": 0.19864817953482675,
      "grad_norm": 0.6846133126736759,
      "learning_rate": 0.0001974902677025234,
      "loss": 12.4405,
      "step": 3648
    },
    {
      "epoch": 0.19870263353140974,
      "grad_norm": 0.7054346767013996,
      "learning_rate": 0.00019748830410171956,
      "loss": 12.6491,
      "step": 3649
    },
    {
      "epoch": 0.19875708752799276,
      "grad_norm": 0.8471597612637782,
      "learning_rate": 0.00019748633974283033,
      "loss": 12.7337,
      "step": 3650
    },
    {
      "epoch": 0.19881154152457578,
      "grad_norm": 0.6812288046264803,
      "learning_rate": 0.00019748437462587096,
      "loss": 12.5475,
      "step": 3651
    },
    {
      "epoch": 0.19886599552115877,
      "grad_norm": 0.8246124260906363,
      "learning_rate": 0.00019748240875085672,
      "loss": 12.6354,
      "step": 3652
    },
    {
      "epoch": 0.1989204495177418,
      "grad_norm": 0.7300175077201014,
      "learning_rate": 0.00019748044211780297,
      "loss": 12.5718,
      "step": 3653
    },
    {
      "epoch": 0.1989749035143248,
      "grad_norm": 0.7580912082168959,
      "learning_rate": 0.00019747847472672493,
      "loss": 12.6165,
      "step": 3654
    },
    {
      "epoch": 0.1990293575109078,
      "grad_norm": 0.6839688816562365,
      "learning_rate": 0.00019747650657763792,
      "loss": 12.6779,
      "step": 3655
    },
    {
      "epoch": 0.19908381150749083,
      "grad_norm": 0.7581402696591821,
      "learning_rate": 0.00019747453767055725,
      "loss": 12.5625,
      "step": 3656
    },
    {
      "epoch": 0.19913826550407385,
      "grad_norm": 0.7254368507070799,
      "learning_rate": 0.00019747256800549824,
      "loss": 12.529,
      "step": 3657
    },
    {
      "epoch": 0.19919271950065684,
      "grad_norm": 0.6925063557379147,
      "learning_rate": 0.00019747059758247617,
      "loss": 12.56,
      "step": 3658
    },
    {
      "epoch": 0.19924717349723986,
      "grad_norm": 0.7606285519649787,
      "learning_rate": 0.00019746862640150642,
      "loss": 12.5587,
      "step": 3659
    },
    {
      "epoch": 0.19930162749382288,
      "grad_norm": 0.726365152862436,
      "learning_rate": 0.00019746665446260426,
      "loss": 12.6198,
      "step": 3660
    },
    {
      "epoch": 0.19935608149040587,
      "grad_norm": 0.6861309571156384,
      "learning_rate": 0.00019746468176578503,
      "loss": 12.7161,
      "step": 3661
    },
    {
      "epoch": 0.1994105354869889,
      "grad_norm": 0.8451387517182866,
      "learning_rate": 0.00019746270831106415,
      "loss": 12.6925,
      "step": 3662
    },
    {
      "epoch": 0.19946498948357191,
      "grad_norm": 0.8460799924334934,
      "learning_rate": 0.00019746073409845685,
      "loss": 12.5613,
      "step": 3663
    },
    {
      "epoch": 0.19951944348015493,
      "grad_norm": 0.8082933138547622,
      "learning_rate": 0.00019745875912797857,
      "loss": 12.4263,
      "step": 3664
    },
    {
      "epoch": 0.19957389747673793,
      "grad_norm": 0.6649570550012347,
      "learning_rate": 0.00019745678339964462,
      "loss": 12.512,
      "step": 3665
    },
    {
      "epoch": 0.19962835147332095,
      "grad_norm": 0.7607425799101774,
      "learning_rate": 0.00019745480691347038,
      "loss": 12.7488,
      "step": 3666
    },
    {
      "epoch": 0.19968280546990397,
      "grad_norm": 0.7754492850947158,
      "learning_rate": 0.00019745282966947123,
      "loss": 12.7272,
      "step": 3667
    },
    {
      "epoch": 0.19973725946648696,
      "grad_norm": 0.7840978078878431,
      "learning_rate": 0.00019745085166766253,
      "loss": 12.6944,
      "step": 3668
    },
    {
      "epoch": 0.19979171346306998,
      "grad_norm": 0.6741926222868774,
      "learning_rate": 0.00019744887290805963,
      "loss": 12.5978,
      "step": 3669
    },
    {
      "epoch": 0.199846167459653,
      "grad_norm": 0.7342454477300046,
      "learning_rate": 0.000197446893390678,
      "loss": 12.6297,
      "step": 3670
    },
    {
      "epoch": 0.199900621456236,
      "grad_norm": 0.8151529838473607,
      "learning_rate": 0.00019744491311553296,
      "loss": 12.8349,
      "step": 3671
    },
    {
      "epoch": 0.19995507545281901,
      "grad_norm": 0.738421652122142,
      "learning_rate": 0.00019744293208263995,
      "loss": 12.7038,
      "step": 3672
    },
    {
      "epoch": 0.20000952944940203,
      "grad_norm": 0.7168652829178292,
      "learning_rate": 0.00019744095029201438,
      "loss": 12.6497,
      "step": 3673
    },
    {
      "epoch": 0.20006398344598503,
      "grad_norm": 0.8215008669818747,
      "learning_rate": 0.0001974389677436716,
      "loss": 12.6954,
      "step": 3674
    },
    {
      "epoch": 0.20011843744256805,
      "grad_norm": 0.8097171866747358,
      "learning_rate": 0.0001974369844376271,
      "loss": 12.5313,
      "step": 3675
    },
    {
      "epoch": 0.20017289143915107,
      "grad_norm": 0.7084186130888639,
      "learning_rate": 0.00019743500037389624,
      "loss": 12.6974,
      "step": 3676
    },
    {
      "epoch": 0.20022734543573406,
      "grad_norm": 0.7524209666082665,
      "learning_rate": 0.00019743301555249446,
      "loss": 12.5611,
      "step": 3677
    },
    {
      "epoch": 0.20028179943231708,
      "grad_norm": 0.7798370366657434,
      "learning_rate": 0.00019743102997343725,
      "loss": 12.5747,
      "step": 3678
    },
    {
      "epoch": 0.2003362534289001,
      "grad_norm": 0.8022141403799491,
      "learning_rate": 0.00019742904363674,
      "loss": 12.7231,
      "step": 3679
    },
    {
      "epoch": 0.20039070742548312,
      "grad_norm": 0.8906558553915508,
      "learning_rate": 0.00019742705654241815,
      "loss": 12.6655,
      "step": 3680
    },
    {
      "epoch": 0.20044516142206611,
      "grad_norm": 0.6849201171068462,
      "learning_rate": 0.00019742506869048718,
      "loss": 12.6287,
      "step": 3681
    },
    {
      "epoch": 0.20049961541864914,
      "grad_norm": 0.7380211492839521,
      "learning_rate": 0.00019742308008096254,
      "loss": 12.806,
      "step": 3682
    },
    {
      "epoch": 0.20055406941523216,
      "grad_norm": 0.7440520912943274,
      "learning_rate": 0.00019742109071385972,
      "loss": 12.6976,
      "step": 3683
    },
    {
      "epoch": 0.20060852341181515,
      "grad_norm": 0.8094904770848835,
      "learning_rate": 0.0001974191005891941,
      "loss": 12.7635,
      "step": 3684
    },
    {
      "epoch": 0.20066297740839817,
      "grad_norm": 1.0039098394519197,
      "learning_rate": 0.0001974171097069813,
      "loss": 12.8509,
      "step": 3685
    },
    {
      "epoch": 0.2007174314049812,
      "grad_norm": 0.720703183546859,
      "learning_rate": 0.00019741511806723664,
      "loss": 12.6221,
      "step": 3686
    },
    {
      "epoch": 0.20077188540156418,
      "grad_norm": 0.8402311327294334,
      "learning_rate": 0.00019741312566997572,
      "loss": 12.6864,
      "step": 3687
    },
    {
      "epoch": 0.2008263393981472,
      "grad_norm": 0.6559263382773919,
      "learning_rate": 0.00019741113251521398,
      "loss": 12.4228,
      "step": 3688
    },
    {
      "epoch": 0.20088079339473022,
      "grad_norm": 0.721615468741822,
      "learning_rate": 0.00019740913860296697,
      "loss": 12.6649,
      "step": 3689
    },
    {
      "epoch": 0.20093524739131322,
      "grad_norm": 0.7573016725666538,
      "learning_rate": 0.00019740714393325014,
      "loss": 12.5624,
      "step": 3690
    },
    {
      "epoch": 0.20098970138789624,
      "grad_norm": 0.8287294148461015,
      "learning_rate": 0.00019740514850607904,
      "loss": 12.7974,
      "step": 3691
    },
    {
      "epoch": 0.20104415538447926,
      "grad_norm": 0.7486278885824178,
      "learning_rate": 0.00019740315232146913,
      "loss": 12.6035,
      "step": 3692
    },
    {
      "epoch": 0.20109860938106225,
      "grad_norm": 0.7321991888679197,
      "learning_rate": 0.00019740115537943603,
      "loss": 12.7047,
      "step": 3693
    },
    {
      "epoch": 0.20115306337764527,
      "grad_norm": 0.7230726632935411,
      "learning_rate": 0.00019739915767999518,
      "loss": 12.6969,
      "step": 3694
    },
    {
      "epoch": 0.2012075173742283,
      "grad_norm": 0.8336807145808396,
      "learning_rate": 0.00019739715922316214,
      "loss": 12.7886,
      "step": 3695
    },
    {
      "epoch": 0.2012619713708113,
      "grad_norm": 0.8261360393578282,
      "learning_rate": 0.00019739516000895246,
      "loss": 12.7838,
      "step": 3696
    },
    {
      "epoch": 0.2013164253673943,
      "grad_norm": 0.7161373279083758,
      "learning_rate": 0.00019739316003738167,
      "loss": 12.6303,
      "step": 3697
    },
    {
      "epoch": 0.20137087936397732,
      "grad_norm": 0.6807051699186523,
      "learning_rate": 0.00019739115930846537,
      "loss": 12.626,
      "step": 3698
    },
    {
      "epoch": 0.20142533336056034,
      "grad_norm": 0.6295910457426575,
      "learning_rate": 0.00019738915782221907,
      "loss": 12.5255,
      "step": 3699
    },
    {
      "epoch": 0.20147978735714334,
      "grad_norm": 0.7683727689702888,
      "learning_rate": 0.00019738715557865834,
      "loss": 12.7087,
      "step": 3700
    },
    {
      "epoch": 0.20153424135372636,
      "grad_norm": 0.744167704136972,
      "learning_rate": 0.00019738515257779877,
      "loss": 12.5179,
      "step": 3701
    },
    {
      "epoch": 0.20158869535030938,
      "grad_norm": 0.69197717897407,
      "learning_rate": 0.0001973831488196559,
      "loss": 12.4385,
      "step": 3702
    },
    {
      "epoch": 0.20164314934689237,
      "grad_norm": 0.7643508370180074,
      "learning_rate": 0.00019738114430424534,
      "loss": 12.5915,
      "step": 3703
    },
    {
      "epoch": 0.2016976033434754,
      "grad_norm": 0.7776169332587081,
      "learning_rate": 0.00019737913903158268,
      "loss": 12.7894,
      "step": 3704
    },
    {
      "epoch": 0.2017520573400584,
      "grad_norm": 0.7447411020234989,
      "learning_rate": 0.0001973771330016835,
      "loss": 12.5682,
      "step": 3705
    },
    {
      "epoch": 0.2018065113366414,
      "grad_norm": 0.7246836967348779,
      "learning_rate": 0.0001973751262145634,
      "loss": 12.6121,
      "step": 3706
    },
    {
      "epoch": 0.20186096533322442,
      "grad_norm": 0.7535649752440313,
      "learning_rate": 0.00019737311867023798,
      "loss": 12.6519,
      "step": 3707
    },
    {
      "epoch": 0.20191541932980744,
      "grad_norm": 0.7390233694094689,
      "learning_rate": 0.0001973711103687229,
      "loss": 12.8212,
      "step": 3708
    },
    {
      "epoch": 0.20196987332639044,
      "grad_norm": 0.8148239607128785,
      "learning_rate": 0.00019736910131003369,
      "loss": 12.6742,
      "step": 3709
    },
    {
      "epoch": 0.20202432732297346,
      "grad_norm": 0.7451607895808458,
      "learning_rate": 0.00019736709149418603,
      "loss": 12.7395,
      "step": 3710
    },
    {
      "epoch": 0.20207878131955648,
      "grad_norm": 0.7093270748909235,
      "learning_rate": 0.00019736508092119554,
      "loss": 12.644,
      "step": 3711
    },
    {
      "epoch": 0.20213323531613947,
      "grad_norm": 0.8758348356705866,
      "learning_rate": 0.00019736306959107787,
      "loss": 12.8661,
      "step": 3712
    },
    {
      "epoch": 0.2021876893127225,
      "grad_norm": 0.6816408808300222,
      "learning_rate": 0.00019736105750384864,
      "loss": 12.6664,
      "step": 3713
    },
    {
      "epoch": 0.2022421433093055,
      "grad_norm": 0.7525269361120305,
      "learning_rate": 0.00019735904465952348,
      "loss": 12.6835,
      "step": 3714
    },
    {
      "epoch": 0.20229659730588853,
      "grad_norm": 0.8220374711376354,
      "learning_rate": 0.00019735703105811807,
      "loss": 12.6514,
      "step": 3715
    },
    {
      "epoch": 0.20235105130247152,
      "grad_norm": 0.6714431280177346,
      "learning_rate": 0.00019735501669964806,
      "loss": 12.6893,
      "step": 3716
    },
    {
      "epoch": 0.20240550529905454,
      "grad_norm": 0.9556940251485908,
      "learning_rate": 0.00019735300158412911,
      "loss": 12.7626,
      "step": 3717
    },
    {
      "epoch": 0.20245995929563756,
      "grad_norm": 0.7312910563842567,
      "learning_rate": 0.0001973509857115769,
      "loss": 12.5876,
      "step": 3718
    },
    {
      "epoch": 0.20251441329222056,
      "grad_norm": 0.7475186407951433,
      "learning_rate": 0.0001973489690820071,
      "loss": 12.6019,
      "step": 3719
    },
    {
      "epoch": 0.20256886728880358,
      "grad_norm": 0.7941381101364623,
      "learning_rate": 0.0001973469516954354,
      "loss": 12.648,
      "step": 3720
    },
    {
      "epoch": 0.2026233212853866,
      "grad_norm": 0.7493618173219636,
      "learning_rate": 0.00019734493355187747,
      "loss": 12.5324,
      "step": 3721
    },
    {
      "epoch": 0.2026777752819696,
      "grad_norm": 0.7908043493982647,
      "learning_rate": 0.00019734291465134903,
      "loss": 12.6918,
      "step": 3722
    },
    {
      "epoch": 0.2027322292785526,
      "grad_norm": 0.9627774794856256,
      "learning_rate": 0.00019734089499386573,
      "loss": 12.6197,
      "step": 3723
    },
    {
      "epoch": 0.20278668327513563,
      "grad_norm": 0.7098037134646387,
      "learning_rate": 0.0001973388745794433,
      "loss": 12.6087,
      "step": 3724
    },
    {
      "epoch": 0.20284113727171862,
      "grad_norm": 0.7636725436853691,
      "learning_rate": 0.0001973368534080975,
      "loss": 12.6153,
      "step": 3725
    },
    {
      "epoch": 0.20289559126830164,
      "grad_norm": 0.7626349542601414,
      "learning_rate": 0.00019733483147984395,
      "loss": 12.6811,
      "step": 3726
    },
    {
      "epoch": 0.20295004526488467,
      "grad_norm": 0.74736286062578,
      "learning_rate": 0.00019733280879469847,
      "loss": 12.6376,
      "step": 3727
    },
    {
      "epoch": 0.20300449926146766,
      "grad_norm": 0.7195779098222057,
      "learning_rate": 0.00019733078535267673,
      "loss": 12.6022,
      "step": 3728
    },
    {
      "epoch": 0.20305895325805068,
      "grad_norm": 0.7584375166225391,
      "learning_rate": 0.00019732876115379449,
      "loss": 12.6474,
      "step": 3729
    },
    {
      "epoch": 0.2031134072546337,
      "grad_norm": 0.6761342916077878,
      "learning_rate": 0.00019732673619806746,
      "loss": 12.5903,
      "step": 3730
    },
    {
      "epoch": 0.20316786125121672,
      "grad_norm": 0.7292770150961099,
      "learning_rate": 0.00019732471048551143,
      "loss": 12.5185,
      "step": 3731
    },
    {
      "epoch": 0.2032223152477997,
      "grad_norm": 0.6717013120758528,
      "learning_rate": 0.00019732268401614214,
      "loss": 12.7058,
      "step": 3732
    },
    {
      "epoch": 0.20327676924438273,
      "grad_norm": 0.7336554411162386,
      "learning_rate": 0.00019732065678997529,
      "loss": 12.6187,
      "step": 3733
    },
    {
      "epoch": 0.20333122324096575,
      "grad_norm": 0.6693441989255688,
      "learning_rate": 0.00019731862880702675,
      "loss": 12.5719,
      "step": 3734
    },
    {
      "epoch": 0.20338567723754875,
      "grad_norm": 0.7673519285327757,
      "learning_rate": 0.0001973166000673122,
      "loss": 12.6662,
      "step": 3735
    },
    {
      "epoch": 0.20344013123413177,
      "grad_norm": 0.6610446427454186,
      "learning_rate": 0.00019731457057084746,
      "loss": 12.6042,
      "step": 3736
    },
    {
      "epoch": 0.2034945852307148,
      "grad_norm": 0.7575110207153776,
      "learning_rate": 0.0001973125403176483,
      "loss": 12.6657,
      "step": 3737
    },
    {
      "epoch": 0.20354903922729778,
      "grad_norm": 0.7236274014841737,
      "learning_rate": 0.00019731050930773048,
      "loss": 12.665,
      "step": 3738
    },
    {
      "epoch": 0.2036034932238808,
      "grad_norm": 0.7145867625466173,
      "learning_rate": 0.00019730847754110983,
      "loss": 12.6669,
      "step": 3739
    },
    {
      "epoch": 0.20365794722046382,
      "grad_norm": 0.7723555587156568,
      "learning_rate": 0.00019730644501780216,
      "loss": 12.5307,
      "step": 3740
    },
    {
      "epoch": 0.2037124012170468,
      "grad_norm": 0.7367181865589261,
      "learning_rate": 0.00019730441173782323,
      "loss": 12.5939,
      "step": 3741
    },
    {
      "epoch": 0.20376685521362983,
      "grad_norm": 0.7185443582889142,
      "learning_rate": 0.0001973023777011889,
      "loss": 12.5633,
      "step": 3742
    },
    {
      "epoch": 0.20382130921021285,
      "grad_norm": 0.9380030962463133,
      "learning_rate": 0.00019730034290791495,
      "loss": 12.8148,
      "step": 3743
    },
    {
      "epoch": 0.20387576320679585,
      "grad_norm": 0.7097460165179456,
      "learning_rate": 0.00019729830735801723,
      "loss": 12.5936,
      "step": 3744
    },
    {
      "epoch": 0.20393021720337887,
      "grad_norm": 0.7630980977859622,
      "learning_rate": 0.00019729627105151157,
      "loss": 12.5326,
      "step": 3745
    },
    {
      "epoch": 0.2039846711999619,
      "grad_norm": 0.7874843041862235,
      "learning_rate": 0.00019729423398841375,
      "loss": 12.7463,
      "step": 3746
    },
    {
      "epoch": 0.2040391251965449,
      "grad_norm": 0.7706619050603203,
      "learning_rate": 0.00019729219616873965,
      "loss": 12.5942,
      "step": 3747
    },
    {
      "epoch": 0.2040935791931279,
      "grad_norm": 0.8042598954022532,
      "learning_rate": 0.00019729015759250516,
      "loss": 12.6013,
      "step": 3748
    },
    {
      "epoch": 0.20414803318971092,
      "grad_norm": 0.7745473764242377,
      "learning_rate": 0.00019728811825972604,
      "loss": 12.5237,
      "step": 3749
    },
    {
      "epoch": 0.20420248718629394,
      "grad_norm": 0.7852220734332034,
      "learning_rate": 0.0001972860781704182,
      "loss": 12.652,
      "step": 3750
    },
    {
      "epoch": 0.20425694118287693,
      "grad_norm": 0.7271940145806121,
      "learning_rate": 0.00019728403732459756,
      "loss": 12.6442,
      "step": 3751
    },
    {
      "epoch": 0.20431139517945995,
      "grad_norm": 0.796056460138645,
      "learning_rate": 0.00019728199572227988,
      "loss": 12.4352,
      "step": 3752
    },
    {
      "epoch": 0.20436584917604297,
      "grad_norm": 0.9419382278691907,
      "learning_rate": 0.0001972799533634811,
      "loss": 12.5889,
      "step": 3753
    },
    {
      "epoch": 0.20442030317262597,
      "grad_norm": 0.687316538090516,
      "learning_rate": 0.0001972779102482171,
      "loss": 12.5692,
      "step": 3754
    },
    {
      "epoch": 0.204474757169209,
      "grad_norm": 0.7396866618011716,
      "learning_rate": 0.00019727586637650373,
      "loss": 12.5719,
      "step": 3755
    },
    {
      "epoch": 0.204529211165792,
      "grad_norm": 0.8042947033928121,
      "learning_rate": 0.00019727382174835692,
      "loss": 12.6433,
      "step": 3756
    },
    {
      "epoch": 0.204583665162375,
      "grad_norm": 0.783706359978815,
      "learning_rate": 0.00019727177636379257,
      "loss": 12.6691,
      "step": 3757
    },
    {
      "epoch": 0.20463811915895802,
      "grad_norm": 0.8823304070658445,
      "learning_rate": 0.00019726973022282657,
      "loss": 12.6565,
      "step": 3758
    },
    {
      "epoch": 0.20469257315554104,
      "grad_norm": 0.7065231139049295,
      "learning_rate": 0.00019726768332547484,
      "loss": 12.642,
      "step": 3759
    },
    {
      "epoch": 0.20474702715212403,
      "grad_norm": 0.8411424177529961,
      "learning_rate": 0.00019726563567175326,
      "loss": 12.6264,
      "step": 3760
    },
    {
      "epoch": 0.20480148114870705,
      "grad_norm": 0.7193541612205515,
      "learning_rate": 0.00019726358726167783,
      "loss": 12.7119,
      "step": 3761
    },
    {
      "epoch": 0.20485593514529007,
      "grad_norm": 0.826046012175022,
      "learning_rate": 0.0001972615380952644,
      "loss": 12.6619,
      "step": 3762
    },
    {
      "epoch": 0.2049103891418731,
      "grad_norm": 0.8003066900229244,
      "learning_rate": 0.00019725948817252896,
      "loss": 12.6848,
      "step": 3763
    },
    {
      "epoch": 0.2049648431384561,
      "grad_norm": 0.707534024551744,
      "learning_rate": 0.00019725743749348743,
      "loss": 12.6277,
      "step": 3764
    },
    {
      "epoch": 0.2050192971350391,
      "grad_norm": 0.7530481312430138,
      "learning_rate": 0.00019725538605815573,
      "loss": 12.6478,
      "step": 3765
    },
    {
      "epoch": 0.20507375113162213,
      "grad_norm": 0.7858723007595092,
      "learning_rate": 0.00019725333386654987,
      "loss": 12.5266,
      "step": 3766
    },
    {
      "epoch": 0.20512820512820512,
      "grad_norm": 0.7056438702746797,
      "learning_rate": 0.00019725128091868576,
      "loss": 12.5997,
      "step": 3767
    },
    {
      "epoch": 0.20518265912478814,
      "grad_norm": 0.8041505336620463,
      "learning_rate": 0.00019724922721457938,
      "loss": 12.7159,
      "step": 3768
    },
    {
      "epoch": 0.20523711312137116,
      "grad_norm": 0.7398059661675201,
      "learning_rate": 0.00019724717275424673,
      "loss": 12.6444,
      "step": 3769
    },
    {
      "epoch": 0.20529156711795415,
      "grad_norm": 0.6868492068938551,
      "learning_rate": 0.00019724511753770374,
      "loss": 12.6264,
      "step": 3770
    },
    {
      "epoch": 0.20534602111453717,
      "grad_norm": 0.7634898216311271,
      "learning_rate": 0.0001972430615649664,
      "loss": 12.629,
      "step": 3771
    },
    {
      "epoch": 0.2054004751111202,
      "grad_norm": 0.8049756446724106,
      "learning_rate": 0.00019724100483605069,
      "loss": 12.7396,
      "step": 3772
    },
    {
      "epoch": 0.2054549291077032,
      "grad_norm": 0.7387584826388791,
      "learning_rate": 0.00019723894735097262,
      "loss": 12.5023,
      "step": 3773
    },
    {
      "epoch": 0.2055093831042862,
      "grad_norm": 0.6968592334321247,
      "learning_rate": 0.00019723688910974822,
      "loss": 12.6262,
      "step": 3774
    },
    {
      "epoch": 0.20556383710086923,
      "grad_norm": 0.7328486248615598,
      "learning_rate": 0.00019723483011239345,
      "loss": 12.6452,
      "step": 3775
    },
    {
      "epoch": 0.20561829109745222,
      "grad_norm": 0.8873669973092565,
      "learning_rate": 0.00019723277035892434,
      "loss": 12.4639,
      "step": 3776
    },
    {
      "epoch": 0.20567274509403524,
      "grad_norm": 0.800906708179299,
      "learning_rate": 0.00019723070984935687,
      "loss": 12.6349,
      "step": 3777
    },
    {
      "epoch": 0.20572719909061826,
      "grad_norm": 0.7353268544823438,
      "learning_rate": 0.00019722864858370714,
      "loss": 12.6264,
      "step": 3778
    },
    {
      "epoch": 0.20578165308720126,
      "grad_norm": 0.7185208506391326,
      "learning_rate": 0.00019722658656199112,
      "loss": 12.7086,
      "step": 3779
    },
    {
      "epoch": 0.20583610708378428,
      "grad_norm": 0.7791559699979477,
      "learning_rate": 0.00019722452378422484,
      "loss": 12.6312,
      "step": 3780
    },
    {
      "epoch": 0.2058905610803673,
      "grad_norm": 0.7679050894103957,
      "learning_rate": 0.00019722246025042438,
      "loss": 12.6771,
      "step": 3781
    },
    {
      "epoch": 0.20594501507695032,
      "grad_norm": 0.7874917981449171,
      "learning_rate": 0.00019722039596060573,
      "loss": 12.5139,
      "step": 3782
    },
    {
      "epoch": 0.2059994690735333,
      "grad_norm": 0.7991378424352726,
      "learning_rate": 0.00019721833091478498,
      "loss": 12.7016,
      "step": 3783
    },
    {
      "epoch": 0.20605392307011633,
      "grad_norm": 0.6959229181476858,
      "learning_rate": 0.0001972162651129782,
      "loss": 12.746,
      "step": 3784
    },
    {
      "epoch": 0.20610837706669935,
      "grad_norm": 0.8788473918123754,
      "learning_rate": 0.0001972141985552015,
      "loss": 12.6087,
      "step": 3785
    },
    {
      "epoch": 0.20616283106328234,
      "grad_norm": 0.7331671459525739,
      "learning_rate": 0.0001972121312414708,
      "loss": 12.3949,
      "step": 3786
    },
    {
      "epoch": 0.20621728505986536,
      "grad_norm": 0.7483671298906623,
      "learning_rate": 0.0001972100631718023,
      "loss": 12.6329,
      "step": 3787
    },
    {
      "epoch": 0.20627173905644838,
      "grad_norm": 0.7858065652580846,
      "learning_rate": 0.00019720799434621206,
      "loss": 12.512,
      "step": 3788
    },
    {
      "epoch": 0.20632619305303138,
      "grad_norm": 0.8633490504031776,
      "learning_rate": 0.00019720592476471613,
      "loss": 12.7427,
      "step": 3789
    },
    {
      "epoch": 0.2063806470496144,
      "grad_norm": 0.7954996022566327,
      "learning_rate": 0.00019720385442733063,
      "loss": 12.7359,
      "step": 3790
    },
    {
      "epoch": 0.20643510104619742,
      "grad_norm": 0.7359411612203388,
      "learning_rate": 0.00019720178333407166,
      "loss": 12.5897,
      "step": 3791
    },
    {
      "epoch": 0.2064895550427804,
      "grad_norm": 0.834858116614354,
      "learning_rate": 0.00019719971148495535,
      "loss": 12.7141,
      "step": 3792
    },
    {
      "epoch": 0.20654400903936343,
      "grad_norm": 0.8172693244285171,
      "learning_rate": 0.00019719763887999774,
      "loss": 12.6519,
      "step": 3793
    },
    {
      "epoch": 0.20659846303594645,
      "grad_norm": 0.7447324971926574,
      "learning_rate": 0.00019719556551921503,
      "loss": 12.7726,
      "step": 3794
    },
    {
      "epoch": 0.20665291703252944,
      "grad_norm": 0.790954552773982,
      "learning_rate": 0.00019719349140262326,
      "loss": 12.648,
      "step": 3795
    },
    {
      "epoch": 0.20670737102911246,
      "grad_norm": 0.7456368748812312,
      "learning_rate": 0.00019719141653023865,
      "loss": 12.5901,
      "step": 3796
    },
    {
      "epoch": 0.20676182502569548,
      "grad_norm": 0.7694652585477508,
      "learning_rate": 0.00019718934090207725,
      "loss": 12.5993,
      "step": 3797
    },
    {
      "epoch": 0.2068162790222785,
      "grad_norm": 0.6836342209484225,
      "learning_rate": 0.00019718726451815524,
      "loss": 12.6304,
      "step": 3798
    },
    {
      "epoch": 0.2068707330188615,
      "grad_norm": 0.7862179588431599,
      "learning_rate": 0.00019718518737848876,
      "loss": 12.5865,
      "step": 3799
    },
    {
      "epoch": 0.20692518701544452,
      "grad_norm": 0.7866118126659232,
      "learning_rate": 0.000197183109483094,
      "loss": 12.7151,
      "step": 3800
    },
    {
      "epoch": 0.20697964101202754,
      "grad_norm": 0.7387733279981449,
      "learning_rate": 0.00019718103083198705,
      "loss": 12.5917,
      "step": 3801
    },
    {
      "epoch": 0.20703409500861053,
      "grad_norm": 0.7673889903132298,
      "learning_rate": 0.0001971789514251841,
      "loss": 12.6231,
      "step": 3802
    },
    {
      "epoch": 0.20708854900519355,
      "grad_norm": 0.7998345043574764,
      "learning_rate": 0.00019717687126270133,
      "loss": 12.556,
      "step": 3803
    },
    {
      "epoch": 0.20714300300177657,
      "grad_norm": 0.6636026554154101,
      "learning_rate": 0.00019717479034455493,
      "loss": 12.5632,
      "step": 3804
    },
    {
      "epoch": 0.20719745699835956,
      "grad_norm": 0.7661527649226619,
      "learning_rate": 0.00019717270867076106,
      "loss": 12.5358,
      "step": 3805
    },
    {
      "epoch": 0.20725191099494258,
      "grad_norm": 0.8041290947239476,
      "learning_rate": 0.00019717062624133593,
      "loss": 12.779,
      "step": 3806
    },
    {
      "epoch": 0.2073063649915256,
      "grad_norm": 0.8016300510471224,
      "learning_rate": 0.00019716854305629569,
      "loss": 12.8196,
      "step": 3807
    },
    {
      "epoch": 0.2073608189881086,
      "grad_norm": 0.7844723455880096,
      "learning_rate": 0.00019716645911565657,
      "loss": 12.7492,
      "step": 3808
    },
    {
      "epoch": 0.20741527298469162,
      "grad_norm": 0.646050380941143,
      "learning_rate": 0.00019716437441943477,
      "loss": 12.6152,
      "step": 3809
    },
    {
      "epoch": 0.20746972698127464,
      "grad_norm": 0.7273011653324999,
      "learning_rate": 0.0001971622889676465,
      "loss": 12.5066,
      "step": 3810
    },
    {
      "epoch": 0.20752418097785763,
      "grad_norm": 0.6956035958581436,
      "learning_rate": 0.00019716020276030796,
      "loss": 12.5895,
      "step": 3811
    },
    {
      "epoch": 0.20757863497444065,
      "grad_norm": 0.68685569219118,
      "learning_rate": 0.00019715811579743543,
      "loss": 12.613,
      "step": 3812
    },
    {
      "epoch": 0.20763308897102367,
      "grad_norm": 0.6921888162595382,
      "learning_rate": 0.00019715602807904504,
      "loss": 12.6016,
      "step": 3813
    },
    {
      "epoch": 0.2076875429676067,
      "grad_norm": 0.7068906851597578,
      "learning_rate": 0.0001971539396051531,
      "loss": 12.5869,
      "step": 3814
    },
    {
      "epoch": 0.20774199696418968,
      "grad_norm": 0.6871538704273896,
      "learning_rate": 0.00019715185037577586,
      "loss": 12.5944,
      "step": 3815
    },
    {
      "epoch": 0.2077964509607727,
      "grad_norm": 0.7600040985296437,
      "learning_rate": 0.00019714976039092954,
      "loss": 12.5654,
      "step": 3816
    },
    {
      "epoch": 0.20785090495735573,
      "grad_norm": 0.6744308421520104,
      "learning_rate": 0.00019714766965063036,
      "loss": 12.5306,
      "step": 3817
    },
    {
      "epoch": 0.20790535895393872,
      "grad_norm": 0.7878816728043071,
      "learning_rate": 0.00019714557815489462,
      "loss": 12.793,
      "step": 3818
    },
    {
      "epoch": 0.20795981295052174,
      "grad_norm": 0.8099819105548215,
      "learning_rate": 0.0001971434859037386,
      "loss": 12.7286,
      "step": 3819
    },
    {
      "epoch": 0.20801426694710476,
      "grad_norm": 0.6456719660154859,
      "learning_rate": 0.0001971413928971785,
      "loss": 12.6326,
      "step": 3820
    },
    {
      "epoch": 0.20806872094368775,
      "grad_norm": 0.8364155602205413,
      "learning_rate": 0.00019713929913523068,
      "loss": 12.6638,
      "step": 3821
    },
    {
      "epoch": 0.20812317494027077,
      "grad_norm": 0.6785665463512941,
      "learning_rate": 0.00019713720461791135,
      "loss": 12.6289,
      "step": 3822
    },
    {
      "epoch": 0.2081776289368538,
      "grad_norm": 0.7327840517026947,
      "learning_rate": 0.00019713510934523683,
      "loss": 12.6189,
      "step": 3823
    },
    {
      "epoch": 0.20823208293343679,
      "grad_norm": 0.7269210860075299,
      "learning_rate": 0.00019713301331722343,
      "loss": 12.749,
      "step": 3824
    },
    {
      "epoch": 0.2082865369300198,
      "grad_norm": 0.7482373158985721,
      "learning_rate": 0.0001971309165338874,
      "loss": 12.6211,
      "step": 3825
    },
    {
      "epoch": 0.20834099092660283,
      "grad_norm": 0.6911904425702748,
      "learning_rate": 0.0001971288189952451,
      "loss": 12.5941,
      "step": 3826
    },
    {
      "epoch": 0.20839544492318582,
      "grad_norm": 0.8193143300411501,
      "learning_rate": 0.0001971267207013128,
      "loss": 12.6969,
      "step": 3827
    },
    {
      "epoch": 0.20844989891976884,
      "grad_norm": 0.711827074497607,
      "learning_rate": 0.00019712462165210684,
      "loss": 12.6972,
      "step": 3828
    },
    {
      "epoch": 0.20850435291635186,
      "grad_norm": 0.7307796661654593,
      "learning_rate": 0.00019712252184764354,
      "loss": 12.6634,
      "step": 3829
    },
    {
      "epoch": 0.20855880691293488,
      "grad_norm": 0.750460068624912,
      "learning_rate": 0.00019712042128793922,
      "loss": 12.6106,
      "step": 3830
    },
    {
      "epoch": 0.20861326090951787,
      "grad_norm": 0.7586157633791372,
      "learning_rate": 0.0001971183199730102,
      "loss": 12.8095,
      "step": 3831
    },
    {
      "epoch": 0.2086677149061009,
      "grad_norm": 0.7641058986350736,
      "learning_rate": 0.00019711621790287286,
      "loss": 12.6426,
      "step": 3832
    },
    {
      "epoch": 0.2087221689026839,
      "grad_norm": 0.7337260905000521,
      "learning_rate": 0.00019711411507754352,
      "loss": 12.4951,
      "step": 3833
    },
    {
      "epoch": 0.2087766228992669,
      "grad_norm": 0.7356825958053724,
      "learning_rate": 0.0001971120114970385,
      "loss": 12.4995,
      "step": 3834
    },
    {
      "epoch": 0.20883107689584993,
      "grad_norm": 0.6250276991757239,
      "learning_rate": 0.00019710990716137423,
      "loss": 12.5067,
      "step": 3835
    },
    {
      "epoch": 0.20888553089243295,
      "grad_norm": 0.7157164003208205,
      "learning_rate": 0.00019710780207056702,
      "loss": 12.5821,
      "step": 3836
    },
    {
      "epoch": 0.20893998488901594,
      "grad_norm": 0.781156813133304,
      "learning_rate": 0.00019710569622463327,
      "loss": 12.7614,
      "step": 3837
    },
    {
      "epoch": 0.20899443888559896,
      "grad_norm": 0.8273223237189452,
      "learning_rate": 0.00019710358962358933,
      "loss": 12.6423,
      "step": 3838
    },
    {
      "epoch": 0.20904889288218198,
      "grad_norm": 0.6768823229002183,
      "learning_rate": 0.0001971014822674516,
      "loss": 12.6379,
      "step": 3839
    },
    {
      "epoch": 0.20910334687876497,
      "grad_norm": 0.7061533212592964,
      "learning_rate": 0.00019709937415623646,
      "loss": 12.7136,
      "step": 3840
    },
    {
      "epoch": 0.209157800875348,
      "grad_norm": 0.686532646992562,
      "learning_rate": 0.00019709726528996027,
      "loss": 12.6175,
      "step": 3841
    },
    {
      "epoch": 0.209212254871931,
      "grad_norm": 0.7922752222688225,
      "learning_rate": 0.00019709515566863951,
      "loss": 12.7968,
      "step": 3842
    },
    {
      "epoch": 0.209266708868514,
      "grad_norm": 0.6801245682410086,
      "learning_rate": 0.00019709304529229053,
      "loss": 12.7147,
      "step": 3843
    },
    {
      "epoch": 0.20932116286509703,
      "grad_norm": 0.6555963750355286,
      "learning_rate": 0.0001970909341609297,
      "loss": 12.5695,
      "step": 3844
    },
    {
      "epoch": 0.20937561686168005,
      "grad_norm": 0.7913863987878447,
      "learning_rate": 0.00019708882227457354,
      "loss": 12.6226,
      "step": 3845
    },
    {
      "epoch": 0.20943007085826304,
      "grad_norm": 0.7239585315891942,
      "learning_rate": 0.00019708670963323842,
      "loss": 12.6389,
      "step": 3846
    },
    {
      "epoch": 0.20948452485484606,
      "grad_norm": 0.7520411193774028,
      "learning_rate": 0.00019708459623694072,
      "loss": 12.7562,
      "step": 3847
    },
    {
      "epoch": 0.20953897885142908,
      "grad_norm": 0.6390701288621873,
      "learning_rate": 0.00019708248208569695,
      "loss": 12.536,
      "step": 3848
    },
    {
      "epoch": 0.2095934328480121,
      "grad_norm": 0.7632552432870136,
      "learning_rate": 0.0001970803671795235,
      "loss": 12.7752,
      "step": 3849
    },
    {
      "epoch": 0.2096478868445951,
      "grad_norm": 0.6980795126129343,
      "learning_rate": 0.00019707825151843683,
      "loss": 12.6178,
      "step": 3850
    },
    {
      "epoch": 0.20970234084117811,
      "grad_norm": 0.6297619428404707,
      "learning_rate": 0.0001970761351024534,
      "loss": 12.6154,
      "step": 3851
    },
    {
      "epoch": 0.20975679483776113,
      "grad_norm": 0.716503403711023,
      "learning_rate": 0.0001970740179315897,
      "loss": 12.6761,
      "step": 3852
    },
    {
      "epoch": 0.20981124883434413,
      "grad_norm": 1.0613728402924163,
      "learning_rate": 0.0001970719000058621,
      "loss": 12.6331,
      "step": 3853
    },
    {
      "epoch": 0.20986570283092715,
      "grad_norm": 0.8591171392678476,
      "learning_rate": 0.00019706978132528718,
      "loss": 12.5784,
      "step": 3854
    },
    {
      "epoch": 0.20992015682751017,
      "grad_norm": 0.714119371116891,
      "learning_rate": 0.00019706766188988133,
      "loss": 12.6617,
      "step": 3855
    },
    {
      "epoch": 0.20997461082409316,
      "grad_norm": 0.7094014273501245,
      "learning_rate": 0.00019706554169966105,
      "loss": 12.6061,
      "step": 3856
    },
    {
      "epoch": 0.21002906482067618,
      "grad_norm": 0.9055915065824638,
      "learning_rate": 0.00019706342075464286,
      "loss": 12.688,
      "step": 3857
    },
    {
      "epoch": 0.2100835188172592,
      "grad_norm": 0.7266838551953319,
      "learning_rate": 0.00019706129905484323,
      "loss": 12.6648,
      "step": 3858
    },
    {
      "epoch": 0.2101379728138422,
      "grad_norm": 0.7897653004405011,
      "learning_rate": 0.00019705917660027867,
      "loss": 12.5621,
      "step": 3859
    },
    {
      "epoch": 0.21019242681042521,
      "grad_norm": 0.6577066357431925,
      "learning_rate": 0.00019705705339096566,
      "loss": 12.6753,
      "step": 3860
    },
    {
      "epoch": 0.21024688080700824,
      "grad_norm": 0.7631655878819147,
      "learning_rate": 0.00019705492942692074,
      "loss": 12.5231,
      "step": 3861
    },
    {
      "epoch": 0.21030133480359123,
      "grad_norm": 0.6823330943710011,
      "learning_rate": 0.00019705280470816043,
      "loss": 12.6195,
      "step": 3862
    },
    {
      "epoch": 0.21035578880017425,
      "grad_norm": 0.6802375107907133,
      "learning_rate": 0.0001970506792347012,
      "loss": 12.5155,
      "step": 3863
    },
    {
      "epoch": 0.21041024279675727,
      "grad_norm": 0.7288057731179113,
      "learning_rate": 0.00019704855300655964,
      "loss": 12.7704,
      "step": 3864
    },
    {
      "epoch": 0.2104646967933403,
      "grad_norm": 0.7067370312565961,
      "learning_rate": 0.00019704642602375223,
      "loss": 12.7041,
      "step": 3865
    },
    {
      "epoch": 0.21051915078992328,
      "grad_norm": 0.7328334988150504,
      "learning_rate": 0.00019704429828629554,
      "loss": 12.793,
      "step": 3866
    },
    {
      "epoch": 0.2105736047865063,
      "grad_norm": 0.743035582800159,
      "learning_rate": 0.00019704216979420612,
      "loss": 12.6079,
      "step": 3867
    },
    {
      "epoch": 0.21062805878308932,
      "grad_norm": 0.6952611456974587,
      "learning_rate": 0.00019704004054750055,
      "loss": 12.6263,
      "step": 3868
    },
    {
      "epoch": 0.21068251277967232,
      "grad_norm": 0.6967072401084503,
      "learning_rate": 0.0001970379105461953,
      "loss": 12.7435,
      "step": 3869
    },
    {
      "epoch": 0.21073696677625534,
      "grad_norm": 0.6810544243648045,
      "learning_rate": 0.00019703577979030698,
      "loss": 12.6692,
      "step": 3870
    },
    {
      "epoch": 0.21079142077283836,
      "grad_norm": 0.7173889510000903,
      "learning_rate": 0.0001970336482798522,
      "loss": 12.4846,
      "step": 3871
    },
    {
      "epoch": 0.21084587476942135,
      "grad_norm": 0.6766072230821139,
      "learning_rate": 0.0001970315160148475,
      "loss": 12.6103,
      "step": 3872
    },
    {
      "epoch": 0.21090032876600437,
      "grad_norm": 0.7340416998263765,
      "learning_rate": 0.00019702938299530942,
      "loss": 12.5992,
      "step": 3873
    },
    {
      "epoch": 0.2109547827625874,
      "grad_norm": 0.7250266860347498,
      "learning_rate": 0.00019702724922125462,
      "loss": 12.596,
      "step": 3874
    },
    {
      "epoch": 0.21100923675917038,
      "grad_norm": 0.7231440439405903,
      "learning_rate": 0.00019702511469269965,
      "loss": 12.6429,
      "step": 3875
    },
    {
      "epoch": 0.2110636907557534,
      "grad_norm": 0.7575607296218159,
      "learning_rate": 0.0001970229794096611,
      "loss": 12.637,
      "step": 3876
    },
    {
      "epoch": 0.21111814475233642,
      "grad_norm": 0.7560500121927132,
      "learning_rate": 0.0001970208433721556,
      "loss": 12.5314,
      "step": 3877
    },
    {
      "epoch": 0.21117259874891942,
      "grad_norm": 0.7222307087053694,
      "learning_rate": 0.00019701870658019976,
      "loss": 12.6902,
      "step": 3878
    },
    {
      "epoch": 0.21122705274550244,
      "grad_norm": 0.6647439342103886,
      "learning_rate": 0.0001970165690338102,
      "loss": 12.5535,
      "step": 3879
    },
    {
      "epoch": 0.21128150674208546,
      "grad_norm": 0.6745547574619641,
      "learning_rate": 0.00019701443073300349,
      "loss": 12.3277,
      "step": 3880
    },
    {
      "epoch": 0.21133596073866848,
      "grad_norm": 0.683311103597084,
      "learning_rate": 0.00019701229167779633,
      "loss": 12.6541,
      "step": 3881
    },
    {
      "epoch": 0.21139041473525147,
      "grad_norm": 0.7031825756420972,
      "learning_rate": 0.0001970101518682053,
      "loss": 12.3421,
      "step": 3882
    },
    {
      "epoch": 0.2114448687318345,
      "grad_norm": 0.6709281810865525,
      "learning_rate": 0.0001970080113042471,
      "loss": 12.6411,
      "step": 3883
    },
    {
      "epoch": 0.2114993227284175,
      "grad_norm": 0.7013708067172494,
      "learning_rate": 0.00019700586998593829,
      "loss": 12.622,
      "step": 3884
    },
    {
      "epoch": 0.2115537767250005,
      "grad_norm": 0.6970959113609223,
      "learning_rate": 0.0001970037279132956,
      "loss": 12.6072,
      "step": 3885
    },
    {
      "epoch": 0.21160823072158352,
      "grad_norm": 0.7172589159645827,
      "learning_rate": 0.00019700158508633564,
      "loss": 12.6771,
      "step": 3886
    },
    {
      "epoch": 0.21166268471816654,
      "grad_norm": 0.6968691606849642,
      "learning_rate": 0.00019699944150507507,
      "loss": 12.6998,
      "step": 3887
    },
    {
      "epoch": 0.21171713871474954,
      "grad_norm": 0.7427929804273878,
      "learning_rate": 0.0001969972971695306,
      "loss": 12.6571,
      "step": 3888
    },
    {
      "epoch": 0.21177159271133256,
      "grad_norm": 0.7081419720758875,
      "learning_rate": 0.00019699515207971885,
      "loss": 12.7094,
      "step": 3889
    },
    {
      "epoch": 0.21182604670791558,
      "grad_norm": 0.826287521696027,
      "learning_rate": 0.00019699300623565657,
      "loss": 12.5314,
      "step": 3890
    },
    {
      "epoch": 0.21188050070449857,
      "grad_norm": 0.7260506698265387,
      "learning_rate": 0.00019699085963736042,
      "loss": 12.6649,
      "step": 3891
    },
    {
      "epoch": 0.2119349547010816,
      "grad_norm": 0.8452203006842348,
      "learning_rate": 0.00019698871228484704,
      "loss": 12.6858,
      "step": 3892
    },
    {
      "epoch": 0.2119894086976646,
      "grad_norm": 0.7775963563602492,
      "learning_rate": 0.00019698656417813318,
      "loss": 12.5169,
      "step": 3893
    },
    {
      "epoch": 0.2120438626942476,
      "grad_norm": 0.7891925501823804,
      "learning_rate": 0.00019698441531723553,
      "loss": 12.7274,
      "step": 3894
    },
    {
      "epoch": 0.21209831669083062,
      "grad_norm": 0.6844535277171849,
      "learning_rate": 0.0001969822657021708,
      "loss": 12.494,
      "step": 3895
    },
    {
      "epoch": 0.21215277068741364,
      "grad_norm": 0.7594849342870383,
      "learning_rate": 0.0001969801153329557,
      "loss": 12.6393,
      "step": 3896
    },
    {
      "epoch": 0.21220722468399666,
      "grad_norm": 0.9077815788828331,
      "learning_rate": 0.000196977964209607,
      "loss": 12.6181,
      "step": 3897
    },
    {
      "epoch": 0.21226167868057966,
      "grad_norm": 0.7621298337666311,
      "learning_rate": 0.00019697581233214134,
      "loss": 12.6645,
      "step": 3898
    },
    {
      "epoch": 0.21231613267716268,
      "grad_norm": 0.8366099030579992,
      "learning_rate": 0.00019697365970057553,
      "loss": 12.6583,
      "step": 3899
    },
    {
      "epoch": 0.2123705866737457,
      "grad_norm": 0.7217507276712453,
      "learning_rate": 0.00019697150631492626,
      "loss": 12.6469,
      "step": 3900
    },
    {
      "epoch": 0.2124250406703287,
      "grad_norm": 0.8339395971536379,
      "learning_rate": 0.00019696935217521032,
      "loss": 12.7546,
      "step": 3901
    },
    {
      "epoch": 0.2124794946669117,
      "grad_norm": 0.7181593477883866,
      "learning_rate": 0.00019696719728144442,
      "loss": 12.5382,
      "step": 3902
    },
    {
      "epoch": 0.21253394866349473,
      "grad_norm": 0.7617599944487994,
      "learning_rate": 0.0001969650416336453,
      "loss": 12.4544,
      "step": 3903
    },
    {
      "epoch": 0.21258840266007772,
      "grad_norm": 0.7392836979080841,
      "learning_rate": 0.0001969628852318298,
      "loss": 12.5215,
      "step": 3904
    },
    {
      "epoch": 0.21264285665666074,
      "grad_norm": 0.7660976492049502,
      "learning_rate": 0.00019696072807601464,
      "loss": 12.5948,
      "step": 3905
    },
    {
      "epoch": 0.21269731065324377,
      "grad_norm": 0.7933437707513886,
      "learning_rate": 0.0001969585701662166,
      "loss": 12.7446,
      "step": 3906
    },
    {
      "epoch": 0.21275176464982676,
      "grad_norm": 0.7222756885902184,
      "learning_rate": 0.00019695641150245242,
      "loss": 12.5878,
      "step": 3907
    },
    {
      "epoch": 0.21280621864640978,
      "grad_norm": 0.7396778956759119,
      "learning_rate": 0.000196954252084739,
      "loss": 12.6364,
      "step": 3908
    },
    {
      "epoch": 0.2128606726429928,
      "grad_norm": 0.7406513542494016,
      "learning_rate": 0.000196952091913093,
      "loss": 12.4174,
      "step": 3909
    },
    {
      "epoch": 0.2129151266395758,
      "grad_norm": 0.801324775162043,
      "learning_rate": 0.00019694993098753126,
      "loss": 12.6281,
      "step": 3910
    },
    {
      "epoch": 0.2129695806361588,
      "grad_norm": 0.7922004514295725,
      "learning_rate": 0.0001969477693080706,
      "loss": 12.711,
      "step": 3911
    },
    {
      "epoch": 0.21302403463274183,
      "grad_norm": 0.7192166126667284,
      "learning_rate": 0.00019694560687472787,
      "loss": 12.713,
      "step": 3912
    },
    {
      "epoch": 0.21307848862932482,
      "grad_norm": 0.8280048325074828,
      "learning_rate": 0.00019694344368751984,
      "loss": 12.7116,
      "step": 3913
    },
    {
      "epoch": 0.21313294262590785,
      "grad_norm": 0.7244318785527158,
      "learning_rate": 0.00019694127974646334,
      "loss": 12.6594,
      "step": 3914
    },
    {
      "epoch": 0.21318739662249087,
      "grad_norm": 0.8168065098975937,
      "learning_rate": 0.00019693911505157515,
      "loss": 12.6092,
      "step": 3915
    },
    {
      "epoch": 0.21324185061907389,
      "grad_norm": 0.6701580431033446,
      "learning_rate": 0.00019693694960287218,
      "loss": 12.5596,
      "step": 3916
    },
    {
      "epoch": 0.21329630461565688,
      "grad_norm": 0.7177138044919541,
      "learning_rate": 0.0001969347834003712,
      "loss": 12.5675,
      "step": 3917
    },
    {
      "epoch": 0.2133507586122399,
      "grad_norm": 0.820418801242099,
      "learning_rate": 0.00019693261644408908,
      "loss": 12.5892,
      "step": 3918
    },
    {
      "epoch": 0.21340521260882292,
      "grad_norm": 0.8034108120593468,
      "learning_rate": 0.00019693044873404274,
      "loss": 12.7667,
      "step": 3919
    },
    {
      "epoch": 0.2134596666054059,
      "grad_norm": 0.7710641017405725,
      "learning_rate": 0.00019692828027024893,
      "loss": 12.7673,
      "step": 3920
    },
    {
      "epoch": 0.21351412060198893,
      "grad_norm": 0.6792728233039044,
      "learning_rate": 0.00019692611105272457,
      "loss": 12.6426,
      "step": 3921
    },
    {
      "epoch": 0.21356857459857195,
      "grad_norm": 0.8051947786505732,
      "learning_rate": 0.0001969239410814865,
      "loss": 12.6935,
      "step": 3922
    },
    {
      "epoch": 0.21362302859515495,
      "grad_norm": 0.8199349179966275,
      "learning_rate": 0.00019692177035655163,
      "loss": 12.5231,
      "step": 3923
    },
    {
      "epoch": 0.21367748259173797,
      "grad_norm": 0.6997119387345971,
      "learning_rate": 0.00019691959887793684,
      "loss": 12.6468,
      "step": 3924
    },
    {
      "epoch": 0.213731936588321,
      "grad_norm": 0.8372167910644972,
      "learning_rate": 0.00019691742664565895,
      "loss": 12.6043,
      "step": 3925
    },
    {
      "epoch": 0.21378639058490398,
      "grad_norm": 0.6962495433547871,
      "learning_rate": 0.0001969152536597349,
      "loss": 12.684,
      "step": 3926
    },
    {
      "epoch": 0.213840844581487,
      "grad_norm": 0.7266854442111711,
      "learning_rate": 0.00019691307992018161,
      "loss": 12.4952,
      "step": 3927
    },
    {
      "epoch": 0.21389529857807002,
      "grad_norm": 0.7655950665136835,
      "learning_rate": 0.00019691090542701595,
      "loss": 12.4292,
      "step": 3928
    },
    {
      "epoch": 0.213949752574653,
      "grad_norm": 0.7277789599591761,
      "learning_rate": 0.00019690873018025483,
      "loss": 12.6611,
      "step": 3929
    },
    {
      "epoch": 0.21400420657123603,
      "grad_norm": 0.8271872279766458,
      "learning_rate": 0.0001969065541799152,
      "loss": 12.6395,
      "step": 3930
    },
    {
      "epoch": 0.21405866056781905,
      "grad_norm": 0.661706779368455,
      "learning_rate": 0.00019690437742601394,
      "loss": 12.6383,
      "step": 3931
    },
    {
      "epoch": 0.21411311456440207,
      "grad_norm": 0.7064738109764515,
      "learning_rate": 0.00019690219991856797,
      "loss": 12.6467,
      "step": 3932
    },
    {
      "epoch": 0.21416756856098507,
      "grad_norm": 0.7982907102311182,
      "learning_rate": 0.00019690002165759424,
      "loss": 12.5564,
      "step": 3933
    },
    {
      "epoch": 0.2142220225575681,
      "grad_norm": 0.7074857690670541,
      "learning_rate": 0.00019689784264310972,
      "loss": 12.5441,
      "step": 3934
    },
    {
      "epoch": 0.2142764765541511,
      "grad_norm": 0.7952675669800652,
      "learning_rate": 0.00019689566287513132,
      "loss": 12.7703,
      "step": 3935
    },
    {
      "epoch": 0.2143309305507341,
      "grad_norm": 0.7824586864295231,
      "learning_rate": 0.00019689348235367598,
      "loss": 12.7611,
      "step": 3936
    },
    {
      "epoch": 0.21438538454731712,
      "grad_norm": 0.7943644437390582,
      "learning_rate": 0.00019689130107876067,
      "loss": 12.6341,
      "step": 3937
    },
    {
      "epoch": 0.21443983854390014,
      "grad_norm": 0.7623897601744621,
      "learning_rate": 0.00019688911905040238,
      "loss": 12.6317,
      "step": 3938
    },
    {
      "epoch": 0.21449429254048313,
      "grad_norm": 0.728279967862069,
      "learning_rate": 0.00019688693626861804,
      "loss": 12.5935,
      "step": 3939
    },
    {
      "epoch": 0.21454874653706615,
      "grad_norm": 0.7511841158265858,
      "learning_rate": 0.00019688475273342464,
      "loss": 12.505,
      "step": 3940
    },
    {
      "epoch": 0.21460320053364917,
      "grad_norm": 0.7418238611000724,
      "learning_rate": 0.00019688256844483914,
      "loss": 12.565,
      "step": 3941
    },
    {
      "epoch": 0.21465765453023217,
      "grad_norm": 0.718602048206933,
      "learning_rate": 0.00019688038340287856,
      "loss": 12.7313,
      "step": 3942
    },
    {
      "epoch": 0.2147121085268152,
      "grad_norm": 0.6239293268350437,
      "learning_rate": 0.00019687819760755987,
      "loss": 12.6087,
      "step": 3943
    },
    {
      "epoch": 0.2147665625233982,
      "grad_norm": 0.7081825399431875,
      "learning_rate": 0.00019687601105890004,
      "loss": 12.5982,
      "step": 3944
    },
    {
      "epoch": 0.2148210165199812,
      "grad_norm": 0.7000307810121966,
      "learning_rate": 0.0001968738237569161,
      "loss": 12.5143,
      "step": 3945
    },
    {
      "epoch": 0.21487547051656422,
      "grad_norm": 0.7241333911326127,
      "learning_rate": 0.0001968716357016251,
      "loss": 12.6107,
      "step": 3946
    },
    {
      "epoch": 0.21492992451314724,
      "grad_norm": 0.7073929167858216,
      "learning_rate": 0.00019686944689304402,
      "loss": 12.6784,
      "step": 3947
    },
    {
      "epoch": 0.21498437850973026,
      "grad_norm": 0.6963246525209683,
      "learning_rate": 0.00019686725733118982,
      "loss": 12.6063,
      "step": 3948
    },
    {
      "epoch": 0.21503883250631325,
      "grad_norm": 0.696928836554792,
      "learning_rate": 0.00019686506701607965,
      "loss": 12.5421,
      "step": 3949
    },
    {
      "epoch": 0.21509328650289627,
      "grad_norm": 0.7947151519515174,
      "learning_rate": 0.00019686287594773043,
      "loss": 12.7203,
      "step": 3950
    },
    {
      "epoch": 0.2151477404994793,
      "grad_norm": 0.6941464137226102,
      "learning_rate": 0.00019686068412615927,
      "loss": 12.651,
      "step": 3951
    },
    {
      "epoch": 0.2152021944960623,
      "grad_norm": 0.7422280709843154,
      "learning_rate": 0.00019685849155138315,
      "loss": 12.6142,
      "step": 3952
    },
    {
      "epoch": 0.2152566484926453,
      "grad_norm": 0.6568159068976495,
      "learning_rate": 0.00019685629822341919,
      "loss": 12.4753,
      "step": 3953
    },
    {
      "epoch": 0.21531110248922833,
      "grad_norm": 0.7266034209404038,
      "learning_rate": 0.0001968541041422844,
      "loss": 12.7158,
      "step": 3954
    },
    {
      "epoch": 0.21536555648581132,
      "grad_norm": 0.6945221783504558,
      "learning_rate": 0.00019685190930799585,
      "loss": 12.6453,
      "step": 3955
    },
    {
      "epoch": 0.21542001048239434,
      "grad_norm": 0.7526163319281612,
      "learning_rate": 0.00019684971372057063,
      "loss": 12.6438,
      "step": 3956
    },
    {
      "epoch": 0.21547446447897736,
      "grad_norm": 0.7752381893997414,
      "learning_rate": 0.00019684751738002575,
      "loss": 12.6173,
      "step": 3957
    },
    {
      "epoch": 0.21552891847556035,
      "grad_norm": 0.6891814177801274,
      "learning_rate": 0.00019684532028637836,
      "loss": 12.5664,
      "step": 3958
    },
    {
      "epoch": 0.21558337247214338,
      "grad_norm": 0.7040966702280157,
      "learning_rate": 0.0001968431224396455,
      "loss": 12.6211,
      "step": 3959
    },
    {
      "epoch": 0.2156378264687264,
      "grad_norm": 0.6086700373400811,
      "learning_rate": 0.00019684092383984433,
      "loss": 12.5274,
      "step": 3960
    },
    {
      "epoch": 0.2156922804653094,
      "grad_norm": 0.7380800816251231,
      "learning_rate": 0.00019683872448699184,
      "loss": 12.683,
      "step": 3961
    },
    {
      "epoch": 0.2157467344618924,
      "grad_norm": 0.7400584232257849,
      "learning_rate": 0.00019683652438110523,
      "loss": 12.6166,
      "step": 3962
    },
    {
      "epoch": 0.21580118845847543,
      "grad_norm": 0.756629350492766,
      "learning_rate": 0.00019683432352220158,
      "loss": 12.6753,
      "step": 3963
    },
    {
      "epoch": 0.21585564245505845,
      "grad_norm": 0.6501914484379949,
      "learning_rate": 0.00019683212191029794,
      "loss": 12.4404,
      "step": 3964
    },
    {
      "epoch": 0.21591009645164144,
      "grad_norm": 0.7284031309302922,
      "learning_rate": 0.0001968299195454115,
      "loss": 12.5702,
      "step": 3965
    },
    {
      "epoch": 0.21596455044822446,
      "grad_norm": 0.7840331234544401,
      "learning_rate": 0.00019682771642755937,
      "loss": 12.8293,
      "step": 3966
    },
    {
      "epoch": 0.21601900444480748,
      "grad_norm": 0.7101177614963934,
      "learning_rate": 0.00019682551255675867,
      "loss": 12.7148,
      "step": 3967
    },
    {
      "epoch": 0.21607345844139048,
      "grad_norm": 0.7470213722881334,
      "learning_rate": 0.00019682330793302657,
      "loss": 12.7036,
      "step": 3968
    },
    {
      "epoch": 0.2161279124379735,
      "grad_norm": 1.004377002918978,
      "learning_rate": 0.00019682110255638018,
      "loss": 12.6216,
      "step": 3969
    },
    {
      "epoch": 0.21618236643455652,
      "grad_norm": 0.7040489342355073,
      "learning_rate": 0.00019681889642683668,
      "loss": 12.648,
      "step": 3970
    },
    {
      "epoch": 0.2162368204311395,
      "grad_norm": 0.7142888356405561,
      "learning_rate": 0.00019681668954441315,
      "loss": 12.5634,
      "step": 3971
    },
    {
      "epoch": 0.21629127442772253,
      "grad_norm": 0.7280247286864808,
      "learning_rate": 0.00019681448190912682,
      "loss": 12.5871,
      "step": 3972
    },
    {
      "epoch": 0.21634572842430555,
      "grad_norm": 0.7804693598806084,
      "learning_rate": 0.00019681227352099487,
      "loss": 12.7291,
      "step": 3973
    },
    {
      "epoch": 0.21640018242088854,
      "grad_norm": 0.7894803374028877,
      "learning_rate": 0.00019681006438003443,
      "loss": 12.6476,
      "step": 3974
    },
    {
      "epoch": 0.21645463641747156,
      "grad_norm": 0.6976484929957576,
      "learning_rate": 0.00019680785448626274,
      "loss": 12.6018,
      "step": 3975
    },
    {
      "epoch": 0.21650909041405458,
      "grad_norm": 0.6978787462272699,
      "learning_rate": 0.00019680564383969687,
      "loss": 12.651,
      "step": 3976
    },
    {
      "epoch": 0.21656354441063758,
      "grad_norm": 0.7023747519995679,
      "learning_rate": 0.00019680343244035412,
      "loss": 12.7279,
      "step": 3977
    },
    {
      "epoch": 0.2166179984072206,
      "grad_norm": 0.7581605914443555,
      "learning_rate": 0.00019680122028825162,
      "loss": 12.5454,
      "step": 3978
    },
    {
      "epoch": 0.21667245240380362,
      "grad_norm": 0.8027276031510364,
      "learning_rate": 0.00019679900738340663,
      "loss": 12.6674,
      "step": 3979
    },
    {
      "epoch": 0.2167269064003866,
      "grad_norm": 0.7443417644929946,
      "learning_rate": 0.0001967967937258363,
      "loss": 12.694,
      "step": 3980
    },
    {
      "epoch": 0.21678136039696963,
      "grad_norm": 0.7614531973040706,
      "learning_rate": 0.00019679457931555787,
      "loss": 12.7878,
      "step": 3981
    },
    {
      "epoch": 0.21683581439355265,
      "grad_norm": 0.7739638001554833,
      "learning_rate": 0.00019679236415258856,
      "loss": 12.672,
      "step": 3982
    },
    {
      "epoch": 0.21689026839013567,
      "grad_norm": 0.7819656619703798,
      "learning_rate": 0.0001967901482369456,
      "loss": 12.5756,
      "step": 3983
    },
    {
      "epoch": 0.21694472238671866,
      "grad_norm": 0.7461819005859127,
      "learning_rate": 0.00019678793156864622,
      "loss": 12.7048,
      "step": 3984
    },
    {
      "epoch": 0.21699917638330168,
      "grad_norm": 0.7125263632988991,
      "learning_rate": 0.00019678571414770763,
      "loss": 12.5498,
      "step": 3985
    },
    {
      "epoch": 0.2170536303798847,
      "grad_norm": 0.7633302060511443,
      "learning_rate": 0.0001967834959741471,
      "loss": 12.5501,
      "step": 3986
    },
    {
      "epoch": 0.2171080843764677,
      "grad_norm": 0.6537109819979228,
      "learning_rate": 0.0001967812770479819,
      "loss": 12.5498,
      "step": 3987
    },
    {
      "epoch": 0.21716253837305072,
      "grad_norm": 0.6956757640626371,
      "learning_rate": 0.00019677905736922928,
      "loss": 12.6805,
      "step": 3988
    },
    {
      "epoch": 0.21721699236963374,
      "grad_norm": 0.9038073036106121,
      "learning_rate": 0.00019677683693790642,
      "loss": 12.5616,
      "step": 3989
    },
    {
      "epoch": 0.21727144636621673,
      "grad_norm": 0.7845108311106693,
      "learning_rate": 0.0001967746157540307,
      "loss": 12.6692,
      "step": 3990
    },
    {
      "epoch": 0.21732590036279975,
      "grad_norm": 0.865577954805613,
      "learning_rate": 0.00019677239381761928,
      "loss": 12.5767,
      "step": 3991
    },
    {
      "epoch": 0.21738035435938277,
      "grad_norm": 0.6931754652824569,
      "learning_rate": 0.00019677017112868956,
      "loss": 12.5629,
      "step": 3992
    },
    {
      "epoch": 0.21743480835596576,
      "grad_norm": 0.7771496190857541,
      "learning_rate": 0.00019676794768725873,
      "loss": 12.6854,
      "step": 3993
    },
    {
      "epoch": 0.21748926235254878,
      "grad_norm": 0.7092020476329501,
      "learning_rate": 0.0001967657234933441,
      "loss": 12.6091,
      "step": 3994
    },
    {
      "epoch": 0.2175437163491318,
      "grad_norm": 0.693179443968465,
      "learning_rate": 0.000196763498546963,
      "loss": 12.7026,
      "step": 3995
    },
    {
      "epoch": 0.2175981703457148,
      "grad_norm": 0.731434066275842,
      "learning_rate": 0.00019676127284813267,
      "loss": 12.6795,
      "step": 3996
    },
    {
      "epoch": 0.21765262434229782,
      "grad_norm": 0.7032078557389057,
      "learning_rate": 0.0001967590463968705,
      "loss": 12.6302,
      "step": 3997
    },
    {
      "epoch": 0.21770707833888084,
      "grad_norm": 0.635970585227238,
      "learning_rate": 0.00019675681919319372,
      "loss": 12.6628,
      "step": 3998
    },
    {
      "epoch": 0.21776153233546386,
      "grad_norm": 0.8211180272514582,
      "learning_rate": 0.00019675459123711972,
      "loss": 12.8123,
      "step": 3999
    },
    {
      "epoch": 0.21781598633204685,
      "grad_norm": 0.7603117513168859,
      "learning_rate": 0.00019675236252866577,
      "loss": 12.578,
      "step": 4000
    },
    {
      "epoch": 0.21787044032862987,
      "grad_norm": 0.7594642151567548,
      "learning_rate": 0.00019675013306784923,
      "loss": 12.5305,
      "step": 4001
    },
    {
      "epoch": 0.2179248943252129,
      "grad_norm": 0.7699914028970279,
      "learning_rate": 0.00019674790285468746,
      "loss": 12.7227,
      "step": 4002
    },
    {
      "epoch": 0.21797934832179588,
      "grad_norm": 0.7397614162642909,
      "learning_rate": 0.00019674567188919775,
      "loss": 12.678,
      "step": 4003
    },
    {
      "epoch": 0.2180338023183789,
      "grad_norm": 0.7175728299201021,
      "learning_rate": 0.00019674344017139744,
      "loss": 12.6919,
      "step": 4004
    },
    {
      "epoch": 0.21808825631496193,
      "grad_norm": 0.7697025967962892,
      "learning_rate": 0.00019674120770130394,
      "loss": 12.7325,
      "step": 4005
    },
    {
      "epoch": 0.21814271031154492,
      "grad_norm": 0.7006179379055654,
      "learning_rate": 0.00019673897447893462,
      "loss": 12.5337,
      "step": 4006
    },
    {
      "epoch": 0.21819716430812794,
      "grad_norm": 0.7105004628449897,
      "learning_rate": 0.00019673674050430678,
      "loss": 12.677,
      "step": 4007
    },
    {
      "epoch": 0.21825161830471096,
      "grad_norm": 0.6571087553060003,
      "learning_rate": 0.0001967345057774378,
      "loss": 12.6363,
      "step": 4008
    },
    {
      "epoch": 0.21830607230129395,
      "grad_norm": 0.7093641040329042,
      "learning_rate": 0.00019673227029834512,
      "loss": 12.5658,
      "step": 4009
    },
    {
      "epoch": 0.21836052629787697,
      "grad_norm": 0.6380887380541235,
      "learning_rate": 0.00019673003406704605,
      "loss": 12.5858,
      "step": 4010
    },
    {
      "epoch": 0.21841498029446,
      "grad_norm": 0.7916333527955144,
      "learning_rate": 0.00019672779708355804,
      "loss": 12.7207,
      "step": 4011
    },
    {
      "epoch": 0.21846943429104299,
      "grad_norm": 0.6850591486656107,
      "learning_rate": 0.00019672555934789845,
      "loss": 12.6482,
      "step": 4012
    },
    {
      "epoch": 0.218523888287626,
      "grad_norm": 0.7198343448962357,
      "learning_rate": 0.0001967233208600847,
      "loss": 12.6282,
      "step": 4013
    },
    {
      "epoch": 0.21857834228420903,
      "grad_norm": 0.7496108215300044,
      "learning_rate": 0.0001967210816201342,
      "loss": 12.6532,
      "step": 4014
    },
    {
      "epoch": 0.21863279628079205,
      "grad_norm": 0.7301845943741914,
      "learning_rate": 0.0001967188416280643,
      "loss": 12.5878,
      "step": 4015
    },
    {
      "epoch": 0.21868725027737504,
      "grad_norm": 0.7480537564589492,
      "learning_rate": 0.0001967166008838925,
      "loss": 12.7344,
      "step": 4016
    },
    {
      "epoch": 0.21874170427395806,
      "grad_norm": 0.7040530822679284,
      "learning_rate": 0.00019671435938763615,
      "loss": 12.6365,
      "step": 4017
    },
    {
      "epoch": 0.21879615827054108,
      "grad_norm": 0.729618764939438,
      "learning_rate": 0.0001967121171393128,
      "loss": 12.6541,
      "step": 4018
    },
    {
      "epoch": 0.21885061226712407,
      "grad_norm": 0.7578991783646115,
      "learning_rate": 0.00019670987413893974,
      "loss": 12.7008,
      "step": 4019
    },
    {
      "epoch": 0.2189050662637071,
      "grad_norm": 0.7139122569181902,
      "learning_rate": 0.00019670763038653452,
      "loss": 12.6024,
      "step": 4020
    },
    {
      "epoch": 0.2189595202602901,
      "grad_norm": 0.7190580028658345,
      "learning_rate": 0.00019670538588211455,
      "loss": 12.6473,
      "step": 4021
    },
    {
      "epoch": 0.2190139742568731,
      "grad_norm": 0.7332441857723779,
      "learning_rate": 0.00019670314062569726,
      "loss": 12.6767,
      "step": 4022
    },
    {
      "epoch": 0.21906842825345613,
      "grad_norm": 0.7084478878468786,
      "learning_rate": 0.00019670089461730012,
      "loss": 12.657,
      "step": 4023
    },
    {
      "epoch": 0.21912288225003915,
      "grad_norm": 0.7499768445803511,
      "learning_rate": 0.00019669864785694063,
      "loss": 12.5707,
      "step": 4024
    },
    {
      "epoch": 0.21917733624662214,
      "grad_norm": 0.8299453060159917,
      "learning_rate": 0.00019669640034463624,
      "loss": 12.8349,
      "step": 4025
    },
    {
      "epoch": 0.21923179024320516,
      "grad_norm": 0.7774346746945877,
      "learning_rate": 0.00019669415208040442,
      "loss": 12.5669,
      "step": 4026
    },
    {
      "epoch": 0.21928624423978818,
      "grad_norm": 0.7413977284793508,
      "learning_rate": 0.00019669190306426264,
      "loss": 12.407,
      "step": 4027
    },
    {
      "epoch": 0.21934069823637117,
      "grad_norm": 0.7179477776342924,
      "learning_rate": 0.00019668965329622845,
      "loss": 12.6059,
      "step": 4028
    },
    {
      "epoch": 0.2193951522329542,
      "grad_norm": 0.7544261333086814,
      "learning_rate": 0.00019668740277631926,
      "loss": 12.5511,
      "step": 4029
    },
    {
      "epoch": 0.21944960622953721,
      "grad_norm": 0.6788519137175846,
      "learning_rate": 0.0001966851515045526,
      "loss": 12.6608,
      "step": 4030
    },
    {
      "epoch": 0.21950406022612023,
      "grad_norm": 0.7230530691838619,
      "learning_rate": 0.00019668289948094602,
      "loss": 12.6335,
      "step": 4031
    },
    {
      "epoch": 0.21955851422270323,
      "grad_norm": 0.7056961235171023,
      "learning_rate": 0.000196680646705517,
      "loss": 12.673,
      "step": 4032
    },
    {
      "epoch": 0.21961296821928625,
      "grad_norm": 0.7378250013137789,
      "learning_rate": 0.00019667839317828305,
      "loss": 12.7235,
      "step": 4033
    },
    {
      "epoch": 0.21966742221586927,
      "grad_norm": 0.7101142061368517,
      "learning_rate": 0.00019667613889926168,
      "loss": 12.6265,
      "step": 4034
    },
    {
      "epoch": 0.21972187621245226,
      "grad_norm": 0.708434740030311,
      "learning_rate": 0.00019667388386847048,
      "loss": 12.6863,
      "step": 4035
    },
    {
      "epoch": 0.21977633020903528,
      "grad_norm": 0.7302184396771872,
      "learning_rate": 0.00019667162808592695,
      "loss": 12.4923,
      "step": 4036
    },
    {
      "epoch": 0.2198307842056183,
      "grad_norm": 0.9297099830656682,
      "learning_rate": 0.0001966693715516486,
      "loss": 12.5281,
      "step": 4037
    },
    {
      "epoch": 0.2198852382022013,
      "grad_norm": 0.7556286735752703,
      "learning_rate": 0.00019666711426565302,
      "loss": 12.6344,
      "step": 4038
    },
    {
      "epoch": 0.21993969219878431,
      "grad_norm": 0.7837233340630051,
      "learning_rate": 0.00019666485622795774,
      "loss": 12.5989,
      "step": 4039
    },
    {
      "epoch": 0.21999414619536733,
      "grad_norm": 0.7453415181216037,
      "learning_rate": 0.00019666259743858033,
      "loss": 12.6815,
      "step": 4040
    },
    {
      "epoch": 0.22004860019195033,
      "grad_norm": 0.800459321357,
      "learning_rate": 0.00019666033789753832,
      "loss": 12.6644,
      "step": 4041
    },
    {
      "epoch": 0.22010305418853335,
      "grad_norm": 0.7277918752524827,
      "learning_rate": 0.00019665807760484936,
      "loss": 12.7014,
      "step": 4042
    },
    {
      "epoch": 0.22015750818511637,
      "grad_norm": 0.7666164066093426,
      "learning_rate": 0.00019665581656053095,
      "loss": 12.5904,
      "step": 4043
    },
    {
      "epoch": 0.22021196218169936,
      "grad_norm": 0.8457032281116588,
      "learning_rate": 0.00019665355476460072,
      "loss": 12.6473,
      "step": 4044
    },
    {
      "epoch": 0.22026641617828238,
      "grad_norm": 0.8591039778091485,
      "learning_rate": 0.00019665129221707625,
      "loss": 12.6753,
      "step": 4045
    },
    {
      "epoch": 0.2203208701748654,
      "grad_norm": 0.8961209104006046,
      "learning_rate": 0.0001966490289179751,
      "loss": 12.3711,
      "step": 4046
    },
    {
      "epoch": 0.2203753241714484,
      "grad_norm": 0.703577480574146,
      "learning_rate": 0.0001966467648673149,
      "loss": 12.4597,
      "step": 4047
    },
    {
      "epoch": 0.22042977816803141,
      "grad_norm": 0.7176771691091773,
      "learning_rate": 0.00019664450006511323,
      "loss": 12.4849,
      "step": 4048
    },
    {
      "epoch": 0.22048423216461444,
      "grad_norm": 0.9445294769580541,
      "learning_rate": 0.00019664223451138776,
      "loss": 12.6456,
      "step": 4049
    },
    {
      "epoch": 0.22053868616119746,
      "grad_norm": 0.79355294350249,
      "learning_rate": 0.00019663996820615604,
      "loss": 12.5788,
      "step": 4050
    },
    {
      "epoch": 0.22059314015778045,
      "grad_norm": 0.7021683884325197,
      "learning_rate": 0.0001966377011494357,
      "loss": 12.628,
      "step": 4051
    },
    {
      "epoch": 0.22064759415436347,
      "grad_norm": 0.8065361759591646,
      "learning_rate": 0.00019663543334124444,
      "loss": 12.6926,
      "step": 4052
    },
    {
      "epoch": 0.2207020481509465,
      "grad_norm": 0.7289818176356541,
      "learning_rate": 0.00019663316478159984,
      "loss": 12.627,
      "step": 4053
    },
    {
      "epoch": 0.22075650214752948,
      "grad_norm": 0.6840374913718438,
      "learning_rate": 0.0001966308954705195,
      "loss": 12.639,
      "step": 4054
    },
    {
      "epoch": 0.2208109561441125,
      "grad_norm": 0.7007821582356654,
      "learning_rate": 0.00019662862540802115,
      "loss": 12.6939,
      "step": 4055
    },
    {
      "epoch": 0.22086541014069552,
      "grad_norm": 0.820489336660405,
      "learning_rate": 0.00019662635459412239,
      "loss": 12.6103,
      "step": 4056
    },
    {
      "epoch": 0.22091986413727852,
      "grad_norm": 0.6884190201784677,
      "learning_rate": 0.0001966240830288409,
      "loss": 12.6194,
      "step": 4057
    },
    {
      "epoch": 0.22097431813386154,
      "grad_norm": 0.7121741594391826,
      "learning_rate": 0.00019662181071219433,
      "loss": 12.6443,
      "step": 4058
    },
    {
      "epoch": 0.22102877213044456,
      "grad_norm": 0.8198708361122542,
      "learning_rate": 0.00019661953764420036,
      "loss": 12.7249,
      "step": 4059
    },
    {
      "epoch": 0.22108322612702755,
      "grad_norm": 0.7579769411330018,
      "learning_rate": 0.00019661726382487666,
      "loss": 12.5446,
      "step": 4060
    },
    {
      "epoch": 0.22113768012361057,
      "grad_norm": 0.7321724612096204,
      "learning_rate": 0.0001966149892542409,
      "loss": 12.6478,
      "step": 4061
    },
    {
      "epoch": 0.2211921341201936,
      "grad_norm": 0.7611930700464427,
      "learning_rate": 0.0001966127139323108,
      "loss": 12.6653,
      "step": 4062
    },
    {
      "epoch": 0.22124658811677658,
      "grad_norm": 0.7843978002683786,
      "learning_rate": 0.00019661043785910404,
      "loss": 12.6038,
      "step": 4063
    },
    {
      "epoch": 0.2213010421133596,
      "grad_norm": 0.6158107207239923,
      "learning_rate": 0.00019660816103463831,
      "loss": 12.5457,
      "step": 4064
    },
    {
      "epoch": 0.22135549610994262,
      "grad_norm": 0.7349160639773851,
      "learning_rate": 0.00019660588345893132,
      "loss": 12.7054,
      "step": 4065
    },
    {
      "epoch": 0.22140995010652564,
      "grad_norm": 0.8064390119251176,
      "learning_rate": 0.00019660360513200074,
      "loss": 12.7049,
      "step": 4066
    },
    {
      "epoch": 0.22146440410310864,
      "grad_norm": 0.8285649273192288,
      "learning_rate": 0.00019660132605386438,
      "loss": 12.7298,
      "step": 4067
    },
    {
      "epoch": 0.22151885809969166,
      "grad_norm": 0.7237315038079152,
      "learning_rate": 0.00019659904622453987,
      "loss": 12.6079,
      "step": 4068
    },
    {
      "epoch": 0.22157331209627468,
      "grad_norm": 0.6493306300297004,
      "learning_rate": 0.00019659676564404502,
      "loss": 12.6085,
      "step": 4069
    },
    {
      "epoch": 0.22162776609285767,
      "grad_norm": 0.6758583142842389,
      "learning_rate": 0.00019659448431239747,
      "loss": 12.6432,
      "step": 4070
    },
    {
      "epoch": 0.2216822200894407,
      "grad_norm": 0.7787709224078277,
      "learning_rate": 0.00019659220222961504,
      "loss": 12.7466,
      "step": 4071
    },
    {
      "epoch": 0.2217366740860237,
      "grad_norm": 0.7290007115312267,
      "learning_rate": 0.00019658991939571543,
      "loss": 12.8019,
      "step": 4072
    },
    {
      "epoch": 0.2217911280826067,
      "grad_norm": 0.8270168433996183,
      "learning_rate": 0.0001965876358107164,
      "loss": 12.5678,
      "step": 4073
    },
    {
      "epoch": 0.22184558207918972,
      "grad_norm": 0.679111176559926,
      "learning_rate": 0.00019658535147463576,
      "loss": 12.5322,
      "step": 4074
    },
    {
      "epoch": 0.22190003607577274,
      "grad_norm": 0.7672042597534164,
      "learning_rate": 0.0001965830663874912,
      "loss": 12.4992,
      "step": 4075
    },
    {
      "epoch": 0.22195449007235574,
      "grad_norm": 0.7822125277954906,
      "learning_rate": 0.00019658078054930053,
      "loss": 12.6805,
      "step": 4076
    },
    {
      "epoch": 0.22200894406893876,
      "grad_norm": 0.708356713729573,
      "learning_rate": 0.0001965784939600815,
      "loss": 12.7529,
      "step": 4077
    },
    {
      "epoch": 0.22206339806552178,
      "grad_norm": 0.9155739250602781,
      "learning_rate": 0.00019657620661985193,
      "loss": 12.8248,
      "step": 4078
    },
    {
      "epoch": 0.22211785206210477,
      "grad_norm": 0.7570284536971716,
      "learning_rate": 0.00019657391852862956,
      "loss": 12.7426,
      "step": 4079
    },
    {
      "epoch": 0.2221723060586878,
      "grad_norm": 0.7462136050334757,
      "learning_rate": 0.0001965716296864322,
      "loss": 12.7032,
      "step": 4080
    },
    {
      "epoch": 0.2222267600552708,
      "grad_norm": 0.7066401415807251,
      "learning_rate": 0.00019656934009327769,
      "loss": 12.587,
      "step": 4081
    },
    {
      "epoch": 0.22228121405185383,
      "grad_norm": 0.7253805256660538,
      "learning_rate": 0.00019656704974918375,
      "loss": 12.511,
      "step": 4082
    },
    {
      "epoch": 0.22233566804843682,
      "grad_norm": 0.7647726895218305,
      "learning_rate": 0.00019656475865416825,
      "loss": 12.6477,
      "step": 4083
    },
    {
      "epoch": 0.22239012204501984,
      "grad_norm": 0.7215912435196542,
      "learning_rate": 0.00019656246680824902,
      "loss": 12.623,
      "step": 4084
    },
    {
      "epoch": 0.22244457604160287,
      "grad_norm": 0.7155861894393772,
      "learning_rate": 0.00019656017421144383,
      "loss": 12.6334,
      "step": 4085
    },
    {
      "epoch": 0.22249903003818586,
      "grad_norm": 0.7067711985235319,
      "learning_rate": 0.00019655788086377052,
      "loss": 12.6266,
      "step": 4086
    },
    {
      "epoch": 0.22255348403476888,
      "grad_norm": 0.7096197888718686,
      "learning_rate": 0.00019655558676524696,
      "loss": 12.6845,
      "step": 4087
    },
    {
      "epoch": 0.2226079380313519,
      "grad_norm": 0.7265919662042608,
      "learning_rate": 0.00019655329191589096,
      "loss": 12.5453,
      "step": 4088
    },
    {
      "epoch": 0.2226623920279349,
      "grad_norm": 0.7522879143900745,
      "learning_rate": 0.00019655099631572036,
      "loss": 12.6943,
      "step": 4089
    },
    {
      "epoch": 0.2227168460245179,
      "grad_norm": 0.6771672916143178,
      "learning_rate": 0.00019654869996475302,
      "loss": 12.5641,
      "step": 4090
    },
    {
      "epoch": 0.22277130002110093,
      "grad_norm": 0.7851229816759847,
      "learning_rate": 0.00019654640286300681,
      "loss": 12.6618,
      "step": 4091
    },
    {
      "epoch": 0.22282575401768392,
      "grad_norm": 0.72448806730371,
      "learning_rate": 0.00019654410501049956,
      "loss": 12.5751,
      "step": 4092
    },
    {
      "epoch": 0.22288020801426695,
      "grad_norm": 0.7173123686937297,
      "learning_rate": 0.0001965418064072492,
      "loss": 12.6472,
      "step": 4093
    },
    {
      "epoch": 0.22293466201084997,
      "grad_norm": 0.6722797988171498,
      "learning_rate": 0.00019653950705327352,
      "loss": 12.6206,
      "step": 4094
    },
    {
      "epoch": 0.22298911600743296,
      "grad_norm": 0.7023256897730658,
      "learning_rate": 0.00019653720694859045,
      "loss": 12.5819,
      "step": 4095
    },
    {
      "epoch": 0.22304357000401598,
      "grad_norm": 0.6768174394192674,
      "learning_rate": 0.0001965349060932179,
      "loss": 12.6696,
      "step": 4096
    },
    {
      "epoch": 0.223098024000599,
      "grad_norm": 0.6975337948647515,
      "learning_rate": 0.0001965326044871737,
      "loss": 12.5153,
      "step": 4097
    },
    {
      "epoch": 0.22315247799718202,
      "grad_norm": 0.6670804020560877,
      "learning_rate": 0.00019653030213047582,
      "loss": 12.5818,
      "step": 4098
    },
    {
      "epoch": 0.223206931993765,
      "grad_norm": 0.7435266309532783,
      "learning_rate": 0.00019652799902314207,
      "loss": 12.6334,
      "step": 4099
    },
    {
      "epoch": 0.22326138599034803,
      "grad_norm": 0.6434074828767778,
      "learning_rate": 0.00019652569516519043,
      "loss": 12.7553,
      "step": 4100
    },
    {
      "epoch": 0.22331583998693105,
      "grad_norm": 0.7702468225912249,
      "learning_rate": 0.00019652339055663883,
      "loss": 12.6044,
      "step": 4101
    },
    {
      "epoch": 0.22337029398351405,
      "grad_norm": 0.6827954463588795,
      "learning_rate": 0.0001965210851975051,
      "loss": 12.6484,
      "step": 4102
    },
    {
      "epoch": 0.22342474798009707,
      "grad_norm": 0.6972485509404794,
      "learning_rate": 0.00019651877908780728,
      "loss": 12.6238,
      "step": 4103
    },
    {
      "epoch": 0.2234792019766801,
      "grad_norm": 0.6520662489966935,
      "learning_rate": 0.0001965164722275632,
      "loss": 12.5524,
      "step": 4104
    },
    {
      "epoch": 0.22353365597326308,
      "grad_norm": 0.7948014592544079,
      "learning_rate": 0.00019651416461679086,
      "loss": 12.7693,
      "step": 4105
    },
    {
      "epoch": 0.2235881099698461,
      "grad_norm": 0.6865331043079922,
      "learning_rate": 0.00019651185625550822,
      "loss": 12.6462,
      "step": 4106
    },
    {
      "epoch": 0.22364256396642912,
      "grad_norm": 0.678447503680815,
      "learning_rate": 0.00019650954714373316,
      "loss": 12.4801,
      "step": 4107
    },
    {
      "epoch": 0.2236970179630121,
      "grad_norm": 0.6861766778644109,
      "learning_rate": 0.0001965072372814837,
      "loss": 12.7116,
      "step": 4108
    },
    {
      "epoch": 0.22375147195959513,
      "grad_norm": 0.6552148787788998,
      "learning_rate": 0.00019650492666877778,
      "loss": 12.5291,
      "step": 4109
    },
    {
      "epoch": 0.22380592595617815,
      "grad_norm": 0.7820906246883227,
      "learning_rate": 0.00019650261530563336,
      "loss": 12.773,
      "step": 4110
    },
    {
      "epoch": 0.22386037995276115,
      "grad_norm": 0.7375117802220131,
      "learning_rate": 0.00019650030319206844,
      "loss": 12.698,
      "step": 4111
    },
    {
      "epoch": 0.22391483394934417,
      "grad_norm": 0.6853124627196624,
      "learning_rate": 0.00019649799032810095,
      "loss": 12.6565,
      "step": 4112
    },
    {
      "epoch": 0.2239692879459272,
      "grad_norm": 0.8280545884927758,
      "learning_rate": 0.0001964956767137489,
      "loss": 12.5425,
      "step": 4113
    },
    {
      "epoch": 0.22402374194251018,
      "grad_norm": 0.7010860592903173,
      "learning_rate": 0.00019649336234903031,
      "loss": 12.6033,
      "step": 4114
    },
    {
      "epoch": 0.2240781959390932,
      "grad_norm": 0.8662239711101437,
      "learning_rate": 0.00019649104723396313,
      "loss": 12.7076,
      "step": 4115
    },
    {
      "epoch": 0.22413264993567622,
      "grad_norm": 0.7016765695316105,
      "learning_rate": 0.0001964887313685654,
      "loss": 12.6596,
      "step": 4116
    },
    {
      "epoch": 0.22418710393225924,
      "grad_norm": 0.7540474476328786,
      "learning_rate": 0.0001964864147528551,
      "loss": 12.6279,
      "step": 4117
    },
    {
      "epoch": 0.22424155792884223,
      "grad_norm": 0.8586664519926759,
      "learning_rate": 0.0001964840973868503,
      "loss": 12.677,
      "step": 4118
    },
    {
      "epoch": 0.22429601192542525,
      "grad_norm": 0.7518180989190282,
      "learning_rate": 0.00019648177927056892,
      "loss": 12.652,
      "step": 4119
    },
    {
      "epoch": 0.22435046592200827,
      "grad_norm": 0.8042141580978586,
      "learning_rate": 0.00019647946040402908,
      "loss": 12.6802,
      "step": 4120
    },
    {
      "epoch": 0.22440491991859127,
      "grad_norm": 0.7799997298026489,
      "learning_rate": 0.00019647714078724877,
      "loss": 12.6783,
      "step": 4121
    },
    {
      "epoch": 0.2244593739151743,
      "grad_norm": 0.7540211261418177,
      "learning_rate": 0.00019647482042024607,
      "loss": 12.5943,
      "step": 4122
    },
    {
      "epoch": 0.2245138279117573,
      "grad_norm": 0.8132608128572707,
      "learning_rate": 0.00019647249930303894,
      "loss": 12.5836,
      "step": 4123
    },
    {
      "epoch": 0.2245682819083403,
      "grad_norm": 0.6984708197327121,
      "learning_rate": 0.0001964701774356455,
      "loss": 12.5904,
      "step": 4124
    },
    {
      "epoch": 0.22462273590492332,
      "grad_norm": 0.8561128483421145,
      "learning_rate": 0.00019646785481808375,
      "loss": 12.5673,
      "step": 4125
    },
    {
      "epoch": 0.22467718990150634,
      "grad_norm": 0.7102278937221332,
      "learning_rate": 0.0001964655314503718,
      "loss": 12.6511,
      "step": 4126
    },
    {
      "epoch": 0.22473164389808933,
      "grad_norm": 0.8040309226278436,
      "learning_rate": 0.0001964632073325277,
      "loss": 12.5945,
      "step": 4127
    },
    {
      "epoch": 0.22478609789467235,
      "grad_norm": 0.7324994720574526,
      "learning_rate": 0.00019646088246456952,
      "loss": 12.5847,
      "step": 4128
    },
    {
      "epoch": 0.22484055189125537,
      "grad_norm": 0.6946331463546693,
      "learning_rate": 0.00019645855684651535,
      "loss": 12.6176,
      "step": 4129
    },
    {
      "epoch": 0.22489500588783837,
      "grad_norm": 0.8127685703864836,
      "learning_rate": 0.00019645623047838325,
      "loss": 12.6553,
      "step": 4130
    },
    {
      "epoch": 0.2249494598844214,
      "grad_norm": 0.7716413856032072,
      "learning_rate": 0.00019645390336019136,
      "loss": 12.6135,
      "step": 4131
    },
    {
      "epoch": 0.2250039138810044,
      "grad_norm": 0.7148837586860729,
      "learning_rate": 0.0001964515754919577,
      "loss": 12.608,
      "step": 4132
    },
    {
      "epoch": 0.22505836787758743,
      "grad_norm": 0.7684066446503315,
      "learning_rate": 0.00019644924687370045,
      "loss": 12.5608,
      "step": 4133
    },
    {
      "epoch": 0.22511282187417042,
      "grad_norm": 0.7642611411607005,
      "learning_rate": 0.00019644691750543767,
      "loss": 12.5928,
      "step": 4134
    },
    {
      "epoch": 0.22516727587075344,
      "grad_norm": 0.8170139904918936,
      "learning_rate": 0.0001964445873871875,
      "loss": 12.588,
      "step": 4135
    },
    {
      "epoch": 0.22522172986733646,
      "grad_norm": 0.754685287492516,
      "learning_rate": 0.000196442256518968,
      "loss": 12.65,
      "step": 4136
    },
    {
      "epoch": 0.22527618386391945,
      "grad_norm": 0.7479712814785009,
      "learning_rate": 0.00019643992490079736,
      "loss": 12.509,
      "step": 4137
    },
    {
      "epoch": 0.22533063786050248,
      "grad_norm": 0.7957960856658557,
      "learning_rate": 0.00019643759253269372,
      "loss": 12.5958,
      "step": 4138
    },
    {
      "epoch": 0.2253850918570855,
      "grad_norm": 0.6433026845291914,
      "learning_rate": 0.00019643525941467516,
      "loss": 12.5226,
      "step": 4139
    },
    {
      "epoch": 0.2254395458536685,
      "grad_norm": 0.7612427732206075,
      "learning_rate": 0.00019643292554675986,
      "loss": 12.536,
      "step": 4140
    },
    {
      "epoch": 0.2254939998502515,
      "grad_norm": 0.7186653394613721,
      "learning_rate": 0.00019643059092896596,
      "loss": 12.5145,
      "step": 4141
    },
    {
      "epoch": 0.22554845384683453,
      "grad_norm": 0.8423148084191446,
      "learning_rate": 0.0001964282555613116,
      "loss": 12.6958,
      "step": 4142
    },
    {
      "epoch": 0.22560290784341752,
      "grad_norm": 0.7648405356031992,
      "learning_rate": 0.00019642591944381497,
      "loss": 12.5947,
      "step": 4143
    },
    {
      "epoch": 0.22565736184000054,
      "grad_norm": 0.7652229852198938,
      "learning_rate": 0.0001964235825764942,
      "loss": 12.6697,
      "step": 4144
    },
    {
      "epoch": 0.22571181583658356,
      "grad_norm": 0.7824181915244368,
      "learning_rate": 0.00019642124495936752,
      "loss": 12.581,
      "step": 4145
    },
    {
      "epoch": 0.22576626983316656,
      "grad_norm": 0.766258138818385,
      "learning_rate": 0.000196418906592453,
      "loss": 12.6982,
      "step": 4146
    },
    {
      "epoch": 0.22582072382974958,
      "grad_norm": 0.7751052091342521,
      "learning_rate": 0.00019641656747576897,
      "loss": 12.6076,
      "step": 4147
    },
    {
      "epoch": 0.2258751778263326,
      "grad_norm": 0.9516446815823854,
      "learning_rate": 0.00019641422760933351,
      "loss": 12.8024,
      "step": 4148
    },
    {
      "epoch": 0.22592963182291562,
      "grad_norm": 0.6470356345065555,
      "learning_rate": 0.00019641188699316483,
      "loss": 12.6156,
      "step": 4149
    },
    {
      "epoch": 0.2259840858194986,
      "grad_norm": 0.7168144094402956,
      "learning_rate": 0.00019640954562728119,
      "loss": 12.5586,
      "step": 4150
    },
    {
      "epoch": 0.22603853981608163,
      "grad_norm": 0.704367688027529,
      "learning_rate": 0.0001964072035117007,
      "loss": 12.7024,
      "step": 4151
    },
    {
      "epoch": 0.22609299381266465,
      "grad_norm": 0.7572302834035705,
      "learning_rate": 0.00019640486064644166,
      "loss": 12.6288,
      "step": 4152
    },
    {
      "epoch": 0.22614744780924764,
      "grad_norm": 0.8705790782861066,
      "learning_rate": 0.00019640251703152229,
      "loss": 12.5788,
      "step": 4153
    },
    {
      "epoch": 0.22620190180583066,
      "grad_norm": 0.719569601862707,
      "learning_rate": 0.00019640017266696073,
      "loss": 12.6585,
      "step": 4154
    },
    {
      "epoch": 0.22625635580241368,
      "grad_norm": 0.8192872285406801,
      "learning_rate": 0.00019639782755277526,
      "loss": 12.8433,
      "step": 4155
    },
    {
      "epoch": 0.22631080979899668,
      "grad_norm": 0.7591508591799239,
      "learning_rate": 0.00019639548168898414,
      "loss": 12.7183,
      "step": 4156
    },
    {
      "epoch": 0.2263652637955797,
      "grad_norm": 0.6542259611617403,
      "learning_rate": 0.0001963931350756056,
      "loss": 12.542,
      "step": 4157
    },
    {
      "epoch": 0.22641971779216272,
      "grad_norm": 0.7187092777667181,
      "learning_rate": 0.00019639078771265783,
      "loss": 12.6347,
      "step": 4158
    },
    {
      "epoch": 0.2264741717887457,
      "grad_norm": 0.7491587955339019,
      "learning_rate": 0.00019638843960015918,
      "loss": 12.6626,
      "step": 4159
    },
    {
      "epoch": 0.22652862578532873,
      "grad_norm": 0.6890269683428409,
      "learning_rate": 0.00019638609073812784,
      "loss": 12.6092,
      "step": 4160
    },
    {
      "epoch": 0.22658307978191175,
      "grad_norm": 0.7204526761660863,
      "learning_rate": 0.0001963837411265821,
      "loss": 12.5484,
      "step": 4161
    },
    {
      "epoch": 0.22663753377849474,
      "grad_norm": 0.6620763697964263,
      "learning_rate": 0.0001963813907655402,
      "loss": 12.5442,
      "step": 4162
    },
    {
      "epoch": 0.22669198777507776,
      "grad_norm": 0.6967383157647706,
      "learning_rate": 0.00019637903965502048,
      "loss": 12.6577,
      "step": 4163
    },
    {
      "epoch": 0.22674644177166078,
      "grad_norm": 0.7205232920046198,
      "learning_rate": 0.00019637668779504114,
      "loss": 12.7114,
      "step": 4164
    },
    {
      "epoch": 0.2268008957682438,
      "grad_norm": 0.8163517909426771,
      "learning_rate": 0.00019637433518562055,
      "loss": 12.673,
      "step": 4165
    },
    {
      "epoch": 0.2268553497648268,
      "grad_norm": 0.7954287675018735,
      "learning_rate": 0.00019637198182677695,
      "loss": 12.5768,
      "step": 4166
    },
    {
      "epoch": 0.22690980376140982,
      "grad_norm": 0.7460431744584577,
      "learning_rate": 0.00019636962771852866,
      "loss": 12.4162,
      "step": 4167
    },
    {
      "epoch": 0.22696425775799284,
      "grad_norm": 0.7643922709161617,
      "learning_rate": 0.00019636727286089398,
      "loss": 12.5648,
      "step": 4168
    },
    {
      "epoch": 0.22701871175457583,
      "grad_norm": 0.6811069165096117,
      "learning_rate": 0.00019636491725389123,
      "loss": 12.5059,
      "step": 4169
    },
    {
      "epoch": 0.22707316575115885,
      "grad_norm": 0.6880108265779822,
      "learning_rate": 0.00019636256089753876,
      "loss": 12.5661,
      "step": 4170
    },
    {
      "epoch": 0.22712761974774187,
      "grad_norm": 0.8314396438696734,
      "learning_rate": 0.0001963602037918548,
      "loss": 12.5678,
      "step": 4171
    },
    {
      "epoch": 0.22718207374432486,
      "grad_norm": 0.8376904101132442,
      "learning_rate": 0.00019635784593685776,
      "loss": 12.8749,
      "step": 4172
    },
    {
      "epoch": 0.22723652774090788,
      "grad_norm": 0.7170796749752048,
      "learning_rate": 0.00019635548733256596,
      "loss": 12.5654,
      "step": 4173
    },
    {
      "epoch": 0.2272909817374909,
      "grad_norm": 0.7785199073373608,
      "learning_rate": 0.00019635312797899773,
      "loss": 12.6354,
      "step": 4174
    },
    {
      "epoch": 0.2273454357340739,
      "grad_norm": 0.683307625656454,
      "learning_rate": 0.0001963507678761714,
      "loss": 12.516,
      "step": 4175
    },
    {
      "epoch": 0.22739988973065692,
      "grad_norm": 0.7206391649286215,
      "learning_rate": 0.00019634840702410534,
      "loss": 12.5217,
      "step": 4176
    },
    {
      "epoch": 0.22745434372723994,
      "grad_norm": 0.7691043392271019,
      "learning_rate": 0.00019634604542281793,
      "loss": 12.6352,
      "step": 4177
    },
    {
      "epoch": 0.22750879772382293,
      "grad_norm": 0.7698192268790165,
      "learning_rate": 0.0001963436830723275,
      "loss": 12.6516,
      "step": 4178
    },
    {
      "epoch": 0.22756325172040595,
      "grad_norm": 0.8346404720305081,
      "learning_rate": 0.00019634131997265243,
      "loss": 12.6,
      "step": 4179
    },
    {
      "epoch": 0.22761770571698897,
      "grad_norm": 0.6818073699154533,
      "learning_rate": 0.0001963389561238111,
      "loss": 12.6393,
      "step": 4180
    },
    {
      "epoch": 0.22767215971357196,
      "grad_norm": 0.8939568346328681,
      "learning_rate": 0.00019633659152582192,
      "loss": 12.6412,
      "step": 4181
    },
    {
      "epoch": 0.22772661371015498,
      "grad_norm": 0.8633836344610171,
      "learning_rate": 0.0001963342261787032,
      "loss": 12.5934,
      "step": 4182
    },
    {
      "epoch": 0.227781067706738,
      "grad_norm": 0.7662553649513152,
      "learning_rate": 0.00019633186008247342,
      "loss": 12.7304,
      "step": 4183
    },
    {
      "epoch": 0.22783552170332103,
      "grad_norm": 0.8173319496752366,
      "learning_rate": 0.00019632949323715093,
      "loss": 12.7181,
      "step": 4184
    },
    {
      "epoch": 0.22788997569990402,
      "grad_norm": 0.7528487750285134,
      "learning_rate": 0.00019632712564275414,
      "loss": 12.574,
      "step": 4185
    },
    {
      "epoch": 0.22794442969648704,
      "grad_norm": 0.9025976985810679,
      "learning_rate": 0.00019632475729930147,
      "loss": 12.3497,
      "step": 4186
    },
    {
      "epoch": 0.22799888369307006,
      "grad_norm": 0.7802567733274954,
      "learning_rate": 0.00019632238820681134,
      "loss": 12.7456,
      "step": 4187
    },
    {
      "epoch": 0.22805333768965305,
      "grad_norm": 0.8131014048200902,
      "learning_rate": 0.0001963200183653022,
      "loss": 12.57,
      "step": 4188
    },
    {
      "epoch": 0.22810779168623607,
      "grad_norm": 0.75760509813138,
      "learning_rate": 0.00019631764777479238,
      "loss": 12.723,
      "step": 4189
    },
    {
      "epoch": 0.2281622456828191,
      "grad_norm": 0.7666646103675416,
      "learning_rate": 0.0001963152764353004,
      "loss": 12.524,
      "step": 4190
    },
    {
      "epoch": 0.22821669967940209,
      "grad_norm": 0.7536721587180825,
      "learning_rate": 0.00019631290434684467,
      "loss": 12.4945,
      "step": 4191
    },
    {
      "epoch": 0.2282711536759851,
      "grad_norm": 0.7901263068923782,
      "learning_rate": 0.00019631053150944366,
      "loss": 12.6857,
      "step": 4192
    },
    {
      "epoch": 0.22832560767256813,
      "grad_norm": 0.7848747736826245,
      "learning_rate": 0.00019630815792311582,
      "loss": 12.6599,
      "step": 4193
    },
    {
      "epoch": 0.22838006166915112,
      "grad_norm": 0.7355671871543848,
      "learning_rate": 0.00019630578358787956,
      "loss": 12.585,
      "step": 4194
    },
    {
      "epoch": 0.22843451566573414,
      "grad_norm": 0.8300002825783853,
      "learning_rate": 0.0001963034085037534,
      "loss": 12.7075,
      "step": 4195
    },
    {
      "epoch": 0.22848896966231716,
      "grad_norm": 0.7966639359103621,
      "learning_rate": 0.00019630103267075577,
      "loss": 12.7292,
      "step": 4196
    },
    {
      "epoch": 0.22854342365890015,
      "grad_norm": 0.7727276636202759,
      "learning_rate": 0.00019629865608890515,
      "loss": 12.648,
      "step": 4197
    },
    {
      "epoch": 0.22859787765548317,
      "grad_norm": 0.7582843931050108,
      "learning_rate": 0.00019629627875822006,
      "loss": 12.6394,
      "step": 4198
    },
    {
      "epoch": 0.2286523316520662,
      "grad_norm": 0.7575204216635002,
      "learning_rate": 0.00019629390067871894,
      "loss": 12.563,
      "step": 4199
    },
    {
      "epoch": 0.2287067856486492,
      "grad_norm": 0.7514351983682588,
      "learning_rate": 0.00019629152185042032,
      "loss": 12.5458,
      "step": 4200
    },
    {
      "epoch": 0.2287612396452322,
      "grad_norm": 0.7343913978310641,
      "learning_rate": 0.00019628914227334265,
      "loss": 12.6761,
      "step": 4201
    },
    {
      "epoch": 0.22881569364181523,
      "grad_norm": 0.7953266280694237,
      "learning_rate": 0.0001962867619475045,
      "loss": 12.4819,
      "step": 4202
    },
    {
      "epoch": 0.22887014763839825,
      "grad_norm": 0.674170778375874,
      "learning_rate": 0.0001962843808729243,
      "loss": 12.6999,
      "step": 4203
    },
    {
      "epoch": 0.22892460163498124,
      "grad_norm": 0.8598325420364537,
      "learning_rate": 0.00019628199904962065,
      "loss": 12.5789,
      "step": 4204
    },
    {
      "epoch": 0.22897905563156426,
      "grad_norm": 0.7080454670134934,
      "learning_rate": 0.00019627961647761198,
      "loss": 12.7293,
      "step": 4205
    },
    {
      "epoch": 0.22903350962814728,
      "grad_norm": 0.8012826117160416,
      "learning_rate": 0.0001962772331569169,
      "loss": 12.5654,
      "step": 4206
    },
    {
      "epoch": 0.22908796362473027,
      "grad_norm": 0.8221366642183998,
      "learning_rate": 0.0001962748490875539,
      "loss": 12.7532,
      "step": 4207
    },
    {
      "epoch": 0.2291424176213133,
      "grad_norm": 0.7487133557436079,
      "learning_rate": 0.00019627246426954153,
      "loss": 12.7589,
      "step": 4208
    },
    {
      "epoch": 0.22919687161789631,
      "grad_norm": 0.8018833033180254,
      "learning_rate": 0.00019627007870289833,
      "loss": 12.5792,
      "step": 4209
    },
    {
      "epoch": 0.2292513256144793,
      "grad_norm": 0.6758886976115429,
      "learning_rate": 0.00019626769238764285,
      "loss": 12.5047,
      "step": 4210
    },
    {
      "epoch": 0.22930577961106233,
      "grad_norm": 0.7721195598513376,
      "learning_rate": 0.00019626530532379366,
      "loss": 12.6491,
      "step": 4211
    },
    {
      "epoch": 0.22936023360764535,
      "grad_norm": 0.7287958100996574,
      "learning_rate": 0.00019626291751136932,
      "loss": 12.7247,
      "step": 4212
    },
    {
      "epoch": 0.22941468760422834,
      "grad_norm": 0.70739885144157,
      "learning_rate": 0.00019626052895038837,
      "loss": 12.6829,
      "step": 4213
    },
    {
      "epoch": 0.22946914160081136,
      "grad_norm": 0.9481536470865546,
      "learning_rate": 0.00019625813964086941,
      "loss": 12.6122,
      "step": 4214
    },
    {
      "epoch": 0.22952359559739438,
      "grad_norm": 0.7477751782480188,
      "learning_rate": 0.000196255749582831,
      "loss": 12.5888,
      "step": 4215
    },
    {
      "epoch": 0.2295780495939774,
      "grad_norm": 0.8458601177972982,
      "learning_rate": 0.00019625335877629176,
      "loss": 12.5862,
      "step": 4216
    },
    {
      "epoch": 0.2296325035905604,
      "grad_norm": 0.7034832614989683,
      "learning_rate": 0.00019625096722127025,
      "loss": 12.6872,
      "step": 4217
    },
    {
      "epoch": 0.22968695758714341,
      "grad_norm": 0.8110735484700187,
      "learning_rate": 0.0001962485749177851,
      "loss": 12.425,
      "step": 4218
    },
    {
      "epoch": 0.22974141158372643,
      "grad_norm": 0.8934675458851745,
      "learning_rate": 0.00019624618186585483,
      "loss": 12.7089,
      "step": 4219
    },
    {
      "epoch": 0.22979586558030943,
      "grad_norm": 0.9536438597146135,
      "learning_rate": 0.00019624378806549816,
      "loss": 12.8108,
      "step": 4220
    },
    {
      "epoch": 0.22985031957689245,
      "grad_norm": 0.7748427675417733,
      "learning_rate": 0.00019624139351673368,
      "loss": 12.592,
      "step": 4221
    },
    {
      "epoch": 0.22990477357347547,
      "grad_norm": 0.7547516755231537,
      "learning_rate": 0.00019623899821957994,
      "loss": 12.6094,
      "step": 4222
    },
    {
      "epoch": 0.22995922757005846,
      "grad_norm": 0.7069906171041361,
      "learning_rate": 0.0001962366021740556,
      "loss": 12.4284,
      "step": 4223
    },
    {
      "epoch": 0.23001368156664148,
      "grad_norm": 0.7339094492319244,
      "learning_rate": 0.00019623420538017933,
      "loss": 12.6512,
      "step": 4224
    },
    {
      "epoch": 0.2300681355632245,
      "grad_norm": 0.7168079873618214,
      "learning_rate": 0.00019623180783796972,
      "loss": 12.7724,
      "step": 4225
    },
    {
      "epoch": 0.2301225895598075,
      "grad_norm": 0.7628963015436998,
      "learning_rate": 0.00019622940954744546,
      "loss": 12.7698,
      "step": 4226
    },
    {
      "epoch": 0.23017704355639051,
      "grad_norm": 0.800313160513469,
      "learning_rate": 0.00019622701050862516,
      "loss": 12.668,
      "step": 4227
    },
    {
      "epoch": 0.23023149755297354,
      "grad_norm": 0.7402059656718677,
      "learning_rate": 0.0001962246107215275,
      "loss": 12.7014,
      "step": 4228
    },
    {
      "epoch": 0.23028595154955653,
      "grad_norm": 0.768773436899772,
      "learning_rate": 0.0001962222101861711,
      "loss": 12.6177,
      "step": 4229
    },
    {
      "epoch": 0.23034040554613955,
      "grad_norm": 0.7433993649806422,
      "learning_rate": 0.00019621980890257467,
      "loss": 12.6767,
      "step": 4230
    },
    {
      "epoch": 0.23039485954272257,
      "grad_norm": 0.7848663476635999,
      "learning_rate": 0.0001962174068707569,
      "loss": 12.6785,
      "step": 4231
    },
    {
      "epoch": 0.2304493135393056,
      "grad_norm": 0.9097231002227915,
      "learning_rate": 0.00019621500409073642,
      "loss": 12.7234,
      "step": 4232
    },
    {
      "epoch": 0.23050376753588858,
      "grad_norm": 0.737656135746251,
      "learning_rate": 0.0001962126005625319,
      "loss": 12.5546,
      "step": 4233
    },
    {
      "epoch": 0.2305582215324716,
      "grad_norm": 0.7434625052577787,
      "learning_rate": 0.0001962101962861621,
      "loss": 12.6206,
      "step": 4234
    },
    {
      "epoch": 0.23061267552905462,
      "grad_norm": 0.7067678014290748,
      "learning_rate": 0.00019620779126164567,
      "loss": 12.5727,
      "step": 4235
    },
    {
      "epoch": 0.23066712952563762,
      "grad_norm": 0.8034904350345814,
      "learning_rate": 0.00019620538548900134,
      "loss": 12.5893,
      "step": 4236
    },
    {
      "epoch": 0.23072158352222064,
      "grad_norm": 0.7364403510918875,
      "learning_rate": 0.00019620297896824778,
      "loss": 12.6274,
      "step": 4237
    },
    {
      "epoch": 0.23077603751880366,
      "grad_norm": 0.7482614937671299,
      "learning_rate": 0.00019620057169940372,
      "loss": 12.6263,
      "step": 4238
    },
    {
      "epoch": 0.23083049151538665,
      "grad_norm": 0.7421278997437073,
      "learning_rate": 0.0001961981636824879,
      "loss": 12.5673,
      "step": 4239
    },
    {
      "epoch": 0.23088494551196967,
      "grad_norm": 0.7276833601350934,
      "learning_rate": 0.000196195754917519,
      "loss": 12.5512,
      "step": 4240
    },
    {
      "epoch": 0.2309393995085527,
      "grad_norm": 0.7876421122370915,
      "learning_rate": 0.00019619334540451578,
      "loss": 12.6634,
      "step": 4241
    },
    {
      "epoch": 0.23099385350513568,
      "grad_norm": 0.6979158172927132,
      "learning_rate": 0.00019619093514349698,
      "loss": 12.5341,
      "step": 4242
    },
    {
      "epoch": 0.2310483075017187,
      "grad_norm": 0.655866451015755,
      "learning_rate": 0.00019618852413448134,
      "loss": 12.5973,
      "step": 4243
    },
    {
      "epoch": 0.23110276149830172,
      "grad_norm": 0.8482436189951839,
      "learning_rate": 0.0001961861123774876,
      "loss": 12.7595,
      "step": 4244
    },
    {
      "epoch": 0.23115721549488472,
      "grad_norm": 0.6987739037734397,
      "learning_rate": 0.00019618369987253452,
      "loss": 12.5868,
      "step": 4245
    },
    {
      "epoch": 0.23121166949146774,
      "grad_norm": 0.7143858283375534,
      "learning_rate": 0.0001961812866196409,
      "loss": 12.563,
      "step": 4246
    },
    {
      "epoch": 0.23126612348805076,
      "grad_norm": 0.7598538747605517,
      "learning_rate": 0.00019617887261882543,
      "loss": 12.6624,
      "step": 4247
    },
    {
      "epoch": 0.23132057748463375,
      "grad_norm": 0.6621791276826825,
      "learning_rate": 0.0001961764578701069,
      "loss": 12.4685,
      "step": 4248
    },
    {
      "epoch": 0.23137503148121677,
      "grad_norm": 0.6874364514880499,
      "learning_rate": 0.00019617404237350412,
      "loss": 12.6243,
      "step": 4249
    },
    {
      "epoch": 0.2314294854777998,
      "grad_norm": 0.6843292576653252,
      "learning_rate": 0.00019617162612903588,
      "loss": 12.6069,
      "step": 4250
    },
    {
      "epoch": 0.2314839394743828,
      "grad_norm": 0.7386083117850814,
      "learning_rate": 0.00019616920913672093,
      "loss": 12.6869,
      "step": 4251
    },
    {
      "epoch": 0.2315383934709658,
      "grad_norm": 0.7430864456399691,
      "learning_rate": 0.00019616679139657808,
      "loss": 12.7086,
      "step": 4252
    },
    {
      "epoch": 0.23159284746754882,
      "grad_norm": 0.7003175792165027,
      "learning_rate": 0.00019616437290862613,
      "loss": 12.5829,
      "step": 4253
    },
    {
      "epoch": 0.23164730146413184,
      "grad_norm": 0.789833782665007,
      "learning_rate": 0.0001961619536728839,
      "loss": 12.6162,
      "step": 4254
    },
    {
      "epoch": 0.23170175546071484,
      "grad_norm": 0.7355859612170126,
      "learning_rate": 0.00019615953368937018,
      "loss": 12.456,
      "step": 4255
    },
    {
      "epoch": 0.23175620945729786,
      "grad_norm": 0.7273326403014023,
      "learning_rate": 0.0001961571129581038,
      "loss": 12.639,
      "step": 4256
    },
    {
      "epoch": 0.23181066345388088,
      "grad_norm": 0.7677796485027044,
      "learning_rate": 0.00019615469147910358,
      "loss": 12.7071,
      "step": 4257
    },
    {
      "epoch": 0.23186511745046387,
      "grad_norm": 0.6403050854559836,
      "learning_rate": 0.00019615226925238837,
      "loss": 12.5584,
      "step": 4258
    },
    {
      "epoch": 0.2319195714470469,
      "grad_norm": 0.7559823101393152,
      "learning_rate": 0.00019614984627797699,
      "loss": 12.5595,
      "step": 4259
    },
    {
      "epoch": 0.2319740254436299,
      "grad_norm": 0.6971580283174519,
      "learning_rate": 0.00019614742255588826,
      "loss": 12.6703,
      "step": 4260
    },
    {
      "epoch": 0.2320284794402129,
      "grad_norm": 0.7388504771360798,
      "learning_rate": 0.00019614499808614106,
      "loss": 12.6664,
      "step": 4261
    },
    {
      "epoch": 0.23208293343679592,
      "grad_norm": 0.8428539947928397,
      "learning_rate": 0.00019614257286875423,
      "loss": 12.64,
      "step": 4262
    },
    {
      "epoch": 0.23213738743337894,
      "grad_norm": 0.6449467533861518,
      "learning_rate": 0.00019614014690374666,
      "loss": 12.5579,
      "step": 4263
    },
    {
      "epoch": 0.23219184142996194,
      "grad_norm": 0.7948607639082977,
      "learning_rate": 0.00019613772019113715,
      "loss": 12.603,
      "step": 4264
    },
    {
      "epoch": 0.23224629542654496,
      "grad_norm": 0.7533943506759686,
      "learning_rate": 0.0001961352927309446,
      "loss": 12.6152,
      "step": 4265
    },
    {
      "epoch": 0.23230074942312798,
      "grad_norm": 0.7791838528326387,
      "learning_rate": 0.0001961328645231879,
      "loss": 12.5939,
      "step": 4266
    },
    {
      "epoch": 0.232355203419711,
      "grad_norm": 0.73616378505053,
      "learning_rate": 0.00019613043556788594,
      "loss": 12.5496,
      "step": 4267
    },
    {
      "epoch": 0.232409657416294,
      "grad_norm": 0.7783342577870753,
      "learning_rate": 0.0001961280058650576,
      "loss": 12.74,
      "step": 4268
    },
    {
      "epoch": 0.232464111412877,
      "grad_norm": 0.7484188737899983,
      "learning_rate": 0.0001961255754147217,
      "loss": 12.5975,
      "step": 4269
    },
    {
      "epoch": 0.23251856540946003,
      "grad_norm": 0.6277519572272489,
      "learning_rate": 0.00019612314421689727,
      "loss": 12.4444,
      "step": 4270
    },
    {
      "epoch": 0.23257301940604302,
      "grad_norm": 0.7895078058046108,
      "learning_rate": 0.00019612071227160315,
      "loss": 12.5838,
      "step": 4271
    },
    {
      "epoch": 0.23262747340262604,
      "grad_norm": 0.7453697623277638,
      "learning_rate": 0.00019611827957885823,
      "loss": 12.4951,
      "step": 4272
    },
    {
      "epoch": 0.23268192739920907,
      "grad_norm": 0.7374459052273574,
      "learning_rate": 0.00019611584613868146,
      "loss": 12.6958,
      "step": 4273
    },
    {
      "epoch": 0.23273638139579206,
      "grad_norm": 0.7871237598849445,
      "learning_rate": 0.00019611341195109174,
      "loss": 12.6594,
      "step": 4274
    },
    {
      "epoch": 0.23279083539237508,
      "grad_norm": 0.7746752949147973,
      "learning_rate": 0.00019611097701610804,
      "loss": 12.7189,
      "step": 4275
    },
    {
      "epoch": 0.2328452893889581,
      "grad_norm": 0.8101023763463789,
      "learning_rate": 0.00019610854133374922,
      "loss": 12.6418,
      "step": 4276
    },
    {
      "epoch": 0.2328997433855411,
      "grad_norm": 0.876845460569168,
      "learning_rate": 0.00019610610490403428,
      "loss": 12.5988,
      "step": 4277
    },
    {
      "epoch": 0.2329541973821241,
      "grad_norm": 0.6829682379865623,
      "learning_rate": 0.00019610366772698218,
      "loss": 12.7112,
      "step": 4278
    },
    {
      "epoch": 0.23300865137870713,
      "grad_norm": 0.8823059947422979,
      "learning_rate": 0.00019610122980261182,
      "loss": 12.7105,
      "step": 4279
    },
    {
      "epoch": 0.23306310537529012,
      "grad_norm": 0.6757896880882112,
      "learning_rate": 0.00019609879113094216,
      "loss": 12.5127,
      "step": 4280
    },
    {
      "epoch": 0.23311755937187315,
      "grad_norm": 0.7799192024165947,
      "learning_rate": 0.0001960963517119922,
      "loss": 12.6248,
      "step": 4281
    },
    {
      "epoch": 0.23317201336845617,
      "grad_norm": 0.7738107472625442,
      "learning_rate": 0.0001960939115457809,
      "loss": 12.6383,
      "step": 4282
    },
    {
      "epoch": 0.2332264673650392,
      "grad_norm": 0.6654857619867839,
      "learning_rate": 0.00019609147063232723,
      "loss": 12.5648,
      "step": 4283
    },
    {
      "epoch": 0.23328092136162218,
      "grad_norm": 0.7831452104370247,
      "learning_rate": 0.0001960890289716502,
      "loss": 12.6137,
      "step": 4284
    },
    {
      "epoch": 0.2333353753582052,
      "grad_norm": 0.7727339044966621,
      "learning_rate": 0.0001960865865637687,
      "loss": 12.7041,
      "step": 4285
    },
    {
      "epoch": 0.23338982935478822,
      "grad_norm": 0.7715279175080322,
      "learning_rate": 0.00019608414340870184,
      "loss": 12.7531,
      "step": 4286
    },
    {
      "epoch": 0.2334442833513712,
      "grad_norm": 0.8001630128369412,
      "learning_rate": 0.00019608169950646859,
      "loss": 12.5734,
      "step": 4287
    },
    {
      "epoch": 0.23349873734795423,
      "grad_norm": 0.787395148810035,
      "learning_rate": 0.00019607925485708787,
      "loss": 12.6792,
      "step": 4288
    },
    {
      "epoch": 0.23355319134453725,
      "grad_norm": 0.842592994942548,
      "learning_rate": 0.00019607680946057875,
      "loss": 12.666,
      "step": 4289
    },
    {
      "epoch": 0.23360764534112025,
      "grad_norm": 0.7512388506883075,
      "learning_rate": 0.0001960743633169603,
      "loss": 12.59,
      "step": 4290
    },
    {
      "epoch": 0.23366209933770327,
      "grad_norm": 0.72247188846382,
      "learning_rate": 0.00019607191642625145,
      "loss": 12.6564,
      "step": 4291
    },
    {
      "epoch": 0.2337165533342863,
      "grad_norm": 0.8745336571336647,
      "learning_rate": 0.0001960694687884713,
      "loss": 12.5119,
      "step": 4292
    },
    {
      "epoch": 0.23377100733086928,
      "grad_norm": 0.6912730098958464,
      "learning_rate": 0.0001960670204036388,
      "loss": 12.5903,
      "step": 4293
    },
    {
      "epoch": 0.2338254613274523,
      "grad_norm": 0.8363161507249239,
      "learning_rate": 0.00019606457127177308,
      "loss": 12.6366,
      "step": 4294
    },
    {
      "epoch": 0.23387991532403532,
      "grad_norm": 0.7493847572979708,
      "learning_rate": 0.00019606212139289313,
      "loss": 12.6625,
      "step": 4295
    },
    {
      "epoch": 0.2339343693206183,
      "grad_norm": 0.7003006750115612,
      "learning_rate": 0.00019605967076701802,
      "loss": 12.5667,
      "step": 4296
    },
    {
      "epoch": 0.23398882331720133,
      "grad_norm": 0.7875107147194136,
      "learning_rate": 0.00019605721939416678,
      "loss": 12.6736,
      "step": 4297
    },
    {
      "epoch": 0.23404327731378435,
      "grad_norm": 0.7435004985456852,
      "learning_rate": 0.00019605476727435855,
      "loss": 12.6327,
      "step": 4298
    },
    {
      "epoch": 0.23409773131036737,
      "grad_norm": 0.9089254508292351,
      "learning_rate": 0.0001960523144076123,
      "loss": 12.7987,
      "step": 4299
    },
    {
      "epoch": 0.23415218530695037,
      "grad_norm": 0.7142398932570184,
      "learning_rate": 0.00019604986079394711,
      "loss": 12.6418,
      "step": 4300
    },
    {
      "epoch": 0.2342066393035334,
      "grad_norm": 0.6312072194048043,
      "learning_rate": 0.00019604740643338215,
      "loss": 12.5074,
      "step": 4301
    },
    {
      "epoch": 0.2342610933001164,
      "grad_norm": 0.6643375200710461,
      "learning_rate": 0.00019604495132593644,
      "loss": 12.3109,
      "step": 4302
    },
    {
      "epoch": 0.2343155472966994,
      "grad_norm": 0.7941833749668644,
      "learning_rate": 0.00019604249547162906,
      "loss": 12.6772,
      "step": 4303
    },
    {
      "epoch": 0.23437000129328242,
      "grad_norm": 0.709594513312813,
      "learning_rate": 0.00019604003887047916,
      "loss": 12.5156,
      "step": 4304
    },
    {
      "epoch": 0.23442445528986544,
      "grad_norm": 0.7762370769912964,
      "learning_rate": 0.00019603758152250577,
      "loss": 12.8532,
      "step": 4305
    },
    {
      "epoch": 0.23447890928644843,
      "grad_norm": 0.7910019480267998,
      "learning_rate": 0.00019603512342772808,
      "loss": 12.5901,
      "step": 4306
    },
    {
      "epoch": 0.23453336328303145,
      "grad_norm": 0.6729287952350281,
      "learning_rate": 0.00019603266458616514,
      "loss": 12.5129,
      "step": 4307
    },
    {
      "epoch": 0.23458781727961447,
      "grad_norm": 0.7864051008022672,
      "learning_rate": 0.00019603020499783612,
      "loss": 12.411,
      "step": 4308
    },
    {
      "epoch": 0.23464227127619747,
      "grad_norm": 0.7335956838909282,
      "learning_rate": 0.00019602774466276007,
      "loss": 12.6596,
      "step": 4309
    },
    {
      "epoch": 0.2346967252727805,
      "grad_norm": 0.7876811390143509,
      "learning_rate": 0.00019602528358095625,
      "loss": 12.6153,
      "step": 4310
    },
    {
      "epoch": 0.2347511792693635,
      "grad_norm": 0.7767342521011587,
      "learning_rate": 0.00019602282175244367,
      "loss": 12.5518,
      "step": 4311
    },
    {
      "epoch": 0.2348056332659465,
      "grad_norm": 0.7399594559128674,
      "learning_rate": 0.00019602035917724153,
      "loss": 12.6122,
      "step": 4312
    },
    {
      "epoch": 0.23486008726252952,
      "grad_norm": 0.8289972152340461,
      "learning_rate": 0.000196017895855369,
      "loss": 12.5061,
      "step": 4313
    },
    {
      "epoch": 0.23491454125911254,
      "grad_norm": 0.7408918207456543,
      "learning_rate": 0.00019601543178684517,
      "loss": 12.5571,
      "step": 4314
    },
    {
      "epoch": 0.23496899525569553,
      "grad_norm": 0.8245995095661995,
      "learning_rate": 0.00019601296697168926,
      "loss": 12.6654,
      "step": 4315
    },
    {
      "epoch": 0.23502344925227855,
      "grad_norm": 0.7435772943999259,
      "learning_rate": 0.00019601050140992044,
      "loss": 12.6058,
      "step": 4316
    },
    {
      "epoch": 0.23507790324886157,
      "grad_norm": 0.7958728150332822,
      "learning_rate": 0.00019600803510155782,
      "loss": 12.7924,
      "step": 4317
    },
    {
      "epoch": 0.2351323572454446,
      "grad_norm": 0.7486224387089223,
      "learning_rate": 0.00019600556804662064,
      "loss": 12.634,
      "step": 4318
    },
    {
      "epoch": 0.2351868112420276,
      "grad_norm": 0.7969138995287975,
      "learning_rate": 0.00019600310024512808,
      "loss": 12.6919,
      "step": 4319
    },
    {
      "epoch": 0.2352412652386106,
      "grad_norm": 0.70620545564187,
      "learning_rate": 0.00019600063169709927,
      "loss": 12.718,
      "step": 4320
    },
    {
      "epoch": 0.23529571923519363,
      "grad_norm": 0.7393046120739886,
      "learning_rate": 0.0001959981624025535,
      "loss": 12.5344,
      "step": 4321
    },
    {
      "epoch": 0.23535017323177662,
      "grad_norm": 0.7254327438539935,
      "learning_rate": 0.00019599569236150986,
      "loss": 12.5618,
      "step": 4322
    },
    {
      "epoch": 0.23540462722835964,
      "grad_norm": 0.6887552576178985,
      "learning_rate": 0.00019599322157398764,
      "loss": 12.6536,
      "step": 4323
    },
    {
      "epoch": 0.23545908122494266,
      "grad_norm": 0.7553241878725601,
      "learning_rate": 0.00019599075004000607,
      "loss": 12.5744,
      "step": 4324
    },
    {
      "epoch": 0.23551353522152566,
      "grad_norm": 0.7499048113007982,
      "learning_rate": 0.00019598827775958432,
      "loss": 12.6286,
      "step": 4325
    },
    {
      "epoch": 0.23556798921810868,
      "grad_norm": 0.8636137911339891,
      "learning_rate": 0.0001959858047327416,
      "loss": 12.7978,
      "step": 4326
    },
    {
      "epoch": 0.2356224432146917,
      "grad_norm": 0.7486050817099174,
      "learning_rate": 0.00019598333095949716,
      "loss": 12.5916,
      "step": 4327
    },
    {
      "epoch": 0.2356768972112747,
      "grad_norm": 0.719927559818087,
      "learning_rate": 0.00019598085643987025,
      "loss": 12.6053,
      "step": 4328
    },
    {
      "epoch": 0.2357313512078577,
      "grad_norm": 0.7384823362730173,
      "learning_rate": 0.0001959783811738801,
      "loss": 12.8224,
      "step": 4329
    },
    {
      "epoch": 0.23578580520444073,
      "grad_norm": 0.6720819302333678,
      "learning_rate": 0.00019597590516154598,
      "loss": 12.5344,
      "step": 4330
    },
    {
      "epoch": 0.23584025920102372,
      "grad_norm": 0.7603169777358452,
      "learning_rate": 0.00019597342840288711,
      "loss": 12.6039,
      "step": 4331
    },
    {
      "epoch": 0.23589471319760674,
      "grad_norm": 0.7062455194557229,
      "learning_rate": 0.00019597095089792278,
      "loss": 12.6485,
      "step": 4332
    },
    {
      "epoch": 0.23594916719418976,
      "grad_norm": 0.735680768440551,
      "learning_rate": 0.0001959684726466722,
      "loss": 12.8305,
      "step": 4333
    },
    {
      "epoch": 0.23600362119077278,
      "grad_norm": 0.6484772548160889,
      "learning_rate": 0.00019596599364915472,
      "loss": 12.4678,
      "step": 4334
    },
    {
      "epoch": 0.23605807518735578,
      "grad_norm": 0.7327670055095351,
      "learning_rate": 0.0001959635139053896,
      "loss": 12.5556,
      "step": 4335
    },
    {
      "epoch": 0.2361125291839388,
      "grad_norm": 0.7395264390154621,
      "learning_rate": 0.00019596103341539608,
      "loss": 12.8079,
      "step": 4336
    },
    {
      "epoch": 0.23616698318052182,
      "grad_norm": 0.6869678528143364,
      "learning_rate": 0.00019595855217919347,
      "loss": 12.5308,
      "step": 4337
    },
    {
      "epoch": 0.2362214371771048,
      "grad_norm": 0.7119042798010712,
      "learning_rate": 0.00019595607019680107,
      "loss": 12.8081,
      "step": 4338
    },
    {
      "epoch": 0.23627589117368783,
      "grad_norm": 0.8493247308149601,
      "learning_rate": 0.00019595358746823819,
      "loss": 12.594,
      "step": 4339
    },
    {
      "epoch": 0.23633034517027085,
      "grad_norm": 0.7133131279686575,
      "learning_rate": 0.0001959511039935241,
      "loss": 12.5086,
      "step": 4340
    },
    {
      "epoch": 0.23638479916685384,
      "grad_norm": 0.905017738612647,
      "learning_rate": 0.00019594861977267813,
      "loss": 12.7374,
      "step": 4341
    },
    {
      "epoch": 0.23643925316343686,
      "grad_norm": 0.7011975944578445,
      "learning_rate": 0.0001959461348057196,
      "loss": 12.6633,
      "step": 4342
    },
    {
      "epoch": 0.23649370716001988,
      "grad_norm": 0.6679431058281032,
      "learning_rate": 0.00019594364909266787,
      "loss": 12.5931,
      "step": 4343
    },
    {
      "epoch": 0.23654816115660288,
      "grad_norm": 0.7666953457547615,
      "learning_rate": 0.0001959411626335422,
      "loss": 12.6656,
      "step": 4344
    },
    {
      "epoch": 0.2366026151531859,
      "grad_norm": 0.7048870608679273,
      "learning_rate": 0.00019593867542836197,
      "loss": 12.6733,
      "step": 4345
    },
    {
      "epoch": 0.23665706914976892,
      "grad_norm": 0.6755394346574072,
      "learning_rate": 0.0001959361874771465,
      "loss": 12.5798,
      "step": 4346
    },
    {
      "epoch": 0.2367115231463519,
      "grad_norm": 0.8117541978725121,
      "learning_rate": 0.0001959336987799152,
      "loss": 12.576,
      "step": 4347
    },
    {
      "epoch": 0.23676597714293493,
      "grad_norm": 0.7168325850886029,
      "learning_rate": 0.00019593120933668733,
      "loss": 12.6734,
      "step": 4348
    },
    {
      "epoch": 0.23682043113951795,
      "grad_norm": 0.7886452783719964,
      "learning_rate": 0.00019592871914748229,
      "loss": 12.8169,
      "step": 4349
    },
    {
      "epoch": 0.23687488513610097,
      "grad_norm": 0.6665072365679392,
      "learning_rate": 0.00019592622821231942,
      "loss": 12.574,
      "step": 4350
    },
    {
      "epoch": 0.23692933913268396,
      "grad_norm": 0.7309078222381065,
      "learning_rate": 0.00019592373653121815,
      "loss": 12.5544,
      "step": 4351
    },
    {
      "epoch": 0.23698379312926698,
      "grad_norm": 0.7226790696502614,
      "learning_rate": 0.00019592124410419782,
      "loss": 12.6653,
      "step": 4352
    },
    {
      "epoch": 0.23703824712585,
      "grad_norm": 0.7930188571842032,
      "learning_rate": 0.00019591875093127778,
      "loss": 12.5467,
      "step": 4353
    },
    {
      "epoch": 0.237092701122433,
      "grad_norm": 0.7132902283812247,
      "learning_rate": 0.00019591625701247743,
      "loss": 12.5414,
      "step": 4354
    },
    {
      "epoch": 0.23714715511901602,
      "grad_norm": 0.71476247205427,
      "learning_rate": 0.00019591376234781623,
      "loss": 12.664,
      "step": 4355
    },
    {
      "epoch": 0.23720160911559904,
      "grad_norm": 0.7012311447820085,
      "learning_rate": 0.0001959112669373135,
      "loss": 12.5862,
      "step": 4356
    },
    {
      "epoch": 0.23725606311218203,
      "grad_norm": 0.6501016857711592,
      "learning_rate": 0.0001959087707809887,
      "loss": 12.6829,
      "step": 4357
    },
    {
      "epoch": 0.23731051710876505,
      "grad_norm": 0.6585230561181619,
      "learning_rate": 0.0001959062738788612,
      "loss": 12.6696,
      "step": 4358
    },
    {
      "epoch": 0.23736497110534807,
      "grad_norm": 0.7190576518237434,
      "learning_rate": 0.00019590377623095043,
      "loss": 12.5348,
      "step": 4359
    },
    {
      "epoch": 0.23741942510193106,
      "grad_norm": 0.7866991075164738,
      "learning_rate": 0.0001959012778372758,
      "loss": 12.6122,
      "step": 4360
    },
    {
      "epoch": 0.23747387909851408,
      "grad_norm": 0.6349344258638765,
      "learning_rate": 0.00019589877869785678,
      "loss": 12.518,
      "step": 4361
    },
    {
      "epoch": 0.2375283330950971,
      "grad_norm": 0.7519536264575788,
      "learning_rate": 0.00019589627881271273,
      "loss": 12.7277,
      "step": 4362
    },
    {
      "epoch": 0.2375827870916801,
      "grad_norm": 0.6524831429079407,
      "learning_rate": 0.00019589377818186318,
      "loss": 12.549,
      "step": 4363
    },
    {
      "epoch": 0.23763724108826312,
      "grad_norm": 0.6799023071625376,
      "learning_rate": 0.0001958912768053275,
      "loss": 12.5668,
      "step": 4364
    },
    {
      "epoch": 0.23769169508484614,
      "grad_norm": 0.6664392562505124,
      "learning_rate": 0.00019588877468312518,
      "loss": 12.7095,
      "step": 4365
    },
    {
      "epoch": 0.23774614908142916,
      "grad_norm": 0.6533629567695574,
      "learning_rate": 0.00019588627181527568,
      "loss": 12.5943,
      "step": 4366
    },
    {
      "epoch": 0.23780060307801215,
      "grad_norm": 0.6506573048539561,
      "learning_rate": 0.00019588376820179845,
      "loss": 12.4782,
      "step": 4367
    },
    {
      "epoch": 0.23785505707459517,
      "grad_norm": 0.7732206437895202,
      "learning_rate": 0.00019588126384271294,
      "loss": 12.7024,
      "step": 4368
    },
    {
      "epoch": 0.2379095110711782,
      "grad_norm": 0.7102121516925589,
      "learning_rate": 0.00019587875873803865,
      "loss": 12.5706,
      "step": 4369
    },
    {
      "epoch": 0.23796396506776119,
      "grad_norm": 0.7655504404935409,
      "learning_rate": 0.00019587625288779506,
      "loss": 12.6266,
      "step": 4370
    },
    {
      "epoch": 0.2380184190643442,
      "grad_norm": 0.7034358027577429,
      "learning_rate": 0.00019587374629200164,
      "loss": 12.665,
      "step": 4371
    },
    {
      "epoch": 0.23807287306092723,
      "grad_norm": 0.7218635469509956,
      "learning_rate": 0.0001958712389506779,
      "loss": 12.4119,
      "step": 4372
    },
    {
      "epoch": 0.23812732705751022,
      "grad_norm": 0.7219544872173718,
      "learning_rate": 0.00019586873086384333,
      "loss": 12.6191,
      "step": 4373
    },
    {
      "epoch": 0.23818178105409324,
      "grad_norm": 0.6753360708985948,
      "learning_rate": 0.0001958662220315174,
      "loss": 12.6027,
      "step": 4374
    },
    {
      "epoch": 0.23823623505067626,
      "grad_norm": 0.7042105075748244,
      "learning_rate": 0.0001958637124537197,
      "loss": 12.678,
      "step": 4375
    },
    {
      "epoch": 0.23829068904725925,
      "grad_norm": 0.6937149586216751,
      "learning_rate": 0.00019586120213046964,
      "loss": 12.6869,
      "step": 4376
    },
    {
      "epoch": 0.23834514304384227,
      "grad_norm": 0.7638703568099982,
      "learning_rate": 0.00019585869106178685,
      "loss": 12.8213,
      "step": 4377
    },
    {
      "epoch": 0.2383995970404253,
      "grad_norm": 0.711513978494034,
      "learning_rate": 0.00019585617924769078,
      "loss": 12.6086,
      "step": 4378
    },
    {
      "epoch": 0.23845405103700829,
      "grad_norm": 0.6760817286826006,
      "learning_rate": 0.00019585366668820097,
      "loss": 12.7248,
      "step": 4379
    },
    {
      "epoch": 0.2385085050335913,
      "grad_norm": 0.8389641014052955,
      "learning_rate": 0.000195851153383337,
      "loss": 12.6197,
      "step": 4380
    },
    {
      "epoch": 0.23856295903017433,
      "grad_norm": 0.7277777233565489,
      "learning_rate": 0.00019584863933311836,
      "loss": 12.514,
      "step": 4381
    },
    {
      "epoch": 0.23861741302675732,
      "grad_norm": 0.7616073170124373,
      "learning_rate": 0.00019584612453756465,
      "loss": 12.6506,
      "step": 4382
    },
    {
      "epoch": 0.23867186702334034,
      "grad_norm": 0.6709118088136903,
      "learning_rate": 0.00019584360899669537,
      "loss": 12.7244,
      "step": 4383
    },
    {
      "epoch": 0.23872632101992336,
      "grad_norm": 0.746987612988271,
      "learning_rate": 0.00019584109271053018,
      "loss": 12.5503,
      "step": 4384
    },
    {
      "epoch": 0.23878077501650638,
      "grad_norm": 0.7572221948189825,
      "learning_rate": 0.0001958385756790885,
      "loss": 12.5647,
      "step": 4385
    },
    {
      "epoch": 0.23883522901308937,
      "grad_norm": 0.6949784375654698,
      "learning_rate": 0.00019583605790239004,
      "loss": 12.561,
      "step": 4386
    },
    {
      "epoch": 0.2388896830096724,
      "grad_norm": 0.741463522239718,
      "learning_rate": 0.00019583353938045433,
      "loss": 12.6955,
      "step": 4387
    },
    {
      "epoch": 0.2389441370062554,
      "grad_norm": 0.6497906124337596,
      "learning_rate": 0.0001958310201133009,
      "loss": 12.5737,
      "step": 4388
    },
    {
      "epoch": 0.2389985910028384,
      "grad_norm": 0.7040373728229842,
      "learning_rate": 0.0001958285001009494,
      "loss": 12.6129,
      "step": 4389
    },
    {
      "epoch": 0.23905304499942143,
      "grad_norm": 0.7392310666609192,
      "learning_rate": 0.00019582597934341943,
      "loss": 12.6106,
      "step": 4390
    },
    {
      "epoch": 0.23910749899600445,
      "grad_norm": 0.7254938521643921,
      "learning_rate": 0.00019582345784073058,
      "loss": 12.5225,
      "step": 4391
    },
    {
      "epoch": 0.23916195299258744,
      "grad_norm": 0.694043218706121,
      "learning_rate": 0.00019582093559290242,
      "loss": 12.5082,
      "step": 4392
    },
    {
      "epoch": 0.23921640698917046,
      "grad_norm": 0.6702855650816293,
      "learning_rate": 0.0001958184125999546,
      "loss": 12.6206,
      "step": 4393
    },
    {
      "epoch": 0.23927086098575348,
      "grad_norm": 0.693921757939034,
      "learning_rate": 0.00019581588886190675,
      "loss": 12.4757,
      "step": 4394
    },
    {
      "epoch": 0.23932531498233647,
      "grad_norm": 0.6787515322047294,
      "learning_rate": 0.00019581336437877848,
      "loss": 12.3603,
      "step": 4395
    },
    {
      "epoch": 0.2393797689789195,
      "grad_norm": 0.7769332202563006,
      "learning_rate": 0.0001958108391505894,
      "loss": 12.4286,
      "step": 4396
    },
    {
      "epoch": 0.23943422297550251,
      "grad_norm": 0.7618523412111193,
      "learning_rate": 0.0001958083131773592,
      "loss": 12.5702,
      "step": 4397
    },
    {
      "epoch": 0.2394886769720855,
      "grad_norm": 0.6980552748721157,
      "learning_rate": 0.0001958057864591075,
      "loss": 12.546,
      "step": 4398
    },
    {
      "epoch": 0.23954313096866853,
      "grad_norm": 0.9703323367362652,
      "learning_rate": 0.00019580325899585388,
      "loss": 12.6259,
      "step": 4399
    },
    {
      "epoch": 0.23959758496525155,
      "grad_norm": 0.6874644663477578,
      "learning_rate": 0.0001958007307876181,
      "loss": 12.6013,
      "step": 4400
    },
    {
      "epoch": 0.23965203896183457,
      "grad_norm": 0.7883185675467992,
      "learning_rate": 0.00019579820183441974,
      "loss": 12.6963,
      "step": 4401
    },
    {
      "epoch": 0.23970649295841756,
      "grad_norm": 0.7134853056312774,
      "learning_rate": 0.0001957956721362785,
      "loss": 12.5664,
      "step": 4402
    },
    {
      "epoch": 0.23976094695500058,
      "grad_norm": 0.6907933935357229,
      "learning_rate": 0.0001957931416932141,
      "loss": 12.6162,
      "step": 4403
    },
    {
      "epoch": 0.2398154009515836,
      "grad_norm": 0.6759806570517407,
      "learning_rate": 0.0001957906105052461,
      "loss": 12.4443,
      "step": 4404
    },
    {
      "epoch": 0.2398698549481666,
      "grad_norm": 0.6749819770696659,
      "learning_rate": 0.00019578807857239427,
      "loss": 12.4787,
      "step": 4405
    },
    {
      "epoch": 0.23992430894474961,
      "grad_norm": 0.6790796285000683,
      "learning_rate": 0.0001957855458946783,
      "loss": 12.5896,
      "step": 4406
    },
    {
      "epoch": 0.23997876294133264,
      "grad_norm": 0.8447902766005312,
      "learning_rate": 0.00019578301247211784,
      "loss": 12.7252,
      "step": 4407
    },
    {
      "epoch": 0.24003321693791563,
      "grad_norm": 0.6929824546055435,
      "learning_rate": 0.00019578047830473263,
      "loss": 12.6199,
      "step": 4408
    },
    {
      "epoch": 0.24008767093449865,
      "grad_norm": 0.684029054330614,
      "learning_rate": 0.00019577794339254234,
      "loss": 12.578,
      "step": 4409
    },
    {
      "epoch": 0.24014212493108167,
      "grad_norm": 0.7338430360544661,
      "learning_rate": 0.00019577540773556672,
      "loss": 12.6144,
      "step": 4410
    },
    {
      "epoch": 0.24019657892766466,
      "grad_norm": 0.6985904077182428,
      "learning_rate": 0.0001957728713338254,
      "loss": 12.6199,
      "step": 4411
    },
    {
      "epoch": 0.24025103292424768,
      "grad_norm": 0.6805492339585597,
      "learning_rate": 0.00019577033418733826,
      "loss": 12.6823,
      "step": 4412
    },
    {
      "epoch": 0.2403054869208307,
      "grad_norm": 0.7185307030195718,
      "learning_rate": 0.0001957677962961249,
      "loss": 12.6543,
      "step": 4413
    },
    {
      "epoch": 0.2403599409174137,
      "grad_norm": 0.6971857831409217,
      "learning_rate": 0.0001957652576602051,
      "loss": 12.5886,
      "step": 4414
    },
    {
      "epoch": 0.24041439491399672,
      "grad_norm": 0.6637590622574503,
      "learning_rate": 0.0001957627182795986,
      "loss": 12.6217,
      "step": 4415
    },
    {
      "epoch": 0.24046884891057974,
      "grad_norm": 0.7310212965379054,
      "learning_rate": 0.00019576017815432515,
      "loss": 12.6621,
      "step": 4416
    },
    {
      "epoch": 0.24052330290716276,
      "grad_norm": 0.7393074452512833,
      "learning_rate": 0.0001957576372844045,
      "loss": 12.5586,
      "step": 4417
    },
    {
      "epoch": 0.24057775690374575,
      "grad_norm": 0.7296197794415451,
      "learning_rate": 0.00019575509566985638,
      "loss": 12.569,
      "step": 4418
    },
    {
      "epoch": 0.24063221090032877,
      "grad_norm": 0.7011152828829472,
      "learning_rate": 0.00019575255331070058,
      "loss": 12.5821,
      "step": 4419
    },
    {
      "epoch": 0.2406866648969118,
      "grad_norm": 0.8052641623839756,
      "learning_rate": 0.0001957500102069569,
      "loss": 12.4746,
      "step": 4420
    },
    {
      "epoch": 0.24074111889349478,
      "grad_norm": 0.6879345374295561,
      "learning_rate": 0.00019574746635864506,
      "loss": 12.5922,
      "step": 4421
    },
    {
      "epoch": 0.2407955728900778,
      "grad_norm": 1.0817216594551267,
      "learning_rate": 0.00019574492176578485,
      "loss": 12.6664,
      "step": 4422
    },
    {
      "epoch": 0.24085002688666082,
      "grad_norm": 0.678943951419676,
      "learning_rate": 0.00019574237642839607,
      "loss": 12.6064,
      "step": 4423
    },
    {
      "epoch": 0.24090448088324382,
      "grad_norm": 0.7892394980152566,
      "learning_rate": 0.00019573983034649854,
      "loss": 12.6721,
      "step": 4424
    },
    {
      "epoch": 0.24095893487982684,
      "grad_norm": 0.6887403346294946,
      "learning_rate": 0.00019573728352011204,
      "loss": 12.5072,
      "step": 4425
    },
    {
      "epoch": 0.24101338887640986,
      "grad_norm": 0.7236036893698674,
      "learning_rate": 0.00019573473594925632,
      "loss": 12.7498,
      "step": 4426
    },
    {
      "epoch": 0.24106784287299285,
      "grad_norm": 0.7725445214988449,
      "learning_rate": 0.0001957321876339513,
      "loss": 12.4948,
      "step": 4427
    },
    {
      "epoch": 0.24112229686957587,
      "grad_norm": 0.7523623474627931,
      "learning_rate": 0.0001957296385742167,
      "loss": 12.6427,
      "step": 4428
    },
    {
      "epoch": 0.2411767508661589,
      "grad_norm": 0.7879142777310196,
      "learning_rate": 0.0001957270887700724,
      "loss": 12.6412,
      "step": 4429
    },
    {
      "epoch": 0.24123120486274188,
      "grad_norm": 0.6593665147814708,
      "learning_rate": 0.00019572453822153818,
      "loss": 12.6723,
      "step": 4430
    },
    {
      "epoch": 0.2412856588593249,
      "grad_norm": 0.7554934253466723,
      "learning_rate": 0.0001957219869286339,
      "loss": 12.7087,
      "step": 4431
    },
    {
      "epoch": 0.24134011285590792,
      "grad_norm": 0.6994668680293442,
      "learning_rate": 0.0001957194348913794,
      "loss": 12.3919,
      "step": 4432
    },
    {
      "epoch": 0.24139456685249094,
      "grad_norm": 0.6874407428394054,
      "learning_rate": 0.00019571688210979452,
      "loss": 12.5445,
      "step": 4433
    },
    {
      "epoch": 0.24144902084907394,
      "grad_norm": 0.667137984555039,
      "learning_rate": 0.00019571432858389912,
      "loss": 12.4685,
      "step": 4434
    },
    {
      "epoch": 0.24150347484565696,
      "grad_norm": 0.7996917414516334,
      "learning_rate": 0.00019571177431371303,
      "loss": 12.6403,
      "step": 4435
    },
    {
      "epoch": 0.24155792884223998,
      "grad_norm": 0.6913213809619198,
      "learning_rate": 0.00019570921929925616,
      "loss": 12.7323,
      "step": 4436
    },
    {
      "epoch": 0.24161238283882297,
      "grad_norm": 0.7221294539274075,
      "learning_rate": 0.00019570666354054835,
      "loss": 12.5116,
      "step": 4437
    },
    {
      "epoch": 0.241666836835406,
      "grad_norm": 0.6955443045709032,
      "learning_rate": 0.00019570410703760946,
      "loss": 12.7694,
      "step": 4438
    },
    {
      "epoch": 0.241721290831989,
      "grad_norm": 0.7939884123941371,
      "learning_rate": 0.0001957015497904594,
      "loss": 12.6727,
      "step": 4439
    },
    {
      "epoch": 0.241775744828572,
      "grad_norm": 0.6435736362998078,
      "learning_rate": 0.00019569899179911802,
      "loss": 12.6093,
      "step": 4440
    },
    {
      "epoch": 0.24183019882515502,
      "grad_norm": 0.6785583000701847,
      "learning_rate": 0.00019569643306360524,
      "loss": 12.6087,
      "step": 4441
    },
    {
      "epoch": 0.24188465282173804,
      "grad_norm": 0.8739610450351043,
      "learning_rate": 0.00019569387358394096,
      "loss": 12.5868,
      "step": 4442
    },
    {
      "epoch": 0.24193910681832104,
      "grad_norm": 0.6785969720317832,
      "learning_rate": 0.00019569131336014502,
      "loss": 12.6191,
      "step": 4443
    },
    {
      "epoch": 0.24199356081490406,
      "grad_norm": 0.6706174843742324,
      "learning_rate": 0.00019568875239223745,
      "loss": 12.6506,
      "step": 4444
    },
    {
      "epoch": 0.24204801481148708,
      "grad_norm": 0.6871753042850144,
      "learning_rate": 0.00019568619068023805,
      "loss": 12.6058,
      "step": 4445
    },
    {
      "epoch": 0.24210246880807007,
      "grad_norm": 0.7924294900049751,
      "learning_rate": 0.0001956836282241668,
      "loss": 12.597,
      "step": 4446
    },
    {
      "epoch": 0.2421569228046531,
      "grad_norm": 0.7458698287245629,
      "learning_rate": 0.0001956810650240436,
      "loss": 12.5591,
      "step": 4447
    },
    {
      "epoch": 0.2422113768012361,
      "grad_norm": 0.7321544092410979,
      "learning_rate": 0.0001956785010798884,
      "loss": 12.5925,
      "step": 4448
    },
    {
      "epoch": 0.2422658307978191,
      "grad_norm": 0.7487279437867312,
      "learning_rate": 0.00019567593639172116,
      "loss": 12.8325,
      "step": 4449
    },
    {
      "epoch": 0.24232028479440212,
      "grad_norm": 0.6905942359080257,
      "learning_rate": 0.00019567337095956173,
      "loss": 12.5908,
      "step": 4450
    },
    {
      "epoch": 0.24237473879098514,
      "grad_norm": 0.7374695447895508,
      "learning_rate": 0.00019567080478343017,
      "loss": 12.727,
      "step": 4451
    },
    {
      "epoch": 0.24242919278756817,
      "grad_norm": 0.6704989604779045,
      "learning_rate": 0.00019566823786334637,
      "loss": 12.5303,
      "step": 4452
    },
    {
      "epoch": 0.24248364678415116,
      "grad_norm": 0.8433399654324624,
      "learning_rate": 0.0001956656701993303,
      "loss": 12.7586,
      "step": 4453
    },
    {
      "epoch": 0.24253810078073418,
      "grad_norm": 0.6902272620976142,
      "learning_rate": 0.00019566310179140194,
      "loss": 12.5854,
      "step": 4454
    },
    {
      "epoch": 0.2425925547773172,
      "grad_norm": 0.6414382311660554,
      "learning_rate": 0.0001956605326395813,
      "loss": 12.5247,
      "step": 4455
    },
    {
      "epoch": 0.2426470087739002,
      "grad_norm": 0.6781107751163772,
      "learning_rate": 0.0001956579627438883,
      "loss": 12.6554,
      "step": 4456
    },
    {
      "epoch": 0.2427014627704832,
      "grad_norm": 0.6529822601179021,
      "learning_rate": 0.0001956553921043429,
      "loss": 12.4097,
      "step": 4457
    },
    {
      "epoch": 0.24275591676706623,
      "grad_norm": 0.64671283128718,
      "learning_rate": 0.00019565282072096514,
      "loss": 12.6114,
      "step": 4458
    },
    {
      "epoch": 0.24281037076364922,
      "grad_norm": 0.7302055774093955,
      "learning_rate": 0.000195650248593775,
      "loss": 12.7056,
      "step": 4459
    },
    {
      "epoch": 0.24286482476023225,
      "grad_norm": 0.6869491716617113,
      "learning_rate": 0.00019564767572279252,
      "loss": 12.5807,
      "step": 4460
    },
    {
      "epoch": 0.24291927875681527,
      "grad_norm": 0.7093695704767358,
      "learning_rate": 0.00019564510210803767,
      "loss": 12.6981,
      "step": 4461
    },
    {
      "epoch": 0.24297373275339826,
      "grad_norm": 0.7599144682393941,
      "learning_rate": 0.00019564252774953046,
      "loss": 12.6494,
      "step": 4462
    },
    {
      "epoch": 0.24302818674998128,
      "grad_norm": 0.6510888654090853,
      "learning_rate": 0.00019563995264729092,
      "loss": 12.5583,
      "step": 4463
    },
    {
      "epoch": 0.2430826407465643,
      "grad_norm": 0.6517012342100406,
      "learning_rate": 0.00019563737680133904,
      "loss": 12.5919,
      "step": 4464
    },
    {
      "epoch": 0.2431370947431473,
      "grad_norm": 0.7345404494271482,
      "learning_rate": 0.0001956348002116949,
      "loss": 12.5523,
      "step": 4465
    },
    {
      "epoch": 0.2431915487397303,
      "grad_norm": 0.6549193855626367,
      "learning_rate": 0.0001956322228783785,
      "loss": 12.6657,
      "step": 4466
    },
    {
      "epoch": 0.24324600273631333,
      "grad_norm": 0.713822158827217,
      "learning_rate": 0.00019562964480140992,
      "loss": 12.6496,
      "step": 4467
    },
    {
      "epoch": 0.24330045673289635,
      "grad_norm": 0.6505310466098979,
      "learning_rate": 0.00019562706598080917,
      "loss": 12.6382,
      "step": 4468
    },
    {
      "epoch": 0.24335491072947935,
      "grad_norm": 0.6734696028472843,
      "learning_rate": 0.00019562448641659633,
      "loss": 12.5192,
      "step": 4469
    },
    {
      "epoch": 0.24340936472606237,
      "grad_norm": 0.7173359735388568,
      "learning_rate": 0.00019562190610879142,
      "loss": 12.6257,
      "step": 4470
    },
    {
      "epoch": 0.2434638187226454,
      "grad_norm": 0.6768112396798218,
      "learning_rate": 0.0001956193250574146,
      "loss": 12.4036,
      "step": 4471
    },
    {
      "epoch": 0.24351827271922838,
      "grad_norm": 0.8482952575831432,
      "learning_rate": 0.0001956167432624858,
      "loss": 12.6744,
      "step": 4472
    },
    {
      "epoch": 0.2435727267158114,
      "grad_norm": 0.6896232479270313,
      "learning_rate": 0.0001956141607240252,
      "loss": 12.5744,
      "step": 4473
    },
    {
      "epoch": 0.24362718071239442,
      "grad_norm": 0.7615411170903834,
      "learning_rate": 0.00019561157744205283,
      "loss": 12.5567,
      "step": 4474
    },
    {
      "epoch": 0.2436816347089774,
      "grad_norm": 0.7651799858552987,
      "learning_rate": 0.00019560899341658882,
      "loss": 12.6451,
      "step": 4475
    },
    {
      "epoch": 0.24373608870556043,
      "grad_norm": 0.7505220181840234,
      "learning_rate": 0.00019560640864765326,
      "loss": 12.6047,
      "step": 4476
    },
    {
      "epoch": 0.24379054270214345,
      "grad_norm": 0.7986243578023776,
      "learning_rate": 0.0001956038231352662,
      "loss": 12.5866,
      "step": 4477
    },
    {
      "epoch": 0.24384499669872645,
      "grad_norm": 0.7308259229200806,
      "learning_rate": 0.0001956012368794478,
      "loss": 12.5419,
      "step": 4478
    },
    {
      "epoch": 0.24389945069530947,
      "grad_norm": 0.7536534796161848,
      "learning_rate": 0.00019559864988021814,
      "loss": 12.4244,
      "step": 4479
    },
    {
      "epoch": 0.2439539046918925,
      "grad_norm": 0.7359555051915565,
      "learning_rate": 0.00019559606213759737,
      "loss": 12.6927,
      "step": 4480
    },
    {
      "epoch": 0.24400835868847548,
      "grad_norm": 0.6842061251039289,
      "learning_rate": 0.00019559347365160555,
      "loss": 12.6481,
      "step": 4481
    },
    {
      "epoch": 0.2440628126850585,
      "grad_norm": 0.7077400251306337,
      "learning_rate": 0.00019559088442226287,
      "loss": 12.4308,
      "step": 4482
    },
    {
      "epoch": 0.24411726668164152,
      "grad_norm": 0.7482556729844912,
      "learning_rate": 0.00019558829444958942,
      "loss": 12.5591,
      "step": 4483
    },
    {
      "epoch": 0.24417172067822454,
      "grad_norm": 0.6449707271355242,
      "learning_rate": 0.0001955857037336054,
      "loss": 12.7376,
      "step": 4484
    },
    {
      "epoch": 0.24422617467480753,
      "grad_norm": 0.7326655502768871,
      "learning_rate": 0.00019558311227433088,
      "loss": 12.4296,
      "step": 4485
    },
    {
      "epoch": 0.24428062867139055,
      "grad_norm": 0.7200166033113317,
      "learning_rate": 0.0001955805200717861,
      "loss": 12.6405,
      "step": 4486
    },
    {
      "epoch": 0.24433508266797357,
      "grad_norm": 0.7553742880217637,
      "learning_rate": 0.00019557792712599113,
      "loss": 12.5288,
      "step": 4487
    },
    {
      "epoch": 0.24438953666455657,
      "grad_norm": 1.3477492623330136,
      "learning_rate": 0.00019557533343696616,
      "loss": 12.7448,
      "step": 4488
    },
    {
      "epoch": 0.2444439906611396,
      "grad_norm": 0.747810008635918,
      "learning_rate": 0.00019557273900473138,
      "loss": 12.6993,
      "step": 4489
    },
    {
      "epoch": 0.2444984446577226,
      "grad_norm": 0.7578599870862847,
      "learning_rate": 0.00019557014382930697,
      "loss": 12.5171,
      "step": 4490
    },
    {
      "epoch": 0.2445528986543056,
      "grad_norm": 0.7716666873016084,
      "learning_rate": 0.0001955675479107131,
      "loss": 12.4974,
      "step": 4491
    },
    {
      "epoch": 0.24460735265088862,
      "grad_norm": 0.7491187824359241,
      "learning_rate": 0.0001955649512489699,
      "loss": 12.6716,
      "step": 4492
    },
    {
      "epoch": 0.24466180664747164,
      "grad_norm": 0.7317652042296284,
      "learning_rate": 0.00019556235384409764,
      "loss": 12.5856,
      "step": 4493
    },
    {
      "epoch": 0.24471626064405463,
      "grad_norm": 0.7157174847925366,
      "learning_rate": 0.00019555975569611654,
      "loss": 12.7635,
      "step": 4494
    },
    {
      "epoch": 0.24477071464063765,
      "grad_norm": 0.7177168932371809,
      "learning_rate": 0.00019555715680504666,
      "loss": 12.5829,
      "step": 4495
    },
    {
      "epoch": 0.24482516863722067,
      "grad_norm": 0.7339058397426691,
      "learning_rate": 0.0001955545571709084,
      "loss": 12.6403,
      "step": 4496
    },
    {
      "epoch": 0.24487962263380367,
      "grad_norm": 0.7295417996740731,
      "learning_rate": 0.0001955519567937218,
      "loss": 12.613,
      "step": 4497
    },
    {
      "epoch": 0.2449340766303867,
      "grad_norm": 0.7351610935959223,
      "learning_rate": 0.00019554935567350721,
      "loss": 12.5433,
      "step": 4498
    },
    {
      "epoch": 0.2449885306269697,
      "grad_norm": 0.6982007901341996,
      "learning_rate": 0.00019554675381028478,
      "loss": 12.7112,
      "step": 4499
    },
    {
      "epoch": 0.24504298462355273,
      "grad_norm": 0.6846838175361749,
      "learning_rate": 0.00019554415120407478,
      "loss": 12.5285,
      "step": 4500
    },
    {
      "epoch": 0.24509743862013572,
      "grad_norm": 0.7254197253188015,
      "learning_rate": 0.00019554154785489744,
      "loss": 12.4814,
      "step": 4501
    },
    {
      "epoch": 0.24515189261671874,
      "grad_norm": 0.709088370584864,
      "learning_rate": 0.000195538943762773,
      "loss": 12.5766,
      "step": 4502
    },
    {
      "epoch": 0.24520634661330176,
      "grad_norm": 0.9468716232091118,
      "learning_rate": 0.00019553633892772172,
      "loss": 12.6409,
      "step": 4503
    },
    {
      "epoch": 0.24526080060988475,
      "grad_norm": 0.7480225416160845,
      "learning_rate": 0.00019553373334976385,
      "loss": 12.6486,
      "step": 4504
    },
    {
      "epoch": 0.24531525460646778,
      "grad_norm": 0.8838370936909044,
      "learning_rate": 0.00019553112702891962,
      "loss": 12.545,
      "step": 4505
    },
    {
      "epoch": 0.2453697086030508,
      "grad_norm": 0.7603176328637311,
      "learning_rate": 0.00019552851996520936,
      "loss": 12.6969,
      "step": 4506
    },
    {
      "epoch": 0.2454241625996338,
      "grad_norm": 0.6745237563930223,
      "learning_rate": 0.00019552591215865327,
      "loss": 12.5745,
      "step": 4507
    },
    {
      "epoch": 0.2454786165962168,
      "grad_norm": 0.7399326419492032,
      "learning_rate": 0.0001955233036092717,
      "loss": 12.6872,
      "step": 4508
    },
    {
      "epoch": 0.24553307059279983,
      "grad_norm": 0.7077478699077312,
      "learning_rate": 0.0001955206943170849,
      "loss": 12.5911,
      "step": 4509
    },
    {
      "epoch": 0.24558752458938282,
      "grad_norm": 0.7307766289166041,
      "learning_rate": 0.00019551808428211313,
      "loss": 12.5886,
      "step": 4510
    },
    {
      "epoch": 0.24564197858596584,
      "grad_norm": 0.6607937087585422,
      "learning_rate": 0.00019551547350437677,
      "loss": 12.6056,
      "step": 4511
    },
    {
      "epoch": 0.24569643258254886,
      "grad_norm": 0.6781420170848789,
      "learning_rate": 0.000195512861983896,
      "loss": 12.5309,
      "step": 4512
    },
    {
      "epoch": 0.24575088657913186,
      "grad_norm": 0.7274451215503108,
      "learning_rate": 0.00019551024972069126,
      "loss": 12.6383,
      "step": 4513
    },
    {
      "epoch": 0.24580534057571488,
      "grad_norm": 0.7328547155168079,
      "learning_rate": 0.00019550763671478277,
      "loss": 12.6395,
      "step": 4514
    },
    {
      "epoch": 0.2458597945722979,
      "grad_norm": 0.7098023166777683,
      "learning_rate": 0.00019550502296619089,
      "loss": 12.5694,
      "step": 4515
    },
    {
      "epoch": 0.2459142485688809,
      "grad_norm": 0.8286201301595062,
      "learning_rate": 0.00019550240847493594,
      "loss": 12.6161,
      "step": 4516
    },
    {
      "epoch": 0.2459687025654639,
      "grad_norm": 0.74030169363629,
      "learning_rate": 0.00019549979324103825,
      "loss": 12.6025,
      "step": 4517
    },
    {
      "epoch": 0.24602315656204693,
      "grad_norm": 0.6559365019851384,
      "learning_rate": 0.00019549717726451813,
      "loss": 12.6041,
      "step": 4518
    },
    {
      "epoch": 0.24607761055862995,
      "grad_norm": 0.7240693992512116,
      "learning_rate": 0.00019549456054539596,
      "loss": 12.6527,
      "step": 4519
    },
    {
      "epoch": 0.24613206455521294,
      "grad_norm": 0.8447464162145961,
      "learning_rate": 0.0001954919430836921,
      "loss": 12.5594,
      "step": 4520
    },
    {
      "epoch": 0.24618651855179596,
      "grad_norm": 0.7760926547157477,
      "learning_rate": 0.00019548932487942683,
      "loss": 12.6414,
      "step": 4521
    },
    {
      "epoch": 0.24624097254837898,
      "grad_norm": 0.7862686525620901,
      "learning_rate": 0.00019548670593262055,
      "loss": 12.5695,
      "step": 4522
    },
    {
      "epoch": 0.24629542654496198,
      "grad_norm": 0.6977819603616029,
      "learning_rate": 0.00019548408624329368,
      "loss": 12.5997,
      "step": 4523
    },
    {
      "epoch": 0.246349880541545,
      "grad_norm": 0.7581882072818931,
      "learning_rate": 0.0001954814658114665,
      "loss": 12.5679,
      "step": 4524
    },
    {
      "epoch": 0.24640433453812802,
      "grad_norm": 0.7417220054167826,
      "learning_rate": 0.00019547884463715944,
      "loss": 12.5288,
      "step": 4525
    },
    {
      "epoch": 0.246458788534711,
      "grad_norm": 0.6571353614508388,
      "learning_rate": 0.00019547622272039287,
      "loss": 12.5123,
      "step": 4526
    },
    {
      "epoch": 0.24651324253129403,
      "grad_norm": 0.9819943625994235,
      "learning_rate": 0.0001954736000611872,
      "loss": 12.6171,
      "step": 4527
    },
    {
      "epoch": 0.24656769652787705,
      "grad_norm": 0.8662000801950319,
      "learning_rate": 0.0001954709766595628,
      "loss": 12.702,
      "step": 4528
    },
    {
      "epoch": 0.24662215052446004,
      "grad_norm": 0.7330313779627509,
      "learning_rate": 0.00019546835251554008,
      "loss": 12.6235,
      "step": 4529
    },
    {
      "epoch": 0.24667660452104306,
      "grad_norm": 0.6708599988369802,
      "learning_rate": 0.00019546572762913942,
      "loss": 12.6261,
      "step": 4530
    },
    {
      "epoch": 0.24673105851762608,
      "grad_norm": 0.6875341786134898,
      "learning_rate": 0.00019546310200038125,
      "loss": 12.648,
      "step": 4531
    },
    {
      "epoch": 0.24678551251420908,
      "grad_norm": 0.6884220381444792,
      "learning_rate": 0.000195460475629286,
      "loss": 12.6227,
      "step": 4532
    },
    {
      "epoch": 0.2468399665107921,
      "grad_norm": 0.7260017148269406,
      "learning_rate": 0.0001954578485158741,
      "loss": 12.4933,
      "step": 4533
    },
    {
      "epoch": 0.24689442050737512,
      "grad_norm": 0.7986836441157589,
      "learning_rate": 0.00019545522066016595,
      "loss": 12.5843,
      "step": 4534
    },
    {
      "epoch": 0.24694887450395814,
      "grad_norm": 0.7316985170974813,
      "learning_rate": 0.00019545259206218198,
      "loss": 12.6712,
      "step": 4535
    },
    {
      "epoch": 0.24700332850054113,
      "grad_norm": 0.716498663788949,
      "learning_rate": 0.00019544996272194266,
      "loss": 12.6374,
      "step": 4536
    },
    {
      "epoch": 0.24705778249712415,
      "grad_norm": 0.6871600074292533,
      "learning_rate": 0.00019544733263946845,
      "loss": 12.5411,
      "step": 4537
    },
    {
      "epoch": 0.24711223649370717,
      "grad_norm": 0.8407670283999826,
      "learning_rate": 0.0001954447018147797,
      "loss": 12.6246,
      "step": 4538
    },
    {
      "epoch": 0.24716669049029016,
      "grad_norm": 0.722063707325831,
      "learning_rate": 0.00019544207024789703,
      "loss": 12.4979,
      "step": 4539
    },
    {
      "epoch": 0.24722114448687318,
      "grad_norm": 0.7595660700514336,
      "learning_rate": 0.00019543943793884076,
      "loss": 12.6016,
      "step": 4540
    },
    {
      "epoch": 0.2472755984834562,
      "grad_norm": 0.7803560129403373,
      "learning_rate": 0.00019543680488763143,
      "loss": 12.6529,
      "step": 4541
    },
    {
      "epoch": 0.2473300524800392,
      "grad_norm": 0.6852586791744784,
      "learning_rate": 0.00019543417109428953,
      "loss": 12.6201,
      "step": 4542
    },
    {
      "epoch": 0.24738450647662222,
      "grad_norm": 0.708842264429155,
      "learning_rate": 0.00019543153655883545,
      "loss": 12.6531,
      "step": 4543
    },
    {
      "epoch": 0.24743896047320524,
      "grad_norm": 0.6785694911685014,
      "learning_rate": 0.0001954289012812898,
      "loss": 12.6396,
      "step": 4544
    },
    {
      "epoch": 0.24749341446978823,
      "grad_norm": 0.6822476492481611,
      "learning_rate": 0.00019542626526167296,
      "loss": 12.5853,
      "step": 4545
    },
    {
      "epoch": 0.24754786846637125,
      "grad_norm": 0.6755924097105872,
      "learning_rate": 0.0001954236285000055,
      "loss": 12.5671,
      "step": 4546
    },
    {
      "epoch": 0.24760232246295427,
      "grad_norm": 0.702809518986082,
      "learning_rate": 0.0001954209909963079,
      "loss": 12.5755,
      "step": 4547
    },
    {
      "epoch": 0.24765677645953726,
      "grad_norm": 0.6967556394981288,
      "learning_rate": 0.0001954183527506007,
      "loss": 12.6763,
      "step": 4548
    },
    {
      "epoch": 0.24771123045612028,
      "grad_norm": 0.662645469888712,
      "learning_rate": 0.00019541571376290436,
      "loss": 12.5686,
      "step": 4549
    },
    {
      "epoch": 0.2477656844527033,
      "grad_norm": 0.74349023652389,
      "learning_rate": 0.00019541307403323944,
      "loss": 12.6814,
      "step": 4550
    },
    {
      "epoch": 0.24782013844928633,
      "grad_norm": 0.6806950441026586,
      "learning_rate": 0.00019541043356162643,
      "loss": 12.5481,
      "step": 4551
    },
    {
      "epoch": 0.24787459244586932,
      "grad_norm": 0.6826692501332223,
      "learning_rate": 0.0001954077923480859,
      "loss": 12.522,
      "step": 4552
    },
    {
      "epoch": 0.24792904644245234,
      "grad_norm": 0.7024226341344949,
      "learning_rate": 0.00019540515039263837,
      "loss": 12.5405,
      "step": 4553
    },
    {
      "epoch": 0.24798350043903536,
      "grad_norm": 0.7953597336673244,
      "learning_rate": 0.00019540250769530443,
      "loss": 12.6771,
      "step": 4554
    },
    {
      "epoch": 0.24803795443561835,
      "grad_norm": 0.6986416150802731,
      "learning_rate": 0.00019539986425610453,
      "loss": 12.5993,
      "step": 4555
    },
    {
      "epoch": 0.24809240843220137,
      "grad_norm": 0.7842682390286129,
      "learning_rate": 0.00019539722007505934,
      "loss": 12.6273,
      "step": 4556
    },
    {
      "epoch": 0.2481468624287844,
      "grad_norm": 0.7765687604639168,
      "learning_rate": 0.00019539457515218932,
      "loss": 12.5093,
      "step": 4557
    },
    {
      "epoch": 0.24820131642536739,
      "grad_norm": 0.8003804060961116,
      "learning_rate": 0.00019539192948751514,
      "loss": 12.5759,
      "step": 4558
    },
    {
      "epoch": 0.2482557704219504,
      "grad_norm": 0.7202326885917015,
      "learning_rate": 0.0001953892830810573,
      "loss": 12.4316,
      "step": 4559
    },
    {
      "epoch": 0.24831022441853343,
      "grad_norm": 0.7474728919149864,
      "learning_rate": 0.00019538663593283637,
      "loss": 12.6285,
      "step": 4560
    },
    {
      "epoch": 0.24836467841511642,
      "grad_norm": 0.7812249465629221,
      "learning_rate": 0.00019538398804287298,
      "loss": 12.6414,
      "step": 4561
    },
    {
      "epoch": 0.24841913241169944,
      "grad_norm": 0.693946563301219,
      "learning_rate": 0.00019538133941118772,
      "loss": 12.4846,
      "step": 4562
    },
    {
      "epoch": 0.24847358640828246,
      "grad_norm": 0.7697627760440247,
      "learning_rate": 0.00019537869003780116,
      "loss": 12.6134,
      "step": 4563
    },
    {
      "epoch": 0.24852804040486545,
      "grad_norm": 0.908532305342887,
      "learning_rate": 0.0001953760399227339,
      "loss": 12.6881,
      "step": 4564
    },
    {
      "epoch": 0.24858249440144847,
      "grad_norm": 0.6788580510320924,
      "learning_rate": 0.00019537338906600659,
      "loss": 12.5017,
      "step": 4565
    },
    {
      "epoch": 0.2486369483980315,
      "grad_norm": 0.7197728170770006,
      "learning_rate": 0.00019537073746763977,
      "loss": 12.5242,
      "step": 4566
    },
    {
      "epoch": 0.2486914023946145,
      "grad_norm": 0.8205666944352907,
      "learning_rate": 0.00019536808512765413,
      "loss": 12.7204,
      "step": 4567
    },
    {
      "epoch": 0.2487458563911975,
      "grad_norm": 0.7434051541276757,
      "learning_rate": 0.00019536543204607025,
      "loss": 12.6962,
      "step": 4568
    },
    {
      "epoch": 0.24880031038778053,
      "grad_norm": 0.6981666345210649,
      "learning_rate": 0.00019536277822290878,
      "loss": 12.645,
      "step": 4569
    },
    {
      "epoch": 0.24885476438436355,
      "grad_norm": 0.7823360214580403,
      "learning_rate": 0.00019536012365819038,
      "loss": 12.6003,
      "step": 4570
    },
    {
      "epoch": 0.24890921838094654,
      "grad_norm": 0.7563770621808997,
      "learning_rate": 0.00019535746835193564,
      "loss": 12.6248,
      "step": 4571
    },
    {
      "epoch": 0.24896367237752956,
      "grad_norm": 0.7685321115617461,
      "learning_rate": 0.00019535481230416524,
      "loss": 12.6604,
      "step": 4572
    },
    {
      "epoch": 0.24901812637411258,
      "grad_norm": 0.8625199287410501,
      "learning_rate": 0.00019535215551489982,
      "loss": 12.7927,
      "step": 4573
    },
    {
      "epoch": 0.24907258037069557,
      "grad_norm": 0.8455092054785058,
      "learning_rate": 0.00019534949798416006,
      "loss": 12.6436,
      "step": 4574
    },
    {
      "epoch": 0.2491270343672786,
      "grad_norm": 0.8004646051656875,
      "learning_rate": 0.00019534683971196662,
      "loss": 12.7403,
      "step": 4575
    },
    {
      "epoch": 0.24918148836386161,
      "grad_norm": 0.7167876264537866,
      "learning_rate": 0.00019534418069834013,
      "loss": 12.6044,
      "step": 4576
    },
    {
      "epoch": 0.2492359423604446,
      "grad_norm": 0.7419510607817421,
      "learning_rate": 0.00019534152094330133,
      "loss": 12.581,
      "step": 4577
    },
    {
      "epoch": 0.24929039635702763,
      "grad_norm": 0.8028717993107427,
      "learning_rate": 0.00019533886044687088,
      "loss": 12.5883,
      "step": 4578
    },
    {
      "epoch": 0.24934485035361065,
      "grad_norm": 0.7294878102992232,
      "learning_rate": 0.00019533619920906946,
      "loss": 12.5647,
      "step": 4579
    },
    {
      "epoch": 0.24939930435019364,
      "grad_norm": 0.7266599602586925,
      "learning_rate": 0.00019533353722991776,
      "loss": 12.6332,
      "step": 4580
    },
    {
      "epoch": 0.24945375834677666,
      "grad_norm": 0.7077771439865166,
      "learning_rate": 0.00019533087450943648,
      "loss": 12.6462,
      "step": 4581
    },
    {
      "epoch": 0.24950821234335968,
      "grad_norm": 0.675721997388249,
      "learning_rate": 0.00019532821104764633,
      "loss": 12.706,
      "step": 4582
    },
    {
      "epoch": 0.24956266633994267,
      "grad_norm": 0.729758379877045,
      "learning_rate": 0.00019532554684456805,
      "loss": 12.7759,
      "step": 4583
    },
    {
      "epoch": 0.2496171203365257,
      "grad_norm": 0.8246553097877232,
      "learning_rate": 0.0001953228819002223,
      "loss": 12.5836,
      "step": 4584
    },
    {
      "epoch": 0.24967157433310871,
      "grad_norm": 0.6787022092779801,
      "learning_rate": 0.00019532021621462988,
      "loss": 12.5742,
      "step": 4585
    },
    {
      "epoch": 0.24972602832969173,
      "grad_norm": 0.7525877839664052,
      "learning_rate": 0.00019531754978781141,
      "loss": 12.528,
      "step": 4586
    },
    {
      "epoch": 0.24978048232627473,
      "grad_norm": 0.7884072027096795,
      "learning_rate": 0.00019531488261978773,
      "loss": 12.5593,
      "step": 4587
    },
    {
      "epoch": 0.24983493632285775,
      "grad_norm": 0.7306243201798892,
      "learning_rate": 0.00019531221471057956,
      "loss": 12.6264,
      "step": 4588
    },
    {
      "epoch": 0.24988939031944077,
      "grad_norm": 0.6650271940846968,
      "learning_rate": 0.00019530954606020759,
      "loss": 12.5565,
      "step": 4589
    },
    {
      "epoch": 0.24994384431602376,
      "grad_norm": 0.6988800412957692,
      "learning_rate": 0.0001953068766686926,
      "loss": 12.5774,
      "step": 4590
    },
    {
      "epoch": 0.24999829831260678,
      "grad_norm": 0.8740708821128162,
      "learning_rate": 0.0001953042065360554,
      "loss": 12.8126,
      "step": 4591
    },
    {
      "epoch": 0.2500527523091898,
      "grad_norm": 0.8266690825077698,
      "learning_rate": 0.00019530153566231666,
      "loss": 12.5194,
      "step": 4592
    },
    {
      "epoch": 0.2501072063057728,
      "grad_norm": 0.6552219976399752,
      "learning_rate": 0.00019529886404749723,
      "loss": 12.4998,
      "step": 4593
    },
    {
      "epoch": 0.2501616603023558,
      "grad_norm": 0.7949170552992534,
      "learning_rate": 0.00019529619169161781,
      "loss": 12.6232,
      "step": 4594
    },
    {
      "epoch": 0.2502161142989388,
      "grad_norm": 0.7088863266575556,
      "learning_rate": 0.00019529351859469928,
      "loss": 12.6323,
      "step": 4595
    },
    {
      "epoch": 0.25027056829552186,
      "grad_norm": 0.841859299254048,
      "learning_rate": 0.0001952908447567623,
      "loss": 12.6773,
      "step": 4596
    },
    {
      "epoch": 0.25032502229210485,
      "grad_norm": 0.7756336581104332,
      "learning_rate": 0.00019528817017782778,
      "loss": 12.4028,
      "step": 4597
    },
    {
      "epoch": 0.25037947628868784,
      "grad_norm": 0.7572196940976758,
      "learning_rate": 0.00019528549485791646,
      "loss": 12.5988,
      "step": 4598
    },
    {
      "epoch": 0.2504339302852709,
      "grad_norm": 0.8035920475602101,
      "learning_rate": 0.00019528281879704912,
      "loss": 12.5705,
      "step": 4599
    },
    {
      "epoch": 0.2504883842818539,
      "grad_norm": 0.6541580909787088,
      "learning_rate": 0.00019528014199524663,
      "loss": 12.53,
      "step": 4600
    },
    {
      "epoch": 0.2505428382784369,
      "grad_norm": 0.7846336945332629,
      "learning_rate": 0.0001952774644525298,
      "loss": 12.6758,
      "step": 4601
    },
    {
      "epoch": 0.2505972922750199,
      "grad_norm": 0.7364919279335622,
      "learning_rate": 0.00019527478616891938,
      "loss": 12.5414,
      "step": 4602
    },
    {
      "epoch": 0.2506517462716029,
      "grad_norm": 0.7374226191410725,
      "learning_rate": 0.00019527210714443628,
      "loss": 12.6017,
      "step": 4603
    },
    {
      "epoch": 0.2507062002681859,
      "grad_norm": 0.6828349388848473,
      "learning_rate": 0.00019526942737910127,
      "loss": 12.6738,
      "step": 4604
    },
    {
      "epoch": 0.25076065426476896,
      "grad_norm": 0.7463262104769107,
      "learning_rate": 0.00019526674687293525,
      "loss": 12.655,
      "step": 4605
    },
    {
      "epoch": 0.25081510826135195,
      "grad_norm": 0.6958274601778409,
      "learning_rate": 0.000195264065625959,
      "loss": 12.4563,
      "step": 4606
    },
    {
      "epoch": 0.25086956225793494,
      "grad_norm": 0.8744339011342619,
      "learning_rate": 0.0001952613836381934,
      "loss": 12.6371,
      "step": 4607
    },
    {
      "epoch": 0.250924016254518,
      "grad_norm": 0.7447951983208602,
      "learning_rate": 0.00019525870090965935,
      "loss": 12.5699,
      "step": 4608
    },
    {
      "epoch": 0.250978470251101,
      "grad_norm": 0.8770057055867937,
      "learning_rate": 0.00019525601744037764,
      "loss": 12.6631,
      "step": 4609
    },
    {
      "epoch": 0.25103292424768403,
      "grad_norm": 0.704687414173674,
      "learning_rate": 0.00019525333323036913,
      "loss": 12.4892,
      "step": 4610
    },
    {
      "epoch": 0.251087378244267,
      "grad_norm": 0.8420204054546363,
      "learning_rate": 0.0001952506482796548,
      "loss": 12.5162,
      "step": 4611
    },
    {
      "epoch": 0.25114183224085,
      "grad_norm": 0.7180873889371547,
      "learning_rate": 0.00019524796258825537,
      "loss": 12.6455,
      "step": 4612
    },
    {
      "epoch": 0.25119628623743306,
      "grad_norm": 0.737165717454661,
      "learning_rate": 0.00019524527615619186,
      "loss": 12.7034,
      "step": 4613
    },
    {
      "epoch": 0.25125074023401606,
      "grad_norm": 0.7384097064777905,
      "learning_rate": 0.0001952425889834851,
      "loss": 12.7689,
      "step": 4614
    },
    {
      "epoch": 0.25130519423059905,
      "grad_norm": 0.8840678095617902,
      "learning_rate": 0.00019523990107015598,
      "loss": 12.7451,
      "step": 4615
    },
    {
      "epoch": 0.2513596482271821,
      "grad_norm": 0.7163607304960726,
      "learning_rate": 0.00019523721241622547,
      "loss": 12.4507,
      "step": 4616
    },
    {
      "epoch": 0.2514141022237651,
      "grad_norm": 0.7197221870928445,
      "learning_rate": 0.0001952345230217144,
      "loss": 12.757,
      "step": 4617
    },
    {
      "epoch": 0.2514685562203481,
      "grad_norm": 0.8539264131334994,
      "learning_rate": 0.0001952318328866437,
      "loss": 12.5577,
      "step": 4618
    },
    {
      "epoch": 0.25152301021693113,
      "grad_norm": 0.7491476006380345,
      "learning_rate": 0.00019522914201103428,
      "loss": 12.6483,
      "step": 4619
    },
    {
      "epoch": 0.2515774642135141,
      "grad_norm": 0.6817222614207579,
      "learning_rate": 0.00019522645039490708,
      "loss": 12.4453,
      "step": 4620
    },
    {
      "epoch": 0.2516319182100971,
      "grad_norm": 0.8458457675512254,
      "learning_rate": 0.00019522375803828306,
      "loss": 12.6245,
      "step": 4621
    },
    {
      "epoch": 0.25168637220668016,
      "grad_norm": 0.8751640549465032,
      "learning_rate": 0.0001952210649411831,
      "loss": 12.6798,
      "step": 4622
    },
    {
      "epoch": 0.25174082620326316,
      "grad_norm": 0.753038570022357,
      "learning_rate": 0.0001952183711036282,
      "loss": 12.5469,
      "step": 4623
    },
    {
      "epoch": 0.25179528019984615,
      "grad_norm": 0.7411325111987628,
      "learning_rate": 0.00019521567652563927,
      "loss": 12.592,
      "step": 4624
    },
    {
      "epoch": 0.2518497341964292,
      "grad_norm": 0.6747257568929965,
      "learning_rate": 0.0001952129812072373,
      "loss": 12.4863,
      "step": 4625
    },
    {
      "epoch": 0.2519041881930122,
      "grad_norm": 0.7675323270556783,
      "learning_rate": 0.00019521028514844316,
      "loss": 12.4248,
      "step": 4626
    },
    {
      "epoch": 0.2519586421895952,
      "grad_norm": 0.6906531701123305,
      "learning_rate": 0.00019520758834927788,
      "loss": 12.6437,
      "step": 4627
    },
    {
      "epoch": 0.25201309618617823,
      "grad_norm": 0.8792626963119561,
      "learning_rate": 0.00019520489080976247,
      "loss": 12.6249,
      "step": 4628
    },
    {
      "epoch": 0.2520675501827612,
      "grad_norm": 0.736497661373532,
      "learning_rate": 0.00019520219252991785,
      "loss": 12.5359,
      "step": 4629
    },
    {
      "epoch": 0.2521220041793442,
      "grad_norm": 0.8371944344160387,
      "learning_rate": 0.000195199493509765,
      "loss": 12.6581,
      "step": 4630
    },
    {
      "epoch": 0.25217645817592726,
      "grad_norm": 0.6801403777773931,
      "learning_rate": 0.00019519679374932494,
      "loss": 12.6248,
      "step": 4631
    },
    {
      "epoch": 0.25223091217251026,
      "grad_norm": 0.7314974324886186,
      "learning_rate": 0.00019519409324861864,
      "loss": 12.5213,
      "step": 4632
    },
    {
      "epoch": 0.25228536616909325,
      "grad_norm": 0.7927647348104103,
      "learning_rate": 0.0001951913920076671,
      "loss": 12.551,
      "step": 4633
    },
    {
      "epoch": 0.2523398201656763,
      "grad_norm": 0.652316124095355,
      "learning_rate": 0.00019518869002649135,
      "loss": 12.5875,
      "step": 4634
    },
    {
      "epoch": 0.2523942741622593,
      "grad_norm": 0.7051434214547742,
      "learning_rate": 0.00019518598730511238,
      "loss": 12.6323,
      "step": 4635
    },
    {
      "epoch": 0.2524487281588423,
      "grad_norm": 0.7956144347172418,
      "learning_rate": 0.00019518328384355118,
      "loss": 12.7241,
      "step": 4636
    },
    {
      "epoch": 0.25250318215542533,
      "grad_norm": 0.7341236561572148,
      "learning_rate": 0.00019518057964182882,
      "loss": 12.6638,
      "step": 4637
    },
    {
      "epoch": 0.2525576361520083,
      "grad_norm": 0.8583819376050722,
      "learning_rate": 0.0001951778746999663,
      "loss": 12.5505,
      "step": 4638
    },
    {
      "epoch": 0.2526120901485913,
      "grad_norm": 0.7002431638214149,
      "learning_rate": 0.00019517516901798468,
      "loss": 12.5828,
      "step": 4639
    },
    {
      "epoch": 0.25266654414517437,
      "grad_norm": 0.6704356853347958,
      "learning_rate": 0.00019517246259590502,
      "loss": 12.3791,
      "step": 4640
    },
    {
      "epoch": 0.25272099814175736,
      "grad_norm": 0.776267355947518,
      "learning_rate": 0.0001951697554337483,
      "loss": 12.6886,
      "step": 4641
    },
    {
      "epoch": 0.25277545213834035,
      "grad_norm": 0.6812531539118747,
      "learning_rate": 0.0001951670475315356,
      "loss": 12.5179,
      "step": 4642
    },
    {
      "epoch": 0.2528299061349234,
      "grad_norm": 0.7270298855868224,
      "learning_rate": 0.00019516433888928795,
      "loss": 12.5398,
      "step": 4643
    },
    {
      "epoch": 0.2528843601315064,
      "grad_norm": 0.6768849151005556,
      "learning_rate": 0.00019516162950702649,
      "loss": 12.5489,
      "step": 4644
    },
    {
      "epoch": 0.25293881412808944,
      "grad_norm": 0.6724977998810044,
      "learning_rate": 0.00019515891938477222,
      "loss": 12.5863,
      "step": 4645
    },
    {
      "epoch": 0.25299326812467243,
      "grad_norm": 0.7309444180220243,
      "learning_rate": 0.00019515620852254625,
      "loss": 12.4596,
      "step": 4646
    },
    {
      "epoch": 0.2530477221212554,
      "grad_norm": 0.7977134686558187,
      "learning_rate": 0.00019515349692036962,
      "loss": 12.7878,
      "step": 4647
    },
    {
      "epoch": 0.2531021761178385,
      "grad_norm": 0.6862794292196578,
      "learning_rate": 0.00019515078457826344,
      "loss": 12.4598,
      "step": 4648
    },
    {
      "epoch": 0.25315663011442147,
      "grad_norm": 0.701878486537764,
      "learning_rate": 0.0001951480714962488,
      "loss": 12.5413,
      "step": 4649
    },
    {
      "epoch": 0.25321108411100446,
      "grad_norm": 0.8180141005964334,
      "learning_rate": 0.0001951453576743468,
      "loss": 12.396,
      "step": 4650
    },
    {
      "epoch": 0.2532655381075875,
      "grad_norm": 0.7346431155727067,
      "learning_rate": 0.00019514264311257858,
      "loss": 12.6374,
      "step": 4651
    },
    {
      "epoch": 0.2533199921041705,
      "grad_norm": 0.6830933567935363,
      "learning_rate": 0.00019513992781096517,
      "loss": 12.6301,
      "step": 4652
    },
    {
      "epoch": 0.2533744461007535,
      "grad_norm": 0.716153198686455,
      "learning_rate": 0.00019513721176952776,
      "loss": 12.568,
      "step": 4653
    },
    {
      "epoch": 0.25342890009733654,
      "grad_norm": 0.6666130419809555,
      "learning_rate": 0.00019513449498828738,
      "loss": 12.5816,
      "step": 4654
    },
    {
      "epoch": 0.25348335409391953,
      "grad_norm": 0.7618240849952472,
      "learning_rate": 0.00019513177746726526,
      "loss": 12.5489,
      "step": 4655
    },
    {
      "epoch": 0.2535378080905025,
      "grad_norm": 0.7861808479939798,
      "learning_rate": 0.0001951290592064825,
      "loss": 12.6229,
      "step": 4656
    },
    {
      "epoch": 0.2535922620870856,
      "grad_norm": 0.7792128962577906,
      "learning_rate": 0.00019512634020596022,
      "loss": 12.6316,
      "step": 4657
    },
    {
      "epoch": 0.25364671608366857,
      "grad_norm": 0.728082447715372,
      "learning_rate": 0.00019512362046571953,
      "loss": 12.6125,
      "step": 4658
    },
    {
      "epoch": 0.25370117008025156,
      "grad_norm": 0.6593637366358605,
      "learning_rate": 0.00019512089998578163,
      "loss": 12.4644,
      "step": 4659
    },
    {
      "epoch": 0.2537556240768346,
      "grad_norm": 0.6873917824355797,
      "learning_rate": 0.00019511817876616765,
      "loss": 12.6692,
      "step": 4660
    },
    {
      "epoch": 0.2538100780734176,
      "grad_norm": 0.6782354154408184,
      "learning_rate": 0.00019511545680689878,
      "loss": 12.506,
      "step": 4661
    },
    {
      "epoch": 0.2538645320700006,
      "grad_norm": 0.7415950041617407,
      "learning_rate": 0.00019511273410799615,
      "loss": 12.5421,
      "step": 4662
    },
    {
      "epoch": 0.25391898606658364,
      "grad_norm": 0.767687803641539,
      "learning_rate": 0.00019511001066948097,
      "loss": 12.6216,
      "step": 4663
    },
    {
      "epoch": 0.25397344006316663,
      "grad_norm": 0.856036373749487,
      "learning_rate": 0.00019510728649137438,
      "loss": 12.651,
      "step": 4664
    },
    {
      "epoch": 0.2540278940597496,
      "grad_norm": 0.7352629375371106,
      "learning_rate": 0.0001951045615736976,
      "loss": 12.5666,
      "step": 4665
    },
    {
      "epoch": 0.2540823480563327,
      "grad_norm": 0.7208212681736773,
      "learning_rate": 0.00019510183591647174,
      "loss": 12.6415,
      "step": 4666
    },
    {
      "epoch": 0.25413680205291567,
      "grad_norm": 0.7564664297026775,
      "learning_rate": 0.00019509910951971812,
      "loss": 12.7613,
      "step": 4667
    },
    {
      "epoch": 0.25419125604949866,
      "grad_norm": 0.6759363508762495,
      "learning_rate": 0.00019509638238345787,
      "loss": 12.5443,
      "step": 4668
    },
    {
      "epoch": 0.2542457100460817,
      "grad_norm": 0.7440815182574181,
      "learning_rate": 0.00019509365450771219,
      "loss": 12.6338,
      "step": 4669
    },
    {
      "epoch": 0.2543001640426647,
      "grad_norm": 0.7406671177032623,
      "learning_rate": 0.00019509092589250232,
      "loss": 12.5788,
      "step": 4670
    },
    {
      "epoch": 0.2543546180392477,
      "grad_norm": 0.7009872427479097,
      "learning_rate": 0.00019508819653784942,
      "loss": 12.6131,
      "step": 4671
    },
    {
      "epoch": 0.25440907203583074,
      "grad_norm": 0.7815802022245177,
      "learning_rate": 0.00019508546644377478,
      "loss": 12.7572,
      "step": 4672
    },
    {
      "epoch": 0.25446352603241373,
      "grad_norm": 0.6203163966102102,
      "learning_rate": 0.00019508273561029963,
      "loss": 12.6344,
      "step": 4673
    },
    {
      "epoch": 0.2545179800289967,
      "grad_norm": 0.6469569272437944,
      "learning_rate": 0.00019508000403744517,
      "loss": 12.7187,
      "step": 4674
    },
    {
      "epoch": 0.2545724340255798,
      "grad_norm": 0.709319102432162,
      "learning_rate": 0.00019507727172523264,
      "loss": 12.6195,
      "step": 4675
    },
    {
      "epoch": 0.25462688802216277,
      "grad_norm": 0.6847848179775061,
      "learning_rate": 0.0001950745386736833,
      "loss": 12.717,
      "step": 4676
    },
    {
      "epoch": 0.2546813420187458,
      "grad_norm": 0.6724445808052889,
      "learning_rate": 0.0001950718048828184,
      "loss": 12.4394,
      "step": 4677
    },
    {
      "epoch": 0.2547357960153288,
      "grad_norm": 0.7258117422289937,
      "learning_rate": 0.00019506907035265924,
      "loss": 12.6126,
      "step": 4678
    },
    {
      "epoch": 0.2547902500119118,
      "grad_norm": 0.7029932545816441,
      "learning_rate": 0.000195066335083227,
      "loss": 12.6328,
      "step": 4679
    },
    {
      "epoch": 0.25484470400849485,
      "grad_norm": 0.7363875572775394,
      "learning_rate": 0.00019506359907454302,
      "loss": 12.6315,
      "step": 4680
    },
    {
      "epoch": 0.25489915800507784,
      "grad_norm": 0.7120951164580571,
      "learning_rate": 0.00019506086232662858,
      "loss": 12.4803,
      "step": 4681
    },
    {
      "epoch": 0.25495361200166083,
      "grad_norm": 0.7888478831502377,
      "learning_rate": 0.00019505812483950488,
      "loss": 12.6586,
      "step": 4682
    },
    {
      "epoch": 0.2550080659982439,
      "grad_norm": 0.6638333450862576,
      "learning_rate": 0.00019505538661319328,
      "loss": 12.5237,
      "step": 4683
    },
    {
      "epoch": 0.2550625199948269,
      "grad_norm": 0.6695752920801711,
      "learning_rate": 0.00019505264764771505,
      "loss": 12.5669,
      "step": 4684
    },
    {
      "epoch": 0.25511697399140987,
      "grad_norm": 0.7552995697491214,
      "learning_rate": 0.00019504990794309151,
      "loss": 12.5755,
      "step": 4685
    },
    {
      "epoch": 0.2551714279879929,
      "grad_norm": 0.7076072556809054,
      "learning_rate": 0.00019504716749934394,
      "loss": 12.5666,
      "step": 4686
    },
    {
      "epoch": 0.2552258819845759,
      "grad_norm": 0.7688772490788264,
      "learning_rate": 0.00019504442631649362,
      "loss": 12.7533,
      "step": 4687
    },
    {
      "epoch": 0.2552803359811589,
      "grad_norm": 0.7287168089619401,
      "learning_rate": 0.00019504168439456193,
      "loss": 12.4461,
      "step": 4688
    },
    {
      "epoch": 0.25533478997774195,
      "grad_norm": 0.7621658441592705,
      "learning_rate": 0.00019503894173357017,
      "loss": 12.4661,
      "step": 4689
    },
    {
      "epoch": 0.25538924397432494,
      "grad_norm": 0.6661750509434194,
      "learning_rate": 0.00019503619833353966,
      "loss": 12.5776,
      "step": 4690
    },
    {
      "epoch": 0.25544369797090793,
      "grad_norm": 0.8341339017909717,
      "learning_rate": 0.00019503345419449172,
      "loss": 12.7292,
      "step": 4691
    },
    {
      "epoch": 0.255498151967491,
      "grad_norm": 0.7358515877426577,
      "learning_rate": 0.0001950307093164477,
      "loss": 12.6913,
      "step": 4692
    },
    {
      "epoch": 0.255552605964074,
      "grad_norm": 0.7633606014871572,
      "learning_rate": 0.00019502796369942895,
      "loss": 12.5919,
      "step": 4693
    },
    {
      "epoch": 0.25560705996065697,
      "grad_norm": 0.7378217364635543,
      "learning_rate": 0.00019502521734345685,
      "loss": 12.6138,
      "step": 4694
    },
    {
      "epoch": 0.25566151395724,
      "grad_norm": 0.7411934737877932,
      "learning_rate": 0.00019502247024855268,
      "loss": 12.5775,
      "step": 4695
    },
    {
      "epoch": 0.255715967953823,
      "grad_norm": 0.6504062988581084,
      "learning_rate": 0.00019501972241473786,
      "loss": 12.7351,
      "step": 4696
    },
    {
      "epoch": 0.255770421950406,
      "grad_norm": 0.7644212366301545,
      "learning_rate": 0.00019501697384203376,
      "loss": 12.5324,
      "step": 4697
    },
    {
      "epoch": 0.25582487594698905,
      "grad_norm": 0.6450263672882387,
      "learning_rate": 0.00019501422453046174,
      "loss": 12.4571,
      "step": 4698
    },
    {
      "epoch": 0.25587932994357204,
      "grad_norm": 0.7020161938921881,
      "learning_rate": 0.00019501147448004318,
      "loss": 12.5161,
      "step": 4699
    },
    {
      "epoch": 0.25593378394015504,
      "grad_norm": 0.6694991425623197,
      "learning_rate": 0.00019500872369079944,
      "loss": 12.5092,
      "step": 4700
    },
    {
      "epoch": 0.2559882379367381,
      "grad_norm": 0.6472421122968421,
      "learning_rate": 0.0001950059721627519,
      "loss": 12.4466,
      "step": 4701
    },
    {
      "epoch": 0.2560426919333211,
      "grad_norm": 0.6765932649098179,
      "learning_rate": 0.00019500321989592204,
      "loss": 12.5659,
      "step": 4702
    },
    {
      "epoch": 0.25609714592990407,
      "grad_norm": 0.7110563531735874,
      "learning_rate": 0.0001950004668903312,
      "loss": 12.5553,
      "step": 4703
    },
    {
      "epoch": 0.2561515999264871,
      "grad_norm": 0.729546587513424,
      "learning_rate": 0.0001949977131460008,
      "loss": 12.5421,
      "step": 4704
    },
    {
      "epoch": 0.2562060539230701,
      "grad_norm": 0.6992897637338722,
      "learning_rate": 0.00019499495866295225,
      "loss": 12.6526,
      "step": 4705
    },
    {
      "epoch": 0.2562605079196531,
      "grad_norm": 0.7222729187165062,
      "learning_rate": 0.00019499220344120697,
      "loss": 12.6912,
      "step": 4706
    },
    {
      "epoch": 0.25631496191623615,
      "grad_norm": 0.8148839595759461,
      "learning_rate": 0.00019498944748078638,
      "loss": 12.6897,
      "step": 4707
    },
    {
      "epoch": 0.25636941591281914,
      "grad_norm": 0.6505194283908418,
      "learning_rate": 0.0001949866907817119,
      "loss": 12.553,
      "step": 4708
    },
    {
      "epoch": 0.25642386990940214,
      "grad_norm": 0.753503763796297,
      "learning_rate": 0.000194983933344005,
      "loss": 12.5926,
      "step": 4709
    },
    {
      "epoch": 0.2564783239059852,
      "grad_norm": 0.7228660846866438,
      "learning_rate": 0.0001949811751676871,
      "loss": 12.5322,
      "step": 4710
    },
    {
      "epoch": 0.2565327779025682,
      "grad_norm": 0.749767652793384,
      "learning_rate": 0.00019497841625277967,
      "loss": 12.5987,
      "step": 4711
    },
    {
      "epoch": 0.2565872318991512,
      "grad_norm": 0.7264752568702961,
      "learning_rate": 0.00019497565659930413,
      "loss": 12.6687,
      "step": 4712
    },
    {
      "epoch": 0.2566416858957342,
      "grad_norm": 0.7555201243456778,
      "learning_rate": 0.00019497289620728196,
      "loss": 12.5931,
      "step": 4713
    },
    {
      "epoch": 0.2566961398923172,
      "grad_norm": 0.7307088174451337,
      "learning_rate": 0.00019497013507673464,
      "loss": 12.6921,
      "step": 4714
    },
    {
      "epoch": 0.25675059388890026,
      "grad_norm": 0.7658765194364088,
      "learning_rate": 0.00019496737320768358,
      "loss": 12.6566,
      "step": 4715
    },
    {
      "epoch": 0.25680504788548325,
      "grad_norm": 0.8126722060344513,
      "learning_rate": 0.00019496461060015036,
      "loss": 12.6151,
      "step": 4716
    },
    {
      "epoch": 0.25685950188206624,
      "grad_norm": 0.6542168082813249,
      "learning_rate": 0.00019496184725415635,
      "loss": 12.4329,
      "step": 4717
    },
    {
      "epoch": 0.2569139558786493,
      "grad_norm": 0.6634464108209316,
      "learning_rate": 0.00019495908316972314,
      "loss": 12.4756,
      "step": 4718
    },
    {
      "epoch": 0.2569684098752323,
      "grad_norm": 0.7463975669736462,
      "learning_rate": 0.00019495631834687212,
      "loss": 12.6905,
      "step": 4719
    },
    {
      "epoch": 0.2570228638718153,
      "grad_norm": 0.7573454303384313,
      "learning_rate": 0.00019495355278562488,
      "loss": 12.7064,
      "step": 4720
    },
    {
      "epoch": 0.2570773178683983,
      "grad_norm": 0.734300825979996,
      "learning_rate": 0.00019495078648600287,
      "loss": 12.6519,
      "step": 4721
    },
    {
      "epoch": 0.2571317718649813,
      "grad_norm": 0.6784505364892656,
      "learning_rate": 0.00019494801944802762,
      "loss": 12.4599,
      "step": 4722
    },
    {
      "epoch": 0.2571862258615643,
      "grad_norm": 0.8001786847577275,
      "learning_rate": 0.00019494525167172068,
      "loss": 12.7742,
      "step": 4723
    },
    {
      "epoch": 0.25724067985814736,
      "grad_norm": 0.6908462600955748,
      "learning_rate": 0.00019494248315710352,
      "loss": 12.6113,
      "step": 4724
    },
    {
      "epoch": 0.25729513385473035,
      "grad_norm": 0.6502127393656697,
      "learning_rate": 0.0001949397139041977,
      "loss": 12.2824,
      "step": 4725
    },
    {
      "epoch": 0.25734958785131334,
      "grad_norm": 0.7030919669677971,
      "learning_rate": 0.00019493694391302472,
      "loss": 12.6083,
      "step": 4726
    },
    {
      "epoch": 0.2574040418478964,
      "grad_norm": 0.6798754788609441,
      "learning_rate": 0.00019493417318360617,
      "loss": 12.4886,
      "step": 4727
    },
    {
      "epoch": 0.2574584958444794,
      "grad_norm": 0.7230420580117327,
      "learning_rate": 0.00019493140171596355,
      "loss": 12.6009,
      "step": 4728
    },
    {
      "epoch": 0.2575129498410624,
      "grad_norm": 0.720929907434017,
      "learning_rate": 0.00019492862951011843,
      "loss": 12.6411,
      "step": 4729
    },
    {
      "epoch": 0.2575674038376454,
      "grad_norm": 0.7651425086653374,
      "learning_rate": 0.00019492585656609237,
      "loss": 12.6533,
      "step": 4730
    },
    {
      "epoch": 0.2576218578342284,
      "grad_norm": 0.8158447242348982,
      "learning_rate": 0.00019492308288390694,
      "loss": 12.69,
      "step": 4731
    },
    {
      "epoch": 0.2576763118308114,
      "grad_norm": 0.7244070048646627,
      "learning_rate": 0.00019492030846358368,
      "loss": 12.6977,
      "step": 4732
    },
    {
      "epoch": 0.25773076582739446,
      "grad_norm": 0.7611098891379391,
      "learning_rate": 0.0001949175333051442,
      "loss": 12.7027,
      "step": 4733
    },
    {
      "epoch": 0.25778521982397745,
      "grad_norm": 0.7838925677761639,
      "learning_rate": 0.00019491475740861006,
      "loss": 12.607,
      "step": 4734
    },
    {
      "epoch": 0.25783967382056044,
      "grad_norm": 0.6918956552390313,
      "learning_rate": 0.00019491198077400284,
      "loss": 12.5875,
      "step": 4735
    },
    {
      "epoch": 0.2578941278171435,
      "grad_norm": 0.8093985749932328,
      "learning_rate": 0.00019490920340134416,
      "loss": 12.691,
      "step": 4736
    },
    {
      "epoch": 0.2579485818137265,
      "grad_norm": 0.7667091966816839,
      "learning_rate": 0.00019490642529065556,
      "loss": 12.6538,
      "step": 4737
    },
    {
      "epoch": 0.2580030358103095,
      "grad_norm": 0.8050816124530826,
      "learning_rate": 0.00019490364644195873,
      "loss": 12.6441,
      "step": 4738
    },
    {
      "epoch": 0.2580574898068925,
      "grad_norm": 0.7796984412669973,
      "learning_rate": 0.0001949008668552752,
      "loss": 12.4625,
      "step": 4739
    },
    {
      "epoch": 0.2581119438034755,
      "grad_norm": 0.7954269326095742,
      "learning_rate": 0.00019489808653062662,
      "loss": 12.6378,
      "step": 4740
    },
    {
      "epoch": 0.2581663978000585,
      "grad_norm": 0.8546297428341599,
      "learning_rate": 0.0001948953054680346,
      "loss": 12.6656,
      "step": 4741
    },
    {
      "epoch": 0.25822085179664156,
      "grad_norm": 0.6836882241215804,
      "learning_rate": 0.0001948925236675208,
      "loss": 12.6027,
      "step": 4742
    },
    {
      "epoch": 0.25827530579322455,
      "grad_norm": 0.775457407414149,
      "learning_rate": 0.00019488974112910677,
      "loss": 12.6133,
      "step": 4743
    },
    {
      "epoch": 0.2583297597898076,
      "grad_norm": 0.6811598006775803,
      "learning_rate": 0.00019488695785281425,
      "loss": 12.6013,
      "step": 4744
    },
    {
      "epoch": 0.2583842137863906,
      "grad_norm": 0.8090432005026313,
      "learning_rate": 0.00019488417383866483,
      "loss": 12.7681,
      "step": 4745
    },
    {
      "epoch": 0.2584386677829736,
      "grad_norm": 0.9336443651864992,
      "learning_rate": 0.00019488138908668013,
      "loss": 12.7671,
      "step": 4746
    },
    {
      "epoch": 0.25849312177955663,
      "grad_norm": 0.6696001900123343,
      "learning_rate": 0.00019487860359688184,
      "loss": 12.6525,
      "step": 4747
    },
    {
      "epoch": 0.2585475757761396,
      "grad_norm": 0.7765604683795253,
      "learning_rate": 0.00019487581736929164,
      "loss": 12.563,
      "step": 4748
    },
    {
      "epoch": 0.2586020297727226,
      "grad_norm": 0.716271800511286,
      "learning_rate": 0.00019487303040393114,
      "loss": 12.6551,
      "step": 4749
    },
    {
      "epoch": 0.25865648376930567,
      "grad_norm": 0.6616733783918025,
      "learning_rate": 0.00019487024270082207,
      "loss": 12.5925,
      "step": 4750
    },
    {
      "epoch": 0.25871093776588866,
      "grad_norm": 0.7618926859107245,
      "learning_rate": 0.0001948674542599861,
      "loss": 12.6893,
      "step": 4751
    },
    {
      "epoch": 0.25876539176247165,
      "grad_norm": 0.747910560094233,
      "learning_rate": 0.00019486466508144488,
      "loss": 12.6486,
      "step": 4752
    },
    {
      "epoch": 0.2588198457590547,
      "grad_norm": 0.8001813888712213,
      "learning_rate": 0.0001948618751652201,
      "loss": 12.5778,
      "step": 4753
    },
    {
      "epoch": 0.2588742997556377,
      "grad_norm": 0.7698063187849554,
      "learning_rate": 0.00019485908451133348,
      "loss": 12.577,
      "step": 4754
    },
    {
      "epoch": 0.2589287537522207,
      "grad_norm": 0.7181274213017971,
      "learning_rate": 0.0001948562931198067,
      "loss": 12.5075,
      "step": 4755
    },
    {
      "epoch": 0.25898320774880373,
      "grad_norm": 0.815996907375043,
      "learning_rate": 0.00019485350099066154,
      "loss": 12.7174,
      "step": 4756
    },
    {
      "epoch": 0.2590376617453867,
      "grad_norm": 0.8277762565865211,
      "learning_rate": 0.00019485070812391957,
      "loss": 12.6259,
      "step": 4757
    },
    {
      "epoch": 0.2590921157419697,
      "grad_norm": 0.8255336053710628,
      "learning_rate": 0.00019484791451960262,
      "loss": 12.5534,
      "step": 4758
    },
    {
      "epoch": 0.25914656973855277,
      "grad_norm": 0.8046986799888131,
      "learning_rate": 0.00019484512017773237,
      "loss": 12.5145,
      "step": 4759
    },
    {
      "epoch": 0.25920102373513576,
      "grad_norm": 0.7738111831193657,
      "learning_rate": 0.00019484232509833058,
      "loss": 12.663,
      "step": 4760
    },
    {
      "epoch": 0.25925547773171875,
      "grad_norm": 0.7868122530602073,
      "learning_rate": 0.00019483952928141894,
      "loss": 12.5728,
      "step": 4761
    },
    {
      "epoch": 0.2593099317283018,
      "grad_norm": 0.8252109638843789,
      "learning_rate": 0.00019483673272701927,
      "loss": 12.6221,
      "step": 4762
    },
    {
      "epoch": 0.2593643857248848,
      "grad_norm": 0.752918560811296,
      "learning_rate": 0.00019483393543515322,
      "loss": 12.5168,
      "step": 4763
    },
    {
      "epoch": 0.2594188397214678,
      "grad_norm": 0.6652628886031119,
      "learning_rate": 0.00019483113740584256,
      "loss": 12.5479,
      "step": 4764
    },
    {
      "epoch": 0.25947329371805083,
      "grad_norm": 0.8406383840699488,
      "learning_rate": 0.0001948283386391091,
      "loss": 12.5782,
      "step": 4765
    },
    {
      "epoch": 0.2595277477146338,
      "grad_norm": 0.6806302328159711,
      "learning_rate": 0.00019482553913497457,
      "loss": 12.5683,
      "step": 4766
    },
    {
      "epoch": 0.2595822017112168,
      "grad_norm": 0.6847797554348255,
      "learning_rate": 0.00019482273889346075,
      "loss": 12.4412,
      "step": 4767
    },
    {
      "epoch": 0.25963665570779987,
      "grad_norm": 0.7825591587553425,
      "learning_rate": 0.0001948199379145894,
      "loss": 12.5548,
      "step": 4768
    },
    {
      "epoch": 0.25969110970438286,
      "grad_norm": 0.6803581758239637,
      "learning_rate": 0.00019481713619838234,
      "loss": 12.6537,
      "step": 4769
    },
    {
      "epoch": 0.25974556370096585,
      "grad_norm": 0.8865961981147894,
      "learning_rate": 0.0001948143337448613,
      "loss": 12.716,
      "step": 4770
    },
    {
      "epoch": 0.2598000176975489,
      "grad_norm": 0.8012560988317154,
      "learning_rate": 0.0001948115305540481,
      "loss": 12.6787,
      "step": 4771
    },
    {
      "epoch": 0.2598544716941319,
      "grad_norm": 0.7402747722806818,
      "learning_rate": 0.00019480872662596457,
      "loss": 12.5813,
      "step": 4772
    },
    {
      "epoch": 0.2599089256907149,
      "grad_norm": 0.7862461301402442,
      "learning_rate": 0.00019480592196063245,
      "loss": 12.6447,
      "step": 4773
    },
    {
      "epoch": 0.25996337968729794,
      "grad_norm": 0.6896585549357185,
      "learning_rate": 0.0001948031165580736,
      "loss": 12.6612,
      "step": 4774
    },
    {
      "epoch": 0.26001783368388093,
      "grad_norm": 0.7165201707189586,
      "learning_rate": 0.0001948003104183098,
      "loss": 12.6166,
      "step": 4775
    },
    {
      "epoch": 0.2600722876804639,
      "grad_norm": 0.772230066475727,
      "learning_rate": 0.0001947975035413629,
      "loss": 12.5939,
      "step": 4776
    },
    {
      "epoch": 0.26012674167704697,
      "grad_norm": 0.6933004562202322,
      "learning_rate": 0.0001947946959272547,
      "loss": 12.6642,
      "step": 4777
    },
    {
      "epoch": 0.26018119567362996,
      "grad_norm": 0.7647195199286283,
      "learning_rate": 0.0001947918875760071,
      "loss": 12.5439,
      "step": 4778
    },
    {
      "epoch": 0.260235649670213,
      "grad_norm": 0.7889058775371521,
      "learning_rate": 0.00019478907848764182,
      "loss": 12.7009,
      "step": 4779
    },
    {
      "epoch": 0.260290103666796,
      "grad_norm": 0.7187506723561059,
      "learning_rate": 0.0001947862686621808,
      "loss": 12.5066,
      "step": 4780
    },
    {
      "epoch": 0.260344557663379,
      "grad_norm": 0.6356252354910998,
      "learning_rate": 0.00019478345809964583,
      "loss": 12.4976,
      "step": 4781
    },
    {
      "epoch": 0.26039901165996204,
      "grad_norm": 0.8288372345791144,
      "learning_rate": 0.00019478064680005885,
      "loss": 12.6177,
      "step": 4782
    },
    {
      "epoch": 0.26045346565654504,
      "grad_norm": 0.7627296418926722,
      "learning_rate": 0.00019477783476344162,
      "loss": 12.6029,
      "step": 4783
    },
    {
      "epoch": 0.26050791965312803,
      "grad_norm": 0.6565336081551778,
      "learning_rate": 0.0001947750219898161,
      "loss": 12.4271,
      "step": 4784
    },
    {
      "epoch": 0.2605623736497111,
      "grad_norm": 0.6840268317999146,
      "learning_rate": 0.00019477220847920405,
      "loss": 12.573,
      "step": 4785
    },
    {
      "epoch": 0.26061682764629407,
      "grad_norm": 0.728386555080175,
      "learning_rate": 0.00019476939423162745,
      "loss": 12.696,
      "step": 4786
    },
    {
      "epoch": 0.26067128164287706,
      "grad_norm": 0.7978933441928162,
      "learning_rate": 0.00019476657924710815,
      "loss": 12.5638,
      "step": 4787
    },
    {
      "epoch": 0.2607257356394601,
      "grad_norm": 0.8473115840042355,
      "learning_rate": 0.00019476376352566804,
      "loss": 12.4955,
      "step": 4788
    },
    {
      "epoch": 0.2607801896360431,
      "grad_norm": 0.7401753572573325,
      "learning_rate": 0.000194760947067329,
      "loss": 12.6471,
      "step": 4789
    },
    {
      "epoch": 0.2608346436326261,
      "grad_norm": 0.7178835390453441,
      "learning_rate": 0.00019475812987211294,
      "loss": 12.4338,
      "step": 4790
    },
    {
      "epoch": 0.26088909762920914,
      "grad_norm": 0.8262283811762116,
      "learning_rate": 0.00019475531194004176,
      "loss": 12.7366,
      "step": 4791
    },
    {
      "epoch": 0.26094355162579214,
      "grad_norm": 0.6845621603784252,
      "learning_rate": 0.00019475249327113742,
      "loss": 12.6441,
      "step": 4792
    },
    {
      "epoch": 0.26099800562237513,
      "grad_norm": 0.7739572980317966,
      "learning_rate": 0.0001947496738654218,
      "loss": 12.493,
      "step": 4793
    },
    {
      "epoch": 0.2610524596189582,
      "grad_norm": 0.6352423477991079,
      "learning_rate": 0.0001947468537229168,
      "loss": 12.5679,
      "step": 4794
    },
    {
      "epoch": 0.26110691361554117,
      "grad_norm": 0.6745795145399036,
      "learning_rate": 0.0001947440328436444,
      "loss": 12.5574,
      "step": 4795
    },
    {
      "epoch": 0.26116136761212416,
      "grad_norm": 0.6633490884387779,
      "learning_rate": 0.0001947412112276265,
      "loss": 12.5556,
      "step": 4796
    },
    {
      "epoch": 0.2612158216087072,
      "grad_norm": 0.6370335086578439,
      "learning_rate": 0.00019473838887488506,
      "loss": 12.4562,
      "step": 4797
    },
    {
      "epoch": 0.2612702756052902,
      "grad_norm": 0.6154624229446888,
      "learning_rate": 0.00019473556578544201,
      "loss": 12.4528,
      "step": 4798
    },
    {
      "epoch": 0.2613247296018732,
      "grad_norm": 0.7574987473958148,
      "learning_rate": 0.00019473274195931932,
      "loss": 12.5948,
      "step": 4799
    },
    {
      "epoch": 0.26137918359845624,
      "grad_norm": 0.660503906328179,
      "learning_rate": 0.00019472991739653893,
      "loss": 12.5449,
      "step": 4800
    },
    {
      "epoch": 0.26143363759503924,
      "grad_norm": 0.680312990012292,
      "learning_rate": 0.00019472709209712282,
      "loss": 12.599,
      "step": 4801
    },
    {
      "epoch": 0.26148809159162223,
      "grad_norm": 0.6525863345468783,
      "learning_rate": 0.00019472426606109299,
      "loss": 12.5575,
      "step": 4802
    },
    {
      "epoch": 0.2615425455882053,
      "grad_norm": 0.719565202973381,
      "learning_rate": 0.00019472143928847134,
      "loss": 12.6364,
      "step": 4803
    },
    {
      "epoch": 0.26159699958478827,
      "grad_norm": 0.6854658367976487,
      "learning_rate": 0.0001947186117792799,
      "loss": 12.6155,
      "step": 4804
    },
    {
      "epoch": 0.26165145358137126,
      "grad_norm": 0.6462389765626788,
      "learning_rate": 0.00019471578353354066,
      "loss": 12.4988,
      "step": 4805
    },
    {
      "epoch": 0.2617059075779543,
      "grad_norm": 0.7059880398408249,
      "learning_rate": 0.0001947129545512756,
      "loss": 12.6068,
      "step": 4806
    },
    {
      "epoch": 0.2617603615745373,
      "grad_norm": 0.6738481299117473,
      "learning_rate": 0.00019471012483250673,
      "loss": 12.6121,
      "step": 4807
    },
    {
      "epoch": 0.2618148155711203,
      "grad_norm": 0.7047780449464357,
      "learning_rate": 0.00019470729437725604,
      "loss": 12.5718,
      "step": 4808
    },
    {
      "epoch": 0.26186926956770334,
      "grad_norm": 0.6574646377296559,
      "learning_rate": 0.00019470446318554553,
      "loss": 12.5995,
      "step": 4809
    },
    {
      "epoch": 0.26192372356428634,
      "grad_norm": 0.9005206763366099,
      "learning_rate": 0.00019470163125739727,
      "loss": 12.6679,
      "step": 4810
    },
    {
      "epoch": 0.2619781775608694,
      "grad_norm": 0.7396838483648821,
      "learning_rate": 0.0001946987985928332,
      "loss": 12.5763,
      "step": 4811
    },
    {
      "epoch": 0.2620326315574524,
      "grad_norm": 0.7708475776756964,
      "learning_rate": 0.00019469596519187542,
      "loss": 12.6198,
      "step": 4812
    },
    {
      "epoch": 0.26208708555403537,
      "grad_norm": 0.7720119216448281,
      "learning_rate": 0.00019469313105454595,
      "loss": 12.6241,
      "step": 4813
    },
    {
      "epoch": 0.2621415395506184,
      "grad_norm": 0.7334360755883025,
      "learning_rate": 0.00019469029618086677,
      "loss": 12.6099,
      "step": 4814
    },
    {
      "epoch": 0.2621959935472014,
      "grad_norm": 0.7044540805944692,
      "learning_rate": 0.00019468746057086002,
      "loss": 12.5668,
      "step": 4815
    },
    {
      "epoch": 0.2622504475437844,
      "grad_norm": 0.7207583424255852,
      "learning_rate": 0.00019468462422454766,
      "loss": 12.6091,
      "step": 4816
    },
    {
      "epoch": 0.26230490154036745,
      "grad_norm": 0.7101695408830705,
      "learning_rate": 0.00019468178714195179,
      "loss": 12.6525,
      "step": 4817
    },
    {
      "epoch": 0.26235935553695044,
      "grad_norm": 0.6242627094997542,
      "learning_rate": 0.00019467894932309444,
      "loss": 12.5819,
      "step": 4818
    },
    {
      "epoch": 0.26241380953353344,
      "grad_norm": 0.757938269214848,
      "learning_rate": 0.00019467611076799774,
      "loss": 12.6262,
      "step": 4819
    },
    {
      "epoch": 0.2624682635301165,
      "grad_norm": 0.6433189196932046,
      "learning_rate": 0.00019467327147668371,
      "loss": 12.502,
      "step": 4820
    },
    {
      "epoch": 0.2625227175266995,
      "grad_norm": 0.6852794377013726,
      "learning_rate": 0.00019467043144917443,
      "loss": 12.6259,
      "step": 4821
    },
    {
      "epoch": 0.26257717152328247,
      "grad_norm": 0.7833380402320976,
      "learning_rate": 0.000194667590685492,
      "loss": 12.6435,
      "step": 4822
    },
    {
      "epoch": 0.2626316255198655,
      "grad_norm": 0.6945606666550747,
      "learning_rate": 0.00019466474918565854,
      "loss": 12.4367,
      "step": 4823
    },
    {
      "epoch": 0.2626860795164485,
      "grad_norm": 0.7123898769033836,
      "learning_rate": 0.00019466190694969612,
      "loss": 12.509,
      "step": 4824
    },
    {
      "epoch": 0.2627405335130315,
      "grad_norm": 0.7654131241421731,
      "learning_rate": 0.00019465906397762682,
      "loss": 12.495,
      "step": 4825
    },
    {
      "epoch": 0.26279498750961455,
      "grad_norm": 0.8301231959722061,
      "learning_rate": 0.00019465622026947275,
      "loss": 12.757,
      "step": 4826
    },
    {
      "epoch": 0.26284944150619755,
      "grad_norm": 0.6288940118080169,
      "learning_rate": 0.00019465337582525604,
      "loss": 12.4903,
      "step": 4827
    },
    {
      "epoch": 0.26290389550278054,
      "grad_norm": 0.7417776345417484,
      "learning_rate": 0.0001946505306449988,
      "loss": 12.5361,
      "step": 4828
    },
    {
      "epoch": 0.2629583494993636,
      "grad_norm": 0.7752400517739513,
      "learning_rate": 0.00019464768472872318,
      "loss": 12.6089,
      "step": 4829
    },
    {
      "epoch": 0.2630128034959466,
      "grad_norm": 0.5998676948980813,
      "learning_rate": 0.00019464483807645128,
      "loss": 12.4756,
      "step": 4830
    },
    {
      "epoch": 0.26306725749252957,
      "grad_norm": 0.7658214146870006,
      "learning_rate": 0.00019464199068820528,
      "loss": 12.6996,
      "step": 4831
    },
    {
      "epoch": 0.2631217114891126,
      "grad_norm": 0.72957856770116,
      "learning_rate": 0.00019463914256400723,
      "loss": 12.4247,
      "step": 4832
    },
    {
      "epoch": 0.2631761654856956,
      "grad_norm": 0.7943362289892251,
      "learning_rate": 0.0001946362937038794,
      "loss": 12.6922,
      "step": 4833
    },
    {
      "epoch": 0.2632306194822786,
      "grad_norm": 0.6523756773773719,
      "learning_rate": 0.00019463344410784383,
      "loss": 12.4808,
      "step": 4834
    },
    {
      "epoch": 0.26328507347886165,
      "grad_norm": 0.6861067893129501,
      "learning_rate": 0.00019463059377592274,
      "loss": 12.5381,
      "step": 4835
    },
    {
      "epoch": 0.26333952747544465,
      "grad_norm": 0.6409161291059233,
      "learning_rate": 0.0001946277427081383,
      "loss": 12.5751,
      "step": 4836
    },
    {
      "epoch": 0.26339398147202764,
      "grad_norm": 0.8461938435521417,
      "learning_rate": 0.00019462489090451266,
      "loss": 12.6144,
      "step": 4837
    },
    {
      "epoch": 0.2634484354686107,
      "grad_norm": 0.6873872509981911,
      "learning_rate": 0.000194622038365068,
      "loss": 12.6681,
      "step": 4838
    },
    {
      "epoch": 0.2635028894651937,
      "grad_norm": 0.809696924328214,
      "learning_rate": 0.00019461918508982646,
      "loss": 12.5898,
      "step": 4839
    },
    {
      "epoch": 0.26355734346177667,
      "grad_norm": 0.7459462348725406,
      "learning_rate": 0.00019461633107881033,
      "loss": 12.7425,
      "step": 4840
    },
    {
      "epoch": 0.2636117974583597,
      "grad_norm": 0.6404604693639886,
      "learning_rate": 0.0001946134763320417,
      "loss": 12.5621,
      "step": 4841
    },
    {
      "epoch": 0.2636662514549427,
      "grad_norm": 0.6782170515842123,
      "learning_rate": 0.00019461062084954285,
      "loss": 12.5363,
      "step": 4842
    },
    {
      "epoch": 0.2637207054515257,
      "grad_norm": 0.7028036018440581,
      "learning_rate": 0.0001946077646313359,
      "loss": 12.5666,
      "step": 4843
    },
    {
      "epoch": 0.26377515944810875,
      "grad_norm": 0.6826374646724563,
      "learning_rate": 0.00019460490767744313,
      "loss": 12.6038,
      "step": 4844
    },
    {
      "epoch": 0.26382961344469175,
      "grad_norm": 0.6777141901476166,
      "learning_rate": 0.00019460204998788673,
      "loss": 12.5829,
      "step": 4845
    },
    {
      "epoch": 0.2638840674412748,
      "grad_norm": 0.6738235373260646,
      "learning_rate": 0.00019459919156268894,
      "loss": 12.5814,
      "step": 4846
    },
    {
      "epoch": 0.2639385214378578,
      "grad_norm": 0.6708886000655676,
      "learning_rate": 0.00019459633240187193,
      "loss": 12.6222,
      "step": 4847
    },
    {
      "epoch": 0.2639929754344408,
      "grad_norm": 0.7305811832108289,
      "learning_rate": 0.00019459347250545803,
      "loss": 12.5633,
      "step": 4848
    },
    {
      "epoch": 0.26404742943102383,
      "grad_norm": 0.7260244255550706,
      "learning_rate": 0.00019459061187346942,
      "loss": 12.5623,
      "step": 4849
    },
    {
      "epoch": 0.2641018834276068,
      "grad_norm": 0.6852163751677094,
      "learning_rate": 0.0001945877505059283,
      "loss": 12.5025,
      "step": 4850
    },
    {
      "epoch": 0.2641563374241898,
      "grad_norm": 0.6700399729180287,
      "learning_rate": 0.000194584888402857,
      "loss": 12.5428,
      "step": 4851
    },
    {
      "epoch": 0.26421079142077286,
      "grad_norm": 0.7012076478375378,
      "learning_rate": 0.00019458202556427775,
      "loss": 12.6202,
      "step": 4852
    },
    {
      "epoch": 0.26426524541735585,
      "grad_norm": 0.767091545667911,
      "learning_rate": 0.0001945791619902128,
      "loss": 12.4397,
      "step": 4853
    },
    {
      "epoch": 0.26431969941393885,
      "grad_norm": 0.7176345157155276,
      "learning_rate": 0.00019457629768068443,
      "loss": 12.5838,
      "step": 4854
    },
    {
      "epoch": 0.2643741534105219,
      "grad_norm": 0.7077784155956962,
      "learning_rate": 0.0001945734326357149,
      "loss": 12.4677,
      "step": 4855
    },
    {
      "epoch": 0.2644286074071049,
      "grad_norm": 0.6837911365687319,
      "learning_rate": 0.00019457056685532652,
      "loss": 12.492,
      "step": 4856
    },
    {
      "epoch": 0.2644830614036879,
      "grad_norm": 0.7466557995988053,
      "learning_rate": 0.0001945677003395415,
      "loss": 12.5368,
      "step": 4857
    },
    {
      "epoch": 0.26453751540027093,
      "grad_norm": 0.7541821594245927,
      "learning_rate": 0.00019456483308838226,
      "loss": 12.7146,
      "step": 4858
    },
    {
      "epoch": 0.2645919693968539,
      "grad_norm": 0.7277307319226684,
      "learning_rate": 0.00019456196510187095,
      "loss": 12.548,
      "step": 4859
    },
    {
      "epoch": 0.2646464233934369,
      "grad_norm": 0.750180857479281,
      "learning_rate": 0.00019455909638002998,
      "loss": 12.6692,
      "step": 4860
    },
    {
      "epoch": 0.26470087739001996,
      "grad_norm": 0.6882113934021395,
      "learning_rate": 0.0001945562269228816,
      "loss": 12.5586,
      "step": 4861
    },
    {
      "epoch": 0.26475533138660295,
      "grad_norm": 0.6876200797232995,
      "learning_rate": 0.00019455335673044814,
      "loss": 12.6751,
      "step": 4862
    },
    {
      "epoch": 0.26480978538318595,
      "grad_norm": 0.7413062407404183,
      "learning_rate": 0.00019455048580275193,
      "loss": 12.56,
      "step": 4863
    },
    {
      "epoch": 0.264864239379769,
      "grad_norm": 0.7997680346304358,
      "learning_rate": 0.0001945476141398153,
      "loss": 12.5148,
      "step": 4864
    },
    {
      "epoch": 0.264918693376352,
      "grad_norm": 0.7037783550914863,
      "learning_rate": 0.00019454474174166055,
      "loss": 12.4402,
      "step": 4865
    },
    {
      "epoch": 0.264973147372935,
      "grad_norm": 0.719253456465287,
      "learning_rate": 0.00019454186860831004,
      "loss": 12.526,
      "step": 4866
    },
    {
      "epoch": 0.26502760136951803,
      "grad_norm": 0.7821364793198318,
      "learning_rate": 0.0001945389947397861,
      "loss": 12.6374,
      "step": 4867
    },
    {
      "epoch": 0.265082055366101,
      "grad_norm": 0.8169352730341746,
      "learning_rate": 0.0001945361201361111,
      "loss": 12.6756,
      "step": 4868
    },
    {
      "epoch": 0.265136509362684,
      "grad_norm": 0.6669004440259494,
      "learning_rate": 0.00019453324479730736,
      "loss": 12.512,
      "step": 4869
    },
    {
      "epoch": 0.26519096335926706,
      "grad_norm": 0.6948054201853392,
      "learning_rate": 0.00019453036872339727,
      "loss": 12.6744,
      "step": 4870
    },
    {
      "epoch": 0.26524541735585005,
      "grad_norm": 0.6595901733830262,
      "learning_rate": 0.00019452749191440315,
      "loss": 12.6445,
      "step": 4871
    },
    {
      "epoch": 0.26529987135243305,
      "grad_norm": 0.7250972141249324,
      "learning_rate": 0.00019452461437034744,
      "loss": 12.6358,
      "step": 4872
    },
    {
      "epoch": 0.2653543253490161,
      "grad_norm": 0.8354176409927749,
      "learning_rate": 0.00019452173609125245,
      "loss": 12.6428,
      "step": 4873
    },
    {
      "epoch": 0.2654087793455991,
      "grad_norm": 0.6850492700866151,
      "learning_rate": 0.0001945188570771406,
      "loss": 12.565,
      "step": 4874
    },
    {
      "epoch": 0.2654632333421821,
      "grad_norm": 0.7474772489873514,
      "learning_rate": 0.00019451597732803426,
      "loss": 12.4716,
      "step": 4875
    },
    {
      "epoch": 0.26551768733876513,
      "grad_norm": 0.7239459348388512,
      "learning_rate": 0.00019451309684395581,
      "loss": 12.5,
      "step": 4876
    },
    {
      "epoch": 0.2655721413353481,
      "grad_norm": 0.7013468148695667,
      "learning_rate": 0.0001945102156249277,
      "loss": 12.6859,
      "step": 4877
    },
    {
      "epoch": 0.26562659533193117,
      "grad_norm": 0.6306546397409388,
      "learning_rate": 0.00019450733367097232,
      "loss": 12.258,
      "step": 4878
    },
    {
      "epoch": 0.26568104932851416,
      "grad_norm": 0.7732678904459518,
      "learning_rate": 0.00019450445098211203,
      "loss": 12.5905,
      "step": 4879
    },
    {
      "epoch": 0.26573550332509716,
      "grad_norm": 0.7102571767926045,
      "learning_rate": 0.00019450156755836928,
      "loss": 12.5944,
      "step": 4880
    },
    {
      "epoch": 0.2657899573216802,
      "grad_norm": 0.7428771349602378,
      "learning_rate": 0.0001944986833997665,
      "loss": 12.7083,
      "step": 4881
    },
    {
      "epoch": 0.2658444113182632,
      "grad_norm": 0.6644542846070517,
      "learning_rate": 0.0001944957985063261,
      "loss": 12.5964,
      "step": 4882
    },
    {
      "epoch": 0.2658988653148462,
      "grad_norm": 0.6076981577547662,
      "learning_rate": 0.00019449291287807055,
      "loss": 12.5323,
      "step": 4883
    },
    {
      "epoch": 0.26595331931142924,
      "grad_norm": 0.6774321021161596,
      "learning_rate": 0.00019449002651502224,
      "loss": 12.628,
      "step": 4884
    },
    {
      "epoch": 0.26600777330801223,
      "grad_norm": 0.6469750254177549,
      "learning_rate": 0.00019448713941720364,
      "loss": 12.5591,
      "step": 4885
    },
    {
      "epoch": 0.2660622273045952,
      "grad_norm": 0.6478552446275883,
      "learning_rate": 0.00019448425158463724,
      "loss": 12.5149,
      "step": 4886
    },
    {
      "epoch": 0.26611668130117827,
      "grad_norm": 0.6353492916660458,
      "learning_rate": 0.0001944813630173454,
      "loss": 12.6437,
      "step": 4887
    },
    {
      "epoch": 0.26617113529776126,
      "grad_norm": 0.7929604305168886,
      "learning_rate": 0.00019447847371535066,
      "loss": 12.6283,
      "step": 4888
    },
    {
      "epoch": 0.26622558929434426,
      "grad_norm": 0.6898416589120709,
      "learning_rate": 0.00019447558367867543,
      "loss": 12.5089,
      "step": 4889
    },
    {
      "epoch": 0.2662800432909273,
      "grad_norm": 0.7295361652237429,
      "learning_rate": 0.0001944726929073423,
      "loss": 12.5732,
      "step": 4890
    },
    {
      "epoch": 0.2663344972875103,
      "grad_norm": 0.7027949921759142,
      "learning_rate": 0.00019446980140137358,
      "loss": 12.6327,
      "step": 4891
    },
    {
      "epoch": 0.2663889512840933,
      "grad_norm": 0.7035390676555576,
      "learning_rate": 0.0001944669091607919,
      "loss": 12.6155,
      "step": 4892
    },
    {
      "epoch": 0.26644340528067634,
      "grad_norm": 0.6927777072601664,
      "learning_rate": 0.00019446401618561967,
      "loss": 12.6609,
      "step": 4893
    },
    {
      "epoch": 0.26649785927725933,
      "grad_norm": 0.7113014488080182,
      "learning_rate": 0.0001944611224758794,
      "loss": 12.6829,
      "step": 4894
    },
    {
      "epoch": 0.2665523132738423,
      "grad_norm": 0.6937576303435324,
      "learning_rate": 0.00019445822803159358,
      "loss": 12.4869,
      "step": 4895
    },
    {
      "epoch": 0.26660676727042537,
      "grad_norm": 0.7508896257482275,
      "learning_rate": 0.00019445533285278478,
      "loss": 12.6736,
      "step": 4896
    },
    {
      "epoch": 0.26666122126700836,
      "grad_norm": 0.6664214127305361,
      "learning_rate": 0.00019445243693947547,
      "loss": 12.7157,
      "step": 4897
    },
    {
      "epoch": 0.26671567526359136,
      "grad_norm": 0.7160016567620926,
      "learning_rate": 0.00019444954029168815,
      "loss": 12.5127,
      "step": 4898
    },
    {
      "epoch": 0.2667701292601744,
      "grad_norm": 0.7095136875260772,
      "learning_rate": 0.00019444664290944538,
      "loss": 12.6951,
      "step": 4899
    },
    {
      "epoch": 0.2668245832567574,
      "grad_norm": 0.7889749179258148,
      "learning_rate": 0.00019444374479276968,
      "loss": 12.5083,
      "step": 4900
    },
    {
      "epoch": 0.2668790372533404,
      "grad_norm": 0.6209670074747341,
      "learning_rate": 0.00019444084594168358,
      "loss": 12.5154,
      "step": 4901
    },
    {
      "epoch": 0.26693349124992344,
      "grad_norm": 0.7192247657918985,
      "learning_rate": 0.0001944379463562096,
      "loss": 12.6027,
      "step": 4902
    },
    {
      "epoch": 0.26698794524650643,
      "grad_norm": 0.877614733642246,
      "learning_rate": 0.00019443504603637032,
      "loss": 12.4859,
      "step": 4903
    },
    {
      "epoch": 0.2670423992430894,
      "grad_norm": 0.622920397255274,
      "learning_rate": 0.0001944321449821883,
      "loss": 12.5212,
      "step": 4904
    },
    {
      "epoch": 0.26709685323967247,
      "grad_norm": 0.8231902555703173,
      "learning_rate": 0.0001944292431936861,
      "loss": 12.5708,
      "step": 4905
    },
    {
      "epoch": 0.26715130723625546,
      "grad_norm": 0.7014509423459101,
      "learning_rate": 0.00019442634067088623,
      "loss": 12.6247,
      "step": 4906
    },
    {
      "epoch": 0.26720576123283846,
      "grad_norm": 0.7338425811141995,
      "learning_rate": 0.00019442343741381133,
      "loss": 12.567,
      "step": 4907
    },
    {
      "epoch": 0.2672602152294215,
      "grad_norm": 0.7128851004410109,
      "learning_rate": 0.00019442053342248392,
      "loss": 12.5589,
      "step": 4908
    },
    {
      "epoch": 0.2673146692260045,
      "grad_norm": 0.7640141068258829,
      "learning_rate": 0.00019441762869692664,
      "loss": 12.5248,
      "step": 4909
    },
    {
      "epoch": 0.2673691232225875,
      "grad_norm": 0.7195831536886108,
      "learning_rate": 0.000194414723237162,
      "loss": 12.6038,
      "step": 4910
    },
    {
      "epoch": 0.26742357721917054,
      "grad_norm": 0.702981087273901,
      "learning_rate": 0.00019441181704321267,
      "loss": 12.6929,
      "step": 4911
    },
    {
      "epoch": 0.26747803121575353,
      "grad_norm": 0.6649595768416755,
      "learning_rate": 0.00019440891011510123,
      "loss": 12.6238,
      "step": 4912
    },
    {
      "epoch": 0.2675324852123366,
      "grad_norm": 0.7465051804702764,
      "learning_rate": 0.00019440600245285023,
      "loss": 12.5796,
      "step": 4913
    },
    {
      "epoch": 0.26758693920891957,
      "grad_norm": 0.6771595729186441,
      "learning_rate": 0.00019440309405648236,
      "loss": 12.6006,
      "step": 4914
    },
    {
      "epoch": 0.26764139320550256,
      "grad_norm": 0.7984906107431816,
      "learning_rate": 0.0001944001849260202,
      "loss": 12.5348,
      "step": 4915
    },
    {
      "epoch": 0.2676958472020856,
      "grad_norm": 0.6987809489400444,
      "learning_rate": 0.00019439727506148635,
      "loss": 12.6028,
      "step": 4916
    },
    {
      "epoch": 0.2677503011986686,
      "grad_norm": 0.7515636815166795,
      "learning_rate": 0.00019439436446290346,
      "loss": 12.5774,
      "step": 4917
    },
    {
      "epoch": 0.2678047551952516,
      "grad_norm": 0.7772563152009838,
      "learning_rate": 0.00019439145313029417,
      "loss": 12.6186,
      "step": 4918
    },
    {
      "epoch": 0.26785920919183465,
      "grad_norm": 0.7773179505401135,
      "learning_rate": 0.00019438854106368112,
      "loss": 12.6273,
      "step": 4919
    },
    {
      "epoch": 0.26791366318841764,
      "grad_norm": 0.8029707125727654,
      "learning_rate": 0.00019438562826308692,
      "loss": 12.7813,
      "step": 4920
    },
    {
      "epoch": 0.26796811718500063,
      "grad_norm": 0.6739494023869455,
      "learning_rate": 0.00019438271472853427,
      "loss": 12.5858,
      "step": 4921
    },
    {
      "epoch": 0.2680225711815837,
      "grad_norm": 0.7172542758143176,
      "learning_rate": 0.0001943798004600458,
      "loss": 12.7093,
      "step": 4922
    },
    {
      "epoch": 0.2680770251781667,
      "grad_norm": 0.8540214909937995,
      "learning_rate": 0.00019437688545764417,
      "loss": 12.6435,
      "step": 4923
    },
    {
      "epoch": 0.26813147917474967,
      "grad_norm": 0.6922381771184539,
      "learning_rate": 0.00019437396972135206,
      "loss": 12.5729,
      "step": 4924
    },
    {
      "epoch": 0.2681859331713327,
      "grad_norm": 0.7388596470211966,
      "learning_rate": 0.00019437105325119212,
      "loss": 12.6941,
      "step": 4925
    },
    {
      "epoch": 0.2682403871679157,
      "grad_norm": 0.6799801483566477,
      "learning_rate": 0.00019436813604718705,
      "loss": 12.58,
      "step": 4926
    },
    {
      "epoch": 0.2682948411644987,
      "grad_norm": 0.677525774257368,
      "learning_rate": 0.00019436521810935954,
      "loss": 12.3613,
      "step": 4927
    },
    {
      "epoch": 0.26834929516108175,
      "grad_norm": 0.7070330522678878,
      "learning_rate": 0.00019436229943773224,
      "loss": 12.5587,
      "step": 4928
    },
    {
      "epoch": 0.26840374915766474,
      "grad_norm": 0.7431247845470497,
      "learning_rate": 0.0001943593800323279,
      "loss": 12.7184,
      "step": 4929
    },
    {
      "epoch": 0.26845820315424773,
      "grad_norm": 0.7448892262812907,
      "learning_rate": 0.00019435645989316917,
      "loss": 12.521,
      "step": 4930
    },
    {
      "epoch": 0.2685126571508308,
      "grad_norm": 0.7563793589303949,
      "learning_rate": 0.00019435353902027882,
      "loss": 12.8593,
      "step": 4931
    },
    {
      "epoch": 0.2685671111474138,
      "grad_norm": 0.6965002870763543,
      "learning_rate": 0.00019435061741367952,
      "loss": 12.5444,
      "step": 4932
    },
    {
      "epoch": 0.26862156514399677,
      "grad_norm": 0.7845469164455278,
      "learning_rate": 0.00019434769507339396,
      "loss": 12.6502,
      "step": 4933
    },
    {
      "epoch": 0.2686760191405798,
      "grad_norm": 0.6084132495946784,
      "learning_rate": 0.00019434477199944494,
      "loss": 12.6393,
      "step": 4934
    },
    {
      "epoch": 0.2687304731371628,
      "grad_norm": 0.7270494062285637,
      "learning_rate": 0.00019434184819185516,
      "loss": 12.5558,
      "step": 4935
    },
    {
      "epoch": 0.2687849271337458,
      "grad_norm": 0.7449005630000488,
      "learning_rate": 0.0001943389236506473,
      "loss": 12.6898,
      "step": 4936
    },
    {
      "epoch": 0.26883938113032885,
      "grad_norm": 0.7542098282712374,
      "learning_rate": 0.0001943359983758442,
      "loss": 12.5602,
      "step": 4937
    },
    {
      "epoch": 0.26889383512691184,
      "grad_norm": 0.6825357213113328,
      "learning_rate": 0.00019433307236746853,
      "loss": 12.6414,
      "step": 4938
    },
    {
      "epoch": 0.26894828912349483,
      "grad_norm": 0.7596587455890367,
      "learning_rate": 0.00019433014562554306,
      "loss": 12.4791,
      "step": 4939
    },
    {
      "epoch": 0.2690027431200779,
      "grad_norm": 0.7335019063446175,
      "learning_rate": 0.00019432721815009057,
      "loss": 12.5679,
      "step": 4940
    },
    {
      "epoch": 0.2690571971166609,
      "grad_norm": 0.8002044902081036,
      "learning_rate": 0.0001943242899411338,
      "loss": 12.6291,
      "step": 4941
    },
    {
      "epoch": 0.26911165111324387,
      "grad_norm": 0.7181184194287559,
      "learning_rate": 0.00019432136099869555,
      "loss": 12.5769,
      "step": 4942
    },
    {
      "epoch": 0.2691661051098269,
      "grad_norm": 0.693469130365641,
      "learning_rate": 0.0001943184313227986,
      "loss": 12.6736,
      "step": 4943
    },
    {
      "epoch": 0.2692205591064099,
      "grad_norm": 0.7883304347264259,
      "learning_rate": 0.00019431550091346565,
      "loss": 12.4797,
      "step": 4944
    },
    {
      "epoch": 0.26927501310299296,
      "grad_norm": 0.7675592218688019,
      "learning_rate": 0.0001943125697707196,
      "loss": 12.599,
      "step": 4945
    },
    {
      "epoch": 0.26932946709957595,
      "grad_norm": 0.721373700751928,
      "learning_rate": 0.0001943096378945832,
      "loss": 12.436,
      "step": 4946
    },
    {
      "epoch": 0.26938392109615894,
      "grad_norm": 0.7432999919007355,
      "learning_rate": 0.0001943067052850792,
      "loss": 12.6031,
      "step": 4947
    },
    {
      "epoch": 0.269438375092742,
      "grad_norm": 0.700930234427065,
      "learning_rate": 0.00019430377194223043,
      "loss": 12.6158,
      "step": 4948
    },
    {
      "epoch": 0.269492829089325,
      "grad_norm": 0.8856859606553761,
      "learning_rate": 0.00019430083786605977,
      "loss": 12.6485,
      "step": 4949
    },
    {
      "epoch": 0.269547283085908,
      "grad_norm": 0.7651445866542139,
      "learning_rate": 0.00019429790305658994,
      "loss": 12.5601,
      "step": 4950
    },
    {
      "epoch": 0.269601737082491,
      "grad_norm": 0.8712471278900041,
      "learning_rate": 0.00019429496751384383,
      "loss": 12.5648,
      "step": 4951
    },
    {
      "epoch": 0.269656191079074,
      "grad_norm": 0.836714647644876,
      "learning_rate": 0.00019429203123784422,
      "loss": 12.5715,
      "step": 4952
    },
    {
      "epoch": 0.269710645075657,
      "grad_norm": 0.7545099563149713,
      "learning_rate": 0.00019428909422861398,
      "loss": 12.4512,
      "step": 4953
    },
    {
      "epoch": 0.26976509907224006,
      "grad_norm": 0.7759249414188034,
      "learning_rate": 0.0001942861564861759,
      "loss": 12.576,
      "step": 4954
    },
    {
      "epoch": 0.26981955306882305,
      "grad_norm": 0.8070255412765706,
      "learning_rate": 0.0001942832180105529,
      "loss": 12.4277,
      "step": 4955
    },
    {
      "epoch": 0.26987400706540604,
      "grad_norm": 0.7411828713292871,
      "learning_rate": 0.00019428027880176777,
      "loss": 12.5935,
      "step": 4956
    },
    {
      "epoch": 0.2699284610619891,
      "grad_norm": 0.8762418560538697,
      "learning_rate": 0.00019427733885984337,
      "loss": 12.7198,
      "step": 4957
    },
    {
      "epoch": 0.2699829150585721,
      "grad_norm": 0.822480922374177,
      "learning_rate": 0.00019427439818480257,
      "loss": 12.5418,
      "step": 4958
    },
    {
      "epoch": 0.2700373690551551,
      "grad_norm": 0.7559488805950916,
      "learning_rate": 0.00019427145677666823,
      "loss": 12.5301,
      "step": 4959
    },
    {
      "epoch": 0.2700918230517381,
      "grad_norm": 0.7075578365647424,
      "learning_rate": 0.00019426851463546325,
      "loss": 12.5837,
      "step": 4960
    },
    {
      "epoch": 0.2701462770483211,
      "grad_norm": 0.7078321812508837,
      "learning_rate": 0.0001942655717612105,
      "loss": 12.6233,
      "step": 4961
    },
    {
      "epoch": 0.2702007310449041,
      "grad_norm": 0.7725000755982798,
      "learning_rate": 0.00019426262815393284,
      "loss": 12.4912,
      "step": 4962
    },
    {
      "epoch": 0.27025518504148716,
      "grad_norm": 0.7809703620738555,
      "learning_rate": 0.00019425968381365317,
      "loss": 12.7238,
      "step": 4963
    },
    {
      "epoch": 0.27030963903807015,
      "grad_norm": 0.7543002957619203,
      "learning_rate": 0.0001942567387403944,
      "loss": 12.5083,
      "step": 4964
    },
    {
      "epoch": 0.27036409303465314,
      "grad_norm": 0.7969983235647131,
      "learning_rate": 0.00019425379293417944,
      "loss": 12.6048,
      "step": 4965
    },
    {
      "epoch": 0.2704185470312362,
      "grad_norm": 0.7196112073402323,
      "learning_rate": 0.00019425084639503116,
      "loss": 12.5924,
      "step": 4966
    },
    {
      "epoch": 0.2704730010278192,
      "grad_norm": 0.7257632197119582,
      "learning_rate": 0.00019424789912297249,
      "loss": 12.6349,
      "step": 4967
    },
    {
      "epoch": 0.2705274550244022,
      "grad_norm": 0.9378435895655606,
      "learning_rate": 0.00019424495111802637,
      "loss": 12.6213,
      "step": 4968
    },
    {
      "epoch": 0.2705819090209852,
      "grad_norm": 0.7084946448579053,
      "learning_rate": 0.00019424200238021567,
      "loss": 12.5467,
      "step": 4969
    },
    {
      "epoch": 0.2706363630175682,
      "grad_norm": 0.7483908594182052,
      "learning_rate": 0.0001942390529095634,
      "loss": 12.6569,
      "step": 4970
    },
    {
      "epoch": 0.2706908170141512,
      "grad_norm": 0.7118408982414774,
      "learning_rate": 0.00019423610270609244,
      "loss": 12.5747,
      "step": 4971
    },
    {
      "epoch": 0.27074527101073426,
      "grad_norm": 0.68660993044931,
      "learning_rate": 0.0001942331517698257,
      "loss": 12.4705,
      "step": 4972
    },
    {
      "epoch": 0.27079972500731725,
      "grad_norm": 0.7802996073573238,
      "learning_rate": 0.00019423020010078622,
      "loss": 12.5696,
      "step": 4973
    },
    {
      "epoch": 0.27085417900390024,
      "grad_norm": 0.674993499837841,
      "learning_rate": 0.00019422724769899686,
      "loss": 12.6346,
      "step": 4974
    },
    {
      "epoch": 0.2709086330004833,
      "grad_norm": 0.9356688643207316,
      "learning_rate": 0.00019422429456448064,
      "loss": 12.8044,
      "step": 4975
    },
    {
      "epoch": 0.2709630869970663,
      "grad_norm": 0.7621397899824885,
      "learning_rate": 0.00019422134069726053,
      "loss": 12.571,
      "step": 4976
    },
    {
      "epoch": 0.2710175409936493,
      "grad_norm": 0.6867695978399014,
      "learning_rate": 0.00019421838609735942,
      "loss": 12.5917,
      "step": 4977
    },
    {
      "epoch": 0.2710719949902323,
      "grad_norm": 0.8231682604535635,
      "learning_rate": 0.0001942154307648004,
      "loss": 12.5752,
      "step": 4978
    },
    {
      "epoch": 0.2711264489868153,
      "grad_norm": 0.8111827291656801,
      "learning_rate": 0.00019421247469960634,
      "loss": 12.6336,
      "step": 4979
    },
    {
      "epoch": 0.27118090298339836,
      "grad_norm": 1.042410249315156,
      "learning_rate": 0.00019420951790180029,
      "loss": 12.6439,
      "step": 4980
    },
    {
      "epoch": 0.27123535697998136,
      "grad_norm": 0.7076263354008396,
      "learning_rate": 0.00019420656037140525,
      "loss": 12.6046,
      "step": 4981
    },
    {
      "epoch": 0.27128981097656435,
      "grad_norm": 0.6515365376827323,
      "learning_rate": 0.00019420360210844418,
      "loss": 12.5624,
      "step": 4982
    },
    {
      "epoch": 0.2713442649731474,
      "grad_norm": 0.7022423293026825,
      "learning_rate": 0.0001942006431129401,
      "loss": 12.5117,
      "step": 4983
    },
    {
      "epoch": 0.2713987189697304,
      "grad_norm": 0.7954526938737386,
      "learning_rate": 0.00019419768338491605,
      "loss": 12.7401,
      "step": 4984
    },
    {
      "epoch": 0.2714531729663134,
      "grad_norm": 0.8688496563455215,
      "learning_rate": 0.00019419472292439498,
      "loss": 12.7052,
      "step": 4985
    },
    {
      "epoch": 0.27150762696289643,
      "grad_norm": 0.6874496497060338,
      "learning_rate": 0.00019419176173139996,
      "loss": 12.653,
      "step": 4986
    },
    {
      "epoch": 0.2715620809594794,
      "grad_norm": 0.7350358087124989,
      "learning_rate": 0.000194188799805954,
      "loss": 12.7017,
      "step": 4987
    },
    {
      "epoch": 0.2716165349560624,
      "grad_norm": 0.7917394937965483,
      "learning_rate": 0.00019418583714808017,
      "loss": 12.5159,
      "step": 4988
    },
    {
      "epoch": 0.27167098895264546,
      "grad_norm": 0.7488765240242293,
      "learning_rate": 0.00019418287375780146,
      "loss": 12.6222,
      "step": 4989
    },
    {
      "epoch": 0.27172544294922846,
      "grad_norm": 0.7146616561206384,
      "learning_rate": 0.00019417990963514086,
      "loss": 12.4543,
      "step": 4990
    },
    {
      "epoch": 0.27177989694581145,
      "grad_norm": 0.7287116095576452,
      "learning_rate": 0.00019417694478012157,
      "loss": 12.5467,
      "step": 4991
    },
    {
      "epoch": 0.2718343509423945,
      "grad_norm": 0.8550421741239044,
      "learning_rate": 0.00019417397919276654,
      "loss": 12.6141,
      "step": 4992
    },
    {
      "epoch": 0.2718888049389775,
      "grad_norm": 0.8027759839067112,
      "learning_rate": 0.00019417101287309886,
      "loss": 12.6347,
      "step": 4993
    },
    {
      "epoch": 0.2719432589355605,
      "grad_norm": 0.7161657765497795,
      "learning_rate": 0.00019416804582114157,
      "loss": 12.4361,
      "step": 4994
    },
    {
      "epoch": 0.27199771293214353,
      "grad_norm": 0.7008462549080949,
      "learning_rate": 0.0001941650780369178,
      "loss": 12.6173,
      "step": 4995
    },
    {
      "epoch": 0.2720521669287265,
      "grad_norm": 0.7668462501840719,
      "learning_rate": 0.00019416210952045057,
      "loss": 12.5077,
      "step": 4996
    },
    {
      "epoch": 0.2721066209253095,
      "grad_norm": 0.7308418987478974,
      "learning_rate": 0.000194159140271763,
      "loss": 12.5051,
      "step": 4997
    },
    {
      "epoch": 0.27216107492189257,
      "grad_norm": 0.7483771822126182,
      "learning_rate": 0.00019415617029087815,
      "loss": 12.7825,
      "step": 4998
    },
    {
      "epoch": 0.27221552891847556,
      "grad_norm": 0.6782389168304814,
      "learning_rate": 0.00019415319957781914,
      "loss": 12.5248,
      "step": 4999
    },
    {
      "epoch": 0.27226998291505855,
      "grad_norm": 0.7489393248917106,
      "learning_rate": 0.00019415022813260903,
      "loss": 12.7125,
      "step": 5000
    },
    {
      "epoch": 0.2723244369116416,
      "grad_norm": 0.679800953165827,
      "learning_rate": 0.000194147255955271,
      "loss": 12.6751,
      "step": 5001
    },
    {
      "epoch": 0.2723788909082246,
      "grad_norm": 0.7888451496769374,
      "learning_rate": 0.0001941442830458281,
      "loss": 12.662,
      "step": 5002
    },
    {
      "epoch": 0.2724333449048076,
      "grad_norm": 0.7325560605775505,
      "learning_rate": 0.00019414130940430347,
      "loss": 12.7182,
      "step": 5003
    },
    {
      "epoch": 0.27248779890139063,
      "grad_norm": 0.7237907170438983,
      "learning_rate": 0.0001941383350307202,
      "loss": 12.4911,
      "step": 5004
    },
    {
      "epoch": 0.2725422528979736,
      "grad_norm": 0.6903596227181271,
      "learning_rate": 0.0001941353599251015,
      "loss": 12.4553,
      "step": 5005
    },
    {
      "epoch": 0.2725967068945566,
      "grad_norm": 0.8111728486824789,
      "learning_rate": 0.00019413238408747042,
      "loss": 12.6992,
      "step": 5006
    },
    {
      "epoch": 0.27265116089113967,
      "grad_norm": 0.8038359449773015,
      "learning_rate": 0.00019412940751785016,
      "loss": 12.5698,
      "step": 5007
    },
    {
      "epoch": 0.27270561488772266,
      "grad_norm": 0.6879193280914043,
      "learning_rate": 0.00019412643021626385,
      "loss": 12.5842,
      "step": 5008
    },
    {
      "epoch": 0.27276006888430565,
      "grad_norm": 0.7823374563346943,
      "learning_rate": 0.0001941234521827346,
      "loss": 12.5844,
      "step": 5009
    },
    {
      "epoch": 0.2728145228808887,
      "grad_norm": 1.109044580363149,
      "learning_rate": 0.00019412047341728562,
      "loss": 12.6108,
      "step": 5010
    },
    {
      "epoch": 0.2728689768774717,
      "grad_norm": 0.6525462085960635,
      "learning_rate": 0.00019411749391994002,
      "loss": 12.5394,
      "step": 5011
    },
    {
      "epoch": 0.27292343087405474,
      "grad_norm": 0.8202131839297926,
      "learning_rate": 0.00019411451369072104,
      "loss": 12.652,
      "step": 5012
    },
    {
      "epoch": 0.27297788487063773,
      "grad_norm": 0.6646217005732907,
      "learning_rate": 0.00019411153272965183,
      "loss": 12.3951,
      "step": 5013
    },
    {
      "epoch": 0.2730323388672207,
      "grad_norm": 0.7073205373515769,
      "learning_rate": 0.00019410855103675552,
      "loss": 12.4288,
      "step": 5014
    },
    {
      "epoch": 0.2730867928638038,
      "grad_norm": 0.7478502526049412,
      "learning_rate": 0.0001941055686120554,
      "loss": 12.6234,
      "step": 5015
    },
    {
      "epoch": 0.27314124686038677,
      "grad_norm": 0.7901002256999474,
      "learning_rate": 0.00019410258545557452,
      "loss": 12.6419,
      "step": 5016
    },
    {
      "epoch": 0.27319570085696976,
      "grad_norm": 0.7292447218005194,
      "learning_rate": 0.0001940996015673362,
      "loss": 12.6508,
      "step": 5017
    },
    {
      "epoch": 0.2732501548535528,
      "grad_norm": 0.7317662343794659,
      "learning_rate": 0.00019409661694736355,
      "loss": 12.6472,
      "step": 5018
    },
    {
      "epoch": 0.2733046088501358,
      "grad_norm": 0.7269954070004496,
      "learning_rate": 0.0001940936315956799,
      "loss": 12.4842,
      "step": 5019
    },
    {
      "epoch": 0.2733590628467188,
      "grad_norm": 0.68273618885286,
      "learning_rate": 0.00019409064551230833,
      "loss": 12.5733,
      "step": 5020
    },
    {
      "epoch": 0.27341351684330184,
      "grad_norm": 0.7839706162049932,
      "learning_rate": 0.00019408765869727214,
      "loss": 12.5424,
      "step": 5021
    },
    {
      "epoch": 0.27346797083988483,
      "grad_norm": 0.7754399766077261,
      "learning_rate": 0.00019408467115059454,
      "loss": 12.6342,
      "step": 5022
    },
    {
      "epoch": 0.2735224248364678,
      "grad_norm": 0.6691754583961804,
      "learning_rate": 0.00019408168287229875,
      "loss": 12.4901,
      "step": 5023
    },
    {
      "epoch": 0.2735768788330509,
      "grad_norm": 0.7275096808427216,
      "learning_rate": 0.00019407869386240805,
      "loss": 12.6807,
      "step": 5024
    },
    {
      "epoch": 0.27363133282963387,
      "grad_norm": 0.7222259300134751,
      "learning_rate": 0.00019407570412094562,
      "loss": 12.687,
      "step": 5025
    },
    {
      "epoch": 0.27368578682621686,
      "grad_norm": 0.6536571577211265,
      "learning_rate": 0.00019407271364793474,
      "loss": 12.5979,
      "step": 5026
    },
    {
      "epoch": 0.2737402408227999,
      "grad_norm": 0.6961205317617686,
      "learning_rate": 0.00019406972244339867,
      "loss": 12.5989,
      "step": 5027
    },
    {
      "epoch": 0.2737946948193829,
      "grad_norm": 0.7038426126899354,
      "learning_rate": 0.00019406673050736067,
      "loss": 12.6524,
      "step": 5028
    },
    {
      "epoch": 0.2738491488159659,
      "grad_norm": 0.7279571986835491,
      "learning_rate": 0.000194063737839844,
      "loss": 12.6209,
      "step": 5029
    },
    {
      "epoch": 0.27390360281254894,
      "grad_norm": 0.7821768847303526,
      "learning_rate": 0.0001940607444408719,
      "loss": 12.657,
      "step": 5030
    },
    {
      "epoch": 0.27395805680913193,
      "grad_norm": 0.7535716318076771,
      "learning_rate": 0.0001940577503104677,
      "loss": 12.5637,
      "step": 5031
    },
    {
      "epoch": 0.2740125108057149,
      "grad_norm": 0.740051362894615,
      "learning_rate": 0.00019405475544865465,
      "loss": 12.616,
      "step": 5032
    },
    {
      "epoch": 0.274066964802298,
      "grad_norm": 0.7922028822553647,
      "learning_rate": 0.00019405175985545605,
      "loss": 12.5247,
      "step": 5033
    },
    {
      "epoch": 0.27412141879888097,
      "grad_norm": 0.6497358398767344,
      "learning_rate": 0.00019404876353089522,
      "loss": 12.4952,
      "step": 5034
    },
    {
      "epoch": 0.27417587279546396,
      "grad_norm": 0.7550312787281879,
      "learning_rate": 0.0001940457664749954,
      "loss": 12.5047,
      "step": 5035
    },
    {
      "epoch": 0.274230326792047,
      "grad_norm": 0.6699670017875384,
      "learning_rate": 0.00019404276868777994,
      "loss": 12.6481,
      "step": 5036
    },
    {
      "epoch": 0.27428478078863,
      "grad_norm": 0.7225857141499656,
      "learning_rate": 0.00019403977016927212,
      "loss": 12.3542,
      "step": 5037
    },
    {
      "epoch": 0.274339234785213,
      "grad_norm": 0.6977057679519985,
      "learning_rate": 0.0001940367709194953,
      "loss": 12.5138,
      "step": 5038
    },
    {
      "epoch": 0.27439368878179604,
      "grad_norm": 0.704198464905272,
      "learning_rate": 0.0001940337709384728,
      "loss": 12.5917,
      "step": 5039
    },
    {
      "epoch": 0.27444814277837903,
      "grad_norm": 0.6711982258147914,
      "learning_rate": 0.0001940307702262279,
      "loss": 12.6132,
      "step": 5040
    },
    {
      "epoch": 0.274502596774962,
      "grad_norm": 0.8079769295912083,
      "learning_rate": 0.00019402776878278395,
      "loss": 12.6885,
      "step": 5041
    },
    {
      "epoch": 0.2745570507715451,
      "grad_norm": 0.6559387708196829,
      "learning_rate": 0.00019402476660816432,
      "loss": 12.5405,
      "step": 5042
    },
    {
      "epoch": 0.27461150476812807,
      "grad_norm": 0.7379381995247378,
      "learning_rate": 0.00019402176370239232,
      "loss": 12.6785,
      "step": 5043
    },
    {
      "epoch": 0.27466595876471106,
      "grad_norm": 0.690471720110709,
      "learning_rate": 0.00019401876006549132,
      "loss": 12.5732,
      "step": 5044
    },
    {
      "epoch": 0.2747204127612941,
      "grad_norm": 0.7080593199826031,
      "learning_rate": 0.0001940157556974847,
      "loss": 12.6293,
      "step": 5045
    },
    {
      "epoch": 0.2747748667578771,
      "grad_norm": 0.6739896979204867,
      "learning_rate": 0.0001940127505983958,
      "loss": 12.5172,
      "step": 5046
    },
    {
      "epoch": 0.27482932075446015,
      "grad_norm": 0.651768469028915,
      "learning_rate": 0.00019400974476824795,
      "loss": 12.5162,
      "step": 5047
    },
    {
      "epoch": 0.27488377475104314,
      "grad_norm": 0.838781952055105,
      "learning_rate": 0.00019400673820706458,
      "loss": 12.6232,
      "step": 5048
    },
    {
      "epoch": 0.27493822874762613,
      "grad_norm": 0.6672933137610362,
      "learning_rate": 0.00019400373091486904,
      "loss": 12.6947,
      "step": 5049
    },
    {
      "epoch": 0.2749926827442092,
      "grad_norm": 0.7835512965878694,
      "learning_rate": 0.00019400072289168474,
      "loss": 12.5005,
      "step": 5050
    },
    {
      "epoch": 0.2750471367407922,
      "grad_norm": 0.74299265506198,
      "learning_rate": 0.00019399771413753506,
      "loss": 12.5162,
      "step": 5051
    },
    {
      "epoch": 0.27510159073737517,
      "grad_norm": 0.7492972464867815,
      "learning_rate": 0.00019399470465244337,
      "loss": 12.6574,
      "step": 5052
    },
    {
      "epoch": 0.2751560447339582,
      "grad_norm": 0.7271180956005475,
      "learning_rate": 0.0001939916944364331,
      "loss": 12.47,
      "step": 5053
    },
    {
      "epoch": 0.2752104987305412,
      "grad_norm": 0.671577089718614,
      "learning_rate": 0.00019398868348952764,
      "loss": 12.6404,
      "step": 5054
    },
    {
      "epoch": 0.2752649527271242,
      "grad_norm": 0.7593124681164001,
      "learning_rate": 0.00019398567181175042,
      "loss": 12.6242,
      "step": 5055
    },
    {
      "epoch": 0.27531940672370725,
      "grad_norm": 0.7147968774089865,
      "learning_rate": 0.0001939826594031249,
      "loss": 12.4892,
      "step": 5056
    },
    {
      "epoch": 0.27537386072029024,
      "grad_norm": 0.6362963231401051,
      "learning_rate": 0.0001939796462636744,
      "loss": 12.5185,
      "step": 5057
    },
    {
      "epoch": 0.27542831471687323,
      "grad_norm": 0.710749446087054,
      "learning_rate": 0.0001939766323934224,
      "loss": 12.5753,
      "step": 5058
    },
    {
      "epoch": 0.2754827687134563,
      "grad_norm": 0.6111662941975942,
      "learning_rate": 0.0001939736177923924,
      "loss": 12.5201,
      "step": 5059
    },
    {
      "epoch": 0.2755372227100393,
      "grad_norm": 0.6880616611302837,
      "learning_rate": 0.00019397060246060776,
      "loss": 12.6284,
      "step": 5060
    },
    {
      "epoch": 0.27559167670662227,
      "grad_norm": 0.6281572869817057,
      "learning_rate": 0.00019396758639809197,
      "loss": 12.5722,
      "step": 5061
    },
    {
      "epoch": 0.2756461307032053,
      "grad_norm": 0.790136974222293,
      "learning_rate": 0.00019396456960486846,
      "loss": 12.6321,
      "step": 5062
    },
    {
      "epoch": 0.2757005846997883,
      "grad_norm": 0.6693878597166492,
      "learning_rate": 0.0001939615520809607,
      "loss": 12.5091,
      "step": 5063
    },
    {
      "epoch": 0.2757550386963713,
      "grad_norm": 0.6646705864772172,
      "learning_rate": 0.00019395853382639215,
      "loss": 12.6344,
      "step": 5064
    },
    {
      "epoch": 0.27580949269295435,
      "grad_norm": 0.7191256074928883,
      "learning_rate": 0.00019395551484118628,
      "loss": 12.6256,
      "step": 5065
    },
    {
      "epoch": 0.27586394668953734,
      "grad_norm": 0.8147113211400999,
      "learning_rate": 0.0001939524951253666,
      "loss": 12.6511,
      "step": 5066
    },
    {
      "epoch": 0.27591840068612034,
      "grad_norm": 0.6417233825294415,
      "learning_rate": 0.00019394947467895652,
      "loss": 12.543,
      "step": 5067
    },
    {
      "epoch": 0.2759728546827034,
      "grad_norm": 0.6366238255470038,
      "learning_rate": 0.0001939464535019796,
      "loss": 12.5125,
      "step": 5068
    },
    {
      "epoch": 0.2760273086792864,
      "grad_norm": 0.6932555039744597,
      "learning_rate": 0.0001939434315944593,
      "loss": 12.6213,
      "step": 5069
    },
    {
      "epoch": 0.27608176267586937,
      "grad_norm": 0.7148171868609757,
      "learning_rate": 0.0001939404089564191,
      "loss": 12.5546,
      "step": 5070
    },
    {
      "epoch": 0.2761362166724524,
      "grad_norm": 0.7662802540895398,
      "learning_rate": 0.00019393738558788254,
      "loss": 12.4323,
      "step": 5071
    },
    {
      "epoch": 0.2761906706690354,
      "grad_norm": 0.7005674022938893,
      "learning_rate": 0.00019393436148887314,
      "loss": 12.7013,
      "step": 5072
    },
    {
      "epoch": 0.2762451246656184,
      "grad_norm": 0.7041144517869332,
      "learning_rate": 0.00019393133665941437,
      "loss": 12.6209,
      "step": 5073
    },
    {
      "epoch": 0.27629957866220145,
      "grad_norm": 0.6777116434436815,
      "learning_rate": 0.00019392831109952977,
      "loss": 12.4813,
      "step": 5074
    },
    {
      "epoch": 0.27635403265878444,
      "grad_norm": 0.6845631530681943,
      "learning_rate": 0.00019392528480924285,
      "loss": 12.5907,
      "step": 5075
    },
    {
      "epoch": 0.27640848665536744,
      "grad_norm": 0.7114657166282016,
      "learning_rate": 0.00019392225778857723,
      "loss": 12.5673,
      "step": 5076
    },
    {
      "epoch": 0.2764629406519505,
      "grad_norm": 0.8107231975439327,
      "learning_rate": 0.00019391923003755633,
      "loss": 12.5245,
      "step": 5077
    },
    {
      "epoch": 0.2765173946485335,
      "grad_norm": 0.8040081885700733,
      "learning_rate": 0.00019391620155620375,
      "loss": 12.5707,
      "step": 5078
    },
    {
      "epoch": 0.2765718486451165,
      "grad_norm": 0.6572804074345328,
      "learning_rate": 0.00019391317234454305,
      "loss": 12.6469,
      "step": 5079
    },
    {
      "epoch": 0.2766263026416995,
      "grad_norm": 0.8236022163967431,
      "learning_rate": 0.00019391014240259776,
      "loss": 12.5572,
      "step": 5080
    },
    {
      "epoch": 0.2766807566382825,
      "grad_norm": 0.658914373309915,
      "learning_rate": 0.00019390711173039146,
      "loss": 12.5925,
      "step": 5081
    },
    {
      "epoch": 0.27673521063486556,
      "grad_norm": 0.7943194386641196,
      "learning_rate": 0.00019390408032794772,
      "loss": 12.6722,
      "step": 5082
    },
    {
      "epoch": 0.27678966463144855,
      "grad_norm": 0.6973239731440578,
      "learning_rate": 0.00019390104819529008,
      "loss": 12.552,
      "step": 5083
    },
    {
      "epoch": 0.27684411862803154,
      "grad_norm": 0.7418809699302191,
      "learning_rate": 0.00019389801533244218,
      "loss": 12.5722,
      "step": 5084
    },
    {
      "epoch": 0.2768985726246146,
      "grad_norm": 0.7422872676063248,
      "learning_rate": 0.00019389498173942756,
      "loss": 12.5761,
      "step": 5085
    },
    {
      "epoch": 0.2769530266211976,
      "grad_norm": 0.703580772909161,
      "learning_rate": 0.0001938919474162698,
      "loss": 12.5502,
      "step": 5086
    },
    {
      "epoch": 0.2770074806177806,
      "grad_norm": 0.7000256855033609,
      "learning_rate": 0.00019388891236299253,
      "loss": 12.6968,
      "step": 5087
    },
    {
      "epoch": 0.2770619346143636,
      "grad_norm": 0.7135620202619926,
      "learning_rate": 0.0001938858765796193,
      "loss": 12.6916,
      "step": 5088
    },
    {
      "epoch": 0.2771163886109466,
      "grad_norm": 0.7652487105619586,
      "learning_rate": 0.00019388284006617375,
      "loss": 12.3941,
      "step": 5089
    },
    {
      "epoch": 0.2771708426075296,
      "grad_norm": 0.690744559913229,
      "learning_rate": 0.0001938798028226795,
      "loss": 12.5351,
      "step": 5090
    },
    {
      "epoch": 0.27722529660411266,
      "grad_norm": 0.8156095234373029,
      "learning_rate": 0.0001938767648491602,
      "loss": 12.4954,
      "step": 5091
    },
    {
      "epoch": 0.27727975060069565,
      "grad_norm": 0.7090953026934278,
      "learning_rate": 0.00019387372614563936,
      "loss": 12.5321,
      "step": 5092
    },
    {
      "epoch": 0.27733420459727864,
      "grad_norm": 0.6409777844639443,
      "learning_rate": 0.00019387068671214072,
      "loss": 12.4822,
      "step": 5093
    },
    {
      "epoch": 0.2773886585938617,
      "grad_norm": 0.7587500601030793,
      "learning_rate": 0.0001938676465486879,
      "loss": 12.6798,
      "step": 5094
    },
    {
      "epoch": 0.2774431125904447,
      "grad_norm": 0.649528915312248,
      "learning_rate": 0.0001938646056553045,
      "loss": 12.502,
      "step": 5095
    },
    {
      "epoch": 0.2774975665870277,
      "grad_norm": 0.747554472997202,
      "learning_rate": 0.00019386156403201416,
      "loss": 12.5573,
      "step": 5096
    },
    {
      "epoch": 0.2775520205836107,
      "grad_norm": 0.8592539276831329,
      "learning_rate": 0.00019385852167884057,
      "loss": 12.476,
      "step": 5097
    },
    {
      "epoch": 0.2776064745801937,
      "grad_norm": 0.7200503490316053,
      "learning_rate": 0.00019385547859580743,
      "loss": 12.5044,
      "step": 5098
    },
    {
      "epoch": 0.2776609285767767,
      "grad_norm": 0.6728688453119079,
      "learning_rate": 0.00019385243478293828,
      "loss": 12.6178,
      "step": 5099
    },
    {
      "epoch": 0.27771538257335976,
      "grad_norm": 0.6660773015222236,
      "learning_rate": 0.0001938493902402569,
      "loss": 12.6218,
      "step": 5100
    },
    {
      "epoch": 0.27776983656994275,
      "grad_norm": 0.6954282062789127,
      "learning_rate": 0.00019384634496778688,
      "loss": 12.4654,
      "step": 5101
    },
    {
      "epoch": 0.27782429056652574,
      "grad_norm": 0.7156888717451666,
      "learning_rate": 0.00019384329896555196,
      "loss": 12.6465,
      "step": 5102
    },
    {
      "epoch": 0.2778787445631088,
      "grad_norm": 0.7179172725399301,
      "learning_rate": 0.00019384025223357582,
      "loss": 12.5168,
      "step": 5103
    },
    {
      "epoch": 0.2779331985596918,
      "grad_norm": 0.7759017642906217,
      "learning_rate": 0.00019383720477188216,
      "loss": 12.602,
      "step": 5104
    },
    {
      "epoch": 0.2779876525562748,
      "grad_norm": 0.6510398797162207,
      "learning_rate": 0.00019383415658049465,
      "loss": 12.4956,
      "step": 5105
    },
    {
      "epoch": 0.2780421065528578,
      "grad_norm": 0.6642562478621538,
      "learning_rate": 0.00019383110765943697,
      "loss": 12.4895,
      "step": 5106
    },
    {
      "epoch": 0.2780965605494408,
      "grad_norm": 0.6856961004079563,
      "learning_rate": 0.00019382805800873288,
      "loss": 12.519,
      "step": 5107
    },
    {
      "epoch": 0.2781510145460238,
      "grad_norm": 0.6423899847602803,
      "learning_rate": 0.0001938250076284061,
      "loss": 12.5387,
      "step": 5108
    },
    {
      "epoch": 0.27820546854260686,
      "grad_norm": 0.636575246825305,
      "learning_rate": 0.00019382195651848028,
      "loss": 12.7195,
      "step": 5109
    },
    {
      "epoch": 0.27825992253918985,
      "grad_norm": 0.6970889423434784,
      "learning_rate": 0.00019381890467897922,
      "loss": 12.6636,
      "step": 5110
    },
    {
      "epoch": 0.27831437653577284,
      "grad_norm": 0.626582862445139,
      "learning_rate": 0.00019381585210992663,
      "loss": 12.5385,
      "step": 5111
    },
    {
      "epoch": 0.2783688305323559,
      "grad_norm": 0.7246283791178956,
      "learning_rate": 0.00019381279881134625,
      "loss": 12.573,
      "step": 5112
    },
    {
      "epoch": 0.2784232845289389,
      "grad_norm": 0.6601903163083505,
      "learning_rate": 0.00019380974478326178,
      "loss": 12.6021,
      "step": 5113
    },
    {
      "epoch": 0.27847773852552193,
      "grad_norm": 0.762912600904577,
      "learning_rate": 0.000193806690025697,
      "loss": 12.6755,
      "step": 5114
    },
    {
      "epoch": 0.2785321925221049,
      "grad_norm": 0.676212979343806,
      "learning_rate": 0.0001938036345386757,
      "loss": 12.5951,
      "step": 5115
    },
    {
      "epoch": 0.2785866465186879,
      "grad_norm": 0.7940947673405626,
      "learning_rate": 0.0001938005783222216,
      "loss": 12.5618,
      "step": 5116
    },
    {
      "epoch": 0.27864110051527097,
      "grad_norm": 0.6346251485123232,
      "learning_rate": 0.00019379752137635848,
      "loss": 12.4973,
      "step": 5117
    },
    {
      "epoch": 0.27869555451185396,
      "grad_norm": 0.8342476054070347,
      "learning_rate": 0.00019379446370111007,
      "loss": 12.6665,
      "step": 5118
    },
    {
      "epoch": 0.27875000850843695,
      "grad_norm": 0.715632773565721,
      "learning_rate": 0.0001937914052965002,
      "loss": 12.641,
      "step": 5119
    },
    {
      "epoch": 0.27880446250502,
      "grad_norm": 0.6382536033591416,
      "learning_rate": 0.00019378834616255264,
      "loss": 12.6341,
      "step": 5120
    },
    {
      "epoch": 0.278858916501603,
      "grad_norm": 0.6833566071913983,
      "learning_rate": 0.0001937852862992912,
      "loss": 12.6692,
      "step": 5121
    },
    {
      "epoch": 0.278913370498186,
      "grad_norm": 0.7187158872693135,
      "learning_rate": 0.00019378222570673955,
      "loss": 12.5714,
      "step": 5122
    },
    {
      "epoch": 0.27896782449476903,
      "grad_norm": 0.7008947274359243,
      "learning_rate": 0.00019377916438492168,
      "loss": 12.577,
      "step": 5123
    },
    {
      "epoch": 0.279022278491352,
      "grad_norm": 0.803851898243202,
      "learning_rate": 0.00019377610233386124,
      "loss": 12.7355,
      "step": 5124
    },
    {
      "epoch": 0.279076732487935,
      "grad_norm": 0.7559303634246921,
      "learning_rate": 0.00019377303955358217,
      "loss": 12.6045,
      "step": 5125
    },
    {
      "epoch": 0.27913118648451807,
      "grad_norm": 0.8192745121555173,
      "learning_rate": 0.00019376997604410816,
      "loss": 12.7426,
      "step": 5126
    },
    {
      "epoch": 0.27918564048110106,
      "grad_norm": 0.7247911238027126,
      "learning_rate": 0.0001937669118054631,
      "loss": 12.598,
      "step": 5127
    },
    {
      "epoch": 0.27924009447768405,
      "grad_norm": 0.7163926957202394,
      "learning_rate": 0.00019376384683767085,
      "loss": 12.5991,
      "step": 5128
    },
    {
      "epoch": 0.2792945484742671,
      "grad_norm": 0.633941715979679,
      "learning_rate": 0.00019376078114075515,
      "loss": 12.509,
      "step": 5129
    },
    {
      "epoch": 0.2793490024708501,
      "grad_norm": 0.8875319613920101,
      "learning_rate": 0.00019375771471473994,
      "loss": 12.5074,
      "step": 5130
    },
    {
      "epoch": 0.2794034564674331,
      "grad_norm": 0.6848286176502122,
      "learning_rate": 0.00019375464755964897,
      "loss": 12.605,
      "step": 5131
    },
    {
      "epoch": 0.27945791046401613,
      "grad_norm": 0.7319492347428458,
      "learning_rate": 0.00019375157967550617,
      "loss": 12.6155,
      "step": 5132
    },
    {
      "epoch": 0.2795123644605991,
      "grad_norm": 0.9091714525701203,
      "learning_rate": 0.00019374851106233534,
      "loss": 12.7187,
      "step": 5133
    },
    {
      "epoch": 0.2795668184571821,
      "grad_norm": 0.7162150438667014,
      "learning_rate": 0.0001937454417201604,
      "loss": 12.5939,
      "step": 5134
    },
    {
      "epoch": 0.27962127245376517,
      "grad_norm": 0.9315398296781449,
      "learning_rate": 0.00019374237164900514,
      "loss": 12.4741,
      "step": 5135
    },
    {
      "epoch": 0.27967572645034816,
      "grad_norm": 0.7612621418120459,
      "learning_rate": 0.00019373930084889352,
      "loss": 12.5645,
      "step": 5136
    },
    {
      "epoch": 0.27973018044693115,
      "grad_norm": 0.8713698804427293,
      "learning_rate": 0.00019373622931984934,
      "loss": 12.4994,
      "step": 5137
    },
    {
      "epoch": 0.2797846344435142,
      "grad_norm": 0.8327126204853291,
      "learning_rate": 0.00019373315706189655,
      "loss": 12.3497,
      "step": 5138
    },
    {
      "epoch": 0.2798390884400972,
      "grad_norm": 0.7275140927615509,
      "learning_rate": 0.000193730084075059,
      "loss": 12.6424,
      "step": 5139
    },
    {
      "epoch": 0.2798935424366802,
      "grad_norm": 0.7794188688974611,
      "learning_rate": 0.0001937270103593606,
      "loss": 12.5054,
      "step": 5140
    },
    {
      "epoch": 0.27994799643326324,
      "grad_norm": 0.6425933749368322,
      "learning_rate": 0.00019372393591482524,
      "loss": 12.5495,
      "step": 5141
    },
    {
      "epoch": 0.28000245042984623,
      "grad_norm": 0.7633781746174405,
      "learning_rate": 0.00019372086074147685,
      "loss": 12.634,
      "step": 5142
    },
    {
      "epoch": 0.2800569044264292,
      "grad_norm": 0.7500822577947447,
      "learning_rate": 0.00019371778483933934,
      "loss": 12.7041,
      "step": 5143
    },
    {
      "epoch": 0.28011135842301227,
      "grad_norm": 0.6833744271936957,
      "learning_rate": 0.0001937147082084366,
      "loss": 12.5521,
      "step": 5144
    },
    {
      "epoch": 0.28016581241959526,
      "grad_norm": 0.7690187012529756,
      "learning_rate": 0.00019371163084879256,
      "loss": 12.4258,
      "step": 5145
    },
    {
      "epoch": 0.2802202664161783,
      "grad_norm": 0.7467336494036554,
      "learning_rate": 0.00019370855276043121,
      "loss": 12.5721,
      "step": 5146
    },
    {
      "epoch": 0.2802747204127613,
      "grad_norm": 0.7625607151593078,
      "learning_rate": 0.0001937054739433764,
      "loss": 12.5985,
      "step": 5147
    },
    {
      "epoch": 0.2803291744093443,
      "grad_norm": 0.8703165636039294,
      "learning_rate": 0.0001937023943976521,
      "loss": 12.638,
      "step": 5148
    },
    {
      "epoch": 0.28038362840592734,
      "grad_norm": 0.725423606749526,
      "learning_rate": 0.0001936993141232823,
      "loss": 12.5103,
      "step": 5149
    },
    {
      "epoch": 0.28043808240251034,
      "grad_norm": 0.763692107957062,
      "learning_rate": 0.0001936962331202909,
      "loss": 12.5039,
      "step": 5150
    },
    {
      "epoch": 0.28049253639909333,
      "grad_norm": 0.6316565367613988,
      "learning_rate": 0.00019369315138870188,
      "loss": 12.4585,
      "step": 5151
    },
    {
      "epoch": 0.2805469903956764,
      "grad_norm": 0.7350835482152517,
      "learning_rate": 0.0001936900689285392,
      "loss": 12.6043,
      "step": 5152
    },
    {
      "epoch": 0.28060144439225937,
      "grad_norm": 0.6690143152131562,
      "learning_rate": 0.00019368698573982686,
      "loss": 12.5098,
      "step": 5153
    },
    {
      "epoch": 0.28065589838884236,
      "grad_norm": 0.7372146097111318,
      "learning_rate": 0.00019368390182258877,
      "loss": 12.5539,
      "step": 5154
    },
    {
      "epoch": 0.2807103523854254,
      "grad_norm": 0.6211436551518016,
      "learning_rate": 0.00019368081717684898,
      "loss": 12.5419,
      "step": 5155
    },
    {
      "epoch": 0.2807648063820084,
      "grad_norm": 0.7063545203040411,
      "learning_rate": 0.00019367773180263141,
      "loss": 12.4829,
      "step": 5156
    },
    {
      "epoch": 0.2808192603785914,
      "grad_norm": 0.7194962328640749,
      "learning_rate": 0.00019367464569996012,
      "loss": 12.6408,
      "step": 5157
    },
    {
      "epoch": 0.28087371437517444,
      "grad_norm": 0.6871823163250769,
      "learning_rate": 0.00019367155886885906,
      "loss": 12.6144,
      "step": 5158
    },
    {
      "epoch": 0.28092816837175744,
      "grad_norm": 0.6661938851213246,
      "learning_rate": 0.00019366847130935225,
      "loss": 12.5857,
      "step": 5159
    },
    {
      "epoch": 0.28098262236834043,
      "grad_norm": 0.6744656602031699,
      "learning_rate": 0.0001936653830214637,
      "loss": 12.5844,
      "step": 5160
    },
    {
      "epoch": 0.2810370763649235,
      "grad_norm": 0.6865644296890181,
      "learning_rate": 0.00019366229400521743,
      "loss": 12.4689,
      "step": 5161
    },
    {
      "epoch": 0.28109153036150647,
      "grad_norm": 0.7895194676119835,
      "learning_rate": 0.00019365920426063745,
      "loss": 12.7955,
      "step": 5162
    },
    {
      "epoch": 0.28114598435808946,
      "grad_norm": 0.6310374554477493,
      "learning_rate": 0.00019365611378774778,
      "loss": 12.6577,
      "step": 5163
    },
    {
      "epoch": 0.2812004383546725,
      "grad_norm": 1.0253218473524535,
      "learning_rate": 0.0001936530225865725,
      "loss": 12.6858,
      "step": 5164
    },
    {
      "epoch": 0.2812548923512555,
      "grad_norm": 0.6754084523502711,
      "learning_rate": 0.0001936499306571356,
      "loss": 12.6863,
      "step": 5165
    },
    {
      "epoch": 0.2813093463478385,
      "grad_norm": 0.6427002592447747,
      "learning_rate": 0.00019364683799946112,
      "loss": 12.5656,
      "step": 5166
    },
    {
      "epoch": 0.28136380034442154,
      "grad_norm": 0.674802174929845,
      "learning_rate": 0.0001936437446135731,
      "loss": 12.4581,
      "step": 5167
    },
    {
      "epoch": 0.28141825434100454,
      "grad_norm": 0.7627046858697148,
      "learning_rate": 0.00019364065049949566,
      "loss": 12.5511,
      "step": 5168
    },
    {
      "epoch": 0.28147270833758753,
      "grad_norm": 0.6432456260881265,
      "learning_rate": 0.00019363755565725277,
      "loss": 12.4937,
      "step": 5169
    },
    {
      "epoch": 0.2815271623341706,
      "grad_norm": 0.6403274159229349,
      "learning_rate": 0.0001936344600868686,
      "loss": 12.6198,
      "step": 5170
    },
    {
      "epoch": 0.28158161633075357,
      "grad_norm": 0.6892405055573441,
      "learning_rate": 0.00019363136378836712,
      "loss": 12.5993,
      "step": 5171
    },
    {
      "epoch": 0.28163607032733656,
      "grad_norm": 0.6442568030583643,
      "learning_rate": 0.00019362826676177247,
      "loss": 12.5813,
      "step": 5172
    },
    {
      "epoch": 0.2816905243239196,
      "grad_norm": 0.6189612613344044,
      "learning_rate": 0.00019362516900710874,
      "loss": 12.56,
      "step": 5173
    },
    {
      "epoch": 0.2817449783205026,
      "grad_norm": 0.6617043807798307,
      "learning_rate": 0.00019362207052439997,
      "loss": 12.5102,
      "step": 5174
    },
    {
      "epoch": 0.2817994323170856,
      "grad_norm": 0.6492188492603169,
      "learning_rate": 0.00019361897131367028,
      "loss": 12.5758,
      "step": 5175
    },
    {
      "epoch": 0.28185388631366864,
      "grad_norm": 0.7473061625828011,
      "learning_rate": 0.00019361587137494378,
      "loss": 12.6179,
      "step": 5176
    },
    {
      "epoch": 0.28190834031025164,
      "grad_norm": 0.6547901694630822,
      "learning_rate": 0.00019361277070824455,
      "loss": 12.539,
      "step": 5177
    },
    {
      "epoch": 0.28196279430683463,
      "grad_norm": 0.7669279291636536,
      "learning_rate": 0.00019360966931359673,
      "loss": 12.5469,
      "step": 5178
    },
    {
      "epoch": 0.2820172483034177,
      "grad_norm": 0.7186642421638226,
      "learning_rate": 0.0001936065671910244,
      "loss": 12.5702,
      "step": 5179
    },
    {
      "epoch": 0.28207170230000067,
      "grad_norm": 0.6731075258080949,
      "learning_rate": 0.00019360346434055172,
      "loss": 12.51,
      "step": 5180
    },
    {
      "epoch": 0.2821261562965837,
      "grad_norm": 0.6667099900813783,
      "learning_rate": 0.00019360036076220282,
      "loss": 12.5498,
      "step": 5181
    },
    {
      "epoch": 0.2821806102931667,
      "grad_norm": 0.7785210314930021,
      "learning_rate": 0.0001935972564560018,
      "loss": 12.6939,
      "step": 5182
    },
    {
      "epoch": 0.2822350642897497,
      "grad_norm": 0.7067372272734197,
      "learning_rate": 0.0001935941514219728,
      "loss": 12.5571,
      "step": 5183
    },
    {
      "epoch": 0.28228951828633275,
      "grad_norm": 0.678421616002225,
      "learning_rate": 0.00019359104566014,
      "loss": 12.599,
      "step": 5184
    },
    {
      "epoch": 0.28234397228291574,
      "grad_norm": 0.6551604392091743,
      "learning_rate": 0.00019358793917052756,
      "loss": 12.5489,
      "step": 5185
    },
    {
      "epoch": 0.28239842627949874,
      "grad_norm": 0.7103555805268724,
      "learning_rate": 0.00019358483195315958,
      "loss": 12.5438,
      "step": 5186
    },
    {
      "epoch": 0.2824528802760818,
      "grad_norm": 0.7069284288372881,
      "learning_rate": 0.00019358172400806028,
      "loss": 12.5381,
      "step": 5187
    },
    {
      "epoch": 0.2825073342726648,
      "grad_norm": 0.6722826182924235,
      "learning_rate": 0.00019357861533525377,
      "loss": 12.586,
      "step": 5188
    },
    {
      "epoch": 0.28256178826924777,
      "grad_norm": 0.699610418795585,
      "learning_rate": 0.00019357550593476425,
      "loss": 12.5382,
      "step": 5189
    },
    {
      "epoch": 0.2826162422658308,
      "grad_norm": 0.8746550655544723,
      "learning_rate": 0.00019357239580661593,
      "loss": 12.4991,
      "step": 5190
    },
    {
      "epoch": 0.2826706962624138,
      "grad_norm": 0.668798828037185,
      "learning_rate": 0.00019356928495083297,
      "loss": 12.6543,
      "step": 5191
    },
    {
      "epoch": 0.2827251502589968,
      "grad_norm": 0.7044621987621039,
      "learning_rate": 0.00019356617336743951,
      "loss": 12.6718,
      "step": 5192
    },
    {
      "epoch": 0.28277960425557985,
      "grad_norm": 0.6422951635514824,
      "learning_rate": 0.00019356306105645983,
      "loss": 12.5407,
      "step": 5193
    },
    {
      "epoch": 0.28283405825216285,
      "grad_norm": 0.684927068062873,
      "learning_rate": 0.00019355994801791812,
      "loss": 12.532,
      "step": 5194
    },
    {
      "epoch": 0.28288851224874584,
      "grad_norm": 0.723283260192857,
      "learning_rate": 0.00019355683425183854,
      "loss": 12.5527,
      "step": 5195
    },
    {
      "epoch": 0.2829429662453289,
      "grad_norm": 0.7360719862470695,
      "learning_rate": 0.0001935537197582453,
      "loss": 12.6376,
      "step": 5196
    },
    {
      "epoch": 0.2829974202419119,
      "grad_norm": 0.6883098067111958,
      "learning_rate": 0.0001935506045371627,
      "loss": 12.5532,
      "step": 5197
    },
    {
      "epoch": 0.28305187423849487,
      "grad_norm": 0.6584212997077407,
      "learning_rate": 0.00019354748858861487,
      "loss": 12.5902,
      "step": 5198
    },
    {
      "epoch": 0.2831063282350779,
      "grad_norm": 0.6992412627582681,
      "learning_rate": 0.0001935443719126261,
      "loss": 12.5555,
      "step": 5199
    },
    {
      "epoch": 0.2831607822316609,
      "grad_norm": 0.7044431102255994,
      "learning_rate": 0.0001935412545092206,
      "loss": 12.5461,
      "step": 5200
    },
    {
      "epoch": 0.2832152362282439,
      "grad_norm": 0.7073752555792435,
      "learning_rate": 0.00019353813637842265,
      "loss": 12.4574,
      "step": 5201
    },
    {
      "epoch": 0.28326969022482695,
      "grad_norm": 0.7015761112063352,
      "learning_rate": 0.00019353501752025643,
      "loss": 12.6205,
      "step": 5202
    },
    {
      "epoch": 0.28332414422140995,
      "grad_norm": 0.6979274453509807,
      "learning_rate": 0.00019353189793474619,
      "loss": 12.7032,
      "step": 5203
    },
    {
      "epoch": 0.28337859821799294,
      "grad_norm": 0.6957153333269988,
      "learning_rate": 0.0001935287776219163,
      "loss": 12.5931,
      "step": 5204
    },
    {
      "epoch": 0.283433052214576,
      "grad_norm": 0.6897842774068781,
      "learning_rate": 0.00019352565658179092,
      "loss": 12.5964,
      "step": 5205
    },
    {
      "epoch": 0.283487506211159,
      "grad_norm": 0.6774486360522814,
      "learning_rate": 0.00019352253481439436,
      "loss": 12.6081,
      "step": 5206
    },
    {
      "epoch": 0.28354196020774197,
      "grad_norm": 0.6775675886557151,
      "learning_rate": 0.00019351941231975087,
      "loss": 12.5461,
      "step": 5207
    },
    {
      "epoch": 0.283596414204325,
      "grad_norm": 0.6820558537395075,
      "learning_rate": 0.00019351628909788473,
      "loss": 12.4608,
      "step": 5208
    },
    {
      "epoch": 0.283650868200908,
      "grad_norm": 0.6534199168216551,
      "learning_rate": 0.00019351316514882027,
      "loss": 12.5736,
      "step": 5209
    },
    {
      "epoch": 0.283705322197491,
      "grad_norm": 0.6448033586536517,
      "learning_rate": 0.00019351004047258176,
      "loss": 12.6143,
      "step": 5210
    },
    {
      "epoch": 0.28375977619407405,
      "grad_norm": 0.6961255459081882,
      "learning_rate": 0.0001935069150691935,
      "loss": 12.5625,
      "step": 5211
    },
    {
      "epoch": 0.28381423019065705,
      "grad_norm": 0.6788903485850866,
      "learning_rate": 0.00019350378893867975,
      "loss": 12.6184,
      "step": 5212
    },
    {
      "epoch": 0.2838686841872401,
      "grad_norm": 0.6912612299861243,
      "learning_rate": 0.00019350066208106487,
      "loss": 12.608,
      "step": 5213
    },
    {
      "epoch": 0.2839231381838231,
      "grad_norm": 0.6830298112202225,
      "learning_rate": 0.00019349753449637318,
      "loss": 12.7429,
      "step": 5214
    },
    {
      "epoch": 0.2839775921804061,
      "grad_norm": 0.6651906648917761,
      "learning_rate": 0.00019349440618462898,
      "loss": 12.5816,
      "step": 5215
    },
    {
      "epoch": 0.28403204617698913,
      "grad_norm": 0.7197247296245165,
      "learning_rate": 0.00019349127714585657,
      "loss": 12.4892,
      "step": 5216
    },
    {
      "epoch": 0.2840865001735721,
      "grad_norm": 0.6601689123815974,
      "learning_rate": 0.00019348814738008035,
      "loss": 12.5705,
      "step": 5217
    },
    {
      "epoch": 0.2841409541701551,
      "grad_norm": 0.8551875521767103,
      "learning_rate": 0.00019348501688732462,
      "loss": 12.6959,
      "step": 5218
    },
    {
      "epoch": 0.28419540816673816,
      "grad_norm": 0.6824147784459194,
      "learning_rate": 0.00019348188566761367,
      "loss": 12.6101,
      "step": 5219
    },
    {
      "epoch": 0.28424986216332115,
      "grad_norm": 0.6444811599986673,
      "learning_rate": 0.00019347875372097194,
      "loss": 12.5973,
      "step": 5220
    },
    {
      "epoch": 0.28430431615990415,
      "grad_norm": 0.7913232678020938,
      "learning_rate": 0.00019347562104742375,
      "loss": 12.5736,
      "step": 5221
    },
    {
      "epoch": 0.2843587701564872,
      "grad_norm": 0.6667682774936895,
      "learning_rate": 0.0001934724876469934,
      "loss": 12.5546,
      "step": 5222
    },
    {
      "epoch": 0.2844132241530702,
      "grad_norm": 0.6818675106959271,
      "learning_rate": 0.00019346935351970536,
      "loss": 12.5956,
      "step": 5223
    },
    {
      "epoch": 0.2844676781496532,
      "grad_norm": 0.741997237016835,
      "learning_rate": 0.00019346621866558395,
      "loss": 12.6094,
      "step": 5224
    },
    {
      "epoch": 0.28452213214623623,
      "grad_norm": 0.7304820963044663,
      "learning_rate": 0.00019346308308465355,
      "loss": 12.5719,
      "step": 5225
    },
    {
      "epoch": 0.2845765861428192,
      "grad_norm": 0.7698383608402133,
      "learning_rate": 0.00019345994677693855,
      "loss": 12.6115,
      "step": 5226
    },
    {
      "epoch": 0.2846310401394022,
      "grad_norm": 0.6658509131843733,
      "learning_rate": 0.0001934568097424633,
      "loss": 12.4676,
      "step": 5227
    },
    {
      "epoch": 0.28468549413598526,
      "grad_norm": 0.7863371286802923,
      "learning_rate": 0.00019345367198125225,
      "loss": 12.4797,
      "step": 5228
    },
    {
      "epoch": 0.28473994813256825,
      "grad_norm": 0.7296124497290241,
      "learning_rate": 0.00019345053349332977,
      "loss": 12.6438,
      "step": 5229
    },
    {
      "epoch": 0.28479440212915125,
      "grad_norm": 0.7198469126781113,
      "learning_rate": 0.00019344739427872026,
      "loss": 12.6073,
      "step": 5230
    },
    {
      "epoch": 0.2848488561257343,
      "grad_norm": 0.7259120593841006,
      "learning_rate": 0.00019344425433744813,
      "loss": 12.4623,
      "step": 5231
    },
    {
      "epoch": 0.2849033101223173,
      "grad_norm": 0.7251504404787701,
      "learning_rate": 0.00019344111366953782,
      "loss": 12.6461,
      "step": 5232
    },
    {
      "epoch": 0.2849577641189003,
      "grad_norm": 0.7413458025890929,
      "learning_rate": 0.00019343797227501375,
      "loss": 12.6042,
      "step": 5233
    },
    {
      "epoch": 0.28501221811548333,
      "grad_norm": 0.7515706622044661,
      "learning_rate": 0.00019343483015390033,
      "loss": 12.6015,
      "step": 5234
    },
    {
      "epoch": 0.2850666721120663,
      "grad_norm": 0.7603678090629596,
      "learning_rate": 0.000193431687306222,
      "loss": 12.6083,
      "step": 5235
    },
    {
      "epoch": 0.2851211261086493,
      "grad_norm": 0.6856133174388271,
      "learning_rate": 0.0001934285437320032,
      "loss": 12.5179,
      "step": 5236
    },
    {
      "epoch": 0.28517558010523236,
      "grad_norm": 0.6688520904320578,
      "learning_rate": 0.0001934253994312684,
      "loss": 12.5741,
      "step": 5237
    },
    {
      "epoch": 0.28523003410181536,
      "grad_norm": 0.6992197464644883,
      "learning_rate": 0.000193422254404042,
      "loss": 12.5731,
      "step": 5238
    },
    {
      "epoch": 0.28528448809839835,
      "grad_norm": 0.7092290730706611,
      "learning_rate": 0.0001934191086503485,
      "loss": 12.649,
      "step": 5239
    },
    {
      "epoch": 0.2853389420949814,
      "grad_norm": 0.7770595485512367,
      "learning_rate": 0.00019341596217021235,
      "loss": 12.5758,
      "step": 5240
    },
    {
      "epoch": 0.2853933960915644,
      "grad_norm": 0.654164035203944,
      "learning_rate": 0.00019341281496365798,
      "loss": 12.6354,
      "step": 5241
    },
    {
      "epoch": 0.2854478500881474,
      "grad_norm": 0.6837349818438978,
      "learning_rate": 0.0001934096670307099,
      "loss": 12.6693,
      "step": 5242
    },
    {
      "epoch": 0.28550230408473043,
      "grad_norm": 0.7154457385795903,
      "learning_rate": 0.00019340651837139257,
      "loss": 12.5421,
      "step": 5243
    },
    {
      "epoch": 0.2855567580813134,
      "grad_norm": 0.7576311967450255,
      "learning_rate": 0.00019340336898573054,
      "loss": 12.6667,
      "step": 5244
    },
    {
      "epoch": 0.2856112120778964,
      "grad_norm": 0.709766937442989,
      "learning_rate": 0.0001934002188737482,
      "loss": 12.6011,
      "step": 5245
    },
    {
      "epoch": 0.28566566607447946,
      "grad_norm": 0.6936480129371836,
      "learning_rate": 0.00019339706803547015,
      "loss": 12.692,
      "step": 5246
    },
    {
      "epoch": 0.28572012007106246,
      "grad_norm": 0.7300076513644107,
      "learning_rate": 0.0001933939164709208,
      "loss": 12.5928,
      "step": 5247
    },
    {
      "epoch": 0.2857745740676455,
      "grad_norm": 0.8340857027312786,
      "learning_rate": 0.0001933907641801247,
      "loss": 12.775,
      "step": 5248
    },
    {
      "epoch": 0.2858290280642285,
      "grad_norm": 0.7522241994359548,
      "learning_rate": 0.00019338761116310634,
      "loss": 12.626,
      "step": 5249
    },
    {
      "epoch": 0.2858834820608115,
      "grad_norm": 0.7167487172392022,
      "learning_rate": 0.0001933844574198903,
      "loss": 12.5693,
      "step": 5250
    },
    {
      "epoch": 0.28593793605739454,
      "grad_norm": 0.6799345059630523,
      "learning_rate": 0.000193381302950501,
      "loss": 12.4642,
      "step": 5251
    },
    {
      "epoch": 0.28599239005397753,
      "grad_norm": 0.8655447933860508,
      "learning_rate": 0.00019337814775496307,
      "loss": 12.6,
      "step": 5252
    },
    {
      "epoch": 0.2860468440505605,
      "grad_norm": 0.6945184162411392,
      "learning_rate": 0.000193374991833301,
      "loss": 12.5597,
      "step": 5253
    },
    {
      "epoch": 0.28610129804714357,
      "grad_norm": 0.7442352463692852,
      "learning_rate": 0.00019337183518553931,
      "loss": 12.5904,
      "step": 5254
    },
    {
      "epoch": 0.28615575204372656,
      "grad_norm": 0.8143236594223469,
      "learning_rate": 0.0001933686778117026,
      "loss": 12.6048,
      "step": 5255
    },
    {
      "epoch": 0.28621020604030956,
      "grad_norm": 0.6790276390562144,
      "learning_rate": 0.0001933655197118154,
      "loss": 12.5126,
      "step": 5256
    },
    {
      "epoch": 0.2862646600368926,
      "grad_norm": 0.6440104432738986,
      "learning_rate": 0.0001933623608859022,
      "loss": 12.532,
      "step": 5257
    },
    {
      "epoch": 0.2863191140334756,
      "grad_norm": 0.6313806996060203,
      "learning_rate": 0.0001933592013339877,
      "loss": 12.5107,
      "step": 5258
    },
    {
      "epoch": 0.2863735680300586,
      "grad_norm": 0.6764511258985463,
      "learning_rate": 0.00019335604105609632,
      "loss": 12.4815,
      "step": 5259
    },
    {
      "epoch": 0.28642802202664164,
      "grad_norm": 0.6467709401645445,
      "learning_rate": 0.00019335288005225277,
      "loss": 12.6475,
      "step": 5260
    },
    {
      "epoch": 0.28648247602322463,
      "grad_norm": 0.6753532669897918,
      "learning_rate": 0.00019334971832248154,
      "loss": 12.5827,
      "step": 5261
    },
    {
      "epoch": 0.2865369300198076,
      "grad_norm": 0.6827578003307204,
      "learning_rate": 0.00019334655586680723,
      "loss": 12.6088,
      "step": 5262
    },
    {
      "epoch": 0.28659138401639067,
      "grad_norm": 0.7450610764899795,
      "learning_rate": 0.0001933433926852545,
      "loss": 12.6325,
      "step": 5263
    },
    {
      "epoch": 0.28664583801297366,
      "grad_norm": 0.7387077864867498,
      "learning_rate": 0.00019334022877784786,
      "loss": 12.5494,
      "step": 5264
    },
    {
      "epoch": 0.28670029200955666,
      "grad_norm": 0.8186123622954318,
      "learning_rate": 0.00019333706414461195,
      "loss": 12.7283,
      "step": 5265
    },
    {
      "epoch": 0.2867547460061397,
      "grad_norm": 0.718727468080562,
      "learning_rate": 0.00019333389878557137,
      "loss": 12.5985,
      "step": 5266
    },
    {
      "epoch": 0.2868092000027227,
      "grad_norm": 0.6681241377843442,
      "learning_rate": 0.00019333073270075076,
      "loss": 12.5538,
      "step": 5267
    },
    {
      "epoch": 0.2868636539993057,
      "grad_norm": 0.6432945678201138,
      "learning_rate": 0.0001933275658901747,
      "loss": 12.5511,
      "step": 5268
    },
    {
      "epoch": 0.28691810799588874,
      "grad_norm": 0.7267145364610287,
      "learning_rate": 0.00019332439835386786,
      "loss": 12.4169,
      "step": 5269
    },
    {
      "epoch": 0.28697256199247173,
      "grad_norm": 0.6547373420430387,
      "learning_rate": 0.00019332123009185482,
      "loss": 12.54,
      "step": 5270
    },
    {
      "epoch": 0.2870270159890547,
      "grad_norm": 0.6464456886624865,
      "learning_rate": 0.00019331806110416027,
      "loss": 12.6696,
      "step": 5271
    },
    {
      "epoch": 0.28708146998563777,
      "grad_norm": 0.639416556182046,
      "learning_rate": 0.0001933148913908088,
      "loss": 12.606,
      "step": 5272
    },
    {
      "epoch": 0.28713592398222076,
      "grad_norm": 0.765463803610551,
      "learning_rate": 0.00019331172095182511,
      "loss": 12.5596,
      "step": 5273
    },
    {
      "epoch": 0.28719037797880376,
      "grad_norm": 0.6475594591585864,
      "learning_rate": 0.00019330854978723383,
      "loss": 12.57,
      "step": 5274
    },
    {
      "epoch": 0.2872448319753868,
      "grad_norm": 0.738671421515384,
      "learning_rate": 0.00019330537789705963,
      "loss": 12.6256,
      "step": 5275
    },
    {
      "epoch": 0.2872992859719698,
      "grad_norm": 0.6778383446733212,
      "learning_rate": 0.00019330220528132713,
      "loss": 12.4934,
      "step": 5276
    },
    {
      "epoch": 0.2873537399685528,
      "grad_norm": 0.7386529725271362,
      "learning_rate": 0.00019329903194006105,
      "loss": 12.5469,
      "step": 5277
    },
    {
      "epoch": 0.28740819396513584,
      "grad_norm": 0.714724000030084,
      "learning_rate": 0.00019329585787328607,
      "loss": 12.6229,
      "step": 5278
    },
    {
      "epoch": 0.28746264796171883,
      "grad_norm": 0.6353740118316481,
      "learning_rate": 0.00019329268308102685,
      "loss": 12.5172,
      "step": 5279
    },
    {
      "epoch": 0.2875171019583019,
      "grad_norm": 0.7049945078654178,
      "learning_rate": 0.00019328950756330803,
      "loss": 12.5439,
      "step": 5280
    },
    {
      "epoch": 0.28757155595488487,
      "grad_norm": 0.7747680558953951,
      "learning_rate": 0.00019328633132015442,
      "loss": 12.6087,
      "step": 5281
    },
    {
      "epoch": 0.28762600995146786,
      "grad_norm": 0.6793194770644609,
      "learning_rate": 0.00019328315435159058,
      "loss": 12.5076,
      "step": 5282
    },
    {
      "epoch": 0.2876804639480509,
      "grad_norm": 0.7425325939186668,
      "learning_rate": 0.00019327997665764137,
      "loss": 12.6259,
      "step": 5283
    },
    {
      "epoch": 0.2877349179446339,
      "grad_norm": 0.6843901494952742,
      "learning_rate": 0.00019327679823833135,
      "loss": 12.6013,
      "step": 5284
    },
    {
      "epoch": 0.2877893719412169,
      "grad_norm": 0.6285854157813631,
      "learning_rate": 0.00019327361909368535,
      "loss": 12.6006,
      "step": 5285
    },
    {
      "epoch": 0.28784382593779995,
      "grad_norm": 0.6295413056282074,
      "learning_rate": 0.00019327043922372802,
      "loss": 12.4413,
      "step": 5286
    },
    {
      "epoch": 0.28789827993438294,
      "grad_norm": 0.7771093608805613,
      "learning_rate": 0.00019326725862848414,
      "loss": 12.5233,
      "step": 5287
    },
    {
      "epoch": 0.28795273393096593,
      "grad_norm": 0.6761632961873533,
      "learning_rate": 0.0001932640773079784,
      "loss": 12.5032,
      "step": 5288
    },
    {
      "epoch": 0.288007187927549,
      "grad_norm": 0.6379880807498205,
      "learning_rate": 0.00019326089526223558,
      "loss": 12.434,
      "step": 5289
    },
    {
      "epoch": 0.288061641924132,
      "grad_norm": 0.6797398725626824,
      "learning_rate": 0.00019325771249128034,
      "loss": 12.6128,
      "step": 5290
    },
    {
      "epoch": 0.28811609592071497,
      "grad_norm": 0.828936749194387,
      "learning_rate": 0.00019325452899513753,
      "loss": 12.6434,
      "step": 5291
    },
    {
      "epoch": 0.288170549917298,
      "grad_norm": 0.6184918134959372,
      "learning_rate": 0.00019325134477383188,
      "loss": 12.5812,
      "step": 5292
    },
    {
      "epoch": 0.288225003913881,
      "grad_norm": 0.750997225216585,
      "learning_rate": 0.0001932481598273881,
      "loss": 12.5099,
      "step": 5293
    },
    {
      "epoch": 0.288279457910464,
      "grad_norm": 0.7042831646975033,
      "learning_rate": 0.000193244974155831,
      "loss": 12.6465,
      "step": 5294
    },
    {
      "epoch": 0.28833391190704705,
      "grad_norm": 0.8397389045457825,
      "learning_rate": 0.00019324178775918536,
      "loss": 12.6856,
      "step": 5295
    },
    {
      "epoch": 0.28838836590363004,
      "grad_norm": 0.6364067470380718,
      "learning_rate": 0.0001932386006374759,
      "loss": 12.604,
      "step": 5296
    },
    {
      "epoch": 0.28844281990021303,
      "grad_norm": 0.697749718400721,
      "learning_rate": 0.00019323541279072748,
      "loss": 12.6742,
      "step": 5297
    },
    {
      "epoch": 0.2884972738967961,
      "grad_norm": 0.6885876972827212,
      "learning_rate": 0.00019323222421896484,
      "loss": 12.5667,
      "step": 5298
    },
    {
      "epoch": 0.2885517278933791,
      "grad_norm": 0.6932645326014508,
      "learning_rate": 0.00019322903492221283,
      "loss": 12.4531,
      "step": 5299
    },
    {
      "epoch": 0.28860618188996207,
      "grad_norm": 0.73021297394276,
      "learning_rate": 0.00019322584490049616,
      "loss": 12.5136,
      "step": 5300
    },
    {
      "epoch": 0.2886606358865451,
      "grad_norm": 0.7518670769526901,
      "learning_rate": 0.00019322265415383969,
      "loss": 12.5631,
      "step": 5301
    },
    {
      "epoch": 0.2887150898831281,
      "grad_norm": 0.7825021435770262,
      "learning_rate": 0.00019321946268226824,
      "loss": 12.4216,
      "step": 5302
    },
    {
      "epoch": 0.2887695438797111,
      "grad_norm": 0.7673677711990196,
      "learning_rate": 0.00019321627048580662,
      "loss": 12.5662,
      "step": 5303
    },
    {
      "epoch": 0.28882399787629415,
      "grad_norm": 0.7472072722363353,
      "learning_rate": 0.00019321307756447963,
      "loss": 12.5825,
      "step": 5304
    },
    {
      "epoch": 0.28887845187287714,
      "grad_norm": 0.8010391264329372,
      "learning_rate": 0.00019320988391831217,
      "loss": 12.4988,
      "step": 5305
    },
    {
      "epoch": 0.28893290586946013,
      "grad_norm": 0.6811847137339496,
      "learning_rate": 0.00019320668954732898,
      "loss": 12.5728,
      "step": 5306
    },
    {
      "epoch": 0.2889873598660432,
      "grad_norm": 0.7944142383842702,
      "learning_rate": 0.00019320349445155492,
      "loss": 12.4664,
      "step": 5307
    },
    {
      "epoch": 0.2890418138626262,
      "grad_norm": 0.792613651605558,
      "learning_rate": 0.0001932002986310149,
      "loss": 12.7375,
      "step": 5308
    },
    {
      "epoch": 0.28909626785920917,
      "grad_norm": 0.7122210755604018,
      "learning_rate": 0.0001931971020857337,
      "loss": 12.7662,
      "step": 5309
    },
    {
      "epoch": 0.2891507218557922,
      "grad_norm": 0.734802658973589,
      "learning_rate": 0.0001931939048157362,
      "loss": 12.627,
      "step": 5310
    },
    {
      "epoch": 0.2892051758523752,
      "grad_norm": 0.6693328648027058,
      "learning_rate": 0.00019319070682104731,
      "loss": 12.601,
      "step": 5311
    },
    {
      "epoch": 0.2892596298489582,
      "grad_norm": 0.7485113286192445,
      "learning_rate": 0.00019318750810169184,
      "loss": 12.5305,
      "step": 5312
    },
    {
      "epoch": 0.28931408384554125,
      "grad_norm": 0.7768309672774144,
      "learning_rate": 0.00019318430865769464,
      "loss": 12.5792,
      "step": 5313
    },
    {
      "epoch": 0.28936853784212424,
      "grad_norm": 0.6765733125020098,
      "learning_rate": 0.00019318110848908065,
      "loss": 12.7013,
      "step": 5314
    },
    {
      "epoch": 0.2894229918387073,
      "grad_norm": 0.8730525986186539,
      "learning_rate": 0.00019317790759587475,
      "loss": 12.5474,
      "step": 5315
    },
    {
      "epoch": 0.2894774458352903,
      "grad_norm": 0.6296560239385222,
      "learning_rate": 0.0001931747059781018,
      "loss": 12.448,
      "step": 5316
    },
    {
      "epoch": 0.2895318998318733,
      "grad_norm": 0.8603087911123338,
      "learning_rate": 0.0001931715036357867,
      "loss": 12.6103,
      "step": 5317
    },
    {
      "epoch": 0.2895863538284563,
      "grad_norm": 1.3755849613015414,
      "learning_rate": 0.0001931683005689544,
      "loss": 12.4477,
      "step": 5318
    },
    {
      "epoch": 0.2896408078250393,
      "grad_norm": 1.2027269928655746,
      "learning_rate": 0.00019316509677762974,
      "loss": 12.5979,
      "step": 5319
    },
    {
      "epoch": 0.2896952618216223,
      "grad_norm": 0.8655129824139789,
      "learning_rate": 0.00019316189226183767,
      "loss": 12.4687,
      "step": 5320
    },
    {
      "epoch": 0.28974971581820536,
      "grad_norm": 3.817913808390665,
      "learning_rate": 0.00019315868702160312,
      "loss": 12.7316,
      "step": 5321
    },
    {
      "epoch": 0.28980416981478835,
      "grad_norm": 0.8343931440789721,
      "learning_rate": 0.00019315548105695098,
      "loss": 12.6442,
      "step": 5322
    },
    {
      "epoch": 0.28985862381137134,
      "grad_norm": 6.952518704615793,
      "learning_rate": 0.00019315227436790623,
      "loss": 12.8677,
      "step": 5323
    },
    {
      "epoch": 0.2899130778079544,
      "grad_norm": 0.9131341069424375,
      "learning_rate": 0.0001931490669544937,
      "loss": 12.7066,
      "step": 5324
    },
    {
      "epoch": 0.2899675318045374,
      "grad_norm": 1.789670303581222,
      "learning_rate": 0.00019314585881673846,
      "loss": 12.8639,
      "step": 5325
    },
    {
      "epoch": 0.2900219858011204,
      "grad_norm": 1.0026867171007399,
      "learning_rate": 0.0001931426499546654,
      "loss": 12.7645,
      "step": 5326
    },
    {
      "epoch": 0.2900764397977034,
      "grad_norm": 1.0203134278952588,
      "learning_rate": 0.00019313944036829944,
      "loss": 12.7672,
      "step": 5327
    },
    {
      "epoch": 0.2901308937942864,
      "grad_norm": 1.0440168423497518,
      "learning_rate": 0.0001931362300576656,
      "loss": 12.6295,
      "step": 5328
    },
    {
      "epoch": 0.2901853477908694,
      "grad_norm": 1.0227780364237082,
      "learning_rate": 0.0001931330190227888,
      "loss": 12.6797,
      "step": 5329
    },
    {
      "epoch": 0.29023980178745246,
      "grad_norm": 1.139990750255168,
      "learning_rate": 0.00019312980726369404,
      "loss": 12.6139,
      "step": 5330
    },
    {
      "epoch": 0.29029425578403545,
      "grad_norm": 1.1023503359236262,
      "learning_rate": 0.00019312659478040628,
      "loss": 12.6556,
      "step": 5331
    },
    {
      "epoch": 0.29034870978061844,
      "grad_norm": 0.8425326234254836,
      "learning_rate": 0.00019312338157295052,
      "loss": 12.7766,
      "step": 5332
    },
    {
      "epoch": 0.2904031637772015,
      "grad_norm": 0.9347222571312567,
      "learning_rate": 0.0001931201676413517,
      "loss": 12.7023,
      "step": 5333
    },
    {
      "epoch": 0.2904576177737845,
      "grad_norm": 0.7686786588022239,
      "learning_rate": 0.00019311695298563484,
      "loss": 12.7143,
      "step": 5334
    },
    {
      "epoch": 0.2905120717703675,
      "grad_norm": 0.892660220273949,
      "learning_rate": 0.00019311373760582494,
      "loss": 12.7365,
      "step": 5335
    },
    {
      "epoch": 0.2905665257669505,
      "grad_norm": 0.7462257993513306,
      "learning_rate": 0.00019311052150194699,
      "loss": 12.7928,
      "step": 5336
    },
    {
      "epoch": 0.2906209797635335,
      "grad_norm": 0.9456338911665619,
      "learning_rate": 0.00019310730467402603,
      "loss": 12.6129,
      "step": 5337
    },
    {
      "epoch": 0.2906754337601165,
      "grad_norm": 0.8339349763476298,
      "learning_rate": 0.00019310408712208706,
      "loss": 12.6664,
      "step": 5338
    },
    {
      "epoch": 0.29072988775669956,
      "grad_norm": 0.7848635052755665,
      "learning_rate": 0.00019310086884615507,
      "loss": 12.6457,
      "step": 5339
    },
    {
      "epoch": 0.29078434175328255,
      "grad_norm": 0.9228880011106196,
      "learning_rate": 0.00019309764984625513,
      "loss": 12.5757,
      "step": 5340
    },
    {
      "epoch": 0.29083879574986554,
      "grad_norm": 0.7168949910507982,
      "learning_rate": 0.00019309443012241226,
      "loss": 12.5764,
      "step": 5341
    },
    {
      "epoch": 0.2908932497464486,
      "grad_norm": 0.8025730823863417,
      "learning_rate": 0.00019309120967465147,
      "loss": 12.6072,
      "step": 5342
    },
    {
      "epoch": 0.2909477037430316,
      "grad_norm": 0.767974913734144,
      "learning_rate": 0.00019308798850299784,
      "loss": 12.7317,
      "step": 5343
    },
    {
      "epoch": 0.2910021577396146,
      "grad_norm": 0.7643361303317291,
      "learning_rate": 0.0001930847666074764,
      "loss": 12.6485,
      "step": 5344
    },
    {
      "epoch": 0.2910566117361976,
      "grad_norm": 0.7860919414875597,
      "learning_rate": 0.00019308154398811218,
      "loss": 12.597,
      "step": 5345
    },
    {
      "epoch": 0.2911110657327806,
      "grad_norm": 0.7617826115160656,
      "learning_rate": 0.00019307832064493027,
      "loss": 12.7591,
      "step": 5346
    },
    {
      "epoch": 0.29116551972936366,
      "grad_norm": 0.6984577542248323,
      "learning_rate": 0.00019307509657795575,
      "loss": 12.519,
      "step": 5347
    },
    {
      "epoch": 0.29121997372594666,
      "grad_norm": 0.7105852071424831,
      "learning_rate": 0.00019307187178721366,
      "loss": 12.6628,
      "step": 5348
    },
    {
      "epoch": 0.29127442772252965,
      "grad_norm": 0.6818024501380333,
      "learning_rate": 0.0001930686462727291,
      "loss": 12.5448,
      "step": 5349
    },
    {
      "epoch": 0.2913288817191127,
      "grad_norm": 0.6718881976533685,
      "learning_rate": 0.00019306542003452712,
      "loss": 12.6965,
      "step": 5350
    },
    {
      "epoch": 0.2913833357156957,
      "grad_norm": 0.7582962470318719,
      "learning_rate": 0.00019306219307263284,
      "loss": 12.6478,
      "step": 5351
    },
    {
      "epoch": 0.2914377897122787,
      "grad_norm": 0.8820400481683287,
      "learning_rate": 0.0001930589653870713,
      "loss": 12.6285,
      "step": 5352
    },
    {
      "epoch": 0.29149224370886173,
      "grad_norm": 0.9586742288093614,
      "learning_rate": 0.0001930557369778677,
      "loss": 12.7713,
      "step": 5353
    },
    {
      "epoch": 0.2915466977054447,
      "grad_norm": 0.7338990360689434,
      "learning_rate": 0.00019305250784504706,
      "loss": 12.6302,
      "step": 5354
    },
    {
      "epoch": 0.2916011517020277,
      "grad_norm": 0.7582419862630847,
      "learning_rate": 0.0001930492779886345,
      "loss": 12.6738,
      "step": 5355
    },
    {
      "epoch": 0.29165560569861076,
      "grad_norm": 0.840787783063441,
      "learning_rate": 0.00019304604740865515,
      "loss": 12.6798,
      "step": 5356
    },
    {
      "epoch": 0.29171005969519376,
      "grad_norm": 0.6931359406914137,
      "learning_rate": 0.00019304281610513414,
      "loss": 12.505,
      "step": 5357
    },
    {
      "epoch": 0.29176451369177675,
      "grad_norm": 0.8046820695731451,
      "learning_rate": 0.00019303958407809656,
      "loss": 12.545,
      "step": 5358
    },
    {
      "epoch": 0.2918189676883598,
      "grad_norm": 0.7504096676621198,
      "learning_rate": 0.00019303635132756762,
      "loss": 12.488,
      "step": 5359
    },
    {
      "epoch": 0.2918734216849428,
      "grad_norm": 0.6977037675551906,
      "learning_rate": 0.0001930331178535724,
      "loss": 12.4553,
      "step": 5360
    },
    {
      "epoch": 0.2919278756815258,
      "grad_norm": 0.7411988969206796,
      "learning_rate": 0.00019302988365613603,
      "loss": 12.6309,
      "step": 5361
    },
    {
      "epoch": 0.29198232967810883,
      "grad_norm": 0.7279922787640098,
      "learning_rate": 0.0001930266487352837,
      "loss": 12.5111,
      "step": 5362
    },
    {
      "epoch": 0.2920367836746918,
      "grad_norm": 0.7811018864235276,
      "learning_rate": 0.00019302341309104055,
      "loss": 12.6204,
      "step": 5363
    },
    {
      "epoch": 0.2920912376712748,
      "grad_norm": 0.7559002866108901,
      "learning_rate": 0.00019302017672343172,
      "loss": 12.4904,
      "step": 5364
    },
    {
      "epoch": 0.29214569166785787,
      "grad_norm": 0.7466752855960884,
      "learning_rate": 0.00019301693963248243,
      "loss": 12.7218,
      "step": 5365
    },
    {
      "epoch": 0.29220014566444086,
      "grad_norm": 0.7993271004037386,
      "learning_rate": 0.00019301370181821782,
      "loss": 12.6146,
      "step": 5366
    },
    {
      "epoch": 0.29225459966102385,
      "grad_norm": 0.6983084821208461,
      "learning_rate": 0.00019301046328066304,
      "loss": 12.5839,
      "step": 5367
    },
    {
      "epoch": 0.2923090536576069,
      "grad_norm": 0.8523665600366496,
      "learning_rate": 0.00019300722401984332,
      "loss": 12.6677,
      "step": 5368
    },
    {
      "epoch": 0.2923635076541899,
      "grad_norm": 0.8207120293181236,
      "learning_rate": 0.0001930039840357838,
      "loss": 12.5995,
      "step": 5369
    },
    {
      "epoch": 0.2924179616507729,
      "grad_norm": 0.6721964030090202,
      "learning_rate": 0.0001930007433285097,
      "loss": 12.4554,
      "step": 5370
    },
    {
      "epoch": 0.29247241564735593,
      "grad_norm": 0.6800829699572934,
      "learning_rate": 0.00019299750189804624,
      "loss": 12.6383,
      "step": 5371
    },
    {
      "epoch": 0.2925268696439389,
      "grad_norm": 0.8289854580672297,
      "learning_rate": 0.00019299425974441862,
      "loss": 12.6417,
      "step": 5372
    },
    {
      "epoch": 0.2925813236405219,
      "grad_norm": 0.7095324885178306,
      "learning_rate": 0.00019299101686765205,
      "loss": 12.5994,
      "step": 5373
    },
    {
      "epoch": 0.29263577763710497,
      "grad_norm": 0.6896328151277851,
      "learning_rate": 0.00019298777326777171,
      "loss": 12.3786,
      "step": 5374
    },
    {
      "epoch": 0.29269023163368796,
      "grad_norm": 0.6994439095041498,
      "learning_rate": 0.00019298452894480286,
      "loss": 12.6018,
      "step": 5375
    },
    {
      "epoch": 0.29274468563027095,
      "grad_norm": 0.80451622927954,
      "learning_rate": 0.00019298128389877073,
      "loss": 12.6109,
      "step": 5376
    },
    {
      "epoch": 0.292799139626854,
      "grad_norm": 0.6689060517208213,
      "learning_rate": 0.00019297803812970052,
      "loss": 12.5704,
      "step": 5377
    },
    {
      "epoch": 0.292853593623437,
      "grad_norm": 0.687941492721622,
      "learning_rate": 0.00019297479163761755,
      "loss": 12.6013,
      "step": 5378
    },
    {
      "epoch": 0.29290804762002,
      "grad_norm": 0.6930940676293417,
      "learning_rate": 0.00019297154442254693,
      "loss": 12.5302,
      "step": 5379
    },
    {
      "epoch": 0.29296250161660303,
      "grad_norm": 0.7573733554612111,
      "learning_rate": 0.00019296829648451404,
      "loss": 12.6814,
      "step": 5380
    },
    {
      "epoch": 0.293016955613186,
      "grad_norm": 0.6483541127785282,
      "learning_rate": 0.00019296504782354408,
      "loss": 12.5151,
      "step": 5381
    },
    {
      "epoch": 0.2930714096097691,
      "grad_norm": 0.6872342229388375,
      "learning_rate": 0.0001929617984396623,
      "loss": 12.5216,
      "step": 5382
    },
    {
      "epoch": 0.29312586360635207,
      "grad_norm": 0.7146685548807061,
      "learning_rate": 0.000192958548332894,
      "loss": 12.8383,
      "step": 5383
    },
    {
      "epoch": 0.29318031760293506,
      "grad_norm": 0.7128076251945845,
      "learning_rate": 0.00019295529750326443,
      "loss": 12.5799,
      "step": 5384
    },
    {
      "epoch": 0.2932347715995181,
      "grad_norm": 0.6913826142462048,
      "learning_rate": 0.0001929520459507989,
      "loss": 12.5626,
      "step": 5385
    },
    {
      "epoch": 0.2932892255961011,
      "grad_norm": 0.7745424972710939,
      "learning_rate": 0.00019294879367552263,
      "loss": 12.6744,
      "step": 5386
    },
    {
      "epoch": 0.2933436795926841,
      "grad_norm": 0.6975593164609588,
      "learning_rate": 0.00019294554067746098,
      "loss": 12.5228,
      "step": 5387
    },
    {
      "epoch": 0.29339813358926714,
      "grad_norm": 0.7678680088261888,
      "learning_rate": 0.0001929422869566392,
      "loss": 12.6008,
      "step": 5388
    },
    {
      "epoch": 0.29345258758585013,
      "grad_norm": 0.692367853277397,
      "learning_rate": 0.00019293903251308266,
      "loss": 12.6218,
      "step": 5389
    },
    {
      "epoch": 0.2935070415824331,
      "grad_norm": 0.8232991412974388,
      "learning_rate": 0.00019293577734681656,
      "loss": 12.7118,
      "step": 5390
    },
    {
      "epoch": 0.2935614955790162,
      "grad_norm": 0.6167356791640606,
      "learning_rate": 0.0001929325214578663,
      "loss": 12.4945,
      "step": 5391
    },
    {
      "epoch": 0.29361594957559917,
      "grad_norm": 0.6438436500011632,
      "learning_rate": 0.00019292926484625714,
      "loss": 12.6473,
      "step": 5392
    },
    {
      "epoch": 0.29367040357218216,
      "grad_norm": 0.7587583883204099,
      "learning_rate": 0.00019292600751201448,
      "loss": 12.5496,
      "step": 5393
    },
    {
      "epoch": 0.2937248575687652,
      "grad_norm": 0.7076624710372246,
      "learning_rate": 0.00019292274945516359,
      "loss": 12.3377,
      "step": 5394
    },
    {
      "epoch": 0.2937793115653482,
      "grad_norm": 0.8745028515304532,
      "learning_rate": 0.00019291949067572978,
      "loss": 12.5201,
      "step": 5395
    },
    {
      "epoch": 0.2938337655619312,
      "grad_norm": 0.7570972500445633,
      "learning_rate": 0.00019291623117373847,
      "loss": 12.5978,
      "step": 5396
    },
    {
      "epoch": 0.29388821955851424,
      "grad_norm": 0.7985090072616512,
      "learning_rate": 0.00019291297094921494,
      "loss": 12.5774,
      "step": 5397
    },
    {
      "epoch": 0.29394267355509723,
      "grad_norm": 0.8410990037792181,
      "learning_rate": 0.00019290971000218457,
      "loss": 12.7778,
      "step": 5398
    },
    {
      "epoch": 0.2939971275516802,
      "grad_norm": 0.6325474282429254,
      "learning_rate": 0.0001929064483326727,
      "loss": 12.5482,
      "step": 5399
    },
    {
      "epoch": 0.2940515815482633,
      "grad_norm": 0.7215869597593331,
      "learning_rate": 0.00019290318594070475,
      "loss": 12.6419,
      "step": 5400
    },
    {
      "epoch": 0.29410603554484627,
      "grad_norm": 0.7388693424632827,
      "learning_rate": 0.00019289992282630602,
      "loss": 12.5937,
      "step": 5401
    },
    {
      "epoch": 0.29416048954142926,
      "grad_norm": 0.9311053731409538,
      "learning_rate": 0.0001928966589895019,
      "loss": 12.673,
      "step": 5402
    },
    {
      "epoch": 0.2942149435380123,
      "grad_norm": 0.7564163948266747,
      "learning_rate": 0.00019289339443031778,
      "loss": 12.5103,
      "step": 5403
    },
    {
      "epoch": 0.2942693975345953,
      "grad_norm": 0.7845040973885391,
      "learning_rate": 0.00019289012914877905,
      "loss": 12.5595,
      "step": 5404
    },
    {
      "epoch": 0.2943238515311783,
      "grad_norm": 0.7268199640406403,
      "learning_rate": 0.00019288686314491115,
      "loss": 12.5784,
      "step": 5405
    },
    {
      "epoch": 0.29437830552776134,
      "grad_norm": 0.7117435618249173,
      "learning_rate": 0.00019288359641873935,
      "loss": 12.5584,
      "step": 5406
    },
    {
      "epoch": 0.29443275952434433,
      "grad_norm": 0.7430669941676982,
      "learning_rate": 0.00019288032897028917,
      "loss": 12.5339,
      "step": 5407
    },
    {
      "epoch": 0.2944872135209273,
      "grad_norm": 0.7309612448126213,
      "learning_rate": 0.00019287706079958595,
      "loss": 12.6017,
      "step": 5408
    },
    {
      "epoch": 0.2945416675175104,
      "grad_norm": 0.7591821592955744,
      "learning_rate": 0.00019287379190665517,
      "loss": 12.5533,
      "step": 5409
    },
    {
      "epoch": 0.29459612151409337,
      "grad_norm": 0.7451068498540538,
      "learning_rate": 0.0001928705222915222,
      "loss": 12.4671,
      "step": 5410
    },
    {
      "epoch": 0.29465057551067636,
      "grad_norm": 0.7720985974444161,
      "learning_rate": 0.00019286725195421243,
      "loss": 12.6444,
      "step": 5411
    },
    {
      "epoch": 0.2947050295072594,
      "grad_norm": 0.686883058089928,
      "learning_rate": 0.00019286398089475134,
      "loss": 12.5692,
      "step": 5412
    },
    {
      "epoch": 0.2947594835038424,
      "grad_norm": 0.7205956053915606,
      "learning_rate": 0.0001928607091131644,
      "loss": 12.7151,
      "step": 5413
    },
    {
      "epoch": 0.29481393750042545,
      "grad_norm": 0.7336566961369465,
      "learning_rate": 0.000192857436609477,
      "loss": 12.5186,
      "step": 5414
    },
    {
      "epoch": 0.29486839149700844,
      "grad_norm": 0.6770882797888417,
      "learning_rate": 0.0001928541633837146,
      "loss": 12.5483,
      "step": 5415
    },
    {
      "epoch": 0.29492284549359143,
      "grad_norm": 0.6642706676974497,
      "learning_rate": 0.00019285088943590267,
      "loss": 12.6158,
      "step": 5416
    },
    {
      "epoch": 0.2949772994901745,
      "grad_norm": 0.6612267667643044,
      "learning_rate": 0.00019284761476606662,
      "loss": 12.6026,
      "step": 5417
    },
    {
      "epoch": 0.2950317534867575,
      "grad_norm": 0.8225131551740218,
      "learning_rate": 0.00019284433937423196,
      "loss": 12.7422,
      "step": 5418
    },
    {
      "epoch": 0.29508620748334047,
      "grad_norm": 0.9312293048136258,
      "learning_rate": 0.00019284106326042415,
      "loss": 12.5606,
      "step": 5419
    },
    {
      "epoch": 0.2951406614799235,
      "grad_norm": 0.686764621159285,
      "learning_rate": 0.00019283778642466864,
      "loss": 12.6587,
      "step": 5420
    },
    {
      "epoch": 0.2951951154765065,
      "grad_norm": 0.7232553800589134,
      "learning_rate": 0.000192834508866991,
      "loss": 12.6868,
      "step": 5421
    },
    {
      "epoch": 0.2952495694730895,
      "grad_norm": 0.7771593086870223,
      "learning_rate": 0.0001928312305874166,
      "loss": 12.6444,
      "step": 5422
    },
    {
      "epoch": 0.29530402346967255,
      "grad_norm": 0.6821004747227135,
      "learning_rate": 0.00019282795158597098,
      "loss": 12.5451,
      "step": 5423
    },
    {
      "epoch": 0.29535847746625554,
      "grad_norm": 0.7158378326219672,
      "learning_rate": 0.00019282467186267966,
      "loss": 12.621,
      "step": 5424
    },
    {
      "epoch": 0.29541293146283853,
      "grad_norm": 0.748752470958464,
      "learning_rate": 0.0001928213914175681,
      "loss": 12.6017,
      "step": 5425
    },
    {
      "epoch": 0.2954673854594216,
      "grad_norm": 0.709105723596565,
      "learning_rate": 0.00019281811025066183,
      "loss": 12.6099,
      "step": 5426
    },
    {
      "epoch": 0.2955218394560046,
      "grad_norm": 0.658589593936804,
      "learning_rate": 0.0001928148283619864,
      "loss": 12.6061,
      "step": 5427
    },
    {
      "epoch": 0.29557629345258757,
      "grad_norm": 0.7269034855071365,
      "learning_rate": 0.0001928115457515673,
      "loss": 12.6086,
      "step": 5428
    },
    {
      "epoch": 0.2956307474491706,
      "grad_norm": 0.8569243295411642,
      "learning_rate": 0.00019280826241943003,
      "loss": 12.7698,
      "step": 5429
    },
    {
      "epoch": 0.2956852014457536,
      "grad_norm": 0.7756779258888888,
      "learning_rate": 0.00019280497836560016,
      "loss": 12.569,
      "step": 5430
    },
    {
      "epoch": 0.2957396554423366,
      "grad_norm": 0.7274438498794342,
      "learning_rate": 0.00019280169359010322,
      "loss": 12.7412,
      "step": 5431
    },
    {
      "epoch": 0.29579410943891965,
      "grad_norm": 0.6509255649787569,
      "learning_rate": 0.00019279840809296474,
      "loss": 12.4884,
      "step": 5432
    },
    {
      "epoch": 0.29584856343550264,
      "grad_norm": 0.6538988898602316,
      "learning_rate": 0.0001927951218742103,
      "loss": 12.525,
      "step": 5433
    },
    {
      "epoch": 0.29590301743208564,
      "grad_norm": 0.678004336147839,
      "learning_rate": 0.00019279183493386542,
      "loss": 12.663,
      "step": 5434
    },
    {
      "epoch": 0.2959574714286687,
      "grad_norm": 0.6604899315721932,
      "learning_rate": 0.00019278854727195564,
      "loss": 12.5462,
      "step": 5435
    },
    {
      "epoch": 0.2960119254252517,
      "grad_norm": 0.650865153177635,
      "learning_rate": 0.00019278525888850658,
      "loss": 12.5251,
      "step": 5436
    },
    {
      "epoch": 0.29606637942183467,
      "grad_norm": 0.6905167141476374,
      "learning_rate": 0.0001927819697835438,
      "loss": 12.4289,
      "step": 5437
    },
    {
      "epoch": 0.2961208334184177,
      "grad_norm": 0.6775595408948848,
      "learning_rate": 0.00019277867995709286,
      "loss": 12.5849,
      "step": 5438
    },
    {
      "epoch": 0.2961752874150007,
      "grad_norm": 0.6750514376962531,
      "learning_rate": 0.0001927753894091793,
      "loss": 12.5668,
      "step": 5439
    },
    {
      "epoch": 0.2962297414115837,
      "grad_norm": 0.7530309678722148,
      "learning_rate": 0.0001927720981398288,
      "loss": 12.4954,
      "step": 5440
    },
    {
      "epoch": 0.29628419540816675,
      "grad_norm": 0.6601666121573037,
      "learning_rate": 0.0001927688061490669,
      "loss": 12.4699,
      "step": 5441
    },
    {
      "epoch": 0.29633864940474974,
      "grad_norm": 0.6954527578087399,
      "learning_rate": 0.0001927655134369192,
      "loss": 12.5202,
      "step": 5442
    },
    {
      "epoch": 0.29639310340133274,
      "grad_norm": 0.6441927111198292,
      "learning_rate": 0.0001927622200034113,
      "loss": 12.5601,
      "step": 5443
    },
    {
      "epoch": 0.2964475573979158,
      "grad_norm": 0.6474561946353591,
      "learning_rate": 0.00019275892584856883,
      "loss": 12.485,
      "step": 5444
    },
    {
      "epoch": 0.2965020113944988,
      "grad_norm": 0.6933453342076746,
      "learning_rate": 0.0001927556309724174,
      "loss": 12.4715,
      "step": 5445
    },
    {
      "epoch": 0.29655646539108177,
      "grad_norm": 0.7682158885479328,
      "learning_rate": 0.00019275233537498264,
      "loss": 12.6883,
      "step": 5446
    },
    {
      "epoch": 0.2966109193876648,
      "grad_norm": 0.6739621499325937,
      "learning_rate": 0.00019274903905629014,
      "loss": 12.6544,
      "step": 5447
    },
    {
      "epoch": 0.2966653733842478,
      "grad_norm": 0.6354207232839544,
      "learning_rate": 0.00019274574201636556,
      "loss": 12.4936,
      "step": 5448
    },
    {
      "epoch": 0.29671982738083086,
      "grad_norm": 0.658172915455953,
      "learning_rate": 0.00019274244425523455,
      "loss": 12.6144,
      "step": 5449
    },
    {
      "epoch": 0.29677428137741385,
      "grad_norm": 0.6766134115227272,
      "learning_rate": 0.00019273914577292274,
      "loss": 12.5919,
      "step": 5450
    },
    {
      "epoch": 0.29682873537399684,
      "grad_norm": 0.6557699542841391,
      "learning_rate": 0.0001927358465694558,
      "loss": 12.6577,
      "step": 5451
    },
    {
      "epoch": 0.2968831893705799,
      "grad_norm": 0.6875351309441434,
      "learning_rate": 0.00019273254664485933,
      "loss": 12.6079,
      "step": 5452
    },
    {
      "epoch": 0.2969376433671629,
      "grad_norm": 0.667682654878768,
      "learning_rate": 0.00019272924599915902,
      "loss": 12.5817,
      "step": 5453
    },
    {
      "epoch": 0.2969920973637459,
      "grad_norm": 0.6779324528523384,
      "learning_rate": 0.00019272594463238057,
      "loss": 12.5989,
      "step": 5454
    },
    {
      "epoch": 0.2970465513603289,
      "grad_norm": 0.624243957794944,
      "learning_rate": 0.00019272264254454962,
      "loss": 12.521,
      "step": 5455
    },
    {
      "epoch": 0.2971010053569119,
      "grad_norm": 1.088983734238408,
      "learning_rate": 0.00019271933973569186,
      "loss": 12.5177,
      "step": 5456
    },
    {
      "epoch": 0.2971554593534949,
      "grad_norm": 0.7190029177790371,
      "learning_rate": 0.00019271603620583293,
      "loss": 12.5043,
      "step": 5457
    },
    {
      "epoch": 0.29720991335007796,
      "grad_norm": 0.6624008667037063,
      "learning_rate": 0.00019271273195499856,
      "loss": 12.5598,
      "step": 5458
    },
    {
      "epoch": 0.29726436734666095,
      "grad_norm": 0.829880367133911,
      "learning_rate": 0.0001927094269832145,
      "loss": 12.7031,
      "step": 5459
    },
    {
      "epoch": 0.29731882134324394,
      "grad_norm": 0.6624179427876619,
      "learning_rate": 0.00019270612129050632,
      "loss": 12.496,
      "step": 5460
    },
    {
      "epoch": 0.297373275339827,
      "grad_norm": 0.7302525172954663,
      "learning_rate": 0.00019270281487689982,
      "loss": 12.6206,
      "step": 5461
    },
    {
      "epoch": 0.29742772933641,
      "grad_norm": 0.7417326761674216,
      "learning_rate": 0.0001926995077424207,
      "loss": 12.7291,
      "step": 5462
    },
    {
      "epoch": 0.297482183332993,
      "grad_norm": 0.6515096566908593,
      "learning_rate": 0.00019269619988709466,
      "loss": 12.4573,
      "step": 5463
    },
    {
      "epoch": 0.297536637329576,
      "grad_norm": 0.660489757020803,
      "learning_rate": 0.0001926928913109474,
      "loss": 12.6159,
      "step": 5464
    },
    {
      "epoch": 0.297591091326159,
      "grad_norm": 0.6976139432730505,
      "learning_rate": 0.00019268958201400466,
      "loss": 12.4181,
      "step": 5465
    },
    {
      "epoch": 0.297645545322742,
      "grad_norm": 0.7180785514358637,
      "learning_rate": 0.0001926862719962922,
      "loss": 12.5845,
      "step": 5466
    },
    {
      "epoch": 0.29769999931932506,
      "grad_norm": 0.6611172259802919,
      "learning_rate": 0.00019268296125783576,
      "loss": 12.6051,
      "step": 5467
    },
    {
      "epoch": 0.29775445331590805,
      "grad_norm": 0.836722423389593,
      "learning_rate": 0.00019267964979866108,
      "loss": 12.5857,
      "step": 5468
    },
    {
      "epoch": 0.29780890731249104,
      "grad_norm": 0.6652025661518768,
      "learning_rate": 0.0001926763376187939,
      "loss": 12.6493,
      "step": 5469
    },
    {
      "epoch": 0.2978633613090741,
      "grad_norm": 0.7509908152819234,
      "learning_rate": 0.00019267302471825994,
      "loss": 12.657,
      "step": 5470
    },
    {
      "epoch": 0.2979178153056571,
      "grad_norm": 0.6440705345244854,
      "learning_rate": 0.00019266971109708502,
      "loss": 12.6235,
      "step": 5471
    },
    {
      "epoch": 0.2979722693022401,
      "grad_norm": 0.7864417869922518,
      "learning_rate": 0.0001926663967552949,
      "loss": 12.6673,
      "step": 5472
    },
    {
      "epoch": 0.2980267232988231,
      "grad_norm": 0.675295748172897,
      "learning_rate": 0.00019266308169291533,
      "loss": 12.5634,
      "step": 5473
    },
    {
      "epoch": 0.2980811772954061,
      "grad_norm": 0.7196578050262258,
      "learning_rate": 0.00019265976590997208,
      "loss": 12.5545,
      "step": 5474
    },
    {
      "epoch": 0.2981356312919891,
      "grad_norm": 0.771184431209636,
      "learning_rate": 0.00019265644940649095,
      "loss": 12.6641,
      "step": 5475
    },
    {
      "epoch": 0.29819008528857216,
      "grad_norm": 0.6625549461496372,
      "learning_rate": 0.00019265313218249776,
      "loss": 12.646,
      "step": 5476
    },
    {
      "epoch": 0.29824453928515515,
      "grad_norm": 0.6714856696134495,
      "learning_rate": 0.00019264981423801824,
      "loss": 12.4311,
      "step": 5477
    },
    {
      "epoch": 0.29829899328173815,
      "grad_norm": 0.6932694439785722,
      "learning_rate": 0.00019264649557307825,
      "loss": 12.66,
      "step": 5478
    },
    {
      "epoch": 0.2983534472783212,
      "grad_norm": 0.7019140285607096,
      "learning_rate": 0.00019264317618770358,
      "loss": 12.5443,
      "step": 5479
    },
    {
      "epoch": 0.2984079012749042,
      "grad_norm": 0.7285887391443051,
      "learning_rate": 0.00019263985608192004,
      "loss": 12.5274,
      "step": 5480
    },
    {
      "epoch": 0.29846235527148723,
      "grad_norm": 0.6743474455149153,
      "learning_rate": 0.0001926365352557534,
      "loss": 12.4967,
      "step": 5481
    },
    {
      "epoch": 0.2985168092680702,
      "grad_norm": 0.7497025945227477,
      "learning_rate": 0.00019263321370922956,
      "loss": 12.6501,
      "step": 5482
    },
    {
      "epoch": 0.2985712632646532,
      "grad_norm": 0.7301840359950457,
      "learning_rate": 0.0001926298914423743,
      "loss": 12.5288,
      "step": 5483
    },
    {
      "epoch": 0.29862571726123627,
      "grad_norm": 0.6743215481536896,
      "learning_rate": 0.0001926265684552135,
      "loss": 12.5529,
      "step": 5484
    },
    {
      "epoch": 0.29868017125781926,
      "grad_norm": 0.7238826933992838,
      "learning_rate": 0.00019262324474777297,
      "loss": 12.5079,
      "step": 5485
    },
    {
      "epoch": 0.29873462525440225,
      "grad_norm": 0.7257838708320444,
      "learning_rate": 0.00019261992032007852,
      "loss": 12.5381,
      "step": 5486
    },
    {
      "epoch": 0.2987890792509853,
      "grad_norm": 0.7158109479170218,
      "learning_rate": 0.00019261659517215608,
      "loss": 12.555,
      "step": 5487
    },
    {
      "epoch": 0.2988435332475683,
      "grad_norm": 0.8007473442264141,
      "learning_rate": 0.00019261326930403142,
      "loss": 12.6532,
      "step": 5488
    },
    {
      "epoch": 0.2988979872441513,
      "grad_norm": 0.6721247192677079,
      "learning_rate": 0.00019260994271573048,
      "loss": 12.6069,
      "step": 5489
    },
    {
      "epoch": 0.29895244124073433,
      "grad_norm": 0.7861205605607333,
      "learning_rate": 0.00019260661540727907,
      "loss": 12.5759,
      "step": 5490
    },
    {
      "epoch": 0.2990068952373173,
      "grad_norm": 0.6645713018669855,
      "learning_rate": 0.0001926032873787031,
      "loss": 12.55,
      "step": 5491
    },
    {
      "epoch": 0.2990613492339003,
      "grad_norm": 0.6227237580241888,
      "learning_rate": 0.00019259995863002844,
      "loss": 12.5693,
      "step": 5492
    },
    {
      "epoch": 0.29911580323048337,
      "grad_norm": 0.7182566580451716,
      "learning_rate": 0.00019259662916128097,
      "loss": 12.6455,
      "step": 5493
    },
    {
      "epoch": 0.29917025722706636,
      "grad_norm": 0.753398157729987,
      "learning_rate": 0.00019259329897248657,
      "loss": 12.6879,
      "step": 5494
    },
    {
      "epoch": 0.29922471122364935,
      "grad_norm": 0.6709225872282627,
      "learning_rate": 0.00019258996806367117,
      "loss": 12.6203,
      "step": 5495
    },
    {
      "epoch": 0.2992791652202324,
      "grad_norm": 0.7117214422038949,
      "learning_rate": 0.00019258663643486067,
      "loss": 12.569,
      "step": 5496
    },
    {
      "epoch": 0.2993336192168154,
      "grad_norm": 0.6641688877669308,
      "learning_rate": 0.00019258330408608088,
      "loss": 12.5621,
      "step": 5497
    },
    {
      "epoch": 0.2993880732133984,
      "grad_norm": 0.753378664160101,
      "learning_rate": 0.00019257997101735787,
      "loss": 12.6489,
      "step": 5498
    },
    {
      "epoch": 0.29944252720998144,
      "grad_norm": 0.6934161511406282,
      "learning_rate": 0.00019257663722871746,
      "loss": 12.584,
      "step": 5499
    },
    {
      "epoch": 0.2994969812065644,
      "grad_norm": 0.6696183362421898,
      "learning_rate": 0.0001925733027201856,
      "loss": 12.6701,
      "step": 5500
    },
    {
      "epoch": 0.2995514352031474,
      "grad_norm": 0.7221308273553433,
      "learning_rate": 0.0001925699674917882,
      "loss": 12.5649,
      "step": 5501
    },
    {
      "epoch": 0.29960588919973047,
      "grad_norm": 0.6253773296377263,
      "learning_rate": 0.00019256663154355118,
      "loss": 12.4283,
      "step": 5502
    },
    {
      "epoch": 0.29966034319631346,
      "grad_norm": 0.6572697114319784,
      "learning_rate": 0.00019256329487550054,
      "loss": 12.6613,
      "step": 5503
    },
    {
      "epoch": 0.29971479719289645,
      "grad_norm": 0.747338700790793,
      "learning_rate": 0.0001925599574876622,
      "loss": 12.5483,
      "step": 5504
    },
    {
      "epoch": 0.2997692511894795,
      "grad_norm": 0.6906470887856688,
      "learning_rate": 0.0001925566193800621,
      "loss": 12.5226,
      "step": 5505
    },
    {
      "epoch": 0.2998237051860625,
      "grad_norm": 0.7248451578017936,
      "learning_rate": 0.00019255328055272624,
      "loss": 12.3915,
      "step": 5506
    },
    {
      "epoch": 0.2998781591826455,
      "grad_norm": 0.6518355865071961,
      "learning_rate": 0.0001925499410056805,
      "loss": 12.5653,
      "step": 5507
    },
    {
      "epoch": 0.29993261317922854,
      "grad_norm": 0.6928218232220432,
      "learning_rate": 0.00019254660073895092,
      "loss": 12.5504,
      "step": 5508
    },
    {
      "epoch": 0.29998706717581153,
      "grad_norm": 0.7400814092127778,
      "learning_rate": 0.00019254325975256344,
      "loss": 12.6619,
      "step": 5509
    },
    {
      "epoch": 0.3000415211723945,
      "grad_norm": 0.6781743427380154,
      "learning_rate": 0.00019253991804654407,
      "loss": 12.4567,
      "step": 5510
    },
    {
      "epoch": 0.30009597516897757,
      "grad_norm": 0.6651358657090074,
      "learning_rate": 0.00019253657562091876,
      "loss": 12.6494,
      "step": 5511
    },
    {
      "epoch": 0.30015042916556056,
      "grad_norm": 0.7184827329752417,
      "learning_rate": 0.00019253323247571356,
      "loss": 12.6796,
      "step": 5512
    },
    {
      "epoch": 0.30020488316214355,
      "grad_norm": 0.9321121206445516,
      "learning_rate": 0.0001925298886109544,
      "loss": 12.7286,
      "step": 5513
    },
    {
      "epoch": 0.3002593371587266,
      "grad_norm": 0.7634100043428852,
      "learning_rate": 0.00019252654402666727,
      "loss": 12.7075,
      "step": 5514
    },
    {
      "epoch": 0.3003137911553096,
      "grad_norm": 0.6500345565420961,
      "learning_rate": 0.00019252319872287824,
      "loss": 12.5077,
      "step": 5515
    },
    {
      "epoch": 0.30036824515189264,
      "grad_norm": 0.7131473979329023,
      "learning_rate": 0.00019251985269961336,
      "loss": 12.6077,
      "step": 5516
    },
    {
      "epoch": 0.30042269914847564,
      "grad_norm": 0.7203919344303918,
      "learning_rate": 0.00019251650595689853,
      "loss": 12.56,
      "step": 5517
    },
    {
      "epoch": 0.30047715314505863,
      "grad_norm": 0.6665760403028634,
      "learning_rate": 0.00019251315849475983,
      "loss": 12.578,
      "step": 5518
    },
    {
      "epoch": 0.3005316071416417,
      "grad_norm": 0.7873972857774381,
      "learning_rate": 0.00019250981031322334,
      "loss": 12.6244,
      "step": 5519
    },
    {
      "epoch": 0.30058606113822467,
      "grad_norm": 0.6749580655656026,
      "learning_rate": 0.00019250646141231502,
      "loss": 12.4792,
      "step": 5520
    },
    {
      "epoch": 0.30064051513480766,
      "grad_norm": 0.670393709534476,
      "learning_rate": 0.00019250311179206092,
      "loss": 12.4619,
      "step": 5521
    },
    {
      "epoch": 0.3006949691313907,
      "grad_norm": 0.6852404075991394,
      "learning_rate": 0.00019249976145248714,
      "loss": 12.5381,
      "step": 5522
    },
    {
      "epoch": 0.3007494231279737,
      "grad_norm": 0.7438713110773497,
      "learning_rate": 0.0001924964103936197,
      "loss": 12.5529,
      "step": 5523
    },
    {
      "epoch": 0.3008038771245567,
      "grad_norm": 0.9422519483715919,
      "learning_rate": 0.00019249305861548466,
      "loss": 12.4879,
      "step": 5524
    },
    {
      "epoch": 0.30085833112113974,
      "grad_norm": 0.7523445937435863,
      "learning_rate": 0.00019248970611810808,
      "loss": 12.572,
      "step": 5525
    },
    {
      "epoch": 0.30091278511772274,
      "grad_norm": 0.7152895724954156,
      "learning_rate": 0.00019248635290151605,
      "loss": 12.5248,
      "step": 5526
    },
    {
      "epoch": 0.30096723911430573,
      "grad_norm": 0.6343277394603853,
      "learning_rate": 0.0001924829989657346,
      "loss": 12.5778,
      "step": 5527
    },
    {
      "epoch": 0.3010216931108888,
      "grad_norm": 0.8243064110016005,
      "learning_rate": 0.00019247964431078987,
      "loss": 12.5656,
      "step": 5528
    },
    {
      "epoch": 0.30107614710747177,
      "grad_norm": 0.691016405989972,
      "learning_rate": 0.0001924762889367079,
      "loss": 12.5543,
      "step": 5529
    },
    {
      "epoch": 0.30113060110405476,
      "grad_norm": 0.6264685213458856,
      "learning_rate": 0.00019247293284351482,
      "loss": 12.443,
      "step": 5530
    },
    {
      "epoch": 0.3011850551006378,
      "grad_norm": 0.7283528943500993,
      "learning_rate": 0.00019246957603123667,
      "loss": 12.7175,
      "step": 5531
    },
    {
      "epoch": 0.3012395090972208,
      "grad_norm": 0.7925468803778901,
      "learning_rate": 0.00019246621849989962,
      "loss": 12.6486,
      "step": 5532
    },
    {
      "epoch": 0.3012939630938038,
      "grad_norm": 0.7083148796953568,
      "learning_rate": 0.00019246286024952975,
      "loss": 12.4917,
      "step": 5533
    },
    {
      "epoch": 0.30134841709038684,
      "grad_norm": 0.6956661001652812,
      "learning_rate": 0.00019245950128015315,
      "loss": 12.6302,
      "step": 5534
    },
    {
      "epoch": 0.30140287108696984,
      "grad_norm": 0.7563057436080856,
      "learning_rate": 0.000192456141591796,
      "loss": 12.224,
      "step": 5535
    },
    {
      "epoch": 0.30145732508355283,
      "grad_norm": 0.6947294851805417,
      "learning_rate": 0.00019245278118448436,
      "loss": 12.5657,
      "step": 5536
    },
    {
      "epoch": 0.3015117790801359,
      "grad_norm": 0.824002618811491,
      "learning_rate": 0.00019244942005824437,
      "loss": 12.6116,
      "step": 5537
    },
    {
      "epoch": 0.30156623307671887,
      "grad_norm": 0.7311410585858291,
      "learning_rate": 0.00019244605821310223,
      "loss": 12.5115,
      "step": 5538
    },
    {
      "epoch": 0.30162068707330186,
      "grad_norm": 0.6903199645204022,
      "learning_rate": 0.000192442695649084,
      "loss": 12.6615,
      "step": 5539
    },
    {
      "epoch": 0.3016751410698849,
      "grad_norm": 0.7494884431333453,
      "learning_rate": 0.0001924393323662159,
      "loss": 12.6129,
      "step": 5540
    },
    {
      "epoch": 0.3017295950664679,
      "grad_norm": 0.6206778446236714,
      "learning_rate": 0.00019243596836452404,
      "loss": 12.5177,
      "step": 5541
    },
    {
      "epoch": 0.3017840490630509,
      "grad_norm": 0.6255558313734257,
      "learning_rate": 0.00019243260364403458,
      "loss": 12.503,
      "step": 5542
    },
    {
      "epoch": 0.30183850305963394,
      "grad_norm": 0.7196447240519136,
      "learning_rate": 0.00019242923820477368,
      "loss": 12.4861,
      "step": 5543
    },
    {
      "epoch": 0.30189295705621694,
      "grad_norm": 0.6812891385581262,
      "learning_rate": 0.00019242587204676754,
      "loss": 12.5954,
      "step": 5544
    },
    {
      "epoch": 0.30194741105279993,
      "grad_norm": 0.7749909003181377,
      "learning_rate": 0.0001924225051700423,
      "loss": 12.7635,
      "step": 5545
    },
    {
      "epoch": 0.302001865049383,
      "grad_norm": 0.6774056944182675,
      "learning_rate": 0.00019241913757462418,
      "loss": 12.6497,
      "step": 5546
    },
    {
      "epoch": 0.30205631904596597,
      "grad_norm": 0.7271543948033461,
      "learning_rate": 0.00019241576926053936,
      "loss": 12.5526,
      "step": 5547
    },
    {
      "epoch": 0.302110773042549,
      "grad_norm": 0.7058606456982123,
      "learning_rate": 0.00019241240022781398,
      "loss": 12.5594,
      "step": 5548
    },
    {
      "epoch": 0.302165227039132,
      "grad_norm": 0.6679446080027783,
      "learning_rate": 0.0001924090304764743,
      "loss": 12.6517,
      "step": 5549
    },
    {
      "epoch": 0.302219681035715,
      "grad_norm": 0.749234625813053,
      "learning_rate": 0.00019240566000654653,
      "loss": 12.6246,
      "step": 5550
    },
    {
      "epoch": 0.30227413503229805,
      "grad_norm": 0.7261391403812401,
      "learning_rate": 0.0001924022888180568,
      "loss": 12.5897,
      "step": 5551
    },
    {
      "epoch": 0.30232858902888105,
      "grad_norm": 0.7054391146170403,
      "learning_rate": 0.00019239891691103143,
      "loss": 12.5725,
      "step": 5552
    },
    {
      "epoch": 0.30238304302546404,
      "grad_norm": 0.7040813463132223,
      "learning_rate": 0.00019239554428549655,
      "loss": 12.5343,
      "step": 5553
    },
    {
      "epoch": 0.3024374970220471,
      "grad_norm": 0.7371343689528318,
      "learning_rate": 0.00019239217094147844,
      "loss": 12.7817,
      "step": 5554
    },
    {
      "epoch": 0.3024919510186301,
      "grad_norm": 0.6871079768052895,
      "learning_rate": 0.0001923887968790033,
      "loss": 12.5964,
      "step": 5555
    },
    {
      "epoch": 0.30254640501521307,
      "grad_norm": 0.789857271839963,
      "learning_rate": 0.0001923854220980974,
      "loss": 12.5383,
      "step": 5556
    },
    {
      "epoch": 0.3026008590117961,
      "grad_norm": 0.640539924809829,
      "learning_rate": 0.00019238204659878692,
      "loss": 12.4824,
      "step": 5557
    },
    {
      "epoch": 0.3026553130083791,
      "grad_norm": 0.727503213368253,
      "learning_rate": 0.00019237867038109823,
      "loss": 12.6039,
      "step": 5558
    },
    {
      "epoch": 0.3027097670049621,
      "grad_norm": 0.773168800791757,
      "learning_rate": 0.00019237529344505745,
      "loss": 12.5989,
      "step": 5559
    },
    {
      "epoch": 0.30276422100154515,
      "grad_norm": 0.6301257451951049,
      "learning_rate": 0.0001923719157906909,
      "loss": 12.4452,
      "step": 5560
    },
    {
      "epoch": 0.30281867499812815,
      "grad_norm": 0.7305946483618813,
      "learning_rate": 0.00019236853741802485,
      "loss": 12.6938,
      "step": 5561
    },
    {
      "epoch": 0.30287312899471114,
      "grad_norm": 0.8053865211101275,
      "learning_rate": 0.00019236515832708558,
      "loss": 12.6093,
      "step": 5562
    },
    {
      "epoch": 0.3029275829912942,
      "grad_norm": 0.9633833175581472,
      "learning_rate": 0.00019236177851789931,
      "loss": 12.6175,
      "step": 5563
    },
    {
      "epoch": 0.3029820369878772,
      "grad_norm": 0.6972993933779662,
      "learning_rate": 0.0001923583979904924,
      "loss": 12.4328,
      "step": 5564
    },
    {
      "epoch": 0.30303649098446017,
      "grad_norm": 0.778723825604669,
      "learning_rate": 0.0001923550167448911,
      "loss": 12.592,
      "step": 5565
    },
    {
      "epoch": 0.3030909449810432,
      "grad_norm": 0.6520394554064838,
      "learning_rate": 0.00019235163478112166,
      "loss": 12.5261,
      "step": 5566
    },
    {
      "epoch": 0.3031453989776262,
      "grad_norm": 0.6738229807373549,
      "learning_rate": 0.00019234825209921047,
      "loss": 12.4971,
      "step": 5567
    },
    {
      "epoch": 0.3031998529742092,
      "grad_norm": 0.7779394650488719,
      "learning_rate": 0.00019234486869918377,
      "loss": 12.5064,
      "step": 5568
    },
    {
      "epoch": 0.30325430697079225,
      "grad_norm": 0.678783994289326,
      "learning_rate": 0.00019234148458106785,
      "loss": 12.5691,
      "step": 5569
    },
    {
      "epoch": 0.30330876096737525,
      "grad_norm": 0.830383539000743,
      "learning_rate": 0.0001923380997448891,
      "loss": 12.559,
      "step": 5570
    },
    {
      "epoch": 0.30336321496395824,
      "grad_norm": 0.7130216579918996,
      "learning_rate": 0.00019233471419067378,
      "loss": 12.5613,
      "step": 5571
    },
    {
      "epoch": 0.3034176689605413,
      "grad_norm": 0.6777881662814074,
      "learning_rate": 0.00019233132791844827,
      "loss": 12.4879,
      "step": 5572
    },
    {
      "epoch": 0.3034721229571243,
      "grad_norm": 0.6787392866295541,
      "learning_rate": 0.00019232794092823884,
      "loss": 12.5894,
      "step": 5573
    },
    {
      "epoch": 0.30352657695370727,
      "grad_norm": 0.694352921112891,
      "learning_rate": 0.00019232455322007184,
      "loss": 12.4459,
      "step": 5574
    },
    {
      "epoch": 0.3035810309502903,
      "grad_norm": 0.7222532512087805,
      "learning_rate": 0.00019232116479397365,
      "loss": 12.6226,
      "step": 5575
    },
    {
      "epoch": 0.3036354849468733,
      "grad_norm": 0.7314508042146828,
      "learning_rate": 0.0001923177756499706,
      "loss": 12.689,
      "step": 5576
    },
    {
      "epoch": 0.3036899389434563,
      "grad_norm": 0.6587607739908065,
      "learning_rate": 0.00019231438578808907,
      "loss": 12.4694,
      "step": 5577
    },
    {
      "epoch": 0.30374439294003935,
      "grad_norm": 0.7635540326543027,
      "learning_rate": 0.00019231099520835535,
      "loss": 12.5989,
      "step": 5578
    },
    {
      "epoch": 0.30379884693662235,
      "grad_norm": 0.7361430250165957,
      "learning_rate": 0.0001923076039107959,
      "loss": 12.5804,
      "step": 5579
    },
    {
      "epoch": 0.30385330093320534,
      "grad_norm": 0.7585903650131236,
      "learning_rate": 0.000192304211895437,
      "loss": 12.5181,
      "step": 5580
    },
    {
      "epoch": 0.3039077549297884,
      "grad_norm": 0.7600365444488244,
      "learning_rate": 0.0001923008191623051,
      "loss": 12.5854,
      "step": 5581
    },
    {
      "epoch": 0.3039622089263714,
      "grad_norm": 0.8182750314826289,
      "learning_rate": 0.00019229742571142655,
      "loss": 12.4639,
      "step": 5582
    },
    {
      "epoch": 0.30401666292295443,
      "grad_norm": 0.7193099733132268,
      "learning_rate": 0.00019229403154282773,
      "loss": 12.4747,
      "step": 5583
    },
    {
      "epoch": 0.3040711169195374,
      "grad_norm": 0.7270686488441531,
      "learning_rate": 0.00019229063665653504,
      "loss": 12.5602,
      "step": 5584
    },
    {
      "epoch": 0.3041255709161204,
      "grad_norm": 0.7135070121903767,
      "learning_rate": 0.00019228724105257487,
      "loss": 12.5432,
      "step": 5585
    },
    {
      "epoch": 0.30418002491270346,
      "grad_norm": 0.6997931480434294,
      "learning_rate": 0.00019228384473097366,
      "loss": 12.6373,
      "step": 5586
    },
    {
      "epoch": 0.30423447890928645,
      "grad_norm": 0.7479379443675384,
      "learning_rate": 0.0001922804476917578,
      "loss": 12.4373,
      "step": 5587
    },
    {
      "epoch": 0.30428893290586945,
      "grad_norm": 0.6981561283860745,
      "learning_rate": 0.0001922770499349537,
      "loss": 12.6439,
      "step": 5588
    },
    {
      "epoch": 0.3043433869024525,
      "grad_norm": 0.6339087812412747,
      "learning_rate": 0.00019227365146058775,
      "loss": 12.5173,
      "step": 5589
    },
    {
      "epoch": 0.3043978408990355,
      "grad_norm": 0.8609219927489593,
      "learning_rate": 0.00019227025226868644,
      "loss": 12.5421,
      "step": 5590
    },
    {
      "epoch": 0.3044522948956185,
      "grad_norm": 0.7118254434995729,
      "learning_rate": 0.00019226685235927617,
      "loss": 12.5478,
      "step": 5591
    },
    {
      "epoch": 0.30450674889220153,
      "grad_norm": 0.6794508342052569,
      "learning_rate": 0.0001922634517323834,
      "loss": 12.5299,
      "step": 5592
    },
    {
      "epoch": 0.3045612028887845,
      "grad_norm": 0.6586240853434104,
      "learning_rate": 0.00019226005038803452,
      "loss": 12.6055,
      "step": 5593
    },
    {
      "epoch": 0.3046156568853675,
      "grad_norm": 0.77236087411886,
      "learning_rate": 0.00019225664832625604,
      "loss": 12.5975,
      "step": 5594
    },
    {
      "epoch": 0.30467011088195056,
      "grad_norm": 0.6719406111328031,
      "learning_rate": 0.0001922532455470744,
      "loss": 12.5498,
      "step": 5595
    },
    {
      "epoch": 0.30472456487853355,
      "grad_norm": 0.6505332143229099,
      "learning_rate": 0.00019224984205051603,
      "loss": 12.5016,
      "step": 5596
    },
    {
      "epoch": 0.30477901887511655,
      "grad_norm": 0.6457023063792332,
      "learning_rate": 0.00019224643783660744,
      "loss": 12.5565,
      "step": 5597
    },
    {
      "epoch": 0.3048334728716996,
      "grad_norm": 0.7990481600920323,
      "learning_rate": 0.00019224303290537508,
      "loss": 12.7244,
      "step": 5598
    },
    {
      "epoch": 0.3048879268682826,
      "grad_norm": 0.6815455908401977,
      "learning_rate": 0.00019223962725684542,
      "loss": 12.5248,
      "step": 5599
    },
    {
      "epoch": 0.3049423808648656,
      "grad_norm": 0.7270449504186137,
      "learning_rate": 0.0001922362208910449,
      "loss": 12.7111,
      "step": 5600
    },
    {
      "epoch": 0.30499683486144863,
      "grad_norm": 0.7763686333017574,
      "learning_rate": 0.0001922328138080001,
      "loss": 12.7218,
      "step": 5601
    },
    {
      "epoch": 0.3050512888580316,
      "grad_norm": 0.7284964808752892,
      "learning_rate": 0.0001922294060077375,
      "loss": 12.5986,
      "step": 5602
    },
    {
      "epoch": 0.3051057428546146,
      "grad_norm": 0.6652999267288491,
      "learning_rate": 0.00019222599749028354,
      "loss": 12.6224,
      "step": 5603
    },
    {
      "epoch": 0.30516019685119766,
      "grad_norm": 0.6816881250974832,
      "learning_rate": 0.00019222258825566478,
      "loss": 12.5041,
      "step": 5604
    },
    {
      "epoch": 0.30521465084778066,
      "grad_norm": 1.0545143198680806,
      "learning_rate": 0.0001922191783039077,
      "loss": 12.4324,
      "step": 5605
    },
    {
      "epoch": 0.30526910484436365,
      "grad_norm": 0.6908851851058719,
      "learning_rate": 0.00019221576763503882,
      "loss": 12.5667,
      "step": 5606
    },
    {
      "epoch": 0.3053235588409467,
      "grad_norm": 0.7155566720462256,
      "learning_rate": 0.00019221235624908466,
      "loss": 12.4893,
      "step": 5607
    },
    {
      "epoch": 0.3053780128375297,
      "grad_norm": 0.7677589294724539,
      "learning_rate": 0.00019220894414607176,
      "loss": 12.5424,
      "step": 5608
    },
    {
      "epoch": 0.3054324668341127,
      "grad_norm": 0.6674925417573994,
      "learning_rate": 0.00019220553132602664,
      "loss": 12.4021,
      "step": 5609
    },
    {
      "epoch": 0.30548692083069573,
      "grad_norm": 0.7080962937903494,
      "learning_rate": 0.00019220211778897585,
      "loss": 12.6609,
      "step": 5610
    },
    {
      "epoch": 0.3055413748272787,
      "grad_norm": 0.6452885530540028,
      "learning_rate": 0.0001921987035349459,
      "loss": 12.4599,
      "step": 5611
    },
    {
      "epoch": 0.3055958288238617,
      "grad_norm": 0.6681628070045235,
      "learning_rate": 0.00019219528856396343,
      "loss": 12.5753,
      "step": 5612
    },
    {
      "epoch": 0.30565028282044476,
      "grad_norm": 0.6669997947535996,
      "learning_rate": 0.00019219187287605491,
      "loss": 12.4837,
      "step": 5613
    },
    {
      "epoch": 0.30570473681702776,
      "grad_norm": 0.7083059515105996,
      "learning_rate": 0.0001921884564712469,
      "loss": 12.6538,
      "step": 5614
    },
    {
      "epoch": 0.3057591908136108,
      "grad_norm": 0.642663359481727,
      "learning_rate": 0.00019218503934956602,
      "loss": 12.5575,
      "step": 5615
    },
    {
      "epoch": 0.3058136448101938,
      "grad_norm": 0.6625459280194421,
      "learning_rate": 0.0001921816215110388,
      "loss": 12.6102,
      "step": 5616
    },
    {
      "epoch": 0.3058680988067768,
      "grad_norm": 0.7182820810684242,
      "learning_rate": 0.00019217820295569185,
      "loss": 12.6836,
      "step": 5617
    },
    {
      "epoch": 0.30592255280335984,
      "grad_norm": 0.753708065496197,
      "learning_rate": 0.00019217478368355173,
      "loss": 12.557,
      "step": 5618
    },
    {
      "epoch": 0.30597700679994283,
      "grad_norm": 0.7136433532288005,
      "learning_rate": 0.00019217136369464503,
      "loss": 12.6235,
      "step": 5619
    },
    {
      "epoch": 0.3060314607965258,
      "grad_norm": 0.7244893505472123,
      "learning_rate": 0.00019216794298899833,
      "loss": 12.5793,
      "step": 5620
    },
    {
      "epoch": 0.30608591479310887,
      "grad_norm": 0.6818378516149921,
      "learning_rate": 0.00019216452156663828,
      "loss": 12.5673,
      "step": 5621
    },
    {
      "epoch": 0.30614036878969186,
      "grad_norm": 0.6811251930879814,
      "learning_rate": 0.00019216109942759145,
      "loss": 12.6225,
      "step": 5622
    },
    {
      "epoch": 0.30619482278627486,
      "grad_norm": 0.6603144888339486,
      "learning_rate": 0.00019215767657188444,
      "loss": 12.4698,
      "step": 5623
    },
    {
      "epoch": 0.3062492767828579,
      "grad_norm": 0.6823421129114194,
      "learning_rate": 0.00019215425299954389,
      "loss": 12.5283,
      "step": 5624
    },
    {
      "epoch": 0.3063037307794409,
      "grad_norm": 0.6790904544641594,
      "learning_rate": 0.0001921508287105964,
      "loss": 12.5425,
      "step": 5625
    },
    {
      "epoch": 0.3063581847760239,
      "grad_norm": 0.6295559442425349,
      "learning_rate": 0.00019214740370506863,
      "loss": 12.554,
      "step": 5626
    },
    {
      "epoch": 0.30641263877260694,
      "grad_norm": 0.6822492643876135,
      "learning_rate": 0.0001921439779829872,
      "loss": 12.5317,
      "step": 5627
    },
    {
      "epoch": 0.30646709276918993,
      "grad_norm": 0.7730070563114577,
      "learning_rate": 0.00019214055154437873,
      "loss": 12.6159,
      "step": 5628
    },
    {
      "epoch": 0.3065215467657729,
      "grad_norm": 0.865180713811351,
      "learning_rate": 0.00019213712438926987,
      "loss": 12.6358,
      "step": 5629
    },
    {
      "epoch": 0.30657600076235597,
      "grad_norm": 0.651411926195066,
      "learning_rate": 0.00019213369651768732,
      "loss": 12.5468,
      "step": 5630
    },
    {
      "epoch": 0.30663045475893896,
      "grad_norm": 0.6433491563576922,
      "learning_rate": 0.00019213026792965767,
      "loss": 12.5849,
      "step": 5631
    },
    {
      "epoch": 0.30668490875552196,
      "grad_norm": 0.7643907663694384,
      "learning_rate": 0.00019212683862520756,
      "loss": 12.7352,
      "step": 5632
    },
    {
      "epoch": 0.306739362752105,
      "grad_norm": 0.6868282637682263,
      "learning_rate": 0.00019212340860436377,
      "loss": 12.5962,
      "step": 5633
    },
    {
      "epoch": 0.306793816748688,
      "grad_norm": 0.7598560634641103,
      "learning_rate": 0.00019211997786715287,
      "loss": 12.6059,
      "step": 5634
    },
    {
      "epoch": 0.306848270745271,
      "grad_norm": 0.7288171464459738,
      "learning_rate": 0.0001921165464136016,
      "loss": 12.5483,
      "step": 5635
    },
    {
      "epoch": 0.30690272474185404,
      "grad_norm": 0.6673891343969206,
      "learning_rate": 0.00019211311424373662,
      "loss": 12.6215,
      "step": 5636
    },
    {
      "epoch": 0.30695717873843703,
      "grad_norm": 0.6744339225596823,
      "learning_rate": 0.00019210968135758457,
      "loss": 12.5486,
      "step": 5637
    },
    {
      "epoch": 0.30701163273502,
      "grad_norm": 0.7259913596649055,
      "learning_rate": 0.00019210624775517225,
      "loss": 12.5398,
      "step": 5638
    },
    {
      "epoch": 0.30706608673160307,
      "grad_norm": 0.6830246301584015,
      "learning_rate": 0.00019210281343652626,
      "loss": 12.7559,
      "step": 5639
    },
    {
      "epoch": 0.30712054072818606,
      "grad_norm": 0.6785998966172879,
      "learning_rate": 0.00019209937840167338,
      "loss": 12.4974,
      "step": 5640
    },
    {
      "epoch": 0.30717499472476906,
      "grad_norm": 0.8090827555182567,
      "learning_rate": 0.00019209594265064025,
      "loss": 12.5941,
      "step": 5641
    },
    {
      "epoch": 0.3072294487213521,
      "grad_norm": 0.7481489237592109,
      "learning_rate": 0.00019209250618345368,
      "loss": 12.5914,
      "step": 5642
    },
    {
      "epoch": 0.3072839027179351,
      "grad_norm": 0.6838197491142659,
      "learning_rate": 0.0001920890690001403,
      "loss": 12.6042,
      "step": 5643
    },
    {
      "epoch": 0.3073383567145181,
      "grad_norm": 0.7250501351581554,
      "learning_rate": 0.00019208563110072687,
      "loss": 12.512,
      "step": 5644
    },
    {
      "epoch": 0.30739281071110114,
      "grad_norm": 0.8233771978321007,
      "learning_rate": 0.00019208219248524014,
      "loss": 12.5859,
      "step": 5645
    },
    {
      "epoch": 0.30744726470768413,
      "grad_norm": 0.6777946678039877,
      "learning_rate": 0.0001920787531537068,
      "loss": 12.6482,
      "step": 5646
    },
    {
      "epoch": 0.3075017187042671,
      "grad_norm": 0.7460758660384199,
      "learning_rate": 0.0001920753131061537,
      "loss": 12.6217,
      "step": 5647
    },
    {
      "epoch": 0.30755617270085017,
      "grad_norm": 0.6870262122278785,
      "learning_rate": 0.0001920718723426075,
      "loss": 12.6333,
      "step": 5648
    },
    {
      "epoch": 0.30761062669743316,
      "grad_norm": 0.7277903111871953,
      "learning_rate": 0.00019206843086309498,
      "loss": 12.511,
      "step": 5649
    },
    {
      "epoch": 0.3076650806940162,
      "grad_norm": 0.698143994457559,
      "learning_rate": 0.00019206498866764288,
      "loss": 12.5323,
      "step": 5650
    },
    {
      "epoch": 0.3077195346905992,
      "grad_norm": 0.6871120504062017,
      "learning_rate": 0.00019206154575627802,
      "loss": 12.3899,
      "step": 5651
    },
    {
      "epoch": 0.3077739886871822,
      "grad_norm": 0.8309989003900052,
      "learning_rate": 0.00019205810212902713,
      "loss": 12.3704,
      "step": 5652
    },
    {
      "epoch": 0.30782844268376525,
      "grad_norm": 0.8425343080407433,
      "learning_rate": 0.00019205465778591698,
      "loss": 12.6307,
      "step": 5653
    },
    {
      "epoch": 0.30788289668034824,
      "grad_norm": 0.6798149257167375,
      "learning_rate": 0.0001920512127269744,
      "loss": 12.6095,
      "step": 5654
    },
    {
      "epoch": 0.30793735067693123,
      "grad_norm": 1.059836564785302,
      "learning_rate": 0.00019204776695222616,
      "loss": 12.655,
      "step": 5655
    },
    {
      "epoch": 0.3079918046735143,
      "grad_norm": 0.7725726412551481,
      "learning_rate": 0.00019204432046169903,
      "loss": 12.6174,
      "step": 5656
    },
    {
      "epoch": 0.3080462586700973,
      "grad_norm": 0.8319369769046256,
      "learning_rate": 0.0001920408732554198,
      "loss": 12.627,
      "step": 5657
    },
    {
      "epoch": 0.30810071266668027,
      "grad_norm": 0.7314135357215268,
      "learning_rate": 0.00019203742533341534,
      "loss": 12.6254,
      "step": 5658
    },
    {
      "epoch": 0.3081551666632633,
      "grad_norm": 0.761529435714816,
      "learning_rate": 0.00019203397669571243,
      "loss": 12.6114,
      "step": 5659
    },
    {
      "epoch": 0.3082096206598463,
      "grad_norm": 0.8174963212126977,
      "learning_rate": 0.00019203052734233786,
      "loss": 12.6282,
      "step": 5660
    },
    {
      "epoch": 0.3082640746564293,
      "grad_norm": 0.6595276890115349,
      "learning_rate": 0.0001920270772733185,
      "loss": 12.5181,
      "step": 5661
    },
    {
      "epoch": 0.30831852865301235,
      "grad_norm": 0.7448298217772923,
      "learning_rate": 0.00019202362648868112,
      "loss": 12.5468,
      "step": 5662
    },
    {
      "epoch": 0.30837298264959534,
      "grad_norm": 0.7681547956272415,
      "learning_rate": 0.00019202017498845265,
      "loss": 12.5377,
      "step": 5663
    },
    {
      "epoch": 0.30842743664617833,
      "grad_norm": 0.6643886990380933,
      "learning_rate": 0.00019201672277265982,
      "loss": 12.6046,
      "step": 5664
    },
    {
      "epoch": 0.3084818906427614,
      "grad_norm": 0.792858194541144,
      "learning_rate": 0.00019201326984132953,
      "loss": 12.5158,
      "step": 5665
    },
    {
      "epoch": 0.3085363446393444,
      "grad_norm": 0.7871328263741095,
      "learning_rate": 0.00019200981619448863,
      "loss": 12.6348,
      "step": 5666
    },
    {
      "epoch": 0.30859079863592737,
      "grad_norm": 0.6989003752746422,
      "learning_rate": 0.00019200636183216397,
      "loss": 12.5612,
      "step": 5667
    },
    {
      "epoch": 0.3086452526325104,
      "grad_norm": 0.840356018473379,
      "learning_rate": 0.0001920029067543824,
      "loss": 12.5499,
      "step": 5668
    },
    {
      "epoch": 0.3086997066290934,
      "grad_norm": 0.6120244188572662,
      "learning_rate": 0.0001919994509611708,
      "loss": 12.5905,
      "step": 5669
    },
    {
      "epoch": 0.3087541606256764,
      "grad_norm": 0.7276319280590692,
      "learning_rate": 0.00019199599445255606,
      "loss": 12.5964,
      "step": 5670
    },
    {
      "epoch": 0.30880861462225945,
      "grad_norm": 0.7024582912295813,
      "learning_rate": 0.000191992537228565,
      "loss": 12.5345,
      "step": 5671
    },
    {
      "epoch": 0.30886306861884244,
      "grad_norm": 0.6446613012579606,
      "learning_rate": 0.00019198907928922457,
      "loss": 12.5057,
      "step": 5672
    },
    {
      "epoch": 0.30891752261542543,
      "grad_norm": 0.7342596906337274,
      "learning_rate": 0.00019198562063456163,
      "loss": 12.5339,
      "step": 5673
    },
    {
      "epoch": 0.3089719766120085,
      "grad_norm": 0.6693775956414059,
      "learning_rate": 0.00019198216126460306,
      "loss": 12.497,
      "step": 5674
    },
    {
      "epoch": 0.3090264306085915,
      "grad_norm": 0.654154004134321,
      "learning_rate": 0.0001919787011793758,
      "loss": 12.5619,
      "step": 5675
    },
    {
      "epoch": 0.30908088460517447,
      "grad_norm": 0.6961529048076709,
      "learning_rate": 0.00019197524037890674,
      "loss": 12.5442,
      "step": 5676
    },
    {
      "epoch": 0.3091353386017575,
      "grad_norm": 0.6364479952408855,
      "learning_rate": 0.0001919717788632228,
      "loss": 12.4765,
      "step": 5677
    },
    {
      "epoch": 0.3091897925983405,
      "grad_norm": 0.7031647693885875,
      "learning_rate": 0.00019196831663235083,
      "loss": 12.496,
      "step": 5678
    },
    {
      "epoch": 0.3092442465949235,
      "grad_norm": 0.6375027088057562,
      "learning_rate": 0.00019196485368631785,
      "loss": 12.4551,
      "step": 5679
    },
    {
      "epoch": 0.30929870059150655,
      "grad_norm": 0.6646869200935568,
      "learning_rate": 0.00019196139002515073,
      "loss": 12.6098,
      "step": 5680
    },
    {
      "epoch": 0.30935315458808954,
      "grad_norm": 0.7643407814102523,
      "learning_rate": 0.00019195792564887643,
      "loss": 12.7625,
      "step": 5681
    },
    {
      "epoch": 0.3094076085846726,
      "grad_norm": 0.6965512867985716,
      "learning_rate": 0.00019195446055752187,
      "loss": 12.6885,
      "step": 5682
    },
    {
      "epoch": 0.3094620625812556,
      "grad_norm": 0.7183214873245715,
      "learning_rate": 0.000191950994751114,
      "loss": 12.671,
      "step": 5683
    },
    {
      "epoch": 0.3095165165778386,
      "grad_norm": 0.7341201497898837,
      "learning_rate": 0.0001919475282296798,
      "loss": 12.5902,
      "step": 5684
    },
    {
      "epoch": 0.3095709705744216,
      "grad_norm": 0.7231880725512029,
      "learning_rate": 0.00019194406099324614,
      "loss": 12.7504,
      "step": 5685
    },
    {
      "epoch": 0.3096254245710046,
      "grad_norm": 0.8481118115905432,
      "learning_rate": 0.0001919405930418401,
      "loss": 12.5931,
      "step": 5686
    },
    {
      "epoch": 0.3096798785675876,
      "grad_norm": 0.60463448001354,
      "learning_rate": 0.00019193712437548858,
      "loss": 12.5207,
      "step": 5687
    },
    {
      "epoch": 0.30973433256417066,
      "grad_norm": 0.8003196827589169,
      "learning_rate": 0.00019193365499421857,
      "loss": 12.5607,
      "step": 5688
    },
    {
      "epoch": 0.30978878656075365,
      "grad_norm": 0.6569732989360393,
      "learning_rate": 0.000191930184898057,
      "loss": 12.5348,
      "step": 5689
    },
    {
      "epoch": 0.30984324055733664,
      "grad_norm": 0.6507547624096993,
      "learning_rate": 0.00019192671408703094,
      "loss": 12.5218,
      "step": 5690
    },
    {
      "epoch": 0.3098976945539197,
      "grad_norm": 0.8016253940452907,
      "learning_rate": 0.00019192324256116732,
      "loss": 12.6849,
      "step": 5691
    },
    {
      "epoch": 0.3099521485505027,
      "grad_norm": 0.6636968700459538,
      "learning_rate": 0.00019191977032049313,
      "loss": 12.4833,
      "step": 5692
    },
    {
      "epoch": 0.3100066025470857,
      "grad_norm": 0.7364071722048001,
      "learning_rate": 0.00019191629736503544,
      "loss": 12.5185,
      "step": 5693
    },
    {
      "epoch": 0.3100610565436687,
      "grad_norm": 0.7051502639556526,
      "learning_rate": 0.00019191282369482115,
      "loss": 12.6904,
      "step": 5694
    },
    {
      "epoch": 0.3101155105402517,
      "grad_norm": 0.6651267585232153,
      "learning_rate": 0.00019190934930987736,
      "loss": 12.6454,
      "step": 5695
    },
    {
      "epoch": 0.3101699645368347,
      "grad_norm": 0.6720576053944531,
      "learning_rate": 0.00019190587421023106,
      "loss": 12.6355,
      "step": 5696
    },
    {
      "epoch": 0.31022441853341776,
      "grad_norm": 0.7403081220658613,
      "learning_rate": 0.00019190239839590926,
      "loss": 12.6477,
      "step": 5697
    },
    {
      "epoch": 0.31027887253000075,
      "grad_norm": 0.6856887471881359,
      "learning_rate": 0.000191898921866939,
      "loss": 12.5374,
      "step": 5698
    },
    {
      "epoch": 0.31033332652658374,
      "grad_norm": 0.7849792672826814,
      "learning_rate": 0.00019189544462334731,
      "loss": 12.5885,
      "step": 5699
    },
    {
      "epoch": 0.3103877805231668,
      "grad_norm": 0.6940421079361554,
      "learning_rate": 0.00019189196666516124,
      "loss": 12.5049,
      "step": 5700
    },
    {
      "epoch": 0.3104422345197498,
      "grad_norm": 0.7815347783955346,
      "learning_rate": 0.00019188848799240782,
      "loss": 12.6357,
      "step": 5701
    },
    {
      "epoch": 0.3104966885163328,
      "grad_norm": 0.7871432775569294,
      "learning_rate": 0.00019188500860511408,
      "loss": 12.5207,
      "step": 5702
    },
    {
      "epoch": 0.3105511425129158,
      "grad_norm": 0.729806048478542,
      "learning_rate": 0.00019188152850330717,
      "loss": 12.6383,
      "step": 5703
    },
    {
      "epoch": 0.3106055965094988,
      "grad_norm": 0.6487578342015209,
      "learning_rate": 0.00019187804768701404,
      "loss": 12.4904,
      "step": 5704
    },
    {
      "epoch": 0.3106600505060818,
      "grad_norm": 0.6137037639031049,
      "learning_rate": 0.0001918745661562618,
      "loss": 12.3875,
      "step": 5705
    },
    {
      "epoch": 0.31071450450266486,
      "grad_norm": 0.6880702484472031,
      "learning_rate": 0.00019187108391107756,
      "loss": 12.5841,
      "step": 5706
    },
    {
      "epoch": 0.31076895849924785,
      "grad_norm": 0.7192930692423495,
      "learning_rate": 0.00019186760095148833,
      "loss": 12.5358,
      "step": 5707
    },
    {
      "epoch": 0.31082341249583084,
      "grad_norm": 0.6777797822609798,
      "learning_rate": 0.00019186411727752125,
      "loss": 12.6532,
      "step": 5708
    },
    {
      "epoch": 0.3108778664924139,
      "grad_norm": 0.6988579087077,
      "learning_rate": 0.00019186063288920336,
      "loss": 12.4749,
      "step": 5709
    },
    {
      "epoch": 0.3109323204889969,
      "grad_norm": 0.8106200711091491,
      "learning_rate": 0.00019185714778656183,
      "loss": 12.7109,
      "step": 5710
    },
    {
      "epoch": 0.3109867744855799,
      "grad_norm": 0.7307713134379897,
      "learning_rate": 0.00019185366196962367,
      "loss": 12.5756,
      "step": 5711
    },
    {
      "epoch": 0.3110412284821629,
      "grad_norm": 0.7003160909230988,
      "learning_rate": 0.00019185017543841605,
      "loss": 12.5781,
      "step": 5712
    },
    {
      "epoch": 0.3110956824787459,
      "grad_norm": 0.7035305537630274,
      "learning_rate": 0.00019184668819296604,
      "loss": 12.592,
      "step": 5713
    },
    {
      "epoch": 0.3111501364753289,
      "grad_norm": 0.6408295726992614,
      "learning_rate": 0.00019184320023330083,
      "loss": 12.4214,
      "step": 5714
    },
    {
      "epoch": 0.31120459047191196,
      "grad_norm": 0.6757175305885018,
      "learning_rate": 0.00019183971155944748,
      "loss": 12.5193,
      "step": 5715
    },
    {
      "epoch": 0.31125904446849495,
      "grad_norm": 0.6714370645069598,
      "learning_rate": 0.0001918362221714331,
      "loss": 12.5045,
      "step": 5716
    },
    {
      "epoch": 0.311313498465078,
      "grad_norm": 0.763641142450625,
      "learning_rate": 0.00019183273206928487,
      "loss": 12.466,
      "step": 5717
    },
    {
      "epoch": 0.311367952461661,
      "grad_norm": 0.6373620796361511,
      "learning_rate": 0.0001918292412530299,
      "loss": 12.4351,
      "step": 5718
    },
    {
      "epoch": 0.311422406458244,
      "grad_norm": 0.773134594117127,
      "learning_rate": 0.00019182574972269537,
      "loss": 12.7381,
      "step": 5719
    },
    {
      "epoch": 0.31147686045482703,
      "grad_norm": 0.6326104987283147,
      "learning_rate": 0.0001918222574783084,
      "loss": 12.5616,
      "step": 5720
    },
    {
      "epoch": 0.31153131445141,
      "grad_norm": 0.7155966148875553,
      "learning_rate": 0.0001918187645198962,
      "loss": 12.6321,
      "step": 5721
    },
    {
      "epoch": 0.311585768447993,
      "grad_norm": 0.7879595861603473,
      "learning_rate": 0.00019181527084748582,
      "loss": 12.5777,
      "step": 5722
    },
    {
      "epoch": 0.31164022244457606,
      "grad_norm": 0.6353999496292784,
      "learning_rate": 0.00019181177646110454,
      "loss": 12.6064,
      "step": 5723
    },
    {
      "epoch": 0.31169467644115906,
      "grad_norm": 0.7480236445199707,
      "learning_rate": 0.00019180828136077947,
      "loss": 12.5245,
      "step": 5724
    },
    {
      "epoch": 0.31174913043774205,
      "grad_norm": 0.7011891796684117,
      "learning_rate": 0.00019180478554653782,
      "loss": 12.6022,
      "step": 5725
    },
    {
      "epoch": 0.3118035844343251,
      "grad_norm": 0.7226576397297431,
      "learning_rate": 0.00019180128901840677,
      "loss": 12.461,
      "step": 5726
    },
    {
      "epoch": 0.3118580384309081,
      "grad_norm": 0.8105551964611454,
      "learning_rate": 0.0001917977917764135,
      "loss": 12.5971,
      "step": 5727
    },
    {
      "epoch": 0.3119124924274911,
      "grad_norm": 0.7105623987225386,
      "learning_rate": 0.00019179429382058517,
      "loss": 12.5568,
      "step": 5728
    },
    {
      "epoch": 0.31196694642407413,
      "grad_norm": 0.7423286623670033,
      "learning_rate": 0.00019179079515094908,
      "loss": 12.5355,
      "step": 5729
    },
    {
      "epoch": 0.3120214004206571,
      "grad_norm": 0.6921990475242991,
      "learning_rate": 0.0001917872957675323,
      "loss": 12.5362,
      "step": 5730
    },
    {
      "epoch": 0.3120758544172401,
      "grad_norm": 0.6326208002108714,
      "learning_rate": 0.00019178379567036217,
      "loss": 12.4887,
      "step": 5731
    },
    {
      "epoch": 0.31213030841382317,
      "grad_norm": 0.8282457364681612,
      "learning_rate": 0.00019178029485946585,
      "loss": 12.6543,
      "step": 5732
    },
    {
      "epoch": 0.31218476241040616,
      "grad_norm": 0.6377328782854843,
      "learning_rate": 0.00019177679333487056,
      "loss": 12.5946,
      "step": 5733
    },
    {
      "epoch": 0.31223921640698915,
      "grad_norm": 0.6920475291113465,
      "learning_rate": 0.0001917732910966035,
      "loss": 12.5284,
      "step": 5734
    },
    {
      "epoch": 0.3122936704035722,
      "grad_norm": 0.7101041584136139,
      "learning_rate": 0.00019176978814469198,
      "loss": 12.5032,
      "step": 5735
    },
    {
      "epoch": 0.3123481244001552,
      "grad_norm": 0.6111176119060407,
      "learning_rate": 0.0001917662844791632,
      "loss": 12.5313,
      "step": 5736
    },
    {
      "epoch": 0.3124025783967382,
      "grad_norm": 0.6681293165538871,
      "learning_rate": 0.00019176278010004435,
      "loss": 12.5214,
      "step": 5737
    },
    {
      "epoch": 0.31245703239332123,
      "grad_norm": 0.7689966914982728,
      "learning_rate": 0.00019175927500736278,
      "loss": 12.5465,
      "step": 5738
    },
    {
      "epoch": 0.3125114863899042,
      "grad_norm": 0.6566846990967126,
      "learning_rate": 0.00019175576920114567,
      "loss": 12.4811,
      "step": 5739
    },
    {
      "epoch": 0.3125659403864872,
      "grad_norm": 0.8515807575793801,
      "learning_rate": 0.00019175226268142032,
      "loss": 12.7253,
      "step": 5740
    },
    {
      "epoch": 0.31262039438307027,
      "grad_norm": 0.8194373785790895,
      "learning_rate": 0.00019174875544821402,
      "loss": 12.5978,
      "step": 5741
    },
    {
      "epoch": 0.31267484837965326,
      "grad_norm": 0.5959364512025092,
      "learning_rate": 0.00019174524750155398,
      "loss": 12.4635,
      "step": 5742
    },
    {
      "epoch": 0.31272930237623625,
      "grad_norm": 0.7182012662497026,
      "learning_rate": 0.00019174173884146752,
      "loss": 12.6004,
      "step": 5743
    },
    {
      "epoch": 0.3127837563728193,
      "grad_norm": 0.6191882074588925,
      "learning_rate": 0.00019173822946798191,
      "loss": 12.5388,
      "step": 5744
    },
    {
      "epoch": 0.3128382103694023,
      "grad_norm": 0.7284139974989752,
      "learning_rate": 0.00019173471938112443,
      "loss": 12.5659,
      "step": 5745
    },
    {
      "epoch": 0.3128926643659853,
      "grad_norm": 0.7308799859614893,
      "learning_rate": 0.00019173120858092242,
      "loss": 12.5344,
      "step": 5746
    },
    {
      "epoch": 0.31294711836256833,
      "grad_norm": 0.7161742868249728,
      "learning_rate": 0.0001917276970674031,
      "loss": 12.6181,
      "step": 5747
    },
    {
      "epoch": 0.3130015723591513,
      "grad_norm": 0.6887630271894791,
      "learning_rate": 0.0001917241848405939,
      "loss": 12.5654,
      "step": 5748
    },
    {
      "epoch": 0.3130560263557344,
      "grad_norm": 0.7334236423854301,
      "learning_rate": 0.000191720671900522,
      "loss": 12.6554,
      "step": 5749
    },
    {
      "epoch": 0.31311048035231737,
      "grad_norm": 0.7280109583218388,
      "learning_rate": 0.00019171715824721478,
      "loss": 12.6293,
      "step": 5750
    },
    {
      "epoch": 0.31316493434890036,
      "grad_norm": 0.7742369837794514,
      "learning_rate": 0.00019171364388069958,
      "loss": 12.537,
      "step": 5751
    },
    {
      "epoch": 0.3132193883454834,
      "grad_norm": 0.6332978382946135,
      "learning_rate": 0.0001917101288010037,
      "loss": 12.5003,
      "step": 5752
    },
    {
      "epoch": 0.3132738423420664,
      "grad_norm": 0.8109885888601659,
      "learning_rate": 0.00019170661300815445,
      "loss": 12.5463,
      "step": 5753
    },
    {
      "epoch": 0.3133282963386494,
      "grad_norm": 0.6776696760063646,
      "learning_rate": 0.0001917030965021792,
      "loss": 12.4903,
      "step": 5754
    },
    {
      "epoch": 0.31338275033523244,
      "grad_norm": 0.7367968361944557,
      "learning_rate": 0.00019169957928310533,
      "loss": 12.5046,
      "step": 5755
    },
    {
      "epoch": 0.31343720433181543,
      "grad_norm": 0.6710299186422611,
      "learning_rate": 0.0001916960613509601,
      "loss": 12.5471,
      "step": 5756
    },
    {
      "epoch": 0.3134916583283984,
      "grad_norm": 0.8285856812284736,
      "learning_rate": 0.00019169254270577098,
      "loss": 12.6954,
      "step": 5757
    },
    {
      "epoch": 0.3135461123249815,
      "grad_norm": 0.6826103991424295,
      "learning_rate": 0.00019168902334756524,
      "loss": 12.3569,
      "step": 5758
    },
    {
      "epoch": 0.31360056632156447,
      "grad_norm": 0.7371646823095679,
      "learning_rate": 0.0001916855032763703,
      "loss": 12.5832,
      "step": 5759
    },
    {
      "epoch": 0.31365502031814746,
      "grad_norm": 0.6338402069507585,
      "learning_rate": 0.00019168198249221348,
      "loss": 12.5793,
      "step": 5760
    },
    {
      "epoch": 0.3137094743147305,
      "grad_norm": 0.6892790788000476,
      "learning_rate": 0.00019167846099512218,
      "loss": 12.5516,
      "step": 5761
    },
    {
      "epoch": 0.3137639283113135,
      "grad_norm": 0.685422960143123,
      "learning_rate": 0.00019167493878512382,
      "loss": 12.6479,
      "step": 5762
    },
    {
      "epoch": 0.3138183823078965,
      "grad_norm": 0.6800269200015937,
      "learning_rate": 0.00019167141586224576,
      "loss": 12.5571,
      "step": 5763
    },
    {
      "epoch": 0.31387283630447954,
      "grad_norm": 0.8151275310663422,
      "learning_rate": 0.00019166789222651537,
      "loss": 12.7188,
      "step": 5764
    },
    {
      "epoch": 0.31392729030106253,
      "grad_norm": 0.69512271148721,
      "learning_rate": 0.0001916643678779601,
      "loss": 12.6066,
      "step": 5765
    },
    {
      "epoch": 0.3139817442976455,
      "grad_norm": 0.8421123782899708,
      "learning_rate": 0.00019166084281660735,
      "loss": 12.5477,
      "step": 5766
    },
    {
      "epoch": 0.3140361982942286,
      "grad_norm": 0.7250909770675339,
      "learning_rate": 0.0001916573170424845,
      "loss": 12.6669,
      "step": 5767
    },
    {
      "epoch": 0.31409065229081157,
      "grad_norm": 0.7765303098532939,
      "learning_rate": 0.00019165379055561895,
      "loss": 12.5521,
      "step": 5768
    },
    {
      "epoch": 0.31414510628739456,
      "grad_norm": 0.757240010661953,
      "learning_rate": 0.0001916502633560382,
      "loss": 12.515,
      "step": 5769
    },
    {
      "epoch": 0.3141995602839776,
      "grad_norm": 0.7924596813156907,
      "learning_rate": 0.00019164673544376962,
      "loss": 12.6334,
      "step": 5770
    },
    {
      "epoch": 0.3142540142805606,
      "grad_norm": 0.6340821796162117,
      "learning_rate": 0.00019164320681884064,
      "loss": 12.5287,
      "step": 5771
    },
    {
      "epoch": 0.3143084682771436,
      "grad_norm": 0.6938443956294725,
      "learning_rate": 0.00019163967748127874,
      "loss": 12.591,
      "step": 5772
    },
    {
      "epoch": 0.31436292227372664,
      "grad_norm": 0.8052442207081102,
      "learning_rate": 0.00019163614743111134,
      "loss": 12.6676,
      "step": 5773
    },
    {
      "epoch": 0.31441737627030963,
      "grad_norm": 0.6897969950775128,
      "learning_rate": 0.00019163261666836588,
      "loss": 12.5717,
      "step": 5774
    },
    {
      "epoch": 0.3144718302668926,
      "grad_norm": 0.7234430598331005,
      "learning_rate": 0.00019162908519306982,
      "loss": 12.7961,
      "step": 5775
    },
    {
      "epoch": 0.3145262842634757,
      "grad_norm": 0.7182030630629905,
      "learning_rate": 0.00019162555300525062,
      "loss": 12.6843,
      "step": 5776
    },
    {
      "epoch": 0.31458073826005867,
      "grad_norm": 0.7212326014240544,
      "learning_rate": 0.00019162202010493577,
      "loss": 12.5077,
      "step": 5777
    },
    {
      "epoch": 0.31463519225664166,
      "grad_norm": 0.7180839647466242,
      "learning_rate": 0.00019161848649215272,
      "loss": 12.6407,
      "step": 5778
    },
    {
      "epoch": 0.3146896462532247,
      "grad_norm": 0.6995245318344423,
      "learning_rate": 0.00019161495216692896,
      "loss": 12.594,
      "step": 5779
    },
    {
      "epoch": 0.3147441002498077,
      "grad_norm": 0.7030108004030603,
      "learning_rate": 0.00019161141712929197,
      "loss": 12.487,
      "step": 5780
    },
    {
      "epoch": 0.3147985542463907,
      "grad_norm": 0.7292919566984026,
      "learning_rate": 0.0001916078813792692,
      "loss": 12.585,
      "step": 5781
    },
    {
      "epoch": 0.31485300824297374,
      "grad_norm": 0.759506562847832,
      "learning_rate": 0.00019160434491688824,
      "loss": 12.4386,
      "step": 5782
    },
    {
      "epoch": 0.31490746223955673,
      "grad_norm": 0.9639081006195463,
      "learning_rate": 0.00019160080774217647,
      "loss": 12.6828,
      "step": 5783
    },
    {
      "epoch": 0.3149619162361398,
      "grad_norm": 0.7434314130278653,
      "learning_rate": 0.00019159726985516152,
      "loss": 12.4645,
      "step": 5784
    },
    {
      "epoch": 0.3150163702327228,
      "grad_norm": 0.6400421001900726,
      "learning_rate": 0.00019159373125587082,
      "loss": 12.3502,
      "step": 5785
    },
    {
      "epoch": 0.31507082422930577,
      "grad_norm": 0.6867545808438131,
      "learning_rate": 0.00019159019194433188,
      "loss": 12.5432,
      "step": 5786
    },
    {
      "epoch": 0.3151252782258888,
      "grad_norm": 0.7896890316318025,
      "learning_rate": 0.00019158665192057229,
      "loss": 12.5461,
      "step": 5787
    },
    {
      "epoch": 0.3151797322224718,
      "grad_norm": 0.6951193183704759,
      "learning_rate": 0.00019158311118461948,
      "loss": 12.6574,
      "step": 5788
    },
    {
      "epoch": 0.3152341862190548,
      "grad_norm": 0.7847001380500844,
      "learning_rate": 0.00019157956973650108,
      "loss": 12.6444,
      "step": 5789
    },
    {
      "epoch": 0.31528864021563785,
      "grad_norm": 0.7231280042579047,
      "learning_rate": 0.0001915760275762446,
      "loss": 12.5887,
      "step": 5790
    },
    {
      "epoch": 0.31534309421222084,
      "grad_norm": 0.710818117137646,
      "learning_rate": 0.00019157248470387753,
      "loss": 12.3551,
      "step": 5791
    },
    {
      "epoch": 0.31539754820880384,
      "grad_norm": 0.8236673096850398,
      "learning_rate": 0.00019156894111942746,
      "loss": 12.6124,
      "step": 5792
    },
    {
      "epoch": 0.3154520022053869,
      "grad_norm": 0.6464910311423513,
      "learning_rate": 0.00019156539682292197,
      "loss": 12.5541,
      "step": 5793
    },
    {
      "epoch": 0.3155064562019699,
      "grad_norm": 0.8011589161346442,
      "learning_rate": 0.00019156185181438861,
      "loss": 12.5346,
      "step": 5794
    },
    {
      "epoch": 0.31556091019855287,
      "grad_norm": 0.7003550705119012,
      "learning_rate": 0.00019155830609385492,
      "loss": 12.5535,
      "step": 5795
    },
    {
      "epoch": 0.3156153641951359,
      "grad_norm": 0.6682420425845063,
      "learning_rate": 0.0001915547596613485,
      "loss": 12.551,
      "step": 5796
    },
    {
      "epoch": 0.3156698181917189,
      "grad_norm": 0.6748930099795875,
      "learning_rate": 0.00019155121251689689,
      "loss": 12.5044,
      "step": 5797
    },
    {
      "epoch": 0.3157242721883019,
      "grad_norm": 0.650553657834858,
      "learning_rate": 0.00019154766466052773,
      "loss": 12.5407,
      "step": 5798
    },
    {
      "epoch": 0.31577872618488495,
      "grad_norm": 0.702942885517404,
      "learning_rate": 0.00019154411609226854,
      "loss": 12.4614,
      "step": 5799
    },
    {
      "epoch": 0.31583318018146794,
      "grad_norm": 0.6481075120474772,
      "learning_rate": 0.000191540566812147,
      "loss": 12.535,
      "step": 5800
    },
    {
      "epoch": 0.31588763417805094,
      "grad_norm": 1.0794965868639268,
      "learning_rate": 0.00019153701682019062,
      "loss": 12.4537,
      "step": 5801
    },
    {
      "epoch": 0.315942088174634,
      "grad_norm": 0.7350571771371223,
      "learning_rate": 0.00019153346611642706,
      "loss": 12.5537,
      "step": 5802
    },
    {
      "epoch": 0.315996542171217,
      "grad_norm": 0.6826378435926155,
      "learning_rate": 0.0001915299147008839,
      "loss": 12.632,
      "step": 5803
    },
    {
      "epoch": 0.31605099616779997,
      "grad_norm": 0.731159866122895,
      "learning_rate": 0.0001915263625735888,
      "loss": 12.564,
      "step": 5804
    },
    {
      "epoch": 0.316105450164383,
      "grad_norm": 0.7153238353345873,
      "learning_rate": 0.00019152280973456934,
      "loss": 12.6477,
      "step": 5805
    },
    {
      "epoch": 0.316159904160966,
      "grad_norm": 0.7530356054837524,
      "learning_rate": 0.0001915192561838532,
      "loss": 12.6045,
      "step": 5806
    },
    {
      "epoch": 0.316214358157549,
      "grad_norm": 0.7768597439292041,
      "learning_rate": 0.00019151570192146793,
      "loss": 12.6109,
      "step": 5807
    },
    {
      "epoch": 0.31626881215413205,
      "grad_norm": 0.7119692150042135,
      "learning_rate": 0.00019151214694744124,
      "loss": 12.5061,
      "step": 5808
    },
    {
      "epoch": 0.31632326615071504,
      "grad_norm": 0.6405228772522945,
      "learning_rate": 0.00019150859126180073,
      "loss": 12.6042,
      "step": 5809
    },
    {
      "epoch": 0.31637772014729804,
      "grad_norm": 0.7223526465464335,
      "learning_rate": 0.00019150503486457408,
      "loss": 12.5468,
      "step": 5810
    },
    {
      "epoch": 0.3164321741438811,
      "grad_norm": 0.783686761881016,
      "learning_rate": 0.00019150147775578893,
      "loss": 12.5976,
      "step": 5811
    },
    {
      "epoch": 0.3164866281404641,
      "grad_norm": 0.6661836937905445,
      "learning_rate": 0.00019149791993547296,
      "loss": 12.4926,
      "step": 5812
    },
    {
      "epoch": 0.31654108213704707,
      "grad_norm": 0.7332835327296134,
      "learning_rate": 0.00019149436140365378,
      "loss": 12.6416,
      "step": 5813
    },
    {
      "epoch": 0.3165955361336301,
      "grad_norm": 0.689079313091683,
      "learning_rate": 0.00019149080216035916,
      "loss": 12.521,
      "step": 5814
    },
    {
      "epoch": 0.3166499901302131,
      "grad_norm": 0.7952304139427612,
      "learning_rate": 0.00019148724220561665,
      "loss": 12.6607,
      "step": 5815
    },
    {
      "epoch": 0.31670444412679616,
      "grad_norm": 0.6523041678617995,
      "learning_rate": 0.00019148368153945407,
      "loss": 12.491,
      "step": 5816
    },
    {
      "epoch": 0.31675889812337915,
      "grad_norm": 0.6227683054782411,
      "learning_rate": 0.000191480120161899,
      "loss": 12.5964,
      "step": 5817
    },
    {
      "epoch": 0.31681335211996214,
      "grad_norm": 0.6687279981265427,
      "learning_rate": 0.00019147655807297918,
      "loss": 12.6218,
      "step": 5818
    },
    {
      "epoch": 0.3168678061165452,
      "grad_norm": 0.6862606107733062,
      "learning_rate": 0.0001914729952727223,
      "loss": 12.5118,
      "step": 5819
    },
    {
      "epoch": 0.3169222601131282,
      "grad_norm": 0.7142573441962577,
      "learning_rate": 0.0001914694317611561,
      "loss": 12.5311,
      "step": 5820
    },
    {
      "epoch": 0.3169767141097112,
      "grad_norm": 0.6552484972372591,
      "learning_rate": 0.0001914658675383082,
      "loss": 12.5544,
      "step": 5821
    },
    {
      "epoch": 0.3170311681062942,
      "grad_norm": 0.9383767153400039,
      "learning_rate": 0.00019146230260420644,
      "loss": 12.6369,
      "step": 5822
    },
    {
      "epoch": 0.3170856221028772,
      "grad_norm": 0.7370541874647787,
      "learning_rate": 0.00019145873695887843,
      "loss": 12.639,
      "step": 5823
    },
    {
      "epoch": 0.3171400760994602,
      "grad_norm": 0.7141483693702555,
      "learning_rate": 0.00019145517060235195,
      "loss": 12.6066,
      "step": 5824
    },
    {
      "epoch": 0.31719453009604326,
      "grad_norm": 0.7333381815447372,
      "learning_rate": 0.00019145160353465474,
      "loss": 12.5564,
      "step": 5825
    },
    {
      "epoch": 0.31724898409262625,
      "grad_norm": 0.6183938441937711,
      "learning_rate": 0.00019144803575581453,
      "loss": 12.5607,
      "step": 5826
    },
    {
      "epoch": 0.31730343808920924,
      "grad_norm": 0.6892142292238302,
      "learning_rate": 0.00019144446726585904,
      "loss": 12.5779,
      "step": 5827
    },
    {
      "epoch": 0.3173578920857923,
      "grad_norm": 0.6630403431973069,
      "learning_rate": 0.00019144089806481606,
      "loss": 12.5383,
      "step": 5828
    },
    {
      "epoch": 0.3174123460823753,
      "grad_norm": 0.6879898272065529,
      "learning_rate": 0.0001914373281527133,
      "loss": 12.6813,
      "step": 5829
    },
    {
      "epoch": 0.3174668000789583,
      "grad_norm": 0.9279147458061492,
      "learning_rate": 0.00019143375752957856,
      "loss": 12.5834,
      "step": 5830
    },
    {
      "epoch": 0.3175212540755413,
      "grad_norm": 0.8261607137141636,
      "learning_rate": 0.0001914301861954396,
      "loss": 12.6757,
      "step": 5831
    },
    {
      "epoch": 0.3175757080721243,
      "grad_norm": 0.7070172220832147,
      "learning_rate": 0.00019142661415032415,
      "loss": 12.5782,
      "step": 5832
    },
    {
      "epoch": 0.3176301620687073,
      "grad_norm": 0.7416069134775134,
      "learning_rate": 0.00019142304139426,
      "loss": 12.4508,
      "step": 5833
    },
    {
      "epoch": 0.31768461606529036,
      "grad_norm": 0.6759809445170305,
      "learning_rate": 0.000191419467927275,
      "loss": 12.4728,
      "step": 5834
    },
    {
      "epoch": 0.31773907006187335,
      "grad_norm": 0.7479228483812418,
      "learning_rate": 0.00019141589374939685,
      "loss": 12.5918,
      "step": 5835
    },
    {
      "epoch": 0.31779352405845634,
      "grad_norm": 0.7018073897109439,
      "learning_rate": 0.0001914123188606534,
      "loss": 12.6481,
      "step": 5836
    },
    {
      "epoch": 0.3178479780550394,
      "grad_norm": 0.6509538967378394,
      "learning_rate": 0.0001914087432610724,
      "loss": 12.6216,
      "step": 5837
    },
    {
      "epoch": 0.3179024320516224,
      "grad_norm": 0.6956653775989627,
      "learning_rate": 0.0001914051669506817,
      "loss": 12.4914,
      "step": 5838
    },
    {
      "epoch": 0.3179568860482054,
      "grad_norm": 0.6488704800793494,
      "learning_rate": 0.0001914015899295091,
      "loss": 12.4812,
      "step": 5839
    },
    {
      "epoch": 0.3180113400447884,
      "grad_norm": 0.6687389598321076,
      "learning_rate": 0.00019139801219758242,
      "loss": 12.5834,
      "step": 5840
    },
    {
      "epoch": 0.3180657940413714,
      "grad_norm": 0.67578294702017,
      "learning_rate": 0.00019139443375492944,
      "loss": 12.5554,
      "step": 5841
    },
    {
      "epoch": 0.3181202480379544,
      "grad_norm": 0.6408985153317925,
      "learning_rate": 0.00019139085460157803,
      "loss": 12.5064,
      "step": 5842
    },
    {
      "epoch": 0.31817470203453746,
      "grad_norm": 0.6129408061020057,
      "learning_rate": 0.00019138727473755603,
      "loss": 12.5581,
      "step": 5843
    },
    {
      "epoch": 0.31822915603112045,
      "grad_norm": 0.7117051176119991,
      "learning_rate": 0.00019138369416289122,
      "loss": 12.6514,
      "step": 5844
    },
    {
      "epoch": 0.31828361002770345,
      "grad_norm": 0.6685305734872339,
      "learning_rate": 0.0001913801128776115,
      "loss": 12.6237,
      "step": 5845
    },
    {
      "epoch": 0.3183380640242865,
      "grad_norm": 0.7935940982999596,
      "learning_rate": 0.0001913765308817447,
      "loss": 12.6416,
      "step": 5846
    },
    {
      "epoch": 0.3183925180208695,
      "grad_norm": 0.6777446131367032,
      "learning_rate": 0.00019137294817531863,
      "loss": 12.4684,
      "step": 5847
    },
    {
      "epoch": 0.3184469720174525,
      "grad_norm": 0.6927049733411546,
      "learning_rate": 0.00019136936475836126,
      "loss": 12.4775,
      "step": 5848
    },
    {
      "epoch": 0.3185014260140355,
      "grad_norm": 0.6803616170636894,
      "learning_rate": 0.00019136578063090034,
      "loss": 12.4689,
      "step": 5849
    },
    {
      "epoch": 0.3185558800106185,
      "grad_norm": 0.6612387529143992,
      "learning_rate": 0.0001913621957929638,
      "loss": 12.61,
      "step": 5850
    },
    {
      "epoch": 0.31861033400720157,
      "grad_norm": 0.6785997938224911,
      "learning_rate": 0.0001913586102445795,
      "loss": 12.6917,
      "step": 5851
    },
    {
      "epoch": 0.31866478800378456,
      "grad_norm": 0.6857052369825005,
      "learning_rate": 0.00019135502398577532,
      "loss": 12.6363,
      "step": 5852
    },
    {
      "epoch": 0.31871924200036755,
      "grad_norm": 0.6627549169132189,
      "learning_rate": 0.00019135143701657915,
      "loss": 12.5624,
      "step": 5853
    },
    {
      "epoch": 0.3187736959969506,
      "grad_norm": 0.6077596656319667,
      "learning_rate": 0.00019134784933701892,
      "loss": 12.4125,
      "step": 5854
    },
    {
      "epoch": 0.3188281499935336,
      "grad_norm": 0.7000816467705235,
      "learning_rate": 0.00019134426094712245,
      "loss": 12.636,
      "step": 5855
    },
    {
      "epoch": 0.3188826039901166,
      "grad_norm": 0.6772191962377853,
      "learning_rate": 0.00019134067184691772,
      "loss": 12.6985,
      "step": 5856
    },
    {
      "epoch": 0.31893705798669963,
      "grad_norm": 0.7770632506722829,
      "learning_rate": 0.0001913370820364326,
      "loss": 12.5842,
      "step": 5857
    },
    {
      "epoch": 0.3189915119832826,
      "grad_norm": 0.7579582324057819,
      "learning_rate": 0.00019133349151569503,
      "loss": 12.4904,
      "step": 5858
    },
    {
      "epoch": 0.3190459659798656,
      "grad_norm": 0.7548108905123265,
      "learning_rate": 0.0001913299002847329,
      "loss": 12.5497,
      "step": 5859
    },
    {
      "epoch": 0.31910041997644867,
      "grad_norm": 0.6842505571032034,
      "learning_rate": 0.00019132630834357413,
      "loss": 12.5264,
      "step": 5860
    },
    {
      "epoch": 0.31915487397303166,
      "grad_norm": 0.7389234910801799,
      "learning_rate": 0.0001913227156922467,
      "loss": 12.5738,
      "step": 5861
    },
    {
      "epoch": 0.31920932796961465,
      "grad_norm": 0.6907600700325771,
      "learning_rate": 0.0001913191223307785,
      "loss": 12.5408,
      "step": 5862
    },
    {
      "epoch": 0.3192637819661977,
      "grad_norm": 0.6885413039328351,
      "learning_rate": 0.0001913155282591975,
      "loss": 12.4737,
      "step": 5863
    },
    {
      "epoch": 0.3193182359627807,
      "grad_norm": 0.7449988965819492,
      "learning_rate": 0.00019131193347753163,
      "loss": 12.6084,
      "step": 5864
    },
    {
      "epoch": 0.3193726899593637,
      "grad_norm": 0.8251864478061979,
      "learning_rate": 0.00019130833798580886,
      "loss": 12.5428,
      "step": 5865
    },
    {
      "epoch": 0.31942714395594674,
      "grad_norm": 0.6841118849731743,
      "learning_rate": 0.00019130474178405714,
      "loss": 12.641,
      "step": 5866
    },
    {
      "epoch": 0.31948159795252973,
      "grad_norm": 0.7122261775745119,
      "learning_rate": 0.00019130114487230442,
      "loss": 12.5943,
      "step": 5867
    },
    {
      "epoch": 0.3195360519491127,
      "grad_norm": 0.6901954748558911,
      "learning_rate": 0.0001912975472505787,
      "loss": 12.5068,
      "step": 5868
    },
    {
      "epoch": 0.31959050594569577,
      "grad_norm": 0.7314037157651961,
      "learning_rate": 0.00019129394891890793,
      "loss": 12.4722,
      "step": 5869
    },
    {
      "epoch": 0.31964495994227876,
      "grad_norm": 0.7117573572189002,
      "learning_rate": 0.00019129034987732012,
      "loss": 12.578,
      "step": 5870
    },
    {
      "epoch": 0.31969941393886175,
      "grad_norm": 0.7010924063861873,
      "learning_rate": 0.00019128675012584326,
      "loss": 12.6654,
      "step": 5871
    },
    {
      "epoch": 0.3197538679354448,
      "grad_norm": 0.6064541420225625,
      "learning_rate": 0.00019128314966450528,
      "loss": 12.55,
      "step": 5872
    },
    {
      "epoch": 0.3198083219320278,
      "grad_norm": 0.6321466814864312,
      "learning_rate": 0.00019127954849333423,
      "loss": 12.5765,
      "step": 5873
    },
    {
      "epoch": 0.3198627759286108,
      "grad_norm": 0.6947632669120551,
      "learning_rate": 0.0001912759466123581,
      "loss": 12.514,
      "step": 5874
    },
    {
      "epoch": 0.31991722992519384,
      "grad_norm": 0.6828212967587132,
      "learning_rate": 0.0001912723440216049,
      "loss": 12.5452,
      "step": 5875
    },
    {
      "epoch": 0.31997168392177683,
      "grad_norm": 0.7090710584294178,
      "learning_rate": 0.00019126874072110267,
      "loss": 12.5601,
      "step": 5876
    },
    {
      "epoch": 0.3200261379183598,
      "grad_norm": 0.641093483322115,
      "learning_rate": 0.0001912651367108794,
      "loss": 12.5624,
      "step": 5877
    },
    {
      "epoch": 0.32008059191494287,
      "grad_norm": 0.6538327455468403,
      "learning_rate": 0.0001912615319909631,
      "loss": 12.5216,
      "step": 5878
    },
    {
      "epoch": 0.32013504591152586,
      "grad_norm": 0.6449607621299951,
      "learning_rate": 0.00019125792656138186,
      "loss": 12.5711,
      "step": 5879
    },
    {
      "epoch": 0.32018949990810885,
      "grad_norm": 0.7118519253223596,
      "learning_rate": 0.00019125432042216365,
      "loss": 12.5681,
      "step": 5880
    },
    {
      "epoch": 0.3202439539046919,
      "grad_norm": 0.6947313023045424,
      "learning_rate": 0.00019125071357333658,
      "loss": 12.5832,
      "step": 5881
    },
    {
      "epoch": 0.3202984079012749,
      "grad_norm": 0.719459911799742,
      "learning_rate": 0.0001912471060149286,
      "loss": 12.5015,
      "step": 5882
    },
    {
      "epoch": 0.32035286189785794,
      "grad_norm": 0.662669731033275,
      "learning_rate": 0.00019124349774696787,
      "loss": 12.5893,
      "step": 5883
    },
    {
      "epoch": 0.32040731589444094,
      "grad_norm": 0.7102087719102528,
      "learning_rate": 0.00019123988876948236,
      "loss": 12.537,
      "step": 5884
    },
    {
      "epoch": 0.32046176989102393,
      "grad_norm": 0.6084472187136899,
      "learning_rate": 0.0001912362790825002,
      "loss": 12.5018,
      "step": 5885
    },
    {
      "epoch": 0.320516223887607,
      "grad_norm": 0.6703985463238232,
      "learning_rate": 0.0001912326686860494,
      "loss": 12.5211,
      "step": 5886
    },
    {
      "epoch": 0.32057067788418997,
      "grad_norm": 0.6735533812325952,
      "learning_rate": 0.00019122905758015812,
      "loss": 12.5738,
      "step": 5887
    },
    {
      "epoch": 0.32062513188077296,
      "grad_norm": 0.8245981215897119,
      "learning_rate": 0.00019122544576485434,
      "loss": 12.6971,
      "step": 5888
    },
    {
      "epoch": 0.320679585877356,
      "grad_norm": 0.7387246587967972,
      "learning_rate": 0.00019122183324016625,
      "loss": 12.7761,
      "step": 5889
    },
    {
      "epoch": 0.320734039873939,
      "grad_norm": 0.6311846714662387,
      "learning_rate": 0.00019121822000612185,
      "loss": 12.6251,
      "step": 5890
    },
    {
      "epoch": 0.320788493870522,
      "grad_norm": 0.6762614014696763,
      "learning_rate": 0.0001912146060627493,
      "loss": 12.5361,
      "step": 5891
    },
    {
      "epoch": 0.32084294786710504,
      "grad_norm": 0.7312297128165434,
      "learning_rate": 0.00019121099141007663,
      "loss": 12.6024,
      "step": 5892
    },
    {
      "epoch": 0.32089740186368804,
      "grad_norm": 0.6691564713248157,
      "learning_rate": 0.00019120737604813205,
      "loss": 12.6048,
      "step": 5893
    },
    {
      "epoch": 0.32095185586027103,
      "grad_norm": 0.6445853731650664,
      "learning_rate": 0.00019120375997694358,
      "loss": 12.5692,
      "step": 5894
    },
    {
      "epoch": 0.3210063098568541,
      "grad_norm": 0.68421869894897,
      "learning_rate": 0.00019120014319653938,
      "loss": 12.583,
      "step": 5895
    },
    {
      "epoch": 0.32106076385343707,
      "grad_norm": 0.8049595047614536,
      "learning_rate": 0.0001911965257069476,
      "loss": 12.7101,
      "step": 5896
    },
    {
      "epoch": 0.32111521785002006,
      "grad_norm": 0.667047744432576,
      "learning_rate": 0.00019119290750819633,
      "loss": 12.5683,
      "step": 5897
    },
    {
      "epoch": 0.3211696718466031,
      "grad_norm": 0.623587825427952,
      "learning_rate": 0.00019118928860031368,
      "loss": 12.5035,
      "step": 5898
    },
    {
      "epoch": 0.3212241258431861,
      "grad_norm": 0.7162232836644334,
      "learning_rate": 0.00019118566898332787,
      "loss": 12.5323,
      "step": 5899
    },
    {
      "epoch": 0.3212785798397691,
      "grad_norm": 0.6843716242945724,
      "learning_rate": 0.000191182048657267,
      "loss": 12.5902,
      "step": 5900
    },
    {
      "epoch": 0.32133303383635214,
      "grad_norm": 0.6313048020288865,
      "learning_rate": 0.00019117842762215922,
      "loss": 12.5645,
      "step": 5901
    },
    {
      "epoch": 0.32138748783293514,
      "grad_norm": 0.6700544330362783,
      "learning_rate": 0.0001911748058780327,
      "loss": 12.5279,
      "step": 5902
    },
    {
      "epoch": 0.32144194182951813,
      "grad_norm": 0.6428991072690676,
      "learning_rate": 0.0001911711834249156,
      "loss": 12.4831,
      "step": 5903
    },
    {
      "epoch": 0.3214963958261012,
      "grad_norm": 0.7827750281315372,
      "learning_rate": 0.00019116756026283608,
      "loss": 12.7236,
      "step": 5904
    },
    {
      "epoch": 0.32155084982268417,
      "grad_norm": 0.655512340891206,
      "learning_rate": 0.00019116393639182232,
      "loss": 12.5124,
      "step": 5905
    },
    {
      "epoch": 0.32160530381926716,
      "grad_norm": 0.6778863662038578,
      "learning_rate": 0.00019116031181190253,
      "loss": 12.488,
      "step": 5906
    },
    {
      "epoch": 0.3216597578158502,
      "grad_norm": 0.7940130979158228,
      "learning_rate": 0.00019115668652310486,
      "loss": 12.6014,
      "step": 5907
    },
    {
      "epoch": 0.3217142118124332,
      "grad_norm": 0.7597082595800774,
      "learning_rate": 0.0001911530605254575,
      "loss": 12.6357,
      "step": 5908
    },
    {
      "epoch": 0.3217686658090162,
      "grad_norm": 0.6902457091288527,
      "learning_rate": 0.00019114943381898865,
      "loss": 12.6692,
      "step": 5909
    },
    {
      "epoch": 0.32182311980559924,
      "grad_norm": 0.7174150884814654,
      "learning_rate": 0.0001911458064037265,
      "loss": 12.5264,
      "step": 5910
    },
    {
      "epoch": 0.32187757380218224,
      "grad_norm": 0.6886629116964688,
      "learning_rate": 0.00019114217827969932,
      "loss": 12.5167,
      "step": 5911
    },
    {
      "epoch": 0.32193202779876523,
      "grad_norm": 0.717546620919752,
      "learning_rate": 0.00019113854944693523,
      "loss": 12.5693,
      "step": 5912
    },
    {
      "epoch": 0.3219864817953483,
      "grad_norm": 0.6188720374301961,
      "learning_rate": 0.00019113491990546252,
      "loss": 12.459,
      "step": 5913
    },
    {
      "epoch": 0.32204093579193127,
      "grad_norm": 0.6450251252181151,
      "learning_rate": 0.00019113128965530943,
      "loss": 12.458,
      "step": 5914
    },
    {
      "epoch": 0.32209538978851426,
      "grad_norm": 0.6846813998165503,
      "learning_rate": 0.00019112765869650405,
      "loss": 12.4775,
      "step": 5915
    },
    {
      "epoch": 0.3221498437850973,
      "grad_norm": 0.6962122434622048,
      "learning_rate": 0.0001911240270290748,
      "loss": 12.6427,
      "step": 5916
    },
    {
      "epoch": 0.3222042977816803,
      "grad_norm": 0.6254373248490377,
      "learning_rate": 0.0001911203946530498,
      "loss": 12.5056,
      "step": 5917
    },
    {
      "epoch": 0.32225875177826335,
      "grad_norm": 0.6217407116597037,
      "learning_rate": 0.00019111676156845735,
      "loss": 12.4804,
      "step": 5918
    },
    {
      "epoch": 0.32231320577484635,
      "grad_norm": 0.7202595630434092,
      "learning_rate": 0.00019111312777532566,
      "loss": 12.5083,
      "step": 5919
    },
    {
      "epoch": 0.32236765977142934,
      "grad_norm": 0.6820159788858451,
      "learning_rate": 0.000191109493273683,
      "loss": 12.4608,
      "step": 5920
    },
    {
      "epoch": 0.3224221137680124,
      "grad_norm": 0.7085165494770425,
      "learning_rate": 0.0001911058580635577,
      "loss": 12.4741,
      "step": 5921
    },
    {
      "epoch": 0.3224765677645954,
      "grad_norm": 0.6757057788462777,
      "learning_rate": 0.0001911022221449779,
      "loss": 12.4002,
      "step": 5922
    },
    {
      "epoch": 0.32253102176117837,
      "grad_norm": 0.7049609976493082,
      "learning_rate": 0.00019109858551797198,
      "loss": 12.5587,
      "step": 5923
    },
    {
      "epoch": 0.3225854757577614,
      "grad_norm": 0.6373738244859181,
      "learning_rate": 0.00019109494818256816,
      "loss": 12.3764,
      "step": 5924
    },
    {
      "epoch": 0.3226399297543444,
      "grad_norm": 0.6628983708514188,
      "learning_rate": 0.00019109131013879475,
      "loss": 12.4796,
      "step": 5925
    },
    {
      "epoch": 0.3226943837509274,
      "grad_norm": 0.6350872853299196,
      "learning_rate": 0.00019108767138668005,
      "loss": 12.4353,
      "step": 5926
    },
    {
      "epoch": 0.32274883774751045,
      "grad_norm": 0.6424512411543644,
      "learning_rate": 0.00019108403192625236,
      "loss": 12.4423,
      "step": 5927
    },
    {
      "epoch": 0.32280329174409345,
      "grad_norm": 0.7047099081442071,
      "learning_rate": 0.00019108039175753992,
      "loss": 12.6113,
      "step": 5928
    },
    {
      "epoch": 0.32285774574067644,
      "grad_norm": 0.6543119269839901,
      "learning_rate": 0.00019107675088057108,
      "loss": 12.6195,
      "step": 5929
    },
    {
      "epoch": 0.3229121997372595,
      "grad_norm": 0.717057102160912,
      "learning_rate": 0.00019107310929537417,
      "loss": 12.3956,
      "step": 5930
    },
    {
      "epoch": 0.3229666537338425,
      "grad_norm": 0.7362917478799178,
      "learning_rate": 0.0001910694670019775,
      "loss": 12.6826,
      "step": 5931
    },
    {
      "epoch": 0.32302110773042547,
      "grad_norm": 0.6330396736954499,
      "learning_rate": 0.00019106582400040938,
      "loss": 12.5617,
      "step": 5932
    },
    {
      "epoch": 0.3230755617270085,
      "grad_norm": 0.6650985491791709,
      "learning_rate": 0.00019106218029069812,
      "loss": 12.6838,
      "step": 5933
    },
    {
      "epoch": 0.3231300157235915,
      "grad_norm": 0.6495988085164426,
      "learning_rate": 0.0001910585358728721,
      "loss": 12.4752,
      "step": 5934
    },
    {
      "epoch": 0.3231844697201745,
      "grad_norm": 0.595367641986107,
      "learning_rate": 0.0001910548907469596,
      "loss": 12.49,
      "step": 5935
    },
    {
      "epoch": 0.32323892371675755,
      "grad_norm": 0.7195058905322125,
      "learning_rate": 0.00019105124491298906,
      "loss": 12.5126,
      "step": 5936
    },
    {
      "epoch": 0.32329337771334055,
      "grad_norm": 0.6263694786258669,
      "learning_rate": 0.00019104759837098868,
      "loss": 12.6228,
      "step": 5937
    },
    {
      "epoch": 0.32334783170992354,
      "grad_norm": 0.6735130785459177,
      "learning_rate": 0.000191043951120987,
      "loss": 12.4233,
      "step": 5938
    },
    {
      "epoch": 0.3234022857065066,
      "grad_norm": 0.6343518628030865,
      "learning_rate": 0.00019104030316301223,
      "loss": 12.5953,
      "step": 5939
    },
    {
      "epoch": 0.3234567397030896,
      "grad_norm": 0.6339707761405831,
      "learning_rate": 0.0001910366544970928,
      "loss": 12.4378,
      "step": 5940
    },
    {
      "epoch": 0.3235111936996726,
      "grad_norm": 0.9740575909840364,
      "learning_rate": 0.00019103300512325708,
      "loss": 12.6041,
      "step": 5941
    },
    {
      "epoch": 0.3235656476962556,
      "grad_norm": 0.6471335562445306,
      "learning_rate": 0.00019102935504153348,
      "loss": 12.3995,
      "step": 5942
    },
    {
      "epoch": 0.3236201016928386,
      "grad_norm": 0.6965000222741758,
      "learning_rate": 0.00019102570425195032,
      "loss": 12.4683,
      "step": 5943
    },
    {
      "epoch": 0.3236745556894216,
      "grad_norm": 0.6845613506290676,
      "learning_rate": 0.000191022052754536,
      "loss": 12.5708,
      "step": 5944
    },
    {
      "epoch": 0.32372900968600465,
      "grad_norm": 1.055426781939791,
      "learning_rate": 0.00019101840054931897,
      "loss": 12.5449,
      "step": 5945
    },
    {
      "epoch": 0.32378346368258765,
      "grad_norm": 0.6478370530528846,
      "learning_rate": 0.0001910147476363276,
      "loss": 12.4678,
      "step": 5946
    },
    {
      "epoch": 0.32383791767917064,
      "grad_norm": 0.7236229953266451,
      "learning_rate": 0.00019101109401559025,
      "loss": 12.45,
      "step": 5947
    },
    {
      "epoch": 0.3238923716757537,
      "grad_norm": 0.7776354757900491,
      "learning_rate": 0.0001910074396871354,
      "loss": 12.7088,
      "step": 5948
    },
    {
      "epoch": 0.3239468256723367,
      "grad_norm": 0.747095157378711,
      "learning_rate": 0.00019100378465099143,
      "loss": 12.4241,
      "step": 5949
    },
    {
      "epoch": 0.32400127966891973,
      "grad_norm": 0.7026491659725063,
      "learning_rate": 0.00019100012890718674,
      "loss": 12.5919,
      "step": 5950
    },
    {
      "epoch": 0.3240557336655027,
      "grad_norm": 0.6995609889373499,
      "learning_rate": 0.00019099647245574981,
      "loss": 12.6247,
      "step": 5951
    },
    {
      "epoch": 0.3241101876620857,
      "grad_norm": 0.8774373263694469,
      "learning_rate": 0.00019099281529670907,
      "loss": 12.4489,
      "step": 5952
    },
    {
      "epoch": 0.32416464165866876,
      "grad_norm": 0.6405476430956029,
      "learning_rate": 0.0001909891574300929,
      "loss": 12.5104,
      "step": 5953
    },
    {
      "epoch": 0.32421909565525175,
      "grad_norm": 0.6562550029524373,
      "learning_rate": 0.00019098549885592983,
      "loss": 12.663,
      "step": 5954
    },
    {
      "epoch": 0.32427354965183475,
      "grad_norm": 0.66271688148055,
      "learning_rate": 0.00019098183957424824,
      "loss": 12.5827,
      "step": 5955
    },
    {
      "epoch": 0.3243280036484178,
      "grad_norm": 0.6147209762320107,
      "learning_rate": 0.0001909781795850766,
      "loss": 12.5188,
      "step": 5956
    },
    {
      "epoch": 0.3243824576450008,
      "grad_norm": 0.7348392864009599,
      "learning_rate": 0.00019097451888844337,
      "loss": 12.5578,
      "step": 5957
    },
    {
      "epoch": 0.3244369116415838,
      "grad_norm": 0.6520863870566146,
      "learning_rate": 0.00019097085748437704,
      "loss": 12.2885,
      "step": 5958
    },
    {
      "epoch": 0.32449136563816683,
      "grad_norm": 0.6079312001278564,
      "learning_rate": 0.00019096719537290606,
      "loss": 12.3898,
      "step": 5959
    },
    {
      "epoch": 0.3245458196347498,
      "grad_norm": 0.6314519112721095,
      "learning_rate": 0.00019096353255405892,
      "loss": 12.4193,
      "step": 5960
    },
    {
      "epoch": 0.3246002736313328,
      "grad_norm": 0.6570340381863177,
      "learning_rate": 0.0001909598690278641,
      "loss": 12.4791,
      "step": 5961
    },
    {
      "epoch": 0.32465472762791586,
      "grad_norm": 0.6753086555455631,
      "learning_rate": 0.0001909562047943501,
      "loss": 12.5465,
      "step": 5962
    },
    {
      "epoch": 0.32470918162449885,
      "grad_norm": 0.730108073855115,
      "learning_rate": 0.00019095253985354534,
      "loss": 12.5237,
      "step": 5963
    },
    {
      "epoch": 0.32476363562108185,
      "grad_norm": 0.6205806638320429,
      "learning_rate": 0.00019094887420547844,
      "loss": 12.5215,
      "step": 5964
    },
    {
      "epoch": 0.3248180896176649,
      "grad_norm": 0.6353200889527242,
      "learning_rate": 0.0001909452078501778,
      "loss": 12.5001,
      "step": 5965
    },
    {
      "epoch": 0.3248725436142479,
      "grad_norm": 0.9432560778936758,
      "learning_rate": 0.000190941540787672,
      "loss": 12.6029,
      "step": 5966
    },
    {
      "epoch": 0.3249269976108309,
      "grad_norm": 0.6341818292480685,
      "learning_rate": 0.00019093787301798952,
      "loss": 12.5276,
      "step": 5967
    },
    {
      "epoch": 0.32498145160741393,
      "grad_norm": 0.6416082367670348,
      "learning_rate": 0.00019093420454115886,
      "loss": 12.593,
      "step": 5968
    },
    {
      "epoch": 0.3250359056039969,
      "grad_norm": 0.8068450582418325,
      "learning_rate": 0.00019093053535720861,
      "loss": 12.5589,
      "step": 5969
    },
    {
      "epoch": 0.3250903596005799,
      "grad_norm": 0.7187539550802996,
      "learning_rate": 0.00019092686546616725,
      "loss": 12.4668,
      "step": 5970
    },
    {
      "epoch": 0.32514481359716296,
      "grad_norm": 0.7413479385730268,
      "learning_rate": 0.00019092319486806335,
      "loss": 12.5876,
      "step": 5971
    },
    {
      "epoch": 0.32519926759374596,
      "grad_norm": 0.7031008087918882,
      "learning_rate": 0.00019091952356292544,
      "loss": 12.5985,
      "step": 5972
    },
    {
      "epoch": 0.32525372159032895,
      "grad_norm": 0.715757330929554,
      "learning_rate": 0.00019091585155078206,
      "loss": 12.5552,
      "step": 5973
    },
    {
      "epoch": 0.325308175586912,
      "grad_norm": 0.8596223200264557,
      "learning_rate": 0.00019091217883166178,
      "loss": 12.5456,
      "step": 5974
    },
    {
      "epoch": 0.325362629583495,
      "grad_norm": 0.607716778638219,
      "learning_rate": 0.00019090850540559316,
      "loss": 12.3831,
      "step": 5975
    },
    {
      "epoch": 0.325417083580078,
      "grad_norm": 0.69632444557811,
      "learning_rate": 0.00019090483127260472,
      "loss": 12.5443,
      "step": 5976
    },
    {
      "epoch": 0.32547153757666103,
      "grad_norm": 0.7908830884226349,
      "learning_rate": 0.00019090115643272508,
      "loss": 12.6423,
      "step": 5977
    },
    {
      "epoch": 0.325525991573244,
      "grad_norm": 0.7317983369210814,
      "learning_rate": 0.00019089748088598282,
      "loss": 12.6053,
      "step": 5978
    },
    {
      "epoch": 0.325580445569827,
      "grad_norm": 0.6860666876617846,
      "learning_rate": 0.0001908938046324065,
      "loss": 12.4089,
      "step": 5979
    },
    {
      "epoch": 0.32563489956641006,
      "grad_norm": 0.7133453663417441,
      "learning_rate": 0.0001908901276720247,
      "loss": 12.65,
      "step": 5980
    },
    {
      "epoch": 0.32568935356299306,
      "grad_norm": 0.6771444424428702,
      "learning_rate": 0.00019088645000486603,
      "loss": 12.6196,
      "step": 5981
    },
    {
      "epoch": 0.32574380755957605,
      "grad_norm": 0.6618186331476609,
      "learning_rate": 0.0001908827716309591,
      "loss": 12.4131,
      "step": 5982
    },
    {
      "epoch": 0.3257982615561591,
      "grad_norm": 0.666251776743328,
      "learning_rate": 0.0001908790925503325,
      "loss": 12.5763,
      "step": 5983
    },
    {
      "epoch": 0.3258527155527421,
      "grad_norm": 0.6423401599328059,
      "learning_rate": 0.00019087541276301479,
      "loss": 12.6437,
      "step": 5984
    },
    {
      "epoch": 0.32590716954932514,
      "grad_norm": 0.684078308546453,
      "learning_rate": 0.00019087173226903467,
      "loss": 12.4504,
      "step": 5985
    },
    {
      "epoch": 0.32596162354590813,
      "grad_norm": 0.679832543699596,
      "learning_rate": 0.00019086805106842072,
      "loss": 12.6542,
      "step": 5986
    },
    {
      "epoch": 0.3260160775424911,
      "grad_norm": 0.6848783699223823,
      "learning_rate": 0.00019086436916120153,
      "loss": 12.6109,
      "step": 5987
    },
    {
      "epoch": 0.32607053153907417,
      "grad_norm": 0.6741532940091434,
      "learning_rate": 0.0001908606865474058,
      "loss": 12.492,
      "step": 5988
    },
    {
      "epoch": 0.32612498553565716,
      "grad_norm": 0.6973214200387228,
      "learning_rate": 0.00019085700322706215,
      "loss": 12.5396,
      "step": 5989
    },
    {
      "epoch": 0.32617943953224016,
      "grad_norm": 0.6819143354267655,
      "learning_rate": 0.00019085331920019917,
      "loss": 12.5189,
      "step": 5990
    },
    {
      "epoch": 0.3262338935288232,
      "grad_norm": 0.7607303886975703,
      "learning_rate": 0.00019084963446684556,
      "loss": 12.5383,
      "step": 5991
    },
    {
      "epoch": 0.3262883475254062,
      "grad_norm": 0.7104535783369613,
      "learning_rate": 0.00019084594902702996,
      "loss": 12.6959,
      "step": 5992
    },
    {
      "epoch": 0.3263428015219892,
      "grad_norm": 0.6869245037086184,
      "learning_rate": 0.000190842262880781,
      "loss": 12.4045,
      "step": 5993
    },
    {
      "epoch": 0.32639725551857224,
      "grad_norm": 0.6891889185433268,
      "learning_rate": 0.0001908385760281274,
      "loss": 12.6741,
      "step": 5994
    },
    {
      "epoch": 0.32645170951515523,
      "grad_norm": 0.7146884740935388,
      "learning_rate": 0.0001908348884690978,
      "loss": 12.5421,
      "step": 5995
    },
    {
      "epoch": 0.3265061635117382,
      "grad_norm": 0.660700226334539,
      "learning_rate": 0.00019083120020372087,
      "loss": 12.4129,
      "step": 5996
    },
    {
      "epoch": 0.32656061750832127,
      "grad_norm": 0.6804830605995829,
      "learning_rate": 0.0001908275112320253,
      "loss": 12.3702,
      "step": 5997
    },
    {
      "epoch": 0.32661507150490426,
      "grad_norm": 0.6558070236451653,
      "learning_rate": 0.00019082382155403976,
      "loss": 12.5424,
      "step": 5998
    },
    {
      "epoch": 0.32666952550148726,
      "grad_norm": 0.6585152289440808,
      "learning_rate": 0.00019082013116979293,
      "loss": 12.6123,
      "step": 5999
    },
    {
      "epoch": 0.3267239794980703,
      "grad_norm": 0.6741903332937523,
      "learning_rate": 0.00019081644007931355,
      "loss": 12.4616,
      "step": 6000
    },
    {
      "epoch": 0.3267784334946533,
      "grad_norm": 0.6166158213759575,
      "learning_rate": 0.0001908127482826303,
      "loss": 12.6367,
      "step": 6001
    },
    {
      "epoch": 0.3268328874912363,
      "grad_norm": 0.6880785638149681,
      "learning_rate": 0.0001908090557797719,
      "loss": 12.4693,
      "step": 6002
    },
    {
      "epoch": 0.32688734148781934,
      "grad_norm": 0.6656609169611086,
      "learning_rate": 0.00019080536257076706,
      "loss": 12.5524,
      "step": 6003
    },
    {
      "epoch": 0.32694179548440233,
      "grad_norm": 0.8859792489670578,
      "learning_rate": 0.00019080166865564446,
      "loss": 12.4053,
      "step": 6004
    },
    {
      "epoch": 0.3269962494809853,
      "grad_norm": 0.6772545056208928,
      "learning_rate": 0.0001907979740344329,
      "loss": 12.5152,
      "step": 6005
    },
    {
      "epoch": 0.32705070347756837,
      "grad_norm": 0.7193161190009105,
      "learning_rate": 0.00019079427870716105,
      "loss": 12.575,
      "step": 6006
    },
    {
      "epoch": 0.32710515747415136,
      "grad_norm": 0.6766156295467368,
      "learning_rate": 0.00019079058267385763,
      "loss": 12.5792,
      "step": 6007
    },
    {
      "epoch": 0.32715961147073436,
      "grad_norm": 0.65637458444529,
      "learning_rate": 0.00019078688593455144,
      "loss": 12.4607,
      "step": 6008
    },
    {
      "epoch": 0.3272140654673174,
      "grad_norm": 0.7477260194565057,
      "learning_rate": 0.0001907831884892712,
      "loss": 12.6327,
      "step": 6009
    },
    {
      "epoch": 0.3272685194639004,
      "grad_norm": 0.7543706752417337,
      "learning_rate": 0.00019077949033804566,
      "loss": 12.5272,
      "step": 6010
    },
    {
      "epoch": 0.3273229734604834,
      "grad_norm": 0.7882627811447465,
      "learning_rate": 0.0001907757914809036,
      "loss": 12.5733,
      "step": 6011
    },
    {
      "epoch": 0.32737742745706644,
      "grad_norm": 0.6942572401963698,
      "learning_rate": 0.00019077209191787375,
      "loss": 12.4568,
      "step": 6012
    },
    {
      "epoch": 0.32743188145364943,
      "grad_norm": 0.71472777836806,
      "learning_rate": 0.00019076839164898488,
      "loss": 12.503,
      "step": 6013
    },
    {
      "epoch": 0.3274863354502324,
      "grad_norm": 0.6458760780360772,
      "learning_rate": 0.00019076469067426578,
      "loss": 12.5897,
      "step": 6014
    },
    {
      "epoch": 0.3275407894468155,
      "grad_norm": 0.6523758206718518,
      "learning_rate": 0.0001907609889937452,
      "loss": 12.5079,
      "step": 6015
    },
    {
      "epoch": 0.32759524344339847,
      "grad_norm": 0.6639159221337562,
      "learning_rate": 0.00019075728660745197,
      "loss": 12.4423,
      "step": 6016
    },
    {
      "epoch": 0.3276496974399815,
      "grad_norm": 0.7536010462227256,
      "learning_rate": 0.00019075358351541488,
      "loss": 12.5858,
      "step": 6017
    },
    {
      "epoch": 0.3277041514365645,
      "grad_norm": 0.7421258119115142,
      "learning_rate": 0.0001907498797176627,
      "loss": 12.6127,
      "step": 6018
    },
    {
      "epoch": 0.3277586054331475,
      "grad_norm": 0.6908788776147277,
      "learning_rate": 0.00019074617521422423,
      "loss": 12.561,
      "step": 6019
    },
    {
      "epoch": 0.32781305942973055,
      "grad_norm": 0.6923049654838966,
      "learning_rate": 0.00019074247000512825,
      "loss": 12.3786,
      "step": 6020
    },
    {
      "epoch": 0.32786751342631354,
      "grad_norm": 0.6881176792638315,
      "learning_rate": 0.00019073876409040366,
      "loss": 12.5307,
      "step": 6021
    },
    {
      "epoch": 0.32792196742289653,
      "grad_norm": 0.6819501022563224,
      "learning_rate": 0.0001907350574700792,
      "loss": 12.5123,
      "step": 6022
    },
    {
      "epoch": 0.3279764214194796,
      "grad_norm": 0.6768439916521216,
      "learning_rate": 0.0001907313501441837,
      "loss": 12.626,
      "step": 6023
    },
    {
      "epoch": 0.3280308754160626,
      "grad_norm": 0.7041701961335308,
      "learning_rate": 0.000190727642112746,
      "loss": 12.4595,
      "step": 6024
    },
    {
      "epoch": 0.32808532941264557,
      "grad_norm": 0.704760083849161,
      "learning_rate": 0.00019072393337579499,
      "loss": 12.5799,
      "step": 6025
    },
    {
      "epoch": 0.3281397834092286,
      "grad_norm": 0.6995993768892956,
      "learning_rate": 0.00019072022393335942,
      "loss": 12.384,
      "step": 6026
    },
    {
      "epoch": 0.3281942374058116,
      "grad_norm": 0.7527545179313321,
      "learning_rate": 0.00019071651378546817,
      "loss": 12.5722,
      "step": 6027
    },
    {
      "epoch": 0.3282486914023946,
      "grad_norm": 0.6389152098343975,
      "learning_rate": 0.0001907128029321501,
      "loss": 12.4403,
      "step": 6028
    },
    {
      "epoch": 0.32830314539897765,
      "grad_norm": 0.6888724064811103,
      "learning_rate": 0.00019070909137343408,
      "loss": 12.6418,
      "step": 6029
    },
    {
      "epoch": 0.32835759939556064,
      "grad_norm": 0.6751648270814726,
      "learning_rate": 0.00019070537910934895,
      "loss": 12.6121,
      "step": 6030
    },
    {
      "epoch": 0.32841205339214363,
      "grad_norm": 0.7575202162096621,
      "learning_rate": 0.00019070166613992357,
      "loss": 12.6031,
      "step": 6031
    },
    {
      "epoch": 0.3284665073887267,
      "grad_norm": 0.6958833801819441,
      "learning_rate": 0.00019069795246518683,
      "loss": 12.6046,
      "step": 6032
    },
    {
      "epoch": 0.3285209613853097,
      "grad_norm": 0.6457389212528368,
      "learning_rate": 0.0001906942380851676,
      "loss": 12.4216,
      "step": 6033
    },
    {
      "epoch": 0.32857541538189267,
      "grad_norm": 0.6765810427732396,
      "learning_rate": 0.00019069052299989475,
      "loss": 12.5173,
      "step": 6034
    },
    {
      "epoch": 0.3286298693784757,
      "grad_norm": 0.6486662081210898,
      "learning_rate": 0.00019068680720939722,
      "loss": 12.5056,
      "step": 6035
    },
    {
      "epoch": 0.3286843233750587,
      "grad_norm": 0.6393177510255944,
      "learning_rate": 0.00019068309071370386,
      "loss": 12.5253,
      "step": 6036
    },
    {
      "epoch": 0.3287387773716417,
      "grad_norm": 0.7043933590676535,
      "learning_rate": 0.00019067937351284356,
      "loss": 12.4581,
      "step": 6037
    },
    {
      "epoch": 0.32879323136822475,
      "grad_norm": 0.7073559719836241,
      "learning_rate": 0.00019067565560684525,
      "loss": 12.6616,
      "step": 6038
    },
    {
      "epoch": 0.32884768536480774,
      "grad_norm": 0.6695162577213956,
      "learning_rate": 0.00019067193699573784,
      "loss": 12.4207,
      "step": 6039
    },
    {
      "epoch": 0.32890213936139073,
      "grad_norm": 0.6957632386249147,
      "learning_rate": 0.00019066821767955023,
      "loss": 12.6009,
      "step": 6040
    },
    {
      "epoch": 0.3289565933579738,
      "grad_norm": 0.6786726297932553,
      "learning_rate": 0.00019066449765831135,
      "loss": 12.5736,
      "step": 6041
    },
    {
      "epoch": 0.3290110473545568,
      "grad_norm": 0.7542246599893674,
      "learning_rate": 0.00019066077693205018,
      "loss": 12.4497,
      "step": 6042
    },
    {
      "epoch": 0.32906550135113977,
      "grad_norm": 0.6266163221327751,
      "learning_rate": 0.00019065705550079556,
      "loss": 12.4521,
      "step": 6043
    },
    {
      "epoch": 0.3291199553477228,
      "grad_norm": 0.6632501992328229,
      "learning_rate": 0.0001906533333645765,
      "loss": 12.5465,
      "step": 6044
    },
    {
      "epoch": 0.3291744093443058,
      "grad_norm": 0.7173246364462722,
      "learning_rate": 0.0001906496105234219,
      "loss": 12.4319,
      "step": 6045
    },
    {
      "epoch": 0.3292288633408888,
      "grad_norm": 0.6601271095282973,
      "learning_rate": 0.00019064588697736073,
      "loss": 12.6277,
      "step": 6046
    },
    {
      "epoch": 0.32928331733747185,
      "grad_norm": 0.7662957908838449,
      "learning_rate": 0.00019064216272642192,
      "loss": 12.4888,
      "step": 6047
    },
    {
      "epoch": 0.32933777133405484,
      "grad_norm": 0.9193653350114078,
      "learning_rate": 0.00019063843777063447,
      "loss": 12.7076,
      "step": 6048
    },
    {
      "epoch": 0.32939222533063783,
      "grad_norm": 0.9441997690647264,
      "learning_rate": 0.00019063471211002732,
      "loss": 12.5137,
      "step": 6049
    },
    {
      "epoch": 0.3294466793272209,
      "grad_norm": 0.7440618082238332,
      "learning_rate": 0.0001906309857446295,
      "loss": 12.6509,
      "step": 6050
    },
    {
      "epoch": 0.3295011333238039,
      "grad_norm": 0.6706221552479642,
      "learning_rate": 0.00019062725867446985,
      "loss": 12.4081,
      "step": 6051
    },
    {
      "epoch": 0.3295555873203869,
      "grad_norm": 0.7669635585081318,
      "learning_rate": 0.0001906235308995775,
      "loss": 12.3935,
      "step": 6052
    },
    {
      "epoch": 0.3296100413169699,
      "grad_norm": 0.8102674228184634,
      "learning_rate": 0.00019061980241998137,
      "loss": 12.6809,
      "step": 6053
    },
    {
      "epoch": 0.3296644953135529,
      "grad_norm": 0.6051046325273823,
      "learning_rate": 0.00019061607323571042,
      "loss": 12.3263,
      "step": 6054
    },
    {
      "epoch": 0.32971894931013596,
      "grad_norm": 0.6912898651248617,
      "learning_rate": 0.00019061234334679373,
      "loss": 12.4846,
      "step": 6055
    },
    {
      "epoch": 0.32977340330671895,
      "grad_norm": 0.7745620986720093,
      "learning_rate": 0.00019060861275326026,
      "loss": 12.6085,
      "step": 6056
    },
    {
      "epoch": 0.32982785730330194,
      "grad_norm": 0.6763436075375078,
      "learning_rate": 0.000190604881455139,
      "loss": 12.6575,
      "step": 6057
    },
    {
      "epoch": 0.329882311299885,
      "grad_norm": 0.7307302740737248,
      "learning_rate": 0.000190601149452459,
      "loss": 12.4515,
      "step": 6058
    },
    {
      "epoch": 0.329936765296468,
      "grad_norm": 0.7506534175175649,
      "learning_rate": 0.00019059741674524924,
      "loss": 12.7086,
      "step": 6059
    },
    {
      "epoch": 0.329991219293051,
      "grad_norm": 0.6310610641892026,
      "learning_rate": 0.0001905936833335388,
      "loss": 12.5889,
      "step": 6060
    },
    {
      "epoch": 0.330045673289634,
      "grad_norm": 0.749115126468661,
      "learning_rate": 0.00019058994921735672,
      "loss": 12.6396,
      "step": 6061
    },
    {
      "epoch": 0.330100127286217,
      "grad_norm": 0.6680972574726077,
      "learning_rate": 0.00019058621439673194,
      "loss": 12.6789,
      "step": 6062
    },
    {
      "epoch": 0.3301545812828,
      "grad_norm": 0.6498361336323334,
      "learning_rate": 0.00019058247887169361,
      "loss": 12.5572,
      "step": 6063
    },
    {
      "epoch": 0.33020903527938306,
      "grad_norm": 0.728884306740224,
      "learning_rate": 0.0001905787426422707,
      "loss": 12.4432,
      "step": 6064
    },
    {
      "epoch": 0.33026348927596605,
      "grad_norm": 0.6381536620973719,
      "learning_rate": 0.00019057500570849234,
      "loss": 12.4457,
      "step": 6065
    },
    {
      "epoch": 0.33031794327254904,
      "grad_norm": 0.6814152295949502,
      "learning_rate": 0.00019057126807038753,
      "loss": 12.4274,
      "step": 6066
    },
    {
      "epoch": 0.3303723972691321,
      "grad_norm": 0.7586037938328459,
      "learning_rate": 0.00019056752972798532,
      "loss": 12.4917,
      "step": 6067
    },
    {
      "epoch": 0.3304268512657151,
      "grad_norm": 0.7013635463082079,
      "learning_rate": 0.00019056379068131484,
      "loss": 12.5281,
      "step": 6068
    },
    {
      "epoch": 0.3304813052622981,
      "grad_norm": 0.749680286113315,
      "learning_rate": 0.00019056005093040512,
      "loss": 12.5624,
      "step": 6069
    },
    {
      "epoch": 0.3305357592588811,
      "grad_norm": 0.7209385704762635,
      "learning_rate": 0.00019055631047528528,
      "loss": 12.5035,
      "step": 6070
    },
    {
      "epoch": 0.3305902132554641,
      "grad_norm": 0.7540812916934254,
      "learning_rate": 0.00019055256931598438,
      "loss": 12.5679,
      "step": 6071
    },
    {
      "epoch": 0.3306446672520471,
      "grad_norm": 0.7230512092731918,
      "learning_rate": 0.0001905488274525315,
      "loss": 12.6442,
      "step": 6072
    },
    {
      "epoch": 0.33069912124863016,
      "grad_norm": 1.0561995452306372,
      "learning_rate": 0.00019054508488495575,
      "loss": 12.5927,
      "step": 6073
    },
    {
      "epoch": 0.33075357524521315,
      "grad_norm": 0.7502687032082127,
      "learning_rate": 0.00019054134161328626,
      "loss": 12.5629,
      "step": 6074
    },
    {
      "epoch": 0.33080802924179614,
      "grad_norm": 0.7999552476061841,
      "learning_rate": 0.00019053759763755209,
      "loss": 12.5188,
      "step": 6075
    },
    {
      "epoch": 0.3308624832383792,
      "grad_norm": 0.7713322123206169,
      "learning_rate": 0.0001905338529577824,
      "loss": 12.5704,
      "step": 6076
    },
    {
      "epoch": 0.3309169372349622,
      "grad_norm": 0.8188983023010487,
      "learning_rate": 0.00019053010757400624,
      "loss": 12.6,
      "step": 6077
    },
    {
      "epoch": 0.3309713912315452,
      "grad_norm": 0.9511635742495728,
      "learning_rate": 0.00019052636148625282,
      "loss": 12.5989,
      "step": 6078
    },
    {
      "epoch": 0.3310258452281282,
      "grad_norm": 0.6767946812635767,
      "learning_rate": 0.00019052261469455122,
      "loss": 12.5205,
      "step": 6079
    },
    {
      "epoch": 0.3310802992247112,
      "grad_norm": 0.9301201404025043,
      "learning_rate": 0.0001905188671989306,
      "loss": 12.7243,
      "step": 6080
    },
    {
      "epoch": 0.3311347532212942,
      "grad_norm": 0.9015049000301367,
      "learning_rate": 0.0001905151189994201,
      "loss": 12.4499,
      "step": 6081
    },
    {
      "epoch": 0.33118920721787726,
      "grad_norm": 0.8382596251722875,
      "learning_rate": 0.0001905113700960488,
      "loss": 12.7573,
      "step": 6082
    },
    {
      "epoch": 0.33124366121446025,
      "grad_norm": 0.7181926507073133,
      "learning_rate": 0.00019050762048884596,
      "loss": 12.473,
      "step": 6083
    },
    {
      "epoch": 0.3312981152110433,
      "grad_norm": 0.8447246018943421,
      "learning_rate": 0.0001905038701778407,
      "loss": 12.515,
      "step": 6084
    },
    {
      "epoch": 0.3313525692076263,
      "grad_norm": 0.752685655651504,
      "learning_rate": 0.00019050011916306212,
      "loss": 12.5246,
      "step": 6085
    },
    {
      "epoch": 0.3314070232042093,
      "grad_norm": 0.8405698619472585,
      "learning_rate": 0.00019049636744453944,
      "loss": 12.5166,
      "step": 6086
    },
    {
      "epoch": 0.33146147720079233,
      "grad_norm": 0.7746313603997682,
      "learning_rate": 0.00019049261502230184,
      "loss": 12.4498,
      "step": 6087
    },
    {
      "epoch": 0.3315159311973753,
      "grad_norm": 0.8372687652387514,
      "learning_rate": 0.00019048886189637848,
      "loss": 12.5878,
      "step": 6088
    },
    {
      "epoch": 0.3315703851939583,
      "grad_norm": 0.814258741770328,
      "learning_rate": 0.0001904851080667986,
      "loss": 12.4877,
      "step": 6089
    },
    {
      "epoch": 0.33162483919054137,
      "grad_norm": 0.7280978774222061,
      "learning_rate": 0.00019048135353359129,
      "loss": 12.5206,
      "step": 6090
    },
    {
      "epoch": 0.33167929318712436,
      "grad_norm": 0.7893481669060788,
      "learning_rate": 0.00019047759829678585,
      "loss": 12.6034,
      "step": 6091
    },
    {
      "epoch": 0.33173374718370735,
      "grad_norm": 0.7136790992150998,
      "learning_rate": 0.00019047384235641138,
      "loss": 12.5442,
      "step": 6092
    },
    {
      "epoch": 0.3317882011802904,
      "grad_norm": 0.725570200940181,
      "learning_rate": 0.00019047008571249717,
      "loss": 12.5172,
      "step": 6093
    },
    {
      "epoch": 0.3318426551768734,
      "grad_norm": 0.7172907098779102,
      "learning_rate": 0.0001904663283650724,
      "loss": 12.4823,
      "step": 6094
    },
    {
      "epoch": 0.3318971091734564,
      "grad_norm": 0.650578796409706,
      "learning_rate": 0.0001904625703141663,
      "loss": 12.4808,
      "step": 6095
    },
    {
      "epoch": 0.33195156317003943,
      "grad_norm": 0.7862084399830296,
      "learning_rate": 0.00019045881155980808,
      "loss": 12.5653,
      "step": 6096
    },
    {
      "epoch": 0.3320060171666224,
      "grad_norm": 0.6625604436200386,
      "learning_rate": 0.00019045505210202698,
      "loss": 12.6665,
      "step": 6097
    },
    {
      "epoch": 0.3320604711632054,
      "grad_norm": 0.6707462864315522,
      "learning_rate": 0.00019045129194085217,
      "loss": 12.6802,
      "step": 6098
    },
    {
      "epoch": 0.33211492515978847,
      "grad_norm": 0.6954853299279311,
      "learning_rate": 0.000190447531076313,
      "loss": 12.6003,
      "step": 6099
    },
    {
      "epoch": 0.33216937915637146,
      "grad_norm": 0.7880675830322864,
      "learning_rate": 0.00019044376950843862,
      "loss": 12.6436,
      "step": 6100
    },
    {
      "epoch": 0.33222383315295445,
      "grad_norm": 0.8561969283621422,
      "learning_rate": 0.00019044000723725837,
      "loss": 12.6777,
      "step": 6101
    },
    {
      "epoch": 0.3322782871495375,
      "grad_norm": 0.7027593267734076,
      "learning_rate": 0.0001904362442628014,
      "loss": 12.5563,
      "step": 6102
    },
    {
      "epoch": 0.3323327411461205,
      "grad_norm": 0.9422237632774334,
      "learning_rate": 0.0001904324805850971,
      "loss": 12.5437,
      "step": 6103
    },
    {
      "epoch": 0.3323871951427035,
      "grad_norm": 0.6967880551838675,
      "learning_rate": 0.0001904287162041746,
      "loss": 12.485,
      "step": 6104
    },
    {
      "epoch": 0.33244164913928653,
      "grad_norm": 0.7064817408333989,
      "learning_rate": 0.00019042495112006326,
      "loss": 12.4872,
      "step": 6105
    },
    {
      "epoch": 0.3324961031358695,
      "grad_norm": 0.861640560379946,
      "learning_rate": 0.00019042118533279235,
      "loss": 12.5858,
      "step": 6106
    },
    {
      "epoch": 0.3325505571324525,
      "grad_norm": 0.7716607183906974,
      "learning_rate": 0.00019041741884239113,
      "loss": 12.5587,
      "step": 6107
    },
    {
      "epoch": 0.33260501112903557,
      "grad_norm": 0.763541649976914,
      "learning_rate": 0.00019041365164888888,
      "loss": 12.7393,
      "step": 6108
    },
    {
      "epoch": 0.33265946512561856,
      "grad_norm": 0.637888834883395,
      "learning_rate": 0.00019040988375231495,
      "loss": 12.5298,
      "step": 6109
    },
    {
      "epoch": 0.33271391912220155,
      "grad_norm": 0.7149384308769094,
      "learning_rate": 0.00019040611515269858,
      "loss": 12.6393,
      "step": 6110
    },
    {
      "epoch": 0.3327683731187846,
      "grad_norm": 0.6569658440919112,
      "learning_rate": 0.0001904023458500691,
      "loss": 12.5275,
      "step": 6111
    },
    {
      "epoch": 0.3328228271153676,
      "grad_norm": 0.6572020856486586,
      "learning_rate": 0.0001903985758444558,
      "loss": 12.6111,
      "step": 6112
    },
    {
      "epoch": 0.3328772811119506,
      "grad_norm": 0.6516354181852332,
      "learning_rate": 0.00019039480513588806,
      "loss": 12.5082,
      "step": 6113
    },
    {
      "epoch": 0.33293173510853363,
      "grad_norm": 0.680454891451818,
      "learning_rate": 0.00019039103372439512,
      "loss": 12.5876,
      "step": 6114
    },
    {
      "epoch": 0.3329861891051166,
      "grad_norm": 0.6523523263440932,
      "learning_rate": 0.00019038726161000634,
      "loss": 12.5589,
      "step": 6115
    },
    {
      "epoch": 0.3330406431016996,
      "grad_norm": 0.6829586009320958,
      "learning_rate": 0.00019038348879275106,
      "loss": 12.6797,
      "step": 6116
    },
    {
      "epoch": 0.33309509709828267,
      "grad_norm": 0.6661967631864086,
      "learning_rate": 0.0001903797152726586,
      "loss": 12.5284,
      "step": 6117
    },
    {
      "epoch": 0.33314955109486566,
      "grad_norm": 0.6074346941940245,
      "learning_rate": 0.00019037594104975836,
      "loss": 12.4863,
      "step": 6118
    },
    {
      "epoch": 0.3332040050914487,
      "grad_norm": 0.7862727576794416,
      "learning_rate": 0.00019037216612407962,
      "loss": 12.5459,
      "step": 6119
    },
    {
      "epoch": 0.3332584590880317,
      "grad_norm": 0.6660531417792179,
      "learning_rate": 0.00019036839049565177,
      "loss": 12.4761,
      "step": 6120
    },
    {
      "epoch": 0.3333129130846147,
      "grad_norm": 0.6365992142476833,
      "learning_rate": 0.00019036461416450416,
      "loss": 12.5827,
      "step": 6121
    },
    {
      "epoch": 0.33336736708119774,
      "grad_norm": 0.6579708556000738,
      "learning_rate": 0.00019036083713066612,
      "loss": 12.6584,
      "step": 6122
    },
    {
      "epoch": 0.33342182107778073,
      "grad_norm": 0.6397074316751428,
      "learning_rate": 0.0001903570593941671,
      "loss": 12.5182,
      "step": 6123
    },
    {
      "epoch": 0.3334762750743637,
      "grad_norm": 0.6753784333459871,
      "learning_rate": 0.00019035328095503643,
      "loss": 12.4446,
      "step": 6124
    },
    {
      "epoch": 0.3335307290709468,
      "grad_norm": 0.6293854289115678,
      "learning_rate": 0.00019034950181330348,
      "loss": 12.4655,
      "step": 6125
    },
    {
      "epoch": 0.33358518306752977,
      "grad_norm": 0.6768015115197763,
      "learning_rate": 0.00019034572196899766,
      "loss": 12.6518,
      "step": 6126
    },
    {
      "epoch": 0.33363963706411276,
      "grad_norm": 0.6028883086935058,
      "learning_rate": 0.00019034194142214834,
      "loss": 12.4527,
      "step": 6127
    },
    {
      "epoch": 0.3336940910606958,
      "grad_norm": 0.7512506851451719,
      "learning_rate": 0.00019033816017278497,
      "loss": 12.6985,
      "step": 6128
    },
    {
      "epoch": 0.3337485450572788,
      "grad_norm": 0.6914936250403905,
      "learning_rate": 0.0001903343782209369,
      "loss": 12.4452,
      "step": 6129
    },
    {
      "epoch": 0.3338029990538618,
      "grad_norm": 0.6491276706247793,
      "learning_rate": 0.00019033059556663353,
      "loss": 12.6077,
      "step": 6130
    },
    {
      "epoch": 0.33385745305044484,
      "grad_norm": 0.6610362402021156,
      "learning_rate": 0.0001903268122099043,
      "loss": 12.4978,
      "step": 6131
    },
    {
      "epoch": 0.33391190704702783,
      "grad_norm": 0.6583859042445529,
      "learning_rate": 0.00019032302815077866,
      "loss": 12.5562,
      "step": 6132
    },
    {
      "epoch": 0.3339663610436108,
      "grad_norm": 0.6849355025464567,
      "learning_rate": 0.000190319243389286,
      "loss": 12.5127,
      "step": 6133
    },
    {
      "epoch": 0.3340208150401939,
      "grad_norm": 0.6238656111143649,
      "learning_rate": 0.00019031545792545576,
      "loss": 12.478,
      "step": 6134
    },
    {
      "epoch": 0.33407526903677687,
      "grad_norm": 0.6796712942676156,
      "learning_rate": 0.00019031167175931736,
      "loss": 12.4635,
      "step": 6135
    },
    {
      "epoch": 0.33412972303335986,
      "grad_norm": 0.6388572352631766,
      "learning_rate": 0.00019030788489090027,
      "loss": 12.6054,
      "step": 6136
    },
    {
      "epoch": 0.3341841770299429,
      "grad_norm": 0.7070234751908636,
      "learning_rate": 0.0001903040973202339,
      "loss": 12.5226,
      "step": 6137
    },
    {
      "epoch": 0.3342386310265259,
      "grad_norm": 0.7598039478084103,
      "learning_rate": 0.00019030030904734774,
      "loss": 12.423,
      "step": 6138
    },
    {
      "epoch": 0.3342930850231089,
      "grad_norm": 0.7158751505478054,
      "learning_rate": 0.00019029652007227123,
      "loss": 12.5786,
      "step": 6139
    },
    {
      "epoch": 0.33434753901969194,
      "grad_norm": 0.6145429949826542,
      "learning_rate": 0.00019029273039503387,
      "loss": 12.4051,
      "step": 6140
    },
    {
      "epoch": 0.33440199301627493,
      "grad_norm": 1.1949463795614987,
      "learning_rate": 0.00019028894001566507,
      "loss": 12.7602,
      "step": 6141
    },
    {
      "epoch": 0.3344564470128579,
      "grad_norm": 0.6342220342865346,
      "learning_rate": 0.00019028514893419432,
      "loss": 12.4909,
      "step": 6142
    },
    {
      "epoch": 0.334510901009441,
      "grad_norm": 0.7442484705837343,
      "learning_rate": 0.0001902813571506511,
      "loss": 12.7064,
      "step": 6143
    },
    {
      "epoch": 0.33456535500602397,
      "grad_norm": 0.6796848417099655,
      "learning_rate": 0.00019027756466506493,
      "loss": 12.5268,
      "step": 6144
    },
    {
      "epoch": 0.33461980900260696,
      "grad_norm": 0.6546437357082819,
      "learning_rate": 0.00019027377147746524,
      "loss": 12.4814,
      "step": 6145
    },
    {
      "epoch": 0.33467426299919,
      "grad_norm": 0.7005982307981831,
      "learning_rate": 0.00019026997758788162,
      "loss": 12.6283,
      "step": 6146
    },
    {
      "epoch": 0.334728716995773,
      "grad_norm": 0.6438492247460936,
      "learning_rate": 0.0001902661829963435,
      "loss": 12.5332,
      "step": 6147
    },
    {
      "epoch": 0.334783170992356,
      "grad_norm": 0.7068632930100087,
      "learning_rate": 0.0001902623877028804,
      "loss": 12.5771,
      "step": 6148
    },
    {
      "epoch": 0.33483762498893904,
      "grad_norm": 0.740471689430596,
      "learning_rate": 0.00019025859170752183,
      "loss": 12.5317,
      "step": 6149
    },
    {
      "epoch": 0.33489207898552203,
      "grad_norm": 0.6222370742701101,
      "learning_rate": 0.0001902547950102973,
      "loss": 12.5265,
      "step": 6150
    },
    {
      "epoch": 0.3349465329821051,
      "grad_norm": 0.683206095891851,
      "learning_rate": 0.00019025099761123637,
      "loss": 12.6167,
      "step": 6151
    },
    {
      "epoch": 0.3350009869786881,
      "grad_norm": 0.7004754573579197,
      "learning_rate": 0.00019024719951036856,
      "loss": 12.6346,
      "step": 6152
    },
    {
      "epoch": 0.33505544097527107,
      "grad_norm": 0.677115356865327,
      "learning_rate": 0.00019024340070772336,
      "loss": 12.5155,
      "step": 6153
    },
    {
      "epoch": 0.3351098949718541,
      "grad_norm": 0.6041487063468916,
      "learning_rate": 0.00019023960120333037,
      "loss": 12.5213,
      "step": 6154
    },
    {
      "epoch": 0.3351643489684371,
      "grad_norm": 0.6440215634195646,
      "learning_rate": 0.00019023580099721907,
      "loss": 12.4842,
      "step": 6155
    },
    {
      "epoch": 0.3352188029650201,
      "grad_norm": 0.7218149914408177,
      "learning_rate": 0.00019023200008941912,
      "loss": 12.5914,
      "step": 6156
    },
    {
      "epoch": 0.33527325696160315,
      "grad_norm": 0.6775480085437992,
      "learning_rate": 0.00019022819847995992,
      "loss": 12.608,
      "step": 6157
    },
    {
      "epoch": 0.33532771095818614,
      "grad_norm": 0.6925979630379857,
      "learning_rate": 0.00019022439616887116,
      "loss": 12.6059,
      "step": 6158
    },
    {
      "epoch": 0.33538216495476914,
      "grad_norm": 0.6479672547492501,
      "learning_rate": 0.00019022059315618238,
      "loss": 12.4175,
      "step": 6159
    },
    {
      "epoch": 0.3354366189513522,
      "grad_norm": 0.767868850600018,
      "learning_rate": 0.0001902167894419231,
      "loss": 12.5171,
      "step": 6160
    },
    {
      "epoch": 0.3354910729479352,
      "grad_norm": 0.6072374536074713,
      "learning_rate": 0.00019021298502612294,
      "loss": 12.4668,
      "step": 6161
    },
    {
      "epoch": 0.33554552694451817,
      "grad_norm": 0.8744038116543146,
      "learning_rate": 0.0001902091799088115,
      "loss": 12.5322,
      "step": 6162
    },
    {
      "epoch": 0.3355999809411012,
      "grad_norm": 0.6870120413339742,
      "learning_rate": 0.00019020537409001836,
      "loss": 12.5823,
      "step": 6163
    },
    {
      "epoch": 0.3356544349376842,
      "grad_norm": 0.6446608768392295,
      "learning_rate": 0.00019020156756977309,
      "loss": 12.5155,
      "step": 6164
    },
    {
      "epoch": 0.3357088889342672,
      "grad_norm": 0.6695021313655353,
      "learning_rate": 0.00019019776034810527,
      "loss": 12.2837,
      "step": 6165
    },
    {
      "epoch": 0.33576334293085025,
      "grad_norm": 0.6905675103604029,
      "learning_rate": 0.00019019395242504458,
      "loss": 12.5365,
      "step": 6166
    },
    {
      "epoch": 0.33581779692743324,
      "grad_norm": 0.6892280744834528,
      "learning_rate": 0.0001901901438006206,
      "loss": 12.5121,
      "step": 6167
    },
    {
      "epoch": 0.33587225092401624,
      "grad_norm": 0.6263608033387138,
      "learning_rate": 0.00019018633447486288,
      "loss": 12.5616,
      "step": 6168
    },
    {
      "epoch": 0.3359267049205993,
      "grad_norm": 0.7290340194306969,
      "learning_rate": 0.00019018252444780116,
      "loss": 12.5323,
      "step": 6169
    },
    {
      "epoch": 0.3359811589171823,
      "grad_norm": 0.6877483949236548,
      "learning_rate": 0.00019017871371946498,
      "loss": 12.5425,
      "step": 6170
    },
    {
      "epoch": 0.33603561291376527,
      "grad_norm": 0.6121751565433697,
      "learning_rate": 0.000190174902289884,
      "loss": 12.4145,
      "step": 6171
    },
    {
      "epoch": 0.3360900669103483,
      "grad_norm": 0.6668410447130814,
      "learning_rate": 0.00019017109015908784,
      "loss": 12.6139,
      "step": 6172
    },
    {
      "epoch": 0.3361445209069313,
      "grad_norm": 0.6548950675385099,
      "learning_rate": 0.0001901672773271062,
      "loss": 12.7411,
      "step": 6173
    },
    {
      "epoch": 0.3361989749035143,
      "grad_norm": 0.6620132852747327,
      "learning_rate": 0.00019016346379396867,
      "loss": 12.5068,
      "step": 6174
    },
    {
      "epoch": 0.33625342890009735,
      "grad_norm": 0.6744886194570338,
      "learning_rate": 0.00019015964955970493,
      "loss": 12.3716,
      "step": 6175
    },
    {
      "epoch": 0.33630788289668034,
      "grad_norm": 0.6460492116216238,
      "learning_rate": 0.00019015583462434464,
      "loss": 12.6015,
      "step": 6176
    },
    {
      "epoch": 0.33636233689326334,
      "grad_norm": 0.7586681691462406,
      "learning_rate": 0.00019015201898791743,
      "loss": 12.5306,
      "step": 6177
    },
    {
      "epoch": 0.3364167908898464,
      "grad_norm": 0.7473788029604737,
      "learning_rate": 0.00019014820265045304,
      "loss": 12.5624,
      "step": 6178
    },
    {
      "epoch": 0.3364712448864294,
      "grad_norm": 0.6473311838707987,
      "learning_rate": 0.0001901443856119811,
      "loss": 12.446,
      "step": 6179
    },
    {
      "epoch": 0.33652569888301237,
      "grad_norm": 0.7744353742659558,
      "learning_rate": 0.0001901405678725313,
      "loss": 12.6546,
      "step": 6180
    },
    {
      "epoch": 0.3365801528795954,
      "grad_norm": 0.6991532130330834,
      "learning_rate": 0.00019013674943213328,
      "loss": 12.5082,
      "step": 6181
    },
    {
      "epoch": 0.3366346068761784,
      "grad_norm": 0.7557822736917306,
      "learning_rate": 0.00019013293029081685,
      "loss": 12.436,
      "step": 6182
    },
    {
      "epoch": 0.3366890608727614,
      "grad_norm": 0.6549872291957558,
      "learning_rate": 0.00019012911044861158,
      "loss": 12.5209,
      "step": 6183
    },
    {
      "epoch": 0.33674351486934445,
      "grad_norm": 0.6970846774936678,
      "learning_rate": 0.00019012528990554727,
      "loss": 12.5207,
      "step": 6184
    },
    {
      "epoch": 0.33679796886592744,
      "grad_norm": 0.8749360840578242,
      "learning_rate": 0.00019012146866165358,
      "loss": 12.5774,
      "step": 6185
    },
    {
      "epoch": 0.3368524228625105,
      "grad_norm": 0.7170587629888132,
      "learning_rate": 0.00019011764671696027,
      "loss": 12.4228,
      "step": 6186
    },
    {
      "epoch": 0.3369068768590935,
      "grad_norm": 0.7733647766370366,
      "learning_rate": 0.000190113824071497,
      "loss": 12.4979,
      "step": 6187
    },
    {
      "epoch": 0.3369613308556765,
      "grad_norm": 0.7267443064106669,
      "learning_rate": 0.00019011000072529348,
      "loss": 12.6086,
      "step": 6188
    },
    {
      "epoch": 0.3370157848522595,
      "grad_norm": 0.656912096319067,
      "learning_rate": 0.00019010617667837953,
      "loss": 12.4963,
      "step": 6189
    },
    {
      "epoch": 0.3370702388488425,
      "grad_norm": 0.7539748798624772,
      "learning_rate": 0.00019010235193078482,
      "loss": 12.5574,
      "step": 6190
    },
    {
      "epoch": 0.3371246928454255,
      "grad_norm": 0.7244926493537558,
      "learning_rate": 0.00019009852648253913,
      "loss": 12.6162,
      "step": 6191
    },
    {
      "epoch": 0.33717914684200856,
      "grad_norm": 0.6673239984305686,
      "learning_rate": 0.00019009470033367218,
      "loss": 12.4831,
      "step": 6192
    },
    {
      "epoch": 0.33723360083859155,
      "grad_norm": 0.6975327703524771,
      "learning_rate": 0.00019009087348421372,
      "loss": 12.5692,
      "step": 6193
    },
    {
      "epoch": 0.33728805483517454,
      "grad_norm": 0.7778432350781286,
      "learning_rate": 0.00019008704593419354,
      "loss": 12.5819,
      "step": 6194
    },
    {
      "epoch": 0.3373425088317576,
      "grad_norm": 0.6864969326899306,
      "learning_rate": 0.0001900832176836414,
      "loss": 12.4972,
      "step": 6195
    },
    {
      "epoch": 0.3373969628283406,
      "grad_norm": 0.6202551937185172,
      "learning_rate": 0.00019007938873258698,
      "loss": 12.5221,
      "step": 6196
    },
    {
      "epoch": 0.3374514168249236,
      "grad_norm": 0.6537274528177562,
      "learning_rate": 0.0001900755590810602,
      "loss": 12.5386,
      "step": 6197
    },
    {
      "epoch": 0.3375058708215066,
      "grad_norm": 0.7079124921438565,
      "learning_rate": 0.00019007172872909073,
      "loss": 12.5338,
      "step": 6198
    },
    {
      "epoch": 0.3375603248180896,
      "grad_norm": 0.6657694746805412,
      "learning_rate": 0.00019006789767670842,
      "loss": 12.4443,
      "step": 6199
    },
    {
      "epoch": 0.3376147788146726,
      "grad_norm": 0.730262940364132,
      "learning_rate": 0.000190064065923943,
      "loss": 12.5203,
      "step": 6200
    },
    {
      "epoch": 0.33766923281125566,
      "grad_norm": 0.630319901736221,
      "learning_rate": 0.0001900602334708243,
      "loss": 12.33,
      "step": 6201
    },
    {
      "epoch": 0.33772368680783865,
      "grad_norm": 0.7177306256126688,
      "learning_rate": 0.00019005640031738216,
      "loss": 12.6088,
      "step": 6202
    },
    {
      "epoch": 0.33777814080442164,
      "grad_norm": 0.7407616600834069,
      "learning_rate": 0.00019005256646364632,
      "loss": 12.551,
      "step": 6203
    },
    {
      "epoch": 0.3378325948010047,
      "grad_norm": 0.6405237750463625,
      "learning_rate": 0.00019004873190964664,
      "loss": 12.7041,
      "step": 6204
    },
    {
      "epoch": 0.3378870487975877,
      "grad_norm": 0.5858757409387871,
      "learning_rate": 0.0001900448966554129,
      "loss": 12.4482,
      "step": 6205
    },
    {
      "epoch": 0.3379415027941707,
      "grad_norm": 0.7294245367979105,
      "learning_rate": 0.00019004106070097496,
      "loss": 12.5935,
      "step": 6206
    },
    {
      "epoch": 0.3379959567907537,
      "grad_norm": 0.6463929455412016,
      "learning_rate": 0.0001900372240463626,
      "loss": 12.4734,
      "step": 6207
    },
    {
      "epoch": 0.3380504107873367,
      "grad_norm": 0.6904944812387054,
      "learning_rate": 0.0001900333866916057,
      "loss": 12.5885,
      "step": 6208
    },
    {
      "epoch": 0.3381048647839197,
      "grad_norm": 0.685170401259956,
      "learning_rate": 0.0001900295486367341,
      "loss": 12.5826,
      "step": 6209
    },
    {
      "epoch": 0.33815931878050276,
      "grad_norm": 0.6357849225796319,
      "learning_rate": 0.00019002570988177763,
      "loss": 12.5158,
      "step": 6210
    },
    {
      "epoch": 0.33821377277708575,
      "grad_norm": 0.6683542317848408,
      "learning_rate": 0.00019002187042676613,
      "loss": 12.5072,
      "step": 6211
    },
    {
      "epoch": 0.33826822677366875,
      "grad_norm": 0.6495159611040929,
      "learning_rate": 0.0001900180302717295,
      "loss": 12.5352,
      "step": 6212
    },
    {
      "epoch": 0.3383226807702518,
      "grad_norm": 0.6391516161165911,
      "learning_rate": 0.00019001418941669754,
      "loss": 12.5244,
      "step": 6213
    },
    {
      "epoch": 0.3383771347668348,
      "grad_norm": 0.7089891987518301,
      "learning_rate": 0.00019001034786170014,
      "loss": 12.8155,
      "step": 6214
    },
    {
      "epoch": 0.3384315887634178,
      "grad_norm": 0.7321755456555036,
      "learning_rate": 0.0001900065056067672,
      "loss": 12.6221,
      "step": 6215
    },
    {
      "epoch": 0.3384860427600008,
      "grad_norm": 0.7487920427898623,
      "learning_rate": 0.0001900026626519286,
      "loss": 12.4454,
      "step": 6216
    },
    {
      "epoch": 0.3385404967565838,
      "grad_norm": 0.7044242558236338,
      "learning_rate": 0.00018999881899721416,
      "loss": 12.5365,
      "step": 6217
    },
    {
      "epoch": 0.33859495075316687,
      "grad_norm": 0.6597639513487883,
      "learning_rate": 0.00018999497464265383,
      "loss": 12.5274,
      "step": 6218
    },
    {
      "epoch": 0.33864940474974986,
      "grad_norm": 0.76686473507453,
      "learning_rate": 0.00018999112958827748,
      "loss": 12.5673,
      "step": 6219
    },
    {
      "epoch": 0.33870385874633285,
      "grad_norm": 0.6161336047920143,
      "learning_rate": 0.000189987283834115,
      "loss": 12.5084,
      "step": 6220
    },
    {
      "epoch": 0.3387583127429159,
      "grad_norm": 0.8074532039395405,
      "learning_rate": 0.00018998343738019634,
      "loss": 12.4663,
      "step": 6221
    },
    {
      "epoch": 0.3388127667394989,
      "grad_norm": 0.6652786762546103,
      "learning_rate": 0.00018997959022655137,
      "loss": 12.4228,
      "step": 6222
    },
    {
      "epoch": 0.3388672207360819,
      "grad_norm": 0.7768352951927515,
      "learning_rate": 0.00018997574237321002,
      "loss": 12.5579,
      "step": 6223
    },
    {
      "epoch": 0.33892167473266493,
      "grad_norm": 0.688925320784393,
      "learning_rate": 0.0001899718938202022,
      "loss": 12.4842,
      "step": 6224
    },
    {
      "epoch": 0.3389761287292479,
      "grad_norm": 0.6166350003751984,
      "learning_rate": 0.00018996804456755784,
      "loss": 12.5745,
      "step": 6225
    },
    {
      "epoch": 0.3390305827258309,
      "grad_norm": 0.7968250998062604,
      "learning_rate": 0.00018996419461530687,
      "loss": 12.4601,
      "step": 6226
    },
    {
      "epoch": 0.33908503672241397,
      "grad_norm": 0.7204375657832244,
      "learning_rate": 0.00018996034396347929,
      "loss": 12.4444,
      "step": 6227
    },
    {
      "epoch": 0.33913949071899696,
      "grad_norm": 0.7435334241715121,
      "learning_rate": 0.00018995649261210497,
      "loss": 12.5888,
      "step": 6228
    },
    {
      "epoch": 0.33919394471557995,
      "grad_norm": 0.7132449890620907,
      "learning_rate": 0.0001899526405612138,
      "loss": 12.4273,
      "step": 6229
    },
    {
      "epoch": 0.339248398712163,
      "grad_norm": 0.647762025838811,
      "learning_rate": 0.00018994878781083589,
      "loss": 12.4589,
      "step": 6230
    },
    {
      "epoch": 0.339302852708746,
      "grad_norm": 0.8074144202446832,
      "learning_rate": 0.00018994493436100108,
      "loss": 12.5515,
      "step": 6231
    },
    {
      "epoch": 0.339357306705329,
      "grad_norm": 0.7421418526430787,
      "learning_rate": 0.00018994108021173943,
      "loss": 12.6725,
      "step": 6232
    },
    {
      "epoch": 0.33941176070191204,
      "grad_norm": 0.6315796994855236,
      "learning_rate": 0.0001899372253630808,
      "loss": 12.5219,
      "step": 6233
    },
    {
      "epoch": 0.33946621469849503,
      "grad_norm": 0.7867514236741496,
      "learning_rate": 0.00018993336981505527,
      "loss": 12.5488,
      "step": 6234
    },
    {
      "epoch": 0.339520668695078,
      "grad_norm": 0.6825545070067882,
      "learning_rate": 0.00018992951356769274,
      "loss": 12.6453,
      "step": 6235
    },
    {
      "epoch": 0.33957512269166107,
      "grad_norm": 0.5581581779692462,
      "learning_rate": 0.00018992565662102323,
      "loss": 12.4608,
      "step": 6236
    },
    {
      "epoch": 0.33962957668824406,
      "grad_norm": 0.6695781279945339,
      "learning_rate": 0.00018992179897507679,
      "loss": 12.5149,
      "step": 6237
    },
    {
      "epoch": 0.33968403068482705,
      "grad_norm": 0.7055852494170761,
      "learning_rate": 0.00018991794062988331,
      "loss": 12.5044,
      "step": 6238
    },
    {
      "epoch": 0.3397384846814101,
      "grad_norm": 0.7226116530536991,
      "learning_rate": 0.00018991408158547285,
      "loss": 12.4421,
      "step": 6239
    },
    {
      "epoch": 0.3397929386779931,
      "grad_norm": 0.7363087436848126,
      "learning_rate": 0.0001899102218418754,
      "loss": 12.499,
      "step": 6240
    },
    {
      "epoch": 0.3398473926745761,
      "grad_norm": 0.6300687166747003,
      "learning_rate": 0.00018990636139912102,
      "loss": 12.5233,
      "step": 6241
    },
    {
      "epoch": 0.33990184667115914,
      "grad_norm": 0.793004713938736,
      "learning_rate": 0.00018990250025723967,
      "loss": 12.5717,
      "step": 6242
    },
    {
      "epoch": 0.33995630066774213,
      "grad_norm": 0.6699161458232288,
      "learning_rate": 0.0001898986384162614,
      "loss": 12.5451,
      "step": 6243
    },
    {
      "epoch": 0.3400107546643251,
      "grad_norm": 0.7459639388831515,
      "learning_rate": 0.00018989477587621627,
      "loss": 12.5257,
      "step": 6244
    },
    {
      "epoch": 0.34006520866090817,
      "grad_norm": 0.7132170474264726,
      "learning_rate": 0.00018989091263713428,
      "loss": 12.5845,
      "step": 6245
    },
    {
      "epoch": 0.34011966265749116,
      "grad_norm": 0.6045019422774155,
      "learning_rate": 0.00018988704869904547,
      "loss": 12.3912,
      "step": 6246
    },
    {
      "epoch": 0.34017411665407415,
      "grad_norm": 0.666948712286019,
      "learning_rate": 0.0001898831840619799,
      "loss": 12.3408,
      "step": 6247
    },
    {
      "epoch": 0.3402285706506572,
      "grad_norm": 0.7105971968384324,
      "learning_rate": 0.0001898793187259676,
      "loss": 12.5605,
      "step": 6248
    },
    {
      "epoch": 0.3402830246472402,
      "grad_norm": 0.628138207855354,
      "learning_rate": 0.00018987545269103865,
      "loss": 12.46,
      "step": 6249
    },
    {
      "epoch": 0.3403374786438232,
      "grad_norm": 0.7078715466744697,
      "learning_rate": 0.00018987158595722313,
      "loss": 12.5245,
      "step": 6250
    },
    {
      "epoch": 0.34039193264040624,
      "grad_norm": 0.659737373649185,
      "learning_rate": 0.00018986771852455109,
      "loss": 12.604,
      "step": 6251
    },
    {
      "epoch": 0.34044638663698923,
      "grad_norm": 0.7146806557781517,
      "learning_rate": 0.00018986385039305255,
      "loss": 12.5897,
      "step": 6252
    },
    {
      "epoch": 0.3405008406335723,
      "grad_norm": 0.6435362637324684,
      "learning_rate": 0.00018985998156275765,
      "loss": 12.5073,
      "step": 6253
    },
    {
      "epoch": 0.34055529463015527,
      "grad_norm": 0.6518305515914652,
      "learning_rate": 0.00018985611203369652,
      "loss": 12.5177,
      "step": 6254
    },
    {
      "epoch": 0.34060974862673826,
      "grad_norm": 0.7293022805036647,
      "learning_rate": 0.00018985224180589913,
      "loss": 12.3901,
      "step": 6255
    },
    {
      "epoch": 0.3406642026233213,
      "grad_norm": 0.7226616833868063,
      "learning_rate": 0.00018984837087939567,
      "loss": 12.5533,
      "step": 6256
    },
    {
      "epoch": 0.3407186566199043,
      "grad_norm": 0.6316932747556198,
      "learning_rate": 0.00018984449925421622,
      "loss": 12.2969,
      "step": 6257
    },
    {
      "epoch": 0.3407731106164873,
      "grad_norm": 0.650964042997637,
      "learning_rate": 0.00018984062693039086,
      "loss": 12.545,
      "step": 6258
    },
    {
      "epoch": 0.34082756461307034,
      "grad_norm": 0.6313622298812942,
      "learning_rate": 0.00018983675390794972,
      "loss": 12.5437,
      "step": 6259
    },
    {
      "epoch": 0.34088201860965334,
      "grad_norm": 0.6856129479064164,
      "learning_rate": 0.0001898328801869229,
      "loss": 12.5644,
      "step": 6260
    },
    {
      "epoch": 0.34093647260623633,
      "grad_norm": 0.6754733609937804,
      "learning_rate": 0.0001898290057673406,
      "loss": 12.5439,
      "step": 6261
    },
    {
      "epoch": 0.3409909266028194,
      "grad_norm": 0.6175904268818588,
      "learning_rate": 0.00018982513064923283,
      "loss": 12.4148,
      "step": 6262
    },
    {
      "epoch": 0.34104538059940237,
      "grad_norm": 0.7029133910496881,
      "learning_rate": 0.00018982125483262978,
      "loss": 12.6239,
      "step": 6263
    },
    {
      "epoch": 0.34109983459598536,
      "grad_norm": 0.6764524896672984,
      "learning_rate": 0.0001898173783175616,
      "loss": 12.4716,
      "step": 6264
    },
    {
      "epoch": 0.3411542885925684,
      "grad_norm": 0.6473075046714889,
      "learning_rate": 0.00018981350110405844,
      "loss": 12.3433,
      "step": 6265
    },
    {
      "epoch": 0.3412087425891514,
      "grad_norm": 0.7340041196610504,
      "learning_rate": 0.0001898096231921504,
      "loss": 12.6536,
      "step": 6266
    },
    {
      "epoch": 0.3412631965857344,
      "grad_norm": 0.6704673154087119,
      "learning_rate": 0.00018980574458186774,
      "loss": 12.5468,
      "step": 6267
    },
    {
      "epoch": 0.34131765058231744,
      "grad_norm": 0.6352924231527934,
      "learning_rate": 0.00018980186527324048,
      "loss": 12.5107,
      "step": 6268
    },
    {
      "epoch": 0.34137210457890044,
      "grad_norm": 0.6933220771845349,
      "learning_rate": 0.00018979798526629887,
      "loss": 12.4702,
      "step": 6269
    },
    {
      "epoch": 0.34142655857548343,
      "grad_norm": 0.6339519467155723,
      "learning_rate": 0.00018979410456107306,
      "loss": 12.3719,
      "step": 6270
    },
    {
      "epoch": 0.3414810125720665,
      "grad_norm": 0.7721477942860392,
      "learning_rate": 0.00018979022315759325,
      "loss": 12.6534,
      "step": 6271
    },
    {
      "epoch": 0.34153546656864947,
      "grad_norm": 0.6153794386548143,
      "learning_rate": 0.00018978634105588961,
      "loss": 12.574,
      "step": 6272
    },
    {
      "epoch": 0.34158992056523246,
      "grad_norm": 0.6769230590319805,
      "learning_rate": 0.00018978245825599234,
      "loss": 12.4373,
      "step": 6273
    },
    {
      "epoch": 0.3416443745618155,
      "grad_norm": 0.8119995792072666,
      "learning_rate": 0.00018977857475793158,
      "loss": 12.5873,
      "step": 6274
    },
    {
      "epoch": 0.3416988285583985,
      "grad_norm": 0.6749571198734036,
      "learning_rate": 0.0001897746905617376,
      "loss": 12.6444,
      "step": 6275
    },
    {
      "epoch": 0.3417532825549815,
      "grad_norm": 0.7317901747313105,
      "learning_rate": 0.00018977080566744055,
      "loss": 12.469,
      "step": 6276
    },
    {
      "epoch": 0.34180773655156454,
      "grad_norm": 0.635670753116196,
      "learning_rate": 0.00018976692007507067,
      "loss": 12.3512,
      "step": 6277
    },
    {
      "epoch": 0.34186219054814754,
      "grad_norm": 0.7243933147841601,
      "learning_rate": 0.00018976303378465814,
      "loss": 12.5583,
      "step": 6278
    },
    {
      "epoch": 0.34191664454473053,
      "grad_norm": 0.6084842547409487,
      "learning_rate": 0.00018975914679623325,
      "loss": 12.538,
      "step": 6279
    },
    {
      "epoch": 0.3419710985413136,
      "grad_norm": 0.7593564909409162,
      "learning_rate": 0.00018975525910982615,
      "loss": 12.6837,
      "step": 6280
    },
    {
      "epoch": 0.34202555253789657,
      "grad_norm": 0.636182795196838,
      "learning_rate": 0.0001897513707254671,
      "loss": 12.5124,
      "step": 6281
    },
    {
      "epoch": 0.34208000653447956,
      "grad_norm": 0.8155321522237161,
      "learning_rate": 0.00018974748164318636,
      "loss": 12.5131,
      "step": 6282
    },
    {
      "epoch": 0.3421344605310626,
      "grad_norm": 0.6426416580788399,
      "learning_rate": 0.00018974359186301417,
      "loss": 12.5234,
      "step": 6283
    },
    {
      "epoch": 0.3421889145276456,
      "grad_norm": 0.8193055100539444,
      "learning_rate": 0.00018973970138498071,
      "loss": 12.5375,
      "step": 6284
    },
    {
      "epoch": 0.34224336852422865,
      "grad_norm": 0.7185734747917539,
      "learning_rate": 0.00018973581020911634,
      "loss": 12.4833,
      "step": 6285
    },
    {
      "epoch": 0.34229782252081165,
      "grad_norm": 0.6288019118690862,
      "learning_rate": 0.0001897319183354512,
      "loss": 12.6266,
      "step": 6286
    },
    {
      "epoch": 0.34235227651739464,
      "grad_norm": 0.7132135055944064,
      "learning_rate": 0.00018972802576401566,
      "loss": 12.6449,
      "step": 6287
    },
    {
      "epoch": 0.3424067305139777,
      "grad_norm": 0.5785610189135727,
      "learning_rate": 0.00018972413249483992,
      "loss": 12.4041,
      "step": 6288
    },
    {
      "epoch": 0.3424611845105607,
      "grad_norm": 0.703198984063344,
      "learning_rate": 0.00018972023852795427,
      "loss": 12.5444,
      "step": 6289
    },
    {
      "epoch": 0.34251563850714367,
      "grad_norm": 0.668041192115363,
      "learning_rate": 0.000189716343863389,
      "loss": 12.5276,
      "step": 6290
    },
    {
      "epoch": 0.3425700925037267,
      "grad_norm": 0.7179533760536456,
      "learning_rate": 0.00018971244850117443,
      "loss": 12.4907,
      "step": 6291
    },
    {
      "epoch": 0.3426245465003097,
      "grad_norm": 0.6547689597350955,
      "learning_rate": 0.0001897085524413408,
      "loss": 12.5585,
      "step": 6292
    },
    {
      "epoch": 0.3426790004968927,
      "grad_norm": 0.6892810246093969,
      "learning_rate": 0.0001897046556839184,
      "loss": 12.6043,
      "step": 6293
    },
    {
      "epoch": 0.34273345449347575,
      "grad_norm": 0.6293810182504478,
      "learning_rate": 0.0001897007582289376,
      "loss": 12.5168,
      "step": 6294
    },
    {
      "epoch": 0.34278790849005875,
      "grad_norm": 0.6465223613402978,
      "learning_rate": 0.0001896968600764286,
      "loss": 12.5322,
      "step": 6295
    },
    {
      "epoch": 0.34284236248664174,
      "grad_norm": 0.6450240234888404,
      "learning_rate": 0.00018969296122642185,
      "loss": 12.5049,
      "step": 6296
    },
    {
      "epoch": 0.3428968164832248,
      "grad_norm": 0.700539836061821,
      "learning_rate": 0.00018968906167894753,
      "loss": 12.5086,
      "step": 6297
    },
    {
      "epoch": 0.3429512704798078,
      "grad_norm": 0.6584639102315182,
      "learning_rate": 0.00018968516143403604,
      "loss": 12.4562,
      "step": 6298
    },
    {
      "epoch": 0.34300572447639077,
      "grad_norm": 0.6823128388229975,
      "learning_rate": 0.00018968126049171772,
      "loss": 12.5732,
      "step": 6299
    },
    {
      "epoch": 0.3430601784729738,
      "grad_norm": 0.6786200126620332,
      "learning_rate": 0.00018967735885202285,
      "loss": 12.353,
      "step": 6300
    },
    {
      "epoch": 0.3431146324695568,
      "grad_norm": 0.7928388839289368,
      "learning_rate": 0.0001896734565149818,
      "loss": 12.5546,
      "step": 6301
    },
    {
      "epoch": 0.3431690864661398,
      "grad_norm": 0.8269561615090272,
      "learning_rate": 0.00018966955348062494,
      "loss": 12.4672,
      "step": 6302
    },
    {
      "epoch": 0.34322354046272285,
      "grad_norm": 0.6666521103507326,
      "learning_rate": 0.00018966564974898256,
      "loss": 12.5375,
      "step": 6303
    },
    {
      "epoch": 0.34327799445930585,
      "grad_norm": 0.6701268520745489,
      "learning_rate": 0.00018966174532008507,
      "loss": 12.6131,
      "step": 6304
    },
    {
      "epoch": 0.34333244845588884,
      "grad_norm": 0.6766200647318046,
      "learning_rate": 0.0001896578401939628,
      "loss": 12.5149,
      "step": 6305
    },
    {
      "epoch": 0.3433869024524719,
      "grad_norm": 0.6599092895292733,
      "learning_rate": 0.00018965393437064614,
      "loss": 12.4065,
      "step": 6306
    },
    {
      "epoch": 0.3434413564490549,
      "grad_norm": 0.6659625059287979,
      "learning_rate": 0.00018965002785016543,
      "loss": 12.5598,
      "step": 6307
    },
    {
      "epoch": 0.3434958104456379,
      "grad_norm": 0.8131428911036774,
      "learning_rate": 0.0001896461206325511,
      "loss": 12.6066,
      "step": 6308
    },
    {
      "epoch": 0.3435502644422209,
      "grad_norm": 0.6293054241618713,
      "learning_rate": 0.00018964221271783349,
      "loss": 12.5573,
      "step": 6309
    },
    {
      "epoch": 0.3436047184388039,
      "grad_norm": 0.618067658836677,
      "learning_rate": 0.000189638304106043,
      "loss": 12.4552,
      "step": 6310
    },
    {
      "epoch": 0.3436591724353869,
      "grad_norm": 0.8179694432192177,
      "learning_rate": 0.00018963439479721003,
      "loss": 12.5569,
      "step": 6311
    },
    {
      "epoch": 0.34371362643196995,
      "grad_norm": 0.6206028828875811,
      "learning_rate": 0.00018963048479136497,
      "loss": 12.5951,
      "step": 6312
    },
    {
      "epoch": 0.34376808042855295,
      "grad_norm": 0.662690160695336,
      "learning_rate": 0.0001896265740885382,
      "loss": 12.555,
      "step": 6313
    },
    {
      "epoch": 0.34382253442513594,
      "grad_norm": 0.692073058932196,
      "learning_rate": 0.0001896226626887602,
      "loss": 12.4739,
      "step": 6314
    },
    {
      "epoch": 0.343876988421719,
      "grad_norm": 0.6636264210544816,
      "learning_rate": 0.00018961875059206136,
      "loss": 12.6116,
      "step": 6315
    },
    {
      "epoch": 0.343931442418302,
      "grad_norm": 0.7960881435996826,
      "learning_rate": 0.00018961483779847204,
      "loss": 12.5392,
      "step": 6316
    },
    {
      "epoch": 0.343985896414885,
      "grad_norm": 0.7024499992984288,
      "learning_rate": 0.00018961092430802275,
      "loss": 12.5088,
      "step": 6317
    },
    {
      "epoch": 0.344040350411468,
      "grad_norm": 0.5998701393444636,
      "learning_rate": 0.00018960701012074387,
      "loss": 12.4843,
      "step": 6318
    },
    {
      "epoch": 0.344094804408051,
      "grad_norm": 0.6386075283677938,
      "learning_rate": 0.00018960309523666585,
      "loss": 12.5588,
      "step": 6319
    },
    {
      "epoch": 0.34414925840463406,
      "grad_norm": 0.6187861240384069,
      "learning_rate": 0.00018959917965581912,
      "loss": 12.5536,
      "step": 6320
    },
    {
      "epoch": 0.34420371240121705,
      "grad_norm": 0.700612269892657,
      "learning_rate": 0.00018959526337823416,
      "loss": 12.5407,
      "step": 6321
    },
    {
      "epoch": 0.34425816639780005,
      "grad_norm": 0.713290950613392,
      "learning_rate": 0.00018959134640394141,
      "loss": 12.5776,
      "step": 6322
    },
    {
      "epoch": 0.3443126203943831,
      "grad_norm": 0.6599590469218795,
      "learning_rate": 0.0001895874287329713,
      "loss": 12.646,
      "step": 6323
    },
    {
      "epoch": 0.3443670743909661,
      "grad_norm": 0.7129764251208779,
      "learning_rate": 0.00018958351036535437,
      "loss": 12.5006,
      "step": 6324
    },
    {
      "epoch": 0.3444215283875491,
      "grad_norm": 0.6853621732756515,
      "learning_rate": 0.000189579591301121,
      "loss": 12.4555,
      "step": 6325
    },
    {
      "epoch": 0.34447598238413213,
      "grad_norm": 0.6806288970030699,
      "learning_rate": 0.00018957567154030173,
      "loss": 12.5047,
      "step": 6326
    },
    {
      "epoch": 0.3445304363807151,
      "grad_norm": 0.694590261163472,
      "learning_rate": 0.000189571751082927,
      "loss": 12.6542,
      "step": 6327
    },
    {
      "epoch": 0.3445848903772981,
      "grad_norm": 0.6753126132952569,
      "learning_rate": 0.0001895678299290273,
      "loss": 12.4212,
      "step": 6328
    },
    {
      "epoch": 0.34463934437388116,
      "grad_norm": 0.7192878280330849,
      "learning_rate": 0.00018956390807863316,
      "loss": 12.3633,
      "step": 6329
    },
    {
      "epoch": 0.34469379837046416,
      "grad_norm": 0.7326573316491729,
      "learning_rate": 0.00018955998553177504,
      "loss": 12.5626,
      "step": 6330
    },
    {
      "epoch": 0.34474825236704715,
      "grad_norm": 0.5918551403736377,
      "learning_rate": 0.00018955606228848347,
      "loss": 12.4205,
      "step": 6331
    },
    {
      "epoch": 0.3448027063636302,
      "grad_norm": 0.7891637583902474,
      "learning_rate": 0.00018955213834878892,
      "loss": 12.5029,
      "step": 6332
    },
    {
      "epoch": 0.3448571603602132,
      "grad_norm": 0.5980834101002779,
      "learning_rate": 0.00018954821371272194,
      "loss": 12.4273,
      "step": 6333
    },
    {
      "epoch": 0.3449116143567962,
      "grad_norm": 0.8125656724378687,
      "learning_rate": 0.000189544288380313,
      "loss": 12.5797,
      "step": 6334
    },
    {
      "epoch": 0.34496606835337923,
      "grad_norm": 0.7509595238380199,
      "learning_rate": 0.0001895403623515927,
      "loss": 12.5089,
      "step": 6335
    },
    {
      "epoch": 0.3450205223499622,
      "grad_norm": 0.7248800926147912,
      "learning_rate": 0.0001895364356265915,
      "loss": 12.5702,
      "step": 6336
    },
    {
      "epoch": 0.3450749763465452,
      "grad_norm": 0.8055583215818384,
      "learning_rate": 0.00018953250820533994,
      "loss": 12.5792,
      "step": 6337
    },
    {
      "epoch": 0.34512943034312826,
      "grad_norm": 0.5895339977865024,
      "learning_rate": 0.00018952858008786861,
      "loss": 12.5369,
      "step": 6338
    },
    {
      "epoch": 0.34518388433971126,
      "grad_norm": 0.7844342988424144,
      "learning_rate": 0.000189524651274208,
      "loss": 12.6507,
      "step": 6339
    },
    {
      "epoch": 0.34523833833629425,
      "grad_norm": 0.641138721002702,
      "learning_rate": 0.00018952072176438875,
      "loss": 12.473,
      "step": 6340
    },
    {
      "epoch": 0.3452927923328773,
      "grad_norm": 0.7272744757161392,
      "learning_rate": 0.0001895167915584413,
      "loss": 12.5863,
      "step": 6341
    },
    {
      "epoch": 0.3453472463294603,
      "grad_norm": 0.7097480150671507,
      "learning_rate": 0.0001895128606563963,
      "loss": 12.6004,
      "step": 6342
    },
    {
      "epoch": 0.3454017003260433,
      "grad_norm": 0.6246608664740043,
      "learning_rate": 0.0001895089290582843,
      "loss": 12.4141,
      "step": 6343
    },
    {
      "epoch": 0.34545615432262633,
      "grad_norm": 0.6930736444561254,
      "learning_rate": 0.0001895049967641358,
      "loss": 12.4728,
      "step": 6344
    },
    {
      "epoch": 0.3455106083192093,
      "grad_norm": 0.6358442922013844,
      "learning_rate": 0.00018950106377398147,
      "loss": 12.601,
      "step": 6345
    },
    {
      "epoch": 0.3455650623157923,
      "grad_norm": 0.6550122925149894,
      "learning_rate": 0.00018949713008785187,
      "loss": 12.6569,
      "step": 6346
    },
    {
      "epoch": 0.34561951631237536,
      "grad_norm": 0.6239208208894378,
      "learning_rate": 0.00018949319570577756,
      "loss": 12.5735,
      "step": 6347
    },
    {
      "epoch": 0.34567397030895836,
      "grad_norm": 0.6506837444783518,
      "learning_rate": 0.0001894892606277891,
      "loss": 12.5761,
      "step": 6348
    },
    {
      "epoch": 0.34572842430554135,
      "grad_norm": 0.6475058151362632,
      "learning_rate": 0.00018948532485391724,
      "loss": 12.516,
      "step": 6349
    },
    {
      "epoch": 0.3457828783021244,
      "grad_norm": 0.6793182543915407,
      "learning_rate": 0.00018948138838419243,
      "loss": 12.3702,
      "step": 6350
    },
    {
      "epoch": 0.3458373322987074,
      "grad_norm": 0.6628263555202152,
      "learning_rate": 0.00018947745121864534,
      "loss": 12.4945,
      "step": 6351
    },
    {
      "epoch": 0.34589178629529044,
      "grad_norm": 0.6251765624796839,
      "learning_rate": 0.0001894735133573066,
      "loss": 12.5348,
      "step": 6352
    },
    {
      "epoch": 0.34594624029187343,
      "grad_norm": 0.6143638650692151,
      "learning_rate": 0.0001894695748002068,
      "loss": 12.4181,
      "step": 6353
    },
    {
      "epoch": 0.3460006942884564,
      "grad_norm": 0.6529588496376133,
      "learning_rate": 0.0001894656355473766,
      "loss": 12.5347,
      "step": 6354
    },
    {
      "epoch": 0.34605514828503947,
      "grad_norm": 0.6120964415426231,
      "learning_rate": 0.0001894616955988466,
      "loss": 12.5119,
      "step": 6355
    },
    {
      "epoch": 0.34610960228162246,
      "grad_norm": 0.7211947140757892,
      "learning_rate": 0.00018945775495464746,
      "loss": 12.5204,
      "step": 6356
    },
    {
      "epoch": 0.34616405627820546,
      "grad_norm": 0.7094435979733846,
      "learning_rate": 0.0001894538136148098,
      "loss": 12.6202,
      "step": 6357
    },
    {
      "epoch": 0.3462185102747885,
      "grad_norm": 0.7143215980668297,
      "learning_rate": 0.00018944987157936433,
      "loss": 12.5306,
      "step": 6358
    },
    {
      "epoch": 0.3462729642713715,
      "grad_norm": 0.594273115652585,
      "learning_rate": 0.00018944592884834158,
      "loss": 12.5041,
      "step": 6359
    },
    {
      "epoch": 0.3463274182679545,
      "grad_norm": 0.6021587890983173,
      "learning_rate": 0.00018944198542177233,
      "loss": 12.4764,
      "step": 6360
    },
    {
      "epoch": 0.34638187226453754,
      "grad_norm": 0.6925011127854152,
      "learning_rate": 0.00018943804129968722,
      "loss": 12.4984,
      "step": 6361
    },
    {
      "epoch": 0.34643632626112053,
      "grad_norm": 0.6341869152621732,
      "learning_rate": 0.00018943409648211688,
      "loss": 12.5632,
      "step": 6362
    },
    {
      "epoch": 0.3464907802577035,
      "grad_norm": 0.6482645542487078,
      "learning_rate": 0.00018943015096909203,
      "loss": 12.539,
      "step": 6363
    },
    {
      "epoch": 0.34654523425428657,
      "grad_norm": 1.0140057634003605,
      "learning_rate": 0.0001894262047606433,
      "loss": 12.5651,
      "step": 6364
    },
    {
      "epoch": 0.34659968825086956,
      "grad_norm": 0.7749586692842871,
      "learning_rate": 0.0001894222578568014,
      "loss": 12.6487,
      "step": 6365
    },
    {
      "epoch": 0.34665414224745256,
      "grad_norm": 0.7511098701115632,
      "learning_rate": 0.00018941831025759705,
      "loss": 12.5013,
      "step": 6366
    },
    {
      "epoch": 0.3467085962440356,
      "grad_norm": 0.6900038024918674,
      "learning_rate": 0.00018941436196306092,
      "loss": 12.4541,
      "step": 6367
    },
    {
      "epoch": 0.3467630502406186,
      "grad_norm": 0.649129436938782,
      "learning_rate": 0.0001894104129732237,
      "loss": 12.5152,
      "step": 6368
    },
    {
      "epoch": 0.3468175042372016,
      "grad_norm": 0.6369552672624779,
      "learning_rate": 0.00018940646328811616,
      "loss": 12.4275,
      "step": 6369
    },
    {
      "epoch": 0.34687195823378464,
      "grad_norm": 0.6559936737005914,
      "learning_rate": 0.0001894025129077689,
      "loss": 12.5292,
      "step": 6370
    },
    {
      "epoch": 0.34692641223036763,
      "grad_norm": 0.6739824166369338,
      "learning_rate": 0.00018939856183221277,
      "loss": 12.5955,
      "step": 6371
    },
    {
      "epoch": 0.3469808662269506,
      "grad_norm": 0.6310087891162702,
      "learning_rate": 0.0001893946100614784,
      "loss": 12.5557,
      "step": 6372
    },
    {
      "epoch": 0.34703532022353367,
      "grad_norm": 0.7462373695583338,
      "learning_rate": 0.00018939065759559655,
      "loss": 12.498,
      "step": 6373
    },
    {
      "epoch": 0.34708977422011666,
      "grad_norm": 0.6208249439174864,
      "learning_rate": 0.00018938670443459797,
      "loss": 12.535,
      "step": 6374
    },
    {
      "epoch": 0.34714422821669966,
      "grad_norm": 0.6676441700139456,
      "learning_rate": 0.0001893827505785134,
      "loss": 12.5842,
      "step": 6375
    },
    {
      "epoch": 0.3471986822132827,
      "grad_norm": 0.6282155039035795,
      "learning_rate": 0.00018937879602737352,
      "loss": 12.5026,
      "step": 6376
    },
    {
      "epoch": 0.3472531362098657,
      "grad_norm": 0.7052680426249938,
      "learning_rate": 0.00018937484078120916,
      "loss": 12.4728,
      "step": 6377
    },
    {
      "epoch": 0.3473075902064487,
      "grad_norm": 0.6920639408081833,
      "learning_rate": 0.00018937088484005107,
      "loss": 12.5618,
      "step": 6378
    },
    {
      "epoch": 0.34736204420303174,
      "grad_norm": 0.6593306649432598,
      "learning_rate": 0.00018936692820392995,
      "loss": 12.5355,
      "step": 6379
    },
    {
      "epoch": 0.34741649819961473,
      "grad_norm": 0.6217853510305813,
      "learning_rate": 0.00018936297087287663,
      "loss": 12.4688,
      "step": 6380
    },
    {
      "epoch": 0.3474709521961977,
      "grad_norm": 0.7318749515701808,
      "learning_rate": 0.00018935901284692188,
      "loss": 12.475,
      "step": 6381
    },
    {
      "epoch": 0.3475254061927808,
      "grad_norm": 0.711444766056599,
      "learning_rate": 0.00018935505412609645,
      "loss": 12.6268,
      "step": 6382
    },
    {
      "epoch": 0.34757986018936377,
      "grad_norm": 0.7406652625637992,
      "learning_rate": 0.0001893510947104311,
      "loss": 12.5164,
      "step": 6383
    },
    {
      "epoch": 0.34763431418594676,
      "grad_norm": 0.7130865533038795,
      "learning_rate": 0.0001893471345999567,
      "loss": 12.5484,
      "step": 6384
    },
    {
      "epoch": 0.3476887681825298,
      "grad_norm": 0.6612167121536203,
      "learning_rate": 0.000189343173794704,
      "loss": 12.4729,
      "step": 6385
    },
    {
      "epoch": 0.3477432221791128,
      "grad_norm": 0.6065860871835124,
      "learning_rate": 0.00018933921229470375,
      "loss": 12.5035,
      "step": 6386
    },
    {
      "epoch": 0.34779767617569585,
      "grad_norm": 0.691402387102845,
      "learning_rate": 0.00018933525009998684,
      "loss": 12.5822,
      "step": 6387
    },
    {
      "epoch": 0.34785213017227884,
      "grad_norm": 0.7433675288628595,
      "learning_rate": 0.00018933128721058403,
      "loss": 12.6504,
      "step": 6388
    },
    {
      "epoch": 0.34790658416886183,
      "grad_norm": 0.6196010968741086,
      "learning_rate": 0.00018932732362652617,
      "loss": 12.4233,
      "step": 6389
    },
    {
      "epoch": 0.3479610381654449,
      "grad_norm": 0.668475042083779,
      "learning_rate": 0.00018932335934784407,
      "loss": 12.5057,
      "step": 6390
    },
    {
      "epoch": 0.3480154921620279,
      "grad_norm": 0.9062156252444816,
      "learning_rate": 0.0001893193943745685,
      "loss": 12.5867,
      "step": 6391
    },
    {
      "epoch": 0.34806994615861087,
      "grad_norm": 0.6297267785296019,
      "learning_rate": 0.0001893154287067304,
      "loss": 12.5889,
      "step": 6392
    },
    {
      "epoch": 0.3481244001551939,
      "grad_norm": 0.6707450566245595,
      "learning_rate": 0.00018931146234436047,
      "loss": 12.457,
      "step": 6393
    },
    {
      "epoch": 0.3481788541517769,
      "grad_norm": 0.7407808248953949,
      "learning_rate": 0.00018930749528748967,
      "loss": 12.5703,
      "step": 6394
    },
    {
      "epoch": 0.3482333081483599,
      "grad_norm": 0.679668366691635,
      "learning_rate": 0.0001893035275361488,
      "loss": 12.5665,
      "step": 6395
    },
    {
      "epoch": 0.34828776214494295,
      "grad_norm": 0.6529951348154104,
      "learning_rate": 0.0001892995590903688,
      "loss": 12.5542,
      "step": 6396
    },
    {
      "epoch": 0.34834221614152594,
      "grad_norm": 0.6383316905634563,
      "learning_rate": 0.00018929558995018036,
      "loss": 12.5076,
      "step": 6397
    },
    {
      "epoch": 0.34839667013810893,
      "grad_norm": 0.7291276863483566,
      "learning_rate": 0.00018929162011561447,
      "loss": 12.5493,
      "step": 6398
    },
    {
      "epoch": 0.348451124134692,
      "grad_norm": 0.6492707381007455,
      "learning_rate": 0.00018928764958670198,
      "loss": 12.4822,
      "step": 6399
    },
    {
      "epoch": 0.348505578131275,
      "grad_norm": 0.7079173277913959,
      "learning_rate": 0.00018928367836347373,
      "loss": 12.5809,
      "step": 6400
    },
    {
      "epoch": 0.34856003212785797,
      "grad_norm": 0.750570468667512,
      "learning_rate": 0.00018927970644596064,
      "loss": 12.481,
      "step": 6401
    },
    {
      "epoch": 0.348614486124441,
      "grad_norm": 0.6244912076523662,
      "learning_rate": 0.00018927573383419356,
      "loss": 12.4996,
      "step": 6402
    },
    {
      "epoch": 0.348668940121024,
      "grad_norm": 0.7211818342696633,
      "learning_rate": 0.0001892717605282034,
      "loss": 12.5767,
      "step": 6403
    },
    {
      "epoch": 0.348723394117607,
      "grad_norm": 0.6794779875901277,
      "learning_rate": 0.00018926778652802111,
      "loss": 12.458,
      "step": 6404
    },
    {
      "epoch": 0.34877784811419005,
      "grad_norm": 0.654629105489235,
      "learning_rate": 0.0001892638118336775,
      "loss": 12.5111,
      "step": 6405
    },
    {
      "epoch": 0.34883230211077304,
      "grad_norm": 0.6827810366255843,
      "learning_rate": 0.00018925983644520352,
      "loss": 12.563,
      "step": 6406
    },
    {
      "epoch": 0.34888675610735603,
      "grad_norm": 0.6691807604313643,
      "learning_rate": 0.0001892558603626301,
      "loss": 12.608,
      "step": 6407
    },
    {
      "epoch": 0.3489412101039391,
      "grad_norm": 0.6556237166333346,
      "learning_rate": 0.00018925188358598813,
      "loss": 12.499,
      "step": 6408
    },
    {
      "epoch": 0.3489956641005221,
      "grad_norm": 0.669163097908646,
      "learning_rate": 0.00018924790611530857,
      "loss": 12.563,
      "step": 6409
    },
    {
      "epoch": 0.34905011809710507,
      "grad_norm": 0.6070951405750545,
      "learning_rate": 0.00018924392795062226,
      "loss": 12.3274,
      "step": 6410
    },
    {
      "epoch": 0.3491045720936881,
      "grad_norm": 0.7249396675579225,
      "learning_rate": 0.00018923994909196025,
      "loss": 12.475,
      "step": 6411
    },
    {
      "epoch": 0.3491590260902711,
      "grad_norm": 0.7266796712993444,
      "learning_rate": 0.00018923596953935342,
      "loss": 12.6071,
      "step": 6412
    },
    {
      "epoch": 0.3492134800868541,
      "grad_norm": 0.6490803692654883,
      "learning_rate": 0.00018923198929283276,
      "loss": 12.5236,
      "step": 6413
    },
    {
      "epoch": 0.34926793408343715,
      "grad_norm": 0.741220300259277,
      "learning_rate": 0.00018922800835242915,
      "loss": 12.635,
      "step": 6414
    },
    {
      "epoch": 0.34932238808002014,
      "grad_norm": 0.7445484770609918,
      "learning_rate": 0.0001892240267181736,
      "loss": 12.6129,
      "step": 6415
    },
    {
      "epoch": 0.34937684207660313,
      "grad_norm": 0.6759528956057483,
      "learning_rate": 0.00018922004439009702,
      "loss": 12.4762,
      "step": 6416
    },
    {
      "epoch": 0.3494312960731862,
      "grad_norm": 0.6745366120055789,
      "learning_rate": 0.00018921606136823046,
      "loss": 12.6715,
      "step": 6417
    },
    {
      "epoch": 0.3494857500697692,
      "grad_norm": 0.8237305176788643,
      "learning_rate": 0.00018921207765260482,
      "loss": 12.5819,
      "step": 6418
    },
    {
      "epoch": 0.3495402040663522,
      "grad_norm": 0.7112539950806442,
      "learning_rate": 0.00018920809324325107,
      "loss": 12.5541,
      "step": 6419
    },
    {
      "epoch": 0.3495946580629352,
      "grad_norm": 0.6667071155633175,
      "learning_rate": 0.0001892041081402003,
      "loss": 12.5276,
      "step": 6420
    },
    {
      "epoch": 0.3496491120595182,
      "grad_norm": 0.6273084531789137,
      "learning_rate": 0.0001892001223434834,
      "loss": 12.5549,
      "step": 6421
    },
    {
      "epoch": 0.34970356605610126,
      "grad_norm": 0.6750038564921762,
      "learning_rate": 0.00018919613585313135,
      "loss": 12.4548,
      "step": 6422
    },
    {
      "epoch": 0.34975802005268425,
      "grad_norm": 0.6252492245083748,
      "learning_rate": 0.00018919214866917522,
      "loss": 12.4935,
      "step": 6423
    },
    {
      "epoch": 0.34981247404926724,
      "grad_norm": 0.6984788655048768,
      "learning_rate": 0.000189188160791646,
      "loss": 12.5693,
      "step": 6424
    },
    {
      "epoch": 0.3498669280458503,
      "grad_norm": 0.6671234434806845,
      "learning_rate": 0.00018918417222057467,
      "loss": 12.4412,
      "step": 6425
    },
    {
      "epoch": 0.3499213820424333,
      "grad_norm": 0.7258796397397406,
      "learning_rate": 0.00018918018295599224,
      "loss": 12.489,
      "step": 6426
    },
    {
      "epoch": 0.3499758360390163,
      "grad_norm": 0.6848281360826799,
      "learning_rate": 0.00018917619299792978,
      "loss": 12.555,
      "step": 6427
    },
    {
      "epoch": 0.3500302900355993,
      "grad_norm": 0.6007537268845069,
      "learning_rate": 0.00018917220234641828,
      "loss": 12.4034,
      "step": 6428
    },
    {
      "epoch": 0.3500847440321823,
      "grad_norm": 0.6459033432263563,
      "learning_rate": 0.00018916821100148877,
      "loss": 12.5938,
      "step": 6429
    },
    {
      "epoch": 0.3501391980287653,
      "grad_norm": 0.6104972833604558,
      "learning_rate": 0.00018916421896317232,
      "loss": 12.5256,
      "step": 6430
    },
    {
      "epoch": 0.35019365202534836,
      "grad_norm": 0.724356568467194,
      "learning_rate": 0.00018916022623149994,
      "loss": 12.5068,
      "step": 6431
    },
    {
      "epoch": 0.35024810602193135,
      "grad_norm": 0.6684551809083941,
      "learning_rate": 0.00018915623280650268,
      "loss": 12.4023,
      "step": 6432
    },
    {
      "epoch": 0.35030256001851434,
      "grad_norm": 0.6606085311908325,
      "learning_rate": 0.00018915223868821158,
      "loss": 12.3153,
      "step": 6433
    },
    {
      "epoch": 0.3503570140150974,
      "grad_norm": 0.6925034443062139,
      "learning_rate": 0.00018914824387665776,
      "loss": 12.5406,
      "step": 6434
    },
    {
      "epoch": 0.3504114680116804,
      "grad_norm": 0.658729625903655,
      "learning_rate": 0.00018914424837187225,
      "loss": 12.4132,
      "step": 6435
    },
    {
      "epoch": 0.3504659220082634,
      "grad_norm": 0.6684606670577931,
      "learning_rate": 0.0001891402521738861,
      "loss": 12.4707,
      "step": 6436
    },
    {
      "epoch": 0.3505203760048464,
      "grad_norm": 0.7201695038485614,
      "learning_rate": 0.0001891362552827304,
      "loss": 12.6224,
      "step": 6437
    },
    {
      "epoch": 0.3505748300014294,
      "grad_norm": 0.7326773653782573,
      "learning_rate": 0.00018913225769843624,
      "loss": 12.5076,
      "step": 6438
    },
    {
      "epoch": 0.3506292839980124,
      "grad_norm": 0.6495451475924698,
      "learning_rate": 0.00018912825942103467,
      "loss": 12.4381,
      "step": 6439
    },
    {
      "epoch": 0.35068373799459546,
      "grad_norm": 0.6452599673897227,
      "learning_rate": 0.00018912426045055683,
      "loss": 12.5916,
      "step": 6440
    },
    {
      "epoch": 0.35073819199117845,
      "grad_norm": 0.6765236346094069,
      "learning_rate": 0.0001891202607870338,
      "loss": 12.5231,
      "step": 6441
    },
    {
      "epoch": 0.35079264598776144,
      "grad_norm": 0.7173390388248864,
      "learning_rate": 0.00018911626043049666,
      "loss": 12.6167,
      "step": 6442
    },
    {
      "epoch": 0.3508470999843445,
      "grad_norm": 0.5881983259786162,
      "learning_rate": 0.0001891122593809765,
      "loss": 12.4536,
      "step": 6443
    },
    {
      "epoch": 0.3509015539809275,
      "grad_norm": 0.6353387106698005,
      "learning_rate": 0.00018910825763850456,
      "loss": 12.4374,
      "step": 6444
    },
    {
      "epoch": 0.3509560079775105,
      "grad_norm": 0.6819914447916594,
      "learning_rate": 0.0001891042552031118,
      "loss": 12.6692,
      "step": 6445
    },
    {
      "epoch": 0.3510104619740935,
      "grad_norm": 0.6744783603916592,
      "learning_rate": 0.00018910025207482942,
      "loss": 12.5202,
      "step": 6446
    },
    {
      "epoch": 0.3510649159706765,
      "grad_norm": 0.9038488748322342,
      "learning_rate": 0.00018909624825368853,
      "loss": 12.4483,
      "step": 6447
    },
    {
      "epoch": 0.3511193699672595,
      "grad_norm": 0.6540032197278525,
      "learning_rate": 0.00018909224373972027,
      "loss": 12.4902,
      "step": 6448
    },
    {
      "epoch": 0.35117382396384256,
      "grad_norm": 0.6431012659983741,
      "learning_rate": 0.0001890882385329558,
      "loss": 12.5164,
      "step": 6449
    },
    {
      "epoch": 0.35122827796042555,
      "grad_norm": 0.6713430875386474,
      "learning_rate": 0.00018908423263342626,
      "loss": 12.5303,
      "step": 6450
    },
    {
      "epoch": 0.35128273195700854,
      "grad_norm": 0.7401191797977572,
      "learning_rate": 0.00018908022604116276,
      "loss": 12.5052,
      "step": 6451
    },
    {
      "epoch": 0.3513371859535916,
      "grad_norm": 0.6505333356783289,
      "learning_rate": 0.00018907621875619647,
      "loss": 12.5267,
      "step": 6452
    },
    {
      "epoch": 0.3513916399501746,
      "grad_norm": 0.6812347254619805,
      "learning_rate": 0.00018907221077855862,
      "loss": 12.4048,
      "step": 6453
    },
    {
      "epoch": 0.35144609394675763,
      "grad_norm": 0.6833135113192058,
      "learning_rate": 0.00018906820210828028,
      "loss": 12.5132,
      "step": 6454
    },
    {
      "epoch": 0.3515005479433406,
      "grad_norm": 0.6871285856235322,
      "learning_rate": 0.00018906419274539266,
      "loss": 12.5483,
      "step": 6455
    },
    {
      "epoch": 0.3515550019399236,
      "grad_norm": 0.7299322555996667,
      "learning_rate": 0.00018906018268992694,
      "loss": 12.396,
      "step": 6456
    },
    {
      "epoch": 0.35160945593650667,
      "grad_norm": 0.6799131452832852,
      "learning_rate": 0.00018905617194191435,
      "loss": 12.4827,
      "step": 6457
    },
    {
      "epoch": 0.35166390993308966,
      "grad_norm": 0.6595334118817054,
      "learning_rate": 0.00018905216050138596,
      "loss": 12.5873,
      "step": 6458
    },
    {
      "epoch": 0.35171836392967265,
      "grad_norm": 0.7508304565503127,
      "learning_rate": 0.00018904814836837307,
      "loss": 12.5145,
      "step": 6459
    },
    {
      "epoch": 0.3517728179262557,
      "grad_norm": 0.6701814667017392,
      "learning_rate": 0.00018904413554290684,
      "loss": 12.5822,
      "step": 6460
    },
    {
      "epoch": 0.3518272719228387,
      "grad_norm": 0.6912066956582165,
      "learning_rate": 0.0001890401220250185,
      "loss": 12.6066,
      "step": 6461
    },
    {
      "epoch": 0.3518817259194217,
      "grad_norm": 0.6390939063604993,
      "learning_rate": 0.00018903610781473927,
      "loss": 12.4256,
      "step": 6462
    },
    {
      "epoch": 0.35193617991600473,
      "grad_norm": 0.7219330604492857,
      "learning_rate": 0.00018903209291210027,
      "loss": 12.6042,
      "step": 6463
    },
    {
      "epoch": 0.3519906339125877,
      "grad_norm": 0.7091152392780032,
      "learning_rate": 0.0001890280773171328,
      "loss": 12.7038,
      "step": 6464
    },
    {
      "epoch": 0.3520450879091707,
      "grad_norm": 0.8294803361278149,
      "learning_rate": 0.0001890240610298681,
      "loss": 12.5393,
      "step": 6465
    },
    {
      "epoch": 0.35209954190575377,
      "grad_norm": 0.7327900394265904,
      "learning_rate": 0.00018902004405033733,
      "loss": 12.6018,
      "step": 6466
    },
    {
      "epoch": 0.35215399590233676,
      "grad_norm": 0.7558412427812986,
      "learning_rate": 0.0001890160263785718,
      "loss": 12.3143,
      "step": 6467
    },
    {
      "epoch": 0.35220844989891975,
      "grad_norm": 0.6485480492908395,
      "learning_rate": 0.0001890120080146027,
      "loss": 12.4945,
      "step": 6468
    },
    {
      "epoch": 0.3522629038955028,
      "grad_norm": 0.7549308050018257,
      "learning_rate": 0.00018900798895846134,
      "loss": 12.4988,
      "step": 6469
    },
    {
      "epoch": 0.3523173578920858,
      "grad_norm": 0.7610598992375268,
      "learning_rate": 0.00018900396921017886,
      "loss": 12.5119,
      "step": 6470
    },
    {
      "epoch": 0.3523718118886688,
      "grad_norm": 0.6593766636270706,
      "learning_rate": 0.00018899994876978664,
      "loss": 12.5072,
      "step": 6471
    },
    {
      "epoch": 0.35242626588525183,
      "grad_norm": 0.5841234835806068,
      "learning_rate": 0.00018899592763731588,
      "loss": 12.5108,
      "step": 6472
    },
    {
      "epoch": 0.3524807198818348,
      "grad_norm": 0.6375815631140179,
      "learning_rate": 0.00018899190581279788,
      "loss": 12.5106,
      "step": 6473
    },
    {
      "epoch": 0.3525351738784178,
      "grad_norm": 0.7265728910133166,
      "learning_rate": 0.00018898788329626388,
      "loss": 12.61,
      "step": 6474
    },
    {
      "epoch": 0.35258962787500087,
      "grad_norm": 0.7442169125235223,
      "learning_rate": 0.00018898386008774515,
      "loss": 12.5473,
      "step": 6475
    },
    {
      "epoch": 0.35264408187158386,
      "grad_norm": 0.7688054198431855,
      "learning_rate": 0.00018897983618727305,
      "loss": 12.5165,
      "step": 6476
    },
    {
      "epoch": 0.35269853586816685,
      "grad_norm": 0.7814867856748376,
      "learning_rate": 0.00018897581159487879,
      "loss": 12.6789,
      "step": 6477
    },
    {
      "epoch": 0.3527529898647499,
      "grad_norm": 0.8538938343014641,
      "learning_rate": 0.00018897178631059372,
      "loss": 12.6351,
      "step": 6478
    },
    {
      "epoch": 0.3528074438613329,
      "grad_norm": 0.6388380559820188,
      "learning_rate": 0.00018896776033444908,
      "loss": 12.5478,
      "step": 6479
    },
    {
      "epoch": 0.3528618978579159,
      "grad_norm": 0.6054334899810893,
      "learning_rate": 0.00018896373366647623,
      "loss": 12.4419,
      "step": 6480
    },
    {
      "epoch": 0.35291635185449893,
      "grad_norm": 0.7531948587362995,
      "learning_rate": 0.00018895970630670653,
      "loss": 12.4213,
      "step": 6481
    },
    {
      "epoch": 0.3529708058510819,
      "grad_norm": 0.7431856849626534,
      "learning_rate": 0.00018895567825517117,
      "loss": 12.4706,
      "step": 6482
    },
    {
      "epoch": 0.3530252598476649,
      "grad_norm": 0.6282523964067674,
      "learning_rate": 0.00018895164951190154,
      "loss": 12.446,
      "step": 6483
    },
    {
      "epoch": 0.35307971384424797,
      "grad_norm": 0.812969482425293,
      "learning_rate": 0.00018894762007692898,
      "loss": 12.5923,
      "step": 6484
    },
    {
      "epoch": 0.35313416784083096,
      "grad_norm": 0.6418835602683038,
      "learning_rate": 0.00018894358995028481,
      "loss": 12.5747,
      "step": 6485
    },
    {
      "epoch": 0.353188621837414,
      "grad_norm": 0.7819665290679157,
      "learning_rate": 0.00018893955913200036,
      "loss": 12.5215,
      "step": 6486
    },
    {
      "epoch": 0.353243075833997,
      "grad_norm": 0.6917473065243283,
      "learning_rate": 0.000188935527622107,
      "loss": 12.4607,
      "step": 6487
    },
    {
      "epoch": 0.35329752983058,
      "grad_norm": 0.7319956830798008,
      "learning_rate": 0.00018893149542063603,
      "loss": 12.5191,
      "step": 6488
    },
    {
      "epoch": 0.35335198382716304,
      "grad_norm": 0.6774051839724597,
      "learning_rate": 0.00018892746252761888,
      "loss": 12.6299,
      "step": 6489
    },
    {
      "epoch": 0.35340643782374603,
      "grad_norm": 0.6421731835455767,
      "learning_rate": 0.00018892342894308683,
      "loss": 12.4119,
      "step": 6490
    },
    {
      "epoch": 0.353460891820329,
      "grad_norm": 0.6386697914025962,
      "learning_rate": 0.0001889193946670713,
      "loss": 12.4614,
      "step": 6491
    },
    {
      "epoch": 0.3535153458169121,
      "grad_norm": 0.694150605451831,
      "learning_rate": 0.00018891535969960368,
      "loss": 12.3783,
      "step": 6492
    },
    {
      "epoch": 0.35356979981349507,
      "grad_norm": 0.6789059976029038,
      "learning_rate": 0.0001889113240407153,
      "loss": 12.5107,
      "step": 6493
    },
    {
      "epoch": 0.35362425381007806,
      "grad_norm": 0.6227305036984996,
      "learning_rate": 0.0001889072876904375,
      "loss": 12.4732,
      "step": 6494
    },
    {
      "epoch": 0.3536787078066611,
      "grad_norm": 0.6914785944572073,
      "learning_rate": 0.00018890325064880177,
      "loss": 12.6268,
      "step": 6495
    },
    {
      "epoch": 0.3537331618032441,
      "grad_norm": 0.8133604433595188,
      "learning_rate": 0.00018889921291583944,
      "loss": 12.4998,
      "step": 6496
    },
    {
      "epoch": 0.3537876157998271,
      "grad_norm": 0.6797182621673394,
      "learning_rate": 0.00018889517449158192,
      "loss": 12.4768,
      "step": 6497
    },
    {
      "epoch": 0.35384206979641014,
      "grad_norm": 0.7804143495476945,
      "learning_rate": 0.0001888911353760606,
      "loss": 12.6219,
      "step": 6498
    },
    {
      "epoch": 0.35389652379299313,
      "grad_norm": 0.6579373319187583,
      "learning_rate": 0.00018888709556930694,
      "loss": 12.4734,
      "step": 6499
    },
    {
      "epoch": 0.3539509777895761,
      "grad_norm": 0.6642270385457227,
      "learning_rate": 0.00018888305507135228,
      "loss": 12.5715,
      "step": 6500
    },
    {
      "epoch": 0.3540054317861592,
      "grad_norm": 0.6342501988731931,
      "learning_rate": 0.0001888790138822281,
      "loss": 12.4374,
      "step": 6501
    },
    {
      "epoch": 0.35405988578274217,
      "grad_norm": 0.6461119173969245,
      "learning_rate": 0.0001888749720019658,
      "loss": 12.4934,
      "step": 6502
    },
    {
      "epoch": 0.35411433977932516,
      "grad_norm": 0.6868152071661218,
      "learning_rate": 0.0001888709294305968,
      "loss": 12.5943,
      "step": 6503
    },
    {
      "epoch": 0.3541687937759082,
      "grad_norm": 0.6077876589490357,
      "learning_rate": 0.00018886688616815258,
      "loss": 12.4525,
      "step": 6504
    },
    {
      "epoch": 0.3542232477724912,
      "grad_norm": 0.6724925347128431,
      "learning_rate": 0.00018886284221466455,
      "loss": 12.3494,
      "step": 6505
    },
    {
      "epoch": 0.3542777017690742,
      "grad_norm": 0.7110664925398358,
      "learning_rate": 0.00018885879757016413,
      "loss": 12.5749,
      "step": 6506
    },
    {
      "epoch": 0.35433215576565724,
      "grad_norm": 0.7291926327690469,
      "learning_rate": 0.00018885475223468282,
      "loss": 12.5879,
      "step": 6507
    },
    {
      "epoch": 0.35438660976224023,
      "grad_norm": 0.7734607293232016,
      "learning_rate": 0.00018885070620825202,
      "loss": 12.5403,
      "step": 6508
    },
    {
      "epoch": 0.3544410637588232,
      "grad_norm": 0.6355688010151955,
      "learning_rate": 0.00018884665949090327,
      "loss": 12.545,
      "step": 6509
    },
    {
      "epoch": 0.3544955177554063,
      "grad_norm": 0.6731065364021674,
      "learning_rate": 0.000188842612082668,
      "loss": 12.4158,
      "step": 6510
    },
    {
      "epoch": 0.35454997175198927,
      "grad_norm": 0.6785630727656202,
      "learning_rate": 0.00018883856398357765,
      "loss": 12.5455,
      "step": 6511
    },
    {
      "epoch": 0.35460442574857226,
      "grad_norm": 0.677040878512404,
      "learning_rate": 0.00018883451519366372,
      "loss": 12.5873,
      "step": 6512
    },
    {
      "epoch": 0.3546588797451553,
      "grad_norm": 0.65287559154684,
      "learning_rate": 0.00018883046571295772,
      "loss": 12.2615,
      "step": 6513
    },
    {
      "epoch": 0.3547133337417383,
      "grad_norm": 0.5831793815322747,
      "learning_rate": 0.00018882641554149112,
      "loss": 12.5245,
      "step": 6514
    },
    {
      "epoch": 0.3547677877383213,
      "grad_norm": 0.7114706506008506,
      "learning_rate": 0.00018882236467929542,
      "loss": 12.4625,
      "step": 6515
    },
    {
      "epoch": 0.35482224173490434,
      "grad_norm": 0.649963966810882,
      "learning_rate": 0.00018881831312640213,
      "loss": 12.5165,
      "step": 6516
    },
    {
      "epoch": 0.35487669573148733,
      "grad_norm": 0.6883907364877639,
      "learning_rate": 0.0001888142608828427,
      "loss": 12.6147,
      "step": 6517
    },
    {
      "epoch": 0.3549311497280703,
      "grad_norm": 0.6099915381134763,
      "learning_rate": 0.00018881020794864873,
      "loss": 12.4195,
      "step": 6518
    },
    {
      "epoch": 0.3549856037246534,
      "grad_norm": 0.6779392941887493,
      "learning_rate": 0.00018880615432385165,
      "loss": 12.5528,
      "step": 6519
    },
    {
      "epoch": 0.35504005772123637,
      "grad_norm": 0.6853390762420948,
      "learning_rate": 0.00018880210000848306,
      "loss": 12.4175,
      "step": 6520
    },
    {
      "epoch": 0.3550945117178194,
      "grad_norm": 0.7099141792622605,
      "learning_rate": 0.0001887980450025744,
      "loss": 12.576,
      "step": 6521
    },
    {
      "epoch": 0.3551489657144024,
      "grad_norm": 0.6741612375122882,
      "learning_rate": 0.0001887939893061573,
      "loss": 12.5621,
      "step": 6522
    },
    {
      "epoch": 0.3552034197109854,
      "grad_norm": 0.6367742915817244,
      "learning_rate": 0.00018878993291926324,
      "loss": 12.4885,
      "step": 6523
    },
    {
      "epoch": 0.35525787370756845,
      "grad_norm": 0.6563563804700083,
      "learning_rate": 0.00018878587584192374,
      "loss": 12.3315,
      "step": 6524
    },
    {
      "epoch": 0.35531232770415144,
      "grad_norm": 0.6755606542985129,
      "learning_rate": 0.00018878181807417042,
      "loss": 12.3851,
      "step": 6525
    },
    {
      "epoch": 0.35536678170073444,
      "grad_norm": 0.6810645317599997,
      "learning_rate": 0.00018877775961603476,
      "loss": 12.4542,
      "step": 6526
    },
    {
      "epoch": 0.3554212356973175,
      "grad_norm": 0.6584896277234938,
      "learning_rate": 0.00018877370046754838,
      "loss": 12.4904,
      "step": 6527
    },
    {
      "epoch": 0.3554756896939005,
      "grad_norm": 0.782531194635525,
      "learning_rate": 0.00018876964062874277,
      "loss": 12.6789,
      "step": 6528
    },
    {
      "epoch": 0.35553014369048347,
      "grad_norm": 0.7047130293938447,
      "learning_rate": 0.0001887655800996496,
      "loss": 12.4975,
      "step": 6529
    },
    {
      "epoch": 0.3555845976870665,
      "grad_norm": 0.6392556211825621,
      "learning_rate": 0.00018876151888030037,
      "loss": 12.4937,
      "step": 6530
    },
    {
      "epoch": 0.3556390516836495,
      "grad_norm": 0.6357688326994323,
      "learning_rate": 0.00018875745697072668,
      "loss": 12.498,
      "step": 6531
    },
    {
      "epoch": 0.3556935056802325,
      "grad_norm": 0.7523241658001276,
      "learning_rate": 0.00018875339437096012,
      "loss": 12.5092,
      "step": 6532
    },
    {
      "epoch": 0.35574795967681555,
      "grad_norm": 0.6611830202180823,
      "learning_rate": 0.0001887493310810323,
      "loss": 12.5987,
      "step": 6533
    },
    {
      "epoch": 0.35580241367339854,
      "grad_norm": 0.6434045027474856,
      "learning_rate": 0.0001887452671009748,
      "loss": 12.4229,
      "step": 6534
    },
    {
      "epoch": 0.35585686766998154,
      "grad_norm": 0.6348599415287733,
      "learning_rate": 0.0001887412024308192,
      "loss": 12.4465,
      "step": 6535
    },
    {
      "epoch": 0.3559113216665646,
      "grad_norm": 0.6012998013242754,
      "learning_rate": 0.00018873713707059716,
      "loss": 12.4638,
      "step": 6536
    },
    {
      "epoch": 0.3559657756631476,
      "grad_norm": 0.6225976072562621,
      "learning_rate": 0.00018873307102034023,
      "loss": 12.4929,
      "step": 6537
    },
    {
      "epoch": 0.35602022965973057,
      "grad_norm": 0.6537989187335137,
      "learning_rate": 0.00018872900428008004,
      "loss": 12.4924,
      "step": 6538
    },
    {
      "epoch": 0.3560746836563136,
      "grad_norm": 0.7310649052768982,
      "learning_rate": 0.00018872493684984827,
      "loss": 12.6,
      "step": 6539
    },
    {
      "epoch": 0.3561291376528966,
      "grad_norm": 0.619228080128155,
      "learning_rate": 0.00018872086872967652,
      "loss": 12.4596,
      "step": 6540
    },
    {
      "epoch": 0.3561835916494796,
      "grad_norm": 0.8602848445604844,
      "learning_rate": 0.0001887167999195964,
      "loss": 12.3719,
      "step": 6541
    },
    {
      "epoch": 0.35623804564606265,
      "grad_norm": 0.6542020442995943,
      "learning_rate": 0.00018871273041963954,
      "loss": 12.5089,
      "step": 6542
    },
    {
      "epoch": 0.35629249964264564,
      "grad_norm": 0.6450493139868647,
      "learning_rate": 0.00018870866022983765,
      "loss": 12.4721,
      "step": 6543
    },
    {
      "epoch": 0.35634695363922864,
      "grad_norm": 0.6627589095066768,
      "learning_rate": 0.00018870458935022234,
      "loss": 12.5715,
      "step": 6544
    },
    {
      "epoch": 0.3564014076358117,
      "grad_norm": 0.6679819323871122,
      "learning_rate": 0.00018870051778082525,
      "loss": 12.4713,
      "step": 6545
    },
    {
      "epoch": 0.3564558616323947,
      "grad_norm": 0.5891025212950688,
      "learning_rate": 0.00018869644552167803,
      "loss": 12.4168,
      "step": 6546
    },
    {
      "epoch": 0.35651031562897767,
      "grad_norm": 0.7443645315187565,
      "learning_rate": 0.00018869237257281243,
      "loss": 12.5215,
      "step": 6547
    },
    {
      "epoch": 0.3565647696255607,
      "grad_norm": 0.675935462558438,
      "learning_rate": 0.00018868829893426003,
      "loss": 12.3665,
      "step": 6548
    },
    {
      "epoch": 0.3566192236221437,
      "grad_norm": 0.6266163892118104,
      "learning_rate": 0.00018868422460605259,
      "loss": 12.4143,
      "step": 6549
    },
    {
      "epoch": 0.3566736776187267,
      "grad_norm": 0.7024481342728484,
      "learning_rate": 0.00018868014958822168,
      "loss": 12.4768,
      "step": 6550
    },
    {
      "epoch": 0.35672813161530975,
      "grad_norm": 0.7236654554637472,
      "learning_rate": 0.0001886760738807991,
      "loss": 12.5488,
      "step": 6551
    },
    {
      "epoch": 0.35678258561189274,
      "grad_norm": 0.6277720760414693,
      "learning_rate": 0.0001886719974838165,
      "loss": 12.5349,
      "step": 6552
    },
    {
      "epoch": 0.3568370396084758,
      "grad_norm": 0.5491554105623445,
      "learning_rate": 0.00018866792039730557,
      "loss": 12.4084,
      "step": 6553
    },
    {
      "epoch": 0.3568914936050588,
      "grad_norm": 0.6574591041317043,
      "learning_rate": 0.000188663842621298,
      "loss": 12.5417,
      "step": 6554
    },
    {
      "epoch": 0.3569459476016418,
      "grad_norm": 0.7825346624677734,
      "learning_rate": 0.00018865976415582557,
      "loss": 12.4618,
      "step": 6555
    },
    {
      "epoch": 0.3570004015982248,
      "grad_norm": 0.6397378451307003,
      "learning_rate": 0.0001886556850009199,
      "loss": 12.4972,
      "step": 6556
    },
    {
      "epoch": 0.3570548555948078,
      "grad_norm": 0.6964582007099602,
      "learning_rate": 0.00018865160515661278,
      "loss": 12.6336,
      "step": 6557
    },
    {
      "epoch": 0.3571093095913908,
      "grad_norm": 0.6838137992784039,
      "learning_rate": 0.0001886475246229359,
      "loss": 12.6239,
      "step": 6558
    },
    {
      "epoch": 0.35716376358797386,
      "grad_norm": 0.7702590269454724,
      "learning_rate": 0.000188643443399921,
      "loss": 12.2703,
      "step": 6559
    },
    {
      "epoch": 0.35721821758455685,
      "grad_norm": 0.6275611356284347,
      "learning_rate": 0.00018863936148759983,
      "loss": 12.487,
      "step": 6560
    },
    {
      "epoch": 0.35727267158113984,
      "grad_norm": 0.7105981531040471,
      "learning_rate": 0.0001886352788860041,
      "loss": 12.4876,
      "step": 6561
    },
    {
      "epoch": 0.3573271255777229,
      "grad_norm": 0.6785012603603872,
      "learning_rate": 0.0001886311955951656,
      "loss": 12.5144,
      "step": 6562
    },
    {
      "epoch": 0.3573815795743059,
      "grad_norm": 0.6480000289225968,
      "learning_rate": 0.00018862711161511607,
      "loss": 12.5371,
      "step": 6563
    },
    {
      "epoch": 0.3574360335708889,
      "grad_norm": 0.6836036858096097,
      "learning_rate": 0.0001886230269458872,
      "loss": 12.484,
      "step": 6564
    },
    {
      "epoch": 0.3574904875674719,
      "grad_norm": 0.6809995375722718,
      "learning_rate": 0.00018861894158751086,
      "loss": 12.5016,
      "step": 6565
    },
    {
      "epoch": 0.3575449415640549,
      "grad_norm": 0.6525177660409279,
      "learning_rate": 0.00018861485554001877,
      "loss": 12.5633,
      "step": 6566
    },
    {
      "epoch": 0.3575993955606379,
      "grad_norm": 0.6576641847219601,
      "learning_rate": 0.00018861076880344267,
      "loss": 12.4833,
      "step": 6567
    },
    {
      "epoch": 0.35765384955722096,
      "grad_norm": 0.5990481118703957,
      "learning_rate": 0.0001886066813778144,
      "loss": 12.6255,
      "step": 6568
    },
    {
      "epoch": 0.35770830355380395,
      "grad_norm": 0.805099690625791,
      "learning_rate": 0.0001886025932631657,
      "loss": 12.4749,
      "step": 6569
    },
    {
      "epoch": 0.35776275755038695,
      "grad_norm": 0.6835538764125585,
      "learning_rate": 0.0001885985044595284,
      "loss": 12.5702,
      "step": 6570
    },
    {
      "epoch": 0.35781721154697,
      "grad_norm": 0.6435072091519588,
      "learning_rate": 0.00018859441496693426,
      "loss": 12.7074,
      "step": 6571
    },
    {
      "epoch": 0.357871665543553,
      "grad_norm": 0.6384713238741512,
      "learning_rate": 0.0001885903247854151,
      "loss": 12.3787,
      "step": 6572
    },
    {
      "epoch": 0.357926119540136,
      "grad_norm": 0.7192638915760756,
      "learning_rate": 0.00018858623391500268,
      "loss": 12.5576,
      "step": 6573
    },
    {
      "epoch": 0.357980573536719,
      "grad_norm": 0.6350450338580725,
      "learning_rate": 0.0001885821423557289,
      "loss": 12.5939,
      "step": 6574
    },
    {
      "epoch": 0.358035027533302,
      "grad_norm": 0.7084619240659714,
      "learning_rate": 0.00018857805010762547,
      "loss": 12.5479,
      "step": 6575
    },
    {
      "epoch": 0.358089481529885,
      "grad_norm": 0.6182206352160496,
      "learning_rate": 0.0001885739571707243,
      "loss": 12.4144,
      "step": 6576
    },
    {
      "epoch": 0.35814393552646806,
      "grad_norm": 0.7050067760240177,
      "learning_rate": 0.0001885698635450572,
      "loss": 12.518,
      "step": 6577
    },
    {
      "epoch": 0.35819838952305105,
      "grad_norm": 0.6444317754366243,
      "learning_rate": 0.00018856576923065597,
      "loss": 12.5393,
      "step": 6578
    },
    {
      "epoch": 0.35825284351963405,
      "grad_norm": 0.6796109302689314,
      "learning_rate": 0.00018856167422755246,
      "loss": 12.3975,
      "step": 6579
    },
    {
      "epoch": 0.3583072975162171,
      "grad_norm": 0.6790355456078522,
      "learning_rate": 0.00018855757853577853,
      "loss": 12.478,
      "step": 6580
    },
    {
      "epoch": 0.3583617515128001,
      "grad_norm": 0.7346590134415578,
      "learning_rate": 0.000188553482155366,
      "loss": 12.5285,
      "step": 6581
    },
    {
      "epoch": 0.3584162055093831,
      "grad_norm": 0.6529363025771956,
      "learning_rate": 0.00018854938508634678,
      "loss": 12.5329,
      "step": 6582
    },
    {
      "epoch": 0.3584706595059661,
      "grad_norm": 0.6270728056967176,
      "learning_rate": 0.00018854528732875265,
      "loss": 12.4526,
      "step": 6583
    },
    {
      "epoch": 0.3585251135025491,
      "grad_norm": 0.6167445436474197,
      "learning_rate": 0.00018854118888261554,
      "loss": 12.5071,
      "step": 6584
    },
    {
      "epoch": 0.3585795674991321,
      "grad_norm": 0.6658190255876781,
      "learning_rate": 0.0001885370897479673,
      "loss": 12.611,
      "step": 6585
    },
    {
      "epoch": 0.35863402149571516,
      "grad_norm": 0.7280245914122708,
      "learning_rate": 0.0001885329899248398,
      "loss": 12.603,
      "step": 6586
    },
    {
      "epoch": 0.35868847549229815,
      "grad_norm": 0.6513125287893339,
      "learning_rate": 0.0001885288894132649,
      "loss": 12.4227,
      "step": 6587
    },
    {
      "epoch": 0.3587429294888812,
      "grad_norm": 0.6482208097248535,
      "learning_rate": 0.00018852478821327452,
      "loss": 12.4279,
      "step": 6588
    },
    {
      "epoch": 0.3587973834854642,
      "grad_norm": 0.7403836328311797,
      "learning_rate": 0.00018852068632490058,
      "loss": 12.5555,
      "step": 6589
    },
    {
      "epoch": 0.3588518374820472,
      "grad_norm": 0.6310367450242624,
      "learning_rate": 0.0001885165837481749,
      "loss": 12.4712,
      "step": 6590
    },
    {
      "epoch": 0.35890629147863023,
      "grad_norm": 0.7276121536456144,
      "learning_rate": 0.0001885124804831294,
      "loss": 12.5856,
      "step": 6591
    },
    {
      "epoch": 0.3589607454752132,
      "grad_norm": 0.6186713256945982,
      "learning_rate": 0.00018850837652979605,
      "loss": 12.2877,
      "step": 6592
    },
    {
      "epoch": 0.3590151994717962,
      "grad_norm": 0.6180211458265332,
      "learning_rate": 0.00018850427188820673,
      "loss": 12.553,
      "step": 6593
    },
    {
      "epoch": 0.35906965346837927,
      "grad_norm": 0.6810319920116736,
      "learning_rate": 0.0001885001665583933,
      "loss": 12.5653,
      "step": 6594
    },
    {
      "epoch": 0.35912410746496226,
      "grad_norm": 0.6478256244548984,
      "learning_rate": 0.00018849606054038777,
      "loss": 12.4771,
      "step": 6595
    },
    {
      "epoch": 0.35917856146154525,
      "grad_norm": 0.6203360210240151,
      "learning_rate": 0.00018849195383422202,
      "loss": 12.6188,
      "step": 6596
    },
    {
      "epoch": 0.3592330154581283,
      "grad_norm": 0.7394212994938031,
      "learning_rate": 0.000188487846439928,
      "loss": 12.4102,
      "step": 6597
    },
    {
      "epoch": 0.3592874694547113,
      "grad_norm": 0.6700090375629415,
      "learning_rate": 0.00018848373835753766,
      "loss": 12.3701,
      "step": 6598
    },
    {
      "epoch": 0.3593419234512943,
      "grad_norm": 0.6763191547744783,
      "learning_rate": 0.0001884796295870829,
      "loss": 12.5191,
      "step": 6599
    },
    {
      "epoch": 0.35939637744787734,
      "grad_norm": 0.7225861178544174,
      "learning_rate": 0.00018847552012859573,
      "loss": 12.5006,
      "step": 6600
    },
    {
      "epoch": 0.35945083144446033,
      "grad_norm": 0.6598424950049372,
      "learning_rate": 0.00018847140998210806,
      "loss": 12.4225,
      "step": 6601
    },
    {
      "epoch": 0.3595052854410433,
      "grad_norm": 0.8022917346730528,
      "learning_rate": 0.00018846729914765188,
      "loss": 12.4835,
      "step": 6602
    },
    {
      "epoch": 0.35955973943762637,
      "grad_norm": 0.7155075380930347,
      "learning_rate": 0.00018846318762525915,
      "loss": 12.6107,
      "step": 6603
    },
    {
      "epoch": 0.35961419343420936,
      "grad_norm": 0.5788593511681562,
      "learning_rate": 0.00018845907541496182,
      "loss": 12.345,
      "step": 6604
    },
    {
      "epoch": 0.35966864743079235,
      "grad_norm": 0.7692664908605247,
      "learning_rate": 0.00018845496251679192,
      "loss": 12.4208,
      "step": 6605
    },
    {
      "epoch": 0.3597231014273754,
      "grad_norm": 0.6398281023425616,
      "learning_rate": 0.00018845084893078136,
      "loss": 12.4587,
      "step": 6606
    },
    {
      "epoch": 0.3597775554239584,
      "grad_norm": 0.6285622503429014,
      "learning_rate": 0.00018844673465696218,
      "loss": 12.5134,
      "step": 6607
    },
    {
      "epoch": 0.3598320094205414,
      "grad_norm": 0.7085105210743461,
      "learning_rate": 0.00018844261969536637,
      "loss": 12.5605,
      "step": 6608
    },
    {
      "epoch": 0.35988646341712444,
      "grad_norm": 0.6771526198094195,
      "learning_rate": 0.00018843850404602587,
      "loss": 12.4348,
      "step": 6609
    },
    {
      "epoch": 0.35994091741370743,
      "grad_norm": 0.6476584687119452,
      "learning_rate": 0.0001884343877089728,
      "loss": 12.4236,
      "step": 6610
    },
    {
      "epoch": 0.3599953714102904,
      "grad_norm": 0.6365661190221357,
      "learning_rate": 0.00018843027068423903,
      "loss": 12.4869,
      "step": 6611
    },
    {
      "epoch": 0.36004982540687347,
      "grad_norm": 0.6116624437632676,
      "learning_rate": 0.0001884261529718567,
      "loss": 12.4376,
      "step": 6612
    },
    {
      "epoch": 0.36010427940345646,
      "grad_norm": 0.6161016949457048,
      "learning_rate": 0.00018842203457185777,
      "loss": 12.3886,
      "step": 6613
    },
    {
      "epoch": 0.36015873340003945,
      "grad_norm": 0.6629262293434118,
      "learning_rate": 0.00018841791548427427,
      "loss": 12.4196,
      "step": 6614
    },
    {
      "epoch": 0.3602131873966225,
      "grad_norm": 0.6301688987629763,
      "learning_rate": 0.00018841379570913821,
      "loss": 12.4263,
      "step": 6615
    },
    {
      "epoch": 0.3602676413932055,
      "grad_norm": 0.6423425681866812,
      "learning_rate": 0.00018840967524648165,
      "loss": 12.5683,
      "step": 6616
    },
    {
      "epoch": 0.3603220953897885,
      "grad_norm": 0.6722468104910277,
      "learning_rate": 0.00018840555409633665,
      "loss": 12.5797,
      "step": 6617
    },
    {
      "epoch": 0.36037654938637154,
      "grad_norm": 0.683282736727712,
      "learning_rate": 0.00018840143225873522,
      "loss": 12.4557,
      "step": 6618
    },
    {
      "epoch": 0.36043100338295453,
      "grad_norm": 0.5689102418426638,
      "learning_rate": 0.00018839730973370942,
      "loss": 12.4297,
      "step": 6619
    },
    {
      "epoch": 0.3604854573795376,
      "grad_norm": 0.6515763103187273,
      "learning_rate": 0.00018839318652129136,
      "loss": 12.6092,
      "step": 6620
    },
    {
      "epoch": 0.36053991137612057,
      "grad_norm": 0.6638057645135687,
      "learning_rate": 0.000188389062621513,
      "loss": 12.5255,
      "step": 6621
    },
    {
      "epoch": 0.36059436537270356,
      "grad_norm": 0.7578143155135398,
      "learning_rate": 0.0001883849380344065,
      "loss": 12.4897,
      "step": 6622
    },
    {
      "epoch": 0.3606488193692866,
      "grad_norm": 0.6929891599006907,
      "learning_rate": 0.00018838081276000387,
      "loss": 12.4673,
      "step": 6623
    },
    {
      "epoch": 0.3607032733658696,
      "grad_norm": 0.6098098382946556,
      "learning_rate": 0.00018837668679833725,
      "loss": 12.4919,
      "step": 6624
    },
    {
      "epoch": 0.3607577273624526,
      "grad_norm": 0.6995080045001343,
      "learning_rate": 0.0001883725601494387,
      "loss": 12.4121,
      "step": 6625
    },
    {
      "epoch": 0.36081218135903564,
      "grad_norm": 0.6393239609282849,
      "learning_rate": 0.00018836843281334024,
      "loss": 12.4995,
      "step": 6626
    },
    {
      "epoch": 0.36086663535561864,
      "grad_norm": 0.6179401434135583,
      "learning_rate": 0.0001883643047900741,
      "loss": 12.4329,
      "step": 6627
    },
    {
      "epoch": 0.36092108935220163,
      "grad_norm": 0.7364688329740939,
      "learning_rate": 0.00018836017607967227,
      "loss": 12.5791,
      "step": 6628
    },
    {
      "epoch": 0.3609755433487847,
      "grad_norm": 0.7445944340871038,
      "learning_rate": 0.0001883560466821669,
      "loss": 12.4741,
      "step": 6629
    },
    {
      "epoch": 0.36102999734536767,
      "grad_norm": 0.6936132071756218,
      "learning_rate": 0.00018835191659759008,
      "loss": 12.5278,
      "step": 6630
    },
    {
      "epoch": 0.36108445134195066,
      "grad_norm": 0.6341963179540083,
      "learning_rate": 0.00018834778582597396,
      "loss": 12.5872,
      "step": 6631
    },
    {
      "epoch": 0.3611389053385337,
      "grad_norm": 0.6296041806316938,
      "learning_rate": 0.00018834365436735064,
      "loss": 12.4316,
      "step": 6632
    },
    {
      "epoch": 0.3611933593351167,
      "grad_norm": 0.6374466468077414,
      "learning_rate": 0.00018833952222175224,
      "loss": 12.5258,
      "step": 6633
    },
    {
      "epoch": 0.3612478133316997,
      "grad_norm": 0.6999344709830713,
      "learning_rate": 0.0001883353893892109,
      "loss": 12.5829,
      "step": 6634
    },
    {
      "epoch": 0.36130226732828274,
      "grad_norm": 0.6904477138465861,
      "learning_rate": 0.00018833125586975878,
      "loss": 12.4849,
      "step": 6635
    },
    {
      "epoch": 0.36135672132486574,
      "grad_norm": 0.662277379149618,
      "learning_rate": 0.00018832712166342796,
      "loss": 12.5352,
      "step": 6636
    },
    {
      "epoch": 0.36141117532144873,
      "grad_norm": 0.821381962854756,
      "learning_rate": 0.00018832298677025068,
      "loss": 12.4751,
      "step": 6637
    },
    {
      "epoch": 0.3614656293180318,
      "grad_norm": 0.6317517204534076,
      "learning_rate": 0.000188318851190259,
      "loss": 12.461,
      "step": 6638
    },
    {
      "epoch": 0.36152008331461477,
      "grad_norm": 0.6556528541474681,
      "learning_rate": 0.00018831471492348513,
      "loss": 12.5782,
      "step": 6639
    },
    {
      "epoch": 0.36157453731119776,
      "grad_norm": 0.6910220177221039,
      "learning_rate": 0.00018831057796996124,
      "loss": 12.5669,
      "step": 6640
    },
    {
      "epoch": 0.3616289913077808,
      "grad_norm": 0.7937937585155153,
      "learning_rate": 0.0001883064403297195,
      "loss": 12.6712,
      "step": 6641
    },
    {
      "epoch": 0.3616834453043638,
      "grad_norm": 0.7329019212487958,
      "learning_rate": 0.00018830230200279205,
      "loss": 12.4129,
      "step": 6642
    },
    {
      "epoch": 0.3617378993009468,
      "grad_norm": 0.7721716116033265,
      "learning_rate": 0.0001882981629892111,
      "loss": 12.4551,
      "step": 6643
    },
    {
      "epoch": 0.36179235329752985,
      "grad_norm": 0.7340558824600818,
      "learning_rate": 0.00018829402328900883,
      "loss": 12.3573,
      "step": 6644
    },
    {
      "epoch": 0.36184680729411284,
      "grad_norm": 0.7346948738214002,
      "learning_rate": 0.0001882898829022174,
      "loss": 12.496,
      "step": 6645
    },
    {
      "epoch": 0.36190126129069583,
      "grad_norm": 0.8316671174602631,
      "learning_rate": 0.00018828574182886903,
      "loss": 12.5202,
      "step": 6646
    },
    {
      "epoch": 0.3619557152872789,
      "grad_norm": 0.7174804265086283,
      "learning_rate": 0.00018828160006899598,
      "loss": 12.4468,
      "step": 6647
    },
    {
      "epoch": 0.36201016928386187,
      "grad_norm": 0.7011147967368828,
      "learning_rate": 0.00018827745762263037,
      "loss": 12.5342,
      "step": 6648
    },
    {
      "epoch": 0.36206462328044486,
      "grad_norm": 0.683462054591662,
      "learning_rate": 0.00018827331448980443,
      "loss": 12.5567,
      "step": 6649
    },
    {
      "epoch": 0.3621190772770279,
      "grad_norm": 0.8090110863198164,
      "learning_rate": 0.00018826917067055044,
      "loss": 12.5956,
      "step": 6650
    },
    {
      "epoch": 0.3621735312736109,
      "grad_norm": 0.6607693188818475,
      "learning_rate": 0.00018826502616490053,
      "loss": 12.5962,
      "step": 6651
    },
    {
      "epoch": 0.3622279852701939,
      "grad_norm": 0.6767689806560603,
      "learning_rate": 0.000188260880972887,
      "loss": 12.5465,
      "step": 6652
    },
    {
      "epoch": 0.36228243926677695,
      "grad_norm": 0.6372209862019359,
      "learning_rate": 0.00018825673509454202,
      "loss": 12.2851,
      "step": 6653
    },
    {
      "epoch": 0.36233689326335994,
      "grad_norm": 0.635820161629741,
      "learning_rate": 0.0001882525885298979,
      "loss": 12.4901,
      "step": 6654
    },
    {
      "epoch": 0.362391347259943,
      "grad_norm": 0.7700231819144508,
      "learning_rate": 0.00018824844127898687,
      "loss": 12.6406,
      "step": 6655
    },
    {
      "epoch": 0.362445801256526,
      "grad_norm": 0.6610835813691188,
      "learning_rate": 0.00018824429334184112,
      "loss": 12.5198,
      "step": 6656
    },
    {
      "epoch": 0.36250025525310897,
      "grad_norm": 0.6923623425981027,
      "learning_rate": 0.000188240144718493,
      "loss": 12.5646,
      "step": 6657
    },
    {
      "epoch": 0.362554709249692,
      "grad_norm": 0.7295852877262856,
      "learning_rate": 0.00018823599540897465,
      "loss": 12.5918,
      "step": 6658
    },
    {
      "epoch": 0.362609163246275,
      "grad_norm": 0.5943802872107438,
      "learning_rate": 0.00018823184541331845,
      "loss": 12.3685,
      "step": 6659
    },
    {
      "epoch": 0.362663617242858,
      "grad_norm": 0.7152397393362633,
      "learning_rate": 0.0001882276947315566,
      "loss": 12.4319,
      "step": 6660
    },
    {
      "epoch": 0.36271807123944105,
      "grad_norm": 0.7250489702731249,
      "learning_rate": 0.0001882235433637214,
      "loss": 12.5264,
      "step": 6661
    },
    {
      "epoch": 0.36277252523602405,
      "grad_norm": 0.5924669756695498,
      "learning_rate": 0.00018821939130984517,
      "loss": 12.2873,
      "step": 6662
    },
    {
      "epoch": 0.36282697923260704,
      "grad_norm": 0.645723736104555,
      "learning_rate": 0.00018821523856996013,
      "loss": 12.5939,
      "step": 6663
    },
    {
      "epoch": 0.3628814332291901,
      "grad_norm": 0.5875140836827224,
      "learning_rate": 0.00018821108514409856,
      "loss": 12.474,
      "step": 6664
    },
    {
      "epoch": 0.3629358872257731,
      "grad_norm": 0.6503600619137254,
      "learning_rate": 0.00018820693103229288,
      "loss": 12.4141,
      "step": 6665
    },
    {
      "epoch": 0.36299034122235607,
      "grad_norm": 0.6845049787509762,
      "learning_rate": 0.00018820277623457526,
      "loss": 12.6379,
      "step": 6666
    },
    {
      "epoch": 0.3630447952189391,
      "grad_norm": 0.636216840869751,
      "learning_rate": 0.00018819862075097806,
      "loss": 12.3587,
      "step": 6667
    },
    {
      "epoch": 0.3630992492155221,
      "grad_norm": 0.6478000866250779,
      "learning_rate": 0.00018819446458153362,
      "loss": 12.4389,
      "step": 6668
    },
    {
      "epoch": 0.3631537032121051,
      "grad_norm": 0.6636993194800846,
      "learning_rate": 0.0001881903077262742,
      "loss": 12.4896,
      "step": 6669
    },
    {
      "epoch": 0.36320815720868815,
      "grad_norm": 0.6230684019836505,
      "learning_rate": 0.0001881861501852322,
      "loss": 12.6979,
      "step": 6670
    },
    {
      "epoch": 0.36326261120527115,
      "grad_norm": 0.6421208304249839,
      "learning_rate": 0.00018818199195843986,
      "loss": 12.4471,
      "step": 6671
    },
    {
      "epoch": 0.36331706520185414,
      "grad_norm": 0.6028078428535656,
      "learning_rate": 0.00018817783304592959,
      "loss": 12.4494,
      "step": 6672
    },
    {
      "epoch": 0.3633715191984372,
      "grad_norm": 0.6221622025387541,
      "learning_rate": 0.00018817367344773372,
      "loss": 12.4518,
      "step": 6673
    },
    {
      "epoch": 0.3634259731950202,
      "grad_norm": 0.6563182110144833,
      "learning_rate": 0.00018816951316388453,
      "loss": 12.6353,
      "step": 6674
    },
    {
      "epoch": 0.3634804271916032,
      "grad_norm": 0.6485631719952629,
      "learning_rate": 0.00018816535219441446,
      "loss": 12.4347,
      "step": 6675
    },
    {
      "epoch": 0.3635348811881862,
      "grad_norm": 0.6252855421360843,
      "learning_rate": 0.0001881611905393558,
      "loss": 12.4695,
      "step": 6676
    },
    {
      "epoch": 0.3635893351847692,
      "grad_norm": 0.6735701127245386,
      "learning_rate": 0.00018815702819874097,
      "loss": 12.5676,
      "step": 6677
    },
    {
      "epoch": 0.3636437891813522,
      "grad_norm": 0.7415341714654862,
      "learning_rate": 0.00018815286517260229,
      "loss": 12.764,
      "step": 6678
    },
    {
      "epoch": 0.36369824317793525,
      "grad_norm": 0.6322693357159745,
      "learning_rate": 0.00018814870146097214,
      "loss": 12.5009,
      "step": 6679
    },
    {
      "epoch": 0.36375269717451825,
      "grad_norm": 0.6458801056972988,
      "learning_rate": 0.00018814453706388287,
      "loss": 12.5047,
      "step": 6680
    },
    {
      "epoch": 0.36380715117110124,
      "grad_norm": 0.7096837355005496,
      "learning_rate": 0.00018814037198136698,
      "loss": 12.4631,
      "step": 6681
    },
    {
      "epoch": 0.3638616051676843,
      "grad_norm": 0.6345182119218137,
      "learning_rate": 0.0001881362062134567,
      "loss": 12.4933,
      "step": 6682
    },
    {
      "epoch": 0.3639160591642673,
      "grad_norm": 0.6692866959051389,
      "learning_rate": 0.00018813203976018455,
      "loss": 12.4697,
      "step": 6683
    },
    {
      "epoch": 0.3639705131608503,
      "grad_norm": 0.6995810148133702,
      "learning_rate": 0.00018812787262158286,
      "loss": 12.5705,
      "step": 6684
    },
    {
      "epoch": 0.3640249671574333,
      "grad_norm": 0.7825305327360113,
      "learning_rate": 0.00018812370479768406,
      "loss": 12.6368,
      "step": 6685
    },
    {
      "epoch": 0.3640794211540163,
      "grad_norm": 0.6714265725827628,
      "learning_rate": 0.00018811953628852056,
      "loss": 12.5814,
      "step": 6686
    },
    {
      "epoch": 0.3641338751505993,
      "grad_norm": 0.8482129985990133,
      "learning_rate": 0.00018811536709412475,
      "loss": 12.4584,
      "step": 6687
    },
    {
      "epoch": 0.36418832914718235,
      "grad_norm": 0.6692639872542804,
      "learning_rate": 0.00018811119721452908,
      "loss": 12.4091,
      "step": 6688
    },
    {
      "epoch": 0.36424278314376535,
      "grad_norm": 0.844682059447251,
      "learning_rate": 0.00018810702664976594,
      "loss": 12.72,
      "step": 6689
    },
    {
      "epoch": 0.3642972371403484,
      "grad_norm": 0.7272075202619086,
      "learning_rate": 0.0001881028553998678,
      "loss": 12.4081,
      "step": 6690
    },
    {
      "epoch": 0.3643516911369314,
      "grad_norm": 0.7004977677490962,
      "learning_rate": 0.0001880986834648671,
      "loss": 12.4538,
      "step": 6691
    },
    {
      "epoch": 0.3644061451335144,
      "grad_norm": 0.8629727368672526,
      "learning_rate": 0.00018809451084479624,
      "loss": 12.6221,
      "step": 6692
    },
    {
      "epoch": 0.36446059913009743,
      "grad_norm": 0.7851367318362038,
      "learning_rate": 0.0001880903375396877,
      "loss": 12.5663,
      "step": 6693
    },
    {
      "epoch": 0.3645150531266804,
      "grad_norm": 0.6854981472693442,
      "learning_rate": 0.0001880861635495739,
      "loss": 12.4976,
      "step": 6694
    },
    {
      "epoch": 0.3645695071232634,
      "grad_norm": 0.7607531587291991,
      "learning_rate": 0.00018808198887448736,
      "loss": 12.5451,
      "step": 6695
    },
    {
      "epoch": 0.36462396111984646,
      "grad_norm": 0.6346208042477164,
      "learning_rate": 0.00018807781351446048,
      "loss": 12.526,
      "step": 6696
    },
    {
      "epoch": 0.36467841511642946,
      "grad_norm": 0.669565982871631,
      "learning_rate": 0.0001880736374695258,
      "loss": 12.4964,
      "step": 6697
    },
    {
      "epoch": 0.36473286911301245,
      "grad_norm": 0.7089403356676109,
      "learning_rate": 0.00018806946073971569,
      "loss": 12.4569,
      "step": 6698
    },
    {
      "epoch": 0.3647873231095955,
      "grad_norm": 0.8414280784882965,
      "learning_rate": 0.0001880652833250627,
      "loss": 12.5536,
      "step": 6699
    },
    {
      "epoch": 0.3648417771061785,
      "grad_norm": 0.6479374855180915,
      "learning_rate": 0.00018806110522559926,
      "loss": 12.4727,
      "step": 6700
    },
    {
      "epoch": 0.3648962311027615,
      "grad_norm": 0.7031712491465153,
      "learning_rate": 0.00018805692644135796,
      "loss": 12.5094,
      "step": 6701
    },
    {
      "epoch": 0.36495068509934453,
      "grad_norm": 0.6561201324217154,
      "learning_rate": 0.00018805274697237119,
      "loss": 12.6912,
      "step": 6702
    },
    {
      "epoch": 0.3650051390959275,
      "grad_norm": 0.8083614773954539,
      "learning_rate": 0.00018804856681867152,
      "loss": 12.5316,
      "step": 6703
    },
    {
      "epoch": 0.3650595930925105,
      "grad_norm": 0.67374064745605,
      "learning_rate": 0.0001880443859802914,
      "loss": 12.6457,
      "step": 6704
    },
    {
      "epoch": 0.36511404708909356,
      "grad_norm": 0.6532324213824884,
      "learning_rate": 0.00018804020445726337,
      "loss": 12.5391,
      "step": 6705
    },
    {
      "epoch": 0.36516850108567656,
      "grad_norm": 0.6757495628646862,
      "learning_rate": 0.00018803602224962,
      "loss": 12.5225,
      "step": 6706
    },
    {
      "epoch": 0.36522295508225955,
      "grad_norm": 0.6432686278770082,
      "learning_rate": 0.0001880318393573937,
      "loss": 12.453,
      "step": 6707
    },
    {
      "epoch": 0.3652774090788426,
      "grad_norm": 0.7883001660334636,
      "learning_rate": 0.00018802765578061705,
      "loss": 12.5551,
      "step": 6708
    },
    {
      "epoch": 0.3653318630754256,
      "grad_norm": 0.6461071475432286,
      "learning_rate": 0.00018802347151932264,
      "loss": 12.55,
      "step": 6709
    },
    {
      "epoch": 0.3653863170720086,
      "grad_norm": 0.6691570814181542,
      "learning_rate": 0.0001880192865735429,
      "loss": 12.6225,
      "step": 6710
    },
    {
      "epoch": 0.36544077106859163,
      "grad_norm": 0.6922546666923483,
      "learning_rate": 0.00018801510094331047,
      "loss": 12.4039,
      "step": 6711
    },
    {
      "epoch": 0.3654952250651746,
      "grad_norm": 0.5936430837492487,
      "learning_rate": 0.00018801091462865784,
      "loss": 12.4361,
      "step": 6712
    },
    {
      "epoch": 0.3655496790617576,
      "grad_norm": 0.6664258393725,
      "learning_rate": 0.00018800672762961758,
      "loss": 12.4437,
      "step": 6713
    },
    {
      "epoch": 0.36560413305834066,
      "grad_norm": 0.5887534964830626,
      "learning_rate": 0.00018800253994622222,
      "loss": 12.476,
      "step": 6714
    },
    {
      "epoch": 0.36565858705492366,
      "grad_norm": 0.6531177932509925,
      "learning_rate": 0.00018799835157850439,
      "loss": 12.6041,
      "step": 6715
    },
    {
      "epoch": 0.36571304105150665,
      "grad_norm": 0.7136114805228129,
      "learning_rate": 0.00018799416252649658,
      "loss": 12.5732,
      "step": 6716
    },
    {
      "epoch": 0.3657674950480897,
      "grad_norm": 0.6176999560468083,
      "learning_rate": 0.00018798997279023143,
      "loss": 12.4697,
      "step": 6717
    },
    {
      "epoch": 0.3658219490446727,
      "grad_norm": 0.641058439634723,
      "learning_rate": 0.00018798578236974153,
      "loss": 12.5075,
      "step": 6718
    },
    {
      "epoch": 0.3658764030412557,
      "grad_norm": 0.693524325390353,
      "learning_rate": 0.00018798159126505936,
      "loss": 12.614,
      "step": 6719
    },
    {
      "epoch": 0.36593085703783873,
      "grad_norm": 0.6367206031439175,
      "learning_rate": 0.00018797739947621763,
      "loss": 12.5712,
      "step": 6720
    },
    {
      "epoch": 0.3659853110344217,
      "grad_norm": 0.6231257744211336,
      "learning_rate": 0.00018797320700324885,
      "loss": 12.4514,
      "step": 6721
    },
    {
      "epoch": 0.36603976503100477,
      "grad_norm": 0.6479835224474679,
      "learning_rate": 0.0001879690138461857,
      "loss": 12.6704,
      "step": 6722
    },
    {
      "epoch": 0.36609421902758776,
      "grad_norm": 0.6621267367927367,
      "learning_rate": 0.00018796482000506072,
      "loss": 12.4577,
      "step": 6723
    },
    {
      "epoch": 0.36614867302417076,
      "grad_norm": 0.6044248186047356,
      "learning_rate": 0.00018796062547990657,
      "loss": 12.3559,
      "step": 6724
    },
    {
      "epoch": 0.3662031270207538,
      "grad_norm": 0.6139721548320676,
      "learning_rate": 0.00018795643027075585,
      "loss": 12.503,
      "step": 6725
    },
    {
      "epoch": 0.3662575810173368,
      "grad_norm": 0.652693152929206,
      "learning_rate": 0.00018795223437764115,
      "loss": 12.4726,
      "step": 6726
    },
    {
      "epoch": 0.3663120350139198,
      "grad_norm": 0.5914756538338115,
      "learning_rate": 0.0001879480378005951,
      "loss": 12.3033,
      "step": 6727
    },
    {
      "epoch": 0.36636648901050284,
      "grad_norm": 0.7254217447618527,
      "learning_rate": 0.00018794384053965043,
      "loss": 12.5628,
      "step": 6728
    },
    {
      "epoch": 0.36642094300708583,
      "grad_norm": 0.5891298822369286,
      "learning_rate": 0.00018793964259483968,
      "loss": 12.4264,
      "step": 6729
    },
    {
      "epoch": 0.3664753970036688,
      "grad_norm": 0.6553598414916912,
      "learning_rate": 0.0001879354439661955,
      "loss": 12.5271,
      "step": 6730
    },
    {
      "epoch": 0.36652985100025187,
      "grad_norm": 0.6145565110030625,
      "learning_rate": 0.0001879312446537506,
      "loss": 12.3724,
      "step": 6731
    },
    {
      "epoch": 0.36658430499683486,
      "grad_norm": 0.664373293286463,
      "learning_rate": 0.00018792704465753755,
      "loss": 12.6268,
      "step": 6732
    },
    {
      "epoch": 0.36663875899341786,
      "grad_norm": 0.6838236991001653,
      "learning_rate": 0.00018792284397758908,
      "loss": 12.6584,
      "step": 6733
    },
    {
      "epoch": 0.3666932129900009,
      "grad_norm": 0.6663236308420681,
      "learning_rate": 0.00018791864261393784,
      "loss": 12.4839,
      "step": 6734
    },
    {
      "epoch": 0.3667476669865839,
      "grad_norm": 0.6418049564560676,
      "learning_rate": 0.00018791444056661646,
      "loss": 12.5051,
      "step": 6735
    },
    {
      "epoch": 0.3668021209831669,
      "grad_norm": 0.5569929085876848,
      "learning_rate": 0.0001879102378356577,
      "loss": 12.5471,
      "step": 6736
    },
    {
      "epoch": 0.36685657497974994,
      "grad_norm": 0.67123615672468,
      "learning_rate": 0.00018790603442109412,
      "loss": 12.4874,
      "step": 6737
    },
    {
      "epoch": 0.36691102897633293,
      "grad_norm": 0.6303929900678815,
      "learning_rate": 0.0001879018303229585,
      "loss": 12.415,
      "step": 6738
    },
    {
      "epoch": 0.3669654829729159,
      "grad_norm": 0.7224967526388294,
      "learning_rate": 0.0001878976255412835,
      "loss": 12.4743,
      "step": 6739
    },
    {
      "epoch": 0.36701993696949897,
      "grad_norm": 0.6074412775014366,
      "learning_rate": 0.00018789342007610187,
      "loss": 12.4987,
      "step": 6740
    },
    {
      "epoch": 0.36707439096608196,
      "grad_norm": 0.6347238589759355,
      "learning_rate": 0.00018788921392744624,
      "loss": 12.517,
      "step": 6741
    },
    {
      "epoch": 0.36712884496266496,
      "grad_norm": 0.6664557545586834,
      "learning_rate": 0.00018788500709534934,
      "loss": 12.55,
      "step": 6742
    },
    {
      "epoch": 0.367183298959248,
      "grad_norm": 0.6477748468933896,
      "learning_rate": 0.00018788079957984385,
      "loss": 12.4114,
      "step": 6743
    },
    {
      "epoch": 0.367237752955831,
      "grad_norm": 0.6270361143383161,
      "learning_rate": 0.00018787659138096258,
      "loss": 12.3699,
      "step": 6744
    },
    {
      "epoch": 0.367292206952414,
      "grad_norm": 0.6315587680247866,
      "learning_rate": 0.0001878723824987382,
      "loss": 12.5573,
      "step": 6745
    },
    {
      "epoch": 0.36734666094899704,
      "grad_norm": 0.6682517723733418,
      "learning_rate": 0.00018786817293320337,
      "loss": 12.4946,
      "step": 6746
    },
    {
      "epoch": 0.36740111494558003,
      "grad_norm": 0.5731857714905204,
      "learning_rate": 0.00018786396268439094,
      "loss": 12.5497,
      "step": 6747
    },
    {
      "epoch": 0.367455568942163,
      "grad_norm": 0.7154608334837631,
      "learning_rate": 0.00018785975175233363,
      "loss": 12.4118,
      "step": 6748
    },
    {
      "epoch": 0.3675100229387461,
      "grad_norm": 0.6314309590611179,
      "learning_rate": 0.00018785554013706413,
      "loss": 12.5279,
      "step": 6749
    },
    {
      "epoch": 0.36756447693532907,
      "grad_norm": 0.7327550590135451,
      "learning_rate": 0.0001878513278386152,
      "loss": 12.644,
      "step": 6750
    },
    {
      "epoch": 0.36761893093191206,
      "grad_norm": 0.6563415578448006,
      "learning_rate": 0.0001878471148570196,
      "loss": 12.4376,
      "step": 6751
    },
    {
      "epoch": 0.3676733849284951,
      "grad_norm": 0.7432858762702095,
      "learning_rate": 0.00018784290119231014,
      "loss": 12.5806,
      "step": 6752
    },
    {
      "epoch": 0.3677278389250781,
      "grad_norm": 0.6496962666478835,
      "learning_rate": 0.00018783868684451953,
      "loss": 12.4287,
      "step": 6753
    },
    {
      "epoch": 0.3677822929216611,
      "grad_norm": 0.5991457432589863,
      "learning_rate": 0.00018783447181368058,
      "loss": 12.4827,
      "step": 6754
    },
    {
      "epoch": 0.36783674691824414,
      "grad_norm": 0.8446082627576676,
      "learning_rate": 0.00018783025609982602,
      "loss": 12.4579,
      "step": 6755
    },
    {
      "epoch": 0.36789120091482713,
      "grad_norm": 0.6363909276299062,
      "learning_rate": 0.00018782603970298869,
      "loss": 12.4383,
      "step": 6756
    },
    {
      "epoch": 0.3679456549114102,
      "grad_norm": 0.6666559724643025,
      "learning_rate": 0.00018782182262320132,
      "loss": 12.6553,
      "step": 6757
    },
    {
      "epoch": 0.3680001089079932,
      "grad_norm": 0.6768404315222415,
      "learning_rate": 0.00018781760486049674,
      "loss": 12.5123,
      "step": 6758
    },
    {
      "epoch": 0.36805456290457617,
      "grad_norm": 0.6747314018648729,
      "learning_rate": 0.00018781338641490772,
      "loss": 12.5436,
      "step": 6759
    },
    {
      "epoch": 0.3681090169011592,
      "grad_norm": 0.7112379071050334,
      "learning_rate": 0.0001878091672864671,
      "loss": 12.5604,
      "step": 6760
    },
    {
      "epoch": 0.3681634708977422,
      "grad_norm": 0.7209862155449003,
      "learning_rate": 0.00018780494747520766,
      "loss": 12.4318,
      "step": 6761
    },
    {
      "epoch": 0.3682179248943252,
      "grad_norm": 0.6681822817895532,
      "learning_rate": 0.00018780072698116224,
      "loss": 12.5159,
      "step": 6762
    },
    {
      "epoch": 0.36827237889090825,
      "grad_norm": 0.6863679955143484,
      "learning_rate": 0.00018779650580436362,
      "loss": 12.5072,
      "step": 6763
    },
    {
      "epoch": 0.36832683288749124,
      "grad_norm": 0.6044769521952945,
      "learning_rate": 0.00018779228394484468,
      "loss": 12.4695,
      "step": 6764
    },
    {
      "epoch": 0.36838128688407423,
      "grad_norm": 0.6071552219295997,
      "learning_rate": 0.0001877880614026382,
      "loss": 12.4761,
      "step": 6765
    },
    {
      "epoch": 0.3684357408806573,
      "grad_norm": 0.7369894755682067,
      "learning_rate": 0.00018778383817777704,
      "loss": 12.614,
      "step": 6766
    },
    {
      "epoch": 0.3684901948772403,
      "grad_norm": 0.6885261413499081,
      "learning_rate": 0.000187779614270294,
      "loss": 12.6022,
      "step": 6767
    },
    {
      "epoch": 0.36854464887382327,
      "grad_norm": 0.6742656846843884,
      "learning_rate": 0.00018777538968022198,
      "loss": 12.3581,
      "step": 6768
    },
    {
      "epoch": 0.3685991028704063,
      "grad_norm": 0.7097374898586997,
      "learning_rate": 0.0001877711644075938,
      "loss": 12.4439,
      "step": 6769
    },
    {
      "epoch": 0.3686535568669893,
      "grad_norm": 0.7041533992553495,
      "learning_rate": 0.00018776693845244235,
      "loss": 12.5134,
      "step": 6770
    },
    {
      "epoch": 0.3687080108635723,
      "grad_norm": 0.6803788854145529,
      "learning_rate": 0.00018776271181480047,
      "loss": 12.5001,
      "step": 6771
    },
    {
      "epoch": 0.36876246486015535,
      "grad_norm": 0.6677003831250875,
      "learning_rate": 0.000187758484494701,
      "loss": 12.4216,
      "step": 6772
    },
    {
      "epoch": 0.36881691885673834,
      "grad_norm": 1.0267058549178372,
      "learning_rate": 0.00018775425649217685,
      "loss": 12.4736,
      "step": 6773
    },
    {
      "epoch": 0.36887137285332133,
      "grad_norm": 0.7316182162253633,
      "learning_rate": 0.0001877500278072609,
      "loss": 12.399,
      "step": 6774
    },
    {
      "epoch": 0.3689258268499044,
      "grad_norm": 0.6166976637914323,
      "learning_rate": 0.000187745798439986,
      "loss": 12.5217,
      "step": 6775
    },
    {
      "epoch": 0.3689802808464874,
      "grad_norm": 0.677681470141228,
      "learning_rate": 0.00018774156839038506,
      "loss": 12.3609,
      "step": 6776
    },
    {
      "epoch": 0.36903473484307037,
      "grad_norm": 0.6906842272517986,
      "learning_rate": 0.00018773733765849095,
      "loss": 12.4438,
      "step": 6777
    },
    {
      "epoch": 0.3690891888396534,
      "grad_norm": 0.636827348006488,
      "learning_rate": 0.0001877331062443366,
      "loss": 12.6218,
      "step": 6778
    },
    {
      "epoch": 0.3691436428362364,
      "grad_norm": 0.6668917508706977,
      "learning_rate": 0.00018772887414795494,
      "loss": 12.4354,
      "step": 6779
    },
    {
      "epoch": 0.3691980968328194,
      "grad_norm": 0.7166015009254231,
      "learning_rate": 0.00018772464136937878,
      "loss": 12.475,
      "step": 6780
    },
    {
      "epoch": 0.36925255082940245,
      "grad_norm": 0.5999520292287912,
      "learning_rate": 0.00018772040790864116,
      "loss": 12.498,
      "step": 6781
    },
    {
      "epoch": 0.36930700482598544,
      "grad_norm": 0.7259940123210213,
      "learning_rate": 0.00018771617376577487,
      "loss": 12.5116,
      "step": 6782
    },
    {
      "epoch": 0.36936145882256843,
      "grad_norm": 0.7414603041330218,
      "learning_rate": 0.00018771193894081295,
      "loss": 12.4914,
      "step": 6783
    },
    {
      "epoch": 0.3694159128191515,
      "grad_norm": 0.6098300155081475,
      "learning_rate": 0.00018770770343378828,
      "loss": 12.3549,
      "step": 6784
    },
    {
      "epoch": 0.3694703668157345,
      "grad_norm": 0.6771752107026059,
      "learning_rate": 0.0001877034672447338,
      "loss": 12.4367,
      "step": 6785
    },
    {
      "epoch": 0.36952482081231747,
      "grad_norm": 0.6127061024770842,
      "learning_rate": 0.00018769923037368244,
      "loss": 12.5099,
      "step": 6786
    },
    {
      "epoch": 0.3695792748089005,
      "grad_norm": 0.6713858656002362,
      "learning_rate": 0.00018769499282066717,
      "loss": 12.5197,
      "step": 6787
    },
    {
      "epoch": 0.3696337288054835,
      "grad_norm": 0.6113076049001404,
      "learning_rate": 0.0001876907545857209,
      "loss": 12.4514,
      "step": 6788
    },
    {
      "epoch": 0.36968818280206656,
      "grad_norm": 0.69200188488012,
      "learning_rate": 0.00018768651566887664,
      "loss": 12.5427,
      "step": 6789
    },
    {
      "epoch": 0.36974263679864955,
      "grad_norm": 0.649783035094314,
      "learning_rate": 0.00018768227607016735,
      "loss": 12.5515,
      "step": 6790
    },
    {
      "epoch": 0.36979709079523254,
      "grad_norm": 0.651731885718608,
      "learning_rate": 0.00018767803578962594,
      "loss": 12.3617,
      "step": 6791
    },
    {
      "epoch": 0.3698515447918156,
      "grad_norm": 0.6236780373250338,
      "learning_rate": 0.00018767379482728544,
      "loss": 12.4804,
      "step": 6792
    },
    {
      "epoch": 0.3699059987883986,
      "grad_norm": 0.6290590856325058,
      "learning_rate": 0.00018766955318317877,
      "loss": 12.5694,
      "step": 6793
    },
    {
      "epoch": 0.3699604527849816,
      "grad_norm": 0.994191559529947,
      "learning_rate": 0.000187665310857339,
      "loss": 12.3978,
      "step": 6794
    },
    {
      "epoch": 0.3700149067815646,
      "grad_norm": 0.6691599006801765,
      "learning_rate": 0.00018766106784979907,
      "loss": 12.5427,
      "step": 6795
    },
    {
      "epoch": 0.3700693607781476,
      "grad_norm": 0.6725550264378077,
      "learning_rate": 0.00018765682416059195,
      "loss": 12.5197,
      "step": 6796
    },
    {
      "epoch": 0.3701238147747306,
      "grad_norm": 0.7775858640825722,
      "learning_rate": 0.00018765257978975067,
      "loss": 12.565,
      "step": 6797
    },
    {
      "epoch": 0.37017826877131366,
      "grad_norm": 0.7820371214677028,
      "learning_rate": 0.00018764833473730824,
      "loss": 12.5492,
      "step": 6798
    },
    {
      "epoch": 0.37023272276789665,
      "grad_norm": 0.6070673240639204,
      "learning_rate": 0.00018764408900329767,
      "loss": 12.3199,
      "step": 6799
    },
    {
      "epoch": 0.37028717676447964,
      "grad_norm": 0.7546031888907322,
      "learning_rate": 0.00018763984258775195,
      "loss": 12.4831,
      "step": 6800
    },
    {
      "epoch": 0.3703416307610627,
      "grad_norm": 0.6614244889442318,
      "learning_rate": 0.00018763559549070413,
      "loss": 12.4839,
      "step": 6801
    },
    {
      "epoch": 0.3703960847576457,
      "grad_norm": 0.66700405636965,
      "learning_rate": 0.0001876313477121872,
      "loss": 12.444,
      "step": 6802
    },
    {
      "epoch": 0.3704505387542287,
      "grad_norm": 0.6430092498863155,
      "learning_rate": 0.00018762709925223422,
      "loss": 12.5453,
      "step": 6803
    },
    {
      "epoch": 0.3705049927508117,
      "grad_norm": 0.6240645971741429,
      "learning_rate": 0.00018762285011087823,
      "loss": 12.5121,
      "step": 6804
    },
    {
      "epoch": 0.3705594467473947,
      "grad_norm": 0.7551045479724068,
      "learning_rate": 0.00018761860028815227,
      "loss": 12.4146,
      "step": 6805
    },
    {
      "epoch": 0.3706139007439777,
      "grad_norm": 0.6744163269445749,
      "learning_rate": 0.00018761434978408937,
      "loss": 12.4616,
      "step": 6806
    },
    {
      "epoch": 0.37066835474056076,
      "grad_norm": 0.6083501446794017,
      "learning_rate": 0.00018761009859872259,
      "loss": 12.4124,
      "step": 6807
    },
    {
      "epoch": 0.37072280873714375,
      "grad_norm": 0.6741727963074209,
      "learning_rate": 0.000187605846732085,
      "loss": 12.4202,
      "step": 6808
    },
    {
      "epoch": 0.37077726273372674,
      "grad_norm": 0.6277622677085708,
      "learning_rate": 0.00018760159418420967,
      "loss": 12.482,
      "step": 6809
    },
    {
      "epoch": 0.3708317167303098,
      "grad_norm": 0.6334500184621313,
      "learning_rate": 0.00018759734095512962,
      "loss": 12.4696,
      "step": 6810
    },
    {
      "epoch": 0.3708861707268928,
      "grad_norm": 0.7533923247549392,
      "learning_rate": 0.00018759308704487796,
      "loss": 12.5491,
      "step": 6811
    },
    {
      "epoch": 0.3709406247234758,
      "grad_norm": 0.6355387749570965,
      "learning_rate": 0.0001875888324534878,
      "loss": 12.4132,
      "step": 6812
    },
    {
      "epoch": 0.3709950787200588,
      "grad_norm": 0.6199304004384589,
      "learning_rate": 0.00018758457718099213,
      "loss": 12.4855,
      "step": 6813
    },
    {
      "epoch": 0.3710495327166418,
      "grad_norm": 0.6562595002491339,
      "learning_rate": 0.00018758032122742415,
      "loss": 12.5171,
      "step": 6814
    },
    {
      "epoch": 0.3711039867132248,
      "grad_norm": 0.7378677389636937,
      "learning_rate": 0.0001875760645928169,
      "loss": 12.6526,
      "step": 6815
    },
    {
      "epoch": 0.37115844070980786,
      "grad_norm": 0.7079173817993047,
      "learning_rate": 0.00018757180727720348,
      "loss": 12.5506,
      "step": 6816
    },
    {
      "epoch": 0.37121289470639085,
      "grad_norm": 0.7743100219942269,
      "learning_rate": 0.000187567549280617,
      "loss": 12.42,
      "step": 6817
    },
    {
      "epoch": 0.37126734870297384,
      "grad_norm": 0.6741818318526057,
      "learning_rate": 0.00018756329060309055,
      "loss": 12.5836,
      "step": 6818
    },
    {
      "epoch": 0.3713218026995569,
      "grad_norm": 0.7489878835475556,
      "learning_rate": 0.0001875590312446573,
      "loss": 12.5678,
      "step": 6819
    },
    {
      "epoch": 0.3713762566961399,
      "grad_norm": 0.7337700505792882,
      "learning_rate": 0.0001875547712053503,
      "loss": 12.3313,
      "step": 6820
    },
    {
      "epoch": 0.3714307106927229,
      "grad_norm": 0.643565639957734,
      "learning_rate": 0.00018755051048520275,
      "loss": 12.3265,
      "step": 6821
    },
    {
      "epoch": 0.3714851646893059,
      "grad_norm": 0.7085073729411158,
      "learning_rate": 0.00018754624908424777,
      "loss": 12.5107,
      "step": 6822
    },
    {
      "epoch": 0.3715396186858889,
      "grad_norm": 0.7446332347373452,
      "learning_rate": 0.00018754198700251842,
      "loss": 12.5754,
      "step": 6823
    },
    {
      "epoch": 0.37159407268247197,
      "grad_norm": 0.7543591298912838,
      "learning_rate": 0.00018753772424004791,
      "loss": 12.5493,
      "step": 6824
    },
    {
      "epoch": 0.37164852667905496,
      "grad_norm": 0.779747215779012,
      "learning_rate": 0.00018753346079686942,
      "loss": 12.6352,
      "step": 6825
    },
    {
      "epoch": 0.37170298067563795,
      "grad_norm": 0.6777786449446311,
      "learning_rate": 0.00018752919667301603,
      "loss": 12.4106,
      "step": 6826
    },
    {
      "epoch": 0.371757434672221,
      "grad_norm": 0.7842274548150086,
      "learning_rate": 0.0001875249318685209,
      "loss": 12.545,
      "step": 6827
    },
    {
      "epoch": 0.371811888668804,
      "grad_norm": 0.6771490065998343,
      "learning_rate": 0.00018752066638341724,
      "loss": 12.4642,
      "step": 6828
    },
    {
      "epoch": 0.371866342665387,
      "grad_norm": 0.6722420889789897,
      "learning_rate": 0.00018751640021773822,
      "loss": 12.4239,
      "step": 6829
    },
    {
      "epoch": 0.37192079666197003,
      "grad_norm": 0.7656829797000084,
      "learning_rate": 0.00018751213337151699,
      "loss": 12.4541,
      "step": 6830
    },
    {
      "epoch": 0.371975250658553,
      "grad_norm": 0.6173941419610085,
      "learning_rate": 0.00018750786584478674,
      "loss": 12.5112,
      "step": 6831
    },
    {
      "epoch": 0.372029704655136,
      "grad_norm": 0.6862551318512679,
      "learning_rate": 0.00018750359763758064,
      "loss": 12.5843,
      "step": 6832
    },
    {
      "epoch": 0.37208415865171907,
      "grad_norm": 0.7438790929735515,
      "learning_rate": 0.00018749932874993191,
      "loss": 12.3711,
      "step": 6833
    },
    {
      "epoch": 0.37213861264830206,
      "grad_norm": 0.7148570406597187,
      "learning_rate": 0.00018749505918187368,
      "loss": 12.6131,
      "step": 6834
    },
    {
      "epoch": 0.37219306664488505,
      "grad_norm": 0.6099944499008384,
      "learning_rate": 0.00018749078893343923,
      "loss": 12.4924,
      "step": 6835
    },
    {
      "epoch": 0.3722475206414681,
      "grad_norm": 0.6416494040504213,
      "learning_rate": 0.00018748651800466176,
      "loss": 12.4666,
      "step": 6836
    },
    {
      "epoch": 0.3723019746380511,
      "grad_norm": 0.5960517211523131,
      "learning_rate": 0.0001874822463955744,
      "loss": 12.4964,
      "step": 6837
    },
    {
      "epoch": 0.3723564286346341,
      "grad_norm": 0.7202137666628853,
      "learning_rate": 0.00018747797410621043,
      "loss": 12.5145,
      "step": 6838
    },
    {
      "epoch": 0.37241088263121713,
      "grad_norm": 0.6686744255372455,
      "learning_rate": 0.0001874737011366031,
      "loss": 12.509,
      "step": 6839
    },
    {
      "epoch": 0.3724653366278001,
      "grad_norm": 0.5923281634322939,
      "learning_rate": 0.00018746942748678556,
      "loss": 12.3796,
      "step": 6840
    },
    {
      "epoch": 0.3725197906243831,
      "grad_norm": 0.6462348839480198,
      "learning_rate": 0.00018746515315679112,
      "loss": 12.4071,
      "step": 6841
    },
    {
      "epoch": 0.37257424462096617,
      "grad_norm": 0.6403103417823736,
      "learning_rate": 0.00018746087814665297,
      "loss": 12.3721,
      "step": 6842
    },
    {
      "epoch": 0.37262869861754916,
      "grad_norm": 0.6452511968744338,
      "learning_rate": 0.00018745660245640433,
      "loss": 12.4392,
      "step": 6843
    },
    {
      "epoch": 0.37268315261413215,
      "grad_norm": 0.8042292528683954,
      "learning_rate": 0.0001874523260860785,
      "loss": 12.56,
      "step": 6844
    },
    {
      "epoch": 0.3727376066107152,
      "grad_norm": 0.6047385552048565,
      "learning_rate": 0.00018744804903570873,
      "loss": 12.4637,
      "step": 6845
    },
    {
      "epoch": 0.3727920606072982,
      "grad_norm": 0.6150113789305852,
      "learning_rate": 0.00018744377130532826,
      "loss": 12.4973,
      "step": 6846
    },
    {
      "epoch": 0.3728465146038812,
      "grad_norm": 0.6082391745298686,
      "learning_rate": 0.00018743949289497035,
      "loss": 12.4028,
      "step": 6847
    },
    {
      "epoch": 0.37290096860046423,
      "grad_norm": 0.7418134326699819,
      "learning_rate": 0.00018743521380466832,
      "loss": 12.5765,
      "step": 6848
    },
    {
      "epoch": 0.3729554225970472,
      "grad_norm": 0.6251421328537706,
      "learning_rate": 0.00018743093403445537,
      "loss": 12.3914,
      "step": 6849
    },
    {
      "epoch": 0.3730098765936302,
      "grad_norm": 0.6643823954217485,
      "learning_rate": 0.00018742665358436483,
      "loss": 12.576,
      "step": 6850
    },
    {
      "epoch": 0.37306433059021327,
      "grad_norm": 0.6821878946854963,
      "learning_rate": 0.00018742237245442995,
      "loss": 12.4979,
      "step": 6851
    },
    {
      "epoch": 0.37311878458679626,
      "grad_norm": 0.6284956737925181,
      "learning_rate": 0.00018741809064468402,
      "loss": 12.3956,
      "step": 6852
    },
    {
      "epoch": 0.37317323858337925,
      "grad_norm": 0.5928727082407248,
      "learning_rate": 0.0001874138081551604,
      "loss": 12.3172,
      "step": 6853
    },
    {
      "epoch": 0.3732276925799623,
      "grad_norm": 0.7075238356745377,
      "learning_rate": 0.00018740952498589236,
      "loss": 12.5216,
      "step": 6854
    },
    {
      "epoch": 0.3732821465765453,
      "grad_norm": 0.8036941645145784,
      "learning_rate": 0.00018740524113691314,
      "loss": 12.7216,
      "step": 6855
    },
    {
      "epoch": 0.37333660057312834,
      "grad_norm": 0.6723133376471627,
      "learning_rate": 0.00018740095660825615,
      "loss": 12.4274,
      "step": 6856
    },
    {
      "epoch": 0.37339105456971133,
      "grad_norm": 0.691765799487332,
      "learning_rate": 0.00018739667139995464,
      "loss": 12.5515,
      "step": 6857
    },
    {
      "epoch": 0.3734455085662943,
      "grad_norm": 0.6374179067955774,
      "learning_rate": 0.00018739238551204198,
      "loss": 12.3448,
      "step": 6858
    },
    {
      "epoch": 0.3734999625628774,
      "grad_norm": 0.736793700634577,
      "learning_rate": 0.00018738809894455147,
      "loss": 12.4875,
      "step": 6859
    },
    {
      "epoch": 0.37355441655946037,
      "grad_norm": 0.714006602323917,
      "learning_rate": 0.00018738381169751644,
      "loss": 12.5406,
      "step": 6860
    },
    {
      "epoch": 0.37360887055604336,
      "grad_norm": 0.7584859501030283,
      "learning_rate": 0.00018737952377097025,
      "loss": 12.4817,
      "step": 6861
    },
    {
      "epoch": 0.3736633245526264,
      "grad_norm": 0.6704402496309448,
      "learning_rate": 0.0001873752351649462,
      "loss": 12.4883,
      "step": 6862
    },
    {
      "epoch": 0.3737177785492094,
      "grad_norm": 0.6354882856330369,
      "learning_rate": 0.0001873709458794777,
      "loss": 12.3731,
      "step": 6863
    },
    {
      "epoch": 0.3737722325457924,
      "grad_norm": 0.7091904296222025,
      "learning_rate": 0.0001873666559145981,
      "loss": 12.5834,
      "step": 6864
    },
    {
      "epoch": 0.37382668654237544,
      "grad_norm": 0.7834726155353686,
      "learning_rate": 0.00018736236527034067,
      "loss": 12.557,
      "step": 6865
    },
    {
      "epoch": 0.37388114053895843,
      "grad_norm": 0.6743082186571374,
      "learning_rate": 0.00018735807394673883,
      "loss": 12.4709,
      "step": 6866
    },
    {
      "epoch": 0.3739355945355414,
      "grad_norm": 0.6070605579786073,
      "learning_rate": 0.000187353781943826,
      "loss": 12.37,
      "step": 6867
    },
    {
      "epoch": 0.3739900485321245,
      "grad_norm": 0.6865918490759351,
      "learning_rate": 0.0001873494892616355,
      "loss": 12.5524,
      "step": 6868
    },
    {
      "epoch": 0.37404450252870747,
      "grad_norm": 0.662629494210567,
      "learning_rate": 0.00018734519590020071,
      "loss": 12.553,
      "step": 6869
    },
    {
      "epoch": 0.37409895652529046,
      "grad_norm": 0.6170089067050327,
      "learning_rate": 0.00018734090185955503,
      "loss": 12.4318,
      "step": 6870
    },
    {
      "epoch": 0.3741534105218735,
      "grad_norm": 0.6673199828990071,
      "learning_rate": 0.00018733660713973188,
      "loss": 12.344,
      "step": 6871
    },
    {
      "epoch": 0.3742078645184565,
      "grad_norm": 0.9030032262663705,
      "learning_rate": 0.0001873323117407646,
      "loss": 12.4672,
      "step": 6872
    },
    {
      "epoch": 0.3742623185150395,
      "grad_norm": 0.6476698207148913,
      "learning_rate": 0.00018732801566268662,
      "loss": 12.4269,
      "step": 6873
    },
    {
      "epoch": 0.37431677251162254,
      "grad_norm": 0.6389493481839894,
      "learning_rate": 0.00018732371890553136,
      "loss": 12.3736,
      "step": 6874
    },
    {
      "epoch": 0.37437122650820553,
      "grad_norm": 0.6581260506637848,
      "learning_rate": 0.0001873194214693322,
      "loss": 12.5084,
      "step": 6875
    },
    {
      "epoch": 0.3744256805047885,
      "grad_norm": 0.7331480328607267,
      "learning_rate": 0.0001873151233541226,
      "loss": 12.4586,
      "step": 6876
    },
    {
      "epoch": 0.3744801345013716,
      "grad_norm": 0.6757682925411165,
      "learning_rate": 0.00018731082455993595,
      "loss": 12.5669,
      "step": 6877
    },
    {
      "epoch": 0.37453458849795457,
      "grad_norm": 0.7647719878421511,
      "learning_rate": 0.00018730652508680567,
      "loss": 12.6505,
      "step": 6878
    },
    {
      "epoch": 0.37458904249453756,
      "grad_norm": 0.8106155568072291,
      "learning_rate": 0.0001873022249347652,
      "loss": 12.5532,
      "step": 6879
    },
    {
      "epoch": 0.3746434964911206,
      "grad_norm": 0.6475029617687691,
      "learning_rate": 0.000187297924103848,
      "loss": 12.3556,
      "step": 6880
    },
    {
      "epoch": 0.3746979504877036,
      "grad_norm": 0.6906212964994336,
      "learning_rate": 0.0001872936225940875,
      "loss": 12.5516,
      "step": 6881
    },
    {
      "epoch": 0.3747524044842866,
      "grad_norm": 0.8103337070967035,
      "learning_rate": 0.00018728932040551718,
      "loss": 12.4691,
      "step": 6882
    },
    {
      "epoch": 0.37480685848086964,
      "grad_norm": 0.6576683290497881,
      "learning_rate": 0.00018728501753817044,
      "loss": 12.5408,
      "step": 6883
    },
    {
      "epoch": 0.37486131247745264,
      "grad_norm": 0.7051994294658306,
      "learning_rate": 0.00018728071399208077,
      "loss": 12.342,
      "step": 6884
    },
    {
      "epoch": 0.3749157664740356,
      "grad_norm": 0.6795758079567125,
      "learning_rate": 0.00018727640976728163,
      "loss": 12.5536,
      "step": 6885
    },
    {
      "epoch": 0.3749702204706187,
      "grad_norm": 0.7042609974729568,
      "learning_rate": 0.00018727210486380649,
      "loss": 12.5584,
      "step": 6886
    },
    {
      "epoch": 0.37502467446720167,
      "grad_norm": 0.6567415913122908,
      "learning_rate": 0.00018726779928168882,
      "loss": 12.6032,
      "step": 6887
    },
    {
      "epoch": 0.37507912846378466,
      "grad_norm": 0.6822256992051228,
      "learning_rate": 0.00018726349302096212,
      "loss": 12.522,
      "step": 6888
    },
    {
      "epoch": 0.3751335824603677,
      "grad_norm": 0.7086191085604834,
      "learning_rate": 0.00018725918608165988,
      "loss": 12.5549,
      "step": 6889
    },
    {
      "epoch": 0.3751880364569507,
      "grad_norm": 0.7026318941518679,
      "learning_rate": 0.00018725487846381556,
      "loss": 12.498,
      "step": 6890
    },
    {
      "epoch": 0.37524249045353375,
      "grad_norm": 0.6746306298341949,
      "learning_rate": 0.0001872505701674627,
      "loss": 12.5457,
      "step": 6891
    },
    {
      "epoch": 0.37529694445011674,
      "grad_norm": 0.9491430206384743,
      "learning_rate": 0.0001872462611926347,
      "loss": 12.5156,
      "step": 6892
    },
    {
      "epoch": 0.37535139844669974,
      "grad_norm": 0.6081121088069118,
      "learning_rate": 0.0001872419515393652,
      "loss": 12.3423,
      "step": 6893
    },
    {
      "epoch": 0.3754058524432828,
      "grad_norm": 0.6289634087585562,
      "learning_rate": 0.00018723764120768762,
      "loss": 12.6402,
      "step": 6894
    },
    {
      "epoch": 0.3754603064398658,
      "grad_norm": 0.766105858999642,
      "learning_rate": 0.00018723333019763554,
      "loss": 12.333,
      "step": 6895
    },
    {
      "epoch": 0.37551476043644877,
      "grad_norm": 0.7784877505068041,
      "learning_rate": 0.00018722901850924247,
      "loss": 12.5083,
      "step": 6896
    },
    {
      "epoch": 0.3755692144330318,
      "grad_norm": 0.6293072113676585,
      "learning_rate": 0.0001872247061425419,
      "loss": 12.4921,
      "step": 6897
    },
    {
      "epoch": 0.3756236684296148,
      "grad_norm": 0.6539959921320518,
      "learning_rate": 0.00018722039309756737,
      "loss": 12.488,
      "step": 6898
    },
    {
      "epoch": 0.3756781224261978,
      "grad_norm": 0.7197826833160897,
      "learning_rate": 0.00018721607937435247,
      "loss": 12.2857,
      "step": 6899
    },
    {
      "epoch": 0.37573257642278085,
      "grad_norm": 0.7122843009448809,
      "learning_rate": 0.00018721176497293068,
      "loss": 12.4334,
      "step": 6900
    },
    {
      "epoch": 0.37578703041936384,
      "grad_norm": 0.6417828679475482,
      "learning_rate": 0.0001872074498933356,
      "loss": 12.4879,
      "step": 6901
    },
    {
      "epoch": 0.37584148441594684,
      "grad_norm": 0.7345247568517772,
      "learning_rate": 0.00018720313413560078,
      "loss": 12.5805,
      "step": 6902
    },
    {
      "epoch": 0.3758959384125299,
      "grad_norm": 0.6627147722447752,
      "learning_rate": 0.00018719881769975973,
      "loss": 12.4623,
      "step": 6903
    },
    {
      "epoch": 0.3759503924091129,
      "grad_norm": 0.7144649763094523,
      "learning_rate": 0.00018719450058584606,
      "loss": 12.5279,
      "step": 6904
    },
    {
      "epoch": 0.37600484640569587,
      "grad_norm": 0.7627784910727949,
      "learning_rate": 0.00018719018279389336,
      "loss": 12.713,
      "step": 6905
    },
    {
      "epoch": 0.3760593004022789,
      "grad_norm": 0.6574616799381727,
      "learning_rate": 0.00018718586432393515,
      "loss": 12.4512,
      "step": 6906
    },
    {
      "epoch": 0.3761137543988619,
      "grad_norm": 0.6138750975409948,
      "learning_rate": 0.00018718154517600503,
      "loss": 12.5478,
      "step": 6907
    },
    {
      "epoch": 0.3761682083954449,
      "grad_norm": 0.6381407623000527,
      "learning_rate": 0.00018717722535013662,
      "loss": 12.523,
      "step": 6908
    },
    {
      "epoch": 0.37622266239202795,
      "grad_norm": 0.5934615819263432,
      "learning_rate": 0.00018717290484636346,
      "loss": 12.4251,
      "step": 6909
    },
    {
      "epoch": 0.37627711638861094,
      "grad_norm": 0.6521590898875078,
      "learning_rate": 0.00018716858366471918,
      "loss": 12.5358,
      "step": 6910
    },
    {
      "epoch": 0.37633157038519394,
      "grad_norm": 0.7510537780545019,
      "learning_rate": 0.00018716426180523737,
      "loss": 12.4232,
      "step": 6911
    },
    {
      "epoch": 0.376386024381777,
      "grad_norm": 0.6803248714221944,
      "learning_rate": 0.00018715993926795167,
      "loss": 12.5625,
      "step": 6912
    },
    {
      "epoch": 0.37644047837836,
      "grad_norm": 0.6825366462464981,
      "learning_rate": 0.00018715561605289565,
      "loss": 12.4024,
      "step": 6913
    },
    {
      "epoch": 0.37649493237494297,
      "grad_norm": 0.6789555150219736,
      "learning_rate": 0.00018715129216010295,
      "loss": 12.5354,
      "step": 6914
    },
    {
      "epoch": 0.376549386371526,
      "grad_norm": 0.7318822770872114,
      "learning_rate": 0.0001871469675896072,
      "loss": 12.5837,
      "step": 6915
    },
    {
      "epoch": 0.376603840368109,
      "grad_norm": 0.6626231088289419,
      "learning_rate": 0.00018714264234144198,
      "loss": 12.5106,
      "step": 6916
    },
    {
      "epoch": 0.376658294364692,
      "grad_norm": 0.7217653612409614,
      "learning_rate": 0.00018713831641564097,
      "loss": 12.5852,
      "step": 6917
    },
    {
      "epoch": 0.37671274836127505,
      "grad_norm": 0.6263709870042876,
      "learning_rate": 0.0001871339898122378,
      "loss": 12.4448,
      "step": 6918
    },
    {
      "epoch": 0.37676720235785804,
      "grad_norm": 0.6811089588998432,
      "learning_rate": 0.0001871296625312661,
      "loss": 12.5556,
      "step": 6919
    },
    {
      "epoch": 0.37682165635444104,
      "grad_norm": 0.6573727791834589,
      "learning_rate": 0.0001871253345727596,
      "loss": 12.4701,
      "step": 6920
    },
    {
      "epoch": 0.3768761103510241,
      "grad_norm": 0.6684755993242253,
      "learning_rate": 0.00018712100593675182,
      "loss": 12.5007,
      "step": 6921
    },
    {
      "epoch": 0.3769305643476071,
      "grad_norm": 0.6443217936756908,
      "learning_rate": 0.0001871166766232765,
      "loss": 12.4511,
      "step": 6922
    },
    {
      "epoch": 0.3769850183441901,
      "grad_norm": 0.6535774774464812,
      "learning_rate": 0.0001871123466323673,
      "loss": 12.549,
      "step": 6923
    },
    {
      "epoch": 0.3770394723407731,
      "grad_norm": 0.6911624086653004,
      "learning_rate": 0.00018710801596405786,
      "loss": 12.6717,
      "step": 6924
    },
    {
      "epoch": 0.3770939263373561,
      "grad_norm": 0.6649678708702024,
      "learning_rate": 0.0001871036846183819,
      "loss": 12.3266,
      "step": 6925
    },
    {
      "epoch": 0.37714838033393916,
      "grad_norm": 0.6838715546443267,
      "learning_rate": 0.00018709935259537307,
      "loss": 12.5144,
      "step": 6926
    },
    {
      "epoch": 0.37720283433052215,
      "grad_norm": 0.6701906131016372,
      "learning_rate": 0.00018709501989506508,
      "loss": 12.6125,
      "step": 6927
    },
    {
      "epoch": 0.37725728832710514,
      "grad_norm": 0.6492567726196441,
      "learning_rate": 0.00018709068651749162,
      "loss": 12.5389,
      "step": 6928
    },
    {
      "epoch": 0.3773117423236882,
      "grad_norm": 0.7128276988468376,
      "learning_rate": 0.0001870863524626864,
      "loss": 12.5214,
      "step": 6929
    },
    {
      "epoch": 0.3773661963202712,
      "grad_norm": 0.7077479095340496,
      "learning_rate": 0.00018708201773068303,
      "loss": 12.5846,
      "step": 6930
    },
    {
      "epoch": 0.3774206503168542,
      "grad_norm": 0.6949116410713745,
      "learning_rate": 0.00018707768232151533,
      "loss": 12.6226,
      "step": 6931
    },
    {
      "epoch": 0.3774751043134372,
      "grad_norm": 0.6231198852264871,
      "learning_rate": 0.00018707334623521696,
      "loss": 12.3808,
      "step": 6932
    },
    {
      "epoch": 0.3775295583100202,
      "grad_norm": 0.6374499503869536,
      "learning_rate": 0.00018706900947182165,
      "loss": 12.378,
      "step": 6933
    },
    {
      "epoch": 0.3775840123066032,
      "grad_norm": 0.6319859637637218,
      "learning_rate": 0.00018706467203136312,
      "loss": 12.5027,
      "step": 6934
    },
    {
      "epoch": 0.37763846630318626,
      "grad_norm": 0.6557296163600854,
      "learning_rate": 0.0001870603339138751,
      "loss": 12.4791,
      "step": 6935
    },
    {
      "epoch": 0.37769292029976925,
      "grad_norm": 0.7869772154386102,
      "learning_rate": 0.0001870559951193913,
      "loss": 12.4481,
      "step": 6936
    },
    {
      "epoch": 0.37774737429635225,
      "grad_norm": 0.6653265908844587,
      "learning_rate": 0.0001870516556479455,
      "loss": 12.4008,
      "step": 6937
    },
    {
      "epoch": 0.3778018282929353,
      "grad_norm": 0.6172815715116846,
      "learning_rate": 0.00018704731549957146,
      "loss": 12.3995,
      "step": 6938
    },
    {
      "epoch": 0.3778562822895183,
      "grad_norm": 0.6468767865845889,
      "learning_rate": 0.00018704297467430286,
      "loss": 12.5116,
      "step": 6939
    },
    {
      "epoch": 0.3779107362861013,
      "grad_norm": 0.759668058569879,
      "learning_rate": 0.00018703863317217349,
      "loss": 12.3574,
      "step": 6940
    },
    {
      "epoch": 0.3779651902826843,
      "grad_norm": 0.7262095640332545,
      "learning_rate": 0.0001870342909932171,
      "loss": 12.5853,
      "step": 6941
    },
    {
      "epoch": 0.3780196442792673,
      "grad_norm": 0.7033921882821391,
      "learning_rate": 0.0001870299481374675,
      "loss": 12.4206,
      "step": 6942
    },
    {
      "epoch": 0.3780740982758503,
      "grad_norm": 0.6548268998794594,
      "learning_rate": 0.00018702560460495844,
      "loss": 12.4,
      "step": 6943
    },
    {
      "epoch": 0.37812855227243336,
      "grad_norm": 0.7040645449818776,
      "learning_rate": 0.00018702126039572364,
      "loss": 12.5144,
      "step": 6944
    },
    {
      "epoch": 0.37818300626901635,
      "grad_norm": 0.619790631113332,
      "learning_rate": 0.00018701691550979696,
      "loss": 12.4999,
      "step": 6945
    },
    {
      "epoch": 0.37823746026559935,
      "grad_norm": 0.6725003511870317,
      "learning_rate": 0.00018701256994721214,
      "loss": 12.4599,
      "step": 6946
    },
    {
      "epoch": 0.3782919142621824,
      "grad_norm": 0.833030015569806,
      "learning_rate": 0.000187008223708003,
      "loss": 12.405,
      "step": 6947
    },
    {
      "epoch": 0.3783463682587654,
      "grad_norm": 0.6901930567573182,
      "learning_rate": 0.00018700387679220328,
      "loss": 12.4261,
      "step": 6948
    },
    {
      "epoch": 0.3784008222553484,
      "grad_norm": 0.6502639678681048,
      "learning_rate": 0.00018699952919984684,
      "loss": 12.6263,
      "step": 6949
    },
    {
      "epoch": 0.3784552762519314,
      "grad_norm": 0.8842314476326208,
      "learning_rate": 0.0001869951809309675,
      "loss": 12.7245,
      "step": 6950
    },
    {
      "epoch": 0.3785097302485144,
      "grad_norm": 0.6350870700668845,
      "learning_rate": 0.00018699083198559904,
      "loss": 12.4577,
      "step": 6951
    },
    {
      "epoch": 0.3785641842450974,
      "grad_norm": 0.6456154832906754,
      "learning_rate": 0.00018698648236377524,
      "loss": 12.4396,
      "step": 6952
    },
    {
      "epoch": 0.37861863824168046,
      "grad_norm": 0.6606268237260083,
      "learning_rate": 0.00018698213206553001,
      "loss": 12.5647,
      "step": 6953
    },
    {
      "epoch": 0.37867309223826345,
      "grad_norm": 0.6516744393667526,
      "learning_rate": 0.00018697778109089713,
      "loss": 12.4016,
      "step": 6954
    },
    {
      "epoch": 0.37872754623484645,
      "grad_norm": 0.627243644658615,
      "learning_rate": 0.00018697342943991042,
      "loss": 12.4443,
      "step": 6955
    },
    {
      "epoch": 0.3787820002314295,
      "grad_norm": 0.5718938120004171,
      "learning_rate": 0.00018696907711260373,
      "loss": 12.3243,
      "step": 6956
    },
    {
      "epoch": 0.3788364542280125,
      "grad_norm": 0.6304957153919665,
      "learning_rate": 0.00018696472410901092,
      "loss": 12.5637,
      "step": 6957
    },
    {
      "epoch": 0.37889090822459554,
      "grad_norm": 0.6404029072045193,
      "learning_rate": 0.00018696037042916582,
      "loss": 12.4211,
      "step": 6958
    },
    {
      "epoch": 0.3789453622211785,
      "grad_norm": 0.5913884159550243,
      "learning_rate": 0.00018695601607310233,
      "loss": 12.4747,
      "step": 6959
    },
    {
      "epoch": 0.3789998162177615,
      "grad_norm": 0.697847973853796,
      "learning_rate": 0.00018695166104085425,
      "loss": 12.5801,
      "step": 6960
    },
    {
      "epoch": 0.37905427021434457,
      "grad_norm": 0.6430583472870945,
      "learning_rate": 0.00018694730533245547,
      "loss": 12.5845,
      "step": 6961
    },
    {
      "epoch": 0.37910872421092756,
      "grad_norm": 0.715986131250449,
      "learning_rate": 0.00018694294894793987,
      "loss": 12.4404,
      "step": 6962
    },
    {
      "epoch": 0.37916317820751055,
      "grad_norm": 0.6646346545669394,
      "learning_rate": 0.00018693859188734132,
      "loss": 12.5251,
      "step": 6963
    },
    {
      "epoch": 0.3792176322040936,
      "grad_norm": 0.6636199953215705,
      "learning_rate": 0.00018693423415069372,
      "loss": 12.5536,
      "step": 6964
    },
    {
      "epoch": 0.3792720862006766,
      "grad_norm": 0.688855654533608,
      "learning_rate": 0.00018692987573803088,
      "loss": 12.5833,
      "step": 6965
    },
    {
      "epoch": 0.3793265401972596,
      "grad_norm": 0.7349669171500565,
      "learning_rate": 0.0001869255166493868,
      "loss": 12.3737,
      "step": 6966
    },
    {
      "epoch": 0.37938099419384264,
      "grad_norm": 0.6417685318125261,
      "learning_rate": 0.00018692115688479532,
      "loss": 12.4549,
      "step": 6967
    },
    {
      "epoch": 0.37943544819042563,
      "grad_norm": 0.6182139840504983,
      "learning_rate": 0.00018691679644429034,
      "loss": 12.3984,
      "step": 6968
    },
    {
      "epoch": 0.3794899021870086,
      "grad_norm": 0.6748562831650746,
      "learning_rate": 0.00018691243532790576,
      "loss": 12.5668,
      "step": 6969
    },
    {
      "epoch": 0.37954435618359167,
      "grad_norm": 0.6298528138760133,
      "learning_rate": 0.00018690807353567552,
      "loss": 12.3831,
      "step": 6970
    },
    {
      "epoch": 0.37959881018017466,
      "grad_norm": 0.6123954408760172,
      "learning_rate": 0.00018690371106763355,
      "loss": 12.3804,
      "step": 6971
    },
    {
      "epoch": 0.37965326417675765,
      "grad_norm": 0.6024966858966703,
      "learning_rate": 0.0001868993479238137,
      "loss": 12.4899,
      "step": 6972
    },
    {
      "epoch": 0.3797077181733407,
      "grad_norm": 0.6696906147905399,
      "learning_rate": 0.00018689498410424997,
      "loss": 12.3153,
      "step": 6973
    },
    {
      "epoch": 0.3797621721699237,
      "grad_norm": 0.6218780017682547,
      "learning_rate": 0.00018689061960897626,
      "loss": 12.2652,
      "step": 6974
    },
    {
      "epoch": 0.3798166261665067,
      "grad_norm": 0.6388311787850542,
      "learning_rate": 0.00018688625443802654,
      "loss": 12.4931,
      "step": 6975
    },
    {
      "epoch": 0.37987108016308974,
      "grad_norm": 0.6041740522690894,
      "learning_rate": 0.00018688188859143474,
      "loss": 12.3498,
      "step": 6976
    },
    {
      "epoch": 0.37992553415967273,
      "grad_norm": 0.6607235444859734,
      "learning_rate": 0.0001868775220692348,
      "loss": 12.5813,
      "step": 6977
    },
    {
      "epoch": 0.3799799881562557,
      "grad_norm": 0.6593464381004138,
      "learning_rate": 0.00018687315487146065,
      "loss": 12.4265,
      "step": 6978
    },
    {
      "epoch": 0.38003444215283877,
      "grad_norm": 0.6444505430730013,
      "learning_rate": 0.00018686878699814629,
      "loss": 12.4545,
      "step": 6979
    },
    {
      "epoch": 0.38008889614942176,
      "grad_norm": 0.6341830350464173,
      "learning_rate": 0.0001868644184493257,
      "loss": 12.4751,
      "step": 6980
    },
    {
      "epoch": 0.38014335014600475,
      "grad_norm": 0.8000018745026319,
      "learning_rate": 0.0001868600492250328,
      "loss": 12.5567,
      "step": 6981
    },
    {
      "epoch": 0.3801978041425878,
      "grad_norm": 0.6539842272711024,
      "learning_rate": 0.00018685567932530162,
      "loss": 12.5653,
      "step": 6982
    },
    {
      "epoch": 0.3802522581391708,
      "grad_norm": 0.6947964022800268,
      "learning_rate": 0.0001868513087501661,
      "loss": 12.3735,
      "step": 6983
    },
    {
      "epoch": 0.3803067121357538,
      "grad_norm": 0.586767907830935,
      "learning_rate": 0.0001868469374996602,
      "loss": 12.4559,
      "step": 6984
    },
    {
      "epoch": 0.38036116613233684,
      "grad_norm": 0.650386122773557,
      "learning_rate": 0.000186842565573818,
      "loss": 12.3953,
      "step": 6985
    },
    {
      "epoch": 0.38041562012891983,
      "grad_norm": 0.6709438303267575,
      "learning_rate": 0.00018683819297267342,
      "loss": 12.4941,
      "step": 6986
    },
    {
      "epoch": 0.3804700741255028,
      "grad_norm": 0.6318954670614686,
      "learning_rate": 0.0001868338196962605,
      "loss": 12.4621,
      "step": 6987
    },
    {
      "epoch": 0.38052452812208587,
      "grad_norm": 0.5801655542248927,
      "learning_rate": 0.00018682944574461324,
      "loss": 12.2644,
      "step": 6988
    },
    {
      "epoch": 0.38057898211866886,
      "grad_norm": 0.6750535916093292,
      "learning_rate": 0.00018682507111776565,
      "loss": 12.5385,
      "step": 6989
    },
    {
      "epoch": 0.3806334361152519,
      "grad_norm": 0.6481330046989016,
      "learning_rate": 0.00018682069581575173,
      "loss": 12.5772,
      "step": 6990
    },
    {
      "epoch": 0.3806878901118349,
      "grad_norm": 0.6622877449441541,
      "learning_rate": 0.00018681631983860552,
      "loss": 12.4641,
      "step": 6991
    },
    {
      "epoch": 0.3807423441084179,
      "grad_norm": 0.6137140645342416,
      "learning_rate": 0.00018681194318636104,
      "loss": 12.4936,
      "step": 6992
    },
    {
      "epoch": 0.38079679810500094,
      "grad_norm": 0.6082401870926571,
      "learning_rate": 0.00018680756585905234,
      "loss": 12.4662,
      "step": 6993
    },
    {
      "epoch": 0.38085125210158394,
      "grad_norm": 0.6308495684452727,
      "learning_rate": 0.00018680318785671348,
      "loss": 12.6563,
      "step": 6994
    },
    {
      "epoch": 0.38090570609816693,
      "grad_norm": 0.7416546041029876,
      "learning_rate": 0.00018679880917937843,
      "loss": 12.5217,
      "step": 6995
    },
    {
      "epoch": 0.38096016009475,
      "grad_norm": 0.7087242752036145,
      "learning_rate": 0.00018679442982708132,
      "loss": 12.4414,
      "step": 6996
    },
    {
      "epoch": 0.38101461409133297,
      "grad_norm": 0.6176554026255653,
      "learning_rate": 0.00018679004979985615,
      "loss": 12.3972,
      "step": 6997
    },
    {
      "epoch": 0.38106906808791596,
      "grad_norm": 0.6720561068355646,
      "learning_rate": 0.00018678566909773698,
      "loss": 12.517,
      "step": 6998
    },
    {
      "epoch": 0.381123522084499,
      "grad_norm": 0.7147146419691247,
      "learning_rate": 0.00018678128772075793,
      "loss": 12.4048,
      "step": 6999
    },
    {
      "epoch": 0.381177976081082,
      "grad_norm": 0.8218624835163882,
      "learning_rate": 0.00018677690566895302,
      "loss": 12.3689,
      "step": 7000
    },
    {
      "epoch": 0.381232430077665,
      "grad_norm": 0.9732985061261495,
      "learning_rate": 0.00018677252294235634,
      "loss": 12.5755,
      "step": 7001
    },
    {
      "epoch": 0.38128688407424804,
      "grad_norm": 0.7098918965887916,
      "learning_rate": 0.00018676813954100196,
      "loss": 12.4009,
      "step": 7002
    },
    {
      "epoch": 0.38134133807083104,
      "grad_norm": 0.7461767152255029,
      "learning_rate": 0.00018676375546492396,
      "loss": 12.57,
      "step": 7003
    },
    {
      "epoch": 0.38139579206741403,
      "grad_norm": 0.8675423218585037,
      "learning_rate": 0.00018675937071415647,
      "loss": 12.3617,
      "step": 7004
    },
    {
      "epoch": 0.3814502460639971,
      "grad_norm": 0.717370084546687,
      "learning_rate": 0.0001867549852887336,
      "loss": 12.5304,
      "step": 7005
    },
    {
      "epoch": 0.38150470006058007,
      "grad_norm": 0.6231210159631485,
      "learning_rate": 0.00018675059918868935,
      "loss": 12.3976,
      "step": 7006
    },
    {
      "epoch": 0.38155915405716306,
      "grad_norm": 0.6817521240331398,
      "learning_rate": 0.00018674621241405792,
      "loss": 12.5481,
      "step": 7007
    },
    {
      "epoch": 0.3816136080537461,
      "grad_norm": 0.7397513221624799,
      "learning_rate": 0.0001867418249648734,
      "loss": 12.6074,
      "step": 7008
    },
    {
      "epoch": 0.3816680620503291,
      "grad_norm": 0.7386122621149642,
      "learning_rate": 0.0001867374368411699,
      "loss": 12.5742,
      "step": 7009
    },
    {
      "epoch": 0.3817225160469121,
      "grad_norm": 0.6308245493328805,
      "learning_rate": 0.00018673304804298156,
      "loss": 12.4646,
      "step": 7010
    },
    {
      "epoch": 0.38177697004349515,
      "grad_norm": 0.6765571167770338,
      "learning_rate": 0.00018672865857034246,
      "loss": 12.391,
      "step": 7011
    },
    {
      "epoch": 0.38183142404007814,
      "grad_norm": 0.6044597658565793,
      "learning_rate": 0.00018672426842328678,
      "loss": 12.4188,
      "step": 7012
    },
    {
      "epoch": 0.38188587803666113,
      "grad_norm": 0.6584114653028906,
      "learning_rate": 0.00018671987760184865,
      "loss": 12.513,
      "step": 7013
    },
    {
      "epoch": 0.3819403320332442,
      "grad_norm": 0.6697607658639393,
      "learning_rate": 0.0001867154861060622,
      "loss": 12.4682,
      "step": 7014
    },
    {
      "epoch": 0.38199478602982717,
      "grad_norm": 0.7005381190713698,
      "learning_rate": 0.00018671109393596157,
      "loss": 12.3883,
      "step": 7015
    },
    {
      "epoch": 0.38204924002641016,
      "grad_norm": 0.6644075644800487,
      "learning_rate": 0.00018670670109158097,
      "loss": 12.5448,
      "step": 7016
    },
    {
      "epoch": 0.3821036940229932,
      "grad_norm": 0.5827707034044087,
      "learning_rate": 0.00018670230757295453,
      "loss": 12.452,
      "step": 7017
    },
    {
      "epoch": 0.3821581480195762,
      "grad_norm": 0.6974112601121403,
      "learning_rate": 0.00018669791338011638,
      "loss": 12.4363,
      "step": 7018
    },
    {
      "epoch": 0.3822126020161592,
      "grad_norm": 0.6065298551477821,
      "learning_rate": 0.00018669351851310074,
      "loss": 12.4149,
      "step": 7019
    },
    {
      "epoch": 0.38226705601274225,
      "grad_norm": 0.649312763578988,
      "learning_rate": 0.0001866891229719417,
      "loss": 12.5342,
      "step": 7020
    },
    {
      "epoch": 0.38232151000932524,
      "grad_norm": 0.6438265673298778,
      "learning_rate": 0.00018668472675667354,
      "loss": 12.4505,
      "step": 7021
    },
    {
      "epoch": 0.38237596400590823,
      "grad_norm": 0.6989175412573948,
      "learning_rate": 0.00018668032986733044,
      "loss": 12.5572,
      "step": 7022
    },
    {
      "epoch": 0.3824304180024913,
      "grad_norm": 0.6689752525509498,
      "learning_rate": 0.0001866759323039465,
      "loss": 12.5695,
      "step": 7023
    },
    {
      "epoch": 0.38248487199907427,
      "grad_norm": 0.6255947812736262,
      "learning_rate": 0.00018667153406655605,
      "loss": 12.4277,
      "step": 7024
    },
    {
      "epoch": 0.3825393259956573,
      "grad_norm": 0.6846448014748062,
      "learning_rate": 0.00018666713515519314,
      "loss": 12.5029,
      "step": 7025
    },
    {
      "epoch": 0.3825937799922403,
      "grad_norm": 0.6256232122015504,
      "learning_rate": 0.0001866627355698921,
      "loss": 12.4771,
      "step": 7026
    },
    {
      "epoch": 0.3826482339888233,
      "grad_norm": 0.6257148059143375,
      "learning_rate": 0.0001866583353106871,
      "loss": 12.4713,
      "step": 7027
    },
    {
      "epoch": 0.38270268798540635,
      "grad_norm": 0.7104753379111266,
      "learning_rate": 0.00018665393437761231,
      "loss": 12.488,
      "step": 7028
    },
    {
      "epoch": 0.38275714198198935,
      "grad_norm": 0.6939144331346863,
      "learning_rate": 0.000186649532770702,
      "loss": 12.5109,
      "step": 7029
    },
    {
      "epoch": 0.38281159597857234,
      "grad_norm": 0.600769214359039,
      "learning_rate": 0.0001866451304899904,
      "loss": 12.4113,
      "step": 7030
    },
    {
      "epoch": 0.3828660499751554,
      "grad_norm": 0.569467143868859,
      "learning_rate": 0.00018664072753551175,
      "loss": 12.2896,
      "step": 7031
    },
    {
      "epoch": 0.3829205039717384,
      "grad_norm": 0.7355576933800445,
      "learning_rate": 0.0001866363239073003,
      "loss": 12.4322,
      "step": 7032
    },
    {
      "epoch": 0.3829749579683214,
      "grad_norm": 0.6909015301874856,
      "learning_rate": 0.00018663191960539022,
      "loss": 12.4023,
      "step": 7033
    },
    {
      "epoch": 0.3830294119649044,
      "grad_norm": 0.7983345034187423,
      "learning_rate": 0.0001866275146298158,
      "loss": 12.4993,
      "step": 7034
    },
    {
      "epoch": 0.3830838659614874,
      "grad_norm": 0.5951362455422403,
      "learning_rate": 0.00018662310898061134,
      "loss": 12.4463,
      "step": 7035
    },
    {
      "epoch": 0.3831383199580704,
      "grad_norm": 0.6606117787777263,
      "learning_rate": 0.00018661870265781103,
      "loss": 12.6477,
      "step": 7036
    },
    {
      "epoch": 0.38319277395465345,
      "grad_norm": 0.6052443542532031,
      "learning_rate": 0.00018661429566144917,
      "loss": 12.3899,
      "step": 7037
    },
    {
      "epoch": 0.38324722795123645,
      "grad_norm": 0.6912374558936648,
      "learning_rate": 0.00018660988799156002,
      "loss": 12.3956,
      "step": 7038
    },
    {
      "epoch": 0.38330168194781944,
      "grad_norm": 0.678500062864124,
      "learning_rate": 0.00018660547964817784,
      "loss": 12.4912,
      "step": 7039
    },
    {
      "epoch": 0.3833561359444025,
      "grad_norm": 0.6451129761422665,
      "learning_rate": 0.00018660107063133693,
      "loss": 12.4821,
      "step": 7040
    },
    {
      "epoch": 0.3834105899409855,
      "grad_norm": 0.7276540261481783,
      "learning_rate": 0.0001865966609410716,
      "loss": 12.5236,
      "step": 7041
    },
    {
      "epoch": 0.3834650439375685,
      "grad_norm": 0.7008677196954861,
      "learning_rate": 0.0001865922505774161,
      "loss": 12.4493,
      "step": 7042
    },
    {
      "epoch": 0.3835194979341515,
      "grad_norm": 0.6704835137426014,
      "learning_rate": 0.00018658783954040472,
      "loss": 12.6249,
      "step": 7043
    },
    {
      "epoch": 0.3835739519307345,
      "grad_norm": 0.7532813678875407,
      "learning_rate": 0.0001865834278300718,
      "loss": 12.4514,
      "step": 7044
    },
    {
      "epoch": 0.3836284059273175,
      "grad_norm": 0.6759469688196224,
      "learning_rate": 0.0001865790154464516,
      "loss": 12.6866,
      "step": 7045
    },
    {
      "epoch": 0.38368285992390055,
      "grad_norm": 0.5925743120960456,
      "learning_rate": 0.0001865746023895785,
      "loss": 12.3878,
      "step": 7046
    },
    {
      "epoch": 0.38373731392048355,
      "grad_norm": 0.7693887197413217,
      "learning_rate": 0.00018657018865948674,
      "loss": 12.6576,
      "step": 7047
    },
    {
      "epoch": 0.38379176791706654,
      "grad_norm": 0.6898680905211517,
      "learning_rate": 0.00018656577425621066,
      "loss": 12.5065,
      "step": 7048
    },
    {
      "epoch": 0.3838462219136496,
      "grad_norm": 0.7156536380503498,
      "learning_rate": 0.00018656135917978462,
      "loss": 12.4546,
      "step": 7049
    },
    {
      "epoch": 0.3839006759102326,
      "grad_norm": 0.6994971638710071,
      "learning_rate": 0.00018655694343024294,
      "loss": 12.6109,
      "step": 7050
    },
    {
      "epoch": 0.3839551299068156,
      "grad_norm": 0.7793553009856318,
      "learning_rate": 0.00018655252700761996,
      "loss": 12.4605,
      "step": 7051
    },
    {
      "epoch": 0.3840095839033986,
      "grad_norm": 0.6759341611546079,
      "learning_rate": 0.00018654810991195001,
      "loss": 12.464,
      "step": 7052
    },
    {
      "epoch": 0.3840640378999816,
      "grad_norm": 0.6301665691505982,
      "learning_rate": 0.00018654369214326746,
      "loss": 12.517,
      "step": 7053
    },
    {
      "epoch": 0.3841184918965646,
      "grad_norm": 0.7135243832459273,
      "learning_rate": 0.0001865392737016066,
      "loss": 12.4516,
      "step": 7054
    },
    {
      "epoch": 0.38417294589314765,
      "grad_norm": 0.6858399057276566,
      "learning_rate": 0.00018653485458700186,
      "loss": 12.5874,
      "step": 7055
    },
    {
      "epoch": 0.38422739988973065,
      "grad_norm": 0.8149053831416443,
      "learning_rate": 0.00018653043479948758,
      "loss": 12.5103,
      "step": 7056
    },
    {
      "epoch": 0.3842818538863137,
      "grad_norm": 0.7046987729073869,
      "learning_rate": 0.00018652601433909814,
      "loss": 12.5548,
      "step": 7057
    },
    {
      "epoch": 0.3843363078828967,
      "grad_norm": 0.6290864552411815,
      "learning_rate": 0.00018652159320586788,
      "loss": 12.499,
      "step": 7058
    },
    {
      "epoch": 0.3843907618794797,
      "grad_norm": 0.6916875542987843,
      "learning_rate": 0.00018651717139983123,
      "loss": 12.4951,
      "step": 7059
    },
    {
      "epoch": 0.38444521587606273,
      "grad_norm": 0.6978979497616465,
      "learning_rate": 0.0001865127489210225,
      "loss": 12.6373,
      "step": 7060
    },
    {
      "epoch": 0.3844996698726457,
      "grad_norm": 0.610515344658147,
      "learning_rate": 0.0001865083257694762,
      "loss": 12.5357,
      "step": 7061
    },
    {
      "epoch": 0.3845541238692287,
      "grad_norm": 0.6823700404070057,
      "learning_rate": 0.00018650390194522657,
      "loss": 12.5222,
      "step": 7062
    },
    {
      "epoch": 0.38460857786581176,
      "grad_norm": 0.6160180562682802,
      "learning_rate": 0.00018649947744830815,
      "loss": 12.4801,
      "step": 7063
    },
    {
      "epoch": 0.38466303186239476,
      "grad_norm": 0.6670446903986819,
      "learning_rate": 0.00018649505227875525,
      "loss": 12.6218,
      "step": 7064
    },
    {
      "epoch": 0.38471748585897775,
      "grad_norm": 0.5787244558977549,
      "learning_rate": 0.00018649062643660234,
      "loss": 12.3791,
      "step": 7065
    },
    {
      "epoch": 0.3847719398555608,
      "grad_norm": 0.6615016138111451,
      "learning_rate": 0.0001864861999218838,
      "loss": 12.4197,
      "step": 7066
    },
    {
      "epoch": 0.3848263938521438,
      "grad_norm": 0.6674308971925494,
      "learning_rate": 0.00018648177273463407,
      "loss": 12.5483,
      "step": 7067
    },
    {
      "epoch": 0.3848808478487268,
      "grad_norm": 0.6715041283823071,
      "learning_rate": 0.0001864773448748876,
      "loss": 12.4357,
      "step": 7068
    },
    {
      "epoch": 0.38493530184530983,
      "grad_norm": 0.6385821710498155,
      "learning_rate": 0.00018647291634267874,
      "loss": 12.45,
      "step": 7069
    },
    {
      "epoch": 0.3849897558418928,
      "grad_norm": 0.667305118023495,
      "learning_rate": 0.00018646848713804203,
      "loss": 12.5501,
      "step": 7070
    },
    {
      "epoch": 0.3850442098384758,
      "grad_norm": 0.6637185088679254,
      "learning_rate": 0.0001864640572610118,
      "loss": 12.5592,
      "step": 7071
    },
    {
      "epoch": 0.38509866383505886,
      "grad_norm": 0.6979028179275554,
      "learning_rate": 0.00018645962671162263,
      "loss": 12.4207,
      "step": 7072
    },
    {
      "epoch": 0.38515311783164186,
      "grad_norm": 0.7793517294375236,
      "learning_rate": 0.00018645519548990888,
      "loss": 12.4839,
      "step": 7073
    },
    {
      "epoch": 0.38520757182822485,
      "grad_norm": 0.6047285988296436,
      "learning_rate": 0.00018645076359590502,
      "loss": 12.4372,
      "step": 7074
    },
    {
      "epoch": 0.3852620258248079,
      "grad_norm": 0.6646751397762883,
      "learning_rate": 0.00018644633102964556,
      "loss": 12.593,
      "step": 7075
    },
    {
      "epoch": 0.3853164798213909,
      "grad_norm": 0.729705495106553,
      "learning_rate": 0.00018644189779116487,
      "loss": 12.4074,
      "step": 7076
    },
    {
      "epoch": 0.3853709338179739,
      "grad_norm": 0.6084925856486768,
      "learning_rate": 0.00018643746388049754,
      "loss": 12.5115,
      "step": 7077
    },
    {
      "epoch": 0.38542538781455693,
      "grad_norm": 0.7647209717282059,
      "learning_rate": 0.00018643302929767795,
      "loss": 12.5896,
      "step": 7078
    },
    {
      "epoch": 0.3854798418111399,
      "grad_norm": 0.5955347064856068,
      "learning_rate": 0.00018642859404274068,
      "loss": 12.5077,
      "step": 7079
    },
    {
      "epoch": 0.3855342958077229,
      "grad_norm": 0.6422483619212731,
      "learning_rate": 0.00018642415811572017,
      "loss": 12.5562,
      "step": 7080
    },
    {
      "epoch": 0.38558874980430596,
      "grad_norm": 0.694516033481041,
      "learning_rate": 0.00018641972151665085,
      "loss": 12.5192,
      "step": 7081
    },
    {
      "epoch": 0.38564320380088896,
      "grad_norm": 0.6587257113621006,
      "learning_rate": 0.00018641528424556735,
      "loss": 12.5755,
      "step": 7082
    },
    {
      "epoch": 0.38569765779747195,
      "grad_norm": 0.6763562072483827,
      "learning_rate": 0.00018641084630250407,
      "loss": 12.4134,
      "step": 7083
    },
    {
      "epoch": 0.385752111794055,
      "grad_norm": 0.6203052034426025,
      "learning_rate": 0.00018640640768749557,
      "loss": 12.5064,
      "step": 7084
    },
    {
      "epoch": 0.385806565790638,
      "grad_norm": 0.659861270783671,
      "learning_rate": 0.00018640196840057636,
      "loss": 12.4955,
      "step": 7085
    },
    {
      "epoch": 0.385861019787221,
      "grad_norm": 0.6437770506543342,
      "learning_rate": 0.00018639752844178093,
      "loss": 12.5114,
      "step": 7086
    },
    {
      "epoch": 0.38591547378380403,
      "grad_norm": 0.7570541110260534,
      "learning_rate": 0.00018639308781114386,
      "loss": 12.6241,
      "step": 7087
    },
    {
      "epoch": 0.385969927780387,
      "grad_norm": 0.7411434827368806,
      "learning_rate": 0.00018638864650869966,
      "loss": 12.4622,
      "step": 7088
    },
    {
      "epoch": 0.38602438177697,
      "grad_norm": 0.6590951056007902,
      "learning_rate": 0.0001863842045344828,
      "loss": 12.564,
      "step": 7089
    },
    {
      "epoch": 0.38607883577355306,
      "grad_norm": 0.6097965134684681,
      "learning_rate": 0.0001863797618885279,
      "loss": 12.5037,
      "step": 7090
    },
    {
      "epoch": 0.38613328977013606,
      "grad_norm": 0.6427854864860034,
      "learning_rate": 0.00018637531857086952,
      "loss": 12.4269,
      "step": 7091
    },
    {
      "epoch": 0.3861877437667191,
      "grad_norm": 0.7155316157969179,
      "learning_rate": 0.00018637087458154214,
      "loss": 12.5963,
      "step": 7092
    },
    {
      "epoch": 0.3862421977633021,
      "grad_norm": 0.6965946192322601,
      "learning_rate": 0.00018636642992058038,
      "loss": 12.5289,
      "step": 7093
    },
    {
      "epoch": 0.3862966517598851,
      "grad_norm": 0.6009755117826611,
      "learning_rate": 0.00018636198458801877,
      "loss": 12.4643,
      "step": 7094
    },
    {
      "epoch": 0.38635110575646814,
      "grad_norm": 0.7223297002920598,
      "learning_rate": 0.0001863575385838919,
      "loss": 12.6021,
      "step": 7095
    },
    {
      "epoch": 0.38640555975305113,
      "grad_norm": 0.6603037776913001,
      "learning_rate": 0.0001863530919082343,
      "loss": 12.5357,
      "step": 7096
    },
    {
      "epoch": 0.3864600137496341,
      "grad_norm": 0.6071533180146176,
      "learning_rate": 0.00018634864456108056,
      "loss": 12.4168,
      "step": 7097
    },
    {
      "epoch": 0.38651446774621717,
      "grad_norm": 0.7172644840488438,
      "learning_rate": 0.00018634419654246532,
      "loss": 12.5584,
      "step": 7098
    },
    {
      "epoch": 0.38656892174280016,
      "grad_norm": 0.7060305880193891,
      "learning_rate": 0.00018633974785242313,
      "loss": 12.4297,
      "step": 7099
    },
    {
      "epoch": 0.38662337573938316,
      "grad_norm": 0.6810705096067343,
      "learning_rate": 0.00018633529849098856,
      "loss": 12.5561,
      "step": 7100
    },
    {
      "epoch": 0.3866778297359662,
      "grad_norm": 0.6530001434960051,
      "learning_rate": 0.0001863308484581962,
      "loss": 12.5864,
      "step": 7101
    },
    {
      "epoch": 0.3867322837325492,
      "grad_norm": 0.6948748912939586,
      "learning_rate": 0.00018632639775408073,
      "loss": 12.4706,
      "step": 7102
    },
    {
      "epoch": 0.3867867377291322,
      "grad_norm": 0.6345644797992628,
      "learning_rate": 0.0001863219463786767,
      "loss": 12.395,
      "step": 7103
    },
    {
      "epoch": 0.38684119172571524,
      "grad_norm": 0.7069631467744489,
      "learning_rate": 0.00018631749433201876,
      "loss": 12.6182,
      "step": 7104
    },
    {
      "epoch": 0.38689564572229823,
      "grad_norm": 0.6557637249767786,
      "learning_rate": 0.0001863130416141415,
      "loss": 12.5655,
      "step": 7105
    },
    {
      "epoch": 0.3869500997188812,
      "grad_norm": 0.5965658094492697,
      "learning_rate": 0.00018630858822507956,
      "loss": 12.4235,
      "step": 7106
    },
    {
      "epoch": 0.3870045537154643,
      "grad_norm": 0.6341298768236069,
      "learning_rate": 0.00018630413416486754,
      "loss": 12.4438,
      "step": 7107
    },
    {
      "epoch": 0.38705900771204726,
      "grad_norm": 0.6500506288593676,
      "learning_rate": 0.0001862996794335401,
      "loss": 12.5798,
      "step": 7108
    },
    {
      "epoch": 0.38711346170863026,
      "grad_norm": 0.6411657990928703,
      "learning_rate": 0.0001862952240311319,
      "loss": 12.4585,
      "step": 7109
    },
    {
      "epoch": 0.3871679157052133,
      "grad_norm": 0.6437618395015029,
      "learning_rate": 0.00018629076795767755,
      "loss": 12.4662,
      "step": 7110
    },
    {
      "epoch": 0.3872223697017963,
      "grad_norm": 0.665479862318005,
      "learning_rate": 0.00018628631121321172,
      "loss": 12.4517,
      "step": 7111
    },
    {
      "epoch": 0.3872768236983793,
      "grad_norm": 0.6880421473175254,
      "learning_rate": 0.00018628185379776909,
      "loss": 12.4043,
      "step": 7112
    },
    {
      "epoch": 0.38733127769496234,
      "grad_norm": 0.6701975253629583,
      "learning_rate": 0.00018627739571138425,
      "loss": 12.4521,
      "step": 7113
    },
    {
      "epoch": 0.38738573169154533,
      "grad_norm": 0.6120212944128985,
      "learning_rate": 0.00018627293695409194,
      "loss": 12.5299,
      "step": 7114
    },
    {
      "epoch": 0.3874401856881283,
      "grad_norm": 0.6194184670856256,
      "learning_rate": 0.0001862684775259268,
      "loss": 12.4645,
      "step": 7115
    },
    {
      "epoch": 0.3874946396847114,
      "grad_norm": 0.7047969299610378,
      "learning_rate": 0.0001862640174269235,
      "loss": 12.6798,
      "step": 7116
    },
    {
      "epoch": 0.38754909368129437,
      "grad_norm": 0.7116460721530385,
      "learning_rate": 0.00018625955665711673,
      "loss": 12.5787,
      "step": 7117
    },
    {
      "epoch": 0.38760354767787736,
      "grad_norm": 0.7729800247148316,
      "learning_rate": 0.00018625509521654122,
      "loss": 12.462,
      "step": 7118
    },
    {
      "epoch": 0.3876580016744604,
      "grad_norm": 0.6599703454795484,
      "learning_rate": 0.0001862506331052316,
      "loss": 12.4132,
      "step": 7119
    },
    {
      "epoch": 0.3877124556710434,
      "grad_norm": 0.7320634611929326,
      "learning_rate": 0.0001862461703232226,
      "loss": 12.6311,
      "step": 7120
    },
    {
      "epoch": 0.3877669096676264,
      "grad_norm": 0.7085179533336992,
      "learning_rate": 0.0001862417068705489,
      "loss": 12.6353,
      "step": 7121
    },
    {
      "epoch": 0.38782136366420944,
      "grad_norm": 0.7532299118642597,
      "learning_rate": 0.00018623724274724522,
      "loss": 12.4862,
      "step": 7122
    },
    {
      "epoch": 0.38787581766079243,
      "grad_norm": 0.6448793596255141,
      "learning_rate": 0.00018623277795334632,
      "loss": 12.4253,
      "step": 7123
    },
    {
      "epoch": 0.3879302716573755,
      "grad_norm": 0.6352504803054885,
      "learning_rate": 0.00018622831248888682,
      "loss": 12.4379,
      "step": 7124
    },
    {
      "epoch": 0.3879847256539585,
      "grad_norm": 0.6218883827318556,
      "learning_rate": 0.00018622384635390152,
      "loss": 12.5006,
      "step": 7125
    },
    {
      "epoch": 0.38803917965054147,
      "grad_norm": 0.6190690293378829,
      "learning_rate": 0.00018621937954842516,
      "loss": 12.4908,
      "step": 7126
    },
    {
      "epoch": 0.3880936336471245,
      "grad_norm": 0.6275365107995106,
      "learning_rate": 0.00018621491207249243,
      "loss": 12.45,
      "step": 7127
    },
    {
      "epoch": 0.3881480876437075,
      "grad_norm": 0.6047005196824049,
      "learning_rate": 0.00018621044392613809,
      "loss": 12.3975,
      "step": 7128
    },
    {
      "epoch": 0.3882025416402905,
      "grad_norm": 0.6524608765328237,
      "learning_rate": 0.00018620597510939687,
      "loss": 12.4844,
      "step": 7129
    },
    {
      "epoch": 0.38825699563687355,
      "grad_norm": 0.7510200705796262,
      "learning_rate": 0.00018620150562230354,
      "loss": 12.6705,
      "step": 7130
    },
    {
      "epoch": 0.38831144963345654,
      "grad_norm": 0.73051352377799,
      "learning_rate": 0.00018619703546489286,
      "loss": 12.4825,
      "step": 7131
    },
    {
      "epoch": 0.38836590363003953,
      "grad_norm": 0.7746106714477208,
      "learning_rate": 0.00018619256463719953,
      "loss": 12.5045,
      "step": 7132
    },
    {
      "epoch": 0.3884203576266226,
      "grad_norm": 0.6517344434189947,
      "learning_rate": 0.0001861880931392584,
      "loss": 12.472,
      "step": 7133
    },
    {
      "epoch": 0.3884748116232056,
      "grad_norm": 0.7236585827856387,
      "learning_rate": 0.00018618362097110418,
      "loss": 12.4292,
      "step": 7134
    },
    {
      "epoch": 0.38852926561978857,
      "grad_norm": 0.7377438184733501,
      "learning_rate": 0.0001861791481327717,
      "loss": 12.52,
      "step": 7135
    },
    {
      "epoch": 0.3885837196163716,
      "grad_norm": 0.701077172486775,
      "learning_rate": 0.0001861746746242957,
      "loss": 12.3832,
      "step": 7136
    },
    {
      "epoch": 0.3886381736129546,
      "grad_norm": 0.7465602021994938,
      "learning_rate": 0.00018617020044571096,
      "loss": 12.477,
      "step": 7137
    },
    {
      "epoch": 0.3886926276095376,
      "grad_norm": 0.6913046599592584,
      "learning_rate": 0.00018616572559705232,
      "loss": 12.494,
      "step": 7138
    },
    {
      "epoch": 0.38874708160612065,
      "grad_norm": 0.798704738006824,
      "learning_rate": 0.00018616125007835454,
      "loss": 12.4876,
      "step": 7139
    },
    {
      "epoch": 0.38880153560270364,
      "grad_norm": 0.7090710177877242,
      "learning_rate": 0.0001861567738896524,
      "loss": 12.518,
      "step": 7140
    },
    {
      "epoch": 0.38885598959928663,
      "grad_norm": 0.6283988372735225,
      "learning_rate": 0.00018615229703098076,
      "loss": 12.612,
      "step": 7141
    },
    {
      "epoch": 0.3889104435958697,
      "grad_norm": 0.7791594674546557,
      "learning_rate": 0.00018614781950237444,
      "loss": 12.6456,
      "step": 7142
    },
    {
      "epoch": 0.3889648975924527,
      "grad_norm": 0.6618988345899924,
      "learning_rate": 0.0001861433413038682,
      "loss": 12.4756,
      "step": 7143
    },
    {
      "epoch": 0.38901935158903567,
      "grad_norm": 0.7549097356269053,
      "learning_rate": 0.0001861388624354969,
      "loss": 12.4658,
      "step": 7144
    },
    {
      "epoch": 0.3890738055856187,
      "grad_norm": 0.6821411593614526,
      "learning_rate": 0.00018613438289729535,
      "loss": 12.5978,
      "step": 7145
    },
    {
      "epoch": 0.3891282595822017,
      "grad_norm": 0.6596391875897409,
      "learning_rate": 0.00018612990268929838,
      "loss": 12.433,
      "step": 7146
    },
    {
      "epoch": 0.3891827135787847,
      "grad_norm": 0.6942742601285982,
      "learning_rate": 0.00018612542181154087,
      "loss": 12.4488,
      "step": 7147
    },
    {
      "epoch": 0.38923716757536775,
      "grad_norm": 0.6441158155215414,
      "learning_rate": 0.00018612094026405763,
      "loss": 12.3776,
      "step": 7148
    },
    {
      "epoch": 0.38929162157195074,
      "grad_norm": 0.6671097050311692,
      "learning_rate": 0.0001861164580468835,
      "loss": 12.6619,
      "step": 7149
    },
    {
      "epoch": 0.38934607556853373,
      "grad_norm": 0.6378455673853604,
      "learning_rate": 0.00018611197516005335,
      "loss": 12.401,
      "step": 7150
    },
    {
      "epoch": 0.3894005295651168,
      "grad_norm": 0.5825367655607089,
      "learning_rate": 0.0001861074916036021,
      "loss": 12.5309,
      "step": 7151
    },
    {
      "epoch": 0.3894549835616998,
      "grad_norm": 0.7563755785105186,
      "learning_rate": 0.00018610300737756448,
      "loss": 12.4888,
      "step": 7152
    },
    {
      "epoch": 0.38950943755828277,
      "grad_norm": 0.7009475975387238,
      "learning_rate": 0.00018609852248197546,
      "loss": 12.5647,
      "step": 7153
    },
    {
      "epoch": 0.3895638915548658,
      "grad_norm": 0.6796488958941794,
      "learning_rate": 0.0001860940369168699,
      "loss": 12.4961,
      "step": 7154
    },
    {
      "epoch": 0.3896183455514488,
      "grad_norm": 0.6557015786790266,
      "learning_rate": 0.00018608955068228267,
      "loss": 12.4476,
      "step": 7155
    },
    {
      "epoch": 0.3896727995480318,
      "grad_norm": 0.7001203977916104,
      "learning_rate": 0.00018608506377824864,
      "loss": 12.5892,
      "step": 7156
    },
    {
      "epoch": 0.38972725354461485,
      "grad_norm": 0.7131040980800413,
      "learning_rate": 0.00018608057620480274,
      "loss": 12.4833,
      "step": 7157
    },
    {
      "epoch": 0.38978170754119784,
      "grad_norm": 0.7123328847352828,
      "learning_rate": 0.00018607608796197982,
      "loss": 12.5396,
      "step": 7158
    },
    {
      "epoch": 0.3898361615377809,
      "grad_norm": 0.6952953693606627,
      "learning_rate": 0.00018607159904981483,
      "loss": 12.3928,
      "step": 7159
    },
    {
      "epoch": 0.3898906155343639,
      "grad_norm": 0.6573785347664906,
      "learning_rate": 0.0001860671094683426,
      "loss": 12.4525,
      "step": 7160
    },
    {
      "epoch": 0.3899450695309469,
      "grad_norm": 0.5640682053021612,
      "learning_rate": 0.00018606261921759814,
      "loss": 12.4185,
      "step": 7161
    },
    {
      "epoch": 0.3899995235275299,
      "grad_norm": 0.6663834979866167,
      "learning_rate": 0.00018605812829761633,
      "loss": 12.4792,
      "step": 7162
    },
    {
      "epoch": 0.3900539775241129,
      "grad_norm": 0.6167571085476997,
      "learning_rate": 0.00018605363670843206,
      "loss": 12.3879,
      "step": 7163
    },
    {
      "epoch": 0.3901084315206959,
      "grad_norm": 0.6064142565549342,
      "learning_rate": 0.0001860491444500803,
      "loss": 12.5378,
      "step": 7164
    },
    {
      "epoch": 0.39016288551727896,
      "grad_norm": 0.5654870240016255,
      "learning_rate": 0.00018604465152259595,
      "loss": 12.4184,
      "step": 7165
    },
    {
      "epoch": 0.39021733951386195,
      "grad_norm": 0.7203300327021557,
      "learning_rate": 0.00018604015792601396,
      "loss": 12.4794,
      "step": 7166
    },
    {
      "epoch": 0.39027179351044494,
      "grad_norm": 0.6751225100880477,
      "learning_rate": 0.00018603566366036923,
      "loss": 12.6103,
      "step": 7167
    },
    {
      "epoch": 0.390326247507028,
      "grad_norm": 0.6162717287525183,
      "learning_rate": 0.00018603116872569682,
      "loss": 12.3781,
      "step": 7168
    },
    {
      "epoch": 0.390380701503611,
      "grad_norm": 0.6081743008463881,
      "learning_rate": 0.00018602667312203158,
      "loss": 12.4616,
      "step": 7169
    },
    {
      "epoch": 0.390435155500194,
      "grad_norm": 0.7200821163330738,
      "learning_rate": 0.0001860221768494085,
      "loss": 12.2856,
      "step": 7170
    },
    {
      "epoch": 0.390489609496777,
      "grad_norm": 0.6801818674283662,
      "learning_rate": 0.00018601767990786256,
      "loss": 12.4423,
      "step": 7171
    },
    {
      "epoch": 0.39054406349336,
      "grad_norm": 0.7458239719186429,
      "learning_rate": 0.00018601318229742874,
      "loss": 12.4531,
      "step": 7172
    },
    {
      "epoch": 0.390598517489943,
      "grad_norm": 0.636396501553983,
      "learning_rate": 0.00018600868401814194,
      "loss": 12.418,
      "step": 7173
    },
    {
      "epoch": 0.39065297148652606,
      "grad_norm": 0.6982245766018101,
      "learning_rate": 0.00018600418507003723,
      "loss": 12.6615,
      "step": 7174
    },
    {
      "epoch": 0.39070742548310905,
      "grad_norm": 0.6725384848733218,
      "learning_rate": 0.00018599968545314951,
      "loss": 12.5772,
      "step": 7175
    },
    {
      "epoch": 0.39076187947969204,
      "grad_norm": 0.6217640478486886,
      "learning_rate": 0.00018599518516751386,
      "loss": 12.4921,
      "step": 7176
    },
    {
      "epoch": 0.3908163334762751,
      "grad_norm": 0.735065262741298,
      "learning_rate": 0.0001859906842131652,
      "loss": 12.606,
      "step": 7177
    },
    {
      "epoch": 0.3908707874728581,
      "grad_norm": 0.6181369701959938,
      "learning_rate": 0.00018598618259013856,
      "loss": 12.3427,
      "step": 7178
    },
    {
      "epoch": 0.3909252414694411,
      "grad_norm": 0.6777786413899306,
      "learning_rate": 0.00018598168029846895,
      "loss": 12.4036,
      "step": 7179
    },
    {
      "epoch": 0.3909796954660241,
      "grad_norm": 0.679731794899129,
      "learning_rate": 0.00018597717733819137,
      "loss": 12.4271,
      "step": 7180
    },
    {
      "epoch": 0.3910341494626071,
      "grad_norm": 0.6111586941549961,
      "learning_rate": 0.00018597267370934085,
      "loss": 12.4846,
      "step": 7181
    },
    {
      "epoch": 0.3910886034591901,
      "grad_norm": 0.6457344417871698,
      "learning_rate": 0.0001859681694119524,
      "loss": 12.5326,
      "step": 7182
    },
    {
      "epoch": 0.39114305745577316,
      "grad_norm": 0.6598166421712464,
      "learning_rate": 0.00018596366444606106,
      "loss": 12.3722,
      "step": 7183
    },
    {
      "epoch": 0.39119751145235615,
      "grad_norm": 0.6505653730095216,
      "learning_rate": 0.00018595915881170183,
      "loss": 12.4866,
      "step": 7184
    },
    {
      "epoch": 0.39125196544893914,
      "grad_norm": 0.8942655461387717,
      "learning_rate": 0.00018595465250890975,
      "loss": 12.6271,
      "step": 7185
    },
    {
      "epoch": 0.3913064194455222,
      "grad_norm": 0.683973537855153,
      "learning_rate": 0.00018595014553771992,
      "loss": 12.4369,
      "step": 7186
    },
    {
      "epoch": 0.3913608734421052,
      "grad_norm": 0.7404856511844542,
      "learning_rate": 0.00018594563789816734,
      "loss": 12.4747,
      "step": 7187
    },
    {
      "epoch": 0.3914153274386882,
      "grad_norm": 0.7896839713826467,
      "learning_rate": 0.00018594112959028706,
      "loss": 12.3406,
      "step": 7188
    },
    {
      "epoch": 0.3914697814352712,
      "grad_norm": 0.6195008716658261,
      "learning_rate": 0.00018593662061411413,
      "loss": 12.4288,
      "step": 7189
    },
    {
      "epoch": 0.3915242354318542,
      "grad_norm": 0.7143577956920403,
      "learning_rate": 0.00018593211096968362,
      "loss": 12.4196,
      "step": 7190
    },
    {
      "epoch": 0.39157868942843727,
      "grad_norm": 0.7425400955747433,
      "learning_rate": 0.0001859276006570306,
      "loss": 12.606,
      "step": 7191
    },
    {
      "epoch": 0.39163314342502026,
      "grad_norm": 0.6396423199370846,
      "learning_rate": 0.00018592308967619017,
      "loss": 12.6347,
      "step": 7192
    },
    {
      "epoch": 0.39168759742160325,
      "grad_norm": 0.6885635209124368,
      "learning_rate": 0.00018591857802719737,
      "loss": 12.5374,
      "step": 7193
    },
    {
      "epoch": 0.3917420514181863,
      "grad_norm": 0.6645824603123485,
      "learning_rate": 0.0001859140657100873,
      "loss": 12.5546,
      "step": 7194
    },
    {
      "epoch": 0.3917965054147693,
      "grad_norm": 0.5879980669589716,
      "learning_rate": 0.00018590955272489504,
      "loss": 12.3882,
      "step": 7195
    },
    {
      "epoch": 0.3918509594113523,
      "grad_norm": 0.7996186092060785,
      "learning_rate": 0.00018590503907165573,
      "loss": 12.5485,
      "step": 7196
    },
    {
      "epoch": 0.39190541340793533,
      "grad_norm": 0.6731604785694854,
      "learning_rate": 0.0001859005247504044,
      "loss": 12.5588,
      "step": 7197
    },
    {
      "epoch": 0.3919598674045183,
      "grad_norm": 0.7331086743184092,
      "learning_rate": 0.00018589600976117617,
      "loss": 12.3912,
      "step": 7198
    },
    {
      "epoch": 0.3920143214011013,
      "grad_norm": 0.6348125847667513,
      "learning_rate": 0.0001858914941040062,
      "loss": 12.5074,
      "step": 7199
    },
    {
      "epoch": 0.39206877539768437,
      "grad_norm": 0.6554620268542093,
      "learning_rate": 0.00018588697777892953,
      "loss": 12.4549,
      "step": 7200
    },
    {
      "epoch": 0.39212322939426736,
      "grad_norm": 0.6242822708657442,
      "learning_rate": 0.00018588246078598135,
      "loss": 12.5546,
      "step": 7201
    },
    {
      "epoch": 0.39217768339085035,
      "grad_norm": 0.762153351802112,
      "learning_rate": 0.00018587794312519674,
      "loss": 12.5397,
      "step": 7202
    },
    {
      "epoch": 0.3922321373874334,
      "grad_norm": 0.6007328039341426,
      "learning_rate": 0.00018587342479661084,
      "loss": 12.3427,
      "step": 7203
    },
    {
      "epoch": 0.3922865913840164,
      "grad_norm": 0.6532959864928064,
      "learning_rate": 0.00018586890580025878,
      "loss": 12.5656,
      "step": 7204
    },
    {
      "epoch": 0.3923410453805994,
      "grad_norm": 0.6905479326225938,
      "learning_rate": 0.0001858643861361757,
      "loss": 12.4903,
      "step": 7205
    },
    {
      "epoch": 0.39239549937718243,
      "grad_norm": 0.851211578891341,
      "learning_rate": 0.00018585986580439682,
      "loss": 12.5757,
      "step": 7206
    },
    {
      "epoch": 0.3924499533737654,
      "grad_norm": 0.683472164164967,
      "learning_rate": 0.00018585534480495714,
      "loss": 12.3313,
      "step": 7207
    },
    {
      "epoch": 0.3925044073703484,
      "grad_norm": 0.607373149852656,
      "learning_rate": 0.00018585082313789196,
      "loss": 12.3781,
      "step": 7208
    },
    {
      "epoch": 0.39255886136693147,
      "grad_norm": 0.6533168461112862,
      "learning_rate": 0.00018584630080323633,
      "loss": 12.5263,
      "step": 7209
    },
    {
      "epoch": 0.39261331536351446,
      "grad_norm": 0.6703700361785917,
      "learning_rate": 0.00018584177780102553,
      "loss": 12.3924,
      "step": 7210
    },
    {
      "epoch": 0.39266776936009745,
      "grad_norm": 0.7017657038020355,
      "learning_rate": 0.00018583725413129462,
      "loss": 12.6166,
      "step": 7211
    },
    {
      "epoch": 0.3927222233566805,
      "grad_norm": 0.6458042616638657,
      "learning_rate": 0.00018583272979407885,
      "loss": 12.4746,
      "step": 7212
    },
    {
      "epoch": 0.3927766773532635,
      "grad_norm": 0.6663919039166154,
      "learning_rate": 0.00018582820478941337,
      "loss": 12.4783,
      "step": 7213
    },
    {
      "epoch": 0.3928311313498465,
      "grad_norm": 0.6678956163369926,
      "learning_rate": 0.00018582367911733339,
      "loss": 12.6578,
      "step": 7214
    },
    {
      "epoch": 0.39288558534642953,
      "grad_norm": 0.6717646126286817,
      "learning_rate": 0.00018581915277787406,
      "loss": 12.5549,
      "step": 7215
    },
    {
      "epoch": 0.3929400393430125,
      "grad_norm": 0.6514951420285618,
      "learning_rate": 0.00018581462577107062,
      "loss": 12.4215,
      "step": 7216
    },
    {
      "epoch": 0.3929944933395955,
      "grad_norm": 0.6131590893663512,
      "learning_rate": 0.00018581009809695828,
      "loss": 12.413,
      "step": 7217
    },
    {
      "epoch": 0.39304894733617857,
      "grad_norm": 0.7061939089881658,
      "learning_rate": 0.0001858055697555722,
      "loss": 12.4528,
      "step": 7218
    },
    {
      "epoch": 0.39310340133276156,
      "grad_norm": 0.6396540860826965,
      "learning_rate": 0.00018580104074694765,
      "loss": 12.4773,
      "step": 7219
    },
    {
      "epoch": 0.39315785532934455,
      "grad_norm": 0.6813268667296958,
      "learning_rate": 0.00018579651107111979,
      "loss": 12.1887,
      "step": 7220
    },
    {
      "epoch": 0.3932123093259276,
      "grad_norm": 0.652895067238895,
      "learning_rate": 0.00018579198072812386,
      "loss": 12.5702,
      "step": 7221
    },
    {
      "epoch": 0.3932667633225106,
      "grad_norm": 0.6312854585604019,
      "learning_rate": 0.00018578744971799513,
      "loss": 12.4433,
      "step": 7222
    },
    {
      "epoch": 0.3933212173190936,
      "grad_norm": 0.7119148682281505,
      "learning_rate": 0.0001857829180407688,
      "loss": 12.4816,
      "step": 7223
    },
    {
      "epoch": 0.39337567131567663,
      "grad_norm": 0.5874724663556574,
      "learning_rate": 0.00018577838569648012,
      "loss": 12.3591,
      "step": 7224
    },
    {
      "epoch": 0.3934301253122596,
      "grad_norm": 0.724354991397679,
      "learning_rate": 0.0001857738526851643,
      "loss": 12.483,
      "step": 7225
    },
    {
      "epoch": 0.3934845793088427,
      "grad_norm": 0.6301934564762179,
      "learning_rate": 0.00018576931900685665,
      "loss": 12.421,
      "step": 7226
    },
    {
      "epoch": 0.39353903330542567,
      "grad_norm": 0.6071793867462213,
      "learning_rate": 0.00018576478466159237,
      "loss": 12.4453,
      "step": 7227
    },
    {
      "epoch": 0.39359348730200866,
      "grad_norm": 0.6099905450348952,
      "learning_rate": 0.00018576024964940673,
      "loss": 12.3518,
      "step": 7228
    },
    {
      "epoch": 0.3936479412985917,
      "grad_norm": 0.6502863285735471,
      "learning_rate": 0.00018575571397033504,
      "loss": 12.2838,
      "step": 7229
    },
    {
      "epoch": 0.3937023952951747,
      "grad_norm": 0.6212062647781128,
      "learning_rate": 0.00018575117762441252,
      "loss": 12.5731,
      "step": 7230
    },
    {
      "epoch": 0.3937568492917577,
      "grad_norm": 0.6241215749927159,
      "learning_rate": 0.00018574664061167447,
      "loss": 12.4702,
      "step": 7231
    },
    {
      "epoch": 0.39381130328834074,
      "grad_norm": 0.6667971540441856,
      "learning_rate": 0.00018574210293215615,
      "loss": 12.5083,
      "step": 7232
    },
    {
      "epoch": 0.39386575728492373,
      "grad_norm": 0.612237029823004,
      "learning_rate": 0.00018573756458589288,
      "loss": 12.4686,
      "step": 7233
    },
    {
      "epoch": 0.3939202112815067,
      "grad_norm": 0.6123887965534771,
      "learning_rate": 0.00018573302557291989,
      "loss": 12.4532,
      "step": 7234
    },
    {
      "epoch": 0.3939746652780898,
      "grad_norm": 0.5806781769345054,
      "learning_rate": 0.00018572848589327255,
      "loss": 12.4606,
      "step": 7235
    },
    {
      "epoch": 0.39402911927467277,
      "grad_norm": 0.6702610377858681,
      "learning_rate": 0.00018572394554698614,
      "loss": 12.5989,
      "step": 7236
    },
    {
      "epoch": 0.39408357327125576,
      "grad_norm": 0.6998756911883727,
      "learning_rate": 0.0001857194045340959,
      "loss": 12.3319,
      "step": 7237
    },
    {
      "epoch": 0.3941380272678388,
      "grad_norm": 0.6148350823215484,
      "learning_rate": 0.00018571486285463723,
      "loss": 12.5016,
      "step": 7238
    },
    {
      "epoch": 0.3941924812644218,
      "grad_norm": 0.6236150418109734,
      "learning_rate": 0.0001857103205086454,
      "loss": 12.5312,
      "step": 7239
    },
    {
      "epoch": 0.3942469352610048,
      "grad_norm": 0.6139467227040993,
      "learning_rate": 0.00018570577749615577,
      "loss": 12.6076,
      "step": 7240
    },
    {
      "epoch": 0.39430138925758784,
      "grad_norm": 0.6040187667043337,
      "learning_rate": 0.00018570123381720364,
      "loss": 12.4072,
      "step": 7241
    },
    {
      "epoch": 0.39435584325417083,
      "grad_norm": 0.6223642956678106,
      "learning_rate": 0.0001856966894718243,
      "loss": 12.4661,
      "step": 7242
    },
    {
      "epoch": 0.3944102972507538,
      "grad_norm": 0.659730055040347,
      "learning_rate": 0.00018569214446005316,
      "loss": 12.3883,
      "step": 7243
    },
    {
      "epoch": 0.3944647512473369,
      "grad_norm": 0.6723695378689196,
      "learning_rate": 0.00018568759878192554,
      "loss": 12.3468,
      "step": 7244
    },
    {
      "epoch": 0.39451920524391987,
      "grad_norm": 0.6392869809761946,
      "learning_rate": 0.00018568305243747677,
      "loss": 12.4045,
      "step": 7245
    },
    {
      "epoch": 0.39457365924050286,
      "grad_norm": 0.6810339540317356,
      "learning_rate": 0.0001856785054267422,
      "loss": 12.5154,
      "step": 7246
    },
    {
      "epoch": 0.3946281132370859,
      "grad_norm": 0.6434168951571108,
      "learning_rate": 0.00018567395774975724,
      "loss": 12.5282,
      "step": 7247
    },
    {
      "epoch": 0.3946825672336689,
      "grad_norm": 0.6729813495271544,
      "learning_rate": 0.0001856694094065572,
      "loss": 12.5041,
      "step": 7248
    },
    {
      "epoch": 0.3947370212302519,
      "grad_norm": 0.6876531212401764,
      "learning_rate": 0.00018566486039717749,
      "loss": 12.6692,
      "step": 7249
    },
    {
      "epoch": 0.39479147522683494,
      "grad_norm": 0.587843823790897,
      "learning_rate": 0.0001856603107216534,
      "loss": 12.3987,
      "step": 7250
    },
    {
      "epoch": 0.39484592922341794,
      "grad_norm": 0.6931917437986364,
      "learning_rate": 0.0001856557603800204,
      "loss": 12.3018,
      "step": 7251
    },
    {
      "epoch": 0.39490038322000093,
      "grad_norm": 0.6971045277469835,
      "learning_rate": 0.00018565120937231387,
      "loss": 12.4893,
      "step": 7252
    },
    {
      "epoch": 0.394954837216584,
      "grad_norm": 0.5889567317818012,
      "learning_rate": 0.00018564665769856914,
      "loss": 12.4292,
      "step": 7253
    },
    {
      "epoch": 0.39500929121316697,
      "grad_norm": 0.5758889736487871,
      "learning_rate": 0.00018564210535882168,
      "loss": 12.3916,
      "step": 7254
    },
    {
      "epoch": 0.39506374520974996,
      "grad_norm": 0.6136402012755989,
      "learning_rate": 0.00018563755235310677,
      "loss": 12.555,
      "step": 7255
    },
    {
      "epoch": 0.395118199206333,
      "grad_norm": 0.6967106844765542,
      "learning_rate": 0.00018563299868145996,
      "loss": 12.3989,
      "step": 7256
    },
    {
      "epoch": 0.395172653202916,
      "grad_norm": 0.6602235149415349,
      "learning_rate": 0.00018562844434391655,
      "loss": 12.5221,
      "step": 7257
    },
    {
      "epoch": 0.39522710719949905,
      "grad_norm": 0.5970856829434794,
      "learning_rate": 0.00018562388934051204,
      "loss": 12.5314,
      "step": 7258
    },
    {
      "epoch": 0.39528156119608204,
      "grad_norm": 0.702231835022106,
      "learning_rate": 0.00018561933367128175,
      "loss": 12.3386,
      "step": 7259
    },
    {
      "epoch": 0.39533601519266504,
      "grad_norm": 0.6349517268734202,
      "learning_rate": 0.0001856147773362612,
      "loss": 12.4374,
      "step": 7260
    },
    {
      "epoch": 0.3953904691892481,
      "grad_norm": 0.5998605612793768,
      "learning_rate": 0.00018561022033548578,
      "loss": 12.5123,
      "step": 7261
    },
    {
      "epoch": 0.3954449231858311,
      "grad_norm": 0.798835307386948,
      "learning_rate": 0.0001856056626689909,
      "loss": 12.5172,
      "step": 7262
    },
    {
      "epoch": 0.39549937718241407,
      "grad_norm": 0.6122143003084994,
      "learning_rate": 0.00018560110433681209,
      "loss": 12.4921,
      "step": 7263
    },
    {
      "epoch": 0.3955538311789971,
      "grad_norm": 0.6398409323570206,
      "learning_rate": 0.0001855965453389847,
      "loss": 12.5301,
      "step": 7264
    },
    {
      "epoch": 0.3956082851755801,
      "grad_norm": 0.6628705007558366,
      "learning_rate": 0.00018559198567554423,
      "loss": 12.5528,
      "step": 7265
    },
    {
      "epoch": 0.3956627391721631,
      "grad_norm": 0.6082496779379879,
      "learning_rate": 0.00018558742534652612,
      "loss": 12.4979,
      "step": 7266
    },
    {
      "epoch": 0.39571719316874615,
      "grad_norm": 0.6679068622193955,
      "learning_rate": 0.00018558286435196584,
      "loss": 12.4034,
      "step": 7267
    },
    {
      "epoch": 0.39577164716532914,
      "grad_norm": 0.6283397647402458,
      "learning_rate": 0.00018557830269189885,
      "loss": 12.4187,
      "step": 7268
    },
    {
      "epoch": 0.39582610116191214,
      "grad_norm": 0.6170647594859415,
      "learning_rate": 0.0001855737403663606,
      "loss": 12.4321,
      "step": 7269
    },
    {
      "epoch": 0.3958805551584952,
      "grad_norm": 0.6354956277518662,
      "learning_rate": 0.00018556917737538663,
      "loss": 12.531,
      "step": 7270
    },
    {
      "epoch": 0.3959350091550782,
      "grad_norm": 0.7097303096725949,
      "learning_rate": 0.0001855646137190124,
      "loss": 12.6097,
      "step": 7271
    },
    {
      "epoch": 0.39598946315166117,
      "grad_norm": 0.634232155487576,
      "learning_rate": 0.00018556004939727333,
      "loss": 12.3796,
      "step": 7272
    },
    {
      "epoch": 0.3960439171482442,
      "grad_norm": 0.6053037252161205,
      "learning_rate": 0.00018555548441020502,
      "loss": 12.4573,
      "step": 7273
    },
    {
      "epoch": 0.3960983711448272,
      "grad_norm": 0.6051827555115559,
      "learning_rate": 0.0001855509187578429,
      "loss": 12.4087,
      "step": 7274
    },
    {
      "epoch": 0.3961528251414102,
      "grad_norm": 0.5997554655096772,
      "learning_rate": 0.00018554635244022246,
      "loss": 12.5251,
      "step": 7275
    },
    {
      "epoch": 0.39620727913799325,
      "grad_norm": 0.5727144957820633,
      "learning_rate": 0.0001855417854573793,
      "loss": 12.3609,
      "step": 7276
    },
    {
      "epoch": 0.39626173313457624,
      "grad_norm": 0.626733753861516,
      "learning_rate": 0.00018553721780934884,
      "loss": 12.4887,
      "step": 7277
    },
    {
      "epoch": 0.39631618713115924,
      "grad_norm": 0.6508171147743776,
      "learning_rate": 0.0001855326494961666,
      "loss": 12.4415,
      "step": 7278
    },
    {
      "epoch": 0.3963706411277423,
      "grad_norm": 0.7338951360507143,
      "learning_rate": 0.00018552808051786816,
      "loss": 12.3495,
      "step": 7279
    },
    {
      "epoch": 0.3964250951243253,
      "grad_norm": 0.6598739504874858,
      "learning_rate": 0.00018552351087448903,
      "loss": 12.4492,
      "step": 7280
    },
    {
      "epoch": 0.39647954912090827,
      "grad_norm": 0.659297850454176,
      "learning_rate": 0.00018551894056606473,
      "loss": 12.5406,
      "step": 7281
    },
    {
      "epoch": 0.3965340031174913,
      "grad_norm": 0.7926530178887912,
      "learning_rate": 0.0001855143695926308,
      "loss": 12.5387,
      "step": 7282
    },
    {
      "epoch": 0.3965884571140743,
      "grad_norm": 0.8547824409637148,
      "learning_rate": 0.00018550979795422281,
      "loss": 12.4184,
      "step": 7283
    },
    {
      "epoch": 0.3966429111106573,
      "grad_norm": 0.6204336063723833,
      "learning_rate": 0.00018550522565087625,
      "loss": 12.409,
      "step": 7284
    },
    {
      "epoch": 0.39669736510724035,
      "grad_norm": 0.6544669397684993,
      "learning_rate": 0.00018550065268262676,
      "loss": 12.4706,
      "step": 7285
    },
    {
      "epoch": 0.39675181910382334,
      "grad_norm": 0.6906323331020744,
      "learning_rate": 0.00018549607904950983,
      "loss": 12.3418,
      "step": 7286
    },
    {
      "epoch": 0.39680627310040634,
      "grad_norm": 0.6146860200521059,
      "learning_rate": 0.00018549150475156108,
      "loss": 12.3718,
      "step": 7287
    },
    {
      "epoch": 0.3968607270969894,
      "grad_norm": 0.7514602294337431,
      "learning_rate": 0.00018548692978881601,
      "loss": 12.6076,
      "step": 7288
    },
    {
      "epoch": 0.3969151810935724,
      "grad_norm": 0.6983633388102153,
      "learning_rate": 0.00018548235416131025,
      "loss": 12.4947,
      "step": 7289
    },
    {
      "epoch": 0.39696963509015537,
      "grad_norm": 0.7084821502082099,
      "learning_rate": 0.0001854777778690794,
      "loss": 12.3876,
      "step": 7290
    },
    {
      "epoch": 0.3970240890867384,
      "grad_norm": 0.6327235755021157,
      "learning_rate": 0.00018547320091215897,
      "loss": 12.4699,
      "step": 7291
    },
    {
      "epoch": 0.3970785430833214,
      "grad_norm": 0.623885800366473,
      "learning_rate": 0.00018546862329058464,
      "loss": 12.4228,
      "step": 7292
    },
    {
      "epoch": 0.39713299707990446,
      "grad_norm": 0.752140264223175,
      "learning_rate": 0.00018546404500439194,
      "loss": 12.5489,
      "step": 7293
    },
    {
      "epoch": 0.39718745107648745,
      "grad_norm": 0.6581033958309009,
      "learning_rate": 0.0001854594660536165,
      "loss": 12.1787,
      "step": 7294
    },
    {
      "epoch": 0.39724190507307044,
      "grad_norm": 0.6692429709274312,
      "learning_rate": 0.0001854548864382939,
      "loss": 12.5248,
      "step": 7295
    },
    {
      "epoch": 0.3972963590696535,
      "grad_norm": 0.6222330407924297,
      "learning_rate": 0.00018545030615845978,
      "loss": 12.3521,
      "step": 7296
    },
    {
      "epoch": 0.3973508130662365,
      "grad_norm": 0.7111786469242476,
      "learning_rate": 0.00018544572521414976,
      "loss": 12.4401,
      "step": 7297
    },
    {
      "epoch": 0.3974052670628195,
      "grad_norm": 0.6415092416534993,
      "learning_rate": 0.00018544114360539947,
      "loss": 12.3767,
      "step": 7298
    },
    {
      "epoch": 0.3974597210594025,
      "grad_norm": 0.6301328211565949,
      "learning_rate": 0.0001854365613322445,
      "loss": 12.4496,
      "step": 7299
    },
    {
      "epoch": 0.3975141750559855,
      "grad_norm": 0.5882984345484891,
      "learning_rate": 0.00018543197839472047,
      "loss": 12.4798,
      "step": 7300
    },
    {
      "epoch": 0.3975686290525685,
      "grad_norm": 0.7185448650080746,
      "learning_rate": 0.00018542739479286309,
      "loss": 12.6752,
      "step": 7301
    },
    {
      "epoch": 0.39762308304915156,
      "grad_norm": 0.5885969522111261,
      "learning_rate": 0.00018542281052670795,
      "loss": 12.3972,
      "step": 7302
    },
    {
      "epoch": 0.39767753704573455,
      "grad_norm": 0.6529425986067049,
      "learning_rate": 0.00018541822559629072,
      "loss": 12.5375,
      "step": 7303
    },
    {
      "epoch": 0.39773199104231755,
      "grad_norm": 0.6470559603557152,
      "learning_rate": 0.00018541364000164702,
      "loss": 12.4317,
      "step": 7304
    },
    {
      "epoch": 0.3977864450389006,
      "grad_norm": 0.7064940258511272,
      "learning_rate": 0.00018540905374281254,
      "loss": 12.3257,
      "step": 7305
    },
    {
      "epoch": 0.3978408990354836,
      "grad_norm": 0.7124757808162602,
      "learning_rate": 0.00018540446681982294,
      "loss": 12.4732,
      "step": 7306
    },
    {
      "epoch": 0.3978953530320666,
      "grad_norm": 0.6063168223396287,
      "learning_rate": 0.0001853998792327139,
      "loss": 12.4021,
      "step": 7307
    },
    {
      "epoch": 0.3979498070286496,
      "grad_norm": 0.6382347270216853,
      "learning_rate": 0.00018539529098152103,
      "loss": 12.49,
      "step": 7308
    },
    {
      "epoch": 0.3980042610252326,
      "grad_norm": 0.6361698118939236,
      "learning_rate": 0.0001853907020662801,
      "loss": 12.4024,
      "step": 7309
    },
    {
      "epoch": 0.3980587150218156,
      "grad_norm": 0.6411437562384148,
      "learning_rate": 0.00018538611248702675,
      "loss": 12.4099,
      "step": 7310
    },
    {
      "epoch": 0.39811316901839866,
      "grad_norm": 0.7301694611151364,
      "learning_rate": 0.00018538152224379666,
      "loss": 12.5862,
      "step": 7311
    },
    {
      "epoch": 0.39816762301498165,
      "grad_norm": 0.5686964007656635,
      "learning_rate": 0.00018537693133662553,
      "loss": 12.3635,
      "step": 7312
    },
    {
      "epoch": 0.39822207701156465,
      "grad_norm": 0.669652844274776,
      "learning_rate": 0.00018537233976554906,
      "loss": 12.5609,
      "step": 7313
    },
    {
      "epoch": 0.3982765310081477,
      "grad_norm": 0.653851941082085,
      "learning_rate": 0.00018536774753060299,
      "loss": 12.4567,
      "step": 7314
    },
    {
      "epoch": 0.3983309850047307,
      "grad_norm": 0.6460005068343455,
      "learning_rate": 0.00018536315463182294,
      "loss": 12.498,
      "step": 7315
    },
    {
      "epoch": 0.3983854390013137,
      "grad_norm": 0.6315188681360169,
      "learning_rate": 0.00018535856106924472,
      "loss": 12.3967,
      "step": 7316
    },
    {
      "epoch": 0.3984398929978967,
      "grad_norm": 0.7126080248284609,
      "learning_rate": 0.00018535396684290402,
      "loss": 12.4171,
      "step": 7317
    },
    {
      "epoch": 0.3984943469944797,
      "grad_norm": 0.6610001603463862,
      "learning_rate": 0.00018534937195283658,
      "loss": 12.3529,
      "step": 7318
    },
    {
      "epoch": 0.3985488009910627,
      "grad_norm": 0.622858169622087,
      "learning_rate": 0.00018534477639907805,
      "loss": 12.4491,
      "step": 7319
    },
    {
      "epoch": 0.39860325498764576,
      "grad_norm": 0.7276069110047648,
      "learning_rate": 0.00018534018018166428,
      "loss": 12.4966,
      "step": 7320
    },
    {
      "epoch": 0.39865770898422875,
      "grad_norm": 0.6699895849428359,
      "learning_rate": 0.00018533558330063095,
      "loss": 12.5369,
      "step": 7321
    },
    {
      "epoch": 0.39871216298081175,
      "grad_norm": 0.7054366604545919,
      "learning_rate": 0.0001853309857560138,
      "loss": 12.4876,
      "step": 7322
    },
    {
      "epoch": 0.3987666169773948,
      "grad_norm": 0.6194120015384266,
      "learning_rate": 0.00018532638754784858,
      "loss": 12.4925,
      "step": 7323
    },
    {
      "epoch": 0.3988210709739778,
      "grad_norm": 0.6784120549830108,
      "learning_rate": 0.00018532178867617107,
      "loss": 12.4642,
      "step": 7324
    },
    {
      "epoch": 0.39887552497056084,
      "grad_norm": 0.619911638459784,
      "learning_rate": 0.00018531718914101703,
      "loss": 12.5518,
      "step": 7325
    },
    {
      "epoch": 0.39892997896714383,
      "grad_norm": 0.5986202288989504,
      "learning_rate": 0.00018531258894242223,
      "loss": 12.4409,
      "step": 7326
    },
    {
      "epoch": 0.3989844329637268,
      "grad_norm": 0.6360512448361856,
      "learning_rate": 0.0001853079880804224,
      "loss": 12.478,
      "step": 7327
    },
    {
      "epoch": 0.39903888696030987,
      "grad_norm": 0.6720223179345861,
      "learning_rate": 0.0001853033865550534,
      "loss": 12.4442,
      "step": 7328
    },
    {
      "epoch": 0.39909334095689286,
      "grad_norm": 0.6887841681858036,
      "learning_rate": 0.0001852987843663509,
      "loss": 12.5266,
      "step": 7329
    },
    {
      "epoch": 0.39914779495347585,
      "grad_norm": 0.6453691945791873,
      "learning_rate": 0.0001852941815143508,
      "loss": 12.4266,
      "step": 7330
    },
    {
      "epoch": 0.3992022489500589,
      "grad_norm": 0.6340510744267103,
      "learning_rate": 0.00018528957799908882,
      "loss": 12.5852,
      "step": 7331
    },
    {
      "epoch": 0.3992567029466419,
      "grad_norm": 0.6540445359669609,
      "learning_rate": 0.00018528497382060076,
      "loss": 12.5478,
      "step": 7332
    },
    {
      "epoch": 0.3993111569432249,
      "grad_norm": 0.6299686222054321,
      "learning_rate": 0.00018528036897892246,
      "loss": 12.5436,
      "step": 7333
    },
    {
      "epoch": 0.39936561093980794,
      "grad_norm": 0.6467089196451699,
      "learning_rate": 0.0001852757634740897,
      "loss": 12.4378,
      "step": 7334
    },
    {
      "epoch": 0.39942006493639093,
      "grad_norm": 0.6326128820218682,
      "learning_rate": 0.0001852711573061383,
      "loss": 12.5551,
      "step": 7335
    },
    {
      "epoch": 0.3994745189329739,
      "grad_norm": 0.6174531091360957,
      "learning_rate": 0.0001852665504751041,
      "loss": 12.4177,
      "step": 7336
    },
    {
      "epoch": 0.39952897292955697,
      "grad_norm": 0.621876904130212,
      "learning_rate": 0.0001852619429810229,
      "loss": 12.3719,
      "step": 7337
    },
    {
      "epoch": 0.39958342692613996,
      "grad_norm": 0.6295489899432883,
      "learning_rate": 0.00018525733482393055,
      "loss": 12.4287,
      "step": 7338
    },
    {
      "epoch": 0.39963788092272295,
      "grad_norm": 0.6920851383728188,
      "learning_rate": 0.00018525272600386283,
      "loss": 12.5537,
      "step": 7339
    },
    {
      "epoch": 0.399692334919306,
      "grad_norm": 0.5565562726035974,
      "learning_rate": 0.00018524811652085563,
      "loss": 12.2755,
      "step": 7340
    },
    {
      "epoch": 0.399746788915889,
      "grad_norm": 0.7016828793560161,
      "learning_rate": 0.0001852435063749448,
      "loss": 12.4688,
      "step": 7341
    },
    {
      "epoch": 0.399801242912472,
      "grad_norm": 0.682530812325876,
      "learning_rate": 0.00018523889556616612,
      "loss": 12.5191,
      "step": 7342
    },
    {
      "epoch": 0.39985569690905504,
      "grad_norm": 0.6143786563448566,
      "learning_rate": 0.00018523428409455555,
      "loss": 12.4387,
      "step": 7343
    },
    {
      "epoch": 0.39991015090563803,
      "grad_norm": 0.5801256496016015,
      "learning_rate": 0.00018522967196014887,
      "loss": 12.4552,
      "step": 7344
    },
    {
      "epoch": 0.399964604902221,
      "grad_norm": 0.6826321396517108,
      "learning_rate": 0.00018522505916298196,
      "loss": 12.5167,
      "step": 7345
    },
    {
      "epoch": 0.40001905889880407,
      "grad_norm": 0.6173642642335142,
      "learning_rate": 0.0001852204457030907,
      "loss": 12.3756,
      "step": 7346
    },
    {
      "epoch": 0.40007351289538706,
      "grad_norm": 0.7812678314690079,
      "learning_rate": 0.00018521583158051093,
      "loss": 12.5332,
      "step": 7347
    },
    {
      "epoch": 0.40012796689197005,
      "grad_norm": 0.6466449436057614,
      "learning_rate": 0.00018521121679527865,
      "loss": 12.4713,
      "step": 7348
    },
    {
      "epoch": 0.4001824208885531,
      "grad_norm": 0.6781616505750094,
      "learning_rate": 0.00018520660134742958,
      "loss": 12.5313,
      "step": 7349
    },
    {
      "epoch": 0.4002368748851361,
      "grad_norm": 0.6356853248099988,
      "learning_rate": 0.00018520198523699972,
      "loss": 12.5831,
      "step": 7350
    },
    {
      "epoch": 0.4002913288817191,
      "grad_norm": 0.6273074559411165,
      "learning_rate": 0.00018519736846402493,
      "loss": 12.5613,
      "step": 7351
    },
    {
      "epoch": 0.40034578287830214,
      "grad_norm": 0.7374084228928206,
      "learning_rate": 0.00018519275102854113,
      "loss": 12.6294,
      "step": 7352
    },
    {
      "epoch": 0.40040023687488513,
      "grad_norm": 0.6254505452815294,
      "learning_rate": 0.00018518813293058419,
      "loss": 12.5451,
      "step": 7353
    },
    {
      "epoch": 0.4004546908714681,
      "grad_norm": 0.6133781391891753,
      "learning_rate": 0.00018518351417019005,
      "loss": 12.3828,
      "step": 7354
    },
    {
      "epoch": 0.40050914486805117,
      "grad_norm": 0.6191631938029492,
      "learning_rate": 0.0001851788947473946,
      "loss": 12.5555,
      "step": 7355
    },
    {
      "epoch": 0.40056359886463416,
      "grad_norm": 0.6316920146031418,
      "learning_rate": 0.0001851742746622338,
      "loss": 12.35,
      "step": 7356
    },
    {
      "epoch": 0.40061805286121716,
      "grad_norm": 0.6218804069826839,
      "learning_rate": 0.00018516965391474354,
      "loss": 12.498,
      "step": 7357
    },
    {
      "epoch": 0.4006725068578002,
      "grad_norm": 0.6234070247722966,
      "learning_rate": 0.0001851650325049598,
      "loss": 12.5307,
      "step": 7358
    },
    {
      "epoch": 0.4007269608543832,
      "grad_norm": 0.6824202368507671,
      "learning_rate": 0.00018516041043291844,
      "loss": 12.5668,
      "step": 7359
    },
    {
      "epoch": 0.40078141485096624,
      "grad_norm": 0.6674063532644566,
      "learning_rate": 0.0001851557876986555,
      "loss": 12.4694,
      "step": 7360
    },
    {
      "epoch": 0.40083586884754924,
      "grad_norm": 0.7075034857406978,
      "learning_rate": 0.00018515116430220684,
      "loss": 12.5614,
      "step": 7361
    },
    {
      "epoch": 0.40089032284413223,
      "grad_norm": 0.6642130474985378,
      "learning_rate": 0.00018514654024360847,
      "loss": 12.4663,
      "step": 7362
    },
    {
      "epoch": 0.4009447768407153,
      "grad_norm": 0.7743771292825637,
      "learning_rate": 0.0001851419155228963,
      "loss": 12.6379,
      "step": 7363
    },
    {
      "epoch": 0.40099923083729827,
      "grad_norm": 0.7207920923280146,
      "learning_rate": 0.00018513729014010632,
      "loss": 12.4966,
      "step": 7364
    },
    {
      "epoch": 0.40105368483388126,
      "grad_norm": 0.6934995219290194,
      "learning_rate": 0.0001851326640952745,
      "loss": 12.4124,
      "step": 7365
    },
    {
      "epoch": 0.4011081388304643,
      "grad_norm": 0.6028082259589631,
      "learning_rate": 0.0001851280373884368,
      "loss": 12.3458,
      "step": 7366
    },
    {
      "epoch": 0.4011625928270473,
      "grad_norm": 0.6905902445618471,
      "learning_rate": 0.0001851234100196292,
      "loss": 12.3137,
      "step": 7367
    },
    {
      "epoch": 0.4012170468236303,
      "grad_norm": 0.73889018111692,
      "learning_rate": 0.0001851187819888877,
      "loss": 12.4944,
      "step": 7368
    },
    {
      "epoch": 0.40127150082021334,
      "grad_norm": 0.6805249740030216,
      "learning_rate": 0.00018511415329624828,
      "loss": 12.51,
      "step": 7369
    },
    {
      "epoch": 0.40132595481679634,
      "grad_norm": 0.6078115817769721,
      "learning_rate": 0.00018510952394174695,
      "loss": 12.3567,
      "step": 7370
    },
    {
      "epoch": 0.40138040881337933,
      "grad_norm": 0.655832926013966,
      "learning_rate": 0.00018510489392541964,
      "loss": 12.421,
      "step": 7371
    },
    {
      "epoch": 0.4014348628099624,
      "grad_norm": 0.6681102398311158,
      "learning_rate": 0.00018510026324730246,
      "loss": 12.5392,
      "step": 7372
    },
    {
      "epoch": 0.40148931680654537,
      "grad_norm": 0.6801679482138852,
      "learning_rate": 0.0001850956319074313,
      "loss": 12.5106,
      "step": 7373
    },
    {
      "epoch": 0.40154377080312836,
      "grad_norm": 0.6802545273425757,
      "learning_rate": 0.00018509099990584227,
      "loss": 12.3287,
      "step": 7374
    },
    {
      "epoch": 0.4015982247997114,
      "grad_norm": 0.6294918505790813,
      "learning_rate": 0.00018508636724257136,
      "loss": 12.4195,
      "step": 7375
    },
    {
      "epoch": 0.4016526787962944,
      "grad_norm": 0.7557878557287484,
      "learning_rate": 0.00018508173391765457,
      "loss": 12.4527,
      "step": 7376
    },
    {
      "epoch": 0.4017071327928774,
      "grad_norm": 0.5810675161036324,
      "learning_rate": 0.00018507709993112795,
      "loss": 12.4873,
      "step": 7377
    },
    {
      "epoch": 0.40176158678946045,
      "grad_norm": 0.7014067302253529,
      "learning_rate": 0.00018507246528302757,
      "loss": 12.5783,
      "step": 7378
    },
    {
      "epoch": 0.40181604078604344,
      "grad_norm": 0.7097681235618133,
      "learning_rate": 0.00018506782997338938,
      "loss": 12.5405,
      "step": 7379
    },
    {
      "epoch": 0.40187049478262643,
      "grad_norm": 0.7227925006604735,
      "learning_rate": 0.00018506319400224953,
      "loss": 12.5748,
      "step": 7380
    },
    {
      "epoch": 0.4019249487792095,
      "grad_norm": 0.6721929918043782,
      "learning_rate": 0.000185058557369644,
      "loss": 12.5959,
      "step": 7381
    },
    {
      "epoch": 0.40197940277579247,
      "grad_norm": 0.6215806863778806,
      "learning_rate": 0.00018505392007560882,
      "loss": 12.6075,
      "step": 7382
    },
    {
      "epoch": 0.40203385677237546,
      "grad_norm": 0.6407617755681561,
      "learning_rate": 0.00018504928212018015,
      "loss": 12.5772,
      "step": 7383
    },
    {
      "epoch": 0.4020883107689585,
      "grad_norm": 0.6070598483952332,
      "learning_rate": 0.00018504464350339398,
      "loss": 12.4512,
      "step": 7384
    },
    {
      "epoch": 0.4021427647655415,
      "grad_norm": 0.715882401990176,
      "learning_rate": 0.0001850400042252864,
      "loss": 12.6478,
      "step": 7385
    },
    {
      "epoch": 0.4021972187621245,
      "grad_norm": 0.7149775663341427,
      "learning_rate": 0.00018503536428589348,
      "loss": 12.6278,
      "step": 7386
    },
    {
      "epoch": 0.40225167275870755,
      "grad_norm": 0.618326261294556,
      "learning_rate": 0.00018503072368525133,
      "loss": 12.4231,
      "step": 7387
    },
    {
      "epoch": 0.40230612675529054,
      "grad_norm": 0.6969232561838311,
      "learning_rate": 0.000185026082423396,
      "loss": 12.5187,
      "step": 7388
    },
    {
      "epoch": 0.40236058075187353,
      "grad_norm": 0.6183554586340254,
      "learning_rate": 0.00018502144050036356,
      "loss": 12.6005,
      "step": 7389
    },
    {
      "epoch": 0.4024150347484566,
      "grad_norm": 0.5869917083721967,
      "learning_rate": 0.00018501679791619018,
      "loss": 12.3899,
      "step": 7390
    },
    {
      "epoch": 0.40246948874503957,
      "grad_norm": 0.6950697425232363,
      "learning_rate": 0.0001850121546709119,
      "loss": 12.5312,
      "step": 7391
    },
    {
      "epoch": 0.4025239427416226,
      "grad_norm": 0.666716020931147,
      "learning_rate": 0.0001850075107645649,
      "loss": 12.5483,
      "step": 7392
    },
    {
      "epoch": 0.4025783967382056,
      "grad_norm": 0.6547101411207449,
      "learning_rate": 0.00018500286619718516,
      "loss": 12.432,
      "step": 7393
    },
    {
      "epoch": 0.4026328507347886,
      "grad_norm": 0.598399101548866,
      "learning_rate": 0.00018499822096880894,
      "loss": 12.3849,
      "step": 7394
    },
    {
      "epoch": 0.40268730473137165,
      "grad_norm": 0.6814945968918992,
      "learning_rate": 0.0001849935750794723,
      "loss": 12.3536,
      "step": 7395
    },
    {
      "epoch": 0.40274175872795465,
      "grad_norm": 0.6626293967806715,
      "learning_rate": 0.00018498892852921134,
      "loss": 12.4122,
      "step": 7396
    },
    {
      "epoch": 0.40279621272453764,
      "grad_norm": 0.6900476293869696,
      "learning_rate": 0.0001849842813180622,
      "loss": 12.55,
      "step": 7397
    },
    {
      "epoch": 0.4028506667211207,
      "grad_norm": 0.6745622136773487,
      "learning_rate": 0.00018497963344606106,
      "loss": 12.4947,
      "step": 7398
    },
    {
      "epoch": 0.4029051207177037,
      "grad_norm": 0.7267336187426208,
      "learning_rate": 0.00018497498491324406,
      "loss": 12.5677,
      "step": 7399
    },
    {
      "epoch": 0.4029595747142867,
      "grad_norm": 0.6591495967434877,
      "learning_rate": 0.00018497033571964727,
      "loss": 12.5507,
      "step": 7400
    },
    {
      "epoch": 0.4030140287108697,
      "grad_norm": 0.6366391704310337,
      "learning_rate": 0.00018496568586530695,
      "loss": 12.5178,
      "step": 7401
    },
    {
      "epoch": 0.4030684827074527,
      "grad_norm": 0.6943015623607408,
      "learning_rate": 0.00018496103535025918,
      "loss": 12.4766,
      "step": 7402
    },
    {
      "epoch": 0.4031229367040357,
      "grad_norm": 0.6496156865013774,
      "learning_rate": 0.00018495638417454017,
      "loss": 12.3915,
      "step": 7403
    },
    {
      "epoch": 0.40317739070061875,
      "grad_norm": 0.7010064864414909,
      "learning_rate": 0.0001849517323381861,
      "loss": 12.4634,
      "step": 7404
    },
    {
      "epoch": 0.40323184469720175,
      "grad_norm": 0.6761556369951273,
      "learning_rate": 0.00018494707984123307,
      "loss": 12.3944,
      "step": 7405
    },
    {
      "epoch": 0.40328629869378474,
      "grad_norm": 0.6087198016655698,
      "learning_rate": 0.0001849424266837173,
      "loss": 12.4833,
      "step": 7406
    },
    {
      "epoch": 0.4033407526903678,
      "grad_norm": 0.7169273106444296,
      "learning_rate": 0.00018493777286567498,
      "loss": 12.5492,
      "step": 7407
    },
    {
      "epoch": 0.4033952066869508,
      "grad_norm": 0.6440946600719528,
      "learning_rate": 0.00018493311838714232,
      "loss": 12.4559,
      "step": 7408
    },
    {
      "epoch": 0.4034496606835338,
      "grad_norm": 0.6450930034300398,
      "learning_rate": 0.00018492846324815547,
      "loss": 12.6247,
      "step": 7409
    },
    {
      "epoch": 0.4035041146801168,
      "grad_norm": 0.6440379111155046,
      "learning_rate": 0.0001849238074487506,
      "loss": 12.4345,
      "step": 7410
    },
    {
      "epoch": 0.4035585686766998,
      "grad_norm": 0.7276183434116561,
      "learning_rate": 0.00018491915098896403,
      "loss": 12.4772,
      "step": 7411
    },
    {
      "epoch": 0.4036130226732828,
      "grad_norm": 0.6384627542574353,
      "learning_rate": 0.0001849144938688319,
      "loss": 12.3952,
      "step": 7412
    },
    {
      "epoch": 0.40366747666986585,
      "grad_norm": 0.6931831770467287,
      "learning_rate": 0.0001849098360883904,
      "loss": 12.3868,
      "step": 7413
    },
    {
      "epoch": 0.40372193066644885,
      "grad_norm": 0.6550735530488863,
      "learning_rate": 0.00018490517764767578,
      "loss": 12.5256,
      "step": 7414
    },
    {
      "epoch": 0.40377638466303184,
      "grad_norm": 0.743756648354596,
      "learning_rate": 0.00018490051854672424,
      "loss": 12.4004,
      "step": 7415
    },
    {
      "epoch": 0.4038308386596149,
      "grad_norm": 0.6575060670971092,
      "learning_rate": 0.00018489585878557206,
      "loss": 12.5674,
      "step": 7416
    },
    {
      "epoch": 0.4038852926561979,
      "grad_norm": 0.6628270930000479,
      "learning_rate": 0.00018489119836425543,
      "loss": 12.3725,
      "step": 7417
    },
    {
      "epoch": 0.4039397466527809,
      "grad_norm": 0.6239260359465777,
      "learning_rate": 0.0001848865372828106,
      "loss": 12.3349,
      "step": 7418
    },
    {
      "epoch": 0.4039942006493639,
      "grad_norm": 0.6977175489948938,
      "learning_rate": 0.00018488187554127383,
      "loss": 12.545,
      "step": 7419
    },
    {
      "epoch": 0.4040486546459469,
      "grad_norm": 0.7582230177928753,
      "learning_rate": 0.00018487721313968137,
      "loss": 12.348,
      "step": 7420
    },
    {
      "epoch": 0.4041031086425299,
      "grad_norm": 0.6080224584126379,
      "learning_rate": 0.00018487255007806945,
      "loss": 12.5413,
      "step": 7421
    },
    {
      "epoch": 0.40415756263911295,
      "grad_norm": 0.6923559615321203,
      "learning_rate": 0.00018486788635647435,
      "loss": 12.5759,
      "step": 7422
    },
    {
      "epoch": 0.40421201663569595,
      "grad_norm": 0.6912263414510522,
      "learning_rate": 0.00018486322197493234,
      "loss": 12.4712,
      "step": 7423
    },
    {
      "epoch": 0.40426647063227894,
      "grad_norm": 0.6863212712827147,
      "learning_rate": 0.00018485855693347968,
      "loss": 12.4547,
      "step": 7424
    },
    {
      "epoch": 0.404320924628862,
      "grad_norm": 0.6403074472885916,
      "learning_rate": 0.00018485389123215265,
      "loss": 12.534,
      "step": 7425
    },
    {
      "epoch": 0.404375378625445,
      "grad_norm": 0.6368847513628307,
      "learning_rate": 0.00018484922487098753,
      "loss": 12.3169,
      "step": 7426
    },
    {
      "epoch": 0.40442983262202803,
      "grad_norm": 0.6936139800096027,
      "learning_rate": 0.00018484455785002063,
      "loss": 12.4682,
      "step": 7427
    },
    {
      "epoch": 0.404484286618611,
      "grad_norm": 0.6983739029881817,
      "learning_rate": 0.00018483989016928817,
      "loss": 12.564,
      "step": 7428
    },
    {
      "epoch": 0.404538740615194,
      "grad_norm": 0.6408473930715737,
      "learning_rate": 0.00018483522182882655,
      "loss": 12.4528,
      "step": 7429
    },
    {
      "epoch": 0.40459319461177706,
      "grad_norm": 0.9015349932579146,
      "learning_rate": 0.000184830552828672,
      "loss": 12.5665,
      "step": 7430
    },
    {
      "epoch": 0.40464764860836006,
      "grad_norm": 0.6763429151139749,
      "learning_rate": 0.00018482588316886083,
      "loss": 12.5239,
      "step": 7431
    },
    {
      "epoch": 0.40470210260494305,
      "grad_norm": 0.6513052832654872,
      "learning_rate": 0.0001848212128494294,
      "loss": 12.3781,
      "step": 7432
    },
    {
      "epoch": 0.4047565566015261,
      "grad_norm": 0.6240966308784069,
      "learning_rate": 0.00018481654187041396,
      "loss": 12.4837,
      "step": 7433
    },
    {
      "epoch": 0.4048110105981091,
      "grad_norm": 0.7165993138055703,
      "learning_rate": 0.00018481187023185086,
      "loss": 12.5838,
      "step": 7434
    },
    {
      "epoch": 0.4048654645946921,
      "grad_norm": 0.5818743862161894,
      "learning_rate": 0.0001848071979337765,
      "loss": 12.3615,
      "step": 7435
    },
    {
      "epoch": 0.40491991859127513,
      "grad_norm": 0.6636559493509343,
      "learning_rate": 0.00018480252497622706,
      "loss": 12.505,
      "step": 7436
    },
    {
      "epoch": 0.4049743725878581,
      "grad_norm": 0.646612671491056,
      "learning_rate": 0.00018479785135923905,
      "loss": 12.5242,
      "step": 7437
    },
    {
      "epoch": 0.4050288265844411,
      "grad_norm": 0.6143601759897361,
      "learning_rate": 0.00018479317708284865,
      "loss": 12.2706,
      "step": 7438
    },
    {
      "epoch": 0.40508328058102416,
      "grad_norm": 0.7183560522499322,
      "learning_rate": 0.00018478850214709232,
      "loss": 12.3756,
      "step": 7439
    },
    {
      "epoch": 0.40513773457760716,
      "grad_norm": 0.6479854547071566,
      "learning_rate": 0.00018478382655200636,
      "loss": 12.4938,
      "step": 7440
    },
    {
      "epoch": 0.40519218857419015,
      "grad_norm": 0.8079384121207523,
      "learning_rate": 0.00018477915029762717,
      "loss": 12.2812,
      "step": 7441
    },
    {
      "epoch": 0.4052466425707732,
      "grad_norm": 0.6780665302072778,
      "learning_rate": 0.00018477447338399107,
      "loss": 12.5561,
      "step": 7442
    },
    {
      "epoch": 0.4053010965673562,
      "grad_norm": 0.6485134302769748,
      "learning_rate": 0.00018476979581113449,
      "loss": 12.3967,
      "step": 7443
    },
    {
      "epoch": 0.4053555505639392,
      "grad_norm": 0.646095910126911,
      "learning_rate": 0.00018476511757909374,
      "loss": 12.5835,
      "step": 7444
    },
    {
      "epoch": 0.40541000456052223,
      "grad_norm": 0.5917501713499067,
      "learning_rate": 0.0001847604386879052,
      "loss": 12.4826,
      "step": 7445
    },
    {
      "epoch": 0.4054644585571052,
      "grad_norm": 0.6986606862332128,
      "learning_rate": 0.00018475575913760528,
      "loss": 12.429,
      "step": 7446
    },
    {
      "epoch": 0.4055189125536882,
      "grad_norm": 0.6715650695644763,
      "learning_rate": 0.00018475107892823039,
      "loss": 12.5835,
      "step": 7447
    },
    {
      "epoch": 0.40557336655027126,
      "grad_norm": 0.6512536707903471,
      "learning_rate": 0.00018474639805981686,
      "loss": 12.4181,
      "step": 7448
    },
    {
      "epoch": 0.40562782054685426,
      "grad_norm": 0.6601484346024479,
      "learning_rate": 0.00018474171653240116,
      "loss": 12.4515,
      "step": 7449
    },
    {
      "epoch": 0.40568227454343725,
      "grad_norm": 0.7259142491346082,
      "learning_rate": 0.00018473703434601963,
      "loss": 12.6206,
      "step": 7450
    },
    {
      "epoch": 0.4057367285400203,
      "grad_norm": 0.7088826468538713,
      "learning_rate": 0.00018473235150070873,
      "loss": 12.4792,
      "step": 7451
    },
    {
      "epoch": 0.4057911825366033,
      "grad_norm": 0.6843809362146842,
      "learning_rate": 0.00018472766799650485,
      "loss": 12.542,
      "step": 7452
    },
    {
      "epoch": 0.4058456365331863,
      "grad_norm": 0.6461760765926458,
      "learning_rate": 0.0001847229838334444,
      "loss": 12.4608,
      "step": 7453
    },
    {
      "epoch": 0.40590009052976933,
      "grad_norm": 0.6445512760636828,
      "learning_rate": 0.00018471829901156386,
      "loss": 12.592,
      "step": 7454
    },
    {
      "epoch": 0.4059545445263523,
      "grad_norm": 0.6575265356329029,
      "learning_rate": 0.0001847136135308996,
      "loss": 12.4246,
      "step": 7455
    },
    {
      "epoch": 0.4060089985229353,
      "grad_norm": 0.6039633442973562,
      "learning_rate": 0.00018470892739148807,
      "loss": 12.4342,
      "step": 7456
    },
    {
      "epoch": 0.40606345251951836,
      "grad_norm": 0.7255842086748665,
      "learning_rate": 0.0001847042405933657,
      "loss": 12.5407,
      "step": 7457
    },
    {
      "epoch": 0.40611790651610136,
      "grad_norm": 0.7658324790055303,
      "learning_rate": 0.000184699553136569,
      "loss": 12.4578,
      "step": 7458
    },
    {
      "epoch": 0.4061723605126844,
      "grad_norm": 0.6736769880323731,
      "learning_rate": 0.00018469486502113432,
      "loss": 12.4223,
      "step": 7459
    },
    {
      "epoch": 0.4062268145092674,
      "grad_norm": 0.646846684475632,
      "learning_rate": 0.00018469017624709818,
      "loss": 12.4377,
      "step": 7460
    },
    {
      "epoch": 0.4062812685058504,
      "grad_norm": 0.643852904553272,
      "learning_rate": 0.00018468548681449702,
      "loss": 12.4088,
      "step": 7461
    },
    {
      "epoch": 0.40633572250243344,
      "grad_norm": 0.6213086606944596,
      "learning_rate": 0.00018468079672336732,
      "loss": 12.474,
      "step": 7462
    },
    {
      "epoch": 0.40639017649901643,
      "grad_norm": 0.7137248569844133,
      "learning_rate": 0.00018467610597374553,
      "loss": 12.5347,
      "step": 7463
    },
    {
      "epoch": 0.4064446304955994,
      "grad_norm": 0.6181355049536756,
      "learning_rate": 0.0001846714145656682,
      "loss": 12.4671,
      "step": 7464
    },
    {
      "epoch": 0.40649908449218247,
      "grad_norm": 0.7793233068249615,
      "learning_rate": 0.0001846667224991717,
      "loss": 12.5739,
      "step": 7465
    },
    {
      "epoch": 0.40655353848876546,
      "grad_norm": 0.6253303433337799,
      "learning_rate": 0.00018466202977429256,
      "loss": 12.4636,
      "step": 7466
    },
    {
      "epoch": 0.40660799248534846,
      "grad_norm": 0.6846509132803774,
      "learning_rate": 0.00018465733639106728,
      "loss": 12.4586,
      "step": 7467
    },
    {
      "epoch": 0.4066624464819315,
      "grad_norm": 0.604509255157627,
      "learning_rate": 0.00018465264234953236,
      "loss": 12.5897,
      "step": 7468
    },
    {
      "epoch": 0.4067169004785145,
      "grad_norm": 0.6414263010065017,
      "learning_rate": 0.00018464794764972434,
      "loss": 12.5562,
      "step": 7469
    },
    {
      "epoch": 0.4067713544750975,
      "grad_norm": 0.6395195367346062,
      "learning_rate": 0.00018464325229167961,
      "loss": 12.5577,
      "step": 7470
    },
    {
      "epoch": 0.40682580847168054,
      "grad_norm": 0.8378691952895028,
      "learning_rate": 0.00018463855627543483,
      "loss": 12.5377,
      "step": 7471
    },
    {
      "epoch": 0.40688026246826353,
      "grad_norm": 0.5838339172660505,
      "learning_rate": 0.0001846338596010264,
      "loss": 12.4138,
      "step": 7472
    },
    {
      "epoch": 0.4069347164648465,
      "grad_norm": 0.7445508194750182,
      "learning_rate": 0.0001846291622684909,
      "loss": 12.5826,
      "step": 7473
    },
    {
      "epoch": 0.4069891704614296,
      "grad_norm": 0.7064093062710597,
      "learning_rate": 0.0001846244642778648,
      "loss": 12.525,
      "step": 7474
    },
    {
      "epoch": 0.40704362445801257,
      "grad_norm": 0.6330305455995636,
      "learning_rate": 0.00018461976562918471,
      "loss": 12.471,
      "step": 7475
    },
    {
      "epoch": 0.40709807845459556,
      "grad_norm": 0.6030418041469815,
      "learning_rate": 0.00018461506632248714,
      "loss": 12.4623,
      "step": 7476
    },
    {
      "epoch": 0.4071525324511786,
      "grad_norm": 0.7572696367663821,
      "learning_rate": 0.00018461036635780863,
      "loss": 12.4131,
      "step": 7477
    },
    {
      "epoch": 0.4072069864477616,
      "grad_norm": 0.7001598074872516,
      "learning_rate": 0.0001846056657351857,
      "loss": 12.4616,
      "step": 7478
    },
    {
      "epoch": 0.4072614404443446,
      "grad_norm": 0.6940485999902011,
      "learning_rate": 0.0001846009644546549,
      "loss": 12.4128,
      "step": 7479
    },
    {
      "epoch": 0.40731589444092764,
      "grad_norm": 0.70654138952586,
      "learning_rate": 0.00018459626251625286,
      "loss": 12.4186,
      "step": 7480
    },
    {
      "epoch": 0.40737034843751063,
      "grad_norm": 0.923957097435168,
      "learning_rate": 0.0001845915599200161,
      "loss": 12.4704,
      "step": 7481
    },
    {
      "epoch": 0.4074248024340936,
      "grad_norm": 0.787704032185711,
      "learning_rate": 0.00018458685666598114,
      "loss": 12.5002,
      "step": 7482
    },
    {
      "epoch": 0.4074792564306767,
      "grad_norm": 0.8237434142055088,
      "learning_rate": 0.00018458215275418463,
      "loss": 12.6203,
      "step": 7483
    },
    {
      "epoch": 0.40753371042725967,
      "grad_norm": 0.7488862790251631,
      "learning_rate": 0.0001845774481846631,
      "loss": 12.3982,
      "step": 7484
    },
    {
      "epoch": 0.40758816442384266,
      "grad_norm": 0.7451931572702298,
      "learning_rate": 0.00018457274295745316,
      "loss": 12.5894,
      "step": 7485
    },
    {
      "epoch": 0.4076426184204257,
      "grad_norm": 0.6763644886168447,
      "learning_rate": 0.0001845680370725914,
      "loss": 12.4027,
      "step": 7486
    },
    {
      "epoch": 0.4076970724170087,
      "grad_norm": 0.8090062704601079,
      "learning_rate": 0.00018456333053011437,
      "loss": 12.3973,
      "step": 7487
    },
    {
      "epoch": 0.4077515264135917,
      "grad_norm": 0.6766492711891675,
      "learning_rate": 0.00018455862333005872,
      "loss": 12.5136,
      "step": 7488
    },
    {
      "epoch": 0.40780598041017474,
      "grad_norm": 0.7354799773717808,
      "learning_rate": 0.000184553915472461,
      "loss": 12.5156,
      "step": 7489
    },
    {
      "epoch": 0.40786043440675773,
      "grad_norm": 0.7358787223055255,
      "learning_rate": 0.0001845492069573579,
      "loss": 12.5716,
      "step": 7490
    },
    {
      "epoch": 0.4079148884033407,
      "grad_norm": 0.7940397944309273,
      "learning_rate": 0.00018454449778478597,
      "loss": 12.6003,
      "step": 7491
    },
    {
      "epoch": 0.4079693423999238,
      "grad_norm": 0.7276279277087789,
      "learning_rate": 0.00018453978795478183,
      "loss": 12.5072,
      "step": 7492
    },
    {
      "epoch": 0.40802379639650677,
      "grad_norm": 0.65622616537559,
      "learning_rate": 0.00018453507746738217,
      "loss": 12.4548,
      "step": 7493
    },
    {
      "epoch": 0.4080782503930898,
      "grad_norm": 0.6544295804803718,
      "learning_rate": 0.00018453036632262352,
      "loss": 12.5034,
      "step": 7494
    },
    {
      "epoch": 0.4081327043896728,
      "grad_norm": 0.6667584813889628,
      "learning_rate": 0.0001845256545205426,
      "loss": 12.3758,
      "step": 7495
    },
    {
      "epoch": 0.4081871583862558,
      "grad_norm": 0.7293652650925002,
      "learning_rate": 0.000184520942061176,
      "loss": 12.4498,
      "step": 7496
    },
    {
      "epoch": 0.40824161238283885,
      "grad_norm": 0.6768095263738398,
      "learning_rate": 0.00018451622894456038,
      "loss": 12.4563,
      "step": 7497
    },
    {
      "epoch": 0.40829606637942184,
      "grad_norm": 0.7193329277422399,
      "learning_rate": 0.0001845115151707324,
      "loss": 12.6072,
      "step": 7498
    },
    {
      "epoch": 0.40835052037600483,
      "grad_norm": 0.6847329382109661,
      "learning_rate": 0.00018450680073972867,
      "loss": 12.4992,
      "step": 7499
    },
    {
      "epoch": 0.4084049743725879,
      "grad_norm": 0.6414975022868673,
      "learning_rate": 0.00018450208565158594,
      "loss": 12.4932,
      "step": 7500
    },
    {
      "epoch": 0.4084594283691709,
      "grad_norm": 0.6553741533966808,
      "learning_rate": 0.0001844973699063408,
      "loss": 12.4239,
      "step": 7501
    },
    {
      "epoch": 0.40851388236575387,
      "grad_norm": 0.6885331621744704,
      "learning_rate": 0.00018449265350402994,
      "loss": 12.4297,
      "step": 7502
    },
    {
      "epoch": 0.4085683363623369,
      "grad_norm": 0.6928052080715973,
      "learning_rate": 0.00018448793644469002,
      "loss": 12.485,
      "step": 7503
    },
    {
      "epoch": 0.4086227903589199,
      "grad_norm": 0.6302817729042274,
      "learning_rate": 0.00018448321872835773,
      "loss": 12.4015,
      "step": 7504
    },
    {
      "epoch": 0.4086772443555029,
      "grad_norm": 0.7513722166710587,
      "learning_rate": 0.0001844785003550698,
      "loss": 12.3403,
      "step": 7505
    },
    {
      "epoch": 0.40873169835208595,
      "grad_norm": 0.6321223134835887,
      "learning_rate": 0.00018447378132486288,
      "loss": 12.3748,
      "step": 7506
    },
    {
      "epoch": 0.40878615234866894,
      "grad_norm": 0.7182095663938053,
      "learning_rate": 0.00018446906163777365,
      "loss": 12.5056,
      "step": 7507
    },
    {
      "epoch": 0.40884060634525193,
      "grad_norm": 0.7045543146464835,
      "learning_rate": 0.00018446434129383885,
      "loss": 12.435,
      "step": 7508
    },
    {
      "epoch": 0.408895060341835,
      "grad_norm": 0.6874610053789847,
      "learning_rate": 0.00018445962029309514,
      "loss": 12.548,
      "step": 7509
    },
    {
      "epoch": 0.408949514338418,
      "grad_norm": 0.6395404723303102,
      "learning_rate": 0.00018445489863557927,
      "loss": 12.5048,
      "step": 7510
    },
    {
      "epoch": 0.40900396833500097,
      "grad_norm": 0.6515443035408928,
      "learning_rate": 0.00018445017632132794,
      "loss": 12.407,
      "step": 7511
    },
    {
      "epoch": 0.409058422331584,
      "grad_norm": 0.6420344734182722,
      "learning_rate": 0.0001844454533503779,
      "loss": 12.4882,
      "step": 7512
    },
    {
      "epoch": 0.409112876328167,
      "grad_norm": 0.6720582995151158,
      "learning_rate": 0.00018444072972276584,
      "loss": 12.7021,
      "step": 7513
    },
    {
      "epoch": 0.40916733032475,
      "grad_norm": 0.6124770996425045,
      "learning_rate": 0.00018443600543852851,
      "loss": 12.3944,
      "step": 7514
    },
    {
      "epoch": 0.40922178432133305,
      "grad_norm": 0.6943339952126241,
      "learning_rate": 0.00018443128049770263,
      "loss": 12.4117,
      "step": 7515
    },
    {
      "epoch": 0.40927623831791604,
      "grad_norm": 0.5849011559469621,
      "learning_rate": 0.00018442655490032498,
      "loss": 12.2855,
      "step": 7516
    },
    {
      "epoch": 0.40933069231449903,
      "grad_norm": 0.7283499647656516,
      "learning_rate": 0.00018442182864643228,
      "loss": 12.4723,
      "step": 7517
    },
    {
      "epoch": 0.4093851463110821,
      "grad_norm": 0.7162779493886587,
      "learning_rate": 0.00018441710173606123,
      "loss": 12.593,
      "step": 7518
    },
    {
      "epoch": 0.4094396003076651,
      "grad_norm": 0.6180021705599357,
      "learning_rate": 0.00018441237416924868,
      "loss": 12.5123,
      "step": 7519
    },
    {
      "epoch": 0.40949405430424807,
      "grad_norm": 0.6582569756408991,
      "learning_rate": 0.00018440764594603135,
      "loss": 12.4382,
      "step": 7520
    },
    {
      "epoch": 0.4095485083008311,
      "grad_norm": 0.6772835849000756,
      "learning_rate": 0.00018440291706644602,
      "loss": 12.5194,
      "step": 7521
    },
    {
      "epoch": 0.4096029622974141,
      "grad_norm": 0.6426675103958575,
      "learning_rate": 0.00018439818753052944,
      "loss": 12.4174,
      "step": 7522
    },
    {
      "epoch": 0.4096574162939971,
      "grad_norm": 0.737761035297852,
      "learning_rate": 0.0001843934573383184,
      "loss": 12.4666,
      "step": 7523
    },
    {
      "epoch": 0.40971187029058015,
      "grad_norm": 0.7037342785824447,
      "learning_rate": 0.00018438872648984965,
      "loss": 12.5767,
      "step": 7524
    },
    {
      "epoch": 0.40976632428716314,
      "grad_norm": 0.7693803436884855,
      "learning_rate": 0.00018438399498516006,
      "loss": 12.6765,
      "step": 7525
    },
    {
      "epoch": 0.4098207782837462,
      "grad_norm": 0.6823533278812086,
      "learning_rate": 0.00018437926282428637,
      "loss": 12.5121,
      "step": 7526
    },
    {
      "epoch": 0.4098752322803292,
      "grad_norm": 0.7116566941304763,
      "learning_rate": 0.00018437453000726538,
      "loss": 12.5272,
      "step": 7527
    },
    {
      "epoch": 0.4099296862769122,
      "grad_norm": 0.6472268275714508,
      "learning_rate": 0.00018436979653413385,
      "loss": 12.619,
      "step": 7528
    },
    {
      "epoch": 0.4099841402734952,
      "grad_norm": 0.625849243204641,
      "learning_rate": 0.0001843650624049287,
      "loss": 12.3751,
      "step": 7529
    },
    {
      "epoch": 0.4100385942700782,
      "grad_norm": 0.6331093988199687,
      "learning_rate": 0.00018436032761968662,
      "loss": 12.4024,
      "step": 7530
    },
    {
      "epoch": 0.4100930482666612,
      "grad_norm": 0.6527662119389117,
      "learning_rate": 0.00018435559217844452,
      "loss": 12.324,
      "step": 7531
    },
    {
      "epoch": 0.41014750226324426,
      "grad_norm": 0.6435280590335135,
      "learning_rate": 0.0001843508560812392,
      "loss": 12.5237,
      "step": 7532
    },
    {
      "epoch": 0.41020195625982725,
      "grad_norm": 0.602130591757125,
      "learning_rate": 0.00018434611932810743,
      "loss": 12.5467,
      "step": 7533
    },
    {
      "epoch": 0.41025641025641024,
      "grad_norm": 0.6666486375237347,
      "learning_rate": 0.00018434138191908615,
      "loss": 12.5347,
      "step": 7534
    },
    {
      "epoch": 0.4103108642529933,
      "grad_norm": 0.63944993563378,
      "learning_rate": 0.0001843366438542121,
      "loss": 12.6362,
      "step": 7535
    },
    {
      "epoch": 0.4103653182495763,
      "grad_norm": 0.6376619610146421,
      "learning_rate": 0.00018433190513352218,
      "loss": 12.4419,
      "step": 7536
    },
    {
      "epoch": 0.4104197722461593,
      "grad_norm": 0.6372513569437053,
      "learning_rate": 0.0001843271657570532,
      "loss": 12.5039,
      "step": 7537
    },
    {
      "epoch": 0.4104742262427423,
      "grad_norm": 0.913361086553666,
      "learning_rate": 0.00018432242572484205,
      "loss": 12.601,
      "step": 7538
    },
    {
      "epoch": 0.4105286802393253,
      "grad_norm": 0.6746220751420223,
      "learning_rate": 0.00018431768503692557,
      "loss": 12.5318,
      "step": 7539
    },
    {
      "epoch": 0.4105831342359083,
      "grad_norm": 0.5908373611740666,
      "learning_rate": 0.00018431294369334065,
      "loss": 12.4197,
      "step": 7540
    },
    {
      "epoch": 0.41063758823249136,
      "grad_norm": 0.5960354209167404,
      "learning_rate": 0.00018430820169412413,
      "loss": 12.3466,
      "step": 7541
    },
    {
      "epoch": 0.41069204222907435,
      "grad_norm": 0.6649575614542876,
      "learning_rate": 0.0001843034590393129,
      "loss": 12.4657,
      "step": 7542
    },
    {
      "epoch": 0.41074649622565734,
      "grad_norm": 0.6638348440625013,
      "learning_rate": 0.0001842987157289438,
      "loss": 12.5338,
      "step": 7543
    },
    {
      "epoch": 0.4108009502222404,
      "grad_norm": 0.6651550518063203,
      "learning_rate": 0.00018429397176305382,
      "loss": 12.337,
      "step": 7544
    },
    {
      "epoch": 0.4108554042188234,
      "grad_norm": 0.6785908004988509,
      "learning_rate": 0.0001842892271416797,
      "loss": 12.4929,
      "step": 7545
    },
    {
      "epoch": 0.4109098582154064,
      "grad_norm": 0.6328267972406906,
      "learning_rate": 0.00018428448186485848,
      "loss": 12.4313,
      "step": 7546
    },
    {
      "epoch": 0.4109643122119894,
      "grad_norm": 0.5950867640188962,
      "learning_rate": 0.00018427973593262696,
      "loss": 12.3173,
      "step": 7547
    },
    {
      "epoch": 0.4110187662085724,
      "grad_norm": 0.7421067335083728,
      "learning_rate": 0.0001842749893450221,
      "loss": 12.7487,
      "step": 7548
    },
    {
      "epoch": 0.4110732202051554,
      "grad_norm": 0.6206223920094129,
      "learning_rate": 0.00018427024210208078,
      "loss": 12.3902,
      "step": 7549
    },
    {
      "epoch": 0.41112767420173846,
      "grad_norm": 0.6604135706333086,
      "learning_rate": 0.0001842654942038399,
      "loss": 12.3789,
      "step": 7550
    },
    {
      "epoch": 0.41118212819832145,
      "grad_norm": 0.6180331174072714,
      "learning_rate": 0.00018426074565033645,
      "loss": 12.5446,
      "step": 7551
    },
    {
      "epoch": 0.41123658219490444,
      "grad_norm": 0.6358907770850538,
      "learning_rate": 0.00018425599644160726,
      "loss": 12.3478,
      "step": 7552
    },
    {
      "epoch": 0.4112910361914875,
      "grad_norm": 0.7649458824755346,
      "learning_rate": 0.0001842512465776894,
      "loss": 12.4298,
      "step": 7553
    },
    {
      "epoch": 0.4113454901880705,
      "grad_norm": 0.661010889082727,
      "learning_rate": 0.0001842464960586196,
      "loss": 12.4604,
      "step": 7554
    },
    {
      "epoch": 0.4113999441846535,
      "grad_norm": 0.5670705848084314,
      "learning_rate": 0.000184241744884435,
      "loss": 12.3934,
      "step": 7555
    },
    {
      "epoch": 0.4114543981812365,
      "grad_norm": 0.6599957599839233,
      "learning_rate": 0.00018423699305517244,
      "loss": 12.5605,
      "step": 7556
    },
    {
      "epoch": 0.4115088521778195,
      "grad_norm": 0.6912314311352966,
      "learning_rate": 0.00018423224057086888,
      "loss": 12.56,
      "step": 7557
    },
    {
      "epoch": 0.4115633061744025,
      "grad_norm": 0.6548538166488451,
      "learning_rate": 0.00018422748743156134,
      "loss": 12.347,
      "step": 7558
    },
    {
      "epoch": 0.41161776017098556,
      "grad_norm": 0.6104316468510993,
      "learning_rate": 0.0001842227336372867,
      "loss": 12.5359,
      "step": 7559
    },
    {
      "epoch": 0.41167221416756855,
      "grad_norm": 0.7374021548975841,
      "learning_rate": 0.00018421797918808194,
      "loss": 12.5152,
      "step": 7560
    },
    {
      "epoch": 0.4117266681641516,
      "grad_norm": 0.6458091136832962,
      "learning_rate": 0.00018421322408398408,
      "loss": 12.5076,
      "step": 7561
    },
    {
      "epoch": 0.4117811221607346,
      "grad_norm": 0.6777619018612451,
      "learning_rate": 0.00018420846832503006,
      "loss": 12.6523,
      "step": 7562
    },
    {
      "epoch": 0.4118355761573176,
      "grad_norm": 0.5836581592065011,
      "learning_rate": 0.00018420371191125686,
      "loss": 12.2898,
      "step": 7563
    },
    {
      "epoch": 0.41189003015390063,
      "grad_norm": 0.728462007314925,
      "learning_rate": 0.0001841989548427015,
      "loss": 12.6287,
      "step": 7564
    },
    {
      "epoch": 0.4119444841504836,
      "grad_norm": 0.7025051926962401,
      "learning_rate": 0.00018419419711940093,
      "loss": 12.4986,
      "step": 7565
    },
    {
      "epoch": 0.4119989381470666,
      "grad_norm": 0.6196466337363593,
      "learning_rate": 0.00018418943874139217,
      "loss": 12.4648,
      "step": 7566
    },
    {
      "epoch": 0.41205339214364967,
      "grad_norm": 0.6256416043269872,
      "learning_rate": 0.00018418467970871222,
      "loss": 12.3932,
      "step": 7567
    },
    {
      "epoch": 0.41210784614023266,
      "grad_norm": 0.636122420202276,
      "learning_rate": 0.00018417992002139807,
      "loss": 12.3566,
      "step": 7568
    },
    {
      "epoch": 0.41216230013681565,
      "grad_norm": 0.6657010523105482,
      "learning_rate": 0.00018417515967948672,
      "loss": 12.3594,
      "step": 7569
    },
    {
      "epoch": 0.4122167541333987,
      "grad_norm": 0.6719196758228524,
      "learning_rate": 0.00018417039868301528,
      "loss": 12.5234,
      "step": 7570
    },
    {
      "epoch": 0.4122712081299817,
      "grad_norm": 0.5998339464207478,
      "learning_rate": 0.00018416563703202064,
      "loss": 12.5577,
      "step": 7571
    },
    {
      "epoch": 0.4123256621265647,
      "grad_norm": 0.8298355480163812,
      "learning_rate": 0.00018416087472653992,
      "loss": 12.6147,
      "step": 7572
    },
    {
      "epoch": 0.41238011612314773,
      "grad_norm": 0.6592033463612728,
      "learning_rate": 0.0001841561117666101,
      "loss": 12.3951,
      "step": 7573
    },
    {
      "epoch": 0.4124345701197307,
      "grad_norm": 0.7207947352174678,
      "learning_rate": 0.00018415134815226826,
      "loss": 12.4531,
      "step": 7574
    },
    {
      "epoch": 0.4124890241163137,
      "grad_norm": 0.6148861802788756,
      "learning_rate": 0.00018414658388355145,
      "loss": 12.4126,
      "step": 7575
    },
    {
      "epoch": 0.41254347811289677,
      "grad_norm": 0.6689104279292235,
      "learning_rate": 0.00018414181896049664,
      "loss": 12.5419,
      "step": 7576
    },
    {
      "epoch": 0.41259793210947976,
      "grad_norm": 0.7014344990842885,
      "learning_rate": 0.00018413705338314097,
      "loss": 12.466,
      "step": 7577
    },
    {
      "epoch": 0.41265238610606275,
      "grad_norm": 0.6714153913382441,
      "learning_rate": 0.00018413228715152147,
      "loss": 12.5583,
      "step": 7578
    },
    {
      "epoch": 0.4127068401026458,
      "grad_norm": 0.7751802041986378,
      "learning_rate": 0.00018412752026567518,
      "loss": 12.6571,
      "step": 7579
    },
    {
      "epoch": 0.4127612940992288,
      "grad_norm": 0.6278998134838513,
      "learning_rate": 0.0001841227527256392,
      "loss": 12.51,
      "step": 7580
    },
    {
      "epoch": 0.4128157480958118,
      "grad_norm": 0.7082243757340546,
      "learning_rate": 0.00018411798453145056,
      "loss": 12.5604,
      "step": 7581
    },
    {
      "epoch": 0.41287020209239483,
      "grad_norm": 0.6326366117703875,
      "learning_rate": 0.00018411321568314638,
      "loss": 12.3858,
      "step": 7582
    },
    {
      "epoch": 0.4129246560889778,
      "grad_norm": 0.6677705390508786,
      "learning_rate": 0.00018410844618076372,
      "loss": 12.5072,
      "step": 7583
    },
    {
      "epoch": 0.4129791100855608,
      "grad_norm": 0.6679816766096263,
      "learning_rate": 0.00018410367602433972,
      "loss": 12.5658,
      "step": 7584
    },
    {
      "epoch": 0.41303356408214387,
      "grad_norm": 0.7019511202137079,
      "learning_rate": 0.00018409890521391137,
      "loss": 12.5263,
      "step": 7585
    },
    {
      "epoch": 0.41308801807872686,
      "grad_norm": 0.6543623414632546,
      "learning_rate": 0.00018409413374951587,
      "loss": 12.4733,
      "step": 7586
    },
    {
      "epoch": 0.41314247207530985,
      "grad_norm": 0.6004737182626931,
      "learning_rate": 0.0001840893616311903,
      "loss": 12.4799,
      "step": 7587
    },
    {
      "epoch": 0.4131969260718929,
      "grad_norm": 0.6714918977825168,
      "learning_rate": 0.0001840845888589717,
      "loss": 12.5035,
      "step": 7588
    },
    {
      "epoch": 0.4132513800684759,
      "grad_norm": 0.6687905292852764,
      "learning_rate": 0.00018407981543289726,
      "loss": 12.4394,
      "step": 7589
    },
    {
      "epoch": 0.4133058340650589,
      "grad_norm": 0.5822307505939365,
      "learning_rate": 0.00018407504135300407,
      "loss": 12.4167,
      "step": 7590
    },
    {
      "epoch": 0.41336028806164193,
      "grad_norm": 0.6797044733620369,
      "learning_rate": 0.00018407026661932928,
      "loss": 12.5009,
      "step": 7591
    },
    {
      "epoch": 0.4134147420582249,
      "grad_norm": 0.6662696811313099,
      "learning_rate": 0.00018406549123190996,
      "loss": 12.4285,
      "step": 7592
    },
    {
      "epoch": 0.413469196054808,
      "grad_norm": 0.6384663985910454,
      "learning_rate": 0.0001840607151907833,
      "loss": 12.4338,
      "step": 7593
    },
    {
      "epoch": 0.41352365005139097,
      "grad_norm": 0.6563451161415367,
      "learning_rate": 0.00018405593849598644,
      "loss": 12.5337,
      "step": 7594
    },
    {
      "epoch": 0.41357810404797396,
      "grad_norm": 0.6007396621690622,
      "learning_rate": 0.00018405116114755647,
      "loss": 12.4726,
      "step": 7595
    },
    {
      "epoch": 0.413632558044557,
      "grad_norm": 0.6356480095698379,
      "learning_rate": 0.00018404638314553062,
      "loss": 12.3176,
      "step": 7596
    },
    {
      "epoch": 0.41368701204114,
      "grad_norm": 0.6362794038608103,
      "learning_rate": 0.00018404160448994597,
      "loss": 12.454,
      "step": 7597
    },
    {
      "epoch": 0.413741466037723,
      "grad_norm": 0.6294171250211859,
      "learning_rate": 0.0001840368251808397,
      "loss": 12.415,
      "step": 7598
    },
    {
      "epoch": 0.41379592003430604,
      "grad_norm": 0.6220974427648912,
      "learning_rate": 0.00018403204521824903,
      "loss": 12.5613,
      "step": 7599
    },
    {
      "epoch": 0.41385037403088903,
      "grad_norm": 0.7140220218064345,
      "learning_rate": 0.00018402726460221104,
      "loss": 12.4242,
      "step": 7600
    },
    {
      "epoch": 0.413904828027472,
      "grad_norm": 0.6264698242072783,
      "learning_rate": 0.00018402248333276297,
      "loss": 12.4845,
      "step": 7601
    },
    {
      "epoch": 0.4139592820240551,
      "grad_norm": 0.6131916443343639,
      "learning_rate": 0.00018401770140994198,
      "loss": 12.4454,
      "step": 7602
    },
    {
      "epoch": 0.41401373602063807,
      "grad_norm": 0.7080235427810602,
      "learning_rate": 0.00018401291883378523,
      "loss": 12.4933,
      "step": 7603
    },
    {
      "epoch": 0.41406819001722106,
      "grad_norm": 0.7726758600328698,
      "learning_rate": 0.00018400813560432997,
      "loss": 12.4119,
      "step": 7604
    },
    {
      "epoch": 0.4141226440138041,
      "grad_norm": 0.6969997285025378,
      "learning_rate": 0.00018400335172161333,
      "loss": 12.4852,
      "step": 7605
    },
    {
      "epoch": 0.4141770980103871,
      "grad_norm": 0.6923251834896971,
      "learning_rate": 0.00018399856718567256,
      "loss": 12.5334,
      "step": 7606
    },
    {
      "epoch": 0.4142315520069701,
      "grad_norm": 0.709215148883989,
      "learning_rate": 0.00018399378199654486,
      "loss": 12.4814,
      "step": 7607
    },
    {
      "epoch": 0.41428600600355314,
      "grad_norm": 0.7182335665032548,
      "learning_rate": 0.00018398899615426737,
      "loss": 12.5802,
      "step": 7608
    },
    {
      "epoch": 0.41434046000013613,
      "grad_norm": 0.7747472579587373,
      "learning_rate": 0.00018398420965887738,
      "loss": 12.3487,
      "step": 7609
    },
    {
      "epoch": 0.4143949139967191,
      "grad_norm": 0.6641584285672352,
      "learning_rate": 0.00018397942251041212,
      "loss": 12.4149,
      "step": 7610
    },
    {
      "epoch": 0.4144493679933022,
      "grad_norm": 0.6559558849248063,
      "learning_rate": 0.00018397463470890877,
      "loss": 12.4061,
      "step": 7611
    },
    {
      "epoch": 0.41450382198988517,
      "grad_norm": 0.6929051117161943,
      "learning_rate": 0.00018396984625440458,
      "loss": 12.4807,
      "step": 7612
    },
    {
      "epoch": 0.41455827598646816,
      "grad_norm": 0.6789925394878573,
      "learning_rate": 0.00018396505714693678,
      "loss": 12.5116,
      "step": 7613
    },
    {
      "epoch": 0.4146127299830512,
      "grad_norm": 0.7897268899930915,
      "learning_rate": 0.00018396026738654264,
      "loss": 12.432,
      "step": 7614
    },
    {
      "epoch": 0.4146671839796342,
      "grad_norm": 0.6807339651106654,
      "learning_rate": 0.00018395547697325933,
      "loss": 12.3411,
      "step": 7615
    },
    {
      "epoch": 0.4147216379762172,
      "grad_norm": 0.6682991423741861,
      "learning_rate": 0.00018395068590712417,
      "loss": 12.3832,
      "step": 7616
    },
    {
      "epoch": 0.41477609197280024,
      "grad_norm": 0.7746811225957453,
      "learning_rate": 0.00018394589418817443,
      "loss": 12.4631,
      "step": 7617
    },
    {
      "epoch": 0.41483054596938324,
      "grad_norm": 0.6964451730482353,
      "learning_rate": 0.00018394110181644733,
      "loss": 12.3519,
      "step": 7618
    },
    {
      "epoch": 0.41488499996596623,
      "grad_norm": 0.6527959058968399,
      "learning_rate": 0.00018393630879198013,
      "loss": 12.4443,
      "step": 7619
    },
    {
      "epoch": 0.4149394539625493,
      "grad_norm": 0.7411057629441736,
      "learning_rate": 0.0001839315151148101,
      "loss": 12.3707,
      "step": 7620
    },
    {
      "epoch": 0.41499390795913227,
      "grad_norm": 0.742397366365453,
      "learning_rate": 0.00018392672078497454,
      "loss": 12.5815,
      "step": 7621
    },
    {
      "epoch": 0.41504836195571526,
      "grad_norm": 0.6845206549741516,
      "learning_rate": 0.00018392192580251075,
      "loss": 12.5448,
      "step": 7622
    },
    {
      "epoch": 0.4151028159522983,
      "grad_norm": 0.8617869176209406,
      "learning_rate": 0.00018391713016745596,
      "loss": 12.5105,
      "step": 7623
    },
    {
      "epoch": 0.4151572699488813,
      "grad_norm": 0.6332044702163018,
      "learning_rate": 0.00018391233387984754,
      "loss": 12.4949,
      "step": 7624
    },
    {
      "epoch": 0.4152117239454643,
      "grad_norm": 0.6921877222300677,
      "learning_rate": 0.0001839075369397227,
      "loss": 12.44,
      "step": 7625
    },
    {
      "epoch": 0.41526617794204734,
      "grad_norm": 0.7304182216130374,
      "learning_rate": 0.0001839027393471188,
      "loss": 12.5507,
      "step": 7626
    },
    {
      "epoch": 0.41532063193863034,
      "grad_norm": 0.6844024377851525,
      "learning_rate": 0.00018389794110207312,
      "loss": 12.5578,
      "step": 7627
    },
    {
      "epoch": 0.4153750859352134,
      "grad_norm": 0.6777229470414581,
      "learning_rate": 0.00018389314220462296,
      "loss": 12.4082,
      "step": 7628
    },
    {
      "epoch": 0.4154295399317964,
      "grad_norm": 0.6432218551933113,
      "learning_rate": 0.0001838883426548057,
      "loss": 12.5191,
      "step": 7629
    },
    {
      "epoch": 0.41548399392837937,
      "grad_norm": 0.6859378464003325,
      "learning_rate": 0.00018388354245265858,
      "loss": 12.5486,
      "step": 7630
    },
    {
      "epoch": 0.4155384479249624,
      "grad_norm": 0.6671747399551233,
      "learning_rate": 0.000183878741598219,
      "loss": 12.5289,
      "step": 7631
    },
    {
      "epoch": 0.4155929019215454,
      "grad_norm": 0.6850227729337616,
      "learning_rate": 0.00018387394009152425,
      "loss": 12.5237,
      "step": 7632
    },
    {
      "epoch": 0.4156473559181284,
      "grad_norm": 0.7907722026545728,
      "learning_rate": 0.00018386913793261167,
      "loss": 12.6377,
      "step": 7633
    },
    {
      "epoch": 0.41570180991471145,
      "grad_norm": 0.7024232324059682,
      "learning_rate": 0.0001838643351215186,
      "loss": 12.4349,
      "step": 7634
    },
    {
      "epoch": 0.41575626391129444,
      "grad_norm": 0.703812480485606,
      "learning_rate": 0.0001838595316582824,
      "loss": 12.5366,
      "step": 7635
    },
    {
      "epoch": 0.41581071790787744,
      "grad_norm": 0.7252726705505731,
      "learning_rate": 0.00018385472754294042,
      "loss": 12.5311,
      "step": 7636
    },
    {
      "epoch": 0.4158651719044605,
      "grad_norm": 0.7160236603812885,
      "learning_rate": 0.00018384992277553001,
      "loss": 12.5574,
      "step": 7637
    },
    {
      "epoch": 0.4159196259010435,
      "grad_norm": 0.6561769782974751,
      "learning_rate": 0.00018384511735608855,
      "loss": 12.3621,
      "step": 7638
    },
    {
      "epoch": 0.41597407989762647,
      "grad_norm": 0.7642371517971974,
      "learning_rate": 0.0001838403112846534,
      "loss": 12.4601,
      "step": 7639
    },
    {
      "epoch": 0.4160285338942095,
      "grad_norm": 0.6832445256349677,
      "learning_rate": 0.00018383550456126192,
      "loss": 12.3823,
      "step": 7640
    },
    {
      "epoch": 0.4160829878907925,
      "grad_norm": 0.9046044716944612,
      "learning_rate": 0.00018383069718595153,
      "loss": 12.681,
      "step": 7641
    },
    {
      "epoch": 0.4161374418873755,
      "grad_norm": 0.6990758835934934,
      "learning_rate": 0.00018382588915875952,
      "loss": 12.5077,
      "step": 7642
    },
    {
      "epoch": 0.41619189588395855,
      "grad_norm": 0.818672399509247,
      "learning_rate": 0.00018382108047972336,
      "loss": 12.4796,
      "step": 7643
    },
    {
      "epoch": 0.41624634988054154,
      "grad_norm": 0.6347657355980698,
      "learning_rate": 0.00018381627114888045,
      "loss": 12.5621,
      "step": 7644
    },
    {
      "epoch": 0.41630080387712454,
      "grad_norm": 0.7051436257130543,
      "learning_rate": 0.00018381146116626816,
      "loss": 12.3868,
      "step": 7645
    },
    {
      "epoch": 0.4163552578737076,
      "grad_norm": 0.6580907573187873,
      "learning_rate": 0.00018380665053192386,
      "loss": 12.5146,
      "step": 7646
    },
    {
      "epoch": 0.4164097118702906,
      "grad_norm": 0.7377588067672651,
      "learning_rate": 0.00018380183924588498,
      "loss": 12.5493,
      "step": 7647
    },
    {
      "epoch": 0.41646416586687357,
      "grad_norm": 0.6434057170985927,
      "learning_rate": 0.000183797027308189,
      "loss": 12.4146,
      "step": 7648
    },
    {
      "epoch": 0.4165186198634566,
      "grad_norm": 0.6201660562859044,
      "learning_rate": 0.00018379221471887325,
      "loss": 12.4045,
      "step": 7649
    },
    {
      "epoch": 0.4165730738600396,
      "grad_norm": 0.7312758361049857,
      "learning_rate": 0.00018378740147797517,
      "loss": 12.5651,
      "step": 7650
    },
    {
      "epoch": 0.4166275278566226,
      "grad_norm": 0.6562949596475576,
      "learning_rate": 0.00018378258758553222,
      "loss": 12.4739,
      "step": 7651
    },
    {
      "epoch": 0.41668198185320565,
      "grad_norm": 0.8109608827473153,
      "learning_rate": 0.00018377777304158182,
      "loss": 12.3613,
      "step": 7652
    },
    {
      "epoch": 0.41673643584978864,
      "grad_norm": 0.6547422099606501,
      "learning_rate": 0.00018377295784616142,
      "loss": 12.5337,
      "step": 7653
    },
    {
      "epoch": 0.41679088984637164,
      "grad_norm": 0.7061811226970356,
      "learning_rate": 0.00018376814199930842,
      "loss": 12.5456,
      "step": 7654
    },
    {
      "epoch": 0.4168453438429547,
      "grad_norm": 0.6012608861041727,
      "learning_rate": 0.00018376332550106033,
      "loss": 12.533,
      "step": 7655
    },
    {
      "epoch": 0.4168997978395377,
      "grad_norm": 0.7006711823775731,
      "learning_rate": 0.00018375850835145456,
      "loss": 12.5389,
      "step": 7656
    },
    {
      "epoch": 0.41695425183612067,
      "grad_norm": 0.6580649051091263,
      "learning_rate": 0.0001837536905505286,
      "loss": 12.4394,
      "step": 7657
    },
    {
      "epoch": 0.4170087058327037,
      "grad_norm": 0.6760449622537299,
      "learning_rate": 0.00018374887209831987,
      "loss": 12.3788,
      "step": 7658
    },
    {
      "epoch": 0.4170631598292867,
      "grad_norm": 0.6497353580956746,
      "learning_rate": 0.00018374405299486588,
      "loss": 12.4658,
      "step": 7659
    },
    {
      "epoch": 0.41711761382586976,
      "grad_norm": 0.652611595647118,
      "learning_rate": 0.0001837392332402041,
      "loss": 12.4612,
      "step": 7660
    },
    {
      "epoch": 0.41717206782245275,
      "grad_norm": 0.6578766692847864,
      "learning_rate": 0.00018373441283437198,
      "loss": 12.4691,
      "step": 7661
    },
    {
      "epoch": 0.41722652181903574,
      "grad_norm": 0.6382698831671395,
      "learning_rate": 0.00018372959177740704,
      "loss": 12.4599,
      "step": 7662
    },
    {
      "epoch": 0.4172809758156188,
      "grad_norm": 0.6850091154967107,
      "learning_rate": 0.00018372477006934674,
      "loss": 12.4671,
      "step": 7663
    },
    {
      "epoch": 0.4173354298122018,
      "grad_norm": 0.6298292742381474,
      "learning_rate": 0.0001837199477102286,
      "loss": 12.4673,
      "step": 7664
    },
    {
      "epoch": 0.4173898838087848,
      "grad_norm": 0.6113991567651845,
      "learning_rate": 0.00018371512470009008,
      "loss": 12.3753,
      "step": 7665
    },
    {
      "epoch": 0.4174443378053678,
      "grad_norm": 0.7055999705123408,
      "learning_rate": 0.00018371030103896872,
      "loss": 12.5082,
      "step": 7666
    },
    {
      "epoch": 0.4174987918019508,
      "grad_norm": 0.6371229065148822,
      "learning_rate": 0.00018370547672690206,
      "loss": 12.4813,
      "step": 7667
    },
    {
      "epoch": 0.4175532457985338,
      "grad_norm": 0.7015226954955563,
      "learning_rate": 0.00018370065176392752,
      "loss": 12.4849,
      "step": 7668
    },
    {
      "epoch": 0.41760769979511686,
      "grad_norm": 0.6310519097840122,
      "learning_rate": 0.00018369582615008272,
      "loss": 12.4796,
      "step": 7669
    },
    {
      "epoch": 0.41766215379169985,
      "grad_norm": 0.5944925755085689,
      "learning_rate": 0.00018369099988540513,
      "loss": 12.2906,
      "step": 7670
    },
    {
      "epoch": 0.41771660778828285,
      "grad_norm": 0.6578983386471594,
      "learning_rate": 0.00018368617296993226,
      "loss": 12.5302,
      "step": 7671
    },
    {
      "epoch": 0.4177710617848659,
      "grad_norm": 0.6437522523863247,
      "learning_rate": 0.00018368134540370173,
      "loss": 12.4882,
      "step": 7672
    },
    {
      "epoch": 0.4178255157814489,
      "grad_norm": 0.6162412646017736,
      "learning_rate": 0.000183676517186751,
      "loss": 12.4843,
      "step": 7673
    },
    {
      "epoch": 0.4178799697780319,
      "grad_norm": 0.6899640431171089,
      "learning_rate": 0.0001836716883191176,
      "loss": 12.5048,
      "step": 7674
    },
    {
      "epoch": 0.4179344237746149,
      "grad_norm": 0.6672292118932008,
      "learning_rate": 0.00018366685880083914,
      "loss": 12.4681,
      "step": 7675
    },
    {
      "epoch": 0.4179888777711979,
      "grad_norm": 0.6079786159449238,
      "learning_rate": 0.00018366202863195316,
      "loss": 12.4824,
      "step": 7676
    },
    {
      "epoch": 0.4180433317677809,
      "grad_norm": 0.7544946764584993,
      "learning_rate": 0.00018365719781249725,
      "loss": 12.5876,
      "step": 7677
    },
    {
      "epoch": 0.41809778576436396,
      "grad_norm": 0.5962533245030253,
      "learning_rate": 0.0001836523663425089,
      "loss": 12.4612,
      "step": 7678
    },
    {
      "epoch": 0.41815223976094695,
      "grad_norm": 0.6123575398733694,
      "learning_rate": 0.00018364753422202575,
      "loss": 12.5313,
      "step": 7679
    },
    {
      "epoch": 0.41820669375752995,
      "grad_norm": 0.6940503398342436,
      "learning_rate": 0.00018364270145108531,
      "loss": 12.5112,
      "step": 7680
    },
    {
      "epoch": 0.418261147754113,
      "grad_norm": 0.6680628632067475,
      "learning_rate": 0.00018363786802972522,
      "loss": 12.4506,
      "step": 7681
    },
    {
      "epoch": 0.418315601750696,
      "grad_norm": 0.6543354862728545,
      "learning_rate": 0.00018363303395798304,
      "loss": 12.4137,
      "step": 7682
    },
    {
      "epoch": 0.418370055747279,
      "grad_norm": 0.6030576150582967,
      "learning_rate": 0.00018362819923589636,
      "loss": 12.4018,
      "step": 7683
    },
    {
      "epoch": 0.418424509743862,
      "grad_norm": 0.6680144702062961,
      "learning_rate": 0.00018362336386350275,
      "loss": 12.4991,
      "step": 7684
    },
    {
      "epoch": 0.418478963740445,
      "grad_norm": 0.6222373278682574,
      "learning_rate": 0.00018361852784083991,
      "loss": 12.3989,
      "step": 7685
    },
    {
      "epoch": 0.418533417737028,
      "grad_norm": 0.6886589633076576,
      "learning_rate": 0.0001836136911679453,
      "loss": 12.402,
      "step": 7686
    },
    {
      "epoch": 0.41858787173361106,
      "grad_norm": 0.6532456915278813,
      "learning_rate": 0.00018360885384485664,
      "loss": 12.4871,
      "step": 7687
    },
    {
      "epoch": 0.41864232573019405,
      "grad_norm": 1.1426480829233576,
      "learning_rate": 0.0001836040158716115,
      "loss": 12.3851,
      "step": 7688
    },
    {
      "epoch": 0.41869677972677705,
      "grad_norm": 0.7287143760029692,
      "learning_rate": 0.00018359917724824752,
      "loss": 12.582,
      "step": 7689
    },
    {
      "epoch": 0.4187512337233601,
      "grad_norm": 0.6414605281108499,
      "learning_rate": 0.00018359433797480234,
      "loss": 12.2483,
      "step": 7690
    },
    {
      "epoch": 0.4188056877199431,
      "grad_norm": 0.5842099261062814,
      "learning_rate": 0.00018358949805131352,
      "loss": 12.373,
      "step": 7691
    },
    {
      "epoch": 0.4188601417165261,
      "grad_norm": 0.6257821953679695,
      "learning_rate": 0.00018358465747781878,
      "loss": 12.4152,
      "step": 7692
    },
    {
      "epoch": 0.41891459571310913,
      "grad_norm": 0.7349478349237475,
      "learning_rate": 0.00018357981625435573,
      "loss": 12.4382,
      "step": 7693
    },
    {
      "epoch": 0.4189690497096921,
      "grad_norm": 0.6866824978335202,
      "learning_rate": 0.000183574974380962,
      "loss": 12.5315,
      "step": 7694
    },
    {
      "epoch": 0.41902350370627517,
      "grad_norm": 0.7215980246716968,
      "learning_rate": 0.00018357013185767526,
      "loss": 12.6287,
      "step": 7695
    },
    {
      "epoch": 0.41907795770285816,
      "grad_norm": 0.7164121518155415,
      "learning_rate": 0.00018356528868453316,
      "loss": 12.5496,
      "step": 7696
    },
    {
      "epoch": 0.41913241169944115,
      "grad_norm": 0.6174833109503086,
      "learning_rate": 0.00018356044486157334,
      "loss": 12.4704,
      "step": 7697
    },
    {
      "epoch": 0.4191868656960242,
      "grad_norm": 0.7061268318151425,
      "learning_rate": 0.00018355560038883353,
      "loss": 12.4084,
      "step": 7698
    },
    {
      "epoch": 0.4192413196926072,
      "grad_norm": 0.7113906455959584,
      "learning_rate": 0.00018355075526635132,
      "loss": 12.5068,
      "step": 7699
    },
    {
      "epoch": 0.4192957736891902,
      "grad_norm": 0.6160925389795541,
      "learning_rate": 0.00018354590949416446,
      "loss": 12.3842,
      "step": 7700
    },
    {
      "epoch": 0.41935022768577324,
      "grad_norm": 0.764056388356585,
      "learning_rate": 0.00018354106307231057,
      "loss": 12.5408,
      "step": 7701
    },
    {
      "epoch": 0.41940468168235623,
      "grad_norm": 0.6184383724182737,
      "learning_rate": 0.00018353621600082737,
      "loss": 12.4184,
      "step": 7702
    },
    {
      "epoch": 0.4194591356789392,
      "grad_norm": 0.7259817883996311,
      "learning_rate": 0.00018353136827975255,
      "loss": 12.4332,
      "step": 7703
    },
    {
      "epoch": 0.41951358967552227,
      "grad_norm": 0.7038668489917369,
      "learning_rate": 0.0001835265199091238,
      "loss": 12.4895,
      "step": 7704
    },
    {
      "epoch": 0.41956804367210526,
      "grad_norm": 0.6629703207999516,
      "learning_rate": 0.0001835216708889788,
      "loss": 12.4573,
      "step": 7705
    },
    {
      "epoch": 0.41962249766868825,
      "grad_norm": 0.6143947927874756,
      "learning_rate": 0.0001835168212193553,
      "loss": 12.5232,
      "step": 7706
    },
    {
      "epoch": 0.4196769516652713,
      "grad_norm": 0.6349297012237477,
      "learning_rate": 0.000183511970900291,
      "loss": 12.567,
      "step": 7707
    },
    {
      "epoch": 0.4197314056618543,
      "grad_norm": 0.824480006731654,
      "learning_rate": 0.0001835071199318236,
      "loss": 12.451,
      "step": 7708
    },
    {
      "epoch": 0.4197858596584373,
      "grad_norm": 0.658652130143729,
      "learning_rate": 0.00018350226831399084,
      "loss": 12.4251,
      "step": 7709
    },
    {
      "epoch": 0.41984031365502034,
      "grad_norm": 0.6313674763579805,
      "learning_rate": 0.00018349741604683045,
      "loss": 12.435,
      "step": 7710
    },
    {
      "epoch": 0.41989476765160333,
      "grad_norm": 0.6637861440684033,
      "learning_rate": 0.00018349256313038013,
      "loss": 12.4185,
      "step": 7711
    },
    {
      "epoch": 0.4199492216481863,
      "grad_norm": 0.7100167755541632,
      "learning_rate": 0.00018348770956467766,
      "loss": 12.5442,
      "step": 7712
    },
    {
      "epoch": 0.42000367564476937,
      "grad_norm": 0.6598267047884887,
      "learning_rate": 0.0001834828553497607,
      "loss": 12.3989,
      "step": 7713
    },
    {
      "epoch": 0.42005812964135236,
      "grad_norm": 0.6461052791122609,
      "learning_rate": 0.0001834780004856671,
      "loss": 12.3846,
      "step": 7714
    },
    {
      "epoch": 0.42011258363793536,
      "grad_norm": 0.6652200852057961,
      "learning_rate": 0.00018347314497243458,
      "loss": 12.634,
      "step": 7715
    },
    {
      "epoch": 0.4201670376345184,
      "grad_norm": 0.6463769020763129,
      "learning_rate": 0.0001834682888101009,
      "loss": 12.3472,
      "step": 7716
    },
    {
      "epoch": 0.4202214916311014,
      "grad_norm": 0.7017130866496301,
      "learning_rate": 0.00018346343199870374,
      "loss": 12.4745,
      "step": 7717
    },
    {
      "epoch": 0.4202759456276844,
      "grad_norm": 0.8069620567128515,
      "learning_rate": 0.000183458574538281,
      "loss": 12.4645,
      "step": 7718
    },
    {
      "epoch": 0.42033039962426744,
      "grad_norm": 0.6694114333255915,
      "learning_rate": 0.00018345371642887034,
      "loss": 12.4017,
      "step": 7719
    },
    {
      "epoch": 0.42038485362085043,
      "grad_norm": 0.6409125471734358,
      "learning_rate": 0.0001834488576705096,
      "loss": 12.4359,
      "step": 7720
    },
    {
      "epoch": 0.4204393076174334,
      "grad_norm": 0.697416875461442,
      "learning_rate": 0.0001834439982632366,
      "loss": 12.5134,
      "step": 7721
    },
    {
      "epoch": 0.42049376161401647,
      "grad_norm": 0.6530676618799656,
      "learning_rate": 0.00018343913820708903,
      "loss": 12.5292,
      "step": 7722
    },
    {
      "epoch": 0.42054821561059946,
      "grad_norm": 0.6277856617161583,
      "learning_rate": 0.00018343427750210474,
      "loss": 12.5292,
      "step": 7723
    },
    {
      "epoch": 0.42060266960718246,
      "grad_norm": 0.6596959812273022,
      "learning_rate": 0.00018342941614832149,
      "loss": 12.5169,
      "step": 7724
    },
    {
      "epoch": 0.4206571236037655,
      "grad_norm": 0.6553076426372618,
      "learning_rate": 0.00018342455414577713,
      "loss": 12.5536,
      "step": 7725
    },
    {
      "epoch": 0.4207115776003485,
      "grad_norm": 0.7052686980586507,
      "learning_rate": 0.00018341969149450943,
      "loss": 12.4405,
      "step": 7726
    },
    {
      "epoch": 0.42076603159693154,
      "grad_norm": 0.6433534034167304,
      "learning_rate": 0.00018341482819455625,
      "loss": 12.5039,
      "step": 7727
    },
    {
      "epoch": 0.42082048559351454,
      "grad_norm": 0.6185445890377947,
      "learning_rate": 0.00018340996424595537,
      "loss": 12.4427,
      "step": 7728
    },
    {
      "epoch": 0.42087493959009753,
      "grad_norm": 0.6349314110007439,
      "learning_rate": 0.0001834050996487446,
      "loss": 12.5068,
      "step": 7729
    },
    {
      "epoch": 0.4209293935866806,
      "grad_norm": 0.6737060455245588,
      "learning_rate": 0.00018340023440296182,
      "loss": 12.4331,
      "step": 7730
    },
    {
      "epoch": 0.42098384758326357,
      "grad_norm": 0.6676306550303723,
      "learning_rate": 0.00018339536850864478,
      "loss": 12.59,
      "step": 7731
    },
    {
      "epoch": 0.42103830157984656,
      "grad_norm": 0.673849100112351,
      "learning_rate": 0.00018339050196583144,
      "loss": 12.4062,
      "step": 7732
    },
    {
      "epoch": 0.4210927555764296,
      "grad_norm": 0.6657925192980901,
      "learning_rate": 0.0001833856347745595,
      "loss": 12.4642,
      "step": 7733
    },
    {
      "epoch": 0.4211472095730126,
      "grad_norm": 0.7768256422367489,
      "learning_rate": 0.00018338076693486693,
      "loss": 12.54,
      "step": 7734
    },
    {
      "epoch": 0.4212016635695956,
      "grad_norm": 0.6839333718371754,
      "learning_rate": 0.00018337589844679152,
      "loss": 12.6317,
      "step": 7735
    },
    {
      "epoch": 0.42125611756617865,
      "grad_norm": 0.7189502447091988,
      "learning_rate": 0.00018337102931037112,
      "loss": 12.4953,
      "step": 7736
    },
    {
      "epoch": 0.42131057156276164,
      "grad_norm": 0.622438423831334,
      "learning_rate": 0.00018336615952564364,
      "loss": 12.4795,
      "step": 7737
    },
    {
      "epoch": 0.42136502555934463,
      "grad_norm": 0.6967805388351358,
      "learning_rate": 0.00018336128909264692,
      "loss": 12.2899,
      "step": 7738
    },
    {
      "epoch": 0.4214194795559277,
      "grad_norm": 0.7283347715661034,
      "learning_rate": 0.00018335641801141883,
      "loss": 12.5149,
      "step": 7739
    },
    {
      "epoch": 0.42147393355251067,
      "grad_norm": 0.7300681925656116,
      "learning_rate": 0.00018335154628199726,
      "loss": 12.4365,
      "step": 7740
    },
    {
      "epoch": 0.42152838754909366,
      "grad_norm": 0.886704235630511,
      "learning_rate": 0.0001833466739044201,
      "loss": 12.4463,
      "step": 7741
    },
    {
      "epoch": 0.4215828415456767,
      "grad_norm": 0.6432836336371802,
      "learning_rate": 0.0001833418008787252,
      "loss": 12.4491,
      "step": 7742
    },
    {
      "epoch": 0.4216372955422597,
      "grad_norm": 0.7927372042689552,
      "learning_rate": 0.0001833369272049505,
      "loss": 12.4499,
      "step": 7743
    },
    {
      "epoch": 0.4216917495388427,
      "grad_norm": 0.6784228826890297,
      "learning_rate": 0.00018333205288313385,
      "loss": 12.4603,
      "step": 7744
    },
    {
      "epoch": 0.42174620353542575,
      "grad_norm": 0.6279210026813303,
      "learning_rate": 0.0001833271779133132,
      "loss": 12.5222,
      "step": 7745
    },
    {
      "epoch": 0.42180065753200874,
      "grad_norm": 0.6901241842062449,
      "learning_rate": 0.00018332230229552645,
      "loss": 12.5653,
      "step": 7746
    },
    {
      "epoch": 0.42185511152859173,
      "grad_norm": 0.7646765823088739,
      "learning_rate": 0.0001833174260298115,
      "loss": 12.4381,
      "step": 7747
    },
    {
      "epoch": 0.4219095655251748,
      "grad_norm": 0.5871207013967358,
      "learning_rate": 0.00018331254911620626,
      "loss": 12.3703,
      "step": 7748
    },
    {
      "epoch": 0.42196401952175777,
      "grad_norm": 0.6341582639526494,
      "learning_rate": 0.00018330767155474867,
      "loss": 12.5204,
      "step": 7749
    },
    {
      "epoch": 0.42201847351834076,
      "grad_norm": 0.7186670813039878,
      "learning_rate": 0.00018330279334547668,
      "loss": 12.5855,
      "step": 7750
    },
    {
      "epoch": 0.4220729275149238,
      "grad_norm": 0.7608887056167734,
      "learning_rate": 0.0001832979144884282,
      "loss": 12.4069,
      "step": 7751
    },
    {
      "epoch": 0.4221273815115068,
      "grad_norm": 0.7286736354425516,
      "learning_rate": 0.00018329303498364113,
      "loss": 12.5926,
      "step": 7752
    },
    {
      "epoch": 0.4221818355080898,
      "grad_norm": 0.6667611301995382,
      "learning_rate": 0.00018328815483115344,
      "loss": 12.364,
      "step": 7753
    },
    {
      "epoch": 0.42223628950467285,
      "grad_norm": 0.8559370596551145,
      "learning_rate": 0.00018328327403100314,
      "loss": 12.4806,
      "step": 7754
    },
    {
      "epoch": 0.42229074350125584,
      "grad_norm": 0.5907731907724925,
      "learning_rate": 0.00018327839258322812,
      "loss": 12.3898,
      "step": 7755
    },
    {
      "epoch": 0.42234519749783883,
      "grad_norm": 0.7229290398251434,
      "learning_rate": 0.00018327351048786635,
      "loss": 12.6165,
      "step": 7756
    },
    {
      "epoch": 0.4223996514944219,
      "grad_norm": 0.7068869446224217,
      "learning_rate": 0.00018326862774495578,
      "loss": 12.5796,
      "step": 7757
    },
    {
      "epoch": 0.42245410549100487,
      "grad_norm": 0.6291494939304955,
      "learning_rate": 0.00018326374435453441,
      "loss": 12.4155,
      "step": 7758
    },
    {
      "epoch": 0.42250855948758786,
      "grad_norm": 0.6855015933078603,
      "learning_rate": 0.00018325886031664022,
      "loss": 12.5138,
      "step": 7759
    },
    {
      "epoch": 0.4225630134841709,
      "grad_norm": 0.6254927977595047,
      "learning_rate": 0.00018325397563131115,
      "loss": 12.4501,
      "step": 7760
    },
    {
      "epoch": 0.4226174674807539,
      "grad_norm": 0.5962974777849228,
      "learning_rate": 0.0001832490902985852,
      "loss": 12.5191,
      "step": 7761
    },
    {
      "epoch": 0.42267192147733695,
      "grad_norm": 0.6723955151607713,
      "learning_rate": 0.00018324420431850037,
      "loss": 12.298,
      "step": 7762
    },
    {
      "epoch": 0.42272637547391995,
      "grad_norm": 0.7398066987283984,
      "learning_rate": 0.00018323931769109465,
      "loss": 12.5349,
      "step": 7763
    },
    {
      "epoch": 0.42278082947050294,
      "grad_norm": 0.6969458920048713,
      "learning_rate": 0.00018323443041640602,
      "loss": 12.4679,
      "step": 7764
    },
    {
      "epoch": 0.422835283467086,
      "grad_norm": 0.6126110579796679,
      "learning_rate": 0.00018322954249447252,
      "loss": 12.4865,
      "step": 7765
    },
    {
      "epoch": 0.422889737463669,
      "grad_norm": 0.7092533999165815,
      "learning_rate": 0.00018322465392533216,
      "loss": 12.5509,
      "step": 7766
    },
    {
      "epoch": 0.422944191460252,
      "grad_norm": 0.6474367495039208,
      "learning_rate": 0.0001832197647090229,
      "loss": 12.3412,
      "step": 7767
    },
    {
      "epoch": 0.422998645456835,
      "grad_norm": 0.6735126182790927,
      "learning_rate": 0.00018321487484558276,
      "loss": 12.508,
      "step": 7768
    },
    {
      "epoch": 0.423053099453418,
      "grad_norm": 0.6488168446659008,
      "learning_rate": 0.00018320998433504987,
      "loss": 12.5817,
      "step": 7769
    },
    {
      "epoch": 0.423107553450001,
      "grad_norm": 0.706447803660833,
      "learning_rate": 0.00018320509317746217,
      "loss": 12.4491,
      "step": 7770
    },
    {
      "epoch": 0.42316200744658405,
      "grad_norm": 0.690598879814916,
      "learning_rate": 0.00018320020137285766,
      "loss": 12.4504,
      "step": 7771
    },
    {
      "epoch": 0.42321646144316705,
      "grad_norm": 0.6087048075104945,
      "learning_rate": 0.0001831953089212745,
      "loss": 12.554,
      "step": 7772
    },
    {
      "epoch": 0.42327091543975004,
      "grad_norm": 0.6844601908693717,
      "learning_rate": 0.00018319041582275062,
      "loss": 12.4907,
      "step": 7773
    },
    {
      "epoch": 0.4233253694363331,
      "grad_norm": 0.6368271217345263,
      "learning_rate": 0.00018318552207732415,
      "loss": 12.5252,
      "step": 7774
    },
    {
      "epoch": 0.4233798234329161,
      "grad_norm": 0.6215840343054171,
      "learning_rate": 0.0001831806276850331,
      "loss": 12.5888,
      "step": 7775
    },
    {
      "epoch": 0.4234342774294991,
      "grad_norm": 0.6224392953398991,
      "learning_rate": 0.00018317573264591553,
      "loss": 12.4872,
      "step": 7776
    },
    {
      "epoch": 0.4234887314260821,
      "grad_norm": 0.7030493089317692,
      "learning_rate": 0.0001831708369600095,
      "loss": 12.5528,
      "step": 7777
    },
    {
      "epoch": 0.4235431854226651,
      "grad_norm": 0.6884725730053268,
      "learning_rate": 0.0001831659406273531,
      "loss": 12.4928,
      "step": 7778
    },
    {
      "epoch": 0.4235976394192481,
      "grad_norm": 0.6763106331873534,
      "learning_rate": 0.00018316104364798444,
      "loss": 12.4183,
      "step": 7779
    },
    {
      "epoch": 0.42365209341583115,
      "grad_norm": 0.6405527677746562,
      "learning_rate": 0.00018315614602194152,
      "loss": 12.4737,
      "step": 7780
    },
    {
      "epoch": 0.42370654741241415,
      "grad_norm": 0.6945702310438716,
      "learning_rate": 0.00018315124774926248,
      "loss": 12.5326,
      "step": 7781
    },
    {
      "epoch": 0.42376100140899714,
      "grad_norm": 0.6044615472501069,
      "learning_rate": 0.00018314634882998538,
      "loss": 12.3923,
      "step": 7782
    },
    {
      "epoch": 0.4238154554055802,
      "grad_norm": 0.6661273133787237,
      "learning_rate": 0.00018314144926414834,
      "loss": 12.399,
      "step": 7783
    },
    {
      "epoch": 0.4238699094021632,
      "grad_norm": 0.7311006197242605,
      "learning_rate": 0.00018313654905178944,
      "loss": 12.4251,
      "step": 7784
    },
    {
      "epoch": 0.4239243633987462,
      "grad_norm": 0.6717220372865804,
      "learning_rate": 0.00018313164819294678,
      "loss": 12.4721,
      "step": 7785
    },
    {
      "epoch": 0.4239788173953292,
      "grad_norm": 0.613043221582493,
      "learning_rate": 0.0001831267466876585,
      "loss": 12.3669,
      "step": 7786
    },
    {
      "epoch": 0.4240332713919122,
      "grad_norm": 1.0101013612937542,
      "learning_rate": 0.00018312184453596269,
      "loss": 12.5856,
      "step": 7787
    },
    {
      "epoch": 0.4240877253884952,
      "grad_norm": 0.6966337078675802,
      "learning_rate": 0.00018311694173789748,
      "loss": 12.5597,
      "step": 7788
    },
    {
      "epoch": 0.42414217938507826,
      "grad_norm": 0.6240157771149879,
      "learning_rate": 0.000183112038293501,
      "loss": 12.4881,
      "step": 7789
    },
    {
      "epoch": 0.42419663338166125,
      "grad_norm": 0.6642561139413364,
      "learning_rate": 0.00018310713420281133,
      "loss": 12.5333,
      "step": 7790
    },
    {
      "epoch": 0.42425108737824424,
      "grad_norm": 0.6291943374062113,
      "learning_rate": 0.00018310222946586667,
      "loss": 12.3764,
      "step": 7791
    },
    {
      "epoch": 0.4243055413748273,
      "grad_norm": 0.6930968219291624,
      "learning_rate": 0.00018309732408270516,
      "loss": 12.4469,
      "step": 7792
    },
    {
      "epoch": 0.4243599953714103,
      "grad_norm": 0.5953476898224157,
      "learning_rate": 0.00018309241805336492,
      "loss": 12.4047,
      "step": 7793
    },
    {
      "epoch": 0.42441444936799333,
      "grad_norm": 0.5844333523660687,
      "learning_rate": 0.0001830875113778841,
      "loss": 12.5129,
      "step": 7794
    },
    {
      "epoch": 0.4244689033645763,
      "grad_norm": 0.7102053139363087,
      "learning_rate": 0.00018308260405630085,
      "loss": 12.4892,
      "step": 7795
    },
    {
      "epoch": 0.4245233573611593,
      "grad_norm": 0.6652312253117758,
      "learning_rate": 0.00018307769608865332,
      "loss": 12.5766,
      "step": 7796
    },
    {
      "epoch": 0.42457781135774236,
      "grad_norm": 0.7115652187451849,
      "learning_rate": 0.00018307278747497973,
      "loss": 12.4086,
      "step": 7797
    },
    {
      "epoch": 0.42463226535432536,
      "grad_norm": 0.735827519599051,
      "learning_rate": 0.00018306787821531818,
      "loss": 12.6014,
      "step": 7798
    },
    {
      "epoch": 0.42468671935090835,
      "grad_norm": 0.6433664841516628,
      "learning_rate": 0.00018306296830970692,
      "loss": 12.3885,
      "step": 7799
    },
    {
      "epoch": 0.4247411733474914,
      "grad_norm": 0.6139720272683106,
      "learning_rate": 0.00018305805775818405,
      "loss": 12.4495,
      "step": 7800
    },
    {
      "epoch": 0.4247956273440744,
      "grad_norm": 0.6668703605246699,
      "learning_rate": 0.0001830531465607878,
      "loss": 12.5359,
      "step": 7801
    },
    {
      "epoch": 0.4248500813406574,
      "grad_norm": 0.630135798486746,
      "learning_rate": 0.00018304823471755637,
      "loss": 12.3706,
      "step": 7802
    },
    {
      "epoch": 0.42490453533724043,
      "grad_norm": 0.6994353450861902,
      "learning_rate": 0.00018304332222852793,
      "loss": 12.5729,
      "step": 7803
    },
    {
      "epoch": 0.4249589893338234,
      "grad_norm": 1.0254774883538031,
      "learning_rate": 0.0001830384090937407,
      "loss": 12.4567,
      "step": 7804
    },
    {
      "epoch": 0.4250134433304064,
      "grad_norm": 0.7018597963711026,
      "learning_rate": 0.00018303349531323287,
      "loss": 12.4913,
      "step": 7805
    },
    {
      "epoch": 0.42506789732698946,
      "grad_norm": 0.6812768819901065,
      "learning_rate": 0.00018302858088704263,
      "loss": 12.4232,
      "step": 7806
    },
    {
      "epoch": 0.42512235132357246,
      "grad_norm": 0.6189707932825785,
      "learning_rate": 0.00018302366581520824,
      "loss": 12.4999,
      "step": 7807
    },
    {
      "epoch": 0.42517680532015545,
      "grad_norm": 0.7008221094901383,
      "learning_rate": 0.00018301875009776793,
      "loss": 12.3748,
      "step": 7808
    },
    {
      "epoch": 0.4252312593167385,
      "grad_norm": 0.685224508822657,
      "learning_rate": 0.00018301383373475988,
      "loss": 12.483,
      "step": 7809
    },
    {
      "epoch": 0.4252857133133215,
      "grad_norm": 0.6439226473701067,
      "learning_rate": 0.00018300891672622232,
      "loss": 12.427,
      "step": 7810
    },
    {
      "epoch": 0.4253401673099045,
      "grad_norm": 0.7611108891203666,
      "learning_rate": 0.0001830039990721935,
      "loss": 12.5046,
      "step": 7811
    },
    {
      "epoch": 0.42539462130648753,
      "grad_norm": 0.67708873861166,
      "learning_rate": 0.0001829990807727117,
      "loss": 12.5311,
      "step": 7812
    },
    {
      "epoch": 0.4254490753030705,
      "grad_norm": 0.5961418714371193,
      "learning_rate": 0.00018299416182781508,
      "loss": 12.3877,
      "step": 7813
    },
    {
      "epoch": 0.4255035292996535,
      "grad_norm": 0.6176383889070957,
      "learning_rate": 0.00018298924223754198,
      "loss": 12.3607,
      "step": 7814
    },
    {
      "epoch": 0.42555798329623656,
      "grad_norm": 0.629423468089237,
      "learning_rate": 0.0001829843220019306,
      "loss": 12.42,
      "step": 7815
    },
    {
      "epoch": 0.42561243729281956,
      "grad_norm": 0.5963945048028937,
      "learning_rate": 0.0001829794011210192,
      "loss": 12.3582,
      "step": 7816
    },
    {
      "epoch": 0.42566689128940255,
      "grad_norm": 0.6719580459725143,
      "learning_rate": 0.00018297447959484607,
      "loss": 12.4179,
      "step": 7817
    },
    {
      "epoch": 0.4257213452859856,
      "grad_norm": 0.6363496362644461,
      "learning_rate": 0.00018296955742344947,
      "loss": 12.5432,
      "step": 7818
    },
    {
      "epoch": 0.4257757992825686,
      "grad_norm": 0.7035535391398001,
      "learning_rate": 0.0001829646346068677,
      "loss": 12.4458,
      "step": 7819
    },
    {
      "epoch": 0.4258302532791516,
      "grad_norm": 0.7125369757670239,
      "learning_rate": 0.00018295971114513901,
      "loss": 12.4325,
      "step": 7820
    },
    {
      "epoch": 0.42588470727573463,
      "grad_norm": 0.5689821498560794,
      "learning_rate": 0.00018295478703830167,
      "loss": 12.4462,
      "step": 7821
    },
    {
      "epoch": 0.4259391612723176,
      "grad_norm": 0.651504419915653,
      "learning_rate": 0.00018294986228639402,
      "loss": 12.5262,
      "step": 7822
    },
    {
      "epoch": 0.4259936152689006,
      "grad_norm": 0.6846627781301227,
      "learning_rate": 0.00018294493688945432,
      "loss": 12.4642,
      "step": 7823
    },
    {
      "epoch": 0.42604806926548366,
      "grad_norm": 0.5897441762051296,
      "learning_rate": 0.0001829400108475209,
      "loss": 12.3916,
      "step": 7824
    },
    {
      "epoch": 0.42610252326206666,
      "grad_norm": 0.7217302351067372,
      "learning_rate": 0.000182935084160632,
      "loss": 12.4843,
      "step": 7825
    },
    {
      "epoch": 0.42615697725864965,
      "grad_norm": 0.6421094690571366,
      "learning_rate": 0.000182930156828826,
      "loss": 12.416,
      "step": 7826
    },
    {
      "epoch": 0.4262114312552327,
      "grad_norm": 0.5840267657072733,
      "learning_rate": 0.00018292522885214122,
      "loss": 12.448,
      "step": 7827
    },
    {
      "epoch": 0.4262658852518157,
      "grad_norm": 0.5861991036359835,
      "learning_rate": 0.00018292030023061594,
      "loss": 12.4205,
      "step": 7828
    },
    {
      "epoch": 0.42632033924839874,
      "grad_norm": 0.7221993662996262,
      "learning_rate": 0.00018291537096428847,
      "loss": 12.3286,
      "step": 7829
    },
    {
      "epoch": 0.42637479324498173,
      "grad_norm": 0.6868999607560279,
      "learning_rate": 0.00018291044105319721,
      "loss": 12.5275,
      "step": 7830
    },
    {
      "epoch": 0.4264292472415647,
      "grad_norm": 0.6912120530089493,
      "learning_rate": 0.00018290551049738042,
      "loss": 12.4169,
      "step": 7831
    },
    {
      "epoch": 0.42648370123814777,
      "grad_norm": 0.6580756550522271,
      "learning_rate": 0.00018290057929687653,
      "loss": 12.6147,
      "step": 7832
    },
    {
      "epoch": 0.42653815523473076,
      "grad_norm": 0.6170758835490263,
      "learning_rate": 0.00018289564745172377,
      "loss": 12.2711,
      "step": 7833
    },
    {
      "epoch": 0.42659260923131376,
      "grad_norm": 0.7117110220895022,
      "learning_rate": 0.0001828907149619606,
      "loss": 12.4925,
      "step": 7834
    },
    {
      "epoch": 0.4266470632278968,
      "grad_norm": 0.6381204388014844,
      "learning_rate": 0.00018288578182762533,
      "loss": 12.5005,
      "step": 7835
    },
    {
      "epoch": 0.4267015172244798,
      "grad_norm": 0.6543463778852519,
      "learning_rate": 0.00018288084804875626,
      "loss": 12.6284,
      "step": 7836
    },
    {
      "epoch": 0.4267559712210628,
      "grad_norm": 0.6685516297944075,
      "learning_rate": 0.00018287591362539188,
      "loss": 12.5296,
      "step": 7837
    },
    {
      "epoch": 0.42681042521764584,
      "grad_norm": 0.8009985268061635,
      "learning_rate": 0.0001828709785575705,
      "loss": 12.407,
      "step": 7838
    },
    {
      "epoch": 0.42686487921422883,
      "grad_norm": 0.6821136473506321,
      "learning_rate": 0.00018286604284533045,
      "loss": 12.5061,
      "step": 7839
    },
    {
      "epoch": 0.4269193332108118,
      "grad_norm": 0.7352034903137306,
      "learning_rate": 0.0001828611064887102,
      "loss": 12.7067,
      "step": 7840
    },
    {
      "epoch": 0.4269737872073949,
      "grad_norm": 0.6683441393984274,
      "learning_rate": 0.00018285616948774807,
      "loss": 12.5754,
      "step": 7841
    },
    {
      "epoch": 0.42702824120397787,
      "grad_norm": 0.6816842155196888,
      "learning_rate": 0.00018285123184248243,
      "loss": 12.338,
      "step": 7842
    },
    {
      "epoch": 0.42708269520056086,
      "grad_norm": 0.6624536746483938,
      "learning_rate": 0.00018284629355295174,
      "loss": 12.5175,
      "step": 7843
    },
    {
      "epoch": 0.4271371491971439,
      "grad_norm": 0.7013469603718336,
      "learning_rate": 0.0001828413546191944,
      "loss": 12.5453,
      "step": 7844
    },
    {
      "epoch": 0.4271916031937269,
      "grad_norm": 0.6046632729996528,
      "learning_rate": 0.0001828364150412488,
      "loss": 12.2746,
      "step": 7845
    },
    {
      "epoch": 0.4272460571903099,
      "grad_norm": 0.625481111917555,
      "learning_rate": 0.00018283147481915334,
      "loss": 12.4797,
      "step": 7846
    },
    {
      "epoch": 0.42730051118689294,
      "grad_norm": 0.6424818253362229,
      "learning_rate": 0.00018282653395294642,
      "loss": 12.5027,
      "step": 7847
    },
    {
      "epoch": 0.42735496518347593,
      "grad_norm": 0.6601364261491262,
      "learning_rate": 0.0001828215924426665,
      "loss": 12.5228,
      "step": 7848
    },
    {
      "epoch": 0.4274094191800589,
      "grad_norm": 0.6606556794480325,
      "learning_rate": 0.00018281665028835198,
      "loss": 12.351,
      "step": 7849
    },
    {
      "epoch": 0.427463873176642,
      "grad_norm": 0.7391279774495332,
      "learning_rate": 0.0001828117074900413,
      "loss": 12.6615,
      "step": 7850
    },
    {
      "epoch": 0.42751832717322497,
      "grad_norm": 0.666721835432075,
      "learning_rate": 0.0001828067640477729,
      "loss": 12.4141,
      "step": 7851
    },
    {
      "epoch": 0.42757278116980796,
      "grad_norm": 0.6556499417303358,
      "learning_rate": 0.00018280181996158522,
      "loss": 12.5608,
      "step": 7852
    },
    {
      "epoch": 0.427627235166391,
      "grad_norm": 0.7136819769413728,
      "learning_rate": 0.0001827968752315167,
      "loss": 12.6186,
      "step": 7853
    },
    {
      "epoch": 0.427681689162974,
      "grad_norm": 0.6494044088806241,
      "learning_rate": 0.00018279192985760578,
      "loss": 12.4636,
      "step": 7854
    },
    {
      "epoch": 0.427736143159557,
      "grad_norm": 0.6681216329881955,
      "learning_rate": 0.00018278698383989092,
      "loss": 12.5388,
      "step": 7855
    },
    {
      "epoch": 0.42779059715614004,
      "grad_norm": 0.6351791202571165,
      "learning_rate": 0.00018278203717841062,
      "loss": 12.3689,
      "step": 7856
    },
    {
      "epoch": 0.42784505115272303,
      "grad_norm": 0.7121964039171422,
      "learning_rate": 0.0001827770898732033,
      "loss": 12.475,
      "step": 7857
    },
    {
      "epoch": 0.427899505149306,
      "grad_norm": 0.7322361257100048,
      "learning_rate": 0.00018277214192430745,
      "loss": 12.3005,
      "step": 7858
    },
    {
      "epoch": 0.4279539591458891,
      "grad_norm": 0.7110469435756508,
      "learning_rate": 0.00018276719333176154,
      "loss": 12.5725,
      "step": 7859
    },
    {
      "epoch": 0.42800841314247207,
      "grad_norm": 0.7114305715474563,
      "learning_rate": 0.00018276224409560405,
      "loss": 12.3835,
      "step": 7860
    },
    {
      "epoch": 0.4280628671390551,
      "grad_norm": 0.6554373743972242,
      "learning_rate": 0.00018275729421587348,
      "loss": 12.4473,
      "step": 7861
    },
    {
      "epoch": 0.4281173211356381,
      "grad_norm": 0.614966738063642,
      "learning_rate": 0.0001827523436926083,
      "loss": 12.5055,
      "step": 7862
    },
    {
      "epoch": 0.4281717751322211,
      "grad_norm": 0.7763242654060049,
      "learning_rate": 0.00018274739252584706,
      "loss": 12.589,
      "step": 7863
    },
    {
      "epoch": 0.42822622912880415,
      "grad_norm": 0.595923750032374,
      "learning_rate": 0.00018274244071562817,
      "loss": 12.4712,
      "step": 7864
    },
    {
      "epoch": 0.42828068312538714,
      "grad_norm": 0.66789073012322,
      "learning_rate": 0.0001827374882619902,
      "loss": 12.5092,
      "step": 7865
    },
    {
      "epoch": 0.42833513712197013,
      "grad_norm": 0.6993718554608768,
      "learning_rate": 0.00018273253516497168,
      "loss": 12.2774,
      "step": 7866
    },
    {
      "epoch": 0.4283895911185532,
      "grad_norm": 0.6853933435979174,
      "learning_rate": 0.00018272758142461103,
      "loss": 12.4401,
      "step": 7867
    },
    {
      "epoch": 0.4284440451151362,
      "grad_norm": 0.6046127331466988,
      "learning_rate": 0.00018272262704094688,
      "loss": 12.4843,
      "step": 7868
    },
    {
      "epoch": 0.42849849911171917,
      "grad_norm": 0.7031626455104183,
      "learning_rate": 0.0001827176720140177,
      "loss": 12.4454,
      "step": 7869
    },
    {
      "epoch": 0.4285529531083022,
      "grad_norm": 0.6102750691205855,
      "learning_rate": 0.000182712716343862,
      "loss": 12.2383,
      "step": 7870
    },
    {
      "epoch": 0.4286074071048852,
      "grad_norm": 0.6430406643586948,
      "learning_rate": 0.0001827077600305184,
      "loss": 12.3706,
      "step": 7871
    },
    {
      "epoch": 0.4286618611014682,
      "grad_norm": 0.7153973691384076,
      "learning_rate": 0.00018270280307402533,
      "loss": 12.4636,
      "step": 7872
    },
    {
      "epoch": 0.42871631509805125,
      "grad_norm": 0.6111169539665425,
      "learning_rate": 0.00018269784547442143,
      "loss": 12.5269,
      "step": 7873
    },
    {
      "epoch": 0.42877076909463424,
      "grad_norm": 0.6605820807457244,
      "learning_rate": 0.00018269288723174522,
      "loss": 12.5237,
      "step": 7874
    },
    {
      "epoch": 0.42882522309121723,
      "grad_norm": 0.7294481892645149,
      "learning_rate": 0.0001826879283460352,
      "loss": 12.3791,
      "step": 7875
    },
    {
      "epoch": 0.4288796770878003,
      "grad_norm": 0.7080461100945729,
      "learning_rate": 0.00018268296881733005,
      "loss": 12.378,
      "step": 7876
    },
    {
      "epoch": 0.4289341310843833,
      "grad_norm": 0.6910783653995367,
      "learning_rate": 0.00018267800864566825,
      "loss": 12.4422,
      "step": 7877
    },
    {
      "epoch": 0.42898858508096627,
      "grad_norm": 0.6612434872957814,
      "learning_rate": 0.00018267304783108838,
      "loss": 12.3455,
      "step": 7878
    },
    {
      "epoch": 0.4290430390775493,
      "grad_norm": 0.6711938850299602,
      "learning_rate": 0.000182668086373629,
      "loss": 12.3548,
      "step": 7879
    },
    {
      "epoch": 0.4290974930741323,
      "grad_norm": 0.626907677513012,
      "learning_rate": 0.00018266312427332878,
      "loss": 12.5163,
      "step": 7880
    },
    {
      "epoch": 0.4291519470707153,
      "grad_norm": 0.6803065095321489,
      "learning_rate": 0.0001826581615302262,
      "loss": 12.3781,
      "step": 7881
    },
    {
      "epoch": 0.42920640106729835,
      "grad_norm": 0.7196134342169394,
      "learning_rate": 0.00018265319814435988,
      "loss": 12.4378,
      "step": 7882
    },
    {
      "epoch": 0.42926085506388134,
      "grad_norm": 0.6925727268427583,
      "learning_rate": 0.00018264823411576846,
      "loss": 12.599,
      "step": 7883
    },
    {
      "epoch": 0.42931530906046433,
      "grad_norm": 0.6023802851561011,
      "learning_rate": 0.00018264326944449048,
      "loss": 12.4611,
      "step": 7884
    },
    {
      "epoch": 0.4293697630570474,
      "grad_norm": 0.7294334658982401,
      "learning_rate": 0.0001826383041305646,
      "loss": 12.4378,
      "step": 7885
    },
    {
      "epoch": 0.4294242170536304,
      "grad_norm": 0.6625800290992043,
      "learning_rate": 0.0001826333381740294,
      "loss": 12.4643,
      "step": 7886
    },
    {
      "epoch": 0.42947867105021337,
      "grad_norm": 0.7471763593863193,
      "learning_rate": 0.00018262837157492353,
      "loss": 12.4879,
      "step": 7887
    },
    {
      "epoch": 0.4295331250467964,
      "grad_norm": 0.6568300177128303,
      "learning_rate": 0.00018262340433328558,
      "loss": 12.3321,
      "step": 7888
    },
    {
      "epoch": 0.4295875790433794,
      "grad_norm": 0.7096946912181274,
      "learning_rate": 0.00018261843644915417,
      "loss": 12.5937,
      "step": 7889
    },
    {
      "epoch": 0.4296420330399624,
      "grad_norm": 0.6339033309753559,
      "learning_rate": 0.00018261346792256794,
      "loss": 12.5053,
      "step": 7890
    },
    {
      "epoch": 0.42969648703654545,
      "grad_norm": 0.6136133084916588,
      "learning_rate": 0.00018260849875356553,
      "loss": 12.4255,
      "step": 7891
    },
    {
      "epoch": 0.42975094103312844,
      "grad_norm": 0.5945688377929977,
      "learning_rate": 0.0001826035289421856,
      "loss": 12.4445,
      "step": 7892
    },
    {
      "epoch": 0.42980539502971143,
      "grad_norm": 0.7462867216348839,
      "learning_rate": 0.00018259855848846675,
      "loss": 12.4613,
      "step": 7893
    },
    {
      "epoch": 0.4298598490262945,
      "grad_norm": 0.6569392321802703,
      "learning_rate": 0.00018259358739244766,
      "loss": 12.5073,
      "step": 7894
    },
    {
      "epoch": 0.4299143030228775,
      "grad_norm": 0.5983744188237977,
      "learning_rate": 0.00018258861565416702,
      "loss": 12.4993,
      "step": 7895
    },
    {
      "epoch": 0.4299687570194605,
      "grad_norm": 0.7053622283322101,
      "learning_rate": 0.0001825836432736634,
      "loss": 12.3607,
      "step": 7896
    },
    {
      "epoch": 0.4300232110160435,
      "grad_norm": 0.7054846914578173,
      "learning_rate": 0.00018257867025097554,
      "loss": 12.5458,
      "step": 7897
    },
    {
      "epoch": 0.4300776650126265,
      "grad_norm": 0.6949317721624084,
      "learning_rate": 0.00018257369658614212,
      "loss": 12.4957,
      "step": 7898
    },
    {
      "epoch": 0.43013211900920956,
      "grad_norm": 0.6836307012239016,
      "learning_rate": 0.00018256872227920173,
      "loss": 12.479,
      "step": 7899
    },
    {
      "epoch": 0.43018657300579255,
      "grad_norm": 0.7228802923883042,
      "learning_rate": 0.00018256374733019315,
      "loss": 12.3583,
      "step": 7900
    },
    {
      "epoch": 0.43024102700237554,
      "grad_norm": 0.9284099241529163,
      "learning_rate": 0.00018255877173915504,
      "loss": 12.5031,
      "step": 7901
    },
    {
      "epoch": 0.4302954809989586,
      "grad_norm": 0.6694914115117954,
      "learning_rate": 0.00018255379550612605,
      "loss": 12.4756,
      "step": 7902
    },
    {
      "epoch": 0.4303499349955416,
      "grad_norm": 0.6478362302886246,
      "learning_rate": 0.0001825488186311449,
      "loss": 12.558,
      "step": 7903
    },
    {
      "epoch": 0.4304043889921246,
      "grad_norm": 0.6546490002593253,
      "learning_rate": 0.0001825438411142503,
      "loss": 12.4417,
      "step": 7904
    },
    {
      "epoch": 0.4304588429887076,
      "grad_norm": 0.7087779462299327,
      "learning_rate": 0.00018253886295548094,
      "loss": 12.5224,
      "step": 7905
    },
    {
      "epoch": 0.4305132969852906,
      "grad_norm": 0.6993477111814185,
      "learning_rate": 0.00018253388415487556,
      "loss": 12.5954,
      "step": 7906
    },
    {
      "epoch": 0.4305677509818736,
      "grad_norm": 0.6535366877281594,
      "learning_rate": 0.00018252890471247285,
      "loss": 12.3998,
      "step": 7907
    },
    {
      "epoch": 0.43062220497845666,
      "grad_norm": 0.6509191262966944,
      "learning_rate": 0.00018252392462831153,
      "loss": 12.3903,
      "step": 7908
    },
    {
      "epoch": 0.43067665897503965,
      "grad_norm": 0.7174800548505674,
      "learning_rate": 0.00018251894390243031,
      "loss": 12.4249,
      "step": 7909
    },
    {
      "epoch": 0.43073111297162264,
      "grad_norm": 0.7117353466917853,
      "learning_rate": 0.00018251396253486798,
      "loss": 12.5,
      "step": 7910
    },
    {
      "epoch": 0.4307855669682057,
      "grad_norm": 0.5842736337850675,
      "learning_rate": 0.00018250898052566322,
      "loss": 12.3333,
      "step": 7911
    },
    {
      "epoch": 0.4308400209647887,
      "grad_norm": 0.626341575150017,
      "learning_rate": 0.0001825039978748548,
      "loss": 12.4355,
      "step": 7912
    },
    {
      "epoch": 0.4308944749613717,
      "grad_norm": 0.6902559185033911,
      "learning_rate": 0.00018249901458248146,
      "loss": 12.5578,
      "step": 7913
    },
    {
      "epoch": 0.4309489289579547,
      "grad_norm": 0.6996091778217305,
      "learning_rate": 0.00018249403064858193,
      "loss": 12.4151,
      "step": 7914
    },
    {
      "epoch": 0.4310033829545377,
      "grad_norm": 0.6589561733097002,
      "learning_rate": 0.000182489046073195,
      "loss": 12.5248,
      "step": 7915
    },
    {
      "epoch": 0.4310578369511207,
      "grad_norm": 0.6274633751809366,
      "learning_rate": 0.00018248406085635943,
      "loss": 12.3801,
      "step": 7916
    },
    {
      "epoch": 0.43111229094770376,
      "grad_norm": 0.7625790300943449,
      "learning_rate": 0.00018247907499811393,
      "loss": 12.5618,
      "step": 7917
    },
    {
      "epoch": 0.43116674494428675,
      "grad_norm": 0.6880714481481783,
      "learning_rate": 0.00018247408849849734,
      "loss": 12.322,
      "step": 7918
    },
    {
      "epoch": 0.43122119894086974,
      "grad_norm": 0.7149825408684415,
      "learning_rate": 0.0001824691013575484,
      "loss": 12.551,
      "step": 7919
    },
    {
      "epoch": 0.4312756529374528,
      "grad_norm": 0.6958054940177517,
      "learning_rate": 0.0001824641135753059,
      "loss": 12.5937,
      "step": 7920
    },
    {
      "epoch": 0.4313301069340358,
      "grad_norm": 0.6496322803509679,
      "learning_rate": 0.00018245912515180862,
      "loss": 12.4297,
      "step": 7921
    },
    {
      "epoch": 0.4313845609306188,
      "grad_norm": 0.7743435913615766,
      "learning_rate": 0.00018245413608709534,
      "loss": 12.3787,
      "step": 7922
    },
    {
      "epoch": 0.4314390149272018,
      "grad_norm": 0.6499155579167446,
      "learning_rate": 0.0001824491463812049,
      "loss": 12.4618,
      "step": 7923
    },
    {
      "epoch": 0.4314934689237848,
      "grad_norm": 0.6323364011027064,
      "learning_rate": 0.00018244415603417603,
      "loss": 12.5198,
      "step": 7924
    },
    {
      "epoch": 0.4315479229203678,
      "grad_norm": 0.6275726046028748,
      "learning_rate": 0.0001824391650460476,
      "loss": 12.4412,
      "step": 7925
    },
    {
      "epoch": 0.43160237691695086,
      "grad_norm": 0.6563516386645423,
      "learning_rate": 0.0001824341734168584,
      "loss": 12.3843,
      "step": 7926
    },
    {
      "epoch": 0.43165683091353385,
      "grad_norm": 0.5944999411676104,
      "learning_rate": 0.00018242918114664725,
      "loss": 12.4227,
      "step": 7927
    },
    {
      "epoch": 0.4317112849101169,
      "grad_norm": 0.6126075950113996,
      "learning_rate": 0.00018242418823545298,
      "loss": 12.5283,
      "step": 7928
    },
    {
      "epoch": 0.4317657389066999,
      "grad_norm": 0.591985788952654,
      "learning_rate": 0.00018241919468331435,
      "loss": 12.4677,
      "step": 7929
    },
    {
      "epoch": 0.4318201929032829,
      "grad_norm": 0.6383852269541036,
      "learning_rate": 0.00018241420049027027,
      "loss": 12.474,
      "step": 7930
    },
    {
      "epoch": 0.43187464689986593,
      "grad_norm": 0.5852504651876139,
      "learning_rate": 0.00018240920565635956,
      "loss": 12.3616,
      "step": 7931
    },
    {
      "epoch": 0.4319291008964489,
      "grad_norm": 0.6130822749181293,
      "learning_rate": 0.00018240421018162102,
      "loss": 12.4868,
      "step": 7932
    },
    {
      "epoch": 0.4319835548930319,
      "grad_norm": 0.6200945418246628,
      "learning_rate": 0.00018239921406609352,
      "loss": 12.3153,
      "step": 7933
    },
    {
      "epoch": 0.43203800888961497,
      "grad_norm": 0.5977586740609638,
      "learning_rate": 0.00018239421730981595,
      "loss": 12.3454,
      "step": 7934
    },
    {
      "epoch": 0.43209246288619796,
      "grad_norm": 0.7620159440437332,
      "learning_rate": 0.00018238921991282708,
      "loss": 12.5599,
      "step": 7935
    },
    {
      "epoch": 0.43214691688278095,
      "grad_norm": 0.6388958721890771,
      "learning_rate": 0.00018238422187516587,
      "loss": 12.4949,
      "step": 7936
    },
    {
      "epoch": 0.432201370879364,
      "grad_norm": 0.6201417753511387,
      "learning_rate": 0.0001823792231968711,
      "loss": 12.4605,
      "step": 7937
    },
    {
      "epoch": 0.432255824875947,
      "grad_norm": 0.7082720364920682,
      "learning_rate": 0.00018237422387798168,
      "loss": 12.5924,
      "step": 7938
    },
    {
      "epoch": 0.43231027887253,
      "grad_norm": 0.6764522855907903,
      "learning_rate": 0.00018236922391853648,
      "loss": 12.5394,
      "step": 7939
    },
    {
      "epoch": 0.43236473286911303,
      "grad_norm": 0.6027504108616288,
      "learning_rate": 0.00018236422331857437,
      "loss": 12.3855,
      "step": 7940
    },
    {
      "epoch": 0.432419186865696,
      "grad_norm": 0.7383961195275949,
      "learning_rate": 0.00018235922207813428,
      "loss": 12.6475,
      "step": 7941
    },
    {
      "epoch": 0.432473640862279,
      "grad_norm": 0.6480989207450246,
      "learning_rate": 0.00018235422019725507,
      "loss": 12.4974,
      "step": 7942
    },
    {
      "epoch": 0.43252809485886207,
      "grad_norm": 0.7626322657347785,
      "learning_rate": 0.00018234921767597558,
      "loss": 12.458,
      "step": 7943
    },
    {
      "epoch": 0.43258254885544506,
      "grad_norm": 0.7024604934391259,
      "learning_rate": 0.00018234421451433482,
      "loss": 12.474,
      "step": 7944
    },
    {
      "epoch": 0.43263700285202805,
      "grad_norm": 0.6236580831047059,
      "learning_rate": 0.0001823392107123716,
      "loss": 12.3832,
      "step": 7945
    },
    {
      "epoch": 0.4326914568486111,
      "grad_norm": 0.7460718232888852,
      "learning_rate": 0.0001823342062701249,
      "loss": 12.639,
      "step": 7946
    },
    {
      "epoch": 0.4327459108451941,
      "grad_norm": 0.6447139498024131,
      "learning_rate": 0.00018232920118763356,
      "loss": 12.5126,
      "step": 7947
    },
    {
      "epoch": 0.4328003648417771,
      "grad_norm": 0.6928880261448563,
      "learning_rate": 0.00018232419546493657,
      "loss": 12.6822,
      "step": 7948
    },
    {
      "epoch": 0.43285481883836013,
      "grad_norm": 0.625071578737955,
      "learning_rate": 0.00018231918910207283,
      "loss": 12.3265,
      "step": 7949
    },
    {
      "epoch": 0.4329092728349431,
      "grad_norm": 0.5987110493663569,
      "learning_rate": 0.0001823141820990813,
      "loss": 12.5354,
      "step": 7950
    },
    {
      "epoch": 0.4329637268315261,
      "grad_norm": 0.620178423324587,
      "learning_rate": 0.00018230917445600085,
      "loss": 12.4393,
      "step": 7951
    },
    {
      "epoch": 0.43301818082810917,
      "grad_norm": 0.6307608473905716,
      "learning_rate": 0.00018230416617287046,
      "loss": 12.4671,
      "step": 7952
    },
    {
      "epoch": 0.43307263482469216,
      "grad_norm": 0.7122894827254327,
      "learning_rate": 0.0001822991572497291,
      "loss": 12.4647,
      "step": 7953
    },
    {
      "epoch": 0.43312708882127515,
      "grad_norm": 0.5671356599600159,
      "learning_rate": 0.00018229414768661565,
      "loss": 12.3425,
      "step": 7954
    },
    {
      "epoch": 0.4331815428178582,
      "grad_norm": 0.6333834878199174,
      "learning_rate": 0.00018228913748356913,
      "loss": 12.4106,
      "step": 7955
    },
    {
      "epoch": 0.4332359968144412,
      "grad_norm": 0.6650636281838411,
      "learning_rate": 0.0001822841266406285,
      "loss": 12.4607,
      "step": 7956
    },
    {
      "epoch": 0.4332904508110242,
      "grad_norm": 0.585052172740277,
      "learning_rate": 0.00018227911515783266,
      "loss": 12.4691,
      "step": 7957
    },
    {
      "epoch": 0.43334490480760723,
      "grad_norm": 0.6751958569542258,
      "learning_rate": 0.00018227410303522064,
      "loss": 12.3421,
      "step": 7958
    },
    {
      "epoch": 0.4333993588041902,
      "grad_norm": 0.6306906606932072,
      "learning_rate": 0.0001822690902728314,
      "loss": 12.306,
      "step": 7959
    },
    {
      "epoch": 0.4334538128007732,
      "grad_norm": 0.6095238580428013,
      "learning_rate": 0.0001822640768707039,
      "loss": 12.4476,
      "step": 7960
    },
    {
      "epoch": 0.43350826679735627,
      "grad_norm": 0.6542022762270828,
      "learning_rate": 0.00018225906282887718,
      "loss": 12.4173,
      "step": 7961
    },
    {
      "epoch": 0.43356272079393926,
      "grad_norm": 0.6344690057285265,
      "learning_rate": 0.00018225404814739012,
      "loss": 12.4391,
      "step": 7962
    },
    {
      "epoch": 0.4336171747905223,
      "grad_norm": 0.7428938770750612,
      "learning_rate": 0.00018224903282628186,
      "loss": 12.4032,
      "step": 7963
    },
    {
      "epoch": 0.4336716287871053,
      "grad_norm": 0.6437215141879193,
      "learning_rate": 0.0001822440168655913,
      "loss": 12.4873,
      "step": 7964
    },
    {
      "epoch": 0.4337260827836883,
      "grad_norm": 0.6611723911944084,
      "learning_rate": 0.00018223900026535748,
      "loss": 12.4769,
      "step": 7965
    },
    {
      "epoch": 0.43378053678027134,
      "grad_norm": 0.6673601417317901,
      "learning_rate": 0.00018223398302561942,
      "loss": 12.4524,
      "step": 7966
    },
    {
      "epoch": 0.43383499077685433,
      "grad_norm": 0.7049890622715992,
      "learning_rate": 0.0001822289651464161,
      "loss": 12.4558,
      "step": 7967
    },
    {
      "epoch": 0.4338894447734373,
      "grad_norm": 0.6010186771172481,
      "learning_rate": 0.0001822239466277865,
      "loss": 12.5184,
      "step": 7968
    },
    {
      "epoch": 0.4339438987700204,
      "grad_norm": 0.6782184525687398,
      "learning_rate": 0.0001822189274697698,
      "loss": 12.5598,
      "step": 7969
    },
    {
      "epoch": 0.43399835276660337,
      "grad_norm": 0.7535593508707078,
      "learning_rate": 0.0001822139076724049,
      "loss": 12.4382,
      "step": 7970
    },
    {
      "epoch": 0.43405280676318636,
      "grad_norm": 0.7476354959202245,
      "learning_rate": 0.00018220888723573082,
      "loss": 12.3689,
      "step": 7971
    },
    {
      "epoch": 0.4341072607597694,
      "grad_norm": 0.6468310309468458,
      "learning_rate": 0.0001822038661597867,
      "loss": 12.4683,
      "step": 7972
    },
    {
      "epoch": 0.4341617147563524,
      "grad_norm": 0.6425677167614992,
      "learning_rate": 0.0001821988444446115,
      "loss": 12.5546,
      "step": 7973
    },
    {
      "epoch": 0.4342161687529354,
      "grad_norm": 0.6616730114954977,
      "learning_rate": 0.0001821938220902443,
      "loss": 12.5815,
      "step": 7974
    },
    {
      "epoch": 0.43427062274951844,
      "grad_norm": 0.6706196272119236,
      "learning_rate": 0.0001821887990967242,
      "loss": 12.4704,
      "step": 7975
    },
    {
      "epoch": 0.43432507674610143,
      "grad_norm": 0.7037258987322563,
      "learning_rate": 0.00018218377546409017,
      "loss": 12.5475,
      "step": 7976
    },
    {
      "epoch": 0.4343795307426844,
      "grad_norm": 0.6489047278433996,
      "learning_rate": 0.00018217875119238132,
      "loss": 12.3292,
      "step": 7977
    },
    {
      "epoch": 0.4344339847392675,
      "grad_norm": 0.6616035776812114,
      "learning_rate": 0.00018217372628163674,
      "loss": 12.2886,
      "step": 7978
    },
    {
      "epoch": 0.43448843873585047,
      "grad_norm": 0.7245041333919243,
      "learning_rate": 0.00018216870073189546,
      "loss": 12.3057,
      "step": 7979
    },
    {
      "epoch": 0.43454289273243346,
      "grad_norm": 0.6823812931676292,
      "learning_rate": 0.00018216367454319663,
      "loss": 12.3488,
      "step": 7980
    },
    {
      "epoch": 0.4345973467290165,
      "grad_norm": 0.6928454505678266,
      "learning_rate": 0.00018215864771557925,
      "loss": 12.4474,
      "step": 7981
    },
    {
      "epoch": 0.4346518007255995,
      "grad_norm": 0.6692762111560462,
      "learning_rate": 0.00018215362024908246,
      "loss": 12.3722,
      "step": 7982
    },
    {
      "epoch": 0.4347062547221825,
      "grad_norm": 0.7217401785786631,
      "learning_rate": 0.0001821485921437453,
      "loss": 12.5339,
      "step": 7983
    },
    {
      "epoch": 0.43476070871876554,
      "grad_norm": 0.7266825040044337,
      "learning_rate": 0.00018214356339960698,
      "loss": 12.3977,
      "step": 7984
    },
    {
      "epoch": 0.43481516271534854,
      "grad_norm": 0.6826875691189774,
      "learning_rate": 0.0001821385340167065,
      "loss": 12.4072,
      "step": 7985
    },
    {
      "epoch": 0.43486961671193153,
      "grad_norm": 0.7064241066509669,
      "learning_rate": 0.000182133503995083,
      "loss": 12.4456,
      "step": 7986
    },
    {
      "epoch": 0.4349240707085146,
      "grad_norm": 0.6175230417065565,
      "learning_rate": 0.0001821284733347756,
      "loss": 12.4795,
      "step": 7987
    },
    {
      "epoch": 0.43497852470509757,
      "grad_norm": 0.6079407316866163,
      "learning_rate": 0.00018212344203582344,
      "loss": 12.3473,
      "step": 7988
    },
    {
      "epoch": 0.43503297870168056,
      "grad_norm": 0.7163691665704485,
      "learning_rate": 0.0001821184100982656,
      "loss": 12.3703,
      "step": 7989
    },
    {
      "epoch": 0.4350874326982636,
      "grad_norm": 0.6258311004616853,
      "learning_rate": 0.00018211337752214123,
      "loss": 12.3942,
      "step": 7990
    },
    {
      "epoch": 0.4351418866948466,
      "grad_norm": 0.671124784124222,
      "learning_rate": 0.0001821083443074895,
      "loss": 12.4334,
      "step": 7991
    },
    {
      "epoch": 0.4351963406914296,
      "grad_norm": 0.6216233607643935,
      "learning_rate": 0.00018210331045434949,
      "loss": 12.4195,
      "step": 7992
    },
    {
      "epoch": 0.43525079468801264,
      "grad_norm": 0.742741508242918,
      "learning_rate": 0.00018209827596276035,
      "loss": 12.4239,
      "step": 7993
    },
    {
      "epoch": 0.43530524868459564,
      "grad_norm": 0.6300698638468436,
      "learning_rate": 0.00018209324083276126,
      "loss": 12.3753,
      "step": 7994
    },
    {
      "epoch": 0.4353597026811787,
      "grad_norm": 0.751409905949325,
      "learning_rate": 0.00018208820506439137,
      "loss": 12.529,
      "step": 7995
    },
    {
      "epoch": 0.4354141566777617,
      "grad_norm": 0.6947725654359397,
      "learning_rate": 0.00018208316865768985,
      "loss": 12.5276,
      "step": 7996
    },
    {
      "epoch": 0.43546861067434467,
      "grad_norm": 0.9447376665430275,
      "learning_rate": 0.00018207813161269583,
      "loss": 12.4355,
      "step": 7997
    },
    {
      "epoch": 0.4355230646709277,
      "grad_norm": 0.6723041201443456,
      "learning_rate": 0.00018207309392944846,
      "loss": 12.4632,
      "step": 7998
    },
    {
      "epoch": 0.4355775186675107,
      "grad_norm": 0.7976902368051231,
      "learning_rate": 0.000182068055607987,
      "loss": 12.6122,
      "step": 7999
    },
    {
      "epoch": 0.4356319726640937,
      "grad_norm": 0.7345131489832951,
      "learning_rate": 0.00018206301664835058,
      "loss": 12.5908,
      "step": 8000
    },
    {
      "epoch": 0.43568642666067675,
      "grad_norm": 0.6454712970992138,
      "learning_rate": 0.00018205797705057834,
      "loss": 12.5204,
      "step": 8001
    },
    {
      "epoch": 0.43574088065725974,
      "grad_norm": 0.6711649663357349,
      "learning_rate": 0.00018205293681470953,
      "loss": 12.5056,
      "step": 8002
    },
    {
      "epoch": 0.43579533465384274,
      "grad_norm": 0.6728369086830224,
      "learning_rate": 0.0001820478959407833,
      "loss": 12.3627,
      "step": 8003
    },
    {
      "epoch": 0.4358497886504258,
      "grad_norm": 0.6597979370043341,
      "learning_rate": 0.0001820428544288389,
      "loss": 12.3749,
      "step": 8004
    },
    {
      "epoch": 0.4359042426470088,
      "grad_norm": 0.6209904486774009,
      "learning_rate": 0.0001820378122789155,
      "loss": 12.4336,
      "step": 8005
    },
    {
      "epoch": 0.43595869664359177,
      "grad_norm": 0.7713948991466982,
      "learning_rate": 0.00018203276949105233,
      "loss": 12.4182,
      "step": 8006
    },
    {
      "epoch": 0.4360131506401748,
      "grad_norm": 0.6469371586470584,
      "learning_rate": 0.00018202772606528856,
      "loss": 12.3034,
      "step": 8007
    },
    {
      "epoch": 0.4360676046367578,
      "grad_norm": 0.6995538373363558,
      "learning_rate": 0.00018202268200166342,
      "loss": 12.4704,
      "step": 8008
    },
    {
      "epoch": 0.4361220586333408,
      "grad_norm": 0.666288730336914,
      "learning_rate": 0.00018201763730021618,
      "loss": 12.4474,
      "step": 8009
    },
    {
      "epoch": 0.43617651262992385,
      "grad_norm": 0.6211419327657228,
      "learning_rate": 0.00018201259196098604,
      "loss": 12.4149,
      "step": 8010
    },
    {
      "epoch": 0.43623096662650684,
      "grad_norm": 0.628883970811886,
      "learning_rate": 0.0001820075459840122,
      "loss": 12.4724,
      "step": 8011
    },
    {
      "epoch": 0.43628542062308984,
      "grad_norm": 0.5950168445018583,
      "learning_rate": 0.00018200249936933398,
      "loss": 12.4613,
      "step": 8012
    },
    {
      "epoch": 0.4363398746196729,
      "grad_norm": 0.6312701310341498,
      "learning_rate": 0.00018199745211699053,
      "loss": 12.4223,
      "step": 8013
    },
    {
      "epoch": 0.4363943286162559,
      "grad_norm": 0.6932151999657433,
      "learning_rate": 0.00018199240422702117,
      "loss": 12.4752,
      "step": 8014
    },
    {
      "epoch": 0.43644878261283887,
      "grad_norm": 0.6294977807573285,
      "learning_rate": 0.0001819873556994651,
      "loss": 12.4771,
      "step": 8015
    },
    {
      "epoch": 0.4365032366094219,
      "grad_norm": 0.6264943199859186,
      "learning_rate": 0.0001819823065343616,
      "loss": 12.484,
      "step": 8016
    },
    {
      "epoch": 0.4365576906060049,
      "grad_norm": 0.6616444184130431,
      "learning_rate": 0.00018197725673174998,
      "loss": 12.5049,
      "step": 8017
    },
    {
      "epoch": 0.4366121446025879,
      "grad_norm": 0.5744172061896488,
      "learning_rate": 0.00018197220629166943,
      "loss": 12.4327,
      "step": 8018
    },
    {
      "epoch": 0.43666659859917095,
      "grad_norm": 0.6289380832372987,
      "learning_rate": 0.00018196715521415926,
      "loss": 12.3483,
      "step": 8019
    },
    {
      "epoch": 0.43672105259575394,
      "grad_norm": 0.6403498509864031,
      "learning_rate": 0.00018196210349925873,
      "loss": 12.438,
      "step": 8020
    },
    {
      "epoch": 0.43677550659233694,
      "grad_norm": 0.579175241208905,
      "learning_rate": 0.00018195705114700713,
      "loss": 12.3951,
      "step": 8021
    },
    {
      "epoch": 0.43682996058892,
      "grad_norm": 0.6219905875288234,
      "learning_rate": 0.00018195199815744382,
      "loss": 12.4101,
      "step": 8022
    },
    {
      "epoch": 0.436884414585503,
      "grad_norm": 0.5955832250530326,
      "learning_rate": 0.000181946944530608,
      "loss": 12.3245,
      "step": 8023
    },
    {
      "epoch": 0.43693886858208597,
      "grad_norm": 0.6072932595615831,
      "learning_rate": 0.00018194189026653898,
      "loss": 12.4663,
      "step": 8024
    },
    {
      "epoch": 0.436993322578669,
      "grad_norm": 0.7165004470076959,
      "learning_rate": 0.00018193683536527609,
      "loss": 12.4848,
      "step": 8025
    },
    {
      "epoch": 0.437047776575252,
      "grad_norm": 0.6513696149725915,
      "learning_rate": 0.00018193177982685858,
      "loss": 12.4224,
      "step": 8026
    },
    {
      "epoch": 0.437102230571835,
      "grad_norm": 0.7428711180542187,
      "learning_rate": 0.00018192672365132588,
      "loss": 12.5178,
      "step": 8027
    },
    {
      "epoch": 0.43715668456841805,
      "grad_norm": 0.6427803787757064,
      "learning_rate": 0.0001819216668387172,
      "loss": 12.4145,
      "step": 8028
    },
    {
      "epoch": 0.43721113856500105,
      "grad_norm": 0.6207320706711867,
      "learning_rate": 0.00018191660938907193,
      "loss": 12.4405,
      "step": 8029
    },
    {
      "epoch": 0.4372655925615841,
      "grad_norm": 0.6852852916241587,
      "learning_rate": 0.00018191155130242934,
      "loss": 12.4623,
      "step": 8030
    },
    {
      "epoch": 0.4373200465581671,
      "grad_norm": 0.6840507058001489,
      "learning_rate": 0.0001819064925788288,
      "loss": 12.4987,
      "step": 8031
    },
    {
      "epoch": 0.4373745005547501,
      "grad_norm": 0.6140790978364469,
      "learning_rate": 0.00018190143321830963,
      "loss": 12.373,
      "step": 8032
    },
    {
      "epoch": 0.4374289545513331,
      "grad_norm": 0.6775222151229636,
      "learning_rate": 0.0001818963732209112,
      "loss": 12.5089,
      "step": 8033
    },
    {
      "epoch": 0.4374834085479161,
      "grad_norm": 0.590068771249146,
      "learning_rate": 0.00018189131258667282,
      "loss": 12.4183,
      "step": 8034
    },
    {
      "epoch": 0.4375378625444991,
      "grad_norm": 0.6183153379587679,
      "learning_rate": 0.00018188625131563386,
      "loss": 12.4238,
      "step": 8035
    },
    {
      "epoch": 0.43759231654108216,
      "grad_norm": 0.7422525928327838,
      "learning_rate": 0.00018188118940783368,
      "loss": 12.5106,
      "step": 8036
    },
    {
      "epoch": 0.43764677053766515,
      "grad_norm": 0.6068305483295992,
      "learning_rate": 0.00018187612686331166,
      "loss": 12.4831,
      "step": 8037
    },
    {
      "epoch": 0.43770122453424815,
      "grad_norm": 0.6158291631235349,
      "learning_rate": 0.00018187106368210712,
      "loss": 12.4306,
      "step": 8038
    },
    {
      "epoch": 0.4377556785308312,
      "grad_norm": 0.6327278698588021,
      "learning_rate": 0.00018186599986425947,
      "loss": 12.4772,
      "step": 8039
    },
    {
      "epoch": 0.4378101325274142,
      "grad_norm": 0.691860427152302,
      "learning_rate": 0.00018186093540980807,
      "loss": 12.5121,
      "step": 8040
    },
    {
      "epoch": 0.4378645865239972,
      "grad_norm": 0.6848902755048658,
      "learning_rate": 0.00018185587031879232,
      "loss": 12.3507,
      "step": 8041
    },
    {
      "epoch": 0.4379190405205802,
      "grad_norm": 0.6024814082483306,
      "learning_rate": 0.0001818508045912516,
      "loss": 12.4528,
      "step": 8042
    },
    {
      "epoch": 0.4379734945171632,
      "grad_norm": 0.6652691577868063,
      "learning_rate": 0.00018184573822722529,
      "loss": 12.3157,
      "step": 8043
    },
    {
      "epoch": 0.4380279485137462,
      "grad_norm": 0.6569694406846195,
      "learning_rate": 0.00018184067122675276,
      "loss": 12.459,
      "step": 8044
    },
    {
      "epoch": 0.43808240251032926,
      "grad_norm": 0.6420980248319899,
      "learning_rate": 0.00018183560358987349,
      "loss": 12.4096,
      "step": 8045
    },
    {
      "epoch": 0.43813685650691225,
      "grad_norm": 0.6627008318367901,
      "learning_rate": 0.00018183053531662684,
      "loss": 12.3847,
      "step": 8046
    },
    {
      "epoch": 0.43819131050349525,
      "grad_norm": 0.7059088848618572,
      "learning_rate": 0.0001818254664070522,
      "loss": 12.5215,
      "step": 8047
    },
    {
      "epoch": 0.4382457645000783,
      "grad_norm": 0.6854425080494145,
      "learning_rate": 0.00018182039686118903,
      "loss": 12.5269,
      "step": 8048
    },
    {
      "epoch": 0.4383002184966613,
      "grad_norm": 0.6535417405105993,
      "learning_rate": 0.00018181532667907671,
      "loss": 12.5775,
      "step": 8049
    },
    {
      "epoch": 0.4383546724932443,
      "grad_norm": 0.6389520951864824,
      "learning_rate": 0.0001818102558607547,
      "loss": 12.2756,
      "step": 8050
    },
    {
      "epoch": 0.4384091264898273,
      "grad_norm": 0.6091592745791563,
      "learning_rate": 0.00018180518440626245,
      "loss": 12.5519,
      "step": 8051
    },
    {
      "epoch": 0.4384635804864103,
      "grad_norm": 0.592034414821979,
      "learning_rate": 0.00018180011231563935,
      "loss": 12.3247,
      "step": 8052
    },
    {
      "epoch": 0.4385180344829933,
      "grad_norm": 0.6822262235027153,
      "learning_rate": 0.00018179503958892483,
      "loss": 12.4094,
      "step": 8053
    },
    {
      "epoch": 0.43857248847957636,
      "grad_norm": 0.6903442897493688,
      "learning_rate": 0.0001817899662261584,
      "loss": 12.4012,
      "step": 8054
    },
    {
      "epoch": 0.43862694247615935,
      "grad_norm": 0.5746722202516459,
      "learning_rate": 0.00018178489222737946,
      "loss": 12.3903,
      "step": 8055
    },
    {
      "epoch": 0.43868139647274235,
      "grad_norm": 0.7103112826336199,
      "learning_rate": 0.00018177981759262747,
      "loss": 12.4608,
      "step": 8056
    },
    {
      "epoch": 0.4387358504693254,
      "grad_norm": 0.7133739481466947,
      "learning_rate": 0.00018177474232194195,
      "loss": 12.3867,
      "step": 8057
    },
    {
      "epoch": 0.4387903044659084,
      "grad_norm": 0.6869948568749044,
      "learning_rate": 0.0001817696664153623,
      "loss": 12.4986,
      "step": 8058
    },
    {
      "epoch": 0.4388447584624914,
      "grad_norm": 0.67170120481675,
      "learning_rate": 0.00018176458987292799,
      "loss": 12.4175,
      "step": 8059
    },
    {
      "epoch": 0.43889921245907443,
      "grad_norm": 0.7517173895741974,
      "learning_rate": 0.00018175951269467855,
      "loss": 12.4281,
      "step": 8060
    },
    {
      "epoch": 0.4389536664556574,
      "grad_norm": 0.7072112142379776,
      "learning_rate": 0.0001817544348806534,
      "loss": 12.4413,
      "step": 8061
    },
    {
      "epoch": 0.43900812045224047,
      "grad_norm": 0.6774523087574283,
      "learning_rate": 0.00018174935643089203,
      "loss": 12.4525,
      "step": 8062
    },
    {
      "epoch": 0.43906257444882346,
      "grad_norm": 0.6340529043477183,
      "learning_rate": 0.000181744277345434,
      "loss": 12.3797,
      "step": 8063
    },
    {
      "epoch": 0.43911702844540645,
      "grad_norm": 0.718405082762463,
      "learning_rate": 0.00018173919762431875,
      "loss": 12.5314,
      "step": 8064
    },
    {
      "epoch": 0.4391714824419895,
      "grad_norm": 0.6952690412740806,
      "learning_rate": 0.0001817341172675858,
      "loss": 12.492,
      "step": 8065
    },
    {
      "epoch": 0.4392259364385725,
      "grad_norm": 0.584979726754951,
      "learning_rate": 0.00018172903627527463,
      "loss": 12.378,
      "step": 8066
    },
    {
      "epoch": 0.4392803904351555,
      "grad_norm": 0.7046022953559745,
      "learning_rate": 0.00018172395464742479,
      "loss": 12.5153,
      "step": 8067
    },
    {
      "epoch": 0.43933484443173854,
      "grad_norm": 0.6895392327708775,
      "learning_rate": 0.00018171887238407574,
      "loss": 12.4722,
      "step": 8068
    },
    {
      "epoch": 0.43938929842832153,
      "grad_norm": 0.593507928029048,
      "learning_rate": 0.00018171378948526705,
      "loss": 12.5153,
      "step": 8069
    },
    {
      "epoch": 0.4394437524249045,
      "grad_norm": 0.6536079851636544,
      "learning_rate": 0.00018170870595103823,
      "loss": 12.5043,
      "step": 8070
    },
    {
      "epoch": 0.43949820642148757,
      "grad_norm": 0.7102749722944064,
      "learning_rate": 0.0001817036217814288,
      "loss": 12.5844,
      "step": 8071
    },
    {
      "epoch": 0.43955266041807056,
      "grad_norm": 0.6250358275118798,
      "learning_rate": 0.0001816985369764783,
      "loss": 12.4398,
      "step": 8072
    },
    {
      "epoch": 0.43960711441465355,
      "grad_norm": 0.6833189631616499,
      "learning_rate": 0.00018169345153622632,
      "loss": 12.5777,
      "step": 8073
    },
    {
      "epoch": 0.4396615684112366,
      "grad_norm": 0.6326543135044489,
      "learning_rate": 0.00018168836546071234,
      "loss": 12.5301,
      "step": 8074
    },
    {
      "epoch": 0.4397160224078196,
      "grad_norm": 0.6599203053997212,
      "learning_rate": 0.00018168327874997593,
      "loss": 12.4672,
      "step": 8075
    },
    {
      "epoch": 0.4397704764044026,
      "grad_norm": 0.67676831770541,
      "learning_rate": 0.00018167819140405662,
      "loss": 12.5454,
      "step": 8076
    },
    {
      "epoch": 0.43982493040098564,
      "grad_norm": 0.6171632602120455,
      "learning_rate": 0.00018167310342299403,
      "loss": 12.3727,
      "step": 8077
    },
    {
      "epoch": 0.43987938439756863,
      "grad_norm": 0.8192129927985445,
      "learning_rate": 0.00018166801480682768,
      "loss": 12.5696,
      "step": 8078
    },
    {
      "epoch": 0.4399338383941516,
      "grad_norm": 0.6841813706928609,
      "learning_rate": 0.00018166292555559714,
      "loss": 12.4212,
      "step": 8079
    },
    {
      "epoch": 0.43998829239073467,
      "grad_norm": 0.6756316788448604,
      "learning_rate": 0.000181657835669342,
      "loss": 12.4523,
      "step": 8080
    },
    {
      "epoch": 0.44004274638731766,
      "grad_norm": 0.7117498961542793,
      "learning_rate": 0.00018165274514810186,
      "loss": 12.4312,
      "step": 8081
    },
    {
      "epoch": 0.44009720038390066,
      "grad_norm": 0.6798671037484373,
      "learning_rate": 0.00018164765399191626,
      "loss": 12.5288,
      "step": 8082
    },
    {
      "epoch": 0.4401516543804837,
      "grad_norm": 0.6398340227930036,
      "learning_rate": 0.0001816425622008248,
      "loss": 12.3392,
      "step": 8083
    },
    {
      "epoch": 0.4402061083770667,
      "grad_norm": 0.7515836639185821,
      "learning_rate": 0.0001816374697748671,
      "loss": 12.5069,
      "step": 8084
    },
    {
      "epoch": 0.4402605623736497,
      "grad_norm": 0.6114018054945423,
      "learning_rate": 0.00018163237671408276,
      "loss": 12.3876,
      "step": 8085
    },
    {
      "epoch": 0.44031501637023274,
      "grad_norm": 0.5820204184592364,
      "learning_rate": 0.00018162728301851134,
      "loss": 12.4475,
      "step": 8086
    },
    {
      "epoch": 0.44036947036681573,
      "grad_norm": 0.7451614160847416,
      "learning_rate": 0.0001816221886881925,
      "loss": 12.4133,
      "step": 8087
    },
    {
      "epoch": 0.4404239243633987,
      "grad_norm": 0.6376617665113111,
      "learning_rate": 0.0001816170937231658,
      "loss": 12.4494,
      "step": 8088
    },
    {
      "epoch": 0.44047837835998177,
      "grad_norm": 0.6485801097204811,
      "learning_rate": 0.00018161199812347093,
      "loss": 12.5021,
      "step": 8089
    },
    {
      "epoch": 0.44053283235656476,
      "grad_norm": 0.6536790997705955,
      "learning_rate": 0.00018160690188914744,
      "loss": 12.486,
      "step": 8090
    },
    {
      "epoch": 0.44058728635314776,
      "grad_norm": 0.6973380046290791,
      "learning_rate": 0.000181601805020235,
      "loss": 12.4213,
      "step": 8091
    },
    {
      "epoch": 0.4406417403497308,
      "grad_norm": 0.6153978517164419,
      "learning_rate": 0.00018159670751677326,
      "loss": 12.4879,
      "step": 8092
    },
    {
      "epoch": 0.4406961943463138,
      "grad_norm": 0.7662004917264686,
      "learning_rate": 0.00018159160937880183,
      "loss": 12.5327,
      "step": 8093
    },
    {
      "epoch": 0.4407506483428968,
      "grad_norm": 0.6315076862191303,
      "learning_rate": 0.00018158651060636038,
      "loss": 12.5029,
      "step": 8094
    },
    {
      "epoch": 0.44080510233947984,
      "grad_norm": 0.6570484528322555,
      "learning_rate": 0.00018158141119948854,
      "loss": 12.5021,
      "step": 8095
    },
    {
      "epoch": 0.44085955633606283,
      "grad_norm": 0.6217452499352483,
      "learning_rate": 0.00018157631115822595,
      "loss": 12.4358,
      "step": 8096
    },
    {
      "epoch": 0.4409140103326459,
      "grad_norm": 0.6097425662470402,
      "learning_rate": 0.0001815712104826123,
      "loss": 12.4138,
      "step": 8097
    },
    {
      "epoch": 0.44096846432922887,
      "grad_norm": 0.5895926618645863,
      "learning_rate": 0.0001815661091726872,
      "loss": 12.3695,
      "step": 8098
    },
    {
      "epoch": 0.44102291832581186,
      "grad_norm": 0.6377967778191228,
      "learning_rate": 0.00018156100722849038,
      "loss": 12.5448,
      "step": 8099
    },
    {
      "epoch": 0.4410773723223949,
      "grad_norm": 0.6401808345387747,
      "learning_rate": 0.0001815559046500615,
      "loss": 12.4537,
      "step": 8100
    },
    {
      "epoch": 0.4411318263189779,
      "grad_norm": 0.6287248883361566,
      "learning_rate": 0.0001815508014374402,
      "loss": 12.5005,
      "step": 8101
    },
    {
      "epoch": 0.4411862803155609,
      "grad_norm": 0.6944250053649565,
      "learning_rate": 0.00018154569759066625,
      "loss": 12.5913,
      "step": 8102
    },
    {
      "epoch": 0.44124073431214395,
      "grad_norm": 0.627346440387178,
      "learning_rate": 0.00018154059310977923,
      "loss": 12.4249,
      "step": 8103
    },
    {
      "epoch": 0.44129518830872694,
      "grad_norm": 0.6420492552682078,
      "learning_rate": 0.00018153548799481887,
      "loss": 12.3946,
      "step": 8104
    },
    {
      "epoch": 0.44134964230530993,
      "grad_norm": 0.681906372443578,
      "learning_rate": 0.00018153038224582492,
      "loss": 12.3387,
      "step": 8105
    },
    {
      "epoch": 0.441404096301893,
      "grad_norm": 0.6210574333231363,
      "learning_rate": 0.00018152527586283703,
      "loss": 12.5162,
      "step": 8106
    },
    {
      "epoch": 0.44145855029847597,
      "grad_norm": 0.64322418419475,
      "learning_rate": 0.0001815201688458949,
      "loss": 12.5622,
      "step": 8107
    },
    {
      "epoch": 0.44151300429505896,
      "grad_norm": 0.8190346317269288,
      "learning_rate": 0.0001815150611950383,
      "loss": 12.4488,
      "step": 8108
    },
    {
      "epoch": 0.441567458291642,
      "grad_norm": 0.6529614749866838,
      "learning_rate": 0.00018150995291030692,
      "loss": 12.4613,
      "step": 8109
    },
    {
      "epoch": 0.441621912288225,
      "grad_norm": 0.7258579649878754,
      "learning_rate": 0.00018150484399174046,
      "loss": 12.4522,
      "step": 8110
    },
    {
      "epoch": 0.441676366284808,
      "grad_norm": 0.6654211401665494,
      "learning_rate": 0.00018149973443937868,
      "loss": 12.3424,
      "step": 8111
    },
    {
      "epoch": 0.44173082028139105,
      "grad_norm": 0.6318212930391481,
      "learning_rate": 0.00018149462425326126,
      "loss": 12.4647,
      "step": 8112
    },
    {
      "epoch": 0.44178527427797404,
      "grad_norm": 0.8207170885860678,
      "learning_rate": 0.00018148951343342803,
      "loss": 12.4289,
      "step": 8113
    },
    {
      "epoch": 0.44183972827455703,
      "grad_norm": 0.6194663346605774,
      "learning_rate": 0.00018148440197991863,
      "loss": 12.3711,
      "step": 8114
    },
    {
      "epoch": 0.4418941822711401,
      "grad_norm": 0.7014168629119157,
      "learning_rate": 0.0001814792898927729,
      "loss": 12.3242,
      "step": 8115
    },
    {
      "epoch": 0.44194863626772307,
      "grad_norm": 0.656875765463606,
      "learning_rate": 0.00018147417717203048,
      "loss": 12.4609,
      "step": 8116
    },
    {
      "epoch": 0.44200309026430606,
      "grad_norm": 0.6422545749068618,
      "learning_rate": 0.00018146906381773126,
      "loss": 12.4707,
      "step": 8117
    },
    {
      "epoch": 0.4420575442608891,
      "grad_norm": 0.6240047670798762,
      "learning_rate": 0.0001814639498299149,
      "loss": 12.4928,
      "step": 8118
    },
    {
      "epoch": 0.4421119982574721,
      "grad_norm": 0.6781025678231234,
      "learning_rate": 0.00018145883520862121,
      "loss": 12.6163,
      "step": 8119
    },
    {
      "epoch": 0.4421664522540551,
      "grad_norm": 0.636615698424067,
      "learning_rate": 0.00018145371995388997,
      "loss": 12.4772,
      "step": 8120
    },
    {
      "epoch": 0.44222090625063815,
      "grad_norm": 0.6454832736621415,
      "learning_rate": 0.00018144860406576094,
      "loss": 12.3847,
      "step": 8121
    },
    {
      "epoch": 0.44227536024722114,
      "grad_norm": 0.7496269835050098,
      "learning_rate": 0.0001814434875442739,
      "loss": 12.2562,
      "step": 8122
    },
    {
      "epoch": 0.44232981424380413,
      "grad_norm": 0.5902049909043856,
      "learning_rate": 0.00018143837038946865,
      "loss": 12.5374,
      "step": 8123
    },
    {
      "epoch": 0.4423842682403872,
      "grad_norm": 0.6287680351391354,
      "learning_rate": 0.00018143325260138498,
      "loss": 12.4129,
      "step": 8124
    },
    {
      "epoch": 0.44243872223697017,
      "grad_norm": 0.6395184786336089,
      "learning_rate": 0.00018142813418006268,
      "loss": 12.5485,
      "step": 8125
    },
    {
      "epoch": 0.44249317623355316,
      "grad_norm": 0.583233484282126,
      "learning_rate": 0.00018142301512554153,
      "loss": 12.4745,
      "step": 8126
    },
    {
      "epoch": 0.4425476302301362,
      "grad_norm": 0.7393265090602186,
      "learning_rate": 0.00018141789543786135,
      "loss": 12.5784,
      "step": 8127
    },
    {
      "epoch": 0.4426020842267192,
      "grad_norm": 0.6099498576642015,
      "learning_rate": 0.000181412775117062,
      "loss": 12.3518,
      "step": 8128
    },
    {
      "epoch": 0.44265653822330225,
      "grad_norm": 0.616859217531763,
      "learning_rate": 0.00018140765416318325,
      "loss": 12.3706,
      "step": 8129
    },
    {
      "epoch": 0.44271099221988525,
      "grad_norm": 0.5738815684564759,
      "learning_rate": 0.00018140253257626493,
      "loss": 12.4688,
      "step": 8130
    },
    {
      "epoch": 0.44276544621646824,
      "grad_norm": 0.6287187273103042,
      "learning_rate": 0.00018139741035634686,
      "loss": 12.4894,
      "step": 8131
    },
    {
      "epoch": 0.4428199002130513,
      "grad_norm": 0.7150333260319802,
      "learning_rate": 0.00018139228750346887,
      "loss": 12.7204,
      "step": 8132
    },
    {
      "epoch": 0.4428743542096343,
      "grad_norm": 0.6655484781426367,
      "learning_rate": 0.0001813871640176708,
      "loss": 12.4733,
      "step": 8133
    },
    {
      "epoch": 0.4429288082062173,
      "grad_norm": 0.638583761056179,
      "learning_rate": 0.0001813820398989925,
      "loss": 12.4027,
      "step": 8134
    },
    {
      "epoch": 0.4429832622028003,
      "grad_norm": 0.6106168427470005,
      "learning_rate": 0.00018137691514747383,
      "loss": 12.35,
      "step": 8135
    },
    {
      "epoch": 0.4430377161993833,
      "grad_norm": 0.7678202216164216,
      "learning_rate": 0.0001813717897631546,
      "loss": 12.3463,
      "step": 8136
    },
    {
      "epoch": 0.4430921701959663,
      "grad_norm": 0.6351437344563395,
      "learning_rate": 0.0001813666637460747,
      "loss": 12.4133,
      "step": 8137
    },
    {
      "epoch": 0.44314662419254935,
      "grad_norm": 0.6594886319959798,
      "learning_rate": 0.00018136153709627398,
      "loss": 12.2562,
      "step": 8138
    },
    {
      "epoch": 0.44320107818913235,
      "grad_norm": 0.6110581042660779,
      "learning_rate": 0.00018135640981379228,
      "loss": 12.4918,
      "step": 8139
    },
    {
      "epoch": 0.44325553218571534,
      "grad_norm": 0.6397466953877733,
      "learning_rate": 0.0001813512818986695,
      "loss": 12.394,
      "step": 8140
    },
    {
      "epoch": 0.4433099861822984,
      "grad_norm": 0.6550859383422465,
      "learning_rate": 0.0001813461533509455,
      "loss": 12.4382,
      "step": 8141
    },
    {
      "epoch": 0.4433644401788814,
      "grad_norm": 0.6917336007640664,
      "learning_rate": 0.00018134102417066022,
      "loss": 12.4055,
      "step": 8142
    },
    {
      "epoch": 0.4434188941754644,
      "grad_norm": 0.5850970346286554,
      "learning_rate": 0.00018133589435785343,
      "loss": 12.4241,
      "step": 8143
    },
    {
      "epoch": 0.4434733481720474,
      "grad_norm": 0.5951134489105335,
      "learning_rate": 0.00018133076391256512,
      "loss": 12.3374,
      "step": 8144
    },
    {
      "epoch": 0.4435278021686304,
      "grad_norm": 0.6551033967638262,
      "learning_rate": 0.00018132563283483516,
      "loss": 12.487,
      "step": 8145
    },
    {
      "epoch": 0.4435822561652134,
      "grad_norm": 0.6538463123605098,
      "learning_rate": 0.0001813205011247034,
      "loss": 12.3122,
      "step": 8146
    },
    {
      "epoch": 0.44363671016179645,
      "grad_norm": 0.5950772399842815,
      "learning_rate": 0.00018131536878220983,
      "loss": 12.4683,
      "step": 8147
    },
    {
      "epoch": 0.44369116415837945,
      "grad_norm": 0.6694727358020924,
      "learning_rate": 0.0001813102358073943,
      "loss": 12.4369,
      "step": 8148
    },
    {
      "epoch": 0.44374561815496244,
      "grad_norm": 0.5837086135181214,
      "learning_rate": 0.00018130510220029674,
      "loss": 12.5195,
      "step": 8149
    },
    {
      "epoch": 0.4438000721515455,
      "grad_norm": 0.6162354476460444,
      "learning_rate": 0.00018129996796095705,
      "loss": 12.4831,
      "step": 8150
    },
    {
      "epoch": 0.4438545261481285,
      "grad_norm": 0.6581910316864154,
      "learning_rate": 0.0001812948330894152,
      "loss": 12.4733,
      "step": 8151
    },
    {
      "epoch": 0.4439089801447115,
      "grad_norm": 0.6213906461100707,
      "learning_rate": 0.0001812896975857111,
      "loss": 12.5239,
      "step": 8152
    },
    {
      "epoch": 0.4439634341412945,
      "grad_norm": 0.606727686748743,
      "learning_rate": 0.00018128456144988467,
      "loss": 12.5373,
      "step": 8153
    },
    {
      "epoch": 0.4440178881378775,
      "grad_norm": 0.735147788381272,
      "learning_rate": 0.00018127942468197582,
      "loss": 12.5172,
      "step": 8154
    },
    {
      "epoch": 0.4440723421344605,
      "grad_norm": 0.5921186091395286,
      "learning_rate": 0.0001812742872820246,
      "loss": 12.3883,
      "step": 8155
    },
    {
      "epoch": 0.44412679613104356,
      "grad_norm": 0.6844273116732905,
      "learning_rate": 0.00018126914925007087,
      "loss": 12.4229,
      "step": 8156
    },
    {
      "epoch": 0.44418125012762655,
      "grad_norm": 0.6966763682784651,
      "learning_rate": 0.00018126401058615458,
      "loss": 12.454,
      "step": 8157
    },
    {
      "epoch": 0.44423570412420954,
      "grad_norm": 0.7599245377713695,
      "learning_rate": 0.00018125887129031573,
      "loss": 12.5486,
      "step": 8158
    },
    {
      "epoch": 0.4442901581207926,
      "grad_norm": 0.5801118497495215,
      "learning_rate": 0.0001812537313625943,
      "loss": 12.4271,
      "step": 8159
    },
    {
      "epoch": 0.4443446121173756,
      "grad_norm": 0.679983449444894,
      "learning_rate": 0.00018124859080303018,
      "loss": 12.3809,
      "step": 8160
    },
    {
      "epoch": 0.4443990661139586,
      "grad_norm": 0.567276752664479,
      "learning_rate": 0.00018124344961166342,
      "loss": 12.3648,
      "step": 8161
    },
    {
      "epoch": 0.4444535201105416,
      "grad_norm": 0.6698341318989772,
      "learning_rate": 0.00018123830778853396,
      "loss": 12.4784,
      "step": 8162
    },
    {
      "epoch": 0.4445079741071246,
      "grad_norm": 0.6111490002980611,
      "learning_rate": 0.0001812331653336818,
      "loss": 12.432,
      "step": 8163
    },
    {
      "epoch": 0.44456242810370766,
      "grad_norm": 0.6109325508698443,
      "learning_rate": 0.00018122802224714693,
      "loss": 12.4912,
      "step": 8164
    },
    {
      "epoch": 0.44461688210029066,
      "grad_norm": 0.5743030968799749,
      "learning_rate": 0.00018122287852896933,
      "loss": 12.4442,
      "step": 8165
    },
    {
      "epoch": 0.44467133609687365,
      "grad_norm": 0.6070693152333769,
      "learning_rate": 0.00018121773417918902,
      "loss": 12.3768,
      "step": 8166
    },
    {
      "epoch": 0.4447257900934567,
      "grad_norm": 0.6136466125925997,
      "learning_rate": 0.00018121258919784596,
      "loss": 12.3442,
      "step": 8167
    },
    {
      "epoch": 0.4447802440900397,
      "grad_norm": 0.6061840102523574,
      "learning_rate": 0.0001812074435849802,
      "loss": 12.3715,
      "step": 8168
    },
    {
      "epoch": 0.4448346980866227,
      "grad_norm": 0.6503836273900797,
      "learning_rate": 0.00018120229734063176,
      "loss": 12.6093,
      "step": 8169
    },
    {
      "epoch": 0.44488915208320573,
      "grad_norm": 0.641563324903526,
      "learning_rate": 0.0001811971504648406,
      "loss": 12.4406,
      "step": 8170
    },
    {
      "epoch": 0.4449436060797887,
      "grad_norm": 0.7577204807990595,
      "learning_rate": 0.00018119200295764682,
      "loss": 12.5485,
      "step": 8171
    },
    {
      "epoch": 0.4449980600763717,
      "grad_norm": 0.6820962027227742,
      "learning_rate": 0.0001811868548190904,
      "loss": 12.5358,
      "step": 8172
    },
    {
      "epoch": 0.44505251407295476,
      "grad_norm": 0.6359158735353171,
      "learning_rate": 0.00018118170604921135,
      "loss": 12.5205,
      "step": 8173
    },
    {
      "epoch": 0.44510696806953776,
      "grad_norm": 0.6111147020344397,
      "learning_rate": 0.00018117655664804977,
      "loss": 12.4213,
      "step": 8174
    },
    {
      "epoch": 0.44516142206612075,
      "grad_norm": 0.6246736894085984,
      "learning_rate": 0.00018117140661564566,
      "loss": 12.4356,
      "step": 8175
    },
    {
      "epoch": 0.4452158760627038,
      "grad_norm": 0.5749923246542837,
      "learning_rate": 0.0001811662559520391,
      "loss": 12.3527,
      "step": 8176
    },
    {
      "epoch": 0.4452703300592868,
      "grad_norm": 0.6530224543649432,
      "learning_rate": 0.0001811611046572701,
      "loss": 12.5725,
      "step": 8177
    },
    {
      "epoch": 0.4453247840558698,
      "grad_norm": 0.5985662252307675,
      "learning_rate": 0.00018115595273137874,
      "loss": 12.4586,
      "step": 8178
    },
    {
      "epoch": 0.44537923805245283,
      "grad_norm": 0.7193898797443692,
      "learning_rate": 0.0001811508001744051,
      "loss": 12.4578,
      "step": 8179
    },
    {
      "epoch": 0.4454336920490358,
      "grad_norm": 0.6444061956976649,
      "learning_rate": 0.0001811456469863892,
      "loss": 12.5317,
      "step": 8180
    },
    {
      "epoch": 0.4454881460456188,
      "grad_norm": 0.5826868228600273,
      "learning_rate": 0.00018114049316737116,
      "loss": 12.5559,
      "step": 8181
    },
    {
      "epoch": 0.44554260004220186,
      "grad_norm": 0.6538082635561483,
      "learning_rate": 0.00018113533871739104,
      "loss": 12.5837,
      "step": 8182
    },
    {
      "epoch": 0.44559705403878486,
      "grad_norm": 0.6200677214450276,
      "learning_rate": 0.00018113018363648893,
      "loss": 12.4668,
      "step": 8183
    },
    {
      "epoch": 0.44565150803536785,
      "grad_norm": 0.6283652497127699,
      "learning_rate": 0.0001811250279247049,
      "loss": 12.4758,
      "step": 8184
    },
    {
      "epoch": 0.4457059620319509,
      "grad_norm": 0.7374868480204493,
      "learning_rate": 0.00018111987158207904,
      "loss": 12.3293,
      "step": 8185
    },
    {
      "epoch": 0.4457604160285339,
      "grad_norm": 0.6923032516638095,
      "learning_rate": 0.00018111471460865146,
      "loss": 12.2851,
      "step": 8186
    },
    {
      "epoch": 0.4458148700251169,
      "grad_norm": 0.6292759013526202,
      "learning_rate": 0.00018110955700446224,
      "loss": 12.4064,
      "step": 8187
    },
    {
      "epoch": 0.44586932402169993,
      "grad_norm": 0.7390498575776396,
      "learning_rate": 0.00018110439876955152,
      "loss": 12.4175,
      "step": 8188
    },
    {
      "epoch": 0.4459237780182829,
      "grad_norm": 0.6180650489407907,
      "learning_rate": 0.00018109923990395938,
      "loss": 12.3875,
      "step": 8189
    },
    {
      "epoch": 0.4459782320148659,
      "grad_norm": 0.6974498935982044,
      "learning_rate": 0.00018109408040772594,
      "loss": 12.6687,
      "step": 8190
    },
    {
      "epoch": 0.44603268601144896,
      "grad_norm": 0.7591191693971592,
      "learning_rate": 0.00018108892028089138,
      "loss": 12.4527,
      "step": 8191
    },
    {
      "epoch": 0.44608714000803196,
      "grad_norm": 0.6208047689810073,
      "learning_rate": 0.00018108375952349573,
      "loss": 12.4469,
      "step": 8192
    },
    {
      "epoch": 0.44614159400461495,
      "grad_norm": 0.781635480903214,
      "learning_rate": 0.0001810785981355792,
      "loss": 12.2055,
      "step": 8193
    },
    {
      "epoch": 0.446196048001198,
      "grad_norm": 0.7186489072431105,
      "learning_rate": 0.00018107343611718188,
      "loss": 12.6631,
      "step": 8194
    },
    {
      "epoch": 0.446250501997781,
      "grad_norm": 0.758734781288091,
      "learning_rate": 0.0001810682734683439,
      "loss": 12.2953,
      "step": 8195
    },
    {
      "epoch": 0.44630495599436404,
      "grad_norm": 0.6159585950601735,
      "learning_rate": 0.00018106311018910548,
      "loss": 12.4242,
      "step": 8196
    },
    {
      "epoch": 0.44635940999094703,
      "grad_norm": 0.7073770563576457,
      "learning_rate": 0.0001810579462795067,
      "loss": 12.5467,
      "step": 8197
    },
    {
      "epoch": 0.44641386398753,
      "grad_norm": 0.6096509478965852,
      "learning_rate": 0.00018105278173958774,
      "loss": 12.2076,
      "step": 8198
    },
    {
      "epoch": 0.44646831798411307,
      "grad_norm": 0.6392576052815786,
      "learning_rate": 0.00018104761656938871,
      "loss": 12.4672,
      "step": 8199
    },
    {
      "epoch": 0.44652277198069606,
      "grad_norm": 0.6049885538084858,
      "learning_rate": 0.00018104245076894988,
      "loss": 12.4106,
      "step": 8200
    },
    {
      "epoch": 0.44657722597727906,
      "grad_norm": 0.6642922311561321,
      "learning_rate": 0.00018103728433831133,
      "loss": 12.4219,
      "step": 8201
    },
    {
      "epoch": 0.4466316799738621,
      "grad_norm": 0.7173739433890999,
      "learning_rate": 0.00018103211727751327,
      "loss": 12.36,
      "step": 8202
    },
    {
      "epoch": 0.4466861339704451,
      "grad_norm": 0.6155354362488491,
      "learning_rate": 0.0001810269495865959,
      "loss": 12.4365,
      "step": 8203
    },
    {
      "epoch": 0.4467405879670281,
      "grad_norm": 0.5944106403284649,
      "learning_rate": 0.00018102178126559937,
      "loss": 12.4833,
      "step": 8204
    },
    {
      "epoch": 0.44679504196361114,
      "grad_norm": 0.6074043433035483,
      "learning_rate": 0.00018101661231456387,
      "loss": 12.4723,
      "step": 8205
    },
    {
      "epoch": 0.44684949596019413,
      "grad_norm": 0.6043312912035862,
      "learning_rate": 0.00018101144273352963,
      "loss": 12.3855,
      "step": 8206
    },
    {
      "epoch": 0.4469039499567771,
      "grad_norm": 0.6270599323257495,
      "learning_rate": 0.0001810062725225368,
      "loss": 12.4546,
      "step": 8207
    },
    {
      "epoch": 0.4469584039533602,
      "grad_norm": 0.6661394924846142,
      "learning_rate": 0.00018100110168162563,
      "loss": 12.3606,
      "step": 8208
    },
    {
      "epoch": 0.44701285794994317,
      "grad_norm": 0.7137124054573656,
      "learning_rate": 0.0001809959302108363,
      "loss": 12.589,
      "step": 8209
    },
    {
      "epoch": 0.44706731194652616,
      "grad_norm": 0.6665658000172924,
      "learning_rate": 0.00018099075811020904,
      "loss": 12.3141,
      "step": 8210
    },
    {
      "epoch": 0.4471217659431092,
      "grad_norm": 0.7127001415690272,
      "learning_rate": 0.00018098558537978405,
      "loss": 12.5324,
      "step": 8211
    },
    {
      "epoch": 0.4471762199396922,
      "grad_norm": 0.6345169385181538,
      "learning_rate": 0.00018098041201960156,
      "loss": 12.44,
      "step": 8212
    },
    {
      "epoch": 0.4472306739362752,
      "grad_norm": 0.6563976622677672,
      "learning_rate": 0.00018097523802970185,
      "loss": 12.3428,
      "step": 8213
    },
    {
      "epoch": 0.44728512793285824,
      "grad_norm": 0.6687606642611973,
      "learning_rate": 0.00018097006341012505,
      "loss": 12.4602,
      "step": 8214
    },
    {
      "epoch": 0.44733958192944123,
      "grad_norm": 0.7164323797540626,
      "learning_rate": 0.0001809648881609115,
      "loss": 12.5104,
      "step": 8215
    },
    {
      "epoch": 0.4473940359260242,
      "grad_norm": 0.6878977953960316,
      "learning_rate": 0.0001809597122821014,
      "loss": 12.4367,
      "step": 8216
    },
    {
      "epoch": 0.4474484899226073,
      "grad_norm": 0.5843306191896047,
      "learning_rate": 0.000180954535773735,
      "loss": 12.3788,
      "step": 8217
    },
    {
      "epoch": 0.44750294391919027,
      "grad_norm": 0.627590152184721,
      "learning_rate": 0.0001809493586358525,
      "loss": 12.335,
      "step": 8218
    },
    {
      "epoch": 0.44755739791577326,
      "grad_norm": 0.6730383060730796,
      "learning_rate": 0.00018094418086849428,
      "loss": 12.4502,
      "step": 8219
    },
    {
      "epoch": 0.4476118519123563,
      "grad_norm": 0.6622308298602019,
      "learning_rate": 0.0001809390024717005,
      "loss": 12.442,
      "step": 8220
    },
    {
      "epoch": 0.4476663059089393,
      "grad_norm": 0.5680820510071307,
      "learning_rate": 0.0001809338234455115,
      "loss": 12.5298,
      "step": 8221
    },
    {
      "epoch": 0.4477207599055223,
      "grad_norm": 0.6048364187339758,
      "learning_rate": 0.00018092864378996748,
      "loss": 12.4445,
      "step": 8222
    },
    {
      "epoch": 0.44777521390210534,
      "grad_norm": 0.7737050573569966,
      "learning_rate": 0.00018092346350510877,
      "loss": 12.5471,
      "step": 8223
    },
    {
      "epoch": 0.44782966789868833,
      "grad_norm": 0.6120992948950894,
      "learning_rate": 0.0001809182825909756,
      "loss": 12.4766,
      "step": 8224
    },
    {
      "epoch": 0.4478841218952713,
      "grad_norm": 0.7214950968754054,
      "learning_rate": 0.00018091310104760834,
      "loss": 12.4849,
      "step": 8225
    },
    {
      "epoch": 0.4479385758918544,
      "grad_norm": 0.6105295559177767,
      "learning_rate": 0.00018090791887504724,
      "loss": 12.4912,
      "step": 8226
    },
    {
      "epoch": 0.44799302988843737,
      "grad_norm": 0.5889511018642188,
      "learning_rate": 0.00018090273607333256,
      "loss": 12.5213,
      "step": 8227
    },
    {
      "epoch": 0.44804748388502036,
      "grad_norm": 0.6737149319485795,
      "learning_rate": 0.00018089755264250466,
      "loss": 12.3463,
      "step": 8228
    },
    {
      "epoch": 0.4481019378816034,
      "grad_norm": 0.6485216645092555,
      "learning_rate": 0.00018089236858260383,
      "loss": 12.4522,
      "step": 8229
    },
    {
      "epoch": 0.4481563918781864,
      "grad_norm": 0.6290966852441273,
      "learning_rate": 0.00018088718389367036,
      "loss": 12.479,
      "step": 8230
    },
    {
      "epoch": 0.44821084587476945,
      "grad_norm": 0.6435504811467402,
      "learning_rate": 0.00018088199857574456,
      "loss": 12.5209,
      "step": 8231
    },
    {
      "epoch": 0.44826529987135244,
      "grad_norm": 0.6477490585103859,
      "learning_rate": 0.00018087681262886682,
      "loss": 12.4887,
      "step": 8232
    },
    {
      "epoch": 0.44831975386793543,
      "grad_norm": 0.7152524441450054,
      "learning_rate": 0.0001808716260530774,
      "loss": 12.5928,
      "step": 8233
    },
    {
      "epoch": 0.4483742078645185,
      "grad_norm": 0.6169314782015952,
      "learning_rate": 0.00018086643884841666,
      "loss": 12.4819,
      "step": 8234
    },
    {
      "epoch": 0.4484286618611015,
      "grad_norm": 0.61995875690478,
      "learning_rate": 0.00018086125101492493,
      "loss": 12.5438,
      "step": 8235
    },
    {
      "epoch": 0.44848311585768447,
      "grad_norm": 0.6410424565512555,
      "learning_rate": 0.00018085606255264255,
      "loss": 12.3693,
      "step": 8236
    },
    {
      "epoch": 0.4485375698542675,
      "grad_norm": 0.675278433061094,
      "learning_rate": 0.00018085087346160988,
      "loss": 12.4955,
      "step": 8237
    },
    {
      "epoch": 0.4485920238508505,
      "grad_norm": 0.6860164489550601,
      "learning_rate": 0.00018084568374186722,
      "loss": 12.3339,
      "step": 8238
    },
    {
      "epoch": 0.4486464778474335,
      "grad_norm": 0.70350072710245,
      "learning_rate": 0.000180840493393455,
      "loss": 12.424,
      "step": 8239
    },
    {
      "epoch": 0.44870093184401655,
      "grad_norm": 0.6038702419154595,
      "learning_rate": 0.00018083530241641357,
      "loss": 12.5123,
      "step": 8240
    },
    {
      "epoch": 0.44875538584059954,
      "grad_norm": 0.6831009948912166,
      "learning_rate": 0.00018083011081078323,
      "loss": 12.4619,
      "step": 8241
    },
    {
      "epoch": 0.44880983983718253,
      "grad_norm": 0.6303418377962721,
      "learning_rate": 0.0001808249185766044,
      "loss": 12.4241,
      "step": 8242
    },
    {
      "epoch": 0.4488642938337656,
      "grad_norm": 0.6777704932279972,
      "learning_rate": 0.00018081972571391749,
      "loss": 12.538,
      "step": 8243
    },
    {
      "epoch": 0.4489187478303486,
      "grad_norm": 0.6216889573416485,
      "learning_rate": 0.00018081453222276278,
      "loss": 12.5019,
      "step": 8244
    },
    {
      "epoch": 0.44897320182693157,
      "grad_norm": 0.827652560719684,
      "learning_rate": 0.00018080933810318072,
      "loss": 12.473,
      "step": 8245
    },
    {
      "epoch": 0.4490276558235146,
      "grad_norm": 0.6836190786922434,
      "learning_rate": 0.00018080414335521172,
      "loss": 12.4538,
      "step": 8246
    },
    {
      "epoch": 0.4490821098200976,
      "grad_norm": 0.7056054941784216,
      "learning_rate": 0.00018079894797889615,
      "loss": 12.5438,
      "step": 8247
    },
    {
      "epoch": 0.4491365638166806,
      "grad_norm": 0.7109116869401612,
      "learning_rate": 0.00018079375197427438,
      "loss": 12.5076,
      "step": 8248
    },
    {
      "epoch": 0.44919101781326365,
      "grad_norm": 0.6384313871607461,
      "learning_rate": 0.00018078855534138687,
      "loss": 12.3335,
      "step": 8249
    },
    {
      "epoch": 0.44924547180984664,
      "grad_norm": 0.6014901136296119,
      "learning_rate": 0.00018078335808027402,
      "loss": 12.4224,
      "step": 8250
    },
    {
      "epoch": 0.44929992580642963,
      "grad_norm": 0.6995417512969619,
      "learning_rate": 0.00018077816019097622,
      "loss": 12.4247,
      "step": 8251
    },
    {
      "epoch": 0.4493543798030127,
      "grad_norm": 0.646866040388919,
      "learning_rate": 0.0001807729616735339,
      "loss": 12.3516,
      "step": 8252
    },
    {
      "epoch": 0.4494088337995957,
      "grad_norm": 0.6323119202761118,
      "learning_rate": 0.00018076776252798747,
      "loss": 12.466,
      "step": 8253
    },
    {
      "epoch": 0.44946328779617867,
      "grad_norm": 0.6573015764103591,
      "learning_rate": 0.0001807625627543774,
      "loss": 12.3854,
      "step": 8254
    },
    {
      "epoch": 0.4495177417927617,
      "grad_norm": 0.7660222464548213,
      "learning_rate": 0.0001807573623527441,
      "loss": 12.6872,
      "step": 8255
    },
    {
      "epoch": 0.4495721957893447,
      "grad_norm": 0.6917988160741411,
      "learning_rate": 0.00018075216132312796,
      "loss": 12.4,
      "step": 8256
    },
    {
      "epoch": 0.4496266497859277,
      "grad_norm": 0.6528921899765494,
      "learning_rate": 0.0001807469596655695,
      "loss": 12.4176,
      "step": 8257
    },
    {
      "epoch": 0.44968110378251075,
      "grad_norm": 0.6387928727990673,
      "learning_rate": 0.00018074175738010915,
      "loss": 12.553,
      "step": 8258
    },
    {
      "epoch": 0.44973555777909374,
      "grad_norm": 0.7032323497462504,
      "learning_rate": 0.00018073655446678734,
      "loss": 12.3964,
      "step": 8259
    },
    {
      "epoch": 0.44979001177567673,
      "grad_norm": 0.6201282183769875,
      "learning_rate": 0.00018073135092564456,
      "loss": 12.4337,
      "step": 8260
    },
    {
      "epoch": 0.4498444657722598,
      "grad_norm": 0.6546034000172873,
      "learning_rate": 0.00018072614675672123,
      "loss": 12.5565,
      "step": 8261
    },
    {
      "epoch": 0.4498989197688428,
      "grad_norm": 0.6167943409390654,
      "learning_rate": 0.00018072094196005788,
      "loss": 12.3856,
      "step": 8262
    },
    {
      "epoch": 0.4499533737654258,
      "grad_norm": 0.6411173044207982,
      "learning_rate": 0.0001807157365356949,
      "loss": 12.4758,
      "step": 8263
    },
    {
      "epoch": 0.4500078277620088,
      "grad_norm": 0.6660852274828428,
      "learning_rate": 0.00018071053048367283,
      "loss": 12.3944,
      "step": 8264
    },
    {
      "epoch": 0.4500622817585918,
      "grad_norm": 0.7313184201567402,
      "learning_rate": 0.00018070532380403217,
      "loss": 12.5177,
      "step": 8265
    },
    {
      "epoch": 0.45011673575517486,
      "grad_norm": 0.5888785217430297,
      "learning_rate": 0.00018070011649681336,
      "loss": 12.4226,
      "step": 8266
    },
    {
      "epoch": 0.45017118975175785,
      "grad_norm": 0.6271369849906314,
      "learning_rate": 0.00018069490856205693,
      "loss": 12.4255,
      "step": 8267
    },
    {
      "epoch": 0.45022564374834084,
      "grad_norm": 0.6877928172862263,
      "learning_rate": 0.00018068969999980334,
      "loss": 12.4457,
      "step": 8268
    },
    {
      "epoch": 0.4502800977449239,
      "grad_norm": 0.7126841747836358,
      "learning_rate": 0.00018068449081009308,
      "loss": 12.4507,
      "step": 8269
    },
    {
      "epoch": 0.4503345517415069,
      "grad_norm": 0.5935590021761236,
      "learning_rate": 0.00018067928099296676,
      "loss": 12.4129,
      "step": 8270
    },
    {
      "epoch": 0.4503890057380899,
      "grad_norm": 0.7332547699397844,
      "learning_rate": 0.00018067407054846477,
      "loss": 12.4179,
      "step": 8271
    },
    {
      "epoch": 0.4504434597346729,
      "grad_norm": 0.6439231739492874,
      "learning_rate": 0.00018066885947662768,
      "loss": 12.4201,
      "step": 8272
    },
    {
      "epoch": 0.4504979137312559,
      "grad_norm": 0.6404282685943609,
      "learning_rate": 0.00018066364777749602,
      "loss": 12.4724,
      "step": 8273
    },
    {
      "epoch": 0.4505523677278389,
      "grad_norm": 0.6757165721643505,
      "learning_rate": 0.00018065843545111033,
      "loss": 12.3736,
      "step": 8274
    },
    {
      "epoch": 0.45060682172442196,
      "grad_norm": 0.6436434529216677,
      "learning_rate": 0.00018065322249751108,
      "loss": 12.5005,
      "step": 8275
    },
    {
      "epoch": 0.45066127572100495,
      "grad_norm": 0.7705414807500481,
      "learning_rate": 0.00018064800891673887,
      "loss": 12.5556,
      "step": 8276
    },
    {
      "epoch": 0.45071572971758794,
      "grad_norm": 0.6253013831319388,
      "learning_rate": 0.0001806427947088342,
      "loss": 12.3188,
      "step": 8277
    },
    {
      "epoch": 0.450770183714171,
      "grad_norm": 0.6693402182742427,
      "learning_rate": 0.00018063757987383765,
      "loss": 12.4618,
      "step": 8278
    },
    {
      "epoch": 0.450824637710754,
      "grad_norm": 0.809093178489467,
      "learning_rate": 0.00018063236441178977,
      "loss": 12.3269,
      "step": 8279
    },
    {
      "epoch": 0.450879091707337,
      "grad_norm": 0.6960771578027255,
      "learning_rate": 0.0001806271483227311,
      "loss": 12.446,
      "step": 8280
    },
    {
      "epoch": 0.45093354570392,
      "grad_norm": 0.6836104620072424,
      "learning_rate": 0.00018062193160670216,
      "loss": 12.4956,
      "step": 8281
    },
    {
      "epoch": 0.450987999700503,
      "grad_norm": 0.6288308608485124,
      "learning_rate": 0.0001806167142637436,
      "loss": 12.4298,
      "step": 8282
    },
    {
      "epoch": 0.451042453697086,
      "grad_norm": 0.6590638845740401,
      "learning_rate": 0.00018061149629389594,
      "loss": 12.6313,
      "step": 8283
    },
    {
      "epoch": 0.45109690769366906,
      "grad_norm": 0.713252599730932,
      "learning_rate": 0.00018060627769719977,
      "loss": 12.5886,
      "step": 8284
    },
    {
      "epoch": 0.45115136169025205,
      "grad_norm": 0.6662160820965829,
      "learning_rate": 0.00018060105847369563,
      "loss": 12.4701,
      "step": 8285
    },
    {
      "epoch": 0.45120581568683504,
      "grad_norm": 0.8098651522302789,
      "learning_rate": 0.0001805958386234242,
      "loss": 12.5582,
      "step": 8286
    },
    {
      "epoch": 0.4512602696834181,
      "grad_norm": 0.5886671797308553,
      "learning_rate": 0.000180590618146426,
      "loss": 12.3877,
      "step": 8287
    },
    {
      "epoch": 0.4513147236800011,
      "grad_norm": 0.6010781282278521,
      "learning_rate": 0.00018058539704274163,
      "loss": 12.4755,
      "step": 8288
    },
    {
      "epoch": 0.4513691776765841,
      "grad_norm": 0.622368369814635,
      "learning_rate": 0.00018058017531241168,
      "loss": 12.4668,
      "step": 8289
    },
    {
      "epoch": 0.4514236316731671,
      "grad_norm": 0.5814870801804356,
      "learning_rate": 0.00018057495295547676,
      "loss": 12.3564,
      "step": 8290
    },
    {
      "epoch": 0.4514780856697501,
      "grad_norm": 0.6701604109652208,
      "learning_rate": 0.00018056972997197753,
      "loss": 12.5295,
      "step": 8291
    },
    {
      "epoch": 0.4515325396663331,
      "grad_norm": 0.6107798920489389,
      "learning_rate": 0.0001805645063619546,
      "loss": 12.4094,
      "step": 8292
    },
    {
      "epoch": 0.45158699366291616,
      "grad_norm": 0.6121065620221543,
      "learning_rate": 0.00018055928212544847,
      "loss": 12.5846,
      "step": 8293
    },
    {
      "epoch": 0.45164144765949915,
      "grad_norm": 0.6625822088074984,
      "learning_rate": 0.00018055405726249992,
      "loss": 12.3911,
      "step": 8294
    },
    {
      "epoch": 0.45169590165608214,
      "grad_norm": 0.6761019614906887,
      "learning_rate": 0.0001805488317731495,
      "loss": 12.5609,
      "step": 8295
    },
    {
      "epoch": 0.4517503556526652,
      "grad_norm": 0.6057534358071944,
      "learning_rate": 0.00018054360565743785,
      "loss": 12.5217,
      "step": 8296
    },
    {
      "epoch": 0.4518048096492482,
      "grad_norm": 0.5904779845769116,
      "learning_rate": 0.0001805383789154056,
      "loss": 12.506,
      "step": 8297
    },
    {
      "epoch": 0.45185926364583123,
      "grad_norm": 0.7203160399786142,
      "learning_rate": 0.00018053315154709342,
      "loss": 12.6001,
      "step": 8298
    },
    {
      "epoch": 0.4519137176424142,
      "grad_norm": 0.7114625436615308,
      "learning_rate": 0.00018052792355254197,
      "loss": 12.4177,
      "step": 8299
    },
    {
      "epoch": 0.4519681716389972,
      "grad_norm": 0.6171990688626064,
      "learning_rate": 0.00018052269493179184,
      "loss": 12.3536,
      "step": 8300
    },
    {
      "epoch": 0.45202262563558027,
      "grad_norm": 0.6358158470366418,
      "learning_rate": 0.00018051746568488377,
      "loss": 12.5709,
      "step": 8301
    },
    {
      "epoch": 0.45207707963216326,
      "grad_norm": 0.6387813901796725,
      "learning_rate": 0.00018051223581185837,
      "loss": 12.3938,
      "step": 8302
    },
    {
      "epoch": 0.45213153362874625,
      "grad_norm": 0.571110397702212,
      "learning_rate": 0.0001805070053127563,
      "loss": 12.4995,
      "step": 8303
    },
    {
      "epoch": 0.4521859876253293,
      "grad_norm": 0.6288736694560902,
      "learning_rate": 0.00018050177418761827,
      "loss": 12.2448,
      "step": 8304
    },
    {
      "epoch": 0.4522404416219123,
      "grad_norm": 0.6325902111994797,
      "learning_rate": 0.00018049654243648496,
      "loss": 12.3974,
      "step": 8305
    },
    {
      "epoch": 0.4522948956184953,
      "grad_norm": 0.6493021827097402,
      "learning_rate": 0.000180491310059397,
      "loss": 12.4227,
      "step": 8306
    },
    {
      "epoch": 0.45234934961507833,
      "grad_norm": 0.5846396599292224,
      "learning_rate": 0.00018048607705639517,
      "loss": 12.3472,
      "step": 8307
    },
    {
      "epoch": 0.4524038036116613,
      "grad_norm": 0.649321265011039,
      "learning_rate": 0.00018048084342752008,
      "loss": 12.5292,
      "step": 8308
    },
    {
      "epoch": 0.4524582576082443,
      "grad_norm": 0.6319409826716816,
      "learning_rate": 0.00018047560917281246,
      "loss": 12.5412,
      "step": 8309
    },
    {
      "epoch": 0.45251271160482737,
      "grad_norm": 0.7202103340396316,
      "learning_rate": 0.00018047037429231298,
      "loss": 12.538,
      "step": 8310
    },
    {
      "epoch": 0.45256716560141036,
      "grad_norm": 0.708175500993768,
      "learning_rate": 0.00018046513878606238,
      "loss": 12.6258,
      "step": 8311
    },
    {
      "epoch": 0.45262161959799335,
      "grad_norm": 0.5807181380628393,
      "learning_rate": 0.0001804599026541014,
      "loss": 12.3849,
      "step": 8312
    },
    {
      "epoch": 0.4526760735945764,
      "grad_norm": 0.6428716877562957,
      "learning_rate": 0.0001804546658964707,
      "loss": 12.4105,
      "step": 8313
    },
    {
      "epoch": 0.4527305275911594,
      "grad_norm": 0.6653068583401828,
      "learning_rate": 0.00018044942851321103,
      "loss": 12.3114,
      "step": 8314
    },
    {
      "epoch": 0.4527849815877424,
      "grad_norm": 0.6454680182879358,
      "learning_rate": 0.0001804441905043631,
      "loss": 12.532,
      "step": 8315
    },
    {
      "epoch": 0.45283943558432543,
      "grad_norm": 0.5780892948692522,
      "learning_rate": 0.00018043895186996766,
      "loss": 12.3504,
      "step": 8316
    },
    {
      "epoch": 0.4528938895809084,
      "grad_norm": 0.69638392730691,
      "learning_rate": 0.00018043371261006546,
      "loss": 12.4106,
      "step": 8317
    },
    {
      "epoch": 0.4529483435774914,
      "grad_norm": 0.635956986948982,
      "learning_rate": 0.0001804284727246972,
      "loss": 12.5283,
      "step": 8318
    },
    {
      "epoch": 0.45300279757407447,
      "grad_norm": 0.6313431542322339,
      "learning_rate": 0.00018042323221390368,
      "loss": 12.5512,
      "step": 8319
    },
    {
      "epoch": 0.45305725157065746,
      "grad_norm": 0.6489741452015877,
      "learning_rate": 0.0001804179910777256,
      "loss": 12.4485,
      "step": 8320
    },
    {
      "epoch": 0.45311170556724045,
      "grad_norm": 0.6315006712394421,
      "learning_rate": 0.00018041274931620372,
      "loss": 12.5214,
      "step": 8321
    },
    {
      "epoch": 0.4531661595638235,
      "grad_norm": 0.5990546472160445,
      "learning_rate": 0.00018040750692937881,
      "loss": 12.525,
      "step": 8322
    },
    {
      "epoch": 0.4532206135604065,
      "grad_norm": 0.5871850278370101,
      "learning_rate": 0.00018040226391729166,
      "loss": 12.3591,
      "step": 8323
    },
    {
      "epoch": 0.4532750675569895,
      "grad_norm": 0.658322856276941,
      "learning_rate": 0.00018039702027998304,
      "loss": 12.5054,
      "step": 8324
    },
    {
      "epoch": 0.45332952155357253,
      "grad_norm": 0.639803097756465,
      "learning_rate": 0.0001803917760174937,
      "loss": 12.5238,
      "step": 8325
    },
    {
      "epoch": 0.4533839755501555,
      "grad_norm": 0.8436479688153324,
      "learning_rate": 0.00018038653112986442,
      "loss": 12.4266,
      "step": 8326
    },
    {
      "epoch": 0.4534384295467385,
      "grad_norm": 0.662813388242909,
      "learning_rate": 0.00018038128561713597,
      "loss": 12.3991,
      "step": 8327
    },
    {
      "epoch": 0.45349288354332157,
      "grad_norm": 0.6802356628734042,
      "learning_rate": 0.00018037603947934917,
      "loss": 12.3814,
      "step": 8328
    },
    {
      "epoch": 0.45354733753990456,
      "grad_norm": 0.7191399813569995,
      "learning_rate": 0.00018037079271654484,
      "loss": 12.5178,
      "step": 8329
    },
    {
      "epoch": 0.4536017915364876,
      "grad_norm": 0.7728736862833322,
      "learning_rate": 0.00018036554532876374,
      "loss": 12.58,
      "step": 8330
    },
    {
      "epoch": 0.4536562455330706,
      "grad_norm": 0.7082727583696239,
      "learning_rate": 0.00018036029731604666,
      "loss": 12.3314,
      "step": 8331
    },
    {
      "epoch": 0.4537106995296536,
      "grad_norm": 0.63936538910493,
      "learning_rate": 0.00018035504867843446,
      "loss": 12.5019,
      "step": 8332
    },
    {
      "epoch": 0.45376515352623664,
      "grad_norm": 0.7309572506200637,
      "learning_rate": 0.0001803497994159679,
      "loss": 12.3034,
      "step": 8333
    },
    {
      "epoch": 0.45381960752281963,
      "grad_norm": 0.7346023329112903,
      "learning_rate": 0.00018034454952868783,
      "loss": 12.493,
      "step": 8334
    },
    {
      "epoch": 0.4538740615194026,
      "grad_norm": 0.6438331942903343,
      "learning_rate": 0.00018033929901663508,
      "loss": 12.4777,
      "step": 8335
    },
    {
      "epoch": 0.4539285155159857,
      "grad_norm": 0.6429282768622242,
      "learning_rate": 0.00018033404787985046,
      "loss": 12.5446,
      "step": 8336
    },
    {
      "epoch": 0.45398296951256867,
      "grad_norm": 0.6470283192872487,
      "learning_rate": 0.00018032879611837484,
      "loss": 12.4359,
      "step": 8337
    },
    {
      "epoch": 0.45403742350915166,
      "grad_norm": 0.6652689528162133,
      "learning_rate": 0.00018032354373224898,
      "loss": 12.4367,
      "step": 8338
    },
    {
      "epoch": 0.4540918775057347,
      "grad_norm": 0.7137004022132828,
      "learning_rate": 0.00018031829072151382,
      "loss": 12.4301,
      "step": 8339
    },
    {
      "epoch": 0.4541463315023177,
      "grad_norm": 0.6441967324262494,
      "learning_rate": 0.0001803130370862101,
      "loss": 12.4321,
      "step": 8340
    },
    {
      "epoch": 0.4542007854989007,
      "grad_norm": 0.7165948198137981,
      "learning_rate": 0.0001803077828263788,
      "loss": 12.3381,
      "step": 8341
    },
    {
      "epoch": 0.45425523949548374,
      "grad_norm": 0.6936232718909088,
      "learning_rate": 0.0001803025279420607,
      "loss": 12.4531,
      "step": 8342
    },
    {
      "epoch": 0.45430969349206674,
      "grad_norm": 0.6793974980558357,
      "learning_rate": 0.00018029727243329665,
      "loss": 12.4894,
      "step": 8343
    },
    {
      "epoch": 0.45436414748864973,
      "grad_norm": 0.7578080572478764,
      "learning_rate": 0.00018029201630012756,
      "loss": 12.4671,
      "step": 8344
    },
    {
      "epoch": 0.4544186014852328,
      "grad_norm": 0.6375758735624671,
      "learning_rate": 0.00018028675954259428,
      "loss": 12.4301,
      "step": 8345
    },
    {
      "epoch": 0.45447305548181577,
      "grad_norm": 0.6403854044238918,
      "learning_rate": 0.00018028150216073768,
      "loss": 12.5084,
      "step": 8346
    },
    {
      "epoch": 0.45452750947839876,
      "grad_norm": 0.744259474887641,
      "learning_rate": 0.00018027624415459869,
      "loss": 12.4074,
      "step": 8347
    },
    {
      "epoch": 0.4545819634749818,
      "grad_norm": 0.6532864185316472,
      "learning_rate": 0.00018027098552421812,
      "loss": 12.1952,
      "step": 8348
    },
    {
      "epoch": 0.4546364174715648,
      "grad_norm": 0.6709441299649114,
      "learning_rate": 0.00018026572626963697,
      "loss": 12.4102,
      "step": 8349
    },
    {
      "epoch": 0.4546908714681478,
      "grad_norm": 0.6527345750406884,
      "learning_rate": 0.000180260466390896,
      "loss": 12.5381,
      "step": 8350
    },
    {
      "epoch": 0.45474532546473084,
      "grad_norm": 0.5993932846023744,
      "learning_rate": 0.00018025520588803624,
      "loss": 12.3806,
      "step": 8351
    },
    {
      "epoch": 0.45479977946131384,
      "grad_norm": 0.6560079767341911,
      "learning_rate": 0.0001802499447610985,
      "loss": 12.426,
      "step": 8352
    },
    {
      "epoch": 0.45485423345789683,
      "grad_norm": 0.6208449905155884,
      "learning_rate": 0.00018024468301012375,
      "loss": 12.4476,
      "step": 8353
    },
    {
      "epoch": 0.4549086874544799,
      "grad_norm": 0.5938706449016852,
      "learning_rate": 0.00018023942063515288,
      "loss": 12.3896,
      "step": 8354
    },
    {
      "epoch": 0.45496314145106287,
      "grad_norm": 0.6840502037543044,
      "learning_rate": 0.00018023415763622684,
      "loss": 12.5287,
      "step": 8355
    },
    {
      "epoch": 0.45501759544764586,
      "grad_norm": 0.6484201237103069,
      "learning_rate": 0.0001802288940133865,
      "loss": 12.5393,
      "step": 8356
    },
    {
      "epoch": 0.4550720494442289,
      "grad_norm": 0.6536567142719277,
      "learning_rate": 0.00018022362976667286,
      "loss": 12.5158,
      "step": 8357
    },
    {
      "epoch": 0.4551265034408119,
      "grad_norm": 0.6495453175136698,
      "learning_rate": 0.00018021836489612682,
      "loss": 12.402,
      "step": 8358
    },
    {
      "epoch": 0.4551809574373949,
      "grad_norm": 0.6546075029096625,
      "learning_rate": 0.0001802130994017893,
      "loss": 12.4056,
      "step": 8359
    },
    {
      "epoch": 0.45523541143397794,
      "grad_norm": 0.6169108403795847,
      "learning_rate": 0.00018020783328370128,
      "loss": 12.3984,
      "step": 8360
    },
    {
      "epoch": 0.45528986543056094,
      "grad_norm": 0.7334837642317463,
      "learning_rate": 0.00018020256654190372,
      "loss": 12.5352,
      "step": 8361
    },
    {
      "epoch": 0.45534431942714393,
      "grad_norm": 0.6355316885852189,
      "learning_rate": 0.0001801972991764375,
      "loss": 12.4515,
      "step": 8362
    },
    {
      "epoch": 0.455398773423727,
      "grad_norm": 0.6762017192687889,
      "learning_rate": 0.00018019203118734364,
      "loss": 12.5213,
      "step": 8363
    },
    {
      "epoch": 0.45545322742030997,
      "grad_norm": 0.652040432310015,
      "learning_rate": 0.00018018676257466315,
      "loss": 12.3384,
      "step": 8364
    },
    {
      "epoch": 0.455507681416893,
      "grad_norm": 0.6052011344639239,
      "learning_rate": 0.0001801814933384369,
      "loss": 12.5054,
      "step": 8365
    },
    {
      "epoch": 0.455562135413476,
      "grad_norm": 0.7345528713970678,
      "learning_rate": 0.00018017622347870594,
      "loss": 12.2335,
      "step": 8366
    },
    {
      "epoch": 0.455616589410059,
      "grad_norm": 0.6469247135524236,
      "learning_rate": 0.0001801709529955112,
      "loss": 12.4192,
      "step": 8367
    },
    {
      "epoch": 0.45567104340664205,
      "grad_norm": 0.562462112853363,
      "learning_rate": 0.00018016568188889366,
      "loss": 12.3651,
      "step": 8368
    },
    {
      "epoch": 0.45572549740322504,
      "grad_norm": 0.7793939972367465,
      "learning_rate": 0.0001801604101588944,
      "loss": 12.5982,
      "step": 8369
    },
    {
      "epoch": 0.45577995139980804,
      "grad_norm": 0.7634582703853848,
      "learning_rate": 0.00018015513780555428,
      "loss": 12.4733,
      "step": 8370
    },
    {
      "epoch": 0.4558344053963911,
      "grad_norm": 0.7433426272281132,
      "learning_rate": 0.00018014986482891443,
      "loss": 12.3873,
      "step": 8371
    },
    {
      "epoch": 0.4558888593929741,
      "grad_norm": 0.7151847951356256,
      "learning_rate": 0.00018014459122901575,
      "loss": 12.4266,
      "step": 8372
    },
    {
      "epoch": 0.45594331338955707,
      "grad_norm": 0.6536945132481292,
      "learning_rate": 0.00018013931700589926,
      "loss": 12.4628,
      "step": 8373
    },
    {
      "epoch": 0.4559977673861401,
      "grad_norm": 0.6432829911573463,
      "learning_rate": 0.00018013404215960608,
      "loss": 12.5578,
      "step": 8374
    },
    {
      "epoch": 0.4560522213827231,
      "grad_norm": 0.7332230865995127,
      "learning_rate": 0.0001801287666901771,
      "loss": 12.4341,
      "step": 8375
    },
    {
      "epoch": 0.4561066753793061,
      "grad_norm": 0.6581836725868025,
      "learning_rate": 0.0001801234905976534,
      "loss": 12.4149,
      "step": 8376
    },
    {
      "epoch": 0.45616112937588915,
      "grad_norm": 0.7295789375243006,
      "learning_rate": 0.000180118213882076,
      "loss": 12.4868,
      "step": 8377
    },
    {
      "epoch": 0.45621558337247214,
      "grad_norm": 0.6554620231047964,
      "learning_rate": 0.00018011293654348592,
      "loss": 12.4085,
      "step": 8378
    },
    {
      "epoch": 0.45627003736905514,
      "grad_norm": 0.6970016512205633,
      "learning_rate": 0.00018010765858192425,
      "loss": 12.5611,
      "step": 8379
    },
    {
      "epoch": 0.4563244913656382,
      "grad_norm": 0.7069624152061406,
      "learning_rate": 0.00018010237999743195,
      "loss": 12.4065,
      "step": 8380
    },
    {
      "epoch": 0.4563789453622212,
      "grad_norm": 0.6534734284592699,
      "learning_rate": 0.00018009710079005013,
      "loss": 12.4033,
      "step": 8381
    },
    {
      "epoch": 0.45643339935880417,
      "grad_norm": 0.6733227018555901,
      "learning_rate": 0.00018009182095981983,
      "loss": 12.6301,
      "step": 8382
    },
    {
      "epoch": 0.4564878533553872,
      "grad_norm": 0.6883935790218004,
      "learning_rate": 0.00018008654050678208,
      "loss": 12.4693,
      "step": 8383
    },
    {
      "epoch": 0.4565423073519702,
      "grad_norm": 0.6744703967734476,
      "learning_rate": 0.000180081259430978,
      "loss": 12.3845,
      "step": 8384
    },
    {
      "epoch": 0.4565967613485532,
      "grad_norm": 0.7186218426749795,
      "learning_rate": 0.00018007597773244855,
      "loss": 12.3878,
      "step": 8385
    },
    {
      "epoch": 0.45665121534513625,
      "grad_norm": 0.6349234443623138,
      "learning_rate": 0.00018007069541123492,
      "loss": 12.4049,
      "step": 8386
    },
    {
      "epoch": 0.45670566934171924,
      "grad_norm": 0.5824037807653083,
      "learning_rate": 0.0001800654124673781,
      "loss": 12.3113,
      "step": 8387
    },
    {
      "epoch": 0.45676012333830224,
      "grad_norm": 0.6222886438219286,
      "learning_rate": 0.00018006012890091924,
      "loss": 12.4073,
      "step": 8388
    },
    {
      "epoch": 0.4568145773348853,
      "grad_norm": 0.7319420156442162,
      "learning_rate": 0.00018005484471189938,
      "loss": 12.5158,
      "step": 8389
    },
    {
      "epoch": 0.4568690313314683,
      "grad_norm": 0.7122592024847274,
      "learning_rate": 0.00018004955990035963,
      "loss": 12.6155,
      "step": 8390
    },
    {
      "epoch": 0.45692348532805127,
      "grad_norm": 0.8861540540847268,
      "learning_rate": 0.00018004427446634104,
      "loss": 12.4889,
      "step": 8391
    },
    {
      "epoch": 0.4569779393246343,
      "grad_norm": 0.5764384661397526,
      "learning_rate": 0.0001800389884098848,
      "loss": 12.3368,
      "step": 8392
    },
    {
      "epoch": 0.4570323933212173,
      "grad_norm": 0.6463869931658384,
      "learning_rate": 0.00018003370173103194,
      "loss": 12.4053,
      "step": 8393
    },
    {
      "epoch": 0.4570868473178003,
      "grad_norm": 0.7557404332112136,
      "learning_rate": 0.0001800284144298236,
      "loss": 12.4105,
      "step": 8394
    },
    {
      "epoch": 0.45714130131438335,
      "grad_norm": 0.6785861968861243,
      "learning_rate": 0.00018002312650630087,
      "loss": 12.5533,
      "step": 8395
    },
    {
      "epoch": 0.45719575531096635,
      "grad_norm": 0.5997132875084884,
      "learning_rate": 0.0001800178379605049,
      "loss": 12.3683,
      "step": 8396
    },
    {
      "epoch": 0.4572502093075494,
      "grad_norm": 0.6723666128208018,
      "learning_rate": 0.0001800125487924768,
      "loss": 12.4165,
      "step": 8397
    },
    {
      "epoch": 0.4573046633041324,
      "grad_norm": 0.604105015372839,
      "learning_rate": 0.00018000725900225767,
      "loss": 12.3035,
      "step": 8398
    },
    {
      "epoch": 0.4573591173007154,
      "grad_norm": 0.6976708969027182,
      "learning_rate": 0.00018000196858988874,
      "loss": 12.4663,
      "step": 8399
    },
    {
      "epoch": 0.4574135712972984,
      "grad_norm": 0.6360785426257929,
      "learning_rate": 0.00017999667755541106,
      "loss": 12.3827,
      "step": 8400
    },
    {
      "epoch": 0.4574680252938814,
      "grad_norm": 0.750727146607152,
      "learning_rate": 0.00017999138589886576,
      "loss": 12.4395,
      "step": 8401
    },
    {
      "epoch": 0.4575224792904644,
      "grad_norm": 0.6165410796762938,
      "learning_rate": 0.0001799860936202941,
      "loss": 12.4172,
      "step": 8402
    },
    {
      "epoch": 0.45757693328704746,
      "grad_norm": 0.664655318925348,
      "learning_rate": 0.00017998080071973712,
      "loss": 12.4702,
      "step": 8403
    },
    {
      "epoch": 0.45763138728363045,
      "grad_norm": 0.6866522603182079,
      "learning_rate": 0.00017997550719723603,
      "loss": 12.3881,
      "step": 8404
    },
    {
      "epoch": 0.45768584128021345,
      "grad_norm": 0.6207786327571557,
      "learning_rate": 0.000179970213052832,
      "loss": 12.5766,
      "step": 8405
    },
    {
      "epoch": 0.4577402952767965,
      "grad_norm": 0.6271054032970457,
      "learning_rate": 0.00017996491828656613,
      "loss": 12.4155,
      "step": 8406
    },
    {
      "epoch": 0.4577947492733795,
      "grad_norm": 0.788733076673143,
      "learning_rate": 0.00017995962289847972,
      "loss": 12.4865,
      "step": 8407
    },
    {
      "epoch": 0.4578492032699625,
      "grad_norm": 0.6451839502353852,
      "learning_rate": 0.00017995432688861383,
      "loss": 12.3857,
      "step": 8408
    },
    {
      "epoch": 0.4579036572665455,
      "grad_norm": 0.7922111769987087,
      "learning_rate": 0.00017994903025700967,
      "loss": 12.3343,
      "step": 8409
    },
    {
      "epoch": 0.4579581112631285,
      "grad_norm": 0.6291459222045038,
      "learning_rate": 0.00017994373300370847,
      "loss": 12.3404,
      "step": 8410
    },
    {
      "epoch": 0.4580125652597115,
      "grad_norm": 0.6257345055875672,
      "learning_rate": 0.0001799384351287514,
      "loss": 12.4469,
      "step": 8411
    },
    {
      "epoch": 0.45806701925629456,
      "grad_norm": 0.6765850397057749,
      "learning_rate": 0.00017993313663217962,
      "loss": 12.3578,
      "step": 8412
    },
    {
      "epoch": 0.45812147325287755,
      "grad_norm": 0.5731249213737087,
      "learning_rate": 0.0001799278375140344,
      "loss": 12.4651,
      "step": 8413
    },
    {
      "epoch": 0.45817592724946055,
      "grad_norm": 0.7520989664714446,
      "learning_rate": 0.00017992253777435687,
      "loss": 12.4347,
      "step": 8414
    },
    {
      "epoch": 0.4582303812460436,
      "grad_norm": 0.6548601863439822,
      "learning_rate": 0.00017991723741318834,
      "loss": 12.4428,
      "step": 8415
    },
    {
      "epoch": 0.4582848352426266,
      "grad_norm": 0.6742770381298215,
      "learning_rate": 0.0001799119364305699,
      "loss": 12.5476,
      "step": 8416
    },
    {
      "epoch": 0.4583392892392096,
      "grad_norm": 0.6649821067512534,
      "learning_rate": 0.00017990663482654292,
      "loss": 12.4915,
      "step": 8417
    },
    {
      "epoch": 0.45839374323579263,
      "grad_norm": 0.7172279380276158,
      "learning_rate": 0.0001799013326011485,
      "loss": 12.5574,
      "step": 8418
    },
    {
      "epoch": 0.4584481972323756,
      "grad_norm": 0.7547962084995847,
      "learning_rate": 0.00017989602975442793,
      "loss": 12.4675,
      "step": 8419
    },
    {
      "epoch": 0.4585026512289586,
      "grad_norm": 0.5859887915230573,
      "learning_rate": 0.00017989072628642243,
      "loss": 12.4824,
      "step": 8420
    },
    {
      "epoch": 0.45855710522554166,
      "grad_norm": 0.7088704213932157,
      "learning_rate": 0.00017988542219717326,
      "loss": 12.6085,
      "step": 8421
    },
    {
      "epoch": 0.45861155922212465,
      "grad_norm": 0.6530669807244944,
      "learning_rate": 0.00017988011748672162,
      "loss": 12.5474,
      "step": 8422
    },
    {
      "epoch": 0.45866601321870765,
      "grad_norm": 0.6012504516424516,
      "learning_rate": 0.00017987481215510882,
      "loss": 12.4466,
      "step": 8423
    },
    {
      "epoch": 0.4587204672152907,
      "grad_norm": 0.6321909432910274,
      "learning_rate": 0.00017986950620237609,
      "loss": 12.4394,
      "step": 8424
    },
    {
      "epoch": 0.4587749212118737,
      "grad_norm": 0.5955062332721126,
      "learning_rate": 0.00017986419962856464,
      "loss": 12.4106,
      "step": 8425
    },
    {
      "epoch": 0.4588293752084567,
      "grad_norm": 0.5834126337825684,
      "learning_rate": 0.00017985889243371582,
      "loss": 12.3952,
      "step": 8426
    },
    {
      "epoch": 0.45888382920503973,
      "grad_norm": 0.6306202539060862,
      "learning_rate": 0.00017985358461787082,
      "loss": 12.4875,
      "step": 8427
    },
    {
      "epoch": 0.4589382832016227,
      "grad_norm": 0.6098987214034494,
      "learning_rate": 0.00017984827618107102,
      "loss": 12.5203,
      "step": 8428
    },
    {
      "epoch": 0.4589927371982057,
      "grad_norm": 0.6298011928808441,
      "learning_rate": 0.00017984296712335757,
      "loss": 12.4491,
      "step": 8429
    },
    {
      "epoch": 0.45904719119478876,
      "grad_norm": 0.6316231234434623,
      "learning_rate": 0.00017983765744477185,
      "loss": 12.3603,
      "step": 8430
    },
    {
      "epoch": 0.45910164519137175,
      "grad_norm": 0.621224439891547,
      "learning_rate": 0.0001798323471453551,
      "loss": 12.5129,
      "step": 8431
    },
    {
      "epoch": 0.4591560991879548,
      "grad_norm": 0.6215508990689357,
      "learning_rate": 0.00017982703622514867,
      "loss": 12.4447,
      "step": 8432
    },
    {
      "epoch": 0.4592105531845378,
      "grad_norm": 0.6373085894914177,
      "learning_rate": 0.00017982172468419377,
      "loss": 12.3633,
      "step": 8433
    },
    {
      "epoch": 0.4592650071811208,
      "grad_norm": 0.5999470984578853,
      "learning_rate": 0.00017981641252253177,
      "loss": 12.3336,
      "step": 8434
    },
    {
      "epoch": 0.45931946117770384,
      "grad_norm": 0.6530057898257308,
      "learning_rate": 0.000179811099740204,
      "loss": 12.5767,
      "step": 8435
    },
    {
      "epoch": 0.45937391517428683,
      "grad_norm": 0.6462165261279232,
      "learning_rate": 0.00017980578633725166,
      "loss": 12.4114,
      "step": 8436
    },
    {
      "epoch": 0.4594283691708698,
      "grad_norm": 0.6665737763342021,
      "learning_rate": 0.0001798004723137162,
      "loss": 12.4862,
      "step": 8437
    },
    {
      "epoch": 0.45948282316745287,
      "grad_norm": 0.5854356255683455,
      "learning_rate": 0.00017979515766963888,
      "loss": 12.3757,
      "step": 8438
    },
    {
      "epoch": 0.45953727716403586,
      "grad_norm": 0.6429628084102862,
      "learning_rate": 0.00017978984240506101,
      "loss": 12.4527,
      "step": 8439
    },
    {
      "epoch": 0.45959173116061885,
      "grad_norm": 0.6224284086182738,
      "learning_rate": 0.00017978452652002392,
      "loss": 12.3866,
      "step": 8440
    },
    {
      "epoch": 0.4596461851572019,
      "grad_norm": 0.7085312267599722,
      "learning_rate": 0.00017977921001456902,
      "loss": 12.5016,
      "step": 8441
    },
    {
      "epoch": 0.4597006391537849,
      "grad_norm": 0.6625040975468844,
      "learning_rate": 0.0001797738928887376,
      "loss": 12.5766,
      "step": 8442
    },
    {
      "epoch": 0.4597550931503679,
      "grad_norm": 0.7112702932006277,
      "learning_rate": 0.000179768575142571,
      "loss": 12.4637,
      "step": 8443
    },
    {
      "epoch": 0.45980954714695094,
      "grad_norm": 0.6543475319952922,
      "learning_rate": 0.0001797632567761106,
      "loss": 12.4763,
      "step": 8444
    },
    {
      "epoch": 0.45986400114353393,
      "grad_norm": 0.6561221437463571,
      "learning_rate": 0.00017975793778939768,
      "loss": 12.2698,
      "step": 8445
    },
    {
      "epoch": 0.4599184551401169,
      "grad_norm": 0.6304476519096577,
      "learning_rate": 0.00017975261818247373,
      "loss": 12.4731,
      "step": 8446
    },
    {
      "epoch": 0.45997290913669997,
      "grad_norm": 0.6737388668545593,
      "learning_rate": 0.00017974729795538,
      "loss": 12.5866,
      "step": 8447
    },
    {
      "epoch": 0.46002736313328296,
      "grad_norm": 0.64445630023729,
      "learning_rate": 0.00017974197710815795,
      "loss": 12.3961,
      "step": 8448
    },
    {
      "epoch": 0.46008181712986596,
      "grad_norm": 0.570660478325581,
      "learning_rate": 0.0001797366556408489,
      "loss": 12.5089,
      "step": 8449
    },
    {
      "epoch": 0.460136271126449,
      "grad_norm": 0.7022380367884817,
      "learning_rate": 0.0001797313335534942,
      "loss": 12.3741,
      "step": 8450
    },
    {
      "epoch": 0.460190725123032,
      "grad_norm": 0.612890241562397,
      "learning_rate": 0.00017972601084613533,
      "loss": 12.4199,
      "step": 8451
    },
    {
      "epoch": 0.460245179119615,
      "grad_norm": 0.6156966291816607,
      "learning_rate": 0.0001797206875188136,
      "loss": 12.5195,
      "step": 8452
    },
    {
      "epoch": 0.46029963311619804,
      "grad_norm": 0.6462544029384523,
      "learning_rate": 0.00017971536357157044,
      "loss": 12.5192,
      "step": 8453
    },
    {
      "epoch": 0.46035408711278103,
      "grad_norm": 0.6231812162951743,
      "learning_rate": 0.00017971003900444727,
      "loss": 12.424,
      "step": 8454
    },
    {
      "epoch": 0.460408541109364,
      "grad_norm": 0.6855647949717323,
      "learning_rate": 0.00017970471381748544,
      "loss": 12.4334,
      "step": 8455
    },
    {
      "epoch": 0.46046299510594707,
      "grad_norm": 0.6311800182849413,
      "learning_rate": 0.0001796993880107264,
      "loss": 12.3238,
      "step": 8456
    },
    {
      "epoch": 0.46051744910253006,
      "grad_norm": 0.6337541676749251,
      "learning_rate": 0.00017969406158421157,
      "loss": 12.4585,
      "step": 8457
    },
    {
      "epoch": 0.46057190309911306,
      "grad_norm": 0.6208653234499459,
      "learning_rate": 0.0001796887345379823,
      "loss": 12.4279,
      "step": 8458
    },
    {
      "epoch": 0.4606263570956961,
      "grad_norm": 0.6330536996832912,
      "learning_rate": 0.00017968340687208012,
      "loss": 12.4568,
      "step": 8459
    },
    {
      "epoch": 0.4606808110922791,
      "grad_norm": 0.6090976625280496,
      "learning_rate": 0.0001796780785865464,
      "loss": 12.3771,
      "step": 8460
    },
    {
      "epoch": 0.4607352650888621,
      "grad_norm": 0.65006715249651,
      "learning_rate": 0.00017967274968142257,
      "loss": 12.5136,
      "step": 8461
    },
    {
      "epoch": 0.46078971908544514,
      "grad_norm": 0.6091210194797642,
      "learning_rate": 0.00017966742015675008,
      "loss": 12.4713,
      "step": 8462
    },
    {
      "epoch": 0.46084417308202813,
      "grad_norm": 0.5394330605868857,
      "learning_rate": 0.00017966209001257032,
      "loss": 12.3411,
      "step": 8463
    },
    {
      "epoch": 0.4608986270786112,
      "grad_norm": 0.6299230807970284,
      "learning_rate": 0.00017965675924892484,
      "loss": 12.3845,
      "step": 8464
    },
    {
      "epoch": 0.46095308107519417,
      "grad_norm": 0.6004734955181203,
      "learning_rate": 0.00017965142786585504,
      "loss": 12.2142,
      "step": 8465
    },
    {
      "epoch": 0.46100753507177716,
      "grad_norm": 0.6013644648463967,
      "learning_rate": 0.0001796460958634024,
      "loss": 12.4469,
      "step": 8466
    },
    {
      "epoch": 0.4610619890683602,
      "grad_norm": 0.6540201882122875,
      "learning_rate": 0.00017964076324160832,
      "loss": 12.4277,
      "step": 8467
    },
    {
      "epoch": 0.4611164430649432,
      "grad_norm": 0.6842417699779186,
      "learning_rate": 0.00017963543000051432,
      "loss": 12.4336,
      "step": 8468
    },
    {
      "epoch": 0.4611708970615262,
      "grad_norm": 0.6259991019467918,
      "learning_rate": 0.00017963009614016187,
      "loss": 12.4175,
      "step": 8469
    },
    {
      "epoch": 0.46122535105810925,
      "grad_norm": 0.7977560619398021,
      "learning_rate": 0.00017962476166059243,
      "loss": 12.6586,
      "step": 8470
    },
    {
      "epoch": 0.46127980505469224,
      "grad_norm": 0.7040812223013768,
      "learning_rate": 0.0001796194265618475,
      "loss": 12.4542,
      "step": 8471
    },
    {
      "epoch": 0.46133425905127523,
      "grad_norm": 0.6170972449417943,
      "learning_rate": 0.00017961409084396856,
      "loss": 12.2917,
      "step": 8472
    },
    {
      "epoch": 0.4613887130478583,
      "grad_norm": 0.5628385005146082,
      "learning_rate": 0.00017960875450699707,
      "loss": 12.4389,
      "step": 8473
    },
    {
      "epoch": 0.46144316704444127,
      "grad_norm": 0.7106466298776539,
      "learning_rate": 0.00017960341755097459,
      "loss": 12.4437,
      "step": 8474
    },
    {
      "epoch": 0.46149762104102426,
      "grad_norm": 0.6325480250324459,
      "learning_rate": 0.00017959807997594256,
      "loss": 12.3271,
      "step": 8475
    },
    {
      "epoch": 0.4615520750376073,
      "grad_norm": 0.6688376322815393,
      "learning_rate": 0.00017959274178194252,
      "loss": 12.5363,
      "step": 8476
    },
    {
      "epoch": 0.4616065290341903,
      "grad_norm": 0.6390089606861243,
      "learning_rate": 0.00017958740296901597,
      "loss": 12.5253,
      "step": 8477
    },
    {
      "epoch": 0.4616609830307733,
      "grad_norm": 0.6483652623580444,
      "learning_rate": 0.00017958206353720443,
      "loss": 12.3701,
      "step": 8478
    },
    {
      "epoch": 0.46171543702735635,
      "grad_norm": 0.5747024925879572,
      "learning_rate": 0.00017957672348654943,
      "loss": 12.3712,
      "step": 8479
    },
    {
      "epoch": 0.46176989102393934,
      "grad_norm": 0.6407691144277222,
      "learning_rate": 0.00017957138281709246,
      "loss": 12.5068,
      "step": 8480
    },
    {
      "epoch": 0.46182434502052233,
      "grad_norm": 0.5804922627948418,
      "learning_rate": 0.00017956604152887507,
      "loss": 12.4644,
      "step": 8481
    },
    {
      "epoch": 0.4618787990171054,
      "grad_norm": 0.7734467267607033,
      "learning_rate": 0.00017956069962193886,
      "loss": 12.5521,
      "step": 8482
    },
    {
      "epoch": 0.46193325301368837,
      "grad_norm": 0.5895896860218451,
      "learning_rate": 0.00017955535709632522,
      "loss": 12.441,
      "step": 8483
    },
    {
      "epoch": 0.46198770701027136,
      "grad_norm": 0.7151105273525564,
      "learning_rate": 0.00017955001395207585,
      "loss": 12.3842,
      "step": 8484
    },
    {
      "epoch": 0.4620421610068544,
      "grad_norm": 0.605451804966123,
      "learning_rate": 0.0001795446701892322,
      "loss": 12.3717,
      "step": 8485
    },
    {
      "epoch": 0.4620966150034374,
      "grad_norm": 0.6796432599907761,
      "learning_rate": 0.00017953932580783586,
      "loss": 12.5606,
      "step": 8486
    },
    {
      "epoch": 0.4621510690000204,
      "grad_norm": 0.6543516269832156,
      "learning_rate": 0.00017953398080792837,
      "loss": 12.3228,
      "step": 8487
    },
    {
      "epoch": 0.46220552299660345,
      "grad_norm": 0.7231693338943618,
      "learning_rate": 0.00017952863518955133,
      "loss": 12.561,
      "step": 8488
    },
    {
      "epoch": 0.46225997699318644,
      "grad_norm": 0.6967078738341969,
      "learning_rate": 0.0001795232889527463,
      "loss": 12.4842,
      "step": 8489
    },
    {
      "epoch": 0.46231443098976943,
      "grad_norm": 0.6900231716925654,
      "learning_rate": 0.0001795179420975548,
      "loss": 12.5378,
      "step": 8490
    },
    {
      "epoch": 0.4623688849863525,
      "grad_norm": 0.6565505110739019,
      "learning_rate": 0.00017951259462401845,
      "loss": 12.3519,
      "step": 8491
    },
    {
      "epoch": 0.4624233389829355,
      "grad_norm": 0.5804690618308558,
      "learning_rate": 0.00017950724653217884,
      "loss": 12.4108,
      "step": 8492
    },
    {
      "epoch": 0.46247779297951846,
      "grad_norm": 0.695013601610202,
      "learning_rate": 0.00017950189782207755,
      "loss": 12.4763,
      "step": 8493
    },
    {
      "epoch": 0.4625322469761015,
      "grad_norm": 0.6476402393977978,
      "learning_rate": 0.00017949654849375616,
      "loss": 12.5224,
      "step": 8494
    },
    {
      "epoch": 0.4625867009726845,
      "grad_norm": 0.650412323007607,
      "learning_rate": 0.00017949119854725632,
      "loss": 12.5714,
      "step": 8495
    },
    {
      "epoch": 0.4626411549692675,
      "grad_norm": 0.6177719603875415,
      "learning_rate": 0.00017948584798261954,
      "loss": 12.4933,
      "step": 8496
    },
    {
      "epoch": 0.46269560896585055,
      "grad_norm": 0.6140868763191671,
      "learning_rate": 0.0001794804967998875,
      "loss": 12.4619,
      "step": 8497
    },
    {
      "epoch": 0.46275006296243354,
      "grad_norm": 0.6042822927413228,
      "learning_rate": 0.00017947514499910177,
      "loss": 12.3968,
      "step": 8498
    },
    {
      "epoch": 0.4628045169590166,
      "grad_norm": 0.614738493564977,
      "learning_rate": 0.00017946979258030399,
      "loss": 12.4828,
      "step": 8499
    },
    {
      "epoch": 0.4628589709555996,
      "grad_norm": 0.5737063644687317,
      "learning_rate": 0.00017946443954353577,
      "loss": 12.3964,
      "step": 8500
    },
    {
      "epoch": 0.4629134249521826,
      "grad_norm": 0.6287436734117607,
      "learning_rate": 0.00017945908588883877,
      "loss": 12.5085,
      "step": 8501
    },
    {
      "epoch": 0.4629678789487656,
      "grad_norm": 0.6167967595743099,
      "learning_rate": 0.00017945373161625455,
      "loss": 12.3852,
      "step": 8502
    },
    {
      "epoch": 0.4630223329453486,
      "grad_norm": 0.6064207477081835,
      "learning_rate": 0.00017944837672582485,
      "loss": 12.3263,
      "step": 8503
    },
    {
      "epoch": 0.4630767869419316,
      "grad_norm": 0.6333852371143583,
      "learning_rate": 0.0001794430212175912,
      "loss": 12.3366,
      "step": 8504
    },
    {
      "epoch": 0.46313124093851465,
      "grad_norm": 0.6246520185592767,
      "learning_rate": 0.0001794376650915953,
      "loss": 12.3339,
      "step": 8505
    },
    {
      "epoch": 0.46318569493509765,
      "grad_norm": 0.6553154091068601,
      "learning_rate": 0.00017943230834787882,
      "loss": 12.3773,
      "step": 8506
    },
    {
      "epoch": 0.46324014893168064,
      "grad_norm": 0.6780757185364424,
      "learning_rate": 0.00017942695098648335,
      "loss": 12.4337,
      "step": 8507
    },
    {
      "epoch": 0.4632946029282637,
      "grad_norm": 0.5992151156206015,
      "learning_rate": 0.00017942159300745063,
      "loss": 12.4676,
      "step": 8508
    },
    {
      "epoch": 0.4633490569248467,
      "grad_norm": 0.7139367597885157,
      "learning_rate": 0.00017941623441082225,
      "loss": 12.4756,
      "step": 8509
    },
    {
      "epoch": 0.4634035109214297,
      "grad_norm": 0.6074562947215172,
      "learning_rate": 0.00017941087519663995,
      "loss": 12.3758,
      "step": 8510
    },
    {
      "epoch": 0.4634579649180127,
      "grad_norm": 0.5919073671557712,
      "learning_rate": 0.00017940551536494535,
      "loss": 12.4662,
      "step": 8511
    },
    {
      "epoch": 0.4635124189145957,
      "grad_norm": 0.6288349721285509,
      "learning_rate": 0.00017940015491578012,
      "loss": 12.4327,
      "step": 8512
    },
    {
      "epoch": 0.4635668729111787,
      "grad_norm": 0.6377686064725293,
      "learning_rate": 0.000179394793849186,
      "loss": 12.4457,
      "step": 8513
    },
    {
      "epoch": 0.46362132690776175,
      "grad_norm": 0.6331751229664899,
      "learning_rate": 0.00017938943216520462,
      "loss": 12.4591,
      "step": 8514
    },
    {
      "epoch": 0.46367578090434475,
      "grad_norm": 0.6205814254067495,
      "learning_rate": 0.00017938406986387775,
      "loss": 12.4356,
      "step": 8515
    },
    {
      "epoch": 0.46373023490092774,
      "grad_norm": 0.6120530025858775,
      "learning_rate": 0.000179378706945247,
      "loss": 12.465,
      "step": 8516
    },
    {
      "epoch": 0.4637846888975108,
      "grad_norm": 0.6576904060379765,
      "learning_rate": 0.00017937334340935413,
      "loss": 12.3719,
      "step": 8517
    },
    {
      "epoch": 0.4638391428940938,
      "grad_norm": 0.6302025635351302,
      "learning_rate": 0.0001793679792562408,
      "loss": 12.3565,
      "step": 8518
    },
    {
      "epoch": 0.4638935968906768,
      "grad_norm": 0.6226271009698168,
      "learning_rate": 0.00017936261448594877,
      "loss": 12.476,
      "step": 8519
    },
    {
      "epoch": 0.4639480508872598,
      "grad_norm": 0.7097366292174755,
      "learning_rate": 0.00017935724909851978,
      "loss": 12.4391,
      "step": 8520
    },
    {
      "epoch": 0.4640025048838428,
      "grad_norm": 0.6343089219332183,
      "learning_rate": 0.00017935188309399546,
      "loss": 12.411,
      "step": 8521
    },
    {
      "epoch": 0.4640569588804258,
      "grad_norm": 0.6863122806284503,
      "learning_rate": 0.00017934651647241764,
      "loss": 12.5285,
      "step": 8522
    },
    {
      "epoch": 0.46411141287700886,
      "grad_norm": 0.6400060317241383,
      "learning_rate": 0.00017934114923382798,
      "loss": 12.4462,
      "step": 8523
    },
    {
      "epoch": 0.46416586687359185,
      "grad_norm": 0.614577460142748,
      "learning_rate": 0.00017933578137826822,
      "loss": 12.4177,
      "step": 8524
    },
    {
      "epoch": 0.46422032087017484,
      "grad_norm": 0.6379323169224547,
      "learning_rate": 0.00017933041290578013,
      "loss": 12.45,
      "step": 8525
    },
    {
      "epoch": 0.4642747748667579,
      "grad_norm": 0.7842891853867016,
      "learning_rate": 0.00017932504381640545,
      "loss": 12.3224,
      "step": 8526
    },
    {
      "epoch": 0.4643292288633409,
      "grad_norm": 0.6253337426516982,
      "learning_rate": 0.0001793196741101859,
      "loss": 12.4774,
      "step": 8527
    },
    {
      "epoch": 0.4643836828599239,
      "grad_norm": 0.7597331992933967,
      "learning_rate": 0.00017931430378716328,
      "loss": 12.5435,
      "step": 8528
    },
    {
      "epoch": 0.4644381368565069,
      "grad_norm": 0.614380364429936,
      "learning_rate": 0.00017930893284737932,
      "loss": 12.3992,
      "step": 8529
    },
    {
      "epoch": 0.4644925908530899,
      "grad_norm": 0.6301733598215885,
      "learning_rate": 0.00017930356129087585,
      "loss": 12.4247,
      "step": 8530
    },
    {
      "epoch": 0.46454704484967296,
      "grad_norm": 0.7021075766021644,
      "learning_rate": 0.00017929818911769453,
      "loss": 12.5615,
      "step": 8531
    },
    {
      "epoch": 0.46460149884625596,
      "grad_norm": 0.6635152504671655,
      "learning_rate": 0.00017929281632787723,
      "loss": 12.4488,
      "step": 8532
    },
    {
      "epoch": 0.46465595284283895,
      "grad_norm": 0.6959204379962824,
      "learning_rate": 0.00017928744292146568,
      "loss": 12.3817,
      "step": 8533
    },
    {
      "epoch": 0.464710406839422,
      "grad_norm": 0.5937204435815098,
      "learning_rate": 0.00017928206889850169,
      "loss": 12.4555,
      "step": 8534
    },
    {
      "epoch": 0.464764860836005,
      "grad_norm": 0.7572250430526706,
      "learning_rate": 0.00017927669425902703,
      "loss": 12.4628,
      "step": 8535
    },
    {
      "epoch": 0.464819314832588,
      "grad_norm": 0.6405099518133247,
      "learning_rate": 0.00017927131900308347,
      "loss": 12.3633,
      "step": 8536
    },
    {
      "epoch": 0.46487376882917103,
      "grad_norm": 0.6410726229961785,
      "learning_rate": 0.0001792659431307129,
      "loss": 12.4437,
      "step": 8537
    },
    {
      "epoch": 0.464928222825754,
      "grad_norm": 0.6256073298072141,
      "learning_rate": 0.000179260566641957,
      "loss": 12.3884,
      "step": 8538
    },
    {
      "epoch": 0.464982676822337,
      "grad_norm": 0.620457959466047,
      "learning_rate": 0.00017925518953685766,
      "loss": 12.3345,
      "step": 8539
    },
    {
      "epoch": 0.46503713081892006,
      "grad_norm": 0.6852978455402273,
      "learning_rate": 0.0001792498118154567,
      "loss": 12.5612,
      "step": 8540
    },
    {
      "epoch": 0.46509158481550306,
      "grad_norm": 0.6724753812631421,
      "learning_rate": 0.0001792444334777959,
      "loss": 12.4961,
      "step": 8541
    },
    {
      "epoch": 0.46514603881208605,
      "grad_norm": 0.6879903993263202,
      "learning_rate": 0.0001792390545239171,
      "loss": 12.5168,
      "step": 8542
    },
    {
      "epoch": 0.4652004928086691,
      "grad_norm": 0.5520030988039244,
      "learning_rate": 0.0001792336749538621,
      "loss": 12.4215,
      "step": 8543
    },
    {
      "epoch": 0.4652549468052521,
      "grad_norm": 0.5474435428367739,
      "learning_rate": 0.00017922829476767278,
      "loss": 12.3678,
      "step": 8544
    },
    {
      "epoch": 0.4653094008018351,
      "grad_norm": 0.7329959158129233,
      "learning_rate": 0.00017922291396539093,
      "loss": 12.3445,
      "step": 8545
    },
    {
      "epoch": 0.46536385479841813,
      "grad_norm": 0.6512212767290448,
      "learning_rate": 0.00017921753254705844,
      "loss": 12.5902,
      "step": 8546
    },
    {
      "epoch": 0.4654183087950011,
      "grad_norm": 0.5896209592313086,
      "learning_rate": 0.0001792121505127171,
      "loss": 12.4124,
      "step": 8547
    },
    {
      "epoch": 0.4654727627915841,
      "grad_norm": 0.5841614441736784,
      "learning_rate": 0.00017920676786240883,
      "loss": 12.3758,
      "step": 8548
    },
    {
      "epoch": 0.46552721678816716,
      "grad_norm": 0.5898442729547501,
      "learning_rate": 0.00017920138459617544,
      "loss": 12.4799,
      "step": 8549
    },
    {
      "epoch": 0.46558167078475016,
      "grad_norm": 0.5748459449326601,
      "learning_rate": 0.00017919600071405881,
      "loss": 12.3012,
      "step": 8550
    },
    {
      "epoch": 0.46563612478133315,
      "grad_norm": 0.608117050757133,
      "learning_rate": 0.0001791906162161008,
      "loss": 12.4454,
      "step": 8551
    },
    {
      "epoch": 0.4656905787779162,
      "grad_norm": 0.6672970333611582,
      "learning_rate": 0.00017918523110234324,
      "loss": 12.3224,
      "step": 8552
    },
    {
      "epoch": 0.4657450327744992,
      "grad_norm": 0.7520661230221706,
      "learning_rate": 0.00017917984537282807,
      "loss": 12.354,
      "step": 8553
    },
    {
      "epoch": 0.4657994867710822,
      "grad_norm": 0.6302720683788045,
      "learning_rate": 0.00017917445902759714,
      "loss": 12.3604,
      "step": 8554
    },
    {
      "epoch": 0.46585394076766523,
      "grad_norm": 0.6579941205866369,
      "learning_rate": 0.00017916907206669237,
      "loss": 12.5587,
      "step": 8555
    },
    {
      "epoch": 0.4659083947642482,
      "grad_norm": 0.6165873606250319,
      "learning_rate": 0.00017916368449015556,
      "loss": 12.431,
      "step": 8556
    },
    {
      "epoch": 0.4659628487608312,
      "grad_norm": 0.7295377109985214,
      "learning_rate": 0.0001791582962980287,
      "loss": 12.3715,
      "step": 8557
    },
    {
      "epoch": 0.46601730275741426,
      "grad_norm": 0.6578045921999068,
      "learning_rate": 0.00017915290749035364,
      "loss": 12.3832,
      "step": 8558
    },
    {
      "epoch": 0.46607175675399726,
      "grad_norm": 0.6093581051436394,
      "learning_rate": 0.0001791475180671723,
      "loss": 12.5183,
      "step": 8559
    },
    {
      "epoch": 0.46612621075058025,
      "grad_norm": 0.6087113495662672,
      "learning_rate": 0.0001791421280285266,
      "loss": 12.5348,
      "step": 8560
    },
    {
      "epoch": 0.4661806647471633,
      "grad_norm": 0.6808914595686044,
      "learning_rate": 0.00017913673737445844,
      "loss": 12.5032,
      "step": 8561
    },
    {
      "epoch": 0.4662351187437463,
      "grad_norm": 0.6960835995102816,
      "learning_rate": 0.0001791313461050097,
      "loss": 12.5811,
      "step": 8562
    },
    {
      "epoch": 0.4662895727403293,
      "grad_norm": 0.7637939174979597,
      "learning_rate": 0.0001791259542202224,
      "loss": 12.4642,
      "step": 8563
    },
    {
      "epoch": 0.46634402673691233,
      "grad_norm": 0.6495379560707983,
      "learning_rate": 0.00017912056172013837,
      "loss": 12.4818,
      "step": 8564
    },
    {
      "epoch": 0.4663984807334953,
      "grad_norm": 0.644911863272382,
      "learning_rate": 0.0001791151686047996,
      "loss": 12.497,
      "step": 8565
    },
    {
      "epoch": 0.4664529347300784,
      "grad_norm": 0.7269018585601411,
      "learning_rate": 0.00017910977487424801,
      "loss": 12.5889,
      "step": 8566
    },
    {
      "epoch": 0.46650738872666137,
      "grad_norm": 0.5786874526945133,
      "learning_rate": 0.00017910438052852557,
      "loss": 12.2868,
      "step": 8567
    },
    {
      "epoch": 0.46656184272324436,
      "grad_norm": 0.6627592741179691,
      "learning_rate": 0.00017909898556767414,
      "loss": 12.4585,
      "step": 8568
    },
    {
      "epoch": 0.4666162967198274,
      "grad_norm": 0.693478438138713,
      "learning_rate": 0.00017909358999173577,
      "loss": 12.3521,
      "step": 8569
    },
    {
      "epoch": 0.4666707507164104,
      "grad_norm": 0.6801219655549972,
      "learning_rate": 0.0001790881938007524,
      "loss": 12.4614,
      "step": 8570
    },
    {
      "epoch": 0.4667252047129934,
      "grad_norm": 0.7301463752425134,
      "learning_rate": 0.00017908279699476592,
      "loss": 12.3549,
      "step": 8571
    },
    {
      "epoch": 0.46677965870957644,
      "grad_norm": 0.6346179560399473,
      "learning_rate": 0.00017907739957381838,
      "loss": 12.3989,
      "step": 8572
    },
    {
      "epoch": 0.46683411270615943,
      "grad_norm": 0.686851017492308,
      "learning_rate": 0.00017907200153795171,
      "loss": 12.3688,
      "step": 8573
    },
    {
      "epoch": 0.4668885667027424,
      "grad_norm": 0.6605494864320264,
      "learning_rate": 0.0001790666028872079,
      "loss": 12.4429,
      "step": 8574
    },
    {
      "epoch": 0.4669430206993255,
      "grad_norm": 0.7145770417283764,
      "learning_rate": 0.00017906120362162894,
      "loss": 12.3289,
      "step": 8575
    },
    {
      "epoch": 0.46699747469590847,
      "grad_norm": 0.6897110873551012,
      "learning_rate": 0.00017905580374125678,
      "loss": 12.4011,
      "step": 8576
    },
    {
      "epoch": 0.46705192869249146,
      "grad_norm": 0.7216304975696103,
      "learning_rate": 0.00017905040324613344,
      "loss": 12.472,
      "step": 8577
    },
    {
      "epoch": 0.4671063826890745,
      "grad_norm": 0.6970788524101713,
      "learning_rate": 0.00017904500213630092,
      "loss": 12.507,
      "step": 8578
    },
    {
      "epoch": 0.4671608366856575,
      "grad_norm": 0.631846296329978,
      "learning_rate": 0.0001790396004118012,
      "loss": 12.3616,
      "step": 8579
    },
    {
      "epoch": 0.4672152906822405,
      "grad_norm": 0.6871599189413196,
      "learning_rate": 0.0001790341980726763,
      "loss": 12.3542,
      "step": 8580
    },
    {
      "epoch": 0.46726974467882354,
      "grad_norm": 0.6227620922656812,
      "learning_rate": 0.0001790287951189682,
      "loss": 12.4218,
      "step": 8581
    },
    {
      "epoch": 0.46732419867540653,
      "grad_norm": 0.6327965646327595,
      "learning_rate": 0.00017902339155071896,
      "loss": 12.5839,
      "step": 8582
    },
    {
      "epoch": 0.4673786526719895,
      "grad_norm": 0.5943442388995044,
      "learning_rate": 0.00017901798736797054,
      "loss": 12.2484,
      "step": 8583
    },
    {
      "epoch": 0.4674331066685726,
      "grad_norm": 0.6691449075498682,
      "learning_rate": 0.00017901258257076503,
      "loss": 12.5749,
      "step": 8584
    },
    {
      "epoch": 0.46748756066515557,
      "grad_norm": 0.6469442060100173,
      "learning_rate": 0.00017900717715914444,
      "loss": 12.4268,
      "step": 8585
    },
    {
      "epoch": 0.46754201466173856,
      "grad_norm": 0.7164673653306719,
      "learning_rate": 0.00017900177113315075,
      "loss": 12.3637,
      "step": 8586
    },
    {
      "epoch": 0.4675964686583216,
      "grad_norm": 0.5329022400651572,
      "learning_rate": 0.00017899636449282606,
      "loss": 12.3726,
      "step": 8587
    },
    {
      "epoch": 0.4676509226549046,
      "grad_norm": 0.7268214005708902,
      "learning_rate": 0.0001789909572382124,
      "loss": 12.4756,
      "step": 8588
    },
    {
      "epoch": 0.4677053766514876,
      "grad_norm": 0.7438340959520051,
      "learning_rate": 0.0001789855493693518,
      "loss": 12.4556,
      "step": 8589
    },
    {
      "epoch": 0.46775983064807064,
      "grad_norm": 0.5711552451773573,
      "learning_rate": 0.00017898014088628633,
      "loss": 12.442,
      "step": 8590
    },
    {
      "epoch": 0.46781428464465363,
      "grad_norm": 0.760380130815662,
      "learning_rate": 0.00017897473178905805,
      "loss": 12.4894,
      "step": 8591
    },
    {
      "epoch": 0.4678687386412366,
      "grad_norm": 0.6237848382862727,
      "learning_rate": 0.00017896932207770895,
      "loss": 12.3818,
      "step": 8592
    },
    {
      "epoch": 0.4679231926378197,
      "grad_norm": 0.6191436468502493,
      "learning_rate": 0.00017896391175228123,
      "loss": 12.4273,
      "step": 8593
    },
    {
      "epoch": 0.46797764663440267,
      "grad_norm": 0.6690538751938576,
      "learning_rate": 0.00017895850081281687,
      "loss": 12.4312,
      "step": 8594
    },
    {
      "epoch": 0.46803210063098566,
      "grad_norm": 0.6807126771202414,
      "learning_rate": 0.00017895308925935794,
      "loss": 12.5251,
      "step": 8595
    },
    {
      "epoch": 0.4680865546275687,
      "grad_norm": 0.6639063805134303,
      "learning_rate": 0.00017894767709194658,
      "loss": 12.4868,
      "step": 8596
    },
    {
      "epoch": 0.4681410086241517,
      "grad_norm": 0.7908769442385336,
      "learning_rate": 0.00017894226431062481,
      "loss": 12.5073,
      "step": 8597
    },
    {
      "epoch": 0.46819546262073475,
      "grad_norm": 0.665810749723039,
      "learning_rate": 0.00017893685091543478,
      "loss": 12.502,
      "step": 8598
    },
    {
      "epoch": 0.46824991661731774,
      "grad_norm": 0.7095793243118013,
      "learning_rate": 0.00017893143690641855,
      "loss": 12.4091,
      "step": 8599
    },
    {
      "epoch": 0.46830437061390073,
      "grad_norm": 0.6582534562533506,
      "learning_rate": 0.00017892602228361824,
      "loss": 12.3877,
      "step": 8600
    },
    {
      "epoch": 0.4683588246104838,
      "grad_norm": 0.6099299580757082,
      "learning_rate": 0.0001789206070470759,
      "loss": 12.3216,
      "step": 8601
    },
    {
      "epoch": 0.4684132786070668,
      "grad_norm": 0.6673127073946039,
      "learning_rate": 0.00017891519119683376,
      "loss": 12.5892,
      "step": 8602
    },
    {
      "epoch": 0.46846773260364977,
      "grad_norm": 0.6467498612037779,
      "learning_rate": 0.0001789097747329338,
      "loss": 12.4686,
      "step": 8603
    },
    {
      "epoch": 0.4685221866002328,
      "grad_norm": 0.6261087762490277,
      "learning_rate": 0.00017890435765541821,
      "loss": 12.5421,
      "step": 8604
    },
    {
      "epoch": 0.4685766405968158,
      "grad_norm": 0.6264196217535764,
      "learning_rate": 0.00017889893996432914,
      "loss": 12.22,
      "step": 8605
    },
    {
      "epoch": 0.4686310945933988,
      "grad_norm": 0.6518593103011723,
      "learning_rate": 0.00017889352165970866,
      "loss": 12.4012,
      "step": 8606
    },
    {
      "epoch": 0.46868554858998185,
      "grad_norm": 0.5694633740086411,
      "learning_rate": 0.0001788881027415989,
      "loss": 12.45,
      "step": 8607
    },
    {
      "epoch": 0.46874000258656484,
      "grad_norm": 0.7154487883983486,
      "learning_rate": 0.00017888268321004203,
      "loss": 12.4673,
      "step": 8608
    },
    {
      "epoch": 0.46879445658314783,
      "grad_norm": 0.6391015557412403,
      "learning_rate": 0.00017887726306508022,
      "loss": 12.515,
      "step": 8609
    },
    {
      "epoch": 0.4688489105797309,
      "grad_norm": 0.6805762056289465,
      "learning_rate": 0.00017887184230675556,
      "loss": 12.6188,
      "step": 8610
    },
    {
      "epoch": 0.4689033645763139,
      "grad_norm": 0.6699372502313612,
      "learning_rate": 0.00017886642093511025,
      "loss": 12.5241,
      "step": 8611
    },
    {
      "epoch": 0.46895781857289687,
      "grad_norm": 0.6015103170519449,
      "learning_rate": 0.0001788609989501864,
      "loss": 12.4285,
      "step": 8612
    },
    {
      "epoch": 0.4690122725694799,
      "grad_norm": 0.6768216487927327,
      "learning_rate": 0.0001788555763520262,
      "loss": 12.3188,
      "step": 8613
    },
    {
      "epoch": 0.4690667265660629,
      "grad_norm": 0.6785633981073629,
      "learning_rate": 0.0001788501531406718,
      "loss": 12.3742,
      "step": 8614
    },
    {
      "epoch": 0.4691211805626459,
      "grad_norm": 0.6077029256627378,
      "learning_rate": 0.00017884472931616543,
      "loss": 12.3687,
      "step": 8615
    },
    {
      "epoch": 0.46917563455922895,
      "grad_norm": 0.6621690317050741,
      "learning_rate": 0.0001788393048785492,
      "loss": 12.3839,
      "step": 8616
    },
    {
      "epoch": 0.46923008855581194,
      "grad_norm": 0.6656474743344936,
      "learning_rate": 0.0001788338798278653,
      "loss": 12.3856,
      "step": 8617
    },
    {
      "epoch": 0.46928454255239493,
      "grad_norm": 0.66881792016721,
      "learning_rate": 0.00017882845416415595,
      "loss": 12.5676,
      "step": 8618
    },
    {
      "epoch": 0.469338996548978,
      "grad_norm": 0.767602290036045,
      "learning_rate": 0.0001788230278874633,
      "loss": 12.3283,
      "step": 8619
    },
    {
      "epoch": 0.469393450545561,
      "grad_norm": 0.7045073223901752,
      "learning_rate": 0.00017881760099782958,
      "loss": 12.5283,
      "step": 8620
    },
    {
      "epoch": 0.46944790454214397,
      "grad_norm": 0.6515931879334268,
      "learning_rate": 0.00017881217349529697,
      "loss": 12.4897,
      "step": 8621
    },
    {
      "epoch": 0.469502358538727,
      "grad_norm": 0.6473209646368532,
      "learning_rate": 0.0001788067453799077,
      "loss": 12.4732,
      "step": 8622
    },
    {
      "epoch": 0.46955681253531,
      "grad_norm": 0.7198939445665117,
      "learning_rate": 0.00017880131665170393,
      "loss": 12.465,
      "step": 8623
    },
    {
      "epoch": 0.469611266531893,
      "grad_norm": 0.6935062918663318,
      "learning_rate": 0.00017879588731072794,
      "loss": 12.4833,
      "step": 8624
    },
    {
      "epoch": 0.46966572052847605,
      "grad_norm": 0.6226308375788506,
      "learning_rate": 0.0001787904573570219,
      "loss": 12.5073,
      "step": 8625
    },
    {
      "epoch": 0.46972017452505904,
      "grad_norm": 0.6849934242523069,
      "learning_rate": 0.00017878502679062806,
      "loss": 12.3697,
      "step": 8626
    },
    {
      "epoch": 0.46977462852164203,
      "grad_norm": 0.6070916164445785,
      "learning_rate": 0.00017877959561158862,
      "loss": 12.4153,
      "step": 8627
    },
    {
      "epoch": 0.4698290825182251,
      "grad_norm": 0.6603216087072952,
      "learning_rate": 0.00017877416381994584,
      "loss": 12.4331,
      "step": 8628
    },
    {
      "epoch": 0.4698835365148081,
      "grad_norm": 0.5860755470881435,
      "learning_rate": 0.00017876873141574198,
      "loss": 12.4932,
      "step": 8629
    },
    {
      "epoch": 0.46993799051139107,
      "grad_norm": 0.607831027666431,
      "learning_rate": 0.00017876329839901922,
      "loss": 12.4657,
      "step": 8630
    },
    {
      "epoch": 0.4699924445079741,
      "grad_norm": 0.6506427431137368,
      "learning_rate": 0.00017875786476981986,
      "loss": 12.4483,
      "step": 8631
    },
    {
      "epoch": 0.4700468985045571,
      "grad_norm": 0.6116981851750112,
      "learning_rate": 0.0001787524305281861,
      "loss": 12.4156,
      "step": 8632
    },
    {
      "epoch": 0.47010135250114016,
      "grad_norm": 0.6386453334944827,
      "learning_rate": 0.00017874699567416028,
      "loss": 12.4169,
      "step": 8633
    },
    {
      "epoch": 0.47015580649772315,
      "grad_norm": 0.5990352000029187,
      "learning_rate": 0.0001787415602077846,
      "loss": 12.4613,
      "step": 8634
    },
    {
      "epoch": 0.47021026049430614,
      "grad_norm": 0.629306239232886,
      "learning_rate": 0.00017873612412910134,
      "loss": 12.3987,
      "step": 8635
    },
    {
      "epoch": 0.4702647144908892,
      "grad_norm": 0.6362931725753519,
      "learning_rate": 0.00017873068743815278,
      "loss": 12.3794,
      "step": 8636
    },
    {
      "epoch": 0.4703191684874722,
      "grad_norm": 0.6621394092121959,
      "learning_rate": 0.00017872525013498122,
      "loss": 12.4266,
      "step": 8637
    },
    {
      "epoch": 0.4703736224840552,
      "grad_norm": 0.6522150718394919,
      "learning_rate": 0.00017871981221962886,
      "loss": 12.4786,
      "step": 8638
    },
    {
      "epoch": 0.4704280764806382,
      "grad_norm": 0.6182797125326692,
      "learning_rate": 0.00017871437369213806,
      "loss": 12.3953,
      "step": 8639
    },
    {
      "epoch": 0.4704825304772212,
      "grad_norm": 0.741453587509592,
      "learning_rate": 0.0001787089345525511,
      "loss": 12.4706,
      "step": 8640
    },
    {
      "epoch": 0.4705369844738042,
      "grad_norm": 0.6616730918075981,
      "learning_rate": 0.00017870349480091023,
      "loss": 12.407,
      "step": 8641
    },
    {
      "epoch": 0.47059143847038726,
      "grad_norm": 0.6721799261869218,
      "learning_rate": 0.00017869805443725782,
      "loss": 12.4711,
      "step": 8642
    },
    {
      "epoch": 0.47064589246697025,
      "grad_norm": 0.6905179211399968,
      "learning_rate": 0.00017869261346163616,
      "loss": 12.4337,
      "step": 8643
    },
    {
      "epoch": 0.47070034646355324,
      "grad_norm": 0.5665557718261672,
      "learning_rate": 0.0001786871718740875,
      "loss": 12.3544,
      "step": 8644
    },
    {
      "epoch": 0.4707548004601363,
      "grad_norm": 0.6720406366448736,
      "learning_rate": 0.0001786817296746542,
      "loss": 12.3461,
      "step": 8645
    },
    {
      "epoch": 0.4708092544567193,
      "grad_norm": 0.7474814381158854,
      "learning_rate": 0.00017867628686337857,
      "loss": 12.5816,
      "step": 8646
    },
    {
      "epoch": 0.4708637084533023,
      "grad_norm": 0.6518981831942263,
      "learning_rate": 0.00017867084344030295,
      "loss": 12.4326,
      "step": 8647
    },
    {
      "epoch": 0.4709181624498853,
      "grad_norm": 0.6892336094441212,
      "learning_rate": 0.00017866539940546966,
      "loss": 12.4625,
      "step": 8648
    },
    {
      "epoch": 0.4709726164464683,
      "grad_norm": 0.68022474893785,
      "learning_rate": 0.00017865995475892105,
      "loss": 12.3368,
      "step": 8649
    },
    {
      "epoch": 0.4710270704430513,
      "grad_norm": 0.7638458776577514,
      "learning_rate": 0.00017865450950069943,
      "loss": 12.5662,
      "step": 8650
    },
    {
      "epoch": 0.47108152443963436,
      "grad_norm": 0.6359881616953548,
      "learning_rate": 0.00017864906363084714,
      "loss": 12.3063,
      "step": 8651
    },
    {
      "epoch": 0.47113597843621735,
      "grad_norm": 0.6350355813886267,
      "learning_rate": 0.00017864361714940653,
      "loss": 12.4142,
      "step": 8652
    },
    {
      "epoch": 0.47119043243280034,
      "grad_norm": 0.6276882384257606,
      "learning_rate": 0.00017863817005642002,
      "loss": 12.3826,
      "step": 8653
    },
    {
      "epoch": 0.4712448864293834,
      "grad_norm": 0.6387207175067731,
      "learning_rate": 0.00017863272235192987,
      "loss": 12.4255,
      "step": 8654
    },
    {
      "epoch": 0.4712993404259664,
      "grad_norm": 0.6642427666470622,
      "learning_rate": 0.00017862727403597848,
      "loss": 12.329,
      "step": 8655
    },
    {
      "epoch": 0.4713537944225494,
      "grad_norm": 0.7565854838138347,
      "learning_rate": 0.00017862182510860827,
      "loss": 12.6547,
      "step": 8656
    },
    {
      "epoch": 0.4714082484191324,
      "grad_norm": 0.655967108899123,
      "learning_rate": 0.00017861637556986152,
      "loss": 12.4853,
      "step": 8657
    },
    {
      "epoch": 0.4714627024157154,
      "grad_norm": 0.5841788295309343,
      "learning_rate": 0.00017861092541978063,
      "loss": 12.4258,
      "step": 8658
    },
    {
      "epoch": 0.4715171564122984,
      "grad_norm": 0.6802712400530733,
      "learning_rate": 0.000178605474658408,
      "loss": 12.164,
      "step": 8659
    },
    {
      "epoch": 0.47157161040888146,
      "grad_norm": 0.5812630193366193,
      "learning_rate": 0.00017860002328578606,
      "loss": 12.4212,
      "step": 8660
    },
    {
      "epoch": 0.47162606440546445,
      "grad_norm": 0.7439484017953842,
      "learning_rate": 0.00017859457130195715,
      "loss": 12.5639,
      "step": 8661
    },
    {
      "epoch": 0.47168051840204744,
      "grad_norm": 0.6444918782437785,
      "learning_rate": 0.00017858911870696366,
      "loss": 12.4413,
      "step": 8662
    },
    {
      "epoch": 0.4717349723986305,
      "grad_norm": 0.6042122509649913,
      "learning_rate": 0.00017858366550084801,
      "loss": 12.2887,
      "step": 8663
    },
    {
      "epoch": 0.4717894263952135,
      "grad_norm": 0.719636542138421,
      "learning_rate": 0.00017857821168365258,
      "loss": 12.3589,
      "step": 8664
    },
    {
      "epoch": 0.47184388039179653,
      "grad_norm": 0.6004943446157107,
      "learning_rate": 0.00017857275725541983,
      "loss": 12.55,
      "step": 8665
    },
    {
      "epoch": 0.4718983343883795,
      "grad_norm": 0.6690559070467175,
      "learning_rate": 0.00017856730221619212,
      "loss": 12.3805,
      "step": 8666
    },
    {
      "epoch": 0.4719527883849625,
      "grad_norm": 0.6520489856594532,
      "learning_rate": 0.00017856184656601189,
      "loss": 12.3617,
      "step": 8667
    },
    {
      "epoch": 0.47200724238154557,
      "grad_norm": 0.5842016736999017,
      "learning_rate": 0.00017855639030492156,
      "loss": 12.3109,
      "step": 8668
    },
    {
      "epoch": 0.47206169637812856,
      "grad_norm": 0.6875535840151342,
      "learning_rate": 0.0001785509334329636,
      "loss": 12.4851,
      "step": 8669
    },
    {
      "epoch": 0.47211615037471155,
      "grad_norm": 0.6728011912417281,
      "learning_rate": 0.0001785454759501804,
      "loss": 12.4484,
      "step": 8670
    },
    {
      "epoch": 0.4721706043712946,
      "grad_norm": 0.6394208019399653,
      "learning_rate": 0.0001785400178566144,
      "loss": 12.5169,
      "step": 8671
    },
    {
      "epoch": 0.4722250583678776,
      "grad_norm": 0.7233003380113386,
      "learning_rate": 0.00017853455915230803,
      "loss": 12.5152,
      "step": 8672
    },
    {
      "epoch": 0.4722795123644606,
      "grad_norm": 0.6577078659945517,
      "learning_rate": 0.00017852909983730376,
      "loss": 12.2267,
      "step": 8673
    },
    {
      "epoch": 0.47233396636104363,
      "grad_norm": 0.6681012762094728,
      "learning_rate": 0.00017852363991164406,
      "loss": 12.4857,
      "step": 8674
    },
    {
      "epoch": 0.4723884203576266,
      "grad_norm": 0.6492669412836646,
      "learning_rate": 0.00017851817937537137,
      "loss": 12.3342,
      "step": 8675
    },
    {
      "epoch": 0.4724428743542096,
      "grad_norm": 0.6258051004677605,
      "learning_rate": 0.00017851271822852817,
      "loss": 12.3255,
      "step": 8676
    },
    {
      "epoch": 0.47249732835079267,
      "grad_norm": 0.6177766546154925,
      "learning_rate": 0.00017850725647115684,
      "loss": 12.453,
      "step": 8677
    },
    {
      "epoch": 0.47255178234737566,
      "grad_norm": 0.5976114556341114,
      "learning_rate": 0.00017850179410329998,
      "loss": 12.354,
      "step": 8678
    },
    {
      "epoch": 0.47260623634395865,
      "grad_norm": 0.6943464826461088,
      "learning_rate": 0.00017849633112499997,
      "loss": 12.4133,
      "step": 8679
    },
    {
      "epoch": 0.4726606903405417,
      "grad_norm": 0.6629539463742055,
      "learning_rate": 0.00017849086753629934,
      "loss": 12.5159,
      "step": 8680
    },
    {
      "epoch": 0.4727151443371247,
      "grad_norm": 0.5857907651074736,
      "learning_rate": 0.00017848540333724054,
      "loss": 12.3979,
      "step": 8681
    },
    {
      "epoch": 0.4727695983337077,
      "grad_norm": 0.66682024503514,
      "learning_rate": 0.0001784799385278661,
      "loss": 12.3639,
      "step": 8682
    },
    {
      "epoch": 0.47282405233029073,
      "grad_norm": 0.6785937455203835,
      "learning_rate": 0.0001784744731082185,
      "loss": 12.7207,
      "step": 8683
    },
    {
      "epoch": 0.4728785063268737,
      "grad_norm": 0.6830103773165097,
      "learning_rate": 0.00017846900707834022,
      "loss": 12.4353,
      "step": 8684
    },
    {
      "epoch": 0.4729329603234567,
      "grad_norm": 0.6678547876212807,
      "learning_rate": 0.0001784635404382738,
      "loss": 12.5259,
      "step": 8685
    },
    {
      "epoch": 0.47298741432003977,
      "grad_norm": 0.6663746763710154,
      "learning_rate": 0.00017845807318806175,
      "loss": 12.3938,
      "step": 8686
    },
    {
      "epoch": 0.47304186831662276,
      "grad_norm": 0.6361131616374217,
      "learning_rate": 0.00017845260532774654,
      "loss": 12.271,
      "step": 8687
    },
    {
      "epoch": 0.47309632231320575,
      "grad_norm": 0.7273696714957184,
      "learning_rate": 0.00017844713685737069,
      "loss": 12.2901,
      "step": 8688
    },
    {
      "epoch": 0.4731507763097888,
      "grad_norm": 0.6327991357395276,
      "learning_rate": 0.00017844166777697678,
      "loss": 12.3317,
      "step": 8689
    },
    {
      "epoch": 0.4732052303063718,
      "grad_norm": 0.6234225324962389,
      "learning_rate": 0.0001784361980866073,
      "loss": 12.4677,
      "step": 8690
    },
    {
      "epoch": 0.4732596843029548,
      "grad_norm": 0.6428092940547633,
      "learning_rate": 0.00017843072778630478,
      "loss": 12.3023,
      "step": 8691
    },
    {
      "epoch": 0.47331413829953783,
      "grad_norm": 0.5903591289493301,
      "learning_rate": 0.00017842525687611179,
      "loss": 12.3877,
      "step": 8692
    },
    {
      "epoch": 0.4733685922961208,
      "grad_norm": 0.6341704384358905,
      "learning_rate": 0.0001784197853560708,
      "loss": 12.3853,
      "step": 8693
    },
    {
      "epoch": 0.4734230462927038,
      "grad_norm": 0.6473153367853635,
      "learning_rate": 0.00017841431322622447,
      "loss": 12.4951,
      "step": 8694
    },
    {
      "epoch": 0.47347750028928687,
      "grad_norm": 0.6562941393001934,
      "learning_rate": 0.00017840884048661527,
      "loss": 12.3385,
      "step": 8695
    },
    {
      "epoch": 0.47353195428586986,
      "grad_norm": 0.5803770674490643,
      "learning_rate": 0.0001784033671372858,
      "loss": 12.5054,
      "step": 8696
    },
    {
      "epoch": 0.47358640828245285,
      "grad_norm": 0.606714580039266,
      "learning_rate": 0.00017839789317827855,
      "loss": 12.4175,
      "step": 8697
    },
    {
      "epoch": 0.4736408622790359,
      "grad_norm": 0.6835740224685091,
      "learning_rate": 0.00017839241860963617,
      "loss": 12.3783,
      "step": 8698
    },
    {
      "epoch": 0.4736953162756189,
      "grad_norm": 0.6513242061593827,
      "learning_rate": 0.00017838694343140117,
      "loss": 12.4464,
      "step": 8699
    },
    {
      "epoch": 0.47374977027220194,
      "grad_norm": 0.7031548801620282,
      "learning_rate": 0.00017838146764361619,
      "loss": 12.3203,
      "step": 8700
    },
    {
      "epoch": 0.47380422426878493,
      "grad_norm": 0.6364147771234054,
      "learning_rate": 0.00017837599124632375,
      "loss": 12.3763,
      "step": 8701
    },
    {
      "epoch": 0.4738586782653679,
      "grad_norm": 0.6186036418075409,
      "learning_rate": 0.00017837051423956644,
      "loss": 12.3576,
      "step": 8702
    },
    {
      "epoch": 0.473913132261951,
      "grad_norm": 0.6176910401272563,
      "learning_rate": 0.00017836503662338688,
      "loss": 12.401,
      "step": 8703
    },
    {
      "epoch": 0.47396758625853397,
      "grad_norm": 0.6096101648023464,
      "learning_rate": 0.00017835955839782766,
      "loss": 12.3551,
      "step": 8704
    },
    {
      "epoch": 0.47402204025511696,
      "grad_norm": 0.5980960499670746,
      "learning_rate": 0.00017835407956293136,
      "loss": 12.3039,
      "step": 8705
    },
    {
      "epoch": 0.4740764942517,
      "grad_norm": 0.6500700845049932,
      "learning_rate": 0.00017834860011874064,
      "loss": 12.4876,
      "step": 8706
    },
    {
      "epoch": 0.474130948248283,
      "grad_norm": 0.6091034212568243,
      "learning_rate": 0.00017834312006529803,
      "loss": 12.4782,
      "step": 8707
    },
    {
      "epoch": 0.474185402244866,
      "grad_norm": 0.6550246652321426,
      "learning_rate": 0.00017833763940264618,
      "loss": 12.4359,
      "step": 8708
    },
    {
      "epoch": 0.47423985624144904,
      "grad_norm": 0.652718777215914,
      "learning_rate": 0.0001783321581308277,
      "loss": 12.5009,
      "step": 8709
    },
    {
      "epoch": 0.47429431023803204,
      "grad_norm": 0.6357131550301056,
      "learning_rate": 0.00017832667624988525,
      "loss": 12.3438,
      "step": 8710
    },
    {
      "epoch": 0.47434876423461503,
      "grad_norm": 0.6056142520938775,
      "learning_rate": 0.00017832119375986143,
      "loss": 12.3436,
      "step": 8711
    },
    {
      "epoch": 0.4744032182311981,
      "grad_norm": 0.6836617173901649,
      "learning_rate": 0.00017831571066079886,
      "loss": 12.4552,
      "step": 8712
    },
    {
      "epoch": 0.47445767222778107,
      "grad_norm": 0.6196554753439548,
      "learning_rate": 0.00017831022695274018,
      "loss": 12.4012,
      "step": 8713
    },
    {
      "epoch": 0.47451212622436406,
      "grad_norm": 0.5652293429605024,
      "learning_rate": 0.00017830474263572804,
      "loss": 12.3303,
      "step": 8714
    },
    {
      "epoch": 0.4745665802209471,
      "grad_norm": 0.6475256810886932,
      "learning_rate": 0.00017829925770980514,
      "loss": 12.5303,
      "step": 8715
    },
    {
      "epoch": 0.4746210342175301,
      "grad_norm": 0.6036166661171558,
      "learning_rate": 0.00017829377217501403,
      "loss": 12.383,
      "step": 8716
    },
    {
      "epoch": 0.4746754882141131,
      "grad_norm": 0.6985522299689985,
      "learning_rate": 0.00017828828603139743,
      "loss": 12.5098,
      "step": 8717
    },
    {
      "epoch": 0.47472994221069614,
      "grad_norm": 0.6070778484391904,
      "learning_rate": 0.00017828279927899798,
      "loss": 12.391,
      "step": 8718
    },
    {
      "epoch": 0.47478439620727914,
      "grad_norm": 0.5823182202407687,
      "learning_rate": 0.00017827731191785836,
      "loss": 12.3258,
      "step": 8719
    },
    {
      "epoch": 0.47483885020386213,
      "grad_norm": 0.5893513288995854,
      "learning_rate": 0.00017827182394802128,
      "loss": 12.2139,
      "step": 8720
    },
    {
      "epoch": 0.4748933042004452,
      "grad_norm": 0.7013688112671357,
      "learning_rate": 0.0001782663353695293,
      "loss": 12.5296,
      "step": 8721
    },
    {
      "epoch": 0.47494775819702817,
      "grad_norm": 0.5962121584629747,
      "learning_rate": 0.0001782608461824252,
      "loss": 12.4412,
      "step": 8722
    },
    {
      "epoch": 0.47500221219361116,
      "grad_norm": 0.658193756347851,
      "learning_rate": 0.00017825535638675165,
      "loss": 12.3744,
      "step": 8723
    },
    {
      "epoch": 0.4750566661901942,
      "grad_norm": 0.5829182717805368,
      "learning_rate": 0.00017824986598255133,
      "loss": 12.5226,
      "step": 8724
    },
    {
      "epoch": 0.4751111201867772,
      "grad_norm": 0.5963759135534142,
      "learning_rate": 0.0001782443749698669,
      "loss": 12.4221,
      "step": 8725
    },
    {
      "epoch": 0.4751655741833602,
      "grad_norm": 0.620977287497132,
      "learning_rate": 0.0001782388833487411,
      "loss": 12.3833,
      "step": 8726
    },
    {
      "epoch": 0.47522002817994324,
      "grad_norm": 0.6899744567652295,
      "learning_rate": 0.00017823339111921663,
      "loss": 12.4674,
      "step": 8727
    },
    {
      "epoch": 0.47527448217652624,
      "grad_norm": 0.6151290935196063,
      "learning_rate": 0.00017822789828133618,
      "loss": 12.5067,
      "step": 8728
    },
    {
      "epoch": 0.47532893617310923,
      "grad_norm": 0.569806095972506,
      "learning_rate": 0.0001782224048351425,
      "loss": 12.4047,
      "step": 8729
    },
    {
      "epoch": 0.4753833901696923,
      "grad_norm": 0.5995478910348392,
      "learning_rate": 0.00017821691078067823,
      "loss": 12.4148,
      "step": 8730
    },
    {
      "epoch": 0.47543784416627527,
      "grad_norm": 0.6286888717890325,
      "learning_rate": 0.00017821141611798618,
      "loss": 12.3511,
      "step": 8731
    },
    {
      "epoch": 0.4754922981628583,
      "grad_norm": 0.5648835797724097,
      "learning_rate": 0.00017820592084710906,
      "loss": 12.2943,
      "step": 8732
    },
    {
      "epoch": 0.4755467521594413,
      "grad_norm": 0.6112850266141724,
      "learning_rate": 0.00017820042496808955,
      "loss": 12.4158,
      "step": 8733
    },
    {
      "epoch": 0.4756012061560243,
      "grad_norm": 0.661691812372672,
      "learning_rate": 0.00017819492848097045,
      "loss": 12.4666,
      "step": 8734
    },
    {
      "epoch": 0.47565566015260735,
      "grad_norm": 0.594886371446287,
      "learning_rate": 0.00017818943138579445,
      "loss": 12.4039,
      "step": 8735
    },
    {
      "epoch": 0.47571011414919034,
      "grad_norm": 0.6531180638084092,
      "learning_rate": 0.00017818393368260432,
      "loss": 12.3811,
      "step": 8736
    },
    {
      "epoch": 0.47576456814577334,
      "grad_norm": 0.7078781658459506,
      "learning_rate": 0.0001781784353714428,
      "loss": 12.5008,
      "step": 8737
    },
    {
      "epoch": 0.4758190221423564,
      "grad_norm": 0.596332682579423,
      "learning_rate": 0.0001781729364523527,
      "loss": 12.466,
      "step": 8738
    },
    {
      "epoch": 0.4758734761389394,
      "grad_norm": 0.6160511904069175,
      "learning_rate": 0.0001781674369253767,
      "loss": 12.4099,
      "step": 8739
    },
    {
      "epoch": 0.47592793013552237,
      "grad_norm": 0.591545992505644,
      "learning_rate": 0.0001781619367905576,
      "loss": 12.3754,
      "step": 8740
    },
    {
      "epoch": 0.4759823841321054,
      "grad_norm": 0.7002224308550122,
      "learning_rate": 0.0001781564360479382,
      "loss": 12.3937,
      "step": 8741
    },
    {
      "epoch": 0.4760368381286884,
      "grad_norm": 0.6085928202462332,
      "learning_rate": 0.0001781509346975612,
      "loss": 12.3617,
      "step": 8742
    },
    {
      "epoch": 0.4760912921252714,
      "grad_norm": 0.6398433181849218,
      "learning_rate": 0.00017814543273946947,
      "loss": 12.4901,
      "step": 8743
    },
    {
      "epoch": 0.47614574612185445,
      "grad_norm": 0.7506931499875767,
      "learning_rate": 0.00017813993017370576,
      "loss": 12.495,
      "step": 8744
    },
    {
      "epoch": 0.47620020011843744,
      "grad_norm": 0.5799727249777821,
      "learning_rate": 0.00017813442700031283,
      "loss": 12.3573,
      "step": 8745
    },
    {
      "epoch": 0.47625465411502044,
      "grad_norm": 0.5869373549653786,
      "learning_rate": 0.00017812892321933345,
      "loss": 12.4455,
      "step": 8746
    },
    {
      "epoch": 0.4763091081116035,
      "grad_norm": 0.5965018552793068,
      "learning_rate": 0.00017812341883081053,
      "loss": 12.4557,
      "step": 8747
    },
    {
      "epoch": 0.4763635621081865,
      "grad_norm": 0.6541023786405988,
      "learning_rate": 0.00017811791383478675,
      "loss": 12.5107,
      "step": 8748
    },
    {
      "epoch": 0.47641801610476947,
      "grad_norm": 0.6206970523605138,
      "learning_rate": 0.000178112408231305,
      "loss": 12.5198,
      "step": 8749
    },
    {
      "epoch": 0.4764724701013525,
      "grad_norm": 0.6499952009932339,
      "learning_rate": 0.00017810690202040805,
      "loss": 12.4502,
      "step": 8750
    },
    {
      "epoch": 0.4765269240979355,
      "grad_norm": 0.6521865596139488,
      "learning_rate": 0.00017810139520213874,
      "loss": 12.4045,
      "step": 8751
    },
    {
      "epoch": 0.4765813780945185,
      "grad_norm": 0.5973388862310863,
      "learning_rate": 0.00017809588777653986,
      "loss": 12.4887,
      "step": 8752
    },
    {
      "epoch": 0.47663583209110155,
      "grad_norm": 0.5928298859565891,
      "learning_rate": 0.0001780903797436543,
      "loss": 12.376,
      "step": 8753
    },
    {
      "epoch": 0.47669028608768454,
      "grad_norm": 0.6457349298774065,
      "learning_rate": 0.00017808487110352483,
      "loss": 12.5605,
      "step": 8754
    },
    {
      "epoch": 0.47674474008426754,
      "grad_norm": 0.6119826626738841,
      "learning_rate": 0.00017807936185619433,
      "loss": 12.4986,
      "step": 8755
    },
    {
      "epoch": 0.4767991940808506,
      "grad_norm": 0.5863087139114767,
      "learning_rate": 0.0001780738520017056,
      "loss": 12.3536,
      "step": 8756
    },
    {
      "epoch": 0.4768536480774336,
      "grad_norm": 0.6428427397487143,
      "learning_rate": 0.0001780683415401015,
      "loss": 12.3435,
      "step": 8757
    },
    {
      "epoch": 0.47690810207401657,
      "grad_norm": 0.6967791103950431,
      "learning_rate": 0.00017806283047142488,
      "loss": 12.5013,
      "step": 8758
    },
    {
      "epoch": 0.4769625560705996,
      "grad_norm": 0.6705016723421287,
      "learning_rate": 0.00017805731879571858,
      "loss": 12.5209,
      "step": 8759
    },
    {
      "epoch": 0.4770170100671826,
      "grad_norm": 0.6148028932835333,
      "learning_rate": 0.00017805180651302553,
      "loss": 12.5144,
      "step": 8760
    },
    {
      "epoch": 0.4770714640637656,
      "grad_norm": 0.5768470841705992,
      "learning_rate": 0.00017804629362338852,
      "loss": 12.4492,
      "step": 8761
    },
    {
      "epoch": 0.47712591806034865,
      "grad_norm": 0.694438897567131,
      "learning_rate": 0.00017804078012685043,
      "loss": 12.4018,
      "step": 8762
    },
    {
      "epoch": 0.47718037205693165,
      "grad_norm": 0.6351675659575002,
      "learning_rate": 0.00017803526602345416,
      "loss": 12.5062,
      "step": 8763
    },
    {
      "epoch": 0.47723482605351464,
      "grad_norm": 0.6840557802934399,
      "learning_rate": 0.0001780297513132426,
      "loss": 12.4258,
      "step": 8764
    },
    {
      "epoch": 0.4772892800500977,
      "grad_norm": 0.6269476064332131,
      "learning_rate": 0.00017802423599625855,
      "loss": 12.3025,
      "step": 8765
    },
    {
      "epoch": 0.4773437340466807,
      "grad_norm": 0.748368075863479,
      "learning_rate": 0.000178018720072545,
      "loss": 12.45,
      "step": 8766
    },
    {
      "epoch": 0.4773981880432637,
      "grad_norm": 0.6079501771640944,
      "learning_rate": 0.00017801320354214476,
      "loss": 12.582,
      "step": 8767
    },
    {
      "epoch": 0.4774526420398467,
      "grad_norm": 0.6021377793201013,
      "learning_rate": 0.0001780076864051008,
      "loss": 12.3335,
      "step": 8768
    },
    {
      "epoch": 0.4775070960364297,
      "grad_norm": 0.6756093490145296,
      "learning_rate": 0.000178002168661456,
      "loss": 12.3745,
      "step": 8769
    },
    {
      "epoch": 0.47756155003301276,
      "grad_norm": 0.6197236151026202,
      "learning_rate": 0.0001779966503112532,
      "loss": 12.3967,
      "step": 8770
    },
    {
      "epoch": 0.47761600402959575,
      "grad_norm": 0.6525965658124591,
      "learning_rate": 0.00017799113135453541,
      "loss": 12.431,
      "step": 8771
    },
    {
      "epoch": 0.47767045802617875,
      "grad_norm": 0.7092955749743148,
      "learning_rate": 0.00017798561179134553,
      "loss": 12.3853,
      "step": 8772
    },
    {
      "epoch": 0.4777249120227618,
      "grad_norm": 0.6196543964995244,
      "learning_rate": 0.0001779800916217264,
      "loss": 12.4704,
      "step": 8773
    },
    {
      "epoch": 0.4777793660193448,
      "grad_norm": 0.7215102850249048,
      "learning_rate": 0.00017797457084572102,
      "loss": 12.4219,
      "step": 8774
    },
    {
      "epoch": 0.4778338200159278,
      "grad_norm": 0.6234504458117953,
      "learning_rate": 0.0001779690494633723,
      "loss": 12.4559,
      "step": 8775
    },
    {
      "epoch": 0.4778882740125108,
      "grad_norm": 0.6269716209072117,
      "learning_rate": 0.0001779635274747232,
      "loss": 12.4085,
      "step": 8776
    },
    {
      "epoch": 0.4779427280090938,
      "grad_norm": 0.7042413882289371,
      "learning_rate": 0.0001779580048798166,
      "loss": 12.2859,
      "step": 8777
    },
    {
      "epoch": 0.4779971820056768,
      "grad_norm": 0.5900419912112504,
      "learning_rate": 0.00017795248167869549,
      "loss": 12.3652,
      "step": 8778
    },
    {
      "epoch": 0.47805163600225986,
      "grad_norm": 0.6511384156373964,
      "learning_rate": 0.0001779469578714028,
      "loss": 12.3114,
      "step": 8779
    },
    {
      "epoch": 0.47810608999884285,
      "grad_norm": 0.6670601280040062,
      "learning_rate": 0.0001779414334579815,
      "loss": 12.3957,
      "step": 8780
    },
    {
      "epoch": 0.47816054399542585,
      "grad_norm": 0.5890634034622276,
      "learning_rate": 0.00017793590843847456,
      "loss": 12.3756,
      "step": 8781
    },
    {
      "epoch": 0.4782149979920089,
      "grad_norm": 0.6218526690363533,
      "learning_rate": 0.00017793038281292494,
      "loss": 12.4279,
      "step": 8782
    },
    {
      "epoch": 0.4782694519885919,
      "grad_norm": 0.6094031786819534,
      "learning_rate": 0.00017792485658137553,
      "loss": 12.3809,
      "step": 8783
    },
    {
      "epoch": 0.4783239059851749,
      "grad_norm": 0.6674236673610006,
      "learning_rate": 0.00017791932974386943,
      "loss": 12.3462,
      "step": 8784
    },
    {
      "epoch": 0.47837835998175793,
      "grad_norm": 0.8290033168357881,
      "learning_rate": 0.00017791380230044955,
      "loss": 12.4798,
      "step": 8785
    },
    {
      "epoch": 0.4784328139783409,
      "grad_norm": 0.5698319474926054,
      "learning_rate": 0.00017790827425115887,
      "loss": 12.3464,
      "step": 8786
    },
    {
      "epoch": 0.4784872679749239,
      "grad_norm": 0.7063867609709856,
      "learning_rate": 0.00017790274559604033,
      "loss": 12.4459,
      "step": 8787
    },
    {
      "epoch": 0.47854172197150696,
      "grad_norm": 0.6248814974218623,
      "learning_rate": 0.00017789721633513703,
      "loss": 12.3679,
      "step": 8788
    },
    {
      "epoch": 0.47859617596808995,
      "grad_norm": 0.6049202524282553,
      "learning_rate": 0.0001778916864684919,
      "loss": 12.4051,
      "step": 8789
    },
    {
      "epoch": 0.47865062996467295,
      "grad_norm": 0.6707745185056192,
      "learning_rate": 0.00017788615599614798,
      "loss": 12.3868,
      "step": 8790
    },
    {
      "epoch": 0.478705083961256,
      "grad_norm": 0.5512079569667026,
      "learning_rate": 0.0001778806249181482,
      "loss": 12.3534,
      "step": 8791
    },
    {
      "epoch": 0.478759537957839,
      "grad_norm": 0.6225053075193171,
      "learning_rate": 0.00017787509323453565,
      "loss": 12.4305,
      "step": 8792
    },
    {
      "epoch": 0.478813991954422,
      "grad_norm": 0.6709506145422299,
      "learning_rate": 0.00017786956094535333,
      "loss": 12.4308,
      "step": 8793
    },
    {
      "epoch": 0.47886844595100503,
      "grad_norm": 0.6751250487279195,
      "learning_rate": 0.0001778640280506442,
      "loss": 12.4982,
      "step": 8794
    },
    {
      "epoch": 0.478922899947588,
      "grad_norm": 0.6378585206417511,
      "learning_rate": 0.00017785849455045138,
      "loss": 12.4145,
      "step": 8795
    },
    {
      "epoch": 0.478977353944171,
      "grad_norm": 0.6245405864362273,
      "learning_rate": 0.0001778529604448178,
      "loss": 12.2687,
      "step": 8796
    },
    {
      "epoch": 0.47903180794075406,
      "grad_norm": 0.6113639663876627,
      "learning_rate": 0.0001778474257337866,
      "loss": 12.4223,
      "step": 8797
    },
    {
      "epoch": 0.47908626193733705,
      "grad_norm": 0.5721990527380328,
      "learning_rate": 0.0001778418904174007,
      "loss": 12.1567,
      "step": 8798
    },
    {
      "epoch": 0.4791407159339201,
      "grad_norm": 0.6724946226809764,
      "learning_rate": 0.00017783635449570326,
      "loss": 12.4662,
      "step": 8799
    },
    {
      "epoch": 0.4791951699305031,
      "grad_norm": 0.6178789728367964,
      "learning_rate": 0.00017783081796873725,
      "loss": 12.431,
      "step": 8800
    },
    {
      "epoch": 0.4792496239270861,
      "grad_norm": 0.6579330034517156,
      "learning_rate": 0.00017782528083654575,
      "loss": 12.2458,
      "step": 8801
    },
    {
      "epoch": 0.47930407792366914,
      "grad_norm": 0.6058708382556764,
      "learning_rate": 0.0001778197430991718,
      "loss": 12.442,
      "step": 8802
    },
    {
      "epoch": 0.47935853192025213,
      "grad_norm": 0.6377720679987039,
      "learning_rate": 0.0001778142047566585,
      "loss": 12.461,
      "step": 8803
    },
    {
      "epoch": 0.4794129859168351,
      "grad_norm": 0.5970891817149896,
      "learning_rate": 0.0001778086658090489,
      "loss": 12.4052,
      "step": 8804
    },
    {
      "epoch": 0.47946743991341817,
      "grad_norm": 0.6461955050830236,
      "learning_rate": 0.00017780312625638605,
      "loss": 12.3717,
      "step": 8805
    },
    {
      "epoch": 0.47952189391000116,
      "grad_norm": 0.6468828453737918,
      "learning_rate": 0.000177797586098713,
      "loss": 12.4399,
      "step": 8806
    },
    {
      "epoch": 0.47957634790658416,
      "grad_norm": 0.6621514434037283,
      "learning_rate": 0.00017779204533607297,
      "loss": 12.5993,
      "step": 8807
    },
    {
      "epoch": 0.4796308019031672,
      "grad_norm": 0.58841400655816,
      "learning_rate": 0.0001777865039685089,
      "loss": 12.4557,
      "step": 8808
    },
    {
      "epoch": 0.4796852558997502,
      "grad_norm": 0.6365515386699335,
      "learning_rate": 0.00017778096199606394,
      "loss": 12.3267,
      "step": 8809
    },
    {
      "epoch": 0.4797397098963332,
      "grad_norm": 0.6823179343753297,
      "learning_rate": 0.00017777541941878114,
      "loss": 12.4056,
      "step": 8810
    },
    {
      "epoch": 0.47979416389291624,
      "grad_norm": 0.5769821845021594,
      "learning_rate": 0.00017776987623670368,
      "loss": 12.4249,
      "step": 8811
    },
    {
      "epoch": 0.47984861788949923,
      "grad_norm": 0.6255846491013269,
      "learning_rate": 0.00017776433244987458,
      "loss": 12.4606,
      "step": 8812
    },
    {
      "epoch": 0.4799030718860822,
      "grad_norm": 0.6799133311882912,
      "learning_rate": 0.000177758788058337,
      "loss": 12.4782,
      "step": 8813
    },
    {
      "epoch": 0.47995752588266527,
      "grad_norm": 0.6058273235144871,
      "learning_rate": 0.00017775324306213406,
      "loss": 12.4778,
      "step": 8814
    },
    {
      "epoch": 0.48001197987924826,
      "grad_norm": 0.6519108191538442,
      "learning_rate": 0.00017774769746130886,
      "loss": 12.4488,
      "step": 8815
    },
    {
      "epoch": 0.48006643387583126,
      "grad_norm": 0.5844059895700789,
      "learning_rate": 0.00017774215125590455,
      "loss": 12.4286,
      "step": 8816
    },
    {
      "epoch": 0.4801208878724143,
      "grad_norm": 0.6788807574460705,
      "learning_rate": 0.00017773660444596418,
      "loss": 12.5376,
      "step": 8817
    },
    {
      "epoch": 0.4801753418689973,
      "grad_norm": 0.6717908703669483,
      "learning_rate": 0.00017773105703153096,
      "loss": 12.4666,
      "step": 8818
    },
    {
      "epoch": 0.4802297958655803,
      "grad_norm": 0.5889511648902014,
      "learning_rate": 0.00017772550901264803,
      "loss": 12.4461,
      "step": 8819
    },
    {
      "epoch": 0.48028424986216334,
      "grad_norm": 0.7001047237575522,
      "learning_rate": 0.00017771996038935846,
      "loss": 12.5566,
      "step": 8820
    },
    {
      "epoch": 0.48033870385874633,
      "grad_norm": 0.6073367192293863,
      "learning_rate": 0.0001777144111617055,
      "loss": 12.4075,
      "step": 8821
    },
    {
      "epoch": 0.4803931578553293,
      "grad_norm": 0.7157381884611609,
      "learning_rate": 0.0001777088613297322,
      "loss": 12.4349,
      "step": 8822
    },
    {
      "epoch": 0.48044761185191237,
      "grad_norm": 0.6451758187148338,
      "learning_rate": 0.0001777033108934818,
      "loss": 12.383,
      "step": 8823
    },
    {
      "epoch": 0.48050206584849536,
      "grad_norm": 0.6259662761328934,
      "learning_rate": 0.00017769775985299738,
      "loss": 12.3062,
      "step": 8824
    },
    {
      "epoch": 0.48055651984507836,
      "grad_norm": 0.5982251100488803,
      "learning_rate": 0.00017769220820832218,
      "loss": 12.2562,
      "step": 8825
    },
    {
      "epoch": 0.4806109738416614,
      "grad_norm": 0.5716512373091801,
      "learning_rate": 0.00017768665595949934,
      "loss": 12.3767,
      "step": 8826
    },
    {
      "epoch": 0.4806654278382444,
      "grad_norm": 0.6944768655221844,
      "learning_rate": 0.00017768110310657204,
      "loss": 12.5173,
      "step": 8827
    },
    {
      "epoch": 0.4807198818348274,
      "grad_norm": 0.7524092316030287,
      "learning_rate": 0.00017767554964958344,
      "loss": 12.2713,
      "step": 8828
    },
    {
      "epoch": 0.48077433583141044,
      "grad_norm": 0.6292076363016551,
      "learning_rate": 0.00017766999558857673,
      "loss": 12.3015,
      "step": 8829
    },
    {
      "epoch": 0.48082878982799343,
      "grad_norm": 0.6170640257155307,
      "learning_rate": 0.00017766444092359512,
      "loss": 12.3525,
      "step": 8830
    },
    {
      "epoch": 0.4808832438245764,
      "grad_norm": 0.63768550877622,
      "learning_rate": 0.00017765888565468178,
      "loss": 12.4197,
      "step": 8831
    },
    {
      "epoch": 0.48093769782115947,
      "grad_norm": 0.6341427907839081,
      "learning_rate": 0.00017765332978187997,
      "loss": 12.4103,
      "step": 8832
    },
    {
      "epoch": 0.48099215181774246,
      "grad_norm": 0.6629413848045236,
      "learning_rate": 0.00017764777330523283,
      "loss": 12.4884,
      "step": 8833
    },
    {
      "epoch": 0.4810466058143255,
      "grad_norm": 0.6164832896056446,
      "learning_rate": 0.00017764221622478354,
      "loss": 12.4355,
      "step": 8834
    },
    {
      "epoch": 0.4811010598109085,
      "grad_norm": 0.5901202811298281,
      "learning_rate": 0.00017763665854057537,
      "loss": 12.3688,
      "step": 8835
    },
    {
      "epoch": 0.4811555138074915,
      "grad_norm": 0.6196660565552202,
      "learning_rate": 0.00017763110025265154,
      "loss": 12.4374,
      "step": 8836
    },
    {
      "epoch": 0.48120996780407455,
      "grad_norm": 0.6734610187372511,
      "learning_rate": 0.00017762554136105524,
      "loss": 12.3679,
      "step": 8837
    },
    {
      "epoch": 0.48126442180065754,
      "grad_norm": 0.8116360626391834,
      "learning_rate": 0.00017761998186582972,
      "loss": 12.5266,
      "step": 8838
    },
    {
      "epoch": 0.48131887579724053,
      "grad_norm": 0.6981569929406607,
      "learning_rate": 0.00017761442176701824,
      "loss": 12.5177,
      "step": 8839
    },
    {
      "epoch": 0.4813733297938236,
      "grad_norm": 0.5768553293400086,
      "learning_rate": 0.00017760886106466396,
      "loss": 12.3916,
      "step": 8840
    },
    {
      "epoch": 0.48142778379040657,
      "grad_norm": 0.7085652244519366,
      "learning_rate": 0.00017760329975881017,
      "loss": 12.3696,
      "step": 8841
    },
    {
      "epoch": 0.48148223778698956,
      "grad_norm": 0.6562927526967182,
      "learning_rate": 0.0001775977378495001,
      "loss": 12.3849,
      "step": 8842
    },
    {
      "epoch": 0.4815366917835726,
      "grad_norm": 0.7220168778666222,
      "learning_rate": 0.00017759217533677702,
      "loss": 12.3901,
      "step": 8843
    },
    {
      "epoch": 0.4815911457801556,
      "grad_norm": 0.5920359098395669,
      "learning_rate": 0.00017758661222068415,
      "loss": 12.3168,
      "step": 8844
    },
    {
      "epoch": 0.4816455997767386,
      "grad_norm": 0.6872763879710608,
      "learning_rate": 0.0001775810485012648,
      "loss": 12.3874,
      "step": 8845
    },
    {
      "epoch": 0.48170005377332165,
      "grad_norm": 0.5996468591319443,
      "learning_rate": 0.00017757548417856217,
      "loss": 12.4146,
      "step": 8846
    },
    {
      "epoch": 0.48175450776990464,
      "grad_norm": 0.5938416569055235,
      "learning_rate": 0.00017756991925261962,
      "loss": 12.3266,
      "step": 8847
    },
    {
      "epoch": 0.48180896176648763,
      "grad_norm": 0.7101647878990114,
      "learning_rate": 0.00017756435372348034,
      "loss": 12.4614,
      "step": 8848
    },
    {
      "epoch": 0.4818634157630707,
      "grad_norm": 0.6568914198752358,
      "learning_rate": 0.00017755878759118763,
      "loss": 12.4564,
      "step": 8849
    },
    {
      "epoch": 0.48191786975965367,
      "grad_norm": 0.6646572763387945,
      "learning_rate": 0.00017755322085578478,
      "loss": 12.2684,
      "step": 8850
    },
    {
      "epoch": 0.48197232375623666,
      "grad_norm": 0.6122037385746242,
      "learning_rate": 0.0001775476535173151,
      "loss": 12.4584,
      "step": 8851
    },
    {
      "epoch": 0.4820267777528197,
      "grad_norm": 0.6091843939075892,
      "learning_rate": 0.00017754208557582186,
      "loss": 12.4766,
      "step": 8852
    },
    {
      "epoch": 0.4820812317494027,
      "grad_norm": 0.7753738208277444,
      "learning_rate": 0.00017753651703134832,
      "loss": 12.3508,
      "step": 8853
    },
    {
      "epoch": 0.4821356857459857,
      "grad_norm": 0.9314633806434001,
      "learning_rate": 0.00017753094788393786,
      "loss": 12.5979,
      "step": 8854
    },
    {
      "epoch": 0.48219013974256875,
      "grad_norm": 0.6731684795342191,
      "learning_rate": 0.00017752537813363372,
      "loss": 12.4653,
      "step": 8855
    },
    {
      "epoch": 0.48224459373915174,
      "grad_norm": 0.6428856940797861,
      "learning_rate": 0.00017751980778047928,
      "loss": 12.5004,
      "step": 8856
    },
    {
      "epoch": 0.48229904773573473,
      "grad_norm": 0.59991219571617,
      "learning_rate": 0.00017751423682451777,
      "loss": 12.3395,
      "step": 8857
    },
    {
      "epoch": 0.4823535017323178,
      "grad_norm": 0.6919336816693169,
      "learning_rate": 0.0001775086652657926,
      "loss": 12.4303,
      "step": 8858
    },
    {
      "epoch": 0.4824079557289008,
      "grad_norm": 0.7949891419946377,
      "learning_rate": 0.000177503093104347,
      "loss": 12.3935,
      "step": 8859
    },
    {
      "epoch": 0.48246240972548377,
      "grad_norm": 0.5947020100389356,
      "learning_rate": 0.0001774975203402244,
      "loss": 12.3886,
      "step": 8860
    },
    {
      "epoch": 0.4825168637220668,
      "grad_norm": 0.5988251917040165,
      "learning_rate": 0.00017749194697346804,
      "loss": 12.342,
      "step": 8861
    },
    {
      "epoch": 0.4825713177186498,
      "grad_norm": 0.6354760855965604,
      "learning_rate": 0.00017748637300412135,
      "loss": 12.4184,
      "step": 8862
    },
    {
      "epoch": 0.4826257717152328,
      "grad_norm": 0.6000525681944205,
      "learning_rate": 0.00017748079843222758,
      "loss": 12.4177,
      "step": 8863
    },
    {
      "epoch": 0.48268022571181585,
      "grad_norm": 0.7000608948503445,
      "learning_rate": 0.00017747522325783016,
      "loss": 12.4286,
      "step": 8864
    },
    {
      "epoch": 0.48273467970839884,
      "grad_norm": 0.6118907742416242,
      "learning_rate": 0.0001774696474809724,
      "loss": 12.3678,
      "step": 8865
    },
    {
      "epoch": 0.4827891337049819,
      "grad_norm": 0.5942633424266189,
      "learning_rate": 0.00017746407110169767,
      "loss": 12.4994,
      "step": 8866
    },
    {
      "epoch": 0.4828435877015649,
      "grad_norm": 0.6539654214633613,
      "learning_rate": 0.00017745849412004937,
      "loss": 12.3596,
      "step": 8867
    },
    {
      "epoch": 0.4828980416981479,
      "grad_norm": 0.6677852512885757,
      "learning_rate": 0.00017745291653607076,
      "loss": 12.3255,
      "step": 8868
    },
    {
      "epoch": 0.4829524956947309,
      "grad_norm": 0.651571708656775,
      "learning_rate": 0.00017744733834980532,
      "loss": 12.327,
      "step": 8869
    },
    {
      "epoch": 0.4830069496913139,
      "grad_norm": 0.6349770322266751,
      "learning_rate": 0.0001774417595612964,
      "loss": 12.487,
      "step": 8870
    },
    {
      "epoch": 0.4830614036878969,
      "grad_norm": 0.6019710222800663,
      "learning_rate": 0.00017743618017058735,
      "loss": 12.4126,
      "step": 8871
    },
    {
      "epoch": 0.48311585768447995,
      "grad_norm": 0.7285677786372385,
      "learning_rate": 0.0001774306001777216,
      "loss": 12.4012,
      "step": 8872
    },
    {
      "epoch": 0.48317031168106295,
      "grad_norm": 0.5809777718047995,
      "learning_rate": 0.00017742501958274248,
      "loss": 12.5225,
      "step": 8873
    },
    {
      "epoch": 0.48322476567764594,
      "grad_norm": 0.6272483646028337,
      "learning_rate": 0.00017741943838569347,
      "loss": 12.4226,
      "step": 8874
    },
    {
      "epoch": 0.483279219674229,
      "grad_norm": 0.6755453012097429,
      "learning_rate": 0.0001774138565866179,
      "loss": 12.3687,
      "step": 8875
    },
    {
      "epoch": 0.483333673670812,
      "grad_norm": 0.6564591893347475,
      "learning_rate": 0.0001774082741855592,
      "loss": 12.3911,
      "step": 8876
    },
    {
      "epoch": 0.483388127667395,
      "grad_norm": 0.6373866759242696,
      "learning_rate": 0.00017740269118256076,
      "loss": 12.5277,
      "step": 8877
    },
    {
      "epoch": 0.483442581663978,
      "grad_norm": 0.5944427936719368,
      "learning_rate": 0.00017739710757766605,
      "loss": 12.5127,
      "step": 8878
    },
    {
      "epoch": 0.483497035660561,
      "grad_norm": 0.6324204783412024,
      "learning_rate": 0.00017739152337091843,
      "loss": 12.4911,
      "step": 8879
    },
    {
      "epoch": 0.483551489657144,
      "grad_norm": 0.6001119109761424,
      "learning_rate": 0.00017738593856236133,
      "loss": 12.384,
      "step": 8880
    },
    {
      "epoch": 0.48360594365372706,
      "grad_norm": 0.6966066832215112,
      "learning_rate": 0.00017738035315203822,
      "loss": 12.3972,
      "step": 8881
    },
    {
      "epoch": 0.48366039765031005,
      "grad_norm": 0.6212816073096596,
      "learning_rate": 0.0001773747671399925,
      "loss": 12.4002,
      "step": 8882
    },
    {
      "epoch": 0.48371485164689304,
      "grad_norm": 0.5724581880947716,
      "learning_rate": 0.0001773691805262676,
      "loss": 12.427,
      "step": 8883
    },
    {
      "epoch": 0.4837693056434761,
      "grad_norm": 0.6334340981535187,
      "learning_rate": 0.000177363593310907,
      "loss": 12.4921,
      "step": 8884
    },
    {
      "epoch": 0.4838237596400591,
      "grad_norm": 0.5685247708491485,
      "learning_rate": 0.00017735800549395413,
      "loss": 12.4575,
      "step": 8885
    },
    {
      "epoch": 0.4838782136366421,
      "grad_norm": 0.5689435269396318,
      "learning_rate": 0.00017735241707545241,
      "loss": 12.3163,
      "step": 8886
    },
    {
      "epoch": 0.4839326676332251,
      "grad_norm": 0.6779432269703239,
      "learning_rate": 0.00017734682805544533,
      "loss": 12.3748,
      "step": 8887
    },
    {
      "epoch": 0.4839871216298081,
      "grad_norm": 0.6796854341083812,
      "learning_rate": 0.00017734123843397636,
      "loss": 12.3531,
      "step": 8888
    },
    {
      "epoch": 0.4840415756263911,
      "grad_norm": 0.5849010737268502,
      "learning_rate": 0.00017733564821108895,
      "loss": 12.5429,
      "step": 8889
    },
    {
      "epoch": 0.48409602962297416,
      "grad_norm": 0.6728716617375359,
      "learning_rate": 0.00017733005738682656,
      "loss": 12.4443,
      "step": 8890
    },
    {
      "epoch": 0.48415048361955715,
      "grad_norm": 0.6222616297984809,
      "learning_rate": 0.00017732446596123268,
      "loss": 12.3582,
      "step": 8891
    },
    {
      "epoch": 0.48420493761614014,
      "grad_norm": 0.6025245468972172,
      "learning_rate": 0.00017731887393435076,
      "loss": 12.3947,
      "step": 8892
    },
    {
      "epoch": 0.4842593916127232,
      "grad_norm": 0.6012661543715989,
      "learning_rate": 0.00017731328130622434,
      "loss": 12.33,
      "step": 8893
    },
    {
      "epoch": 0.4843138456093062,
      "grad_norm": 0.6221467903127814,
      "learning_rate": 0.00017730768807689687,
      "loss": 12.4557,
      "step": 8894
    },
    {
      "epoch": 0.4843682996058892,
      "grad_norm": 0.7082201340878812,
      "learning_rate": 0.00017730209424641187,
      "loss": 12.5036,
      "step": 8895
    },
    {
      "epoch": 0.4844227536024722,
      "grad_norm": 0.5989029302995673,
      "learning_rate": 0.00017729649981481277,
      "loss": 12.4931,
      "step": 8896
    },
    {
      "epoch": 0.4844772075990552,
      "grad_norm": 0.6217807482682669,
      "learning_rate": 0.00017729090478214317,
      "loss": 12.4975,
      "step": 8897
    },
    {
      "epoch": 0.4845316615956382,
      "grad_norm": 0.7228601677700004,
      "learning_rate": 0.00017728530914844654,
      "loss": 12.4275,
      "step": 8898
    },
    {
      "epoch": 0.48458611559222126,
      "grad_norm": 0.5895667014499496,
      "learning_rate": 0.00017727971291376635,
      "loss": 12.3745,
      "step": 8899
    },
    {
      "epoch": 0.48464056958880425,
      "grad_norm": 0.5981741009577994,
      "learning_rate": 0.0001772741160781462,
      "loss": 12.4311,
      "step": 8900
    },
    {
      "epoch": 0.4846950235853873,
      "grad_norm": 0.632341233499706,
      "learning_rate": 0.00017726851864162952,
      "loss": 12.5217,
      "step": 8901
    },
    {
      "epoch": 0.4847494775819703,
      "grad_norm": 0.5661173025837181,
      "learning_rate": 0.0001772629206042599,
      "loss": 12.2706,
      "step": 8902
    },
    {
      "epoch": 0.4848039315785533,
      "grad_norm": 0.5429235944245946,
      "learning_rate": 0.00017725732196608086,
      "loss": 12.3377,
      "step": 8903
    },
    {
      "epoch": 0.48485838557513633,
      "grad_norm": 0.5779148781970922,
      "learning_rate": 0.00017725172272713588,
      "loss": 12.4942,
      "step": 8904
    },
    {
      "epoch": 0.4849128395717193,
      "grad_norm": 0.5910247841671594,
      "learning_rate": 0.0001772461228874686,
      "loss": 12.4021,
      "step": 8905
    },
    {
      "epoch": 0.4849672935683023,
      "grad_norm": 0.5960390491561129,
      "learning_rate": 0.00017724052244712251,
      "loss": 12.476,
      "step": 8906
    },
    {
      "epoch": 0.48502174756488536,
      "grad_norm": 0.5979535906650704,
      "learning_rate": 0.00017723492140614115,
      "loss": 12.1532,
      "step": 8907
    },
    {
      "epoch": 0.48507620156146836,
      "grad_norm": 0.5835970639327157,
      "learning_rate": 0.0001772293197645681,
      "loss": 12.426,
      "step": 8908
    },
    {
      "epoch": 0.48513065555805135,
      "grad_norm": 0.578047193217132,
      "learning_rate": 0.00017722371752244687,
      "loss": 12.2607,
      "step": 8909
    },
    {
      "epoch": 0.4851851095546344,
      "grad_norm": 0.6360885630082748,
      "learning_rate": 0.00017721811467982112,
      "loss": 12.4493,
      "step": 8910
    },
    {
      "epoch": 0.4852395635512174,
      "grad_norm": 0.5881949679055104,
      "learning_rate": 0.0001772125112367343,
      "loss": 12.3724,
      "step": 8911
    },
    {
      "epoch": 0.4852940175478004,
      "grad_norm": 0.6163484189248782,
      "learning_rate": 0.0001772069071932301,
      "loss": 12.4753,
      "step": 8912
    },
    {
      "epoch": 0.48534847154438343,
      "grad_norm": 0.5804625073859097,
      "learning_rate": 0.000177201302549352,
      "loss": 12.4493,
      "step": 8913
    },
    {
      "epoch": 0.4854029255409664,
      "grad_norm": 0.5913429595420551,
      "learning_rate": 0.00017719569730514367,
      "loss": 12.4647,
      "step": 8914
    },
    {
      "epoch": 0.4854573795375494,
      "grad_norm": 0.6458546645180917,
      "learning_rate": 0.00017719009146064863,
      "loss": 12.402,
      "step": 8915
    },
    {
      "epoch": 0.48551183353413246,
      "grad_norm": 0.6266882741409003,
      "learning_rate": 0.00017718448501591048,
      "loss": 12.427,
      "step": 8916
    },
    {
      "epoch": 0.48556628753071546,
      "grad_norm": 0.7313824026603742,
      "learning_rate": 0.00017717887797097284,
      "loss": 12.3917,
      "step": 8917
    },
    {
      "epoch": 0.48562074152729845,
      "grad_norm": 0.6531470914819183,
      "learning_rate": 0.0001771732703258793,
      "loss": 12.4148,
      "step": 8918
    },
    {
      "epoch": 0.4856751955238815,
      "grad_norm": 0.5753462283529436,
      "learning_rate": 0.00017716766208067348,
      "loss": 12.2808,
      "step": 8919
    },
    {
      "epoch": 0.4857296495204645,
      "grad_norm": 0.6291815969960752,
      "learning_rate": 0.00017716205323539897,
      "loss": 12.4196,
      "step": 8920
    },
    {
      "epoch": 0.4857841035170475,
      "grad_norm": 0.6655582182857843,
      "learning_rate": 0.00017715644379009938,
      "loss": 12.6431,
      "step": 8921
    },
    {
      "epoch": 0.48583855751363053,
      "grad_norm": 0.6351500978892104,
      "learning_rate": 0.00017715083374481835,
      "loss": 12.1745,
      "step": 8922
    },
    {
      "epoch": 0.4858930115102135,
      "grad_norm": 0.7014680673123271,
      "learning_rate": 0.00017714522309959953,
      "loss": 12.5735,
      "step": 8923
    },
    {
      "epoch": 0.4859474655067965,
      "grad_norm": 0.5961403293878741,
      "learning_rate": 0.0001771396118544865,
      "loss": 12.3263,
      "step": 8924
    },
    {
      "epoch": 0.48600191950337956,
      "grad_norm": 0.6381696181668911,
      "learning_rate": 0.00017713400000952292,
      "loss": 12.5789,
      "step": 8925
    },
    {
      "epoch": 0.48605637349996256,
      "grad_norm": 0.5781356158913226,
      "learning_rate": 0.00017712838756475237,
      "loss": 12.3219,
      "step": 8926
    },
    {
      "epoch": 0.48611082749654555,
      "grad_norm": 0.6133175971860488,
      "learning_rate": 0.0001771227745202186,
      "loss": 12.5355,
      "step": 8927
    },
    {
      "epoch": 0.4861652814931286,
      "grad_norm": 0.6944077918723269,
      "learning_rate": 0.00017711716087596518,
      "loss": 12.4948,
      "step": 8928
    },
    {
      "epoch": 0.4862197354897116,
      "grad_norm": 0.6446646845681596,
      "learning_rate": 0.00017711154663203578,
      "loss": 12.5193,
      "step": 8929
    },
    {
      "epoch": 0.4862741894862946,
      "grad_norm": 0.6286728175168206,
      "learning_rate": 0.0001771059317884741,
      "loss": 12.5637,
      "step": 8930
    },
    {
      "epoch": 0.48632864348287763,
      "grad_norm": 0.6662026021201488,
      "learning_rate": 0.0001771003163453237,
      "loss": 12.4626,
      "step": 8931
    },
    {
      "epoch": 0.4863830974794606,
      "grad_norm": 0.6301674124609281,
      "learning_rate": 0.00017709470030262834,
      "loss": 12.5071,
      "step": 8932
    },
    {
      "epoch": 0.4864375514760437,
      "grad_norm": 0.5735200957479709,
      "learning_rate": 0.00017708908366043168,
      "loss": 12.4277,
      "step": 8933
    },
    {
      "epoch": 0.48649200547262667,
      "grad_norm": 0.616838235274588,
      "learning_rate": 0.00017708346641877735,
      "loss": 12.4254,
      "step": 8934
    },
    {
      "epoch": 0.48654645946920966,
      "grad_norm": 0.6986291067091521,
      "learning_rate": 0.00017707784857770906,
      "loss": 12.5306,
      "step": 8935
    },
    {
      "epoch": 0.4866009134657927,
      "grad_norm": 0.6414495094248273,
      "learning_rate": 0.00017707223013727053,
      "loss": 12.438,
      "step": 8936
    },
    {
      "epoch": 0.4866553674623757,
      "grad_norm": 0.7109706319146375,
      "learning_rate": 0.00017706661109750536,
      "loss": 12.3861,
      "step": 8937
    },
    {
      "epoch": 0.4867098214589587,
      "grad_norm": 0.6688809895812369,
      "learning_rate": 0.00017706099145845734,
      "loss": 12.3729,
      "step": 8938
    },
    {
      "epoch": 0.48676427545554174,
      "grad_norm": 0.6153038549431864,
      "learning_rate": 0.0001770553712201701,
      "loss": 12.4929,
      "step": 8939
    },
    {
      "epoch": 0.48681872945212473,
      "grad_norm": 0.7329807332996898,
      "learning_rate": 0.0001770497503826874,
      "loss": 12.4184,
      "step": 8940
    },
    {
      "epoch": 0.4868731834487077,
      "grad_norm": 0.7084924363855114,
      "learning_rate": 0.0001770441289460529,
      "loss": 12.3417,
      "step": 8941
    },
    {
      "epoch": 0.4869276374452908,
      "grad_norm": 0.6815236874363345,
      "learning_rate": 0.00017703850691031035,
      "loss": 12.4823,
      "step": 8942
    },
    {
      "epoch": 0.48698209144187377,
      "grad_norm": 0.577985264202812,
      "learning_rate": 0.00017703288427550342,
      "loss": 12.4893,
      "step": 8943
    },
    {
      "epoch": 0.48703654543845676,
      "grad_norm": 0.5997747016250154,
      "learning_rate": 0.0001770272610416759,
      "loss": 12.4236,
      "step": 8944
    },
    {
      "epoch": 0.4870909994350398,
      "grad_norm": 0.6303074713542345,
      "learning_rate": 0.00017702163720887144,
      "loss": 12.3766,
      "step": 8945
    },
    {
      "epoch": 0.4871454534316228,
      "grad_norm": 0.6148182246803541,
      "learning_rate": 0.00017701601277713382,
      "loss": 12.4035,
      "step": 8946
    },
    {
      "epoch": 0.4871999074282058,
      "grad_norm": 0.6129823580686586,
      "learning_rate": 0.0001770103877465068,
      "loss": 12.3825,
      "step": 8947
    },
    {
      "epoch": 0.48725436142478884,
      "grad_norm": 0.623778798045754,
      "learning_rate": 0.00017700476211703406,
      "loss": 12.4081,
      "step": 8948
    },
    {
      "epoch": 0.48730881542137183,
      "grad_norm": 0.6131756723701572,
      "learning_rate": 0.0001769991358887594,
      "loss": 12.346,
      "step": 8949
    },
    {
      "epoch": 0.4873632694179548,
      "grad_norm": 0.6497155540267017,
      "learning_rate": 0.00017699350906172655,
      "loss": 12.4192,
      "step": 8950
    },
    {
      "epoch": 0.4874177234145379,
      "grad_norm": 0.7017317517650884,
      "learning_rate": 0.00017698788163597923,
      "loss": 12.6197,
      "step": 8951
    },
    {
      "epoch": 0.48747217741112087,
      "grad_norm": 0.5495695820318853,
      "learning_rate": 0.00017698225361156126,
      "loss": 12.4219,
      "step": 8952
    },
    {
      "epoch": 0.48752663140770386,
      "grad_norm": 0.600004068120246,
      "learning_rate": 0.00017697662498851634,
      "loss": 12.4167,
      "step": 8953
    },
    {
      "epoch": 0.4875810854042869,
      "grad_norm": 0.5677519445784427,
      "learning_rate": 0.0001769709957668883,
      "loss": 12.4206,
      "step": 8954
    },
    {
      "epoch": 0.4876355394008699,
      "grad_norm": 0.6933956699220511,
      "learning_rate": 0.00017696536594672093,
      "loss": 12.4454,
      "step": 8955
    },
    {
      "epoch": 0.4876899933974529,
      "grad_norm": 0.5852886126317223,
      "learning_rate": 0.0001769597355280579,
      "loss": 12.4294,
      "step": 8956
    },
    {
      "epoch": 0.48774444739403594,
      "grad_norm": 0.6707874767566225,
      "learning_rate": 0.00017695410451094309,
      "loss": 12.3555,
      "step": 8957
    },
    {
      "epoch": 0.48779890139061893,
      "grad_norm": 0.6275576481656014,
      "learning_rate": 0.00017694847289542027,
      "loss": 12.3984,
      "step": 8958
    },
    {
      "epoch": 0.4878533553872019,
      "grad_norm": 0.7453159192884955,
      "learning_rate": 0.0001769428406815332,
      "loss": 12.3675,
      "step": 8959
    },
    {
      "epoch": 0.487907809383785,
      "grad_norm": 0.7149271919206965,
      "learning_rate": 0.0001769372078693257,
      "loss": 12.4132,
      "step": 8960
    },
    {
      "epoch": 0.48796226338036797,
      "grad_norm": 0.6762737105423414,
      "learning_rate": 0.00017693157445884157,
      "loss": 12.3941,
      "step": 8961
    },
    {
      "epoch": 0.48801671737695096,
      "grad_norm": 0.6419090179100578,
      "learning_rate": 0.00017692594045012463,
      "loss": 12.304,
      "step": 8962
    },
    {
      "epoch": 0.488071171373534,
      "grad_norm": 0.6168353578562813,
      "learning_rate": 0.00017692030584321862,
      "loss": 12.43,
      "step": 8963
    },
    {
      "epoch": 0.488125625370117,
      "grad_norm": 0.7407722298231637,
      "learning_rate": 0.0001769146706381675,
      "loss": 12.4277,
      "step": 8964
    },
    {
      "epoch": 0.4881800793667,
      "grad_norm": 0.6123927513441908,
      "learning_rate": 0.00017690903483501494,
      "loss": 12.3529,
      "step": 8965
    },
    {
      "epoch": 0.48823453336328304,
      "grad_norm": 0.6927569113790386,
      "learning_rate": 0.00017690339843380487,
      "loss": 12.3037,
      "step": 8966
    },
    {
      "epoch": 0.48828898735986603,
      "grad_norm": 0.7282325256842512,
      "learning_rate": 0.00017689776143458102,
      "loss": 12.4561,
      "step": 8967
    },
    {
      "epoch": 0.4883434413564491,
      "grad_norm": 0.5931461718947436,
      "learning_rate": 0.00017689212383738734,
      "loss": 12.5021,
      "step": 8968
    },
    {
      "epoch": 0.4883978953530321,
      "grad_norm": 0.6057001158655123,
      "learning_rate": 0.00017688648564226757,
      "loss": 12.3242,
      "step": 8969
    },
    {
      "epoch": 0.48845234934961507,
      "grad_norm": 0.6998284602710015,
      "learning_rate": 0.00017688084684926563,
      "loss": 12.3638,
      "step": 8970
    },
    {
      "epoch": 0.4885068033461981,
      "grad_norm": 0.6223628650538694,
      "learning_rate": 0.0001768752074584253,
      "loss": 12.3924,
      "step": 8971
    },
    {
      "epoch": 0.4885612573427811,
      "grad_norm": 0.6172249769850962,
      "learning_rate": 0.0001768695674697905,
      "loss": 12.3722,
      "step": 8972
    },
    {
      "epoch": 0.4886157113393641,
      "grad_norm": 0.6434240293370114,
      "learning_rate": 0.00017686392688340502,
      "loss": 12.4554,
      "step": 8973
    },
    {
      "epoch": 0.48867016533594715,
      "grad_norm": 0.6284914832648406,
      "learning_rate": 0.00017685828569931277,
      "loss": 12.3951,
      "step": 8974
    },
    {
      "epoch": 0.48872461933253014,
      "grad_norm": 0.6584633649303565,
      "learning_rate": 0.00017685264391755756,
      "loss": 12.4403,
      "step": 8975
    },
    {
      "epoch": 0.48877907332911313,
      "grad_norm": 0.6239764584426922,
      "learning_rate": 0.00017684700153818334,
      "loss": 12.445,
      "step": 8976
    },
    {
      "epoch": 0.4888335273256962,
      "grad_norm": 0.604252889534663,
      "learning_rate": 0.00017684135856123394,
      "loss": 12.3662,
      "step": 8977
    },
    {
      "epoch": 0.4888879813222792,
      "grad_norm": 0.6768630849815542,
      "learning_rate": 0.00017683571498675326,
      "loss": 12.4315,
      "step": 8978
    },
    {
      "epoch": 0.48894243531886217,
      "grad_norm": 0.561773748949832,
      "learning_rate": 0.0001768300708147852,
      "loss": 12.4046,
      "step": 8979
    },
    {
      "epoch": 0.4889968893154452,
      "grad_norm": 0.6191608849730387,
      "learning_rate": 0.00017682442604537358,
      "loss": 12.3913,
      "step": 8980
    },
    {
      "epoch": 0.4890513433120282,
      "grad_norm": 0.668467651031518,
      "learning_rate": 0.00017681878067856235,
      "loss": 12.427,
      "step": 8981
    },
    {
      "epoch": 0.4891057973086112,
      "grad_norm": 0.5852875868444413,
      "learning_rate": 0.0001768131347143954,
      "loss": 12.2236,
      "step": 8982
    },
    {
      "epoch": 0.48916025130519425,
      "grad_norm": 0.6495359491080023,
      "learning_rate": 0.00017680748815291662,
      "loss": 12.2845,
      "step": 8983
    },
    {
      "epoch": 0.48921470530177724,
      "grad_norm": 0.7043442632709124,
      "learning_rate": 0.00017680184099416995,
      "loss": 12.3416,
      "step": 8984
    },
    {
      "epoch": 0.48926915929836023,
      "grad_norm": 0.6821440893967539,
      "learning_rate": 0.0001767961932381993,
      "loss": 12.488,
      "step": 8985
    },
    {
      "epoch": 0.4893236132949433,
      "grad_norm": 0.7518808059202897,
      "learning_rate": 0.00017679054488504856,
      "loss": 12.4599,
      "step": 8986
    },
    {
      "epoch": 0.4893780672915263,
      "grad_norm": 0.6489067588764982,
      "learning_rate": 0.00017678489593476164,
      "loss": 12.3983,
      "step": 8987
    },
    {
      "epoch": 0.48943252128810927,
      "grad_norm": 0.5932349853729175,
      "learning_rate": 0.0001767792463873825,
      "loss": 12.3459,
      "step": 8988
    },
    {
      "epoch": 0.4894869752846923,
      "grad_norm": 0.6926889837411475,
      "learning_rate": 0.0001767735962429551,
      "loss": 12.4644,
      "step": 8989
    },
    {
      "epoch": 0.4895414292812753,
      "grad_norm": 0.6420778151456797,
      "learning_rate": 0.0001767679455015233,
      "loss": 12.3545,
      "step": 8990
    },
    {
      "epoch": 0.4895958832778583,
      "grad_norm": 0.6720821949354343,
      "learning_rate": 0.0001767622941631311,
      "loss": 12.4321,
      "step": 8991
    },
    {
      "epoch": 0.48965033727444135,
      "grad_norm": 0.6577540205055701,
      "learning_rate": 0.00017675664222782244,
      "loss": 12.503,
      "step": 8992
    },
    {
      "epoch": 0.48970479127102434,
      "grad_norm": 0.6245459109383175,
      "learning_rate": 0.00017675098969564124,
      "loss": 12.4008,
      "step": 8993
    },
    {
      "epoch": 0.48975924526760733,
      "grad_norm": 0.6799189419701791,
      "learning_rate": 0.0001767453365666315,
      "loss": 12.566,
      "step": 8994
    },
    {
      "epoch": 0.4898136992641904,
      "grad_norm": 0.6190799234397225,
      "learning_rate": 0.0001767396828408371,
      "loss": 12.4133,
      "step": 8995
    },
    {
      "epoch": 0.4898681532607734,
      "grad_norm": 0.5648051502723331,
      "learning_rate": 0.0001767340285183021,
      "loss": 12.3986,
      "step": 8996
    },
    {
      "epoch": 0.48992260725735637,
      "grad_norm": 0.7832200571204988,
      "learning_rate": 0.0001767283735990704,
      "loss": 12.4317,
      "step": 8997
    },
    {
      "epoch": 0.4899770612539394,
      "grad_norm": 0.5896019405323225,
      "learning_rate": 0.00017672271808318605,
      "loss": 12.3887,
      "step": 8998
    },
    {
      "epoch": 0.4900315152505224,
      "grad_norm": 0.5847485314578779,
      "learning_rate": 0.0001767170619706929,
      "loss": 12.2743,
      "step": 8999
    },
    {
      "epoch": 0.49008596924710546,
      "grad_norm": 0.6561473704564957,
      "learning_rate": 0.00017671140526163506,
      "loss": 12.3998,
      "step": 9000
    },
    {
      "epoch": 0.49014042324368845,
      "grad_norm": 0.6031748325909405,
      "learning_rate": 0.00017670574795605645,
      "loss": 12.3966,
      "step": 9001
    },
    {
      "epoch": 0.49019487724027144,
      "grad_norm": 0.5969040835414722,
      "learning_rate": 0.0001767000900540011,
      "loss": 12.4078,
      "step": 9002
    },
    {
      "epoch": 0.4902493312368545,
      "grad_norm": 0.7009410013108908,
      "learning_rate": 0.00017669443155551298,
      "loss": 12.3908,
      "step": 9003
    },
    {
      "epoch": 0.4903037852334375,
      "grad_norm": 0.6204813872425611,
      "learning_rate": 0.00017668877246063608,
      "loss": 12.3522,
      "step": 9004
    },
    {
      "epoch": 0.4903582392300205,
      "grad_norm": 0.6725773712069133,
      "learning_rate": 0.00017668311276941445,
      "loss": 12.4103,
      "step": 9005
    },
    {
      "epoch": 0.4904126932266035,
      "grad_norm": 0.5686997026451569,
      "learning_rate": 0.00017667745248189207,
      "loss": 12.1809,
      "step": 9006
    },
    {
      "epoch": 0.4904671472231865,
      "grad_norm": 0.6337705792607752,
      "learning_rate": 0.00017667179159811295,
      "loss": 12.3331,
      "step": 9007
    },
    {
      "epoch": 0.4905216012197695,
      "grad_norm": 0.6859319984556346,
      "learning_rate": 0.00017666613011812113,
      "loss": 12.4774,
      "step": 9008
    },
    {
      "epoch": 0.49057605521635256,
      "grad_norm": 0.68712970750493,
      "learning_rate": 0.0001766604680419606,
      "loss": 12.4661,
      "step": 9009
    },
    {
      "epoch": 0.49063050921293555,
      "grad_norm": 0.6337798565063828,
      "learning_rate": 0.00017665480536967546,
      "loss": 12.5433,
      "step": 9010
    },
    {
      "epoch": 0.49068496320951854,
      "grad_norm": 0.9691188287958652,
      "learning_rate": 0.00017664914210130966,
      "loss": 12.5172,
      "step": 9011
    },
    {
      "epoch": 0.4907394172061016,
      "grad_norm": 0.624849717001331,
      "learning_rate": 0.0001766434782369073,
      "loss": 12.4614,
      "step": 9012
    },
    {
      "epoch": 0.4907938712026846,
      "grad_norm": 0.6255963762613319,
      "learning_rate": 0.0001766378137765124,
      "loss": 12.4613,
      "step": 9013
    },
    {
      "epoch": 0.4908483251992676,
      "grad_norm": 0.6966420063072758,
      "learning_rate": 0.000176632148720169,
      "loss": 12.3903,
      "step": 9014
    },
    {
      "epoch": 0.4909027791958506,
      "grad_norm": 0.6181596567503763,
      "learning_rate": 0.00017662648306792118,
      "loss": 12.4234,
      "step": 9015
    },
    {
      "epoch": 0.4909572331924336,
      "grad_norm": 0.6472714093995697,
      "learning_rate": 0.00017662081681981296,
      "loss": 12.4579,
      "step": 9016
    },
    {
      "epoch": 0.4910116871890166,
      "grad_norm": 0.6307280492239105,
      "learning_rate": 0.0001766151499758884,
      "loss": 12.3809,
      "step": 9017
    },
    {
      "epoch": 0.49106614118559966,
      "grad_norm": 0.7145252730355572,
      "learning_rate": 0.0001766094825361916,
      "loss": 12.4365,
      "step": 9018
    },
    {
      "epoch": 0.49112059518218265,
      "grad_norm": 0.5779293889447368,
      "learning_rate": 0.0001766038145007666,
      "loss": 12.3208,
      "step": 9019
    },
    {
      "epoch": 0.49117504917876564,
      "grad_norm": 0.5970030202119183,
      "learning_rate": 0.00017659814586965753,
      "loss": 12.4748,
      "step": 9020
    },
    {
      "epoch": 0.4912295031753487,
      "grad_norm": 0.6713233594796585,
      "learning_rate": 0.00017659247664290843,
      "loss": 12.5403,
      "step": 9021
    },
    {
      "epoch": 0.4912839571719317,
      "grad_norm": 0.6389280094245742,
      "learning_rate": 0.00017658680682056336,
      "loss": 12.3417,
      "step": 9022
    },
    {
      "epoch": 0.4913384111685147,
      "grad_norm": 0.5997464523487829,
      "learning_rate": 0.00017658113640266646,
      "loss": 12.2971,
      "step": 9023
    },
    {
      "epoch": 0.4913928651650977,
      "grad_norm": 0.5998221515239658,
      "learning_rate": 0.00017657546538926176,
      "loss": 12.4103,
      "step": 9024
    },
    {
      "epoch": 0.4914473191616807,
      "grad_norm": 0.6621423715721438,
      "learning_rate": 0.00017656979378039345,
      "loss": 12.2766,
      "step": 9025
    },
    {
      "epoch": 0.4915017731582637,
      "grad_norm": 0.641780152243786,
      "learning_rate": 0.00017656412157610553,
      "loss": 12.4449,
      "step": 9026
    },
    {
      "epoch": 0.49155622715484676,
      "grad_norm": 0.630345320868909,
      "learning_rate": 0.00017655844877644222,
      "loss": 12.3306,
      "step": 9027
    },
    {
      "epoch": 0.49161068115142975,
      "grad_norm": 0.6483735385691813,
      "learning_rate": 0.00017655277538144754,
      "loss": 12.376,
      "step": 9028
    },
    {
      "epoch": 0.49166513514801274,
      "grad_norm": 0.6309651726102987,
      "learning_rate": 0.00017654710139116563,
      "loss": 12.5355,
      "step": 9029
    },
    {
      "epoch": 0.4917195891445958,
      "grad_norm": 0.600940677237466,
      "learning_rate": 0.00017654142680564064,
      "loss": 12.334,
      "step": 9030
    },
    {
      "epoch": 0.4917740431411788,
      "grad_norm": 0.5921764876537235,
      "learning_rate": 0.0001765357516249167,
      "loss": 12.2664,
      "step": 9031
    },
    {
      "epoch": 0.4918284971377618,
      "grad_norm": 0.544108968363809,
      "learning_rate": 0.0001765300758490379,
      "loss": 12.4143,
      "step": 9032
    },
    {
      "epoch": 0.4918829511343448,
      "grad_norm": 0.7068826009519629,
      "learning_rate": 0.00017652439947804838,
      "loss": 12.3461,
      "step": 9033
    },
    {
      "epoch": 0.4919374051309278,
      "grad_norm": 0.6469820503926543,
      "learning_rate": 0.00017651872251199229,
      "loss": 12.3202,
      "step": 9034
    },
    {
      "epoch": 0.49199185912751087,
      "grad_norm": 0.6601942119120693,
      "learning_rate": 0.0001765130449509138,
      "loss": 12.4268,
      "step": 9035
    },
    {
      "epoch": 0.49204631312409386,
      "grad_norm": 0.5950297306819458,
      "learning_rate": 0.00017650736679485706,
      "loss": 12.4843,
      "step": 9036
    },
    {
      "epoch": 0.49210076712067685,
      "grad_norm": 0.6617665756952725,
      "learning_rate": 0.00017650168804386618,
      "loss": 12.3563,
      "step": 9037
    },
    {
      "epoch": 0.4921552211172599,
      "grad_norm": 0.6127139296457449,
      "learning_rate": 0.00017649600869798534,
      "loss": 12.4384,
      "step": 9038
    },
    {
      "epoch": 0.4922096751138429,
      "grad_norm": 0.6596836507747746,
      "learning_rate": 0.00017649032875725873,
      "loss": 12.3869,
      "step": 9039
    },
    {
      "epoch": 0.4922641291104259,
      "grad_norm": 0.7322953833439002,
      "learning_rate": 0.00017648464822173047,
      "loss": 12.4327,
      "step": 9040
    },
    {
      "epoch": 0.49231858310700893,
      "grad_norm": 0.6495668511280798,
      "learning_rate": 0.0001764789670914448,
      "loss": 12.3549,
      "step": 9041
    },
    {
      "epoch": 0.4923730371035919,
      "grad_norm": 0.6118572627218264,
      "learning_rate": 0.00017647328536644582,
      "loss": 12.4796,
      "step": 9042
    },
    {
      "epoch": 0.4924274911001749,
      "grad_norm": 0.8118395035290029,
      "learning_rate": 0.00017646760304677775,
      "loss": 12.3732,
      "step": 9043
    },
    {
      "epoch": 0.49248194509675797,
      "grad_norm": 0.6024457061523186,
      "learning_rate": 0.0001764619201324848,
      "loss": 12.3939,
      "step": 9044
    },
    {
      "epoch": 0.49253639909334096,
      "grad_norm": 0.622607292398364,
      "learning_rate": 0.0001764562366236111,
      "loss": 12.4687,
      "step": 9045
    },
    {
      "epoch": 0.49259085308992395,
      "grad_norm": 0.6375419946487572,
      "learning_rate": 0.00017645055252020093,
      "loss": 12.4258,
      "step": 9046
    },
    {
      "epoch": 0.492645307086507,
      "grad_norm": 0.6396330427611137,
      "learning_rate": 0.0001764448678222984,
      "loss": 12.4937,
      "step": 9047
    },
    {
      "epoch": 0.49269976108309,
      "grad_norm": 0.6192244845607952,
      "learning_rate": 0.00017643918252994778,
      "loss": 12.1713,
      "step": 9048
    },
    {
      "epoch": 0.492754215079673,
      "grad_norm": 0.6223106593402868,
      "learning_rate": 0.00017643349664319327,
      "loss": 12.5063,
      "step": 9049
    },
    {
      "epoch": 0.49280866907625603,
      "grad_norm": 0.6008700229121132,
      "learning_rate": 0.00017642781016207902,
      "loss": 12.4324,
      "step": 9050
    },
    {
      "epoch": 0.492863123072839,
      "grad_norm": 0.6685584067925143,
      "learning_rate": 0.00017642212308664935,
      "loss": 12.4043,
      "step": 9051
    },
    {
      "epoch": 0.492917577069422,
      "grad_norm": 0.652765870070671,
      "learning_rate": 0.00017641643541694841,
      "loss": 12.2954,
      "step": 9052
    },
    {
      "epoch": 0.49297203106600507,
      "grad_norm": 0.5641445539183918,
      "learning_rate": 0.0001764107471530205,
      "loss": 12.2922,
      "step": 9053
    },
    {
      "epoch": 0.49302648506258806,
      "grad_norm": 0.7735158333144617,
      "learning_rate": 0.00017640505829490975,
      "loss": 12.4085,
      "step": 9054
    },
    {
      "epoch": 0.49308093905917105,
      "grad_norm": 0.7470430066736975,
      "learning_rate": 0.00017639936884266045,
      "loss": 12.349,
      "step": 9055
    },
    {
      "epoch": 0.4931353930557541,
      "grad_norm": 0.7206711098167883,
      "learning_rate": 0.00017639367879631687,
      "loss": 12.299,
      "step": 9056
    },
    {
      "epoch": 0.4931898470523371,
      "grad_norm": 0.7371347771651026,
      "learning_rate": 0.00017638798815592326,
      "loss": 12.4882,
      "step": 9057
    },
    {
      "epoch": 0.4932443010489201,
      "grad_norm": 0.595033333497349,
      "learning_rate": 0.0001763822969215238,
      "loss": 12.3871,
      "step": 9058
    },
    {
      "epoch": 0.49329875504550313,
      "grad_norm": 0.8054972406286517,
      "learning_rate": 0.0001763766050931628,
      "loss": 12.405,
      "step": 9059
    },
    {
      "epoch": 0.4933532090420861,
      "grad_norm": 0.6098101737336612,
      "learning_rate": 0.00017637091267088453,
      "loss": 12.4194,
      "step": 9060
    },
    {
      "epoch": 0.4934076630386691,
      "grad_norm": 0.838036559345574,
      "learning_rate": 0.00017636521965473323,
      "loss": 12.4134,
      "step": 9061
    },
    {
      "epoch": 0.49346211703525217,
      "grad_norm": 0.6159087356357568,
      "learning_rate": 0.00017635952604475318,
      "loss": 12.4425,
      "step": 9062
    },
    {
      "epoch": 0.49351657103183516,
      "grad_norm": 0.5772658512197129,
      "learning_rate": 0.0001763538318409886,
      "loss": 12.3848,
      "step": 9063
    },
    {
      "epoch": 0.49357102502841815,
      "grad_norm": 0.648307453696023,
      "learning_rate": 0.00017634813704348385,
      "loss": 12.4,
      "step": 9064
    },
    {
      "epoch": 0.4936254790250012,
      "grad_norm": 0.797999057121607,
      "learning_rate": 0.0001763424416522832,
      "loss": 12.4749,
      "step": 9065
    },
    {
      "epoch": 0.4936799330215842,
      "grad_norm": 0.6154663448292866,
      "learning_rate": 0.00017633674566743093,
      "loss": 12.4569,
      "step": 9066
    },
    {
      "epoch": 0.49373438701816724,
      "grad_norm": 0.6053841272700453,
      "learning_rate": 0.00017633104908897128,
      "loss": 12.4407,
      "step": 9067
    },
    {
      "epoch": 0.49378884101475023,
      "grad_norm": 0.6125434288709458,
      "learning_rate": 0.00017632535191694864,
      "loss": 12.3684,
      "step": 9068
    },
    {
      "epoch": 0.4938432950113332,
      "grad_norm": 0.5478148758765967,
      "learning_rate": 0.0001763196541514072,
      "loss": 12.466,
      "step": 9069
    },
    {
      "epoch": 0.4938977490079163,
      "grad_norm": 0.6767775722685464,
      "learning_rate": 0.00017631395579239138,
      "loss": 12.4683,
      "step": 9070
    },
    {
      "epoch": 0.49395220300449927,
      "grad_norm": 0.6565295047603227,
      "learning_rate": 0.00017630825683994546,
      "loss": 12.4402,
      "step": 9071
    },
    {
      "epoch": 0.49400665700108226,
      "grad_norm": 0.6186660536186829,
      "learning_rate": 0.0001763025572941137,
      "loss": 12.4184,
      "step": 9072
    },
    {
      "epoch": 0.4940611109976653,
      "grad_norm": 0.574512487793142,
      "learning_rate": 0.00017629685715494047,
      "loss": 12.3434,
      "step": 9073
    },
    {
      "epoch": 0.4941155649942483,
      "grad_norm": 0.5786501159465615,
      "learning_rate": 0.00017629115642247008,
      "loss": 12.2469,
      "step": 9074
    },
    {
      "epoch": 0.4941700189908313,
      "grad_norm": 0.6344422448863354,
      "learning_rate": 0.00017628545509674686,
      "loss": 12.3322,
      "step": 9075
    },
    {
      "epoch": 0.49422447298741434,
      "grad_norm": 0.5947827175109859,
      "learning_rate": 0.00017627975317781514,
      "loss": 12.3867,
      "step": 9076
    },
    {
      "epoch": 0.49427892698399734,
      "grad_norm": 0.7234156368573843,
      "learning_rate": 0.00017627405066571926,
      "loss": 12.4595,
      "step": 9077
    },
    {
      "epoch": 0.49433338098058033,
      "grad_norm": 0.5716608289086761,
      "learning_rate": 0.0001762683475605036,
      "loss": 12.3055,
      "step": 9078
    },
    {
      "epoch": 0.4943878349771634,
      "grad_norm": 0.6516898534652547,
      "learning_rate": 0.0001762626438622125,
      "loss": 12.2219,
      "step": 9079
    },
    {
      "epoch": 0.49444228897374637,
      "grad_norm": 0.5531591292155007,
      "learning_rate": 0.00017625693957089023,
      "loss": 12.2898,
      "step": 9080
    },
    {
      "epoch": 0.49449674297032936,
      "grad_norm": 0.6540053412369687,
      "learning_rate": 0.00017625123468658128,
      "loss": 12.3422,
      "step": 9081
    },
    {
      "epoch": 0.4945511969669124,
      "grad_norm": 0.5822114864652976,
      "learning_rate": 0.00017624552920932987,
      "loss": 12.3534,
      "step": 9082
    },
    {
      "epoch": 0.4946056509634954,
      "grad_norm": 0.647468952988309,
      "learning_rate": 0.00017623982313918048,
      "loss": 12.2724,
      "step": 9083
    },
    {
      "epoch": 0.4946601049600784,
      "grad_norm": 0.6538053463120066,
      "learning_rate": 0.00017623411647617744,
      "loss": 12.4656,
      "step": 9084
    },
    {
      "epoch": 0.49471455895666144,
      "grad_norm": 0.5912115377511528,
      "learning_rate": 0.00017622840922036512,
      "loss": 12.3684,
      "step": 9085
    },
    {
      "epoch": 0.49476901295324444,
      "grad_norm": 0.5638563271249786,
      "learning_rate": 0.00017622270137178793,
      "loss": 12.2539,
      "step": 9086
    },
    {
      "epoch": 0.49482346694982743,
      "grad_norm": 0.6885206221172212,
      "learning_rate": 0.00017621699293049017,
      "loss": 12.5263,
      "step": 9087
    },
    {
      "epoch": 0.4948779209464105,
      "grad_norm": 0.6695687649039862,
      "learning_rate": 0.00017621128389651633,
      "loss": 12.4807,
      "step": 9088
    },
    {
      "epoch": 0.49493237494299347,
      "grad_norm": 0.6267441366242752,
      "learning_rate": 0.00017620557426991078,
      "loss": 12.4484,
      "step": 9089
    },
    {
      "epoch": 0.49498682893957646,
      "grad_norm": 0.7064926132108872,
      "learning_rate": 0.00017619986405071787,
      "loss": 12.4488,
      "step": 9090
    },
    {
      "epoch": 0.4950412829361595,
      "grad_norm": 0.56968390851137,
      "learning_rate": 0.00017619415323898206,
      "loss": 12.3079,
      "step": 9091
    },
    {
      "epoch": 0.4950957369327425,
      "grad_norm": 0.6304071941633153,
      "learning_rate": 0.00017618844183474774,
      "loss": 12.512,
      "step": 9092
    },
    {
      "epoch": 0.4951501909293255,
      "grad_norm": 0.6627967334696621,
      "learning_rate": 0.0001761827298380593,
      "loss": 12.444,
      "step": 9093
    },
    {
      "epoch": 0.49520464492590854,
      "grad_norm": 0.6285211262951765,
      "learning_rate": 0.00017617701724896117,
      "loss": 12.3301,
      "step": 9094
    },
    {
      "epoch": 0.49525909892249154,
      "grad_norm": 0.6608859614262765,
      "learning_rate": 0.00017617130406749783,
      "loss": 12.4028,
      "step": 9095
    },
    {
      "epoch": 0.49531355291907453,
      "grad_norm": 0.6317947756451693,
      "learning_rate": 0.0001761655902937136,
      "loss": 12.4618,
      "step": 9096
    },
    {
      "epoch": 0.4953680069156576,
      "grad_norm": 0.7670429802659895,
      "learning_rate": 0.000176159875927653,
      "loss": 12.4435,
      "step": 9097
    },
    {
      "epoch": 0.49542246091224057,
      "grad_norm": 0.7483558764694811,
      "learning_rate": 0.00017615416096936043,
      "loss": 12.4271,
      "step": 9098
    },
    {
      "epoch": 0.49547691490882356,
      "grad_norm": 0.727556789374051,
      "learning_rate": 0.00017614844541888035,
      "loss": 12.2108,
      "step": 9099
    },
    {
      "epoch": 0.4955313689054066,
      "grad_norm": 0.9103992034491821,
      "learning_rate": 0.00017614272927625715,
      "loss": 12.4841,
      "step": 9100
    },
    {
      "epoch": 0.4955858229019896,
      "grad_norm": 0.593670948636598,
      "learning_rate": 0.0001761370125415353,
      "loss": 12.4133,
      "step": 9101
    },
    {
      "epoch": 0.49564027689857265,
      "grad_norm": 0.7023290626503857,
      "learning_rate": 0.00017613129521475932,
      "loss": 12.3761,
      "step": 9102
    },
    {
      "epoch": 0.49569473089515564,
      "grad_norm": 0.6374200651285543,
      "learning_rate": 0.0001761255772959736,
      "loss": 12.4001,
      "step": 9103
    },
    {
      "epoch": 0.49574918489173864,
      "grad_norm": 0.7084811765004031,
      "learning_rate": 0.00017611985878522265,
      "loss": 12.4626,
      "step": 9104
    },
    {
      "epoch": 0.4958036388883217,
      "grad_norm": 0.7987390974405981,
      "learning_rate": 0.00017611413968255087,
      "loss": 12.3393,
      "step": 9105
    },
    {
      "epoch": 0.4958580928849047,
      "grad_norm": 0.6722889300823013,
      "learning_rate": 0.00017610841998800283,
      "loss": 12.3658,
      "step": 9106
    },
    {
      "epoch": 0.49591254688148767,
      "grad_norm": 0.7000109463103993,
      "learning_rate": 0.0001761026997016229,
      "loss": 12.4182,
      "step": 9107
    },
    {
      "epoch": 0.4959670008780707,
      "grad_norm": 0.6721741169406789,
      "learning_rate": 0.00017609697882345565,
      "loss": 12.5067,
      "step": 9108
    },
    {
      "epoch": 0.4960214548746537,
      "grad_norm": 0.7401697808763813,
      "learning_rate": 0.0001760912573535455,
      "loss": 12.5111,
      "step": 9109
    },
    {
      "epoch": 0.4960759088712367,
      "grad_norm": 0.718441618807066,
      "learning_rate": 0.000176085535291937,
      "loss": 12.3193,
      "step": 9110
    },
    {
      "epoch": 0.49613036286781975,
      "grad_norm": 0.6037022793016471,
      "learning_rate": 0.0001760798126386746,
      "loss": 12.3863,
      "step": 9111
    },
    {
      "epoch": 0.49618481686440274,
      "grad_norm": 0.6269927356974456,
      "learning_rate": 0.00017607408939380282,
      "loss": 12.3418,
      "step": 9112
    },
    {
      "epoch": 0.49623927086098574,
      "grad_norm": 0.6458263636719211,
      "learning_rate": 0.00017606836555736618,
      "loss": 12.4122,
      "step": 9113
    },
    {
      "epoch": 0.4962937248575688,
      "grad_norm": 0.6077674815766129,
      "learning_rate": 0.0001760626411294092,
      "loss": 12.3997,
      "step": 9114
    },
    {
      "epoch": 0.4963481788541518,
      "grad_norm": 0.6827924582993539,
      "learning_rate": 0.00017605691610997633,
      "loss": 12.5016,
      "step": 9115
    },
    {
      "epoch": 0.49640263285073477,
      "grad_norm": 1.061163683682852,
      "learning_rate": 0.0001760511904991121,
      "loss": 12.4663,
      "step": 9116
    },
    {
      "epoch": 0.4964570868473178,
      "grad_norm": 0.6579332992213393,
      "learning_rate": 0.0001760454642968611,
      "loss": 12.4702,
      "step": 9117
    },
    {
      "epoch": 0.4965115408439008,
      "grad_norm": 0.6817706542101409,
      "learning_rate": 0.00017603973750326778,
      "loss": 12.4496,
      "step": 9118
    },
    {
      "epoch": 0.4965659948404838,
      "grad_norm": 0.6617212495815381,
      "learning_rate": 0.00017603401011837675,
      "loss": 12.4429,
      "step": 9119
    },
    {
      "epoch": 0.49662044883706685,
      "grad_norm": 0.7923159528782391,
      "learning_rate": 0.00017602828214223249,
      "loss": 12.459,
      "step": 9120
    },
    {
      "epoch": 0.49667490283364985,
      "grad_norm": 0.6616900559517679,
      "learning_rate": 0.00017602255357487953,
      "loss": 12.4547,
      "step": 9121
    },
    {
      "epoch": 0.49672935683023284,
      "grad_norm": 0.6740391179889882,
      "learning_rate": 0.00017601682441636246,
      "loss": 12.3776,
      "step": 9122
    },
    {
      "epoch": 0.4967838108268159,
      "grad_norm": 0.6543330850905006,
      "learning_rate": 0.00017601109466672585,
      "loss": 12.3831,
      "step": 9123
    },
    {
      "epoch": 0.4968382648233989,
      "grad_norm": 0.6138310059376589,
      "learning_rate": 0.00017600536432601417,
      "loss": 12.4033,
      "step": 9124
    },
    {
      "epoch": 0.49689271881998187,
      "grad_norm": 0.6241853205466411,
      "learning_rate": 0.00017599963339427207,
      "loss": 12.1249,
      "step": 9125
    },
    {
      "epoch": 0.4969471728165649,
      "grad_norm": 0.654784791879004,
      "learning_rate": 0.00017599390187154405,
      "loss": 12.34,
      "step": 9126
    },
    {
      "epoch": 0.4970016268131479,
      "grad_norm": 0.6626102053470619,
      "learning_rate": 0.00017598816975787475,
      "loss": 12.524,
      "step": 9127
    },
    {
      "epoch": 0.4970560808097309,
      "grad_norm": 0.7690448781237839,
      "learning_rate": 0.00017598243705330865,
      "loss": 12.3843,
      "step": 9128
    },
    {
      "epoch": 0.49711053480631395,
      "grad_norm": 0.5921755860108986,
      "learning_rate": 0.0001759767037578904,
      "loss": 12.3137,
      "step": 9129
    },
    {
      "epoch": 0.49716498880289695,
      "grad_norm": 0.6724512511904382,
      "learning_rate": 0.00017597096987166454,
      "loss": 12.4212,
      "step": 9130
    },
    {
      "epoch": 0.49721944279947994,
      "grad_norm": 0.6060506869183341,
      "learning_rate": 0.0001759652353946757,
      "loss": 12.4006,
      "step": 9131
    },
    {
      "epoch": 0.497273896796063,
      "grad_norm": 0.5649231816877845,
      "learning_rate": 0.00017595950032696844,
      "loss": 12.3956,
      "step": 9132
    },
    {
      "epoch": 0.497328350792646,
      "grad_norm": 0.7233811405518149,
      "learning_rate": 0.00017595376466858736,
      "loss": 12.4503,
      "step": 9133
    },
    {
      "epoch": 0.497382804789229,
      "grad_norm": 0.6628960967687958,
      "learning_rate": 0.0001759480284195771,
      "loss": 12.4771,
      "step": 9134
    },
    {
      "epoch": 0.497437258785812,
      "grad_norm": 0.68131829983016,
      "learning_rate": 0.00017594229157998222,
      "loss": 12.3811,
      "step": 9135
    },
    {
      "epoch": 0.497491712782395,
      "grad_norm": 0.6134912800656412,
      "learning_rate": 0.0001759365541498473,
      "loss": 12.3098,
      "step": 9136
    },
    {
      "epoch": 0.49754616677897806,
      "grad_norm": 0.6714589554358765,
      "learning_rate": 0.00017593081612921705,
      "loss": 12.411,
      "step": 9137
    },
    {
      "epoch": 0.49760062077556105,
      "grad_norm": 0.6379318436445025,
      "learning_rate": 0.00017592507751813601,
      "loss": 12.4893,
      "step": 9138
    },
    {
      "epoch": 0.49765507477214405,
      "grad_norm": 0.5473509143522587,
      "learning_rate": 0.00017591933831664886,
      "loss": 12.2698,
      "step": 9139
    },
    {
      "epoch": 0.4977095287687271,
      "grad_norm": 0.6009501607651135,
      "learning_rate": 0.0001759135985248002,
      "loss": 12.2894,
      "step": 9140
    },
    {
      "epoch": 0.4977639827653101,
      "grad_norm": 0.6799834772024416,
      "learning_rate": 0.00017590785814263466,
      "loss": 12.4481,
      "step": 9141
    },
    {
      "epoch": 0.4978184367618931,
      "grad_norm": 0.6630782242113525,
      "learning_rate": 0.00017590211717019688,
      "loss": 12.3452,
      "step": 9142
    },
    {
      "epoch": 0.4978728907584761,
      "grad_norm": 0.7008834707717136,
      "learning_rate": 0.00017589637560753153,
      "loss": 12.4784,
      "step": 9143
    },
    {
      "epoch": 0.4979273447550591,
      "grad_norm": 0.6515565727629259,
      "learning_rate": 0.0001758906334546832,
      "loss": 12.3688,
      "step": 9144
    },
    {
      "epoch": 0.4979817987516421,
      "grad_norm": 0.626932835931393,
      "learning_rate": 0.0001758848907116966,
      "loss": 12.3809,
      "step": 9145
    },
    {
      "epoch": 0.49803625274822516,
      "grad_norm": 0.6795840967085799,
      "learning_rate": 0.00017587914737861638,
      "loss": 12.5972,
      "step": 9146
    },
    {
      "epoch": 0.49809070674480815,
      "grad_norm": 0.6063465269047075,
      "learning_rate": 0.00017587340345548717,
      "loss": 12.4608,
      "step": 9147
    },
    {
      "epoch": 0.49814516074139115,
      "grad_norm": 0.5828488884039613,
      "learning_rate": 0.00017586765894235363,
      "loss": 12.3942,
      "step": 9148
    },
    {
      "epoch": 0.4981996147379742,
      "grad_norm": 0.5762418528638724,
      "learning_rate": 0.0001758619138392605,
      "loss": 12.3874,
      "step": 9149
    },
    {
      "epoch": 0.4982540687345572,
      "grad_norm": 0.7426995154173507,
      "learning_rate": 0.00017585616814625235,
      "loss": 12.3388,
      "step": 9150
    },
    {
      "epoch": 0.4983085227311402,
      "grad_norm": 0.6992365012527693,
      "learning_rate": 0.00017585042186337395,
      "loss": 12.4163,
      "step": 9151
    },
    {
      "epoch": 0.49836297672772323,
      "grad_norm": 0.6667318576146535,
      "learning_rate": 0.00017584467499066994,
      "loss": 12.486,
      "step": 9152
    },
    {
      "epoch": 0.4984174307243062,
      "grad_norm": 0.607362379750229,
      "learning_rate": 0.000175838927528185,
      "loss": 12.437,
      "step": 9153
    },
    {
      "epoch": 0.4984718847208892,
      "grad_norm": 0.6540776468373742,
      "learning_rate": 0.00017583317947596387,
      "loss": 12.5109,
      "step": 9154
    },
    {
      "epoch": 0.49852633871747226,
      "grad_norm": 0.6659725463738717,
      "learning_rate": 0.00017582743083405122,
      "loss": 12.3915,
      "step": 9155
    },
    {
      "epoch": 0.49858079271405525,
      "grad_norm": 0.6255060583128188,
      "learning_rate": 0.00017582168160249173,
      "loss": 12.2338,
      "step": 9156
    },
    {
      "epoch": 0.49863524671063825,
      "grad_norm": 0.5953756443739128,
      "learning_rate": 0.00017581593178133016,
      "loss": 12.3541,
      "step": 9157
    },
    {
      "epoch": 0.4986897007072213,
      "grad_norm": 0.6483309366016954,
      "learning_rate": 0.00017581018137061114,
      "loss": 12.2435,
      "step": 9158
    },
    {
      "epoch": 0.4987441547038043,
      "grad_norm": 0.5991782175764987,
      "learning_rate": 0.00017580443037037947,
      "loss": 12.4159,
      "step": 9159
    },
    {
      "epoch": 0.4987986087003873,
      "grad_norm": 0.6104988237916549,
      "learning_rate": 0.00017579867878067984,
      "loss": 12.4037,
      "step": 9160
    },
    {
      "epoch": 0.49885306269697033,
      "grad_norm": 0.6523693446003387,
      "learning_rate": 0.00017579292660155696,
      "loss": 12.4619,
      "step": 9161
    },
    {
      "epoch": 0.4989075166935533,
      "grad_norm": 0.5850126153423292,
      "learning_rate": 0.0001757871738330556,
      "loss": 12.3452,
      "step": 9162
    },
    {
      "epoch": 0.4989619706901363,
      "grad_norm": 0.6481344573786663,
      "learning_rate": 0.00017578142047522044,
      "loss": 12.4467,
      "step": 9163
    },
    {
      "epoch": 0.49901642468671936,
      "grad_norm": 0.6272323440376265,
      "learning_rate": 0.00017577566652809626,
      "loss": 12.3522,
      "step": 9164
    },
    {
      "epoch": 0.49907087868330235,
      "grad_norm": 0.6205903085328062,
      "learning_rate": 0.00017576991199172776,
      "loss": 12.5213,
      "step": 9165
    },
    {
      "epoch": 0.49912533267988535,
      "grad_norm": 0.6477408614756291,
      "learning_rate": 0.00017576415686615976,
      "loss": 12.4615,
      "step": 9166
    },
    {
      "epoch": 0.4991797866764684,
      "grad_norm": 0.7118857219939702,
      "learning_rate": 0.00017575840115143692,
      "loss": 12.5615,
      "step": 9167
    },
    {
      "epoch": 0.4992342406730514,
      "grad_norm": 0.5831572365350796,
      "learning_rate": 0.00017575264484760407,
      "loss": 12.4012,
      "step": 9168
    },
    {
      "epoch": 0.49928869466963444,
      "grad_norm": 0.6296968500770712,
      "learning_rate": 0.00017574688795470595,
      "loss": 12.5436,
      "step": 9169
    },
    {
      "epoch": 0.49934314866621743,
      "grad_norm": 0.6197900937902061,
      "learning_rate": 0.00017574113047278734,
      "loss": 12.3763,
      "step": 9170
    },
    {
      "epoch": 0.4993976026628004,
      "grad_norm": 0.695940103894005,
      "learning_rate": 0.00017573537240189298,
      "loss": 12.3472,
      "step": 9171
    },
    {
      "epoch": 0.49945205665938347,
      "grad_norm": 0.6499827974246789,
      "learning_rate": 0.00017572961374206768,
      "loss": 12.4526,
      "step": 9172
    },
    {
      "epoch": 0.49950651065596646,
      "grad_norm": 0.6880144057586078,
      "learning_rate": 0.0001757238544933562,
      "loss": 12.4985,
      "step": 9173
    },
    {
      "epoch": 0.49956096465254946,
      "grad_norm": 0.7149621700845834,
      "learning_rate": 0.00017571809465580332,
      "loss": 12.4272,
      "step": 9174
    },
    {
      "epoch": 0.4996154186491325,
      "grad_norm": 0.6413703173304228,
      "learning_rate": 0.00017571233422945386,
      "loss": 12.4654,
      "step": 9175
    },
    {
      "epoch": 0.4996698726457155,
      "grad_norm": 0.6444690707512226,
      "learning_rate": 0.00017570657321435256,
      "loss": 12.3721,
      "step": 9176
    },
    {
      "epoch": 0.4997243266422985,
      "grad_norm": 0.6632077405948433,
      "learning_rate": 0.00017570081161054425,
      "loss": 12.4597,
      "step": 9177
    },
    {
      "epoch": 0.49977878063888154,
      "grad_norm": 0.6920115258147176,
      "learning_rate": 0.00017569504941807376,
      "loss": 12.4457,
      "step": 9178
    },
    {
      "epoch": 0.49983323463546453,
      "grad_norm": 0.6958336434558215,
      "learning_rate": 0.00017568928663698588,
      "loss": 12.4001,
      "step": 9179
    },
    {
      "epoch": 0.4998876886320475,
      "grad_norm": 0.6470874095176246,
      "learning_rate": 0.00017568352326732538,
      "loss": 12.3732,
      "step": 9180
    },
    {
      "epoch": 0.49994214262863057,
      "grad_norm": 0.6266020668283614,
      "learning_rate": 0.00017567775930913714,
      "loss": 12.3153,
      "step": 9181
    },
    {
      "epoch": 0.49999659662521356,
      "grad_norm": 0.6116119982186802,
      "learning_rate": 0.00017567199476246591,
      "loss": 12.4759,
      "step": 9182
    },
    {
      "epoch": 0.5000510506217966,
      "grad_norm": 0.6341883591867439,
      "learning_rate": 0.00017566622962735662,
      "loss": 12.4495,
      "step": 9183
    },
    {
      "epoch": 0.5001055046183795,
      "grad_norm": 0.5901644660053987,
      "learning_rate": 0.00017566046390385398,
      "loss": 12.433,
      "step": 9184
    },
    {
      "epoch": 0.5001599586149627,
      "grad_norm": 0.6640999529367435,
      "learning_rate": 0.00017565469759200293,
      "loss": 12.2458,
      "step": 9185
    },
    {
      "epoch": 0.5002144126115456,
      "grad_norm": 0.6842580247261328,
      "learning_rate": 0.00017564893069184825,
      "loss": 12.5172,
      "step": 9186
    },
    {
      "epoch": 0.5002688666081286,
      "grad_norm": 0.5644833573755198,
      "learning_rate": 0.00017564316320343477,
      "loss": 12.3548,
      "step": 9187
    },
    {
      "epoch": 0.5003233206047116,
      "grad_norm": 0.6150274179512815,
      "learning_rate": 0.0001756373951268074,
      "loss": 12.2932,
      "step": 9188
    },
    {
      "epoch": 0.5003777746012946,
      "grad_norm": 0.5598091040693869,
      "learning_rate": 0.00017563162646201094,
      "loss": 12.4621,
      "step": 9189
    },
    {
      "epoch": 0.5004322285978776,
      "grad_norm": 0.6553818557013275,
      "learning_rate": 0.0001756258572090903,
      "loss": 12.5473,
      "step": 9190
    },
    {
      "epoch": 0.5004866825944607,
      "grad_norm": 0.8404702869084849,
      "learning_rate": 0.0001756200873680903,
      "loss": 12.4445,
      "step": 9191
    },
    {
      "epoch": 0.5005411365910437,
      "grad_norm": 0.5642230661996501,
      "learning_rate": 0.0001756143169390558,
      "loss": 12.4559,
      "step": 9192
    },
    {
      "epoch": 0.5005955905876267,
      "grad_norm": 0.6603679295883715,
      "learning_rate": 0.0001756085459220317,
      "loss": 12.4716,
      "step": 9193
    },
    {
      "epoch": 0.5006500445842097,
      "grad_norm": 0.7212196764091794,
      "learning_rate": 0.00017560277431706288,
      "loss": 12.5586,
      "step": 9194
    },
    {
      "epoch": 0.5007044985807927,
      "grad_norm": 0.6153404274085386,
      "learning_rate": 0.0001755970021241942,
      "loss": 12.3659,
      "step": 9195
    },
    {
      "epoch": 0.5007589525773757,
      "grad_norm": 0.7446022353173876,
      "learning_rate": 0.00017559122934347055,
      "loss": 12.5325,
      "step": 9196
    },
    {
      "epoch": 0.5008134065739588,
      "grad_norm": 0.6822957531520627,
      "learning_rate": 0.0001755854559749368,
      "loss": 12.284,
      "step": 9197
    },
    {
      "epoch": 0.5008678605705418,
      "grad_norm": 0.5839316572008065,
      "learning_rate": 0.00017557968201863792,
      "loss": 12.4315,
      "step": 9198
    },
    {
      "epoch": 0.5009223145671248,
      "grad_norm": 0.5920598310314543,
      "learning_rate": 0.00017557390747461872,
      "loss": 12.4369,
      "step": 9199
    },
    {
      "epoch": 0.5009767685637078,
      "grad_norm": 0.5746963240067243,
      "learning_rate": 0.0001755681323429242,
      "loss": 12.4049,
      "step": 9200
    },
    {
      "epoch": 0.5010312225602908,
      "grad_norm": 0.6387688032969914,
      "learning_rate": 0.00017556235662359915,
      "loss": 12.3825,
      "step": 9201
    },
    {
      "epoch": 0.5010856765568737,
      "grad_norm": 0.6220273180989577,
      "learning_rate": 0.00017555658031668853,
      "loss": 12.3686,
      "step": 9202
    },
    {
      "epoch": 0.5011401305534569,
      "grad_norm": 0.5955115588017998,
      "learning_rate": 0.0001755508034222373,
      "loss": 12.4406,
      "step": 9203
    },
    {
      "epoch": 0.5011945845500398,
      "grad_norm": 0.6381359428691479,
      "learning_rate": 0.00017554502594029038,
      "loss": 12.3493,
      "step": 9204
    },
    {
      "epoch": 0.5012490385466228,
      "grad_norm": 0.6268485567466849,
      "learning_rate": 0.00017553924787089262,
      "loss": 12.3149,
      "step": 9205
    },
    {
      "epoch": 0.5013034925432058,
      "grad_norm": 0.5994229081680599,
      "learning_rate": 0.000175533469214089,
      "loss": 12.3849,
      "step": 9206
    },
    {
      "epoch": 0.5013579465397888,
      "grad_norm": 0.6879692862918647,
      "learning_rate": 0.0001755276899699245,
      "loss": 12.3399,
      "step": 9207
    },
    {
      "epoch": 0.5014124005363718,
      "grad_norm": 0.6485348858208008,
      "learning_rate": 0.00017552191013844398,
      "loss": 12.37,
      "step": 9208
    },
    {
      "epoch": 0.5014668545329549,
      "grad_norm": 0.6290445813308777,
      "learning_rate": 0.00017551612971969247,
      "loss": 12.4224,
      "step": 9209
    },
    {
      "epoch": 0.5015213085295379,
      "grad_norm": 0.7839468838039373,
      "learning_rate": 0.0001755103487137148,
      "loss": 12.5299,
      "step": 9210
    },
    {
      "epoch": 0.5015757625261209,
      "grad_norm": 0.6271535640482853,
      "learning_rate": 0.00017550456712055605,
      "loss": 12.3695,
      "step": 9211
    },
    {
      "epoch": 0.5016302165227039,
      "grad_norm": 0.6818818059207238,
      "learning_rate": 0.00017549878494026112,
      "loss": 12.2653,
      "step": 9212
    },
    {
      "epoch": 0.5016846705192869,
      "grad_norm": 0.7178157128174575,
      "learning_rate": 0.00017549300217287495,
      "loss": 12.5114,
      "step": 9213
    },
    {
      "epoch": 0.5017391245158699,
      "grad_norm": 0.6376885520060535,
      "learning_rate": 0.00017548721881844255,
      "loss": 12.3963,
      "step": 9214
    },
    {
      "epoch": 0.501793578512453,
      "grad_norm": 0.7370977772095955,
      "learning_rate": 0.0001754814348770089,
      "loss": 12.5435,
      "step": 9215
    },
    {
      "epoch": 0.501848032509036,
      "grad_norm": 0.636511787689217,
      "learning_rate": 0.00017547565034861892,
      "loss": 12.396,
      "step": 9216
    },
    {
      "epoch": 0.501902486505619,
      "grad_norm": 0.8710913375149281,
      "learning_rate": 0.00017546986523331762,
      "loss": 12.4173,
      "step": 9217
    },
    {
      "epoch": 0.501956940502202,
      "grad_norm": 0.6260169853757912,
      "learning_rate": 0.00017546407953115002,
      "loss": 12.455,
      "step": 9218
    },
    {
      "epoch": 0.502011394498785,
      "grad_norm": 0.6306583641898258,
      "learning_rate": 0.00017545829324216108,
      "loss": 12.4029,
      "step": 9219
    },
    {
      "epoch": 0.5020658484953681,
      "grad_norm": 0.8238270270686283,
      "learning_rate": 0.0001754525063663958,
      "loss": 12.5189,
      "step": 9220
    },
    {
      "epoch": 0.502120302491951,
      "grad_norm": 0.5536911448623411,
      "learning_rate": 0.00017544671890389914,
      "loss": 12.3506,
      "step": 9221
    },
    {
      "epoch": 0.502174756488534,
      "grad_norm": 0.6744248023901999,
      "learning_rate": 0.00017544093085471616,
      "loss": 12.4723,
      "step": 9222
    },
    {
      "epoch": 0.502229210485117,
      "grad_norm": 0.6133959900954926,
      "learning_rate": 0.00017543514221889187,
      "loss": 12.4893,
      "step": 9223
    },
    {
      "epoch": 0.5022836644817,
      "grad_norm": 0.5929697847285679,
      "learning_rate": 0.00017542935299647122,
      "loss": 12.4983,
      "step": 9224
    },
    {
      "epoch": 0.502338118478283,
      "grad_norm": 0.6482710041361429,
      "learning_rate": 0.00017542356318749934,
      "loss": 12.287,
      "step": 9225
    },
    {
      "epoch": 0.5023925724748661,
      "grad_norm": 0.6812798436350104,
      "learning_rate": 0.0001754177727920211,
      "loss": 12.422,
      "step": 9226
    },
    {
      "epoch": 0.5024470264714491,
      "grad_norm": 0.5734535128505137,
      "learning_rate": 0.00017541198181008169,
      "loss": 12.3762,
      "step": 9227
    },
    {
      "epoch": 0.5025014804680321,
      "grad_norm": 0.7298704517473773,
      "learning_rate": 0.000175406190241726,
      "loss": 12.3909,
      "step": 9228
    },
    {
      "epoch": 0.5025559344646151,
      "grad_norm": 0.6211270568866112,
      "learning_rate": 0.0001754003980869992,
      "loss": 12.514,
      "step": 9229
    },
    {
      "epoch": 0.5026103884611981,
      "grad_norm": 0.6538190401861905,
      "learning_rate": 0.00017539460534594622,
      "loss": 12.4927,
      "step": 9230
    },
    {
      "epoch": 0.5026648424577811,
      "grad_norm": 0.649123256226189,
      "learning_rate": 0.00017538881201861214,
      "loss": 12.4245,
      "step": 9231
    },
    {
      "epoch": 0.5027192964543642,
      "grad_norm": 0.5442248840080184,
      "learning_rate": 0.00017538301810504203,
      "loss": 12.3251,
      "step": 9232
    },
    {
      "epoch": 0.5027737504509472,
      "grad_norm": 0.5847750120836829,
      "learning_rate": 0.0001753772236052809,
      "loss": 12.4215,
      "step": 9233
    },
    {
      "epoch": 0.5028282044475302,
      "grad_norm": 0.6489129014356729,
      "learning_rate": 0.00017537142851937386,
      "loss": 12.4467,
      "step": 9234
    },
    {
      "epoch": 0.5028826584441132,
      "grad_norm": 0.6464215828254797,
      "learning_rate": 0.00017536563284736593,
      "loss": 12.357,
      "step": 9235
    },
    {
      "epoch": 0.5029371124406962,
      "grad_norm": 0.5998675678204343,
      "learning_rate": 0.00017535983658930224,
      "loss": 12.4077,
      "step": 9236
    },
    {
      "epoch": 0.5029915664372792,
      "grad_norm": 0.6711699450059563,
      "learning_rate": 0.0001753540397452278,
      "loss": 12.4252,
      "step": 9237
    },
    {
      "epoch": 0.5030460204338623,
      "grad_norm": 0.6565266462253379,
      "learning_rate": 0.00017534824231518772,
      "loss": 12.5112,
      "step": 9238
    },
    {
      "epoch": 0.5031004744304453,
      "grad_norm": 0.7106207839496469,
      "learning_rate": 0.00017534244429922704,
      "loss": 12.3185,
      "step": 9239
    },
    {
      "epoch": 0.5031549284270282,
      "grad_norm": 0.6746786233771818,
      "learning_rate": 0.00017533664569739093,
      "loss": 12.4076,
      "step": 9240
    },
    {
      "epoch": 0.5032093824236112,
      "grad_norm": 0.7287877456392946,
      "learning_rate": 0.00017533084650972437,
      "loss": 12.4355,
      "step": 9241
    },
    {
      "epoch": 0.5032638364201942,
      "grad_norm": 0.7183288979000483,
      "learning_rate": 0.00017532504673627256,
      "loss": 12.3689,
      "step": 9242
    },
    {
      "epoch": 0.5033182904167772,
      "grad_norm": 0.5913008733418407,
      "learning_rate": 0.00017531924637708053,
      "loss": 12.3267,
      "step": 9243
    },
    {
      "epoch": 0.5033727444133603,
      "grad_norm": 0.7399657849568171,
      "learning_rate": 0.0001753134454321934,
      "loss": 12.3304,
      "step": 9244
    },
    {
      "epoch": 0.5034271984099433,
      "grad_norm": 0.6330603467608257,
      "learning_rate": 0.00017530764390165634,
      "loss": 12.3574,
      "step": 9245
    },
    {
      "epoch": 0.5034816524065263,
      "grad_norm": 0.6249736543216847,
      "learning_rate": 0.0001753018417855144,
      "loss": 12.4009,
      "step": 9246
    },
    {
      "epoch": 0.5035361064031093,
      "grad_norm": 0.6501824218490548,
      "learning_rate": 0.00017529603908381268,
      "loss": 12.4118,
      "step": 9247
    },
    {
      "epoch": 0.5035905603996923,
      "grad_norm": 0.5848558170314551,
      "learning_rate": 0.00017529023579659634,
      "loss": 12.3934,
      "step": 9248
    },
    {
      "epoch": 0.5036450143962753,
      "grad_norm": 0.6547297192921312,
      "learning_rate": 0.00017528443192391048,
      "loss": 12.4775,
      "step": 9249
    },
    {
      "epoch": 0.5036994683928584,
      "grad_norm": 0.9148181678227723,
      "learning_rate": 0.00017527862746580028,
      "loss": 12.502,
      "step": 9250
    },
    {
      "epoch": 0.5037539223894414,
      "grad_norm": 0.6406385901042989,
      "learning_rate": 0.00017527282242231083,
      "loss": 12.4692,
      "step": 9251
    },
    {
      "epoch": 0.5038083763860244,
      "grad_norm": 0.639199761303182,
      "learning_rate": 0.0001752670167934873,
      "loss": 12.4294,
      "step": 9252
    },
    {
      "epoch": 0.5038628303826074,
      "grad_norm": 0.6108791181730224,
      "learning_rate": 0.00017526121057937486,
      "loss": 12.566,
      "step": 9253
    },
    {
      "epoch": 0.5039172843791904,
      "grad_norm": 0.6462775981127912,
      "learning_rate": 0.00017525540378001856,
      "loss": 12.4095,
      "step": 9254
    },
    {
      "epoch": 0.5039717383757735,
      "grad_norm": 0.6117529957039444,
      "learning_rate": 0.00017524959639546368,
      "loss": 12.3417,
      "step": 9255
    },
    {
      "epoch": 0.5040261923723565,
      "grad_norm": 0.6653108100707444,
      "learning_rate": 0.0001752437884257553,
      "loss": 12.5033,
      "step": 9256
    },
    {
      "epoch": 0.5040806463689395,
      "grad_norm": 0.5761033208886187,
      "learning_rate": 0.00017523797987093858,
      "loss": 12.3672,
      "step": 9257
    },
    {
      "epoch": 0.5041351003655224,
      "grad_norm": 0.6018386004957551,
      "learning_rate": 0.00017523217073105873,
      "loss": 12.2592,
      "step": 9258
    },
    {
      "epoch": 0.5041895543621054,
      "grad_norm": 0.6977867384115372,
      "learning_rate": 0.0001752263610061609,
      "loss": 12.5126,
      "step": 9259
    },
    {
      "epoch": 0.5042440083586884,
      "grad_norm": 0.6070018997053899,
      "learning_rate": 0.00017522055069629027,
      "loss": 12.4393,
      "step": 9260
    },
    {
      "epoch": 0.5042984623552715,
      "grad_norm": 0.6458477465628019,
      "learning_rate": 0.00017521473980149202,
      "loss": 12.4836,
      "step": 9261
    },
    {
      "epoch": 0.5043529163518545,
      "grad_norm": 0.5913666593119621,
      "learning_rate": 0.00017520892832181134,
      "loss": 12.3034,
      "step": 9262
    },
    {
      "epoch": 0.5044073703484375,
      "grad_norm": 0.6387974436651321,
      "learning_rate": 0.00017520311625729345,
      "loss": 12.4383,
      "step": 9263
    },
    {
      "epoch": 0.5044618243450205,
      "grad_norm": 0.6663629819407951,
      "learning_rate": 0.00017519730360798346,
      "loss": 12.3579,
      "step": 9264
    },
    {
      "epoch": 0.5045162783416035,
      "grad_norm": 0.6743783218918941,
      "learning_rate": 0.00017519149037392668,
      "loss": 12.3556,
      "step": 9265
    },
    {
      "epoch": 0.5045707323381865,
      "grad_norm": 0.5934684892792085,
      "learning_rate": 0.0001751856765551682,
      "loss": 12.6217,
      "step": 9266
    },
    {
      "epoch": 0.5046251863347696,
      "grad_norm": 0.6542403394106991,
      "learning_rate": 0.00017517986215175336,
      "loss": 12.4348,
      "step": 9267
    },
    {
      "epoch": 0.5046796403313526,
      "grad_norm": 0.6140821063343879,
      "learning_rate": 0.00017517404716372727,
      "loss": 12.2976,
      "step": 9268
    },
    {
      "epoch": 0.5047340943279356,
      "grad_norm": 0.6302783978472125,
      "learning_rate": 0.0001751682315911352,
      "loss": 12.3735,
      "step": 9269
    },
    {
      "epoch": 0.5047885483245186,
      "grad_norm": 0.617713894873678,
      "learning_rate": 0.00017516241543402233,
      "loss": 12.4546,
      "step": 9270
    },
    {
      "epoch": 0.5048430023211016,
      "grad_norm": 0.6316766316871215,
      "learning_rate": 0.00017515659869243391,
      "loss": 12.3908,
      "step": 9271
    },
    {
      "epoch": 0.5048974563176846,
      "grad_norm": 0.6274417258446252,
      "learning_rate": 0.0001751507813664152,
      "loss": 12.4156,
      "step": 9272
    },
    {
      "epoch": 0.5049519103142677,
      "grad_norm": 0.6180856210866346,
      "learning_rate": 0.00017514496345601144,
      "loss": 12.4906,
      "step": 9273
    },
    {
      "epoch": 0.5050063643108507,
      "grad_norm": 0.6618825336368301,
      "learning_rate": 0.00017513914496126778,
      "loss": 12.4492,
      "step": 9274
    },
    {
      "epoch": 0.5050608183074337,
      "grad_norm": 0.6346775520196072,
      "learning_rate": 0.00017513332588222955,
      "loss": 12.4255,
      "step": 9275
    },
    {
      "epoch": 0.5051152723040166,
      "grad_norm": 0.5713866022890768,
      "learning_rate": 0.00017512750621894197,
      "loss": 12.3917,
      "step": 9276
    },
    {
      "epoch": 0.5051697263005996,
      "grad_norm": 0.6644568918871256,
      "learning_rate": 0.0001751216859714503,
      "loss": 12.5056,
      "step": 9277
    },
    {
      "epoch": 0.5052241802971826,
      "grad_norm": 0.6559421524036552,
      "learning_rate": 0.00017511586513979985,
      "loss": 12.4277,
      "step": 9278
    },
    {
      "epoch": 0.5052786342937657,
      "grad_norm": 0.5870845703090499,
      "learning_rate": 0.00017511004372403578,
      "loss": 12.5096,
      "step": 9279
    },
    {
      "epoch": 0.5053330882903487,
      "grad_norm": 0.706795464340835,
      "learning_rate": 0.00017510422172420343,
      "loss": 12.3108,
      "step": 9280
    },
    {
      "epoch": 0.5053875422869317,
      "grad_norm": 0.5700640206283255,
      "learning_rate": 0.00017509839914034808,
      "loss": 12.2902,
      "step": 9281
    },
    {
      "epoch": 0.5054419962835147,
      "grad_norm": 0.6128756126723002,
      "learning_rate": 0.00017509257597251497,
      "loss": 12.4242,
      "step": 9282
    },
    {
      "epoch": 0.5054964502800977,
      "grad_norm": 0.6653571480654208,
      "learning_rate": 0.00017508675222074936,
      "loss": 12.4248,
      "step": 9283
    },
    {
      "epoch": 0.5055509042766807,
      "grad_norm": 0.6167569171939026,
      "learning_rate": 0.00017508092788509662,
      "loss": 12.4268,
      "step": 9284
    },
    {
      "epoch": 0.5056053582732638,
      "grad_norm": 0.6094727507670129,
      "learning_rate": 0.00017507510296560196,
      "loss": 12.4102,
      "step": 9285
    },
    {
      "epoch": 0.5056598122698468,
      "grad_norm": 0.6261696978568257,
      "learning_rate": 0.00017506927746231075,
      "loss": 12.3727,
      "step": 9286
    },
    {
      "epoch": 0.5057142662664298,
      "grad_norm": 0.5853889054617658,
      "learning_rate": 0.00017506345137526823,
      "loss": 12.2828,
      "step": 9287
    },
    {
      "epoch": 0.5057687202630128,
      "grad_norm": 0.6989389592903248,
      "learning_rate": 0.00017505762470451972,
      "loss": 12.5096,
      "step": 9288
    },
    {
      "epoch": 0.5058231742595958,
      "grad_norm": 0.8410187354062965,
      "learning_rate": 0.00017505179745011055,
      "loss": 12.4025,
      "step": 9289
    },
    {
      "epoch": 0.5058776282561789,
      "grad_norm": 0.6356104259521266,
      "learning_rate": 0.000175045969612086,
      "loss": 12.4196,
      "step": 9290
    },
    {
      "epoch": 0.5059320822527619,
      "grad_norm": 0.8491671153494073,
      "learning_rate": 0.0001750401411904914,
      "loss": 12.3207,
      "step": 9291
    },
    {
      "epoch": 0.5059865362493449,
      "grad_norm": 0.6281540370395706,
      "learning_rate": 0.00017503431218537205,
      "loss": 12.4043,
      "step": 9292
    },
    {
      "epoch": 0.5060409902459279,
      "grad_norm": 0.7105148750537855,
      "learning_rate": 0.00017502848259677337,
      "loss": 12.4283,
      "step": 9293
    },
    {
      "epoch": 0.5060954442425109,
      "grad_norm": 0.7066195639781923,
      "learning_rate": 0.00017502265242474058,
      "loss": 12.4412,
      "step": 9294
    },
    {
      "epoch": 0.5061498982390938,
      "grad_norm": 0.7110921055568227,
      "learning_rate": 0.0001750168216693191,
      "loss": 12.4691,
      "step": 9295
    },
    {
      "epoch": 0.506204352235677,
      "grad_norm": 0.6143960464742542,
      "learning_rate": 0.00017501099033055418,
      "loss": 12.4342,
      "step": 9296
    },
    {
      "epoch": 0.5062588062322599,
      "grad_norm": 0.6211083975799827,
      "learning_rate": 0.00017500515840849126,
      "loss": 12.444,
      "step": 9297
    },
    {
      "epoch": 0.5063132602288429,
      "grad_norm": 0.6150695172377717,
      "learning_rate": 0.00017499932590317566,
      "loss": 12.4752,
      "step": 9298
    },
    {
      "epoch": 0.5063677142254259,
      "grad_norm": 0.5526792990362289,
      "learning_rate": 0.0001749934928146527,
      "loss": 12.4155,
      "step": 9299
    },
    {
      "epoch": 0.5064221682220089,
      "grad_norm": 0.6248669052704635,
      "learning_rate": 0.00017498765914296778,
      "loss": 12.4054,
      "step": 9300
    },
    {
      "epoch": 0.5064766222185919,
      "grad_norm": 0.5910774952254157,
      "learning_rate": 0.00017498182488816623,
      "loss": 12.3861,
      "step": 9301
    },
    {
      "epoch": 0.506531076215175,
      "grad_norm": 0.5865642470842327,
      "learning_rate": 0.00017497599005029344,
      "loss": 12.2948,
      "step": 9302
    },
    {
      "epoch": 0.506585530211758,
      "grad_norm": 0.6118141150631882,
      "learning_rate": 0.00017497015462939478,
      "loss": 12.4216,
      "step": 9303
    },
    {
      "epoch": 0.506639984208341,
      "grad_norm": 0.6419707275290791,
      "learning_rate": 0.00017496431862551562,
      "loss": 12.1743,
      "step": 9304
    },
    {
      "epoch": 0.506694438204924,
      "grad_norm": 0.6011774662281038,
      "learning_rate": 0.00017495848203870134,
      "loss": 12.3669,
      "step": 9305
    },
    {
      "epoch": 0.506748892201507,
      "grad_norm": 0.65660842217026,
      "learning_rate": 0.00017495264486899737,
      "loss": 12.3027,
      "step": 9306
    },
    {
      "epoch": 0.50680334619809,
      "grad_norm": 0.6817634176980636,
      "learning_rate": 0.00017494680711644902,
      "loss": 12.4326,
      "step": 9307
    },
    {
      "epoch": 0.5068578001946731,
      "grad_norm": 0.561583501971307,
      "learning_rate": 0.00017494096878110176,
      "loss": 12.4116,
      "step": 9308
    },
    {
      "epoch": 0.5069122541912561,
      "grad_norm": 0.5435703750824127,
      "learning_rate": 0.00017493512986300095,
      "loss": 12.3742,
      "step": 9309
    },
    {
      "epoch": 0.5069667081878391,
      "grad_norm": 0.6022867408072567,
      "learning_rate": 0.00017492929036219202,
      "loss": 12.3662,
      "step": 9310
    },
    {
      "epoch": 0.5070211621844221,
      "grad_norm": 0.7724953334573791,
      "learning_rate": 0.00017492345027872035,
      "loss": 12.5138,
      "step": 9311
    },
    {
      "epoch": 0.507075616181005,
      "grad_norm": 0.6585920328470743,
      "learning_rate": 0.00017491760961263136,
      "loss": 12.3403,
      "step": 9312
    },
    {
      "epoch": 0.507130070177588,
      "grad_norm": 0.6466273952386169,
      "learning_rate": 0.00017491176836397046,
      "loss": 12.4771,
      "step": 9313
    },
    {
      "epoch": 0.5071845241741711,
      "grad_norm": 0.6009781768170371,
      "learning_rate": 0.0001749059265327831,
      "loss": 12.4441,
      "step": 9314
    },
    {
      "epoch": 0.5072389781707541,
      "grad_norm": 0.6840403280864805,
      "learning_rate": 0.0001749000841191147,
      "loss": 12.5198,
      "step": 9315
    },
    {
      "epoch": 0.5072934321673371,
      "grad_norm": 0.6337536226529283,
      "learning_rate": 0.00017489424112301067,
      "loss": 12.4631,
      "step": 9316
    },
    {
      "epoch": 0.5073478861639201,
      "grad_norm": 0.6630045328127501,
      "learning_rate": 0.00017488839754451648,
      "loss": 12.4571,
      "step": 9317
    },
    {
      "epoch": 0.5074023401605031,
      "grad_norm": 0.674610953829062,
      "learning_rate": 0.00017488255338367754,
      "loss": 12.3755,
      "step": 9318
    },
    {
      "epoch": 0.5074567941570862,
      "grad_norm": 0.7346706962784108,
      "learning_rate": 0.0001748767086405393,
      "loss": 12.4546,
      "step": 9319
    },
    {
      "epoch": 0.5075112481536692,
      "grad_norm": 0.6197681874525446,
      "learning_rate": 0.00017487086331514725,
      "loss": 12.3235,
      "step": 9320
    },
    {
      "epoch": 0.5075657021502522,
      "grad_norm": 0.5401263427026369,
      "learning_rate": 0.00017486501740754677,
      "loss": 12.327,
      "step": 9321
    },
    {
      "epoch": 0.5076201561468352,
      "grad_norm": 0.6569790738085491,
      "learning_rate": 0.00017485917091778337,
      "loss": 12.2445,
      "step": 9322
    },
    {
      "epoch": 0.5076746101434182,
      "grad_norm": 0.6418331149863142,
      "learning_rate": 0.00017485332384590253,
      "loss": 12.3183,
      "step": 9323
    },
    {
      "epoch": 0.5077290641400012,
      "grad_norm": 0.6243981903237095,
      "learning_rate": 0.00017484747619194964,
      "loss": 12.3514,
      "step": 9324
    },
    {
      "epoch": 0.5077835181365843,
      "grad_norm": 0.714394137512393,
      "learning_rate": 0.00017484162795597028,
      "loss": 12.4041,
      "step": 9325
    },
    {
      "epoch": 0.5078379721331673,
      "grad_norm": 0.6336579308981977,
      "learning_rate": 0.00017483577913800984,
      "loss": 12.4572,
      "step": 9326
    },
    {
      "epoch": 0.5078924261297503,
      "grad_norm": 0.648306200933378,
      "learning_rate": 0.00017482992973811382,
      "loss": 12.4322,
      "step": 9327
    },
    {
      "epoch": 0.5079468801263333,
      "grad_norm": 0.6449632020500176,
      "learning_rate": 0.00017482407975632775,
      "loss": 12.2739,
      "step": 9328
    },
    {
      "epoch": 0.5080013341229163,
      "grad_norm": 0.560273711009082,
      "learning_rate": 0.00017481822919269705,
      "loss": 12.3498,
      "step": 9329
    },
    {
      "epoch": 0.5080557881194993,
      "grad_norm": 0.7294627898226387,
      "learning_rate": 0.00017481237804726728,
      "loss": 12.3475,
      "step": 9330
    },
    {
      "epoch": 0.5081102421160824,
      "grad_norm": 0.6666923064781625,
      "learning_rate": 0.00017480652632008393,
      "loss": 12.3618,
      "step": 9331
    },
    {
      "epoch": 0.5081646961126653,
      "grad_norm": 0.611908423405081,
      "learning_rate": 0.00017480067401119245,
      "loss": 12.3883,
      "step": 9332
    },
    {
      "epoch": 0.5082191501092483,
      "grad_norm": 0.5789056783313532,
      "learning_rate": 0.0001747948211206384,
      "loss": 12.3337,
      "step": 9333
    },
    {
      "epoch": 0.5082736041058313,
      "grad_norm": 0.6528298517402605,
      "learning_rate": 0.00017478896764846726,
      "loss": 12.3418,
      "step": 9334
    },
    {
      "epoch": 0.5083280581024143,
      "grad_norm": 0.586154787199463,
      "learning_rate": 0.00017478311359472458,
      "loss": 12.346,
      "step": 9335
    },
    {
      "epoch": 0.5083825120989973,
      "grad_norm": 0.7375603848486024,
      "learning_rate": 0.00017477725895945589,
      "loss": 12.4789,
      "step": 9336
    },
    {
      "epoch": 0.5084369660955804,
      "grad_norm": 0.5622036931182314,
      "learning_rate": 0.00017477140374270666,
      "loss": 12.3521,
      "step": 9337
    },
    {
      "epoch": 0.5084914200921634,
      "grad_norm": 0.7094470049438123,
      "learning_rate": 0.00017476554794452248,
      "loss": 12.3785,
      "step": 9338
    },
    {
      "epoch": 0.5085458740887464,
      "grad_norm": 0.5535802003127333,
      "learning_rate": 0.00017475969156494883,
      "loss": 12.3477,
      "step": 9339
    },
    {
      "epoch": 0.5086003280853294,
      "grad_norm": 0.6060587087426393,
      "learning_rate": 0.0001747538346040313,
      "loss": 12.4283,
      "step": 9340
    },
    {
      "epoch": 0.5086547820819124,
      "grad_norm": 0.6486830014576412,
      "learning_rate": 0.00017474797706181546,
      "loss": 12.5064,
      "step": 9341
    },
    {
      "epoch": 0.5087092360784954,
      "grad_norm": 0.650171933766708,
      "learning_rate": 0.00017474211893834672,
      "loss": 12.3925,
      "step": 9342
    },
    {
      "epoch": 0.5087636900750785,
      "grad_norm": 0.6178982347558384,
      "learning_rate": 0.0001747362602336708,
      "loss": 12.4251,
      "step": 9343
    },
    {
      "epoch": 0.5088181440716615,
      "grad_norm": 0.6098889569054531,
      "learning_rate": 0.00017473040094783318,
      "loss": 12.4814,
      "step": 9344
    },
    {
      "epoch": 0.5088725980682445,
      "grad_norm": 0.6397229844431023,
      "learning_rate": 0.00017472454108087942,
      "loss": 12.5545,
      "step": 9345
    },
    {
      "epoch": 0.5089270520648275,
      "grad_norm": 0.662132294090361,
      "learning_rate": 0.0001747186806328551,
      "loss": 12.5107,
      "step": 9346
    },
    {
      "epoch": 0.5089815060614105,
      "grad_norm": 0.7496901176361102,
      "learning_rate": 0.00017471281960380577,
      "loss": 12.4908,
      "step": 9347
    },
    {
      "epoch": 0.5090359600579935,
      "grad_norm": 0.5732062504922385,
      "learning_rate": 0.00017470695799377703,
      "loss": 12.3264,
      "step": 9348
    },
    {
      "epoch": 0.5090904140545766,
      "grad_norm": 0.6151031159507464,
      "learning_rate": 0.00017470109580281447,
      "loss": 12.304,
      "step": 9349
    },
    {
      "epoch": 0.5091448680511595,
      "grad_norm": 0.6959356182660499,
      "learning_rate": 0.00017469523303096363,
      "loss": 12.3683,
      "step": 9350
    },
    {
      "epoch": 0.5091993220477425,
      "grad_norm": 0.6226389570908005,
      "learning_rate": 0.00017468936967827018,
      "loss": 12.4791,
      "step": 9351
    },
    {
      "epoch": 0.5092537760443255,
      "grad_norm": 0.6946591122987917,
      "learning_rate": 0.00017468350574477965,
      "loss": 12.3415,
      "step": 9352
    },
    {
      "epoch": 0.5093082300409085,
      "grad_norm": 0.565289908611492,
      "learning_rate": 0.00017467764123053764,
      "loss": 12.3893,
      "step": 9353
    },
    {
      "epoch": 0.5093626840374916,
      "grad_norm": 0.589008060870395,
      "learning_rate": 0.0001746717761355898,
      "loss": 12.3961,
      "step": 9354
    },
    {
      "epoch": 0.5094171380340746,
      "grad_norm": 0.6423667999843898,
      "learning_rate": 0.00017466591045998167,
      "loss": 12.2483,
      "step": 9355
    },
    {
      "epoch": 0.5094715920306576,
      "grad_norm": 0.5878197206271527,
      "learning_rate": 0.0001746600442037589,
      "loss": 12.2345,
      "step": 9356
    },
    {
      "epoch": 0.5095260460272406,
      "grad_norm": 0.611952765287439,
      "learning_rate": 0.0001746541773669671,
      "loss": 12.3193,
      "step": 9357
    },
    {
      "epoch": 0.5095805000238236,
      "grad_norm": 0.7025299355851088,
      "learning_rate": 0.00017464830994965195,
      "loss": 12.6056,
      "step": 9358
    },
    {
      "epoch": 0.5096349540204066,
      "grad_norm": 0.5844492524972718,
      "learning_rate": 0.000174642441951859,
      "loss": 12.3313,
      "step": 9359
    },
    {
      "epoch": 0.5096894080169897,
      "grad_norm": 0.5870217567121213,
      "learning_rate": 0.00017463657337363388,
      "loss": 12.2468,
      "step": 9360
    },
    {
      "epoch": 0.5097438620135727,
      "grad_norm": 0.544716100069704,
      "learning_rate": 0.00017463070421502226,
      "loss": 12.4479,
      "step": 9361
    },
    {
      "epoch": 0.5097983160101557,
      "grad_norm": 0.709392125597355,
      "learning_rate": 0.00017462483447606977,
      "loss": 12.5405,
      "step": 9362
    },
    {
      "epoch": 0.5098527700067387,
      "grad_norm": 0.6876129323729496,
      "learning_rate": 0.00017461896415682206,
      "loss": 12.6349,
      "step": 9363
    },
    {
      "epoch": 0.5099072240033217,
      "grad_norm": 0.5646918644140049,
      "learning_rate": 0.00017461309325732474,
      "loss": 12.4339,
      "step": 9364
    },
    {
      "epoch": 0.5099616779999047,
      "grad_norm": 0.6483491971824025,
      "learning_rate": 0.00017460722177762352,
      "loss": 12.53,
      "step": 9365
    },
    {
      "epoch": 0.5100161319964878,
      "grad_norm": 0.5661502815396339,
      "learning_rate": 0.00017460134971776406,
      "loss": 12.3638,
      "step": 9366
    },
    {
      "epoch": 0.5100705859930708,
      "grad_norm": 0.6473770934053301,
      "learning_rate": 0.00017459547707779195,
      "loss": 12.4482,
      "step": 9367
    },
    {
      "epoch": 0.5101250399896538,
      "grad_norm": 0.5870198823363907,
      "learning_rate": 0.00017458960385775292,
      "loss": 12.3818,
      "step": 9368
    },
    {
      "epoch": 0.5101794939862367,
      "grad_norm": 0.622478533573811,
      "learning_rate": 0.0001745837300576926,
      "loss": 12.3057,
      "step": 9369
    },
    {
      "epoch": 0.5102339479828197,
      "grad_norm": 0.6348645706981006,
      "learning_rate": 0.00017457785567765672,
      "loss": 12.4243,
      "step": 9370
    },
    {
      "epoch": 0.5102884019794027,
      "grad_norm": 0.5427899853169943,
      "learning_rate": 0.00017457198071769091,
      "loss": 12.4867,
      "step": 9371
    },
    {
      "epoch": 0.5103428559759858,
      "grad_norm": 0.9573474443809805,
      "learning_rate": 0.00017456610517784088,
      "loss": 12.3764,
      "step": 9372
    },
    {
      "epoch": 0.5103973099725688,
      "grad_norm": 0.6348206524178484,
      "learning_rate": 0.00017456022905815228,
      "loss": 12.3435,
      "step": 9373
    },
    {
      "epoch": 0.5104517639691518,
      "grad_norm": 0.6032852796079554,
      "learning_rate": 0.00017455435235867088,
      "loss": 12.5172,
      "step": 9374
    },
    {
      "epoch": 0.5105062179657348,
      "grad_norm": 0.6296409528090846,
      "learning_rate": 0.0001745484750794423,
      "loss": 12.3565,
      "step": 9375
    },
    {
      "epoch": 0.5105606719623178,
      "grad_norm": 0.5653690358017915,
      "learning_rate": 0.0001745425972205123,
      "loss": 12.2817,
      "step": 9376
    },
    {
      "epoch": 0.5106151259589008,
      "grad_norm": 0.600763765793065,
      "learning_rate": 0.00017453671878192654,
      "loss": 12.3462,
      "step": 9377
    },
    {
      "epoch": 0.5106695799554839,
      "grad_norm": 0.5921711828910851,
      "learning_rate": 0.00017453083976373077,
      "loss": 12.4408,
      "step": 9378
    },
    {
      "epoch": 0.5107240339520669,
      "grad_norm": 0.653706867357765,
      "learning_rate": 0.0001745249601659707,
      "loss": 12.3856,
      "step": 9379
    },
    {
      "epoch": 0.5107784879486499,
      "grad_norm": 0.6498084542557725,
      "learning_rate": 0.00017451907998869204,
      "loss": 12.5023,
      "step": 9380
    },
    {
      "epoch": 0.5108329419452329,
      "grad_norm": 0.6621391340956829,
      "learning_rate": 0.00017451319923194046,
      "loss": 12.4081,
      "step": 9381
    },
    {
      "epoch": 0.5108873959418159,
      "grad_norm": 0.5955740821428355,
      "learning_rate": 0.0001745073178957618,
      "loss": 12.4427,
      "step": 9382
    },
    {
      "epoch": 0.5109418499383989,
      "grad_norm": 0.629955623693329,
      "learning_rate": 0.00017450143598020173,
      "loss": 12.4182,
      "step": 9383
    },
    {
      "epoch": 0.510996303934982,
      "grad_norm": 0.620450415710074,
      "learning_rate": 0.00017449555348530602,
      "loss": 12.4561,
      "step": 9384
    },
    {
      "epoch": 0.511050757931565,
      "grad_norm": 0.6694978927482998,
      "learning_rate": 0.0001744896704111204,
      "loss": 12.3695,
      "step": 9385
    },
    {
      "epoch": 0.511105211928148,
      "grad_norm": 0.5680924165840874,
      "learning_rate": 0.0001744837867576906,
      "loss": 12.3348,
      "step": 9386
    },
    {
      "epoch": 0.5111596659247309,
      "grad_norm": 0.6566945684404797,
      "learning_rate": 0.00017447790252506237,
      "loss": 12.3644,
      "step": 9387
    },
    {
      "epoch": 0.5112141199213139,
      "grad_norm": 0.6437455314024562,
      "learning_rate": 0.00017447201771328148,
      "loss": 12.4965,
      "step": 9388
    },
    {
      "epoch": 0.511268573917897,
      "grad_norm": 0.5732268946297673,
      "learning_rate": 0.00017446613232239368,
      "loss": 12.4202,
      "step": 9389
    },
    {
      "epoch": 0.51132302791448,
      "grad_norm": 0.5722288630732446,
      "learning_rate": 0.00017446024635244474,
      "loss": 12.3932,
      "step": 9390
    },
    {
      "epoch": 0.511377481911063,
      "grad_norm": 0.5989469527624901,
      "learning_rate": 0.00017445435980348046,
      "loss": 12.4469,
      "step": 9391
    },
    {
      "epoch": 0.511431935907646,
      "grad_norm": 0.5669576255239827,
      "learning_rate": 0.0001744484726755466,
      "loss": 12.2855,
      "step": 9392
    },
    {
      "epoch": 0.511486389904229,
      "grad_norm": 0.6190733787516781,
      "learning_rate": 0.00017444258496868896,
      "loss": 12.5061,
      "step": 9393
    },
    {
      "epoch": 0.511540843900812,
      "grad_norm": 0.5913540005436615,
      "learning_rate": 0.00017443669668295322,
      "loss": 12.4111,
      "step": 9394
    },
    {
      "epoch": 0.5115952978973951,
      "grad_norm": 0.6273924044684714,
      "learning_rate": 0.0001744308078183853,
      "loss": 12.4899,
      "step": 9395
    },
    {
      "epoch": 0.5116497518939781,
      "grad_norm": 0.7100710497105639,
      "learning_rate": 0.00017442491837503093,
      "loss": 12.4267,
      "step": 9396
    },
    {
      "epoch": 0.5117042058905611,
      "grad_norm": 0.6071059916931937,
      "learning_rate": 0.0001744190283529359,
      "loss": 12.342,
      "step": 9397
    },
    {
      "epoch": 0.5117586598871441,
      "grad_norm": 0.6198389253712389,
      "learning_rate": 0.000174413137752146,
      "loss": 12.345,
      "step": 9398
    },
    {
      "epoch": 0.5118131138837271,
      "grad_norm": 0.6068728098719426,
      "learning_rate": 0.0001744072465727071,
      "loss": 12.3248,
      "step": 9399
    },
    {
      "epoch": 0.5118675678803101,
      "grad_norm": 0.5786485279810574,
      "learning_rate": 0.00017440135481466495,
      "loss": 12.3396,
      "step": 9400
    },
    {
      "epoch": 0.5119220218768932,
      "grad_norm": 0.6021258048098241,
      "learning_rate": 0.0001743954624780654,
      "loss": 12.41,
      "step": 9401
    },
    {
      "epoch": 0.5119764758734762,
      "grad_norm": 0.5835711794474728,
      "learning_rate": 0.00017438956956295428,
      "loss": 12.3808,
      "step": 9402
    },
    {
      "epoch": 0.5120309298700592,
      "grad_norm": 0.6170318593463341,
      "learning_rate": 0.00017438367606937733,
      "loss": 12.4245,
      "step": 9403
    },
    {
      "epoch": 0.5120853838666422,
      "grad_norm": 0.6225798029500026,
      "learning_rate": 0.0001743777819973805,
      "loss": 12.2819,
      "step": 9404
    },
    {
      "epoch": 0.5121398378632251,
      "grad_norm": 0.6924692025557676,
      "learning_rate": 0.00017437188734700952,
      "loss": 12.3891,
      "step": 9405
    },
    {
      "epoch": 0.5121942918598081,
      "grad_norm": 0.6360638878142899,
      "learning_rate": 0.00017436599211831033,
      "loss": 12.5471,
      "step": 9406
    },
    {
      "epoch": 0.5122487458563912,
      "grad_norm": 0.6901262928698174,
      "learning_rate": 0.00017436009631132865,
      "loss": 12.4722,
      "step": 9407
    },
    {
      "epoch": 0.5123031998529742,
      "grad_norm": 0.5930470801184227,
      "learning_rate": 0.0001743541999261104,
      "loss": 12.3287,
      "step": 9408
    },
    {
      "epoch": 0.5123576538495572,
      "grad_norm": 0.6015812554557252,
      "learning_rate": 0.00017434830296270145,
      "loss": 12.4187,
      "step": 9409
    },
    {
      "epoch": 0.5124121078461402,
      "grad_norm": 0.6768925144012764,
      "learning_rate": 0.00017434240542114763,
      "loss": 12.3967,
      "step": 9410
    },
    {
      "epoch": 0.5124665618427232,
      "grad_norm": 0.5927749895779888,
      "learning_rate": 0.0001743365073014948,
      "loss": 12.329,
      "step": 9411
    },
    {
      "epoch": 0.5125210158393062,
      "grad_norm": 0.6560057717129133,
      "learning_rate": 0.0001743306086037888,
      "loss": 12.4162,
      "step": 9412
    },
    {
      "epoch": 0.5125754698358893,
      "grad_norm": 0.5679942053866345,
      "learning_rate": 0.00017432470932807553,
      "loss": 12.3746,
      "step": 9413
    },
    {
      "epoch": 0.5126299238324723,
      "grad_norm": 0.5901890825124537,
      "learning_rate": 0.00017431880947440086,
      "loss": 12.3553,
      "step": 9414
    },
    {
      "epoch": 0.5126843778290553,
      "grad_norm": 0.611893604121528,
      "learning_rate": 0.00017431290904281068,
      "loss": 12.4703,
      "step": 9415
    },
    {
      "epoch": 0.5127388318256383,
      "grad_norm": 0.7235144795413267,
      "learning_rate": 0.00017430700803335082,
      "loss": 12.3188,
      "step": 9416
    },
    {
      "epoch": 0.5127932858222213,
      "grad_norm": 0.5831544036131251,
      "learning_rate": 0.00017430110644606726,
      "loss": 12.4885,
      "step": 9417
    },
    {
      "epoch": 0.5128477398188043,
      "grad_norm": 0.6350100789777187,
      "learning_rate": 0.00017429520428100578,
      "loss": 12.289,
      "step": 9418
    },
    {
      "epoch": 0.5129021938153874,
      "grad_norm": 0.6235061151953825,
      "learning_rate": 0.00017428930153821237,
      "loss": 12.3618,
      "step": 9419
    },
    {
      "epoch": 0.5129566478119704,
      "grad_norm": 0.73164702015834,
      "learning_rate": 0.00017428339821773286,
      "loss": 12.4974,
      "step": 9420
    },
    {
      "epoch": 0.5130111018085534,
      "grad_norm": 0.5782408752389678,
      "learning_rate": 0.0001742774943196132,
      "loss": 12.2829,
      "step": 9421
    },
    {
      "epoch": 0.5130655558051364,
      "grad_norm": 0.6331879886470847,
      "learning_rate": 0.0001742715898438993,
      "loss": 12.3868,
      "step": 9422
    },
    {
      "epoch": 0.5131200098017193,
      "grad_norm": 0.667002434791157,
      "learning_rate": 0.00017426568479063706,
      "loss": 12.2425,
      "step": 9423
    },
    {
      "epoch": 0.5131744637983024,
      "grad_norm": 0.7599167814718416,
      "learning_rate": 0.0001742597791598724,
      "loss": 12.2946,
      "step": 9424
    },
    {
      "epoch": 0.5132289177948854,
      "grad_norm": 0.6900616178179273,
      "learning_rate": 0.00017425387295165123,
      "loss": 12.6106,
      "step": 9425
    },
    {
      "epoch": 0.5132833717914684,
      "grad_norm": 0.629354602395147,
      "learning_rate": 0.00017424796616601952,
      "loss": 12.4014,
      "step": 9426
    },
    {
      "epoch": 0.5133378257880514,
      "grad_norm": 0.6277706451765541,
      "learning_rate": 0.00017424205880302312,
      "loss": 12.4716,
      "step": 9427
    },
    {
      "epoch": 0.5133922797846344,
      "grad_norm": 0.7167530697844063,
      "learning_rate": 0.00017423615086270807,
      "loss": 12.4673,
      "step": 9428
    },
    {
      "epoch": 0.5134467337812174,
      "grad_norm": 0.5933066126575434,
      "learning_rate": 0.00017423024234512025,
      "loss": 12.3945,
      "step": 9429
    },
    {
      "epoch": 0.5135011877778005,
      "grad_norm": 0.6372901686720958,
      "learning_rate": 0.0001742243332503056,
      "loss": 12.522,
      "step": 9430
    },
    {
      "epoch": 0.5135556417743835,
      "grad_norm": 0.6803090489789331,
      "learning_rate": 0.0001742184235783101,
      "loss": 12.2965,
      "step": 9431
    },
    {
      "epoch": 0.5136100957709665,
      "grad_norm": 0.6042308174334214,
      "learning_rate": 0.0001742125133291797,
      "loss": 12.2807,
      "step": 9432
    },
    {
      "epoch": 0.5136645497675495,
      "grad_norm": 0.6633908205859076,
      "learning_rate": 0.00017420660250296033,
      "loss": 12.5998,
      "step": 9433
    },
    {
      "epoch": 0.5137190037641325,
      "grad_norm": 0.6149312761371097,
      "learning_rate": 0.000174200691099698,
      "loss": 12.4411,
      "step": 9434
    },
    {
      "epoch": 0.5137734577607155,
      "grad_norm": 0.5980568927740617,
      "learning_rate": 0.00017419477911943864,
      "loss": 12.2741,
      "step": 9435
    },
    {
      "epoch": 0.5138279117572986,
      "grad_norm": 0.7464389865428256,
      "learning_rate": 0.0001741888665622282,
      "loss": 12.4924,
      "step": 9436
    },
    {
      "epoch": 0.5138823657538816,
      "grad_norm": 0.6250246789630683,
      "learning_rate": 0.00017418295342811274,
      "loss": 12.4717,
      "step": 9437
    },
    {
      "epoch": 0.5139368197504646,
      "grad_norm": 0.7034476056047101,
      "learning_rate": 0.00017417703971713818,
      "loss": 12.499,
      "step": 9438
    },
    {
      "epoch": 0.5139912737470476,
      "grad_norm": 0.6775469200024059,
      "learning_rate": 0.00017417112542935054,
      "loss": 12.5894,
      "step": 9439
    },
    {
      "epoch": 0.5140457277436306,
      "grad_norm": 0.6495361024701459,
      "learning_rate": 0.00017416521056479577,
      "loss": 12.4821,
      "step": 9440
    },
    {
      "epoch": 0.5141001817402135,
      "grad_norm": 0.6612506125438407,
      "learning_rate": 0.0001741592951235199,
      "loss": 12.4193,
      "step": 9441
    },
    {
      "epoch": 0.5141546357367967,
      "grad_norm": 0.5411831783413918,
      "learning_rate": 0.00017415337910556888,
      "loss": 12.2917,
      "step": 9442
    },
    {
      "epoch": 0.5142090897333796,
      "grad_norm": 0.6008695783759355,
      "learning_rate": 0.00017414746251098878,
      "loss": 12.2273,
      "step": 9443
    },
    {
      "epoch": 0.5142635437299626,
      "grad_norm": 0.6741696018952588,
      "learning_rate": 0.00017414154533982554,
      "loss": 12.3734,
      "step": 9444
    },
    {
      "epoch": 0.5143179977265456,
      "grad_norm": 0.5877583006812053,
      "learning_rate": 0.00017413562759212525,
      "loss": 12.2825,
      "step": 9445
    },
    {
      "epoch": 0.5143724517231286,
      "grad_norm": 0.6357369551078746,
      "learning_rate": 0.00017412970926793388,
      "loss": 12.4142,
      "step": 9446
    },
    {
      "epoch": 0.5144269057197116,
      "grad_norm": 0.6243208273579955,
      "learning_rate": 0.00017412379036729745,
      "loss": 12.2237,
      "step": 9447
    },
    {
      "epoch": 0.5144813597162947,
      "grad_norm": 0.6491307238798837,
      "learning_rate": 0.00017411787089026198,
      "loss": 12.3881,
      "step": 9448
    },
    {
      "epoch": 0.5145358137128777,
      "grad_norm": 0.582036403334767,
      "learning_rate": 0.00017411195083687355,
      "loss": 12.3083,
      "step": 9449
    },
    {
      "epoch": 0.5145902677094607,
      "grad_norm": 0.6185441686619526,
      "learning_rate": 0.00017410603020717814,
      "loss": 12.4296,
      "step": 9450
    },
    {
      "epoch": 0.5146447217060437,
      "grad_norm": 0.557063880310617,
      "learning_rate": 0.00017410010900122182,
      "loss": 12.3116,
      "step": 9451
    },
    {
      "epoch": 0.5146991757026267,
      "grad_norm": 0.6471912907317706,
      "learning_rate": 0.0001740941872190506,
      "loss": 12.4337,
      "step": 9452
    },
    {
      "epoch": 0.5147536296992098,
      "grad_norm": 0.7343020253109421,
      "learning_rate": 0.0001740882648607106,
      "loss": 12.4328,
      "step": 9453
    },
    {
      "epoch": 0.5148080836957928,
      "grad_norm": 0.5955622730731929,
      "learning_rate": 0.0001740823419262478,
      "loss": 12.384,
      "step": 9454
    },
    {
      "epoch": 0.5148625376923758,
      "grad_norm": 0.6158589032229121,
      "learning_rate": 0.0001740764184157083,
      "loss": 12.3919,
      "step": 9455
    },
    {
      "epoch": 0.5149169916889588,
      "grad_norm": 0.5839526669803079,
      "learning_rate": 0.00017407049432913813,
      "loss": 12.3303,
      "step": 9456
    },
    {
      "epoch": 0.5149714456855418,
      "grad_norm": 0.6667413250980353,
      "learning_rate": 0.0001740645696665834,
      "loss": 12.4874,
      "step": 9457
    },
    {
      "epoch": 0.5150258996821248,
      "grad_norm": 0.5917080616601487,
      "learning_rate": 0.00017405864442809012,
      "loss": 12.4685,
      "step": 9458
    },
    {
      "epoch": 0.5150803536787079,
      "grad_norm": 0.6153657606574432,
      "learning_rate": 0.0001740527186137044,
      "loss": 12.4207,
      "step": 9459
    },
    {
      "epoch": 0.5151348076752909,
      "grad_norm": 0.6172502537558318,
      "learning_rate": 0.00017404679222347236,
      "loss": 12.4641,
      "step": 9460
    },
    {
      "epoch": 0.5151892616718738,
      "grad_norm": 0.5826857856034895,
      "learning_rate": 0.00017404086525744003,
      "loss": 12.2974,
      "step": 9461
    },
    {
      "epoch": 0.5152437156684568,
      "grad_norm": 0.676492290899336,
      "learning_rate": 0.00017403493771565352,
      "loss": 12.2323,
      "step": 9462
    },
    {
      "epoch": 0.5152981696650398,
      "grad_norm": 0.7872839686355796,
      "learning_rate": 0.0001740290095981589,
      "loss": 12.5144,
      "step": 9463
    },
    {
      "epoch": 0.5153526236616228,
      "grad_norm": 0.6425476955150632,
      "learning_rate": 0.0001740230809050023,
      "loss": 12.489,
      "step": 9464
    },
    {
      "epoch": 0.5154070776582059,
      "grad_norm": 0.5880650070662702,
      "learning_rate": 0.0001740171516362298,
      "loss": 12.3891,
      "step": 9465
    },
    {
      "epoch": 0.5154615316547889,
      "grad_norm": 0.6126832975083883,
      "learning_rate": 0.00017401122179188755,
      "loss": 12.3447,
      "step": 9466
    },
    {
      "epoch": 0.5155159856513719,
      "grad_norm": 0.6782429959572243,
      "learning_rate": 0.0001740052913720216,
      "loss": 12.3695,
      "step": 9467
    },
    {
      "epoch": 0.5155704396479549,
      "grad_norm": 0.5803531624554442,
      "learning_rate": 0.00017399936037667808,
      "loss": 12.3674,
      "step": 9468
    },
    {
      "epoch": 0.5156248936445379,
      "grad_norm": 0.5637358898329805,
      "learning_rate": 0.00017399342880590314,
      "loss": 12.4194,
      "step": 9469
    },
    {
      "epoch": 0.5156793476411209,
      "grad_norm": 0.6286820639773689,
      "learning_rate": 0.0001739874966597429,
      "loss": 12.4978,
      "step": 9470
    },
    {
      "epoch": 0.515733801637704,
      "grad_norm": 0.6761119351876996,
      "learning_rate": 0.00017398156393824346,
      "loss": 12.4509,
      "step": 9471
    },
    {
      "epoch": 0.515788255634287,
      "grad_norm": 0.6664367437684052,
      "learning_rate": 0.00017397563064145097,
      "loss": 12.3517,
      "step": 9472
    },
    {
      "epoch": 0.51584270963087,
      "grad_norm": 0.6423591597123725,
      "learning_rate": 0.0001739696967694116,
      "loss": 12.3267,
      "step": 9473
    },
    {
      "epoch": 0.515897163627453,
      "grad_norm": 0.5995208433571146,
      "learning_rate": 0.0001739637623221714,
      "loss": 12.5092,
      "step": 9474
    },
    {
      "epoch": 0.515951617624036,
      "grad_norm": 0.660880752926758,
      "learning_rate": 0.00017395782729977665,
      "loss": 12.4551,
      "step": 9475
    },
    {
      "epoch": 0.516006071620619,
      "grad_norm": 0.5987676654785823,
      "learning_rate": 0.0001739518917022734,
      "loss": 12.3006,
      "step": 9476
    },
    {
      "epoch": 0.5160605256172021,
      "grad_norm": 0.7086942101539682,
      "learning_rate": 0.00017394595552970782,
      "loss": 12.5718,
      "step": 9477
    },
    {
      "epoch": 0.516114979613785,
      "grad_norm": 0.7278895586541723,
      "learning_rate": 0.00017394001878212614,
      "loss": 12.6292,
      "step": 9478
    },
    {
      "epoch": 0.516169433610368,
      "grad_norm": 0.6225783838940412,
      "learning_rate": 0.00017393408145957443,
      "loss": 12.3287,
      "step": 9479
    },
    {
      "epoch": 0.516223887606951,
      "grad_norm": 0.6250124899804488,
      "learning_rate": 0.00017392814356209893,
      "loss": 12.3882,
      "step": 9480
    },
    {
      "epoch": 0.516278341603534,
      "grad_norm": 0.5823883657923915,
      "learning_rate": 0.00017392220508974577,
      "loss": 12.3334,
      "step": 9481
    },
    {
      "epoch": 0.516332795600117,
      "grad_norm": 0.6525559938701772,
      "learning_rate": 0.00017391626604256114,
      "loss": 12.4486,
      "step": 9482
    },
    {
      "epoch": 0.5163872495967001,
      "grad_norm": 0.6078119554876852,
      "learning_rate": 0.00017391032642059126,
      "loss": 12.4749,
      "step": 9483
    },
    {
      "epoch": 0.5164417035932831,
      "grad_norm": 0.6965773196307412,
      "learning_rate": 0.00017390438622388228,
      "loss": 12.3629,
      "step": 9484
    },
    {
      "epoch": 0.5164961575898661,
      "grad_norm": 0.648942053086012,
      "learning_rate": 0.0001738984454524804,
      "loss": 12.1682,
      "step": 9485
    },
    {
      "epoch": 0.5165506115864491,
      "grad_norm": 0.6131617496363042,
      "learning_rate": 0.0001738925041064318,
      "loss": 12.3335,
      "step": 9486
    },
    {
      "epoch": 0.5166050655830321,
      "grad_norm": 0.5930281381763447,
      "learning_rate": 0.00017388656218578268,
      "loss": 12.4831,
      "step": 9487
    },
    {
      "epoch": 0.5166595195796152,
      "grad_norm": 0.6726746559978551,
      "learning_rate": 0.0001738806196905793,
      "loss": 12.3556,
      "step": 9488
    },
    {
      "epoch": 0.5167139735761982,
      "grad_norm": 0.6729462439247543,
      "learning_rate": 0.00017387467662086784,
      "loss": 12.3627,
      "step": 9489
    },
    {
      "epoch": 0.5167684275727812,
      "grad_norm": 0.6840423787451996,
      "learning_rate": 0.00017386873297669446,
      "loss": 12.4291,
      "step": 9490
    },
    {
      "epoch": 0.5168228815693642,
      "grad_norm": 0.727027584770321,
      "learning_rate": 0.00017386278875810543,
      "loss": 12.3196,
      "step": 9491
    },
    {
      "epoch": 0.5168773355659472,
      "grad_norm": 0.618238532380933,
      "learning_rate": 0.000173856843965147,
      "loss": 12.3743,
      "step": 9492
    },
    {
      "epoch": 0.5169317895625302,
      "grad_norm": 0.633616462992701,
      "learning_rate": 0.00017385089859786533,
      "loss": 12.3825,
      "step": 9493
    },
    {
      "epoch": 0.5169862435591133,
      "grad_norm": 0.6852956095611265,
      "learning_rate": 0.00017384495265630672,
      "loss": 12.3855,
      "step": 9494
    },
    {
      "epoch": 0.5170406975556963,
      "grad_norm": 0.6353901270857567,
      "learning_rate": 0.00017383900614051733,
      "loss": 12.3742,
      "step": 9495
    },
    {
      "epoch": 0.5170951515522793,
      "grad_norm": 0.6365417873794621,
      "learning_rate": 0.00017383305905054348,
      "loss": 12.3855,
      "step": 9496
    },
    {
      "epoch": 0.5171496055488622,
      "grad_norm": 0.6392101507202402,
      "learning_rate": 0.0001738271113864314,
      "loss": 12.3915,
      "step": 9497
    },
    {
      "epoch": 0.5172040595454452,
      "grad_norm": 0.636169211044633,
      "learning_rate": 0.0001738211631482273,
      "loss": 12.364,
      "step": 9498
    },
    {
      "epoch": 0.5172585135420282,
      "grad_norm": 0.591556032814142,
      "learning_rate": 0.00017381521433597742,
      "loss": 12.3746,
      "step": 9499
    },
    {
      "epoch": 0.5173129675386113,
      "grad_norm": 0.5989499850668308,
      "learning_rate": 0.00017380926494972812,
      "loss": 12.4657,
      "step": 9500
    },
    {
      "epoch": 0.5173674215351943,
      "grad_norm": 0.5810893580636307,
      "learning_rate": 0.00017380331498952555,
      "loss": 12.3945,
      "step": 9501
    },
    {
      "epoch": 0.5174218755317773,
      "grad_norm": 0.6071717173521719,
      "learning_rate": 0.00017379736445541604,
      "loss": 12.4434,
      "step": 9502
    },
    {
      "epoch": 0.5174763295283603,
      "grad_norm": 0.5678552708006944,
      "learning_rate": 0.00017379141334744586,
      "loss": 12.3299,
      "step": 9503
    },
    {
      "epoch": 0.5175307835249433,
      "grad_norm": 0.5897790600042302,
      "learning_rate": 0.00017378546166566124,
      "loss": 12.3707,
      "step": 9504
    },
    {
      "epoch": 0.5175852375215263,
      "grad_norm": 0.722172830820239,
      "learning_rate": 0.00017377950941010854,
      "loss": 12.5973,
      "step": 9505
    },
    {
      "epoch": 0.5176396915181094,
      "grad_norm": 0.6169290540123696,
      "learning_rate": 0.000173773556580834,
      "loss": 12.4218,
      "step": 9506
    },
    {
      "epoch": 0.5176941455146924,
      "grad_norm": 0.5773241373097306,
      "learning_rate": 0.00017376760317788388,
      "loss": 12.5499,
      "step": 9507
    },
    {
      "epoch": 0.5177485995112754,
      "grad_norm": 0.7301449323538458,
      "learning_rate": 0.0001737616492013045,
      "loss": 12.3069,
      "step": 9508
    },
    {
      "epoch": 0.5178030535078584,
      "grad_norm": 0.5773852222233629,
      "learning_rate": 0.0001737556946511422,
      "loss": 12.3977,
      "step": 9509
    },
    {
      "epoch": 0.5178575075044414,
      "grad_norm": 0.6155233235574765,
      "learning_rate": 0.00017374973952744325,
      "loss": 12.4442,
      "step": 9510
    },
    {
      "epoch": 0.5179119615010244,
      "grad_norm": 0.6083803059234462,
      "learning_rate": 0.00017374378383025392,
      "loss": 12.4416,
      "step": 9511
    },
    {
      "epoch": 0.5179664154976075,
      "grad_norm": 0.6844408691725016,
      "learning_rate": 0.0001737378275596206,
      "loss": 12.3738,
      "step": 9512
    },
    {
      "epoch": 0.5180208694941905,
      "grad_norm": 0.6869586845702121,
      "learning_rate": 0.00017373187071558955,
      "loss": 12.5113,
      "step": 9513
    },
    {
      "epoch": 0.5180753234907735,
      "grad_norm": 0.6628685359190278,
      "learning_rate": 0.00017372591329820712,
      "loss": 12.3677,
      "step": 9514
    },
    {
      "epoch": 0.5181297774873564,
      "grad_norm": 0.6463424280414908,
      "learning_rate": 0.0001737199553075196,
      "loss": 12.6094,
      "step": 9515
    },
    {
      "epoch": 0.5181842314839394,
      "grad_norm": 0.7198397510770467,
      "learning_rate": 0.00017371399674357337,
      "loss": 12.4187,
      "step": 9516
    },
    {
      "epoch": 0.5182386854805224,
      "grad_norm": 0.5829815790571139,
      "learning_rate": 0.0001737080376064147,
      "loss": 12.4891,
      "step": 9517
    },
    {
      "epoch": 0.5182931394771055,
      "grad_norm": 0.6760100979816667,
      "learning_rate": 0.00017370207789609002,
      "loss": 12.3712,
      "step": 9518
    },
    {
      "epoch": 0.5183475934736885,
      "grad_norm": 0.6053151611373706,
      "learning_rate": 0.00017369611761264563,
      "loss": 12.3323,
      "step": 9519
    },
    {
      "epoch": 0.5184020474702715,
      "grad_norm": 0.6811088166907076,
      "learning_rate": 0.0001736901567561278,
      "loss": 12.5549,
      "step": 9520
    },
    {
      "epoch": 0.5184565014668545,
      "grad_norm": 0.6683517540045927,
      "learning_rate": 0.000173684195326583,
      "loss": 12.5542,
      "step": 9521
    },
    {
      "epoch": 0.5185109554634375,
      "grad_norm": 0.6680094107438264,
      "learning_rate": 0.00017367823332405756,
      "loss": 12.4991,
      "step": 9522
    },
    {
      "epoch": 0.5185654094600206,
      "grad_norm": 0.6365234439685792,
      "learning_rate": 0.00017367227074859776,
      "loss": 12.4027,
      "step": 9523
    },
    {
      "epoch": 0.5186198634566036,
      "grad_norm": 0.6213685256896416,
      "learning_rate": 0.0001736663076002501,
      "loss": 12.2578,
      "step": 9524
    },
    {
      "epoch": 0.5186743174531866,
      "grad_norm": 0.673525027278588,
      "learning_rate": 0.00017366034387906085,
      "loss": 12.6111,
      "step": 9525
    },
    {
      "epoch": 0.5187287714497696,
      "grad_norm": 0.6045280088778902,
      "learning_rate": 0.00017365437958507637,
      "loss": 12.2815,
      "step": 9526
    },
    {
      "epoch": 0.5187832254463526,
      "grad_norm": 0.6146377983697311,
      "learning_rate": 0.00017364841471834312,
      "loss": 12.3847,
      "step": 9527
    },
    {
      "epoch": 0.5188376794429356,
      "grad_norm": 0.6224125406584722,
      "learning_rate": 0.00017364244927890746,
      "loss": 12.4285,
      "step": 9528
    },
    {
      "epoch": 0.5188921334395187,
      "grad_norm": 0.6804900827370974,
      "learning_rate": 0.00017363648326681577,
      "loss": 12.3915,
      "step": 9529
    },
    {
      "epoch": 0.5189465874361017,
      "grad_norm": 0.6187006734875188,
      "learning_rate": 0.0001736305166821144,
      "loss": 12.3694,
      "step": 9530
    },
    {
      "epoch": 0.5190010414326847,
      "grad_norm": 0.6055164489750849,
      "learning_rate": 0.0001736245495248498,
      "loss": 12.5032,
      "step": 9531
    },
    {
      "epoch": 0.5190554954292677,
      "grad_norm": 0.6791913560434254,
      "learning_rate": 0.00017361858179506837,
      "loss": 12.4238,
      "step": 9532
    },
    {
      "epoch": 0.5191099494258506,
      "grad_norm": 0.6553159563366687,
      "learning_rate": 0.0001736126134928165,
      "loss": 12.3905,
      "step": 9533
    },
    {
      "epoch": 0.5191644034224336,
      "grad_norm": 0.659497625157703,
      "learning_rate": 0.00017360664461814058,
      "loss": 12.3904,
      "step": 9534
    },
    {
      "epoch": 0.5192188574190167,
      "grad_norm": 0.5837269312568923,
      "learning_rate": 0.00017360067517108705,
      "loss": 12.2439,
      "step": 9535
    },
    {
      "epoch": 0.5192733114155997,
      "grad_norm": 0.606262104777017,
      "learning_rate": 0.00017359470515170233,
      "loss": 12.3929,
      "step": 9536
    },
    {
      "epoch": 0.5193277654121827,
      "grad_norm": 0.6481663145224712,
      "learning_rate": 0.00017358873456003283,
      "loss": 12.3585,
      "step": 9537
    },
    {
      "epoch": 0.5193822194087657,
      "grad_norm": 0.5731228553672747,
      "learning_rate": 0.00017358276339612502,
      "loss": 12.2875,
      "step": 9538
    },
    {
      "epoch": 0.5194366734053487,
      "grad_norm": 0.641688670337266,
      "learning_rate": 0.00017357679166002526,
      "loss": 12.5662,
      "step": 9539
    },
    {
      "epoch": 0.5194911274019317,
      "grad_norm": 0.6234379087948,
      "learning_rate": 0.0001735708193517801,
      "loss": 12.3924,
      "step": 9540
    },
    {
      "epoch": 0.5195455813985148,
      "grad_norm": 0.6020449493619848,
      "learning_rate": 0.00017356484647143585,
      "loss": 12.6541,
      "step": 9541
    },
    {
      "epoch": 0.5196000353950978,
      "grad_norm": 0.5694624446011876,
      "learning_rate": 0.00017355887301903904,
      "loss": 12.3805,
      "step": 9542
    },
    {
      "epoch": 0.5196544893916808,
      "grad_norm": 0.66644121793506,
      "learning_rate": 0.00017355289899463608,
      "loss": 12.4674,
      "step": 9543
    },
    {
      "epoch": 0.5197089433882638,
      "grad_norm": 0.6037137741273337,
      "learning_rate": 0.00017354692439827346,
      "loss": 12.3857,
      "step": 9544
    },
    {
      "epoch": 0.5197633973848468,
      "grad_norm": 0.6096151676019201,
      "learning_rate": 0.0001735409492299976,
      "loss": 12.3534,
      "step": 9545
    },
    {
      "epoch": 0.5198178513814298,
      "grad_norm": 0.5819488973845875,
      "learning_rate": 0.00017353497348985503,
      "loss": 12.3819,
      "step": 9546
    },
    {
      "epoch": 0.5198723053780129,
      "grad_norm": 0.7652426584827898,
      "learning_rate": 0.00017352899717789212,
      "loss": 12.4079,
      "step": 9547
    },
    {
      "epoch": 0.5199267593745959,
      "grad_norm": 0.5772110029012527,
      "learning_rate": 0.00017352302029415543,
      "loss": 12.3658,
      "step": 9548
    },
    {
      "epoch": 0.5199812133711789,
      "grad_norm": 0.5710471850503702,
      "learning_rate": 0.0001735170428386914,
      "loss": 12.4582,
      "step": 9549
    },
    {
      "epoch": 0.5200356673677619,
      "grad_norm": 0.5670726519031118,
      "learning_rate": 0.00017351106481154654,
      "loss": 12.2941,
      "step": 9550
    },
    {
      "epoch": 0.5200901213643448,
      "grad_norm": 0.7440285830960065,
      "learning_rate": 0.00017350508621276727,
      "loss": 12.4657,
      "step": 9551
    },
    {
      "epoch": 0.5201445753609278,
      "grad_norm": 0.577867669932965,
      "learning_rate": 0.00017349910704240014,
      "loss": 12.2572,
      "step": 9552
    },
    {
      "epoch": 0.520199029357511,
      "grad_norm": 0.5761404019054847,
      "learning_rate": 0.00017349312730049164,
      "loss": 12.4117,
      "step": 9553
    },
    {
      "epoch": 0.5202534833540939,
      "grad_norm": 0.6203076031921424,
      "learning_rate": 0.00017348714698708825,
      "loss": 12.4216,
      "step": 9554
    },
    {
      "epoch": 0.5203079373506769,
      "grad_norm": 0.6088496568906002,
      "learning_rate": 0.00017348116610223647,
      "loss": 12.4522,
      "step": 9555
    },
    {
      "epoch": 0.5203623913472599,
      "grad_norm": 0.6520271374564597,
      "learning_rate": 0.00017347518464598283,
      "loss": 12.4652,
      "step": 9556
    },
    {
      "epoch": 0.5204168453438429,
      "grad_norm": 0.6710044566860749,
      "learning_rate": 0.00017346920261837384,
      "loss": 12.4546,
      "step": 9557
    },
    {
      "epoch": 0.520471299340426,
      "grad_norm": 0.6386795522228851,
      "learning_rate": 0.000173463220019456,
      "loss": 12.5289,
      "step": 9558
    },
    {
      "epoch": 0.520525753337009,
      "grad_norm": 0.5617419711380065,
      "learning_rate": 0.0001734572368492758,
      "loss": 12.3144,
      "step": 9559
    },
    {
      "epoch": 0.520580207333592,
      "grad_norm": 0.6464213236433154,
      "learning_rate": 0.00017345125310787987,
      "loss": 12.5115,
      "step": 9560
    },
    {
      "epoch": 0.520634661330175,
      "grad_norm": 0.5952176049191022,
      "learning_rate": 0.00017344526879531466,
      "loss": 12.3311,
      "step": 9561
    },
    {
      "epoch": 0.520689115326758,
      "grad_norm": 0.635195905401357,
      "learning_rate": 0.00017343928391162672,
      "loss": 12.4314,
      "step": 9562
    },
    {
      "epoch": 0.520743569323341,
      "grad_norm": 0.7401122077992383,
      "learning_rate": 0.0001734332984568626,
      "loss": 12.5615,
      "step": 9563
    },
    {
      "epoch": 0.5207980233199241,
      "grad_norm": 0.6328437476921704,
      "learning_rate": 0.0001734273124310688,
      "loss": 12.3782,
      "step": 9564
    },
    {
      "epoch": 0.5208524773165071,
      "grad_norm": 0.6492627760999075,
      "learning_rate": 0.00017342132583429197,
      "loss": 12.2496,
      "step": 9565
    },
    {
      "epoch": 0.5209069313130901,
      "grad_norm": 0.6882328974866949,
      "learning_rate": 0.00017341533866657854,
      "loss": 12.5301,
      "step": 9566
    },
    {
      "epoch": 0.5209613853096731,
      "grad_norm": 0.5539351239324724,
      "learning_rate": 0.00017340935092797515,
      "loss": 12.3704,
      "step": 9567
    },
    {
      "epoch": 0.5210158393062561,
      "grad_norm": 0.6120330624036531,
      "learning_rate": 0.0001734033626185283,
      "loss": 12.4241,
      "step": 9568
    },
    {
      "epoch": 0.521070293302839,
      "grad_norm": 0.6746644635526738,
      "learning_rate": 0.00017339737373828464,
      "loss": 12.3715,
      "step": 9569
    },
    {
      "epoch": 0.5211247472994222,
      "grad_norm": 0.6251845311684804,
      "learning_rate": 0.0001733913842872907,
      "loss": 12.3197,
      "step": 9570
    },
    {
      "epoch": 0.5211792012960051,
      "grad_norm": 0.6403476161548947,
      "learning_rate": 0.00017338539426559301,
      "loss": 12.4137,
      "step": 9571
    },
    {
      "epoch": 0.5212336552925881,
      "grad_norm": 0.6340704373689795,
      "learning_rate": 0.00017337940367323818,
      "loss": 12.3527,
      "step": 9572
    },
    {
      "epoch": 0.5212881092891711,
      "grad_norm": 0.693150477056537,
      "learning_rate": 0.00017337341251027284,
      "loss": 12.5483,
      "step": 9573
    },
    {
      "epoch": 0.5213425632857541,
      "grad_norm": 0.6887097515799869,
      "learning_rate": 0.00017336742077674354,
      "loss": 12.4556,
      "step": 9574
    },
    {
      "epoch": 0.5213970172823371,
      "grad_norm": 0.6742663211247635,
      "learning_rate": 0.00017336142847269685,
      "loss": 12.1687,
      "step": 9575
    },
    {
      "epoch": 0.5214514712789202,
      "grad_norm": 0.6037349611938911,
      "learning_rate": 0.00017335543559817936,
      "loss": 12.43,
      "step": 9576
    },
    {
      "epoch": 0.5215059252755032,
      "grad_norm": 0.7071535351856332,
      "learning_rate": 0.00017334944215323777,
      "loss": 12.3907,
      "step": 9577
    },
    {
      "epoch": 0.5215603792720862,
      "grad_norm": 0.6063097686747864,
      "learning_rate": 0.00017334344813791858,
      "loss": 12.4358,
      "step": 9578
    },
    {
      "epoch": 0.5216148332686692,
      "grad_norm": 0.6568445631972262,
      "learning_rate": 0.0001733374535522684,
      "loss": 12.3488,
      "step": 9579
    },
    {
      "epoch": 0.5216692872652522,
      "grad_norm": 0.6086562414750473,
      "learning_rate": 0.00017333145839633395,
      "loss": 12.3118,
      "step": 9580
    },
    {
      "epoch": 0.5217237412618352,
      "grad_norm": 0.6720034758914826,
      "learning_rate": 0.00017332546267016173,
      "loss": 12.3787,
      "step": 9581
    },
    {
      "epoch": 0.5217781952584183,
      "grad_norm": 0.6747441173462498,
      "learning_rate": 0.00017331946637379848,
      "loss": 12.3641,
      "step": 9582
    },
    {
      "epoch": 0.5218326492550013,
      "grad_norm": 0.6362744321253769,
      "learning_rate": 0.0001733134695072907,
      "loss": 12.3927,
      "step": 9583
    },
    {
      "epoch": 0.5218871032515843,
      "grad_norm": 0.584120914581976,
      "learning_rate": 0.00017330747207068513,
      "loss": 12.3041,
      "step": 9584
    },
    {
      "epoch": 0.5219415572481673,
      "grad_norm": 0.6898276146568693,
      "learning_rate": 0.00017330147406402835,
      "loss": 12.5067,
      "step": 9585
    },
    {
      "epoch": 0.5219960112447503,
      "grad_norm": 0.6092771119092754,
      "learning_rate": 0.000173295475487367,
      "loss": 12.3398,
      "step": 9586
    },
    {
      "epoch": 0.5220504652413334,
      "grad_norm": 0.6183883034009364,
      "learning_rate": 0.00017328947634074774,
      "loss": 12.514,
      "step": 9587
    },
    {
      "epoch": 0.5221049192379164,
      "grad_norm": 0.5953627715745956,
      "learning_rate": 0.00017328347662421724,
      "loss": 12.2748,
      "step": 9588
    },
    {
      "epoch": 0.5221593732344993,
      "grad_norm": 0.6705637092696672,
      "learning_rate": 0.00017327747633782213,
      "loss": 12.388,
      "step": 9589
    },
    {
      "epoch": 0.5222138272310823,
      "grad_norm": 0.5346323503006604,
      "learning_rate": 0.00017327147548160907,
      "loss": 12.384,
      "step": 9590
    },
    {
      "epoch": 0.5222682812276653,
      "grad_norm": 0.5853878712849345,
      "learning_rate": 0.00017326547405562473,
      "loss": 12.286,
      "step": 9591
    },
    {
      "epoch": 0.5223227352242483,
      "grad_norm": 0.5975763630320771,
      "learning_rate": 0.0001732594720599158,
      "loss": 12.4346,
      "step": 9592
    },
    {
      "epoch": 0.5223771892208314,
      "grad_norm": 0.6822777284872105,
      "learning_rate": 0.00017325346949452887,
      "loss": 12.3276,
      "step": 9593
    },
    {
      "epoch": 0.5224316432174144,
      "grad_norm": 0.5940144725337047,
      "learning_rate": 0.00017324746635951073,
      "loss": 12.4804,
      "step": 9594
    },
    {
      "epoch": 0.5224860972139974,
      "grad_norm": 0.6963264100080574,
      "learning_rate": 0.00017324146265490799,
      "loss": 12.4868,
      "step": 9595
    },
    {
      "epoch": 0.5225405512105804,
      "grad_norm": 0.6454399581047955,
      "learning_rate": 0.0001732354583807674,
      "loss": 12.4155,
      "step": 9596
    },
    {
      "epoch": 0.5225950052071634,
      "grad_norm": 0.5968822395243275,
      "learning_rate": 0.00017322945353713553,
      "loss": 12.4344,
      "step": 9597
    },
    {
      "epoch": 0.5226494592037464,
      "grad_norm": 0.5772180794144417,
      "learning_rate": 0.00017322344812405917,
      "loss": 12.4009,
      "step": 9598
    },
    {
      "epoch": 0.5227039132003295,
      "grad_norm": 0.6661396756168132,
      "learning_rate": 0.000173217442141585,
      "loss": 12.5368,
      "step": 9599
    },
    {
      "epoch": 0.5227583671969125,
      "grad_norm": 0.5916943710540453,
      "learning_rate": 0.00017321143558975973,
      "loss": 12.3441,
      "step": 9600
    },
    {
      "epoch": 0.5228128211934955,
      "grad_norm": 0.6980471395490951,
      "learning_rate": 0.00017320542846863005,
      "loss": 12.4568,
      "step": 9601
    },
    {
      "epoch": 0.5228672751900785,
      "grad_norm": 0.6134060077125584,
      "learning_rate": 0.00017319942077824273,
      "loss": 12.3968,
      "step": 9602
    },
    {
      "epoch": 0.5229217291866615,
      "grad_norm": 0.6653622553223292,
      "learning_rate": 0.00017319341251864439,
      "loss": 12.3345,
      "step": 9603
    },
    {
      "epoch": 0.5229761831832445,
      "grad_norm": 0.6385386789833732,
      "learning_rate": 0.00017318740368988178,
      "loss": 12.457,
      "step": 9604
    },
    {
      "epoch": 0.5230306371798276,
      "grad_norm": 0.6778974500953844,
      "learning_rate": 0.0001731813942920017,
      "loss": 12.3487,
      "step": 9605
    },
    {
      "epoch": 0.5230850911764106,
      "grad_norm": 0.6852089521188045,
      "learning_rate": 0.00017317538432505078,
      "loss": 12.576,
      "step": 9606
    },
    {
      "epoch": 0.5231395451729935,
      "grad_norm": 0.6892428008205174,
      "learning_rate": 0.00017316937378907582,
      "loss": 12.4506,
      "step": 9607
    },
    {
      "epoch": 0.5231939991695765,
      "grad_norm": 0.6417345906514075,
      "learning_rate": 0.00017316336268412353,
      "loss": 12.3974,
      "step": 9608
    },
    {
      "epoch": 0.5232484531661595,
      "grad_norm": 0.6643182481858203,
      "learning_rate": 0.00017315735101024066,
      "loss": 12.4555,
      "step": 9609
    },
    {
      "epoch": 0.5233029071627425,
      "grad_norm": 0.7006586374584878,
      "learning_rate": 0.000173151338767474,
      "loss": 12.3776,
      "step": 9610
    },
    {
      "epoch": 0.5233573611593256,
      "grad_norm": 0.6279718640944982,
      "learning_rate": 0.00017314532595587024,
      "loss": 12.4017,
      "step": 9611
    },
    {
      "epoch": 0.5234118151559086,
      "grad_norm": 0.7037251866782321,
      "learning_rate": 0.00017313931257547613,
      "loss": 12.4936,
      "step": 9612
    },
    {
      "epoch": 0.5234662691524916,
      "grad_norm": 0.6205506357515014,
      "learning_rate": 0.00017313329862633848,
      "loss": 12.4313,
      "step": 9613
    },
    {
      "epoch": 0.5235207231490746,
      "grad_norm": 0.6943185242804921,
      "learning_rate": 0.00017312728410850405,
      "loss": 12.4645,
      "step": 9614
    },
    {
      "epoch": 0.5235751771456576,
      "grad_norm": 0.6045707069949517,
      "learning_rate": 0.0001731212690220196,
      "loss": 12.4103,
      "step": 9615
    },
    {
      "epoch": 0.5236296311422406,
      "grad_norm": 0.5648009255297014,
      "learning_rate": 0.00017311525336693184,
      "loss": 12.1542,
      "step": 9616
    },
    {
      "epoch": 0.5236840851388237,
      "grad_norm": 0.6593904156040367,
      "learning_rate": 0.00017310923714328766,
      "loss": 12.3242,
      "step": 9617
    },
    {
      "epoch": 0.5237385391354067,
      "grad_norm": 0.7689015290127676,
      "learning_rate": 0.0001731032203511338,
      "loss": 12.3943,
      "step": 9618
    },
    {
      "epoch": 0.5237929931319897,
      "grad_norm": 0.6050682600697579,
      "learning_rate": 0.000173097202990517,
      "loss": 12.3524,
      "step": 9619
    },
    {
      "epoch": 0.5238474471285727,
      "grad_norm": 0.6355378601271892,
      "learning_rate": 0.00017309118506148412,
      "loss": 12.4862,
      "step": 9620
    },
    {
      "epoch": 0.5239019011251557,
      "grad_norm": 0.625236672043179,
      "learning_rate": 0.00017308516656408193,
      "loss": 12.3745,
      "step": 9621
    },
    {
      "epoch": 0.5239563551217388,
      "grad_norm": 0.6220329845669438,
      "learning_rate": 0.0001730791474983572,
      "loss": 12.4589,
      "step": 9622
    },
    {
      "epoch": 0.5240108091183218,
      "grad_norm": 0.6320089105516793,
      "learning_rate": 0.00017307312786435679,
      "loss": 12.5162,
      "step": 9623
    },
    {
      "epoch": 0.5240652631149048,
      "grad_norm": 0.617134722308437,
      "learning_rate": 0.00017306710766212748,
      "loss": 12.4909,
      "step": 9624
    },
    {
      "epoch": 0.5241197171114877,
      "grad_norm": 0.6240965637008747,
      "learning_rate": 0.00017306108689171606,
      "loss": 12.4501,
      "step": 9625
    },
    {
      "epoch": 0.5241741711080707,
      "grad_norm": 0.5989645747623337,
      "learning_rate": 0.00017305506555316942,
      "loss": 12.495,
      "step": 9626
    },
    {
      "epoch": 0.5242286251046537,
      "grad_norm": 0.6775237364987579,
      "learning_rate": 0.00017304904364653432,
      "loss": 12.372,
      "step": 9627
    },
    {
      "epoch": 0.5242830791012368,
      "grad_norm": 0.6764909063661962,
      "learning_rate": 0.0001730430211718576,
      "loss": 12.4157,
      "step": 9628
    },
    {
      "epoch": 0.5243375330978198,
      "grad_norm": 0.5941504029768336,
      "learning_rate": 0.0001730369981291861,
      "loss": 12.2141,
      "step": 9629
    },
    {
      "epoch": 0.5243919870944028,
      "grad_norm": 0.6937835193473201,
      "learning_rate": 0.00017303097451856666,
      "loss": 12.4662,
      "step": 9630
    },
    {
      "epoch": 0.5244464410909858,
      "grad_norm": 0.5794049013015833,
      "learning_rate": 0.0001730249503400461,
      "loss": 12.391,
      "step": 9631
    },
    {
      "epoch": 0.5245008950875688,
      "grad_norm": 0.5993816672770692,
      "learning_rate": 0.0001730189255936713,
      "loss": 12.4769,
      "step": 9632
    },
    {
      "epoch": 0.5245553490841518,
      "grad_norm": 0.6214200612444263,
      "learning_rate": 0.0001730129002794891,
      "loss": 12.3002,
      "step": 9633
    },
    {
      "epoch": 0.5246098030807349,
      "grad_norm": 0.6614838174941197,
      "learning_rate": 0.00017300687439754627,
      "loss": 12.4977,
      "step": 9634
    },
    {
      "epoch": 0.5246642570773179,
      "grad_norm": 0.6361890413636067,
      "learning_rate": 0.00017300084794788983,
      "loss": 12.432,
      "step": 9635
    },
    {
      "epoch": 0.5247187110739009,
      "grad_norm": 0.6870685161675998,
      "learning_rate": 0.00017299482093056652,
      "loss": 12.509,
      "step": 9636
    },
    {
      "epoch": 0.5247731650704839,
      "grad_norm": 0.6120446501238316,
      "learning_rate": 0.00017298879334562322,
      "loss": 12.3235,
      "step": 9637
    },
    {
      "epoch": 0.5248276190670669,
      "grad_norm": 0.782081122499058,
      "learning_rate": 0.00017298276519310687,
      "loss": 12.2507,
      "step": 9638
    },
    {
      "epoch": 0.5248820730636499,
      "grad_norm": 0.6666692624757277,
      "learning_rate": 0.00017297673647306426,
      "loss": 12.5445,
      "step": 9639
    },
    {
      "epoch": 0.524936527060233,
      "grad_norm": 0.5996083202027668,
      "learning_rate": 0.00017297070718554232,
      "loss": 12.3263,
      "step": 9640
    },
    {
      "epoch": 0.524990981056816,
      "grad_norm": 0.6918954093983186,
      "learning_rate": 0.00017296467733058792,
      "loss": 12.4189,
      "step": 9641
    },
    {
      "epoch": 0.525045435053399,
      "grad_norm": 0.5641966877474066,
      "learning_rate": 0.00017295864690824794,
      "loss": 12.4129,
      "step": 9642
    },
    {
      "epoch": 0.525099889049982,
      "grad_norm": 0.6507912778509167,
      "learning_rate": 0.0001729526159185693,
      "loss": 12.5057,
      "step": 9643
    },
    {
      "epoch": 0.5251543430465649,
      "grad_norm": 0.6194648051953271,
      "learning_rate": 0.0001729465843615989,
      "loss": 12.4627,
      "step": 9644
    },
    {
      "epoch": 0.5252087970431479,
      "grad_norm": 0.6426678578240965,
      "learning_rate": 0.00017294055223738362,
      "loss": 12.4603,
      "step": 9645
    },
    {
      "epoch": 0.525263251039731,
      "grad_norm": 0.7172559439221332,
      "learning_rate": 0.00017293451954597034,
      "loss": 12.3775,
      "step": 9646
    },
    {
      "epoch": 0.525317705036314,
      "grad_norm": 0.6184352607295612,
      "learning_rate": 0.00017292848628740605,
      "loss": 12.4496,
      "step": 9647
    },
    {
      "epoch": 0.525372159032897,
      "grad_norm": 0.66662565383955,
      "learning_rate": 0.00017292245246173761,
      "loss": 12.3351,
      "step": 9648
    },
    {
      "epoch": 0.52542661302948,
      "grad_norm": 0.6044453012503522,
      "learning_rate": 0.00017291641806901193,
      "loss": 12.3472,
      "step": 9649
    },
    {
      "epoch": 0.525481067026063,
      "grad_norm": 0.6094155769366099,
      "learning_rate": 0.000172910383109276,
      "loss": 12.2924,
      "step": 9650
    },
    {
      "epoch": 0.525535521022646,
      "grad_norm": 0.6619147532997315,
      "learning_rate": 0.00017290434758257666,
      "loss": 12.3786,
      "step": 9651
    },
    {
      "epoch": 0.5255899750192291,
      "grad_norm": 0.654740006199492,
      "learning_rate": 0.0001728983114889609,
      "loss": 12.5304,
      "step": 9652
    },
    {
      "epoch": 0.5256444290158121,
      "grad_norm": 0.6360731990373002,
      "learning_rate": 0.00017289227482847563,
      "loss": 12.4277,
      "step": 9653
    },
    {
      "epoch": 0.5256988830123951,
      "grad_norm": 0.7177684346359464,
      "learning_rate": 0.00017288623760116782,
      "loss": 12.2692,
      "step": 9654
    },
    {
      "epoch": 0.5257533370089781,
      "grad_norm": 0.594606411338279,
      "learning_rate": 0.0001728801998070844,
      "loss": 12.5538,
      "step": 9655
    },
    {
      "epoch": 0.5258077910055611,
      "grad_norm": 0.61978373942687,
      "learning_rate": 0.00017287416144627237,
      "loss": 12.4174,
      "step": 9656
    },
    {
      "epoch": 0.5258622450021442,
      "grad_norm": 0.6187951813746733,
      "learning_rate": 0.00017286812251877858,
      "loss": 12.34,
      "step": 9657
    },
    {
      "epoch": 0.5259166989987272,
      "grad_norm": 0.8047359145637732,
      "learning_rate": 0.0001728620830246501,
      "loss": 12.4556,
      "step": 9658
    },
    {
      "epoch": 0.5259711529953102,
      "grad_norm": 0.6399250732558073,
      "learning_rate": 0.00017285604296393377,
      "loss": 12.4849,
      "step": 9659
    },
    {
      "epoch": 0.5260256069918932,
      "grad_norm": 0.8625060189544411,
      "learning_rate": 0.00017285000233667667,
      "loss": 12.3901,
      "step": 9660
    },
    {
      "epoch": 0.5260800609884762,
      "grad_norm": 0.589516031343896,
      "learning_rate": 0.00017284396114292574,
      "loss": 12.2904,
      "step": 9661
    },
    {
      "epoch": 0.5261345149850591,
      "grad_norm": 0.6029145395554105,
      "learning_rate": 0.00017283791938272795,
      "loss": 12.2947,
      "step": 9662
    },
    {
      "epoch": 0.5261889689816422,
      "grad_norm": 0.6933086437955884,
      "learning_rate": 0.0001728318770561303,
      "loss": 12.4769,
      "step": 9663
    },
    {
      "epoch": 0.5262434229782252,
      "grad_norm": 0.7179526791292549,
      "learning_rate": 0.00017282583416317973,
      "loss": 12.5321,
      "step": 9664
    },
    {
      "epoch": 0.5262978769748082,
      "grad_norm": 0.5920105115456481,
      "learning_rate": 0.0001728197907039233,
      "loss": 12.358,
      "step": 9665
    },
    {
      "epoch": 0.5263523309713912,
      "grad_norm": 0.6865859863106162,
      "learning_rate": 0.00017281374667840792,
      "loss": 12.4423,
      "step": 9666
    },
    {
      "epoch": 0.5264067849679742,
      "grad_norm": 0.6678078695824575,
      "learning_rate": 0.00017280770208668065,
      "loss": 12.4424,
      "step": 9667
    },
    {
      "epoch": 0.5264612389645572,
      "grad_norm": 0.607173425284794,
      "learning_rate": 0.00017280165692878848,
      "loss": 12.378,
      "step": 9668
    },
    {
      "epoch": 0.5265156929611403,
      "grad_norm": 0.6363930188981803,
      "learning_rate": 0.00017279561120477844,
      "loss": 12.3906,
      "step": 9669
    },
    {
      "epoch": 0.5265701469577233,
      "grad_norm": 0.6234394296399584,
      "learning_rate": 0.0001727895649146975,
      "loss": 12.4683,
      "step": 9670
    },
    {
      "epoch": 0.5266246009543063,
      "grad_norm": 0.5749083886394103,
      "learning_rate": 0.0001727835180585927,
      "loss": 12.4641,
      "step": 9671
    },
    {
      "epoch": 0.5266790549508893,
      "grad_norm": 0.6452980338903855,
      "learning_rate": 0.00017277747063651106,
      "loss": 12.5147,
      "step": 9672
    },
    {
      "epoch": 0.5267335089474723,
      "grad_norm": 0.5563195642011151,
      "learning_rate": 0.0001727714226484996,
      "loss": 12.3238,
      "step": 9673
    },
    {
      "epoch": 0.5267879629440553,
      "grad_norm": 0.710159581261464,
      "learning_rate": 0.00017276537409460533,
      "loss": 12.4948,
      "step": 9674
    },
    {
      "epoch": 0.5268424169406384,
      "grad_norm": 0.588714729325823,
      "learning_rate": 0.00017275932497487536,
      "loss": 12.3773,
      "step": 9675
    },
    {
      "epoch": 0.5268968709372214,
      "grad_norm": 0.6465746286814458,
      "learning_rate": 0.00017275327528935662,
      "loss": 12.3818,
      "step": 9676
    },
    {
      "epoch": 0.5269513249338044,
      "grad_norm": 0.6322551956059813,
      "learning_rate": 0.00017274722503809624,
      "loss": 12.4437,
      "step": 9677
    },
    {
      "epoch": 0.5270057789303874,
      "grad_norm": 0.5644116231357436,
      "learning_rate": 0.00017274117422114125,
      "loss": 12.3734,
      "step": 9678
    },
    {
      "epoch": 0.5270602329269704,
      "grad_norm": 0.6304177061479034,
      "learning_rate": 0.0001727351228385387,
      "loss": 12.4745,
      "step": 9679
    },
    {
      "epoch": 0.5271146869235533,
      "grad_norm": 0.567406429966043,
      "learning_rate": 0.00017272907089033559,
      "loss": 12.4316,
      "step": 9680
    },
    {
      "epoch": 0.5271691409201364,
      "grad_norm": 0.5429050725645992,
      "learning_rate": 0.00017272301837657905,
      "loss": 12.3625,
      "step": 9681
    },
    {
      "epoch": 0.5272235949167194,
      "grad_norm": 0.6655180482040186,
      "learning_rate": 0.00017271696529731612,
      "loss": 12.5425,
      "step": 9682
    },
    {
      "epoch": 0.5272780489133024,
      "grad_norm": 0.6268527508329086,
      "learning_rate": 0.00017271091165259387,
      "loss": 12.5065,
      "step": 9683
    },
    {
      "epoch": 0.5273325029098854,
      "grad_norm": 0.5693725248078682,
      "learning_rate": 0.0001727048574424594,
      "loss": 12.3665,
      "step": 9684
    },
    {
      "epoch": 0.5273869569064684,
      "grad_norm": 0.6043402385345585,
      "learning_rate": 0.00017269880266695975,
      "loss": 12.4099,
      "step": 9685
    },
    {
      "epoch": 0.5274414109030514,
      "grad_norm": 0.6818348321281018,
      "learning_rate": 0.00017269274732614203,
      "loss": 12.2967,
      "step": 9686
    },
    {
      "epoch": 0.5274958648996345,
      "grad_norm": 0.7589572456632353,
      "learning_rate": 0.00017268669142005328,
      "loss": 12.3437,
      "step": 9687
    },
    {
      "epoch": 0.5275503188962175,
      "grad_norm": 0.701238628571967,
      "learning_rate": 0.0001726806349487407,
      "loss": 12.2364,
      "step": 9688
    },
    {
      "epoch": 0.5276047728928005,
      "grad_norm": 0.6545375754439466,
      "learning_rate": 0.00017267457791225125,
      "loss": 12.5199,
      "step": 9689
    },
    {
      "epoch": 0.5276592268893835,
      "grad_norm": 0.6524621393043788,
      "learning_rate": 0.0001726685203106321,
      "loss": 12.3886,
      "step": 9690
    },
    {
      "epoch": 0.5277136808859665,
      "grad_norm": 0.6160583806710476,
      "learning_rate": 0.00017266246214393038,
      "loss": 12.3516,
      "step": 9691
    },
    {
      "epoch": 0.5277681348825496,
      "grad_norm": 0.5785027843233695,
      "learning_rate": 0.00017265640341219314,
      "loss": 12.4351,
      "step": 9692
    },
    {
      "epoch": 0.5278225888791326,
      "grad_norm": 0.6324907448675565,
      "learning_rate": 0.00017265034411546753,
      "loss": 12.4376,
      "step": 9693
    },
    {
      "epoch": 0.5278770428757156,
      "grad_norm": 0.6337729746200753,
      "learning_rate": 0.00017264428425380068,
      "loss": 12.3728,
      "step": 9694
    },
    {
      "epoch": 0.5279314968722986,
      "grad_norm": 0.5697052599210165,
      "learning_rate": 0.0001726382238272397,
      "loss": 12.3349,
      "step": 9695
    },
    {
      "epoch": 0.5279859508688816,
      "grad_norm": 0.7966711094861342,
      "learning_rate": 0.00017263216283583166,
      "loss": 12.3451,
      "step": 9696
    },
    {
      "epoch": 0.5280404048654646,
      "grad_norm": 0.5574509545166038,
      "learning_rate": 0.0001726261012796238,
      "loss": 12.4332,
      "step": 9697
    },
    {
      "epoch": 0.5280948588620477,
      "grad_norm": 0.7417944850177249,
      "learning_rate": 0.00017262003915866317,
      "loss": 12.3422,
      "step": 9698
    },
    {
      "epoch": 0.5281493128586306,
      "grad_norm": 0.7367901278701788,
      "learning_rate": 0.0001726139764729969,
      "loss": 12.3471,
      "step": 9699
    },
    {
      "epoch": 0.5282037668552136,
      "grad_norm": 0.6468213805767332,
      "learning_rate": 0.00017260791322267223,
      "loss": 12.3907,
      "step": 9700
    },
    {
      "epoch": 0.5282582208517966,
      "grad_norm": 0.5935120222843647,
      "learning_rate": 0.00017260184940773621,
      "loss": 12.3381,
      "step": 9701
    },
    {
      "epoch": 0.5283126748483796,
      "grad_norm": 0.6189022965552061,
      "learning_rate": 0.00017259578502823604,
      "loss": 12.4009,
      "step": 9702
    },
    {
      "epoch": 0.5283671288449626,
      "grad_norm": 0.6342628207844949,
      "learning_rate": 0.0001725897200842189,
      "loss": 12.3978,
      "step": 9703
    },
    {
      "epoch": 0.5284215828415457,
      "grad_norm": 0.7407930942569653,
      "learning_rate": 0.0001725836545757319,
      "loss": 12.5439,
      "step": 9704
    },
    {
      "epoch": 0.5284760368381287,
      "grad_norm": 0.6286378879070083,
      "learning_rate": 0.00017257758850282225,
      "loss": 12.3976,
      "step": 9705
    },
    {
      "epoch": 0.5285304908347117,
      "grad_norm": 0.6695499155829157,
      "learning_rate": 0.00017257152186553708,
      "loss": 12.3466,
      "step": 9706
    },
    {
      "epoch": 0.5285849448312947,
      "grad_norm": 0.6015846987937623,
      "learning_rate": 0.00017256545466392357,
      "loss": 12.2866,
      "step": 9707
    },
    {
      "epoch": 0.5286393988278777,
      "grad_norm": 0.6078027280739121,
      "learning_rate": 0.00017255938689802894,
      "loss": 12.4379,
      "step": 9708
    },
    {
      "epoch": 0.5286938528244607,
      "grad_norm": 0.638874690111006,
      "learning_rate": 0.00017255331856790034,
      "loss": 12.4772,
      "step": 9709
    },
    {
      "epoch": 0.5287483068210438,
      "grad_norm": 0.672426844846369,
      "learning_rate": 0.00017254724967358497,
      "loss": 12.3671,
      "step": 9710
    },
    {
      "epoch": 0.5288027608176268,
      "grad_norm": 0.6827308364374988,
      "learning_rate": 0.00017254118021513005,
      "loss": 12.405,
      "step": 9711
    },
    {
      "epoch": 0.5288572148142098,
      "grad_norm": 0.6486871442666722,
      "learning_rate": 0.00017253511019258273,
      "loss": 12.4867,
      "step": 9712
    },
    {
      "epoch": 0.5289116688107928,
      "grad_norm": 0.7229718641019813,
      "learning_rate": 0.00017252903960599021,
      "loss": 12.6646,
      "step": 9713
    },
    {
      "epoch": 0.5289661228073758,
      "grad_norm": 0.5710992592647909,
      "learning_rate": 0.00017252296845539974,
      "loss": 12.3733,
      "step": 9714
    },
    {
      "epoch": 0.5290205768039588,
      "grad_norm": 0.5865725811347708,
      "learning_rate": 0.00017251689674085846,
      "loss": 12.4057,
      "step": 9715
    },
    {
      "epoch": 0.5290750308005419,
      "grad_norm": 0.574147815297626,
      "learning_rate": 0.00017251082446241366,
      "loss": 12.3963,
      "step": 9716
    },
    {
      "epoch": 0.5291294847971248,
      "grad_norm": 0.5596637567178822,
      "learning_rate": 0.00017250475162011256,
      "loss": 12.31,
      "step": 9717
    },
    {
      "epoch": 0.5291839387937078,
      "grad_norm": 0.83964741392583,
      "learning_rate": 0.0001724986782140023,
      "loss": 12.428,
      "step": 9718
    },
    {
      "epoch": 0.5292383927902908,
      "grad_norm": 0.5941991496411796,
      "learning_rate": 0.00017249260424413018,
      "loss": 12.2539,
      "step": 9719
    },
    {
      "epoch": 0.5292928467868738,
      "grad_norm": 0.6831774494563473,
      "learning_rate": 0.0001724865297105434,
      "loss": 12.4378,
      "step": 9720
    },
    {
      "epoch": 0.5293473007834569,
      "grad_norm": 0.6542378395756087,
      "learning_rate": 0.00017248045461328926,
      "loss": 12.4948,
      "step": 9721
    },
    {
      "epoch": 0.5294017547800399,
      "grad_norm": 0.8125360337002615,
      "learning_rate": 0.0001724743789524149,
      "loss": 12.5094,
      "step": 9722
    },
    {
      "epoch": 0.5294562087766229,
      "grad_norm": 0.6991273418397774,
      "learning_rate": 0.00017246830272796762,
      "loss": 12.5143,
      "step": 9723
    },
    {
      "epoch": 0.5295106627732059,
      "grad_norm": 0.5880633336171539,
      "learning_rate": 0.00017246222593999468,
      "loss": 12.3835,
      "step": 9724
    },
    {
      "epoch": 0.5295651167697889,
      "grad_norm": 0.5804574850995645,
      "learning_rate": 0.0001724561485885433,
      "loss": 12.2713,
      "step": 9725
    },
    {
      "epoch": 0.5296195707663719,
      "grad_norm": 0.7335910759132532,
      "learning_rate": 0.0001724500706736608,
      "loss": 12.3935,
      "step": 9726
    },
    {
      "epoch": 0.529674024762955,
      "grad_norm": 0.5965335798603199,
      "learning_rate": 0.00017244399219539436,
      "loss": 12.3415,
      "step": 9727
    },
    {
      "epoch": 0.529728478759538,
      "grad_norm": 0.5773773691831027,
      "learning_rate": 0.0001724379131537913,
      "loss": 12.3071,
      "step": 9728
    },
    {
      "epoch": 0.529782932756121,
      "grad_norm": 0.5851148451591887,
      "learning_rate": 0.00017243183354889887,
      "loss": 12.4117,
      "step": 9729
    },
    {
      "epoch": 0.529837386752704,
      "grad_norm": 0.8309450756095572,
      "learning_rate": 0.00017242575338076435,
      "loss": 12.4456,
      "step": 9730
    },
    {
      "epoch": 0.529891840749287,
      "grad_norm": 0.6938917740148158,
      "learning_rate": 0.00017241967264943506,
      "loss": 12.3174,
      "step": 9731
    },
    {
      "epoch": 0.52994629474587,
      "grad_norm": 0.5908224767547462,
      "learning_rate": 0.00017241359135495822,
      "loss": 12.4635,
      "step": 9732
    },
    {
      "epoch": 0.5300007487424531,
      "grad_norm": 0.7278377385476976,
      "learning_rate": 0.00017240750949738115,
      "loss": 12.5118,
      "step": 9733
    },
    {
      "epoch": 0.5300552027390361,
      "grad_norm": 0.6470128783211144,
      "learning_rate": 0.00017240142707675117,
      "loss": 12.4494,
      "step": 9734
    },
    {
      "epoch": 0.530109656735619,
      "grad_norm": 0.6891827007678933,
      "learning_rate": 0.00017239534409311554,
      "loss": 12.4469,
      "step": 9735
    },
    {
      "epoch": 0.530164110732202,
      "grad_norm": 0.6342013943166225,
      "learning_rate": 0.00017238926054652157,
      "loss": 12.3601,
      "step": 9736
    },
    {
      "epoch": 0.530218564728785,
      "grad_norm": 0.6399841926597989,
      "learning_rate": 0.00017238317643701655,
      "loss": 12.425,
      "step": 9737
    },
    {
      "epoch": 0.530273018725368,
      "grad_norm": 0.6107766450079387,
      "learning_rate": 0.00017237709176464783,
      "loss": 12.4542,
      "step": 9738
    },
    {
      "epoch": 0.5303274727219511,
      "grad_norm": 0.5742461087396051,
      "learning_rate": 0.00017237100652946273,
      "loss": 12.3553,
      "step": 9739
    },
    {
      "epoch": 0.5303819267185341,
      "grad_norm": 0.645047834726115,
      "learning_rate": 0.00017236492073150852,
      "loss": 12.306,
      "step": 9740
    },
    {
      "epoch": 0.5304363807151171,
      "grad_norm": 1.1717071985382235,
      "learning_rate": 0.00017235883437083252,
      "loss": 12.4642,
      "step": 9741
    },
    {
      "epoch": 0.5304908347117001,
      "grad_norm": 0.5822341358166635,
      "learning_rate": 0.00017235274744748216,
      "loss": 12.3577,
      "step": 9742
    },
    {
      "epoch": 0.5305452887082831,
      "grad_norm": 0.6196614983883354,
      "learning_rate": 0.00017234665996150464,
      "loss": 12.4747,
      "step": 9743
    },
    {
      "epoch": 0.5305997427048661,
      "grad_norm": 0.7803176794795792,
      "learning_rate": 0.00017234057191294739,
      "loss": 12.3551,
      "step": 9744
    },
    {
      "epoch": 0.5306541967014492,
      "grad_norm": 0.5862024454149802,
      "learning_rate": 0.00017233448330185774,
      "loss": 12.2299,
      "step": 9745
    },
    {
      "epoch": 0.5307086506980322,
      "grad_norm": 0.7186574421139408,
      "learning_rate": 0.00017232839412828298,
      "loss": 12.3433,
      "step": 9746
    },
    {
      "epoch": 0.5307631046946152,
      "grad_norm": 0.7315895849141418,
      "learning_rate": 0.0001723223043922705,
      "loss": 12.3404,
      "step": 9747
    },
    {
      "epoch": 0.5308175586911982,
      "grad_norm": 0.667927166615274,
      "learning_rate": 0.00017231621409386766,
      "loss": 12.5153,
      "step": 9748
    },
    {
      "epoch": 0.5308720126877812,
      "grad_norm": 0.7132958588280709,
      "learning_rate": 0.00017231012323312182,
      "loss": 12.4401,
      "step": 9749
    },
    {
      "epoch": 0.5309264666843642,
      "grad_norm": 0.9607096123762794,
      "learning_rate": 0.00017230403181008034,
      "loss": 12.4612,
      "step": 9750
    },
    {
      "epoch": 0.5309809206809473,
      "grad_norm": 0.5842123349197016,
      "learning_rate": 0.00017229793982479058,
      "loss": 12.3212,
      "step": 9751
    },
    {
      "epoch": 0.5310353746775303,
      "grad_norm": 0.6038032722093865,
      "learning_rate": 0.0001722918472772999,
      "loss": 12.3025,
      "step": 9752
    },
    {
      "epoch": 0.5310898286741133,
      "grad_norm": 0.6729744395491271,
      "learning_rate": 0.0001722857541676557,
      "loss": 12.4765,
      "step": 9753
    },
    {
      "epoch": 0.5311442826706962,
      "grad_norm": 0.6006912560201526,
      "learning_rate": 0.00017227966049590535,
      "loss": 12.4064,
      "step": 9754
    },
    {
      "epoch": 0.5311987366672792,
      "grad_norm": 0.6781132871835124,
      "learning_rate": 0.00017227356626209626,
      "loss": 12.3,
      "step": 9755
    },
    {
      "epoch": 0.5312531906638623,
      "grad_norm": 0.6127591248377919,
      "learning_rate": 0.0001722674714662758,
      "loss": 12.4081,
      "step": 9756
    },
    {
      "epoch": 0.5313076446604453,
      "grad_norm": 0.5918857358398967,
      "learning_rate": 0.00017226137610849133,
      "loss": 12.3187,
      "step": 9757
    },
    {
      "epoch": 0.5313620986570283,
      "grad_norm": 0.6277227353304781,
      "learning_rate": 0.00017225528018879028,
      "loss": 12.3859,
      "step": 9758
    },
    {
      "epoch": 0.5314165526536113,
      "grad_norm": 0.634504921022194,
      "learning_rate": 0.0001722491837072201,
      "loss": 12.3973,
      "step": 9759
    },
    {
      "epoch": 0.5314710066501943,
      "grad_norm": 0.662465830818154,
      "learning_rate": 0.0001722430866638281,
      "loss": 12.3916,
      "step": 9760
    },
    {
      "epoch": 0.5315254606467773,
      "grad_norm": 0.5891039439024768,
      "learning_rate": 0.00017223698905866178,
      "loss": 12.3908,
      "step": 9761
    },
    {
      "epoch": 0.5315799146433604,
      "grad_norm": 0.6549544766014891,
      "learning_rate": 0.0001722308908917685,
      "loss": 12.4351,
      "step": 9762
    },
    {
      "epoch": 0.5316343686399434,
      "grad_norm": 0.6505523821972475,
      "learning_rate": 0.0001722247921631957,
      "loss": 12.3605,
      "step": 9763
    },
    {
      "epoch": 0.5316888226365264,
      "grad_norm": 0.6311228049331078,
      "learning_rate": 0.0001722186928729908,
      "loss": 12.342,
      "step": 9764
    },
    {
      "epoch": 0.5317432766331094,
      "grad_norm": 1.0094269914095666,
      "learning_rate": 0.00017221259302120125,
      "loss": 12.3917,
      "step": 9765
    },
    {
      "epoch": 0.5317977306296924,
      "grad_norm": 0.6350641093722831,
      "learning_rate": 0.00017220649260787444,
      "loss": 12.3297,
      "step": 9766
    },
    {
      "epoch": 0.5318521846262754,
      "grad_norm": 0.5972895412440087,
      "learning_rate": 0.00017220039163305786,
      "loss": 12.3677,
      "step": 9767
    },
    {
      "epoch": 0.5319066386228585,
      "grad_norm": 0.5975628558195795,
      "learning_rate": 0.0001721942900967989,
      "loss": 12.3943,
      "step": 9768
    },
    {
      "epoch": 0.5319610926194415,
      "grad_norm": 0.6235024665581594,
      "learning_rate": 0.00017218818799914507,
      "loss": 12.4072,
      "step": 9769
    },
    {
      "epoch": 0.5320155466160245,
      "grad_norm": 0.5994943679323929,
      "learning_rate": 0.00017218208534014378,
      "loss": 12.4792,
      "step": 9770
    },
    {
      "epoch": 0.5320700006126075,
      "grad_norm": 0.663856021291008,
      "learning_rate": 0.00017217598211984248,
      "loss": 12.1824,
      "step": 9771
    },
    {
      "epoch": 0.5321244546091904,
      "grad_norm": 0.6260830279897032,
      "learning_rate": 0.00017216987833828861,
      "loss": 12.2323,
      "step": 9772
    },
    {
      "epoch": 0.5321789086057734,
      "grad_norm": 0.6039866668740309,
      "learning_rate": 0.0001721637739955297,
      "loss": 12.4352,
      "step": 9773
    },
    {
      "epoch": 0.5322333626023565,
      "grad_norm": 0.6712571763891201,
      "learning_rate": 0.00017215766909161318,
      "loss": 12.4429,
      "step": 9774
    },
    {
      "epoch": 0.5322878165989395,
      "grad_norm": 0.5770603415893366,
      "learning_rate": 0.0001721515636265865,
      "loss": 12.2967,
      "step": 9775
    },
    {
      "epoch": 0.5323422705955225,
      "grad_norm": 0.6857828943633366,
      "learning_rate": 0.0001721454576004972,
      "loss": 12.5149,
      "step": 9776
    },
    {
      "epoch": 0.5323967245921055,
      "grad_norm": 0.6672425660206375,
      "learning_rate": 0.00017213935101339267,
      "loss": 12.4305,
      "step": 9777
    },
    {
      "epoch": 0.5324511785886885,
      "grad_norm": 0.7012997332857596,
      "learning_rate": 0.00017213324386532047,
      "loss": 12.259,
      "step": 9778
    },
    {
      "epoch": 0.5325056325852715,
      "grad_norm": 0.6089271523873652,
      "learning_rate": 0.0001721271361563281,
      "loss": 12.3164,
      "step": 9779
    },
    {
      "epoch": 0.5325600865818546,
      "grad_norm": 0.6629686692249517,
      "learning_rate": 0.00017212102788646298,
      "loss": 12.4173,
      "step": 9780
    },
    {
      "epoch": 0.5326145405784376,
      "grad_norm": 0.5903774965381847,
      "learning_rate": 0.00017211491905577266,
      "loss": 12.2993,
      "step": 9781
    },
    {
      "epoch": 0.5326689945750206,
      "grad_norm": 0.6902349289697002,
      "learning_rate": 0.00017210880966430465,
      "loss": 12.3484,
      "step": 9782
    },
    {
      "epoch": 0.5327234485716036,
      "grad_norm": 0.6697846969405893,
      "learning_rate": 0.00017210269971210644,
      "loss": 12.4456,
      "step": 9783
    },
    {
      "epoch": 0.5327779025681866,
      "grad_norm": 0.6201544903874386,
      "learning_rate": 0.00017209658919922554,
      "loss": 12.4618,
      "step": 9784
    },
    {
      "epoch": 0.5328323565647696,
      "grad_norm": 0.6438666747424316,
      "learning_rate": 0.00017209047812570948,
      "loss": 12.3211,
      "step": 9785
    },
    {
      "epoch": 0.5328868105613527,
      "grad_norm": 0.7061246526826793,
      "learning_rate": 0.00017208436649160578,
      "loss": 12.5599,
      "step": 9786
    },
    {
      "epoch": 0.5329412645579357,
      "grad_norm": 0.6512120724465694,
      "learning_rate": 0.00017207825429696195,
      "loss": 12.3956,
      "step": 9787
    },
    {
      "epoch": 0.5329957185545187,
      "grad_norm": 0.6008826656968721,
      "learning_rate": 0.00017207214154182552,
      "loss": 12.3315,
      "step": 9788
    },
    {
      "epoch": 0.5330501725511017,
      "grad_norm": 0.6151984110148345,
      "learning_rate": 0.00017206602822624403,
      "loss": 12.3331,
      "step": 9789
    },
    {
      "epoch": 0.5331046265476846,
      "grad_norm": 0.7381432926540967,
      "learning_rate": 0.00017205991435026503,
      "loss": 12.2305,
      "step": 9790
    },
    {
      "epoch": 0.5331590805442677,
      "grad_norm": 0.5831377313507006,
      "learning_rate": 0.00017205379991393603,
      "loss": 12.4046,
      "step": 9791
    },
    {
      "epoch": 0.5332135345408507,
      "grad_norm": 0.7278580392286879,
      "learning_rate": 0.00017204768491730464,
      "loss": 12.3752,
      "step": 9792
    },
    {
      "epoch": 0.5332679885374337,
      "grad_norm": 0.6868404243386019,
      "learning_rate": 0.00017204156936041832,
      "loss": 12.475,
      "step": 9793
    },
    {
      "epoch": 0.5333224425340167,
      "grad_norm": 0.6386140397717891,
      "learning_rate": 0.00017203545324332472,
      "loss": 12.3795,
      "step": 9794
    },
    {
      "epoch": 0.5333768965305997,
      "grad_norm": 0.741678294977336,
      "learning_rate": 0.00017202933656607132,
      "loss": 12.3816,
      "step": 9795
    },
    {
      "epoch": 0.5334313505271827,
      "grad_norm": 0.6622029228786622,
      "learning_rate": 0.00017202321932870577,
      "loss": 12.5115,
      "step": 9796
    },
    {
      "epoch": 0.5334858045237658,
      "grad_norm": 0.6693469010395893,
      "learning_rate": 0.00017201710153127552,
      "loss": 12.2848,
      "step": 9797
    },
    {
      "epoch": 0.5335402585203488,
      "grad_norm": 0.6947314114981221,
      "learning_rate": 0.00017201098317382824,
      "loss": 12.4751,
      "step": 9798
    },
    {
      "epoch": 0.5335947125169318,
      "grad_norm": 0.6436666397792562,
      "learning_rate": 0.0001720048642564115,
      "loss": 12.4354,
      "step": 9799
    },
    {
      "epoch": 0.5336491665135148,
      "grad_norm": 0.587771081254023,
      "learning_rate": 0.00017199874477907283,
      "loss": 12.3793,
      "step": 9800
    },
    {
      "epoch": 0.5337036205100978,
      "grad_norm": 0.6547024454832397,
      "learning_rate": 0.00017199262474185988,
      "loss": 12.4499,
      "step": 9801
    },
    {
      "epoch": 0.5337580745066808,
      "grad_norm": 0.6429260236613371,
      "learning_rate": 0.00017198650414482019,
      "loss": 12.3718,
      "step": 9802
    },
    {
      "epoch": 0.5338125285032639,
      "grad_norm": 0.6317710786618224,
      "learning_rate": 0.00017198038298800136,
      "loss": 12.3614,
      "step": 9803
    },
    {
      "epoch": 0.5338669824998469,
      "grad_norm": 0.6850677083619697,
      "learning_rate": 0.000171974261271451,
      "loss": 12.3093,
      "step": 9804
    },
    {
      "epoch": 0.5339214364964299,
      "grad_norm": 0.6419486047294934,
      "learning_rate": 0.00017196813899521672,
      "loss": 12.3757,
      "step": 9805
    },
    {
      "epoch": 0.5339758904930129,
      "grad_norm": 0.6135614860091514,
      "learning_rate": 0.00017196201615934614,
      "loss": 12.3681,
      "step": 9806
    },
    {
      "epoch": 0.5340303444895959,
      "grad_norm": 0.7531851068565291,
      "learning_rate": 0.00017195589276388683,
      "loss": 12.5129,
      "step": 9807
    },
    {
      "epoch": 0.5340847984861788,
      "grad_norm": 0.6420256498410256,
      "learning_rate": 0.00017194976880888642,
      "loss": 12.4711,
      "step": 9808
    },
    {
      "epoch": 0.534139252482762,
      "grad_norm": 0.5705655579112456,
      "learning_rate": 0.00017194364429439258,
      "loss": 12.3657,
      "step": 9809
    },
    {
      "epoch": 0.5341937064793449,
      "grad_norm": 0.6295191539717218,
      "learning_rate": 0.00017193751922045286,
      "loss": 12.4479,
      "step": 9810
    },
    {
      "epoch": 0.5342481604759279,
      "grad_norm": 0.6177632784310391,
      "learning_rate": 0.00017193139358711497,
      "loss": 12.477,
      "step": 9811
    },
    {
      "epoch": 0.5343026144725109,
      "grad_norm": 0.6854705931839781,
      "learning_rate": 0.00017192526739442647,
      "loss": 12.4645,
      "step": 9812
    },
    {
      "epoch": 0.5343570684690939,
      "grad_norm": 0.613644425046222,
      "learning_rate": 0.000171919140642435,
      "loss": 12.4764,
      "step": 9813
    },
    {
      "epoch": 0.5344115224656769,
      "grad_norm": 0.5929606787046219,
      "learning_rate": 0.0001719130133311883,
      "loss": 12.4075,
      "step": 9814
    },
    {
      "epoch": 0.53446597646226,
      "grad_norm": 0.6068627512682941,
      "learning_rate": 0.0001719068854607339,
      "loss": 12.3778,
      "step": 9815
    },
    {
      "epoch": 0.534520430458843,
      "grad_norm": 0.6383825000704191,
      "learning_rate": 0.00017190075703111952,
      "loss": 12.4257,
      "step": 9816
    },
    {
      "epoch": 0.534574884455426,
      "grad_norm": 0.630913310539644,
      "learning_rate": 0.0001718946280423928,
      "loss": 12.3683,
      "step": 9817
    },
    {
      "epoch": 0.534629338452009,
      "grad_norm": 0.6614522218218793,
      "learning_rate": 0.00017188849849460137,
      "loss": 12.3796,
      "step": 9818
    },
    {
      "epoch": 0.534683792448592,
      "grad_norm": 0.6034520952779118,
      "learning_rate": 0.00017188236838779295,
      "loss": 12.3671,
      "step": 9819
    },
    {
      "epoch": 0.534738246445175,
      "grad_norm": 0.6531549455633217,
      "learning_rate": 0.00017187623772201515,
      "loss": 12.368,
      "step": 9820
    },
    {
      "epoch": 0.5347927004417581,
      "grad_norm": 0.5538706811573829,
      "learning_rate": 0.0001718701064973157,
      "loss": 12.3199,
      "step": 9821
    },
    {
      "epoch": 0.5348471544383411,
      "grad_norm": 0.6744276782703781,
      "learning_rate": 0.00017186397471374222,
      "loss": 12.5405,
      "step": 9822
    },
    {
      "epoch": 0.5349016084349241,
      "grad_norm": 0.6165108152495673,
      "learning_rate": 0.00017185784237134244,
      "loss": 12.4017,
      "step": 9823
    },
    {
      "epoch": 0.5349560624315071,
      "grad_norm": 0.6047927725391143,
      "learning_rate": 0.00017185170947016403,
      "loss": 12.4021,
      "step": 9824
    },
    {
      "epoch": 0.53501051642809,
      "grad_norm": 0.677679534013877,
      "learning_rate": 0.0001718455760102547,
      "loss": 12.3044,
      "step": 9825
    },
    {
      "epoch": 0.5350649704246732,
      "grad_norm": 0.5815455964482686,
      "learning_rate": 0.00017183944199166207,
      "loss": 12.4053,
      "step": 9826
    },
    {
      "epoch": 0.5351194244212562,
      "grad_norm": 0.6780133563415122,
      "learning_rate": 0.00017183330741443392,
      "loss": 12.285,
      "step": 9827
    },
    {
      "epoch": 0.5351738784178391,
      "grad_norm": 0.662949576246597,
      "learning_rate": 0.0001718271722786179,
      "loss": 12.4557,
      "step": 9828
    },
    {
      "epoch": 0.5352283324144221,
      "grad_norm": 0.6269948079958946,
      "learning_rate": 0.00017182103658426175,
      "loss": 12.3465,
      "step": 9829
    },
    {
      "epoch": 0.5352827864110051,
      "grad_norm": 0.846084103696285,
      "learning_rate": 0.00017181490033141322,
      "loss": 12.31,
      "step": 9830
    },
    {
      "epoch": 0.5353372404075881,
      "grad_norm": 0.54992669181176,
      "learning_rate": 0.00017180876352011995,
      "loss": 12.2473,
      "step": 9831
    },
    {
      "epoch": 0.5353916944041712,
      "grad_norm": 0.7478928913354632,
      "learning_rate": 0.0001718026261504297,
      "loss": 12.3651,
      "step": 9832
    },
    {
      "epoch": 0.5354461484007542,
      "grad_norm": 0.7367492449815537,
      "learning_rate": 0.00017179648822239016,
      "loss": 12.4024,
      "step": 9833
    },
    {
      "epoch": 0.5355006023973372,
      "grad_norm": 0.5685099530599846,
      "learning_rate": 0.00017179034973604913,
      "loss": 12.4241,
      "step": 9834
    },
    {
      "epoch": 0.5355550563939202,
      "grad_norm": 0.849747866660649,
      "learning_rate": 0.00017178421069145427,
      "loss": 12.4344,
      "step": 9835
    },
    {
      "epoch": 0.5356095103905032,
      "grad_norm": 0.6246212097436993,
      "learning_rate": 0.00017177807108865336,
      "loss": 12.2952,
      "step": 9836
    },
    {
      "epoch": 0.5356639643870862,
      "grad_norm": 0.6160026522839797,
      "learning_rate": 0.00017177193092769412,
      "loss": 12.3952,
      "step": 9837
    },
    {
      "epoch": 0.5357184183836693,
      "grad_norm": 0.6626744999205778,
      "learning_rate": 0.0001717657902086243,
      "loss": 12.23,
      "step": 9838
    },
    {
      "epoch": 0.5357728723802523,
      "grad_norm": 0.6631853663477942,
      "learning_rate": 0.0001717596489314917,
      "loss": 12.4028,
      "step": 9839
    },
    {
      "epoch": 0.5358273263768353,
      "grad_norm": 0.603045212597613,
      "learning_rate": 0.00017175350709634402,
      "loss": 12.3879,
      "step": 9840
    },
    {
      "epoch": 0.5358817803734183,
      "grad_norm": 0.7324693325573454,
      "learning_rate": 0.000171747364703229,
      "loss": 12.345,
      "step": 9841
    },
    {
      "epoch": 0.5359362343700013,
      "grad_norm": 0.6550016515664254,
      "learning_rate": 0.00017174122175219448,
      "loss": 12.4589,
      "step": 9842
    },
    {
      "epoch": 0.5359906883665843,
      "grad_norm": 0.641150171504333,
      "learning_rate": 0.00017173507824328819,
      "loss": 12.3711,
      "step": 9843
    },
    {
      "epoch": 0.5360451423631674,
      "grad_norm": 0.686298156521777,
      "learning_rate": 0.00017172893417655792,
      "loss": 12.2025,
      "step": 9844
    },
    {
      "epoch": 0.5360995963597504,
      "grad_norm": 0.7136610616206259,
      "learning_rate": 0.00017172278955205136,
      "loss": 12.6081,
      "step": 9845
    },
    {
      "epoch": 0.5361540503563333,
      "grad_norm": 0.600625392297172,
      "learning_rate": 0.00017171664436981644,
      "loss": 12.3953,
      "step": 9846
    },
    {
      "epoch": 0.5362085043529163,
      "grad_norm": 0.6301836378631536,
      "learning_rate": 0.00017171049862990082,
      "loss": 12.0426,
      "step": 9847
    },
    {
      "epoch": 0.5362629583494993,
      "grad_norm": 0.6714064351021483,
      "learning_rate": 0.00017170435233235235,
      "loss": 12.2334,
      "step": 9848
    },
    {
      "epoch": 0.5363174123460823,
      "grad_norm": 0.706595530166399,
      "learning_rate": 0.0001716982054772188,
      "loss": 12.3248,
      "step": 9849
    },
    {
      "epoch": 0.5363718663426654,
      "grad_norm": 0.9192217290029557,
      "learning_rate": 0.00017169205806454797,
      "loss": 12.5167,
      "step": 9850
    },
    {
      "epoch": 0.5364263203392484,
      "grad_norm": 0.6517521190636777,
      "learning_rate": 0.0001716859100943877,
      "loss": 12.416,
      "step": 9851
    },
    {
      "epoch": 0.5364807743358314,
      "grad_norm": 0.5578077998753249,
      "learning_rate": 0.00017167976156678576,
      "loss": 12.3878,
      "step": 9852
    },
    {
      "epoch": 0.5365352283324144,
      "grad_norm": 0.5882604834303083,
      "learning_rate": 0.00017167361248178996,
      "loss": 12.3207,
      "step": 9853
    },
    {
      "epoch": 0.5365896823289974,
      "grad_norm": 0.609667657721763,
      "learning_rate": 0.00017166746283944816,
      "loss": 12.4551,
      "step": 9854
    },
    {
      "epoch": 0.5366441363255805,
      "grad_norm": 0.582794617796146,
      "learning_rate": 0.00017166131263980812,
      "loss": 12.4657,
      "step": 9855
    },
    {
      "epoch": 0.5366985903221635,
      "grad_norm": 0.6743278602101175,
      "learning_rate": 0.00017165516188291774,
      "loss": 12.4754,
      "step": 9856
    },
    {
      "epoch": 0.5367530443187465,
      "grad_norm": 0.615568670619058,
      "learning_rate": 0.00017164901056882474,
      "loss": 12.4393,
      "step": 9857
    },
    {
      "epoch": 0.5368074983153295,
      "grad_norm": 0.6096022612379464,
      "learning_rate": 0.00017164285869757705,
      "loss": 12.3968,
      "step": 9858
    },
    {
      "epoch": 0.5368619523119125,
      "grad_norm": 0.6849749417805655,
      "learning_rate": 0.0001716367062692225,
      "loss": 12.3788,
      "step": 9859
    },
    {
      "epoch": 0.5369164063084955,
      "grad_norm": 0.6798822650795383,
      "learning_rate": 0.0001716305532838089,
      "loss": 12.3282,
      "step": 9860
    },
    {
      "epoch": 0.5369708603050786,
      "grad_norm": 0.6290777902183159,
      "learning_rate": 0.00017162439974138406,
      "loss": 12.2592,
      "step": 9861
    },
    {
      "epoch": 0.5370253143016616,
      "grad_norm": 0.6034768067187262,
      "learning_rate": 0.0001716182456419959,
      "loss": 12.2475,
      "step": 9862
    },
    {
      "epoch": 0.5370797682982446,
      "grad_norm": 0.5760830659366515,
      "learning_rate": 0.00017161209098569228,
      "loss": 12.2946,
      "step": 9863
    },
    {
      "epoch": 0.5371342222948275,
      "grad_norm": 0.7272255772367627,
      "learning_rate": 0.00017160593577252102,
      "loss": 12.2768,
      "step": 9864
    },
    {
      "epoch": 0.5371886762914105,
      "grad_norm": 0.6931591166323151,
      "learning_rate": 0.00017159978000252997,
      "loss": 12.4305,
      "step": 9865
    },
    {
      "epoch": 0.5372431302879935,
      "grad_norm": 0.5948832222548013,
      "learning_rate": 0.00017159362367576706,
      "loss": 12.4153,
      "step": 9866
    },
    {
      "epoch": 0.5372975842845766,
      "grad_norm": 0.5757398920114913,
      "learning_rate": 0.0001715874667922801,
      "loss": 12.4143,
      "step": 9867
    },
    {
      "epoch": 0.5373520382811596,
      "grad_norm": 0.6099368336620453,
      "learning_rate": 0.00017158130935211697,
      "loss": 12.4317,
      "step": 9868
    },
    {
      "epoch": 0.5374064922777426,
      "grad_norm": 0.577534146116071,
      "learning_rate": 0.00017157515135532563,
      "loss": 12.4152,
      "step": 9869
    },
    {
      "epoch": 0.5374609462743256,
      "grad_norm": 0.6264027513040179,
      "learning_rate": 0.00017156899280195388,
      "loss": 12.4539,
      "step": 9870
    },
    {
      "epoch": 0.5375154002709086,
      "grad_norm": 0.5972610104177685,
      "learning_rate": 0.00017156283369204964,
      "loss": 12.4466,
      "step": 9871
    },
    {
      "epoch": 0.5375698542674916,
      "grad_norm": 0.6230837192559272,
      "learning_rate": 0.00017155667402566081,
      "loss": 12.3657,
      "step": 9872
    },
    {
      "epoch": 0.5376243082640747,
      "grad_norm": 0.587195318824459,
      "learning_rate": 0.00017155051380283526,
      "loss": 12.4126,
      "step": 9873
    },
    {
      "epoch": 0.5376787622606577,
      "grad_norm": 0.6135653893384941,
      "learning_rate": 0.00017154435302362098,
      "loss": 12.5026,
      "step": 9874
    },
    {
      "epoch": 0.5377332162572407,
      "grad_norm": 0.7244856558646927,
      "learning_rate": 0.00017153819168806575,
      "loss": 12.3259,
      "step": 9875
    },
    {
      "epoch": 0.5377876702538237,
      "grad_norm": 0.6456028771473306,
      "learning_rate": 0.00017153202979621756,
      "loss": 12.4228,
      "step": 9876
    },
    {
      "epoch": 0.5378421242504067,
      "grad_norm": 0.6730816084709884,
      "learning_rate": 0.00017152586734812432,
      "loss": 12.3757,
      "step": 9877
    },
    {
      "epoch": 0.5378965782469897,
      "grad_norm": 0.6794971975431958,
      "learning_rate": 0.00017151970434383393,
      "loss": 12.2723,
      "step": 9878
    },
    {
      "epoch": 0.5379510322435728,
      "grad_norm": 0.6763756419411008,
      "learning_rate": 0.0001715135407833943,
      "loss": 12.2668,
      "step": 9879
    },
    {
      "epoch": 0.5380054862401558,
      "grad_norm": 0.6493172433037002,
      "learning_rate": 0.00017150737666685344,
      "loss": 12.2866,
      "step": 9880
    },
    {
      "epoch": 0.5380599402367388,
      "grad_norm": 0.59240170327952,
      "learning_rate": 0.0001715012119942592,
      "loss": 12.3289,
      "step": 9881
    },
    {
      "epoch": 0.5381143942333217,
      "grad_norm": 0.7180125834990535,
      "learning_rate": 0.00017149504676565954,
      "loss": 12.4702,
      "step": 9882
    },
    {
      "epoch": 0.5381688482299047,
      "grad_norm": 0.6399415374368775,
      "learning_rate": 0.00017148888098110244,
      "loss": 12.333,
      "step": 9883
    },
    {
      "epoch": 0.5382233022264877,
      "grad_norm": 0.6806379391970908,
      "learning_rate": 0.00017148271464063574,
      "loss": 12.4301,
      "step": 9884
    },
    {
      "epoch": 0.5382777562230708,
      "grad_norm": 0.687678562545243,
      "learning_rate": 0.00017147654774430753,
      "loss": 12.289,
      "step": 9885
    },
    {
      "epoch": 0.5383322102196538,
      "grad_norm": 0.7406076855620836,
      "learning_rate": 0.00017147038029216566,
      "loss": 12.5046,
      "step": 9886
    },
    {
      "epoch": 0.5383866642162368,
      "grad_norm": 0.7058845572303966,
      "learning_rate": 0.00017146421228425815,
      "loss": 12.3471,
      "step": 9887
    },
    {
      "epoch": 0.5384411182128198,
      "grad_norm": 0.6568949054306508,
      "learning_rate": 0.00017145804372063295,
      "loss": 12.4707,
      "step": 9888
    },
    {
      "epoch": 0.5384955722094028,
      "grad_norm": 0.8311696208643002,
      "learning_rate": 0.000171451874601338,
      "loss": 12.5002,
      "step": 9889
    },
    {
      "epoch": 0.5385500262059859,
      "grad_norm": 0.7688399949416185,
      "learning_rate": 0.00017144570492642127,
      "loss": 12.1362,
      "step": 9890
    },
    {
      "epoch": 0.5386044802025689,
      "grad_norm": 0.7953307230082599,
      "learning_rate": 0.0001714395346959308,
      "loss": 12.5111,
      "step": 9891
    },
    {
      "epoch": 0.5386589341991519,
      "grad_norm": 0.6708137368403178,
      "learning_rate": 0.00017143336390991451,
      "loss": 12.3523,
      "step": 9892
    },
    {
      "epoch": 0.5387133881957349,
      "grad_norm": 0.6617697777443491,
      "learning_rate": 0.0001714271925684204,
      "loss": 12.3565,
      "step": 9893
    },
    {
      "epoch": 0.5387678421923179,
      "grad_norm": 0.6677088180209257,
      "learning_rate": 0.00017142102067149647,
      "loss": 12.4431,
      "step": 9894
    },
    {
      "epoch": 0.5388222961889009,
      "grad_norm": 0.583972439542916,
      "learning_rate": 0.00017141484821919068,
      "loss": 12.4952,
      "step": 9895
    },
    {
      "epoch": 0.538876750185484,
      "grad_norm": 0.717347607231282,
      "learning_rate": 0.0001714086752115511,
      "loss": 12.3808,
      "step": 9896
    },
    {
      "epoch": 0.538931204182067,
      "grad_norm": 0.6595704544058548,
      "learning_rate": 0.00017140250164862563,
      "loss": 12.4887,
      "step": 9897
    },
    {
      "epoch": 0.53898565817865,
      "grad_norm": 0.6045318794148249,
      "learning_rate": 0.00017139632753046237,
      "loss": 12.3262,
      "step": 9898
    },
    {
      "epoch": 0.539040112175233,
      "grad_norm": 0.6248683294042436,
      "learning_rate": 0.0001713901528571093,
      "loss": 12.3629,
      "step": 9899
    },
    {
      "epoch": 0.539094566171816,
      "grad_norm": 0.6558946098725765,
      "learning_rate": 0.0001713839776286144,
      "loss": 12.5053,
      "step": 9900
    },
    {
      "epoch": 0.5391490201683989,
      "grad_norm": 0.6491145532761589,
      "learning_rate": 0.00017137780184502574,
      "loss": 12.4372,
      "step": 9901
    },
    {
      "epoch": 0.539203474164982,
      "grad_norm": 0.6929592259968642,
      "learning_rate": 0.0001713716255063913,
      "loss": 12.3385,
      "step": 9902
    },
    {
      "epoch": 0.539257928161565,
      "grad_norm": 0.6085672631263342,
      "learning_rate": 0.00017136544861275917,
      "loss": 12.3886,
      "step": 9903
    },
    {
      "epoch": 0.539312382158148,
      "grad_norm": 0.5951696288345016,
      "learning_rate": 0.0001713592711641773,
      "loss": 12.2177,
      "step": 9904
    },
    {
      "epoch": 0.539366836154731,
      "grad_norm": 0.6179769387240053,
      "learning_rate": 0.00017135309316069382,
      "loss": 12.3241,
      "step": 9905
    },
    {
      "epoch": 0.539421290151314,
      "grad_norm": 0.6229249331386542,
      "learning_rate": 0.00017134691460235667,
      "loss": 12.386,
      "step": 9906
    },
    {
      "epoch": 0.539475744147897,
      "grad_norm": 0.6238149637276051,
      "learning_rate": 0.000171340735489214,
      "loss": 12.4721,
      "step": 9907
    },
    {
      "epoch": 0.5395301981444801,
      "grad_norm": 0.6983604951623149,
      "learning_rate": 0.00017133455582131374,
      "loss": 12.4289,
      "step": 9908
    },
    {
      "epoch": 0.5395846521410631,
      "grad_norm": 0.6441007491976519,
      "learning_rate": 0.00017132837559870407,
      "loss": 12.4204,
      "step": 9909
    },
    {
      "epoch": 0.5396391061376461,
      "grad_norm": 0.5910950723368182,
      "learning_rate": 0.00017132219482143298,
      "loss": 12.5011,
      "step": 9910
    },
    {
      "epoch": 0.5396935601342291,
      "grad_norm": 0.6216212968450316,
      "learning_rate": 0.00017131601348954853,
      "loss": 12.4437,
      "step": 9911
    },
    {
      "epoch": 0.5397480141308121,
      "grad_norm": 0.5852352225909889,
      "learning_rate": 0.0001713098316030988,
      "loss": 12.2003,
      "step": 9912
    },
    {
      "epoch": 0.5398024681273951,
      "grad_norm": 0.6166938464575757,
      "learning_rate": 0.00017130364916213186,
      "loss": 12.3618,
      "step": 9913
    },
    {
      "epoch": 0.5398569221239782,
      "grad_norm": 0.6585778033684443,
      "learning_rate": 0.00017129746616669576,
      "loss": 12.4934,
      "step": 9914
    },
    {
      "epoch": 0.5399113761205612,
      "grad_norm": 0.5912480969666162,
      "learning_rate": 0.00017129128261683863,
      "loss": 12.3631,
      "step": 9915
    },
    {
      "epoch": 0.5399658301171442,
      "grad_norm": 0.580616927395157,
      "learning_rate": 0.00017128509851260858,
      "loss": 12.4729,
      "step": 9916
    },
    {
      "epoch": 0.5400202841137272,
      "grad_norm": 0.6584369629337092,
      "learning_rate": 0.0001712789138540536,
      "loss": 12.4848,
      "step": 9917
    },
    {
      "epoch": 0.5400747381103101,
      "grad_norm": 0.6410516271365128,
      "learning_rate": 0.0001712727286412218,
      "loss": 12.1675,
      "step": 9918
    },
    {
      "epoch": 0.5401291921068931,
      "grad_norm": 0.6732803345789796,
      "learning_rate": 0.00017126654287416137,
      "loss": 12.3771,
      "step": 9919
    },
    {
      "epoch": 0.5401836461034762,
      "grad_norm": 0.733088340455232,
      "learning_rate": 0.0001712603565529203,
      "loss": 12.5004,
      "step": 9920
    },
    {
      "epoch": 0.5402381001000592,
      "grad_norm": 0.580679752904814,
      "learning_rate": 0.0001712541696775468,
      "loss": 12.3632,
      "step": 9921
    },
    {
      "epoch": 0.5402925540966422,
      "grad_norm": 0.6136950209253756,
      "learning_rate": 0.00017124798224808888,
      "loss": 12.3828,
      "step": 9922
    },
    {
      "epoch": 0.5403470080932252,
      "grad_norm": 0.666839386087224,
      "learning_rate": 0.00017124179426459475,
      "loss": 12.3126,
      "step": 9923
    },
    {
      "epoch": 0.5404014620898082,
      "grad_norm": 0.5709082525476437,
      "learning_rate": 0.00017123560572711245,
      "loss": 12.3103,
      "step": 9924
    },
    {
      "epoch": 0.5404559160863913,
      "grad_norm": 0.7116277072509676,
      "learning_rate": 0.00017122941663569012,
      "loss": 12.4852,
      "step": 9925
    },
    {
      "epoch": 0.5405103700829743,
      "grad_norm": 0.6497855971920484,
      "learning_rate": 0.00017122322699037593,
      "loss": 12.441,
      "step": 9926
    },
    {
      "epoch": 0.5405648240795573,
      "grad_norm": 0.671279944463193,
      "learning_rate": 0.00017121703679121798,
      "loss": 12.3126,
      "step": 9927
    },
    {
      "epoch": 0.5406192780761403,
      "grad_norm": 0.5639881985662691,
      "learning_rate": 0.00017121084603826438,
      "loss": 12.3626,
      "step": 9928
    },
    {
      "epoch": 0.5406737320727233,
      "grad_norm": 0.6615971952124131,
      "learning_rate": 0.00017120465473156334,
      "loss": 12.501,
      "step": 9929
    },
    {
      "epoch": 0.5407281860693063,
      "grad_norm": 0.617198644141775,
      "learning_rate": 0.00017119846287116296,
      "loss": 12.466,
      "step": 9930
    },
    {
      "epoch": 0.5407826400658894,
      "grad_norm": 0.6471490533856712,
      "learning_rate": 0.00017119227045711135,
      "loss": 12.3989,
      "step": 9931
    },
    {
      "epoch": 0.5408370940624724,
      "grad_norm": 0.5967197680845255,
      "learning_rate": 0.00017118607748945673,
      "loss": 12.3919,
      "step": 9932
    },
    {
      "epoch": 0.5408915480590554,
      "grad_norm": 0.6255433894022774,
      "learning_rate": 0.00017117988396824724,
      "loss": 12.3807,
      "step": 9933
    },
    {
      "epoch": 0.5409460020556384,
      "grad_norm": 0.6593193505343585,
      "learning_rate": 0.00017117368989353105,
      "loss": 12.4509,
      "step": 9934
    },
    {
      "epoch": 0.5410004560522214,
      "grad_norm": 0.6377169358835266,
      "learning_rate": 0.00017116749526535627,
      "loss": 12.3728,
      "step": 9935
    },
    {
      "epoch": 0.5410549100488043,
      "grad_norm": 0.5641378918184573,
      "learning_rate": 0.00017116130008377117,
      "loss": 12.3931,
      "step": 9936
    },
    {
      "epoch": 0.5411093640453875,
      "grad_norm": 0.5968616645753267,
      "learning_rate": 0.0001711551043488238,
      "loss": 12.4199,
      "step": 9937
    },
    {
      "epoch": 0.5411638180419704,
      "grad_norm": 0.6743990900177875,
      "learning_rate": 0.00017114890806056243,
      "loss": 12.4669,
      "step": 9938
    },
    {
      "epoch": 0.5412182720385534,
      "grad_norm": 0.6191577597038501,
      "learning_rate": 0.0001711427112190352,
      "loss": 12.5016,
      "step": 9939
    },
    {
      "epoch": 0.5412727260351364,
      "grad_norm": 0.5782102735631727,
      "learning_rate": 0.0001711365138242904,
      "loss": 12.2758,
      "step": 9940
    },
    {
      "epoch": 0.5413271800317194,
      "grad_norm": 0.6275492666648163,
      "learning_rate": 0.00017113031587637608,
      "loss": 12.3231,
      "step": 9941
    },
    {
      "epoch": 0.5413816340283024,
      "grad_norm": 0.5737239024137321,
      "learning_rate": 0.0001711241173753405,
      "loss": 12.2454,
      "step": 9942
    },
    {
      "epoch": 0.5414360880248855,
      "grad_norm": 0.5841077817920884,
      "learning_rate": 0.00017111791832123184,
      "loss": 12.3617,
      "step": 9943
    },
    {
      "epoch": 0.5414905420214685,
      "grad_norm": 0.5125432900399025,
      "learning_rate": 0.00017111171871409835,
      "loss": 12.2638,
      "step": 9944
    },
    {
      "epoch": 0.5415449960180515,
      "grad_norm": 0.627035765756675,
      "learning_rate": 0.00017110551855398817,
      "loss": 12.427,
      "step": 9945
    },
    {
      "epoch": 0.5415994500146345,
      "grad_norm": 0.6375482313018308,
      "learning_rate": 0.0001710993178409496,
      "loss": 12.319,
      "step": 9946
    },
    {
      "epoch": 0.5416539040112175,
      "grad_norm": 0.6194514620473593,
      "learning_rate": 0.00017109311657503078,
      "loss": 12.3979,
      "step": 9947
    },
    {
      "epoch": 0.5417083580078005,
      "grad_norm": 0.6710169185222208,
      "learning_rate": 0.00017108691475627996,
      "loss": 12.3065,
      "step": 9948
    },
    {
      "epoch": 0.5417628120043836,
      "grad_norm": 0.6494699821871501,
      "learning_rate": 0.00017108071238474537,
      "loss": 12.4168,
      "step": 9949
    },
    {
      "epoch": 0.5418172660009666,
      "grad_norm": 0.6433398470683385,
      "learning_rate": 0.00017107450946047528,
      "loss": 12.3035,
      "step": 9950
    },
    {
      "epoch": 0.5418717199975496,
      "grad_norm": 0.5649414899964764,
      "learning_rate": 0.00017106830598351784,
      "loss": 12.3674,
      "step": 9951
    },
    {
      "epoch": 0.5419261739941326,
      "grad_norm": 0.5838873388720432,
      "learning_rate": 0.00017106210195392136,
      "loss": 12.4877,
      "step": 9952
    },
    {
      "epoch": 0.5419806279907156,
      "grad_norm": 0.6215618900230296,
      "learning_rate": 0.00017105589737173403,
      "loss": 12.3699,
      "step": 9953
    },
    {
      "epoch": 0.5420350819872986,
      "grad_norm": 0.5427453094602321,
      "learning_rate": 0.00017104969223700415,
      "loss": 12.2128,
      "step": 9954
    },
    {
      "epoch": 0.5420895359838817,
      "grad_norm": 0.5825640076004421,
      "learning_rate": 0.00017104348654977994,
      "loss": 12.3704,
      "step": 9955
    },
    {
      "epoch": 0.5421439899804646,
      "grad_norm": 0.611865041734815,
      "learning_rate": 0.00017103728031010967,
      "loss": 12.3571,
      "step": 9956
    },
    {
      "epoch": 0.5421984439770476,
      "grad_norm": 0.6584718062782783,
      "learning_rate": 0.0001710310735180416,
      "loss": 12.4224,
      "step": 9957
    },
    {
      "epoch": 0.5422528979736306,
      "grad_norm": 0.5394590262995481,
      "learning_rate": 0.000171024866173624,
      "loss": 12.3904,
      "step": 9958
    },
    {
      "epoch": 0.5423073519702136,
      "grad_norm": 0.6251349184276233,
      "learning_rate": 0.00017101865827690512,
      "loss": 12.4797,
      "step": 9959
    },
    {
      "epoch": 0.5423618059667967,
      "grad_norm": 0.6043745031335845,
      "learning_rate": 0.0001710124498279332,
      "loss": 12.4913,
      "step": 9960
    },
    {
      "epoch": 0.5424162599633797,
      "grad_norm": 0.6416028422300966,
      "learning_rate": 0.00017100624082675662,
      "loss": 12.3101,
      "step": 9961
    },
    {
      "epoch": 0.5424707139599627,
      "grad_norm": 0.586956971175586,
      "learning_rate": 0.00017100003127342358,
      "loss": 12.4128,
      "step": 9962
    },
    {
      "epoch": 0.5425251679565457,
      "grad_norm": 0.5895692580881068,
      "learning_rate": 0.0001709938211679824,
      "loss": 12.3444,
      "step": 9963
    },
    {
      "epoch": 0.5425796219531287,
      "grad_norm": 0.6892529704657986,
      "learning_rate": 0.00017098761051048133,
      "loss": 12.6901,
      "step": 9964
    },
    {
      "epoch": 0.5426340759497117,
      "grad_norm": 0.5922876095576688,
      "learning_rate": 0.00017098139930096874,
      "loss": 12.1998,
      "step": 9965
    },
    {
      "epoch": 0.5426885299462948,
      "grad_norm": 0.6353639822693112,
      "learning_rate": 0.00017097518753949286,
      "loss": 12.3397,
      "step": 9966
    },
    {
      "epoch": 0.5427429839428778,
      "grad_norm": 0.6014554204328489,
      "learning_rate": 0.00017096897522610202,
      "loss": 12.4212,
      "step": 9967
    },
    {
      "epoch": 0.5427974379394608,
      "grad_norm": 0.6129355783996963,
      "learning_rate": 0.00017096276236084452,
      "loss": 12.3241,
      "step": 9968
    },
    {
      "epoch": 0.5428518919360438,
      "grad_norm": 0.6168784082239562,
      "learning_rate": 0.0001709565489437687,
      "loss": 12.3898,
      "step": 9969
    },
    {
      "epoch": 0.5429063459326268,
      "grad_norm": 0.5455036782520887,
      "learning_rate": 0.00017095033497492286,
      "loss": 12.2893,
      "step": 9970
    },
    {
      "epoch": 0.5429607999292098,
      "grad_norm": 0.6710022912122282,
      "learning_rate": 0.00017094412045435528,
      "loss": 12.3584,
      "step": 9971
    },
    {
      "epoch": 0.5430152539257929,
      "grad_norm": 0.6179083424353565,
      "learning_rate": 0.00017093790538211433,
      "loss": 12.579,
      "step": 9972
    },
    {
      "epoch": 0.5430697079223759,
      "grad_norm": 0.5654658232435088,
      "learning_rate": 0.00017093168975824838,
      "loss": 12.3976,
      "step": 9973
    },
    {
      "epoch": 0.5431241619189588,
      "grad_norm": 0.5869301776598929,
      "learning_rate": 0.0001709254735828057,
      "loss": 12.4018,
      "step": 9974
    },
    {
      "epoch": 0.5431786159155418,
      "grad_norm": 0.5699624764075849,
      "learning_rate": 0.0001709192568558346,
      "loss": 12.3251,
      "step": 9975
    },
    {
      "epoch": 0.5432330699121248,
      "grad_norm": 0.6164089422163238,
      "learning_rate": 0.00017091303957738347,
      "loss": 12.4561,
      "step": 9976
    },
    {
      "epoch": 0.5432875239087078,
      "grad_norm": 0.6110377746277542,
      "learning_rate": 0.00017090682174750069,
      "loss": 12.3636,
      "step": 9977
    },
    {
      "epoch": 0.5433419779052909,
      "grad_norm": 0.6089179595565541,
      "learning_rate": 0.00017090060336623456,
      "loss": 12.4271,
      "step": 9978
    },
    {
      "epoch": 0.5433964319018739,
      "grad_norm": 0.6118550994671074,
      "learning_rate": 0.00017089438443363344,
      "loss": 12.4179,
      "step": 9979
    },
    {
      "epoch": 0.5434508858984569,
      "grad_norm": 0.7076959617147793,
      "learning_rate": 0.0001708881649497457,
      "loss": 12.4697,
      "step": 9980
    },
    {
      "epoch": 0.5435053398950399,
      "grad_norm": 0.6383711008143893,
      "learning_rate": 0.00017088194491461967,
      "loss": 12.1758,
      "step": 9981
    },
    {
      "epoch": 0.5435597938916229,
      "grad_norm": 0.6280069425682963,
      "learning_rate": 0.00017087572432830382,
      "loss": 12.3434,
      "step": 9982
    },
    {
      "epoch": 0.5436142478882059,
      "grad_norm": 0.6599043049497079,
      "learning_rate": 0.0001708695031908464,
      "loss": 12.4343,
      "step": 9983
    },
    {
      "epoch": 0.543668701884789,
      "grad_norm": 0.5897655562410326,
      "learning_rate": 0.00017086328150229586,
      "loss": 12.3205,
      "step": 9984
    },
    {
      "epoch": 0.543723155881372,
      "grad_norm": 0.7437039719219413,
      "learning_rate": 0.00017085705926270058,
      "loss": 12.4402,
      "step": 9985
    },
    {
      "epoch": 0.543777609877955,
      "grad_norm": 0.6357530106706325,
      "learning_rate": 0.00017085083647210887,
      "loss": 12.3601,
      "step": 9986
    },
    {
      "epoch": 0.543832063874538,
      "grad_norm": 0.6374285771150914,
      "learning_rate": 0.0001708446131305692,
      "loss": 12.5205,
      "step": 9987
    },
    {
      "epoch": 0.543886517871121,
      "grad_norm": 0.5896956056677658,
      "learning_rate": 0.00017083838923812993,
      "loss": 12.3106,
      "step": 9988
    },
    {
      "epoch": 0.5439409718677041,
      "grad_norm": 0.6115372665262078,
      "learning_rate": 0.00017083216479483947,
      "loss": 12.3934,
      "step": 9989
    },
    {
      "epoch": 0.5439954258642871,
      "grad_norm": 0.6755730245071239,
      "learning_rate": 0.00017082593980074625,
      "loss": 12.3121,
      "step": 9990
    },
    {
      "epoch": 0.5440498798608701,
      "grad_norm": 0.6178738428597439,
      "learning_rate": 0.00017081971425589857,
      "loss": 12.3926,
      "step": 9991
    },
    {
      "epoch": 0.544104333857453,
      "grad_norm": 0.6346592570491943,
      "learning_rate": 0.00017081348816034496,
      "loss": 12.5108,
      "step": 9992
    },
    {
      "epoch": 0.544158787854036,
      "grad_norm": 0.8524667556741318,
      "learning_rate": 0.00017080726151413381,
      "loss": 12.514,
      "step": 9993
    },
    {
      "epoch": 0.544213241850619,
      "grad_norm": 0.6071086314951164,
      "learning_rate": 0.00017080103431731352,
      "loss": 12.494,
      "step": 9994
    },
    {
      "epoch": 0.5442676958472021,
      "grad_norm": 0.5870626705245917,
      "learning_rate": 0.00017079480656993247,
      "loss": 12.3616,
      "step": 9995
    },
    {
      "epoch": 0.5443221498437851,
      "grad_norm": 0.6119188183797124,
      "learning_rate": 0.00017078857827203917,
      "loss": 12.5749,
      "step": 9996
    },
    {
      "epoch": 0.5443766038403681,
      "grad_norm": 0.6068760617708026,
      "learning_rate": 0.00017078234942368198,
      "loss": 12.4392,
      "step": 9997
    },
    {
      "epoch": 0.5444310578369511,
      "grad_norm": 0.6126603120868624,
      "learning_rate": 0.00017077612002490942,
      "loss": 12.3884,
      "step": 9998
    },
    {
      "epoch": 0.5444855118335341,
      "grad_norm": 0.5868375210438797,
      "learning_rate": 0.00017076989007576985,
      "loss": 12.3587,
      "step": 9999
    },
    {
      "epoch": 0.5445399658301171,
      "grad_norm": 0.6864597119491456,
      "learning_rate": 0.00017076365957631174,
      "loss": 12.4161,
      "step": 10000
    },
    {
      "epoch": 0.5445944198267002,
      "grad_norm": 0.622011095040674,
      "learning_rate": 0.00017075742852658355,
      "loss": 12.3495,
      "step": 10001
    },
    {
      "epoch": 0.5446488738232832,
      "grad_norm": 0.5766789005686187,
      "learning_rate": 0.00017075119692663374,
      "loss": 12.4083,
      "step": 10002
    },
    {
      "epoch": 0.5447033278198662,
      "grad_norm": 0.5888590620506866,
      "learning_rate": 0.0001707449647765108,
      "loss": 12.2743,
      "step": 10003
    },
    {
      "epoch": 0.5447577818164492,
      "grad_norm": 0.5877421821137575,
      "learning_rate": 0.00017073873207626309,
      "loss": 12.3029,
      "step": 10004
    },
    {
      "epoch": 0.5448122358130322,
      "grad_norm": 0.618434215481345,
      "learning_rate": 0.00017073249882593912,
      "loss": 12.2655,
      "step": 10005
    },
    {
      "epoch": 0.5448666898096152,
      "grad_norm": 0.6291436904487675,
      "learning_rate": 0.00017072626502558742,
      "loss": 12.3931,
      "step": 10006
    },
    {
      "epoch": 0.5449211438061983,
      "grad_norm": 0.5865487556526245,
      "learning_rate": 0.0001707200306752564,
      "loss": 12.2635,
      "step": 10007
    },
    {
      "epoch": 0.5449755978027813,
      "grad_norm": 0.6079297100272073,
      "learning_rate": 0.00017071379577499458,
      "loss": 12.4051,
      "step": 10008
    },
    {
      "epoch": 0.5450300517993643,
      "grad_norm": 0.6016245121971452,
      "learning_rate": 0.00017070756032485043,
      "loss": 12.3416,
      "step": 10009
    },
    {
      "epoch": 0.5450845057959472,
      "grad_norm": 0.5856170250429272,
      "learning_rate": 0.00017070132432487242,
      "loss": 12.3265,
      "step": 10010
    },
    {
      "epoch": 0.5451389597925302,
      "grad_norm": 0.6551742421540866,
      "learning_rate": 0.00017069508777510904,
      "loss": 12.4072,
      "step": 10011
    },
    {
      "epoch": 0.5451934137891132,
      "grad_norm": 0.666724847008867,
      "learning_rate": 0.00017068885067560884,
      "loss": 12.453,
      "step": 10012
    },
    {
      "epoch": 0.5452478677856963,
      "grad_norm": 0.5980244266505121,
      "learning_rate": 0.00017068261302642025,
      "loss": 12.4869,
      "step": 10013
    },
    {
      "epoch": 0.5453023217822793,
      "grad_norm": 0.6442562753056685,
      "learning_rate": 0.00017067637482759182,
      "loss": 12.4074,
      "step": 10014
    },
    {
      "epoch": 0.5453567757788623,
      "grad_norm": 0.5781895247879832,
      "learning_rate": 0.00017067013607917204,
      "loss": 12.2852,
      "step": 10015
    },
    {
      "epoch": 0.5454112297754453,
      "grad_norm": 0.5659964634909223,
      "learning_rate": 0.00017066389678120942,
      "loss": 12.3117,
      "step": 10016
    },
    {
      "epoch": 0.5454656837720283,
      "grad_norm": 0.5984647462305562,
      "learning_rate": 0.00017065765693375254,
      "loss": 12.2896,
      "step": 10017
    },
    {
      "epoch": 0.5455201377686113,
      "grad_norm": 0.5927320661463614,
      "learning_rate": 0.0001706514165368498,
      "loss": 12.3072,
      "step": 10018
    },
    {
      "epoch": 0.5455745917651944,
      "grad_norm": 0.6037064365778668,
      "learning_rate": 0.00017064517559054983,
      "loss": 12.3729,
      "step": 10019
    },
    {
      "epoch": 0.5456290457617774,
      "grad_norm": 0.5755367621481371,
      "learning_rate": 0.00017063893409490115,
      "loss": 12.409,
      "step": 10020
    },
    {
      "epoch": 0.5456834997583604,
      "grad_norm": 0.6320289142883861,
      "learning_rate": 0.00017063269204995222,
      "loss": 12.3564,
      "step": 10021
    },
    {
      "epoch": 0.5457379537549434,
      "grad_norm": 0.6002473882539896,
      "learning_rate": 0.00017062644945575167,
      "loss": 12.3511,
      "step": 10022
    },
    {
      "epoch": 0.5457924077515264,
      "grad_norm": 0.5886482899613877,
      "learning_rate": 0.000170620206312348,
      "loss": 12.3213,
      "step": 10023
    },
    {
      "epoch": 0.5458468617481095,
      "grad_norm": 0.5881266992555014,
      "learning_rate": 0.00017061396261978978,
      "loss": 12.3273,
      "step": 10024
    },
    {
      "epoch": 0.5459013157446925,
      "grad_norm": 0.6677865513934675,
      "learning_rate": 0.00017060771837812552,
      "loss": 12.448,
      "step": 10025
    },
    {
      "epoch": 0.5459557697412755,
      "grad_norm": 0.5722455480945151,
      "learning_rate": 0.0001706014735874038,
      "loss": 12.4596,
      "step": 10026
    },
    {
      "epoch": 0.5460102237378585,
      "grad_norm": 0.657568952206474,
      "learning_rate": 0.00017059522824767318,
      "loss": 12.4349,
      "step": 10027
    },
    {
      "epoch": 0.5460646777344415,
      "grad_norm": 0.6101702886040364,
      "learning_rate": 0.00017058898235898225,
      "loss": 12.1885,
      "step": 10028
    },
    {
      "epoch": 0.5461191317310244,
      "grad_norm": 0.6164690188553941,
      "learning_rate": 0.00017058273592137954,
      "loss": 12.3377,
      "step": 10029
    },
    {
      "epoch": 0.5461735857276075,
      "grad_norm": 0.5743311102393344,
      "learning_rate": 0.00017057648893491363,
      "loss": 12.3184,
      "step": 10030
    },
    {
      "epoch": 0.5462280397241905,
      "grad_norm": 0.6036523761968633,
      "learning_rate": 0.0001705702413996331,
      "loss": 12.3641,
      "step": 10031
    },
    {
      "epoch": 0.5462824937207735,
      "grad_norm": 0.6380111248095689,
      "learning_rate": 0.00017056399331558656,
      "loss": 12.3799,
      "step": 10032
    },
    {
      "epoch": 0.5463369477173565,
      "grad_norm": 0.7449245804750874,
      "learning_rate": 0.00017055774468282257,
      "loss": 12.4631,
      "step": 10033
    },
    {
      "epoch": 0.5463914017139395,
      "grad_norm": 0.602894598301536,
      "learning_rate": 0.00017055149550138974,
      "loss": 12.4042,
      "step": 10034
    },
    {
      "epoch": 0.5464458557105225,
      "grad_norm": 0.5657734285602912,
      "learning_rate": 0.0001705452457713366,
      "loss": 12.451,
      "step": 10035
    },
    {
      "epoch": 0.5465003097071056,
      "grad_norm": 0.576113903088083,
      "learning_rate": 0.0001705389954927118,
      "loss": 12.3255,
      "step": 10036
    },
    {
      "epoch": 0.5465547637036886,
      "grad_norm": 0.6616302304908845,
      "learning_rate": 0.000170532744665564,
      "loss": 12.4435,
      "step": 10037
    },
    {
      "epoch": 0.5466092177002716,
      "grad_norm": 0.6821592199096708,
      "learning_rate": 0.00017052649328994174,
      "loss": 12.5157,
      "step": 10038
    },
    {
      "epoch": 0.5466636716968546,
      "grad_norm": 0.6096340643614813,
      "learning_rate": 0.0001705202413658936,
      "loss": 12.4992,
      "step": 10039
    },
    {
      "epoch": 0.5467181256934376,
      "grad_norm": 0.6176713315553298,
      "learning_rate": 0.0001705139888934683,
      "loss": 12.2644,
      "step": 10040
    },
    {
      "epoch": 0.5467725796900206,
      "grad_norm": 0.5861837007873087,
      "learning_rate": 0.00017050773587271433,
      "loss": 12.2729,
      "step": 10041
    },
    {
      "epoch": 0.5468270336866037,
      "grad_norm": 0.5764941831254127,
      "learning_rate": 0.0001705014823036804,
      "loss": 12.3468,
      "step": 10042
    },
    {
      "epoch": 0.5468814876831867,
      "grad_norm": 0.6265747671891215,
      "learning_rate": 0.00017049522818641513,
      "loss": 12.3629,
      "step": 10043
    },
    {
      "epoch": 0.5469359416797697,
      "grad_norm": 0.8012125745132999,
      "learning_rate": 0.00017048897352096713,
      "loss": 12.3787,
      "step": 10044
    },
    {
      "epoch": 0.5469903956763527,
      "grad_norm": 0.6351309622415717,
      "learning_rate": 0.00017048271830738507,
      "loss": 12.4865,
      "step": 10045
    },
    {
      "epoch": 0.5470448496729357,
      "grad_norm": 0.5905862443496168,
      "learning_rate": 0.00017047646254571755,
      "loss": 12.322,
      "step": 10046
    },
    {
      "epoch": 0.5470993036695186,
      "grad_norm": 0.6815468712793522,
      "learning_rate": 0.00017047020623601328,
      "loss": 12.4611,
      "step": 10047
    },
    {
      "epoch": 0.5471537576661017,
      "grad_norm": 0.6174565560323156,
      "learning_rate": 0.00017046394937832084,
      "loss": 12.329,
      "step": 10048
    },
    {
      "epoch": 0.5472082116626847,
      "grad_norm": 0.5276648368815748,
      "learning_rate": 0.00017045769197268892,
      "loss": 12.2771,
      "step": 10049
    },
    {
      "epoch": 0.5472626656592677,
      "grad_norm": 0.6482574075836264,
      "learning_rate": 0.00017045143401916613,
      "loss": 12.3731,
      "step": 10050
    },
    {
      "epoch": 0.5473171196558507,
      "grad_norm": 0.7609587337450041,
      "learning_rate": 0.00017044517551780125,
      "loss": 12.3264,
      "step": 10051
    },
    {
      "epoch": 0.5473715736524337,
      "grad_norm": 0.5531086997696827,
      "learning_rate": 0.0001704389164686428,
      "loss": 12.3699,
      "step": 10052
    },
    {
      "epoch": 0.5474260276490167,
      "grad_norm": 0.6909288910934123,
      "learning_rate": 0.00017043265687173955,
      "loss": 12.4163,
      "step": 10053
    },
    {
      "epoch": 0.5474804816455998,
      "grad_norm": 0.6600937201127297,
      "learning_rate": 0.00017042639672714015,
      "loss": 12.3825,
      "step": 10054
    },
    {
      "epoch": 0.5475349356421828,
      "grad_norm": 0.69648958811755,
      "learning_rate": 0.0001704201360348933,
      "loss": 12.2439,
      "step": 10055
    },
    {
      "epoch": 0.5475893896387658,
      "grad_norm": 0.5877556274036734,
      "learning_rate": 0.00017041387479504764,
      "loss": 12.2162,
      "step": 10056
    },
    {
      "epoch": 0.5476438436353488,
      "grad_norm": 0.5761911274527789,
      "learning_rate": 0.00017040761300765188,
      "loss": 12.3922,
      "step": 10057
    },
    {
      "epoch": 0.5476982976319318,
      "grad_norm": 0.7246599447399307,
      "learning_rate": 0.00017040135067275473,
      "loss": 12.3671,
      "step": 10058
    },
    {
      "epoch": 0.5477527516285149,
      "grad_norm": 0.5770976085961329,
      "learning_rate": 0.00017039508779040485,
      "loss": 12.3825,
      "step": 10059
    },
    {
      "epoch": 0.5478072056250979,
      "grad_norm": 0.6303142158584125,
      "learning_rate": 0.00017038882436065097,
      "loss": 12.4369,
      "step": 10060
    },
    {
      "epoch": 0.5478616596216809,
      "grad_norm": 0.6858517319715612,
      "learning_rate": 0.0001703825603835418,
      "loss": 12.349,
      "step": 10061
    },
    {
      "epoch": 0.5479161136182639,
      "grad_norm": 0.6793237037550006,
      "learning_rate": 0.000170376295859126,
      "loss": 12.3091,
      "step": 10062
    },
    {
      "epoch": 0.5479705676148469,
      "grad_norm": 0.6272479682486032,
      "learning_rate": 0.00017037003078745238,
      "loss": 12.3884,
      "step": 10063
    },
    {
      "epoch": 0.5480250216114299,
      "grad_norm": 0.6504731696059936,
      "learning_rate": 0.00017036376516856955,
      "loss": 12.387,
      "step": 10064
    },
    {
      "epoch": 0.548079475608013,
      "grad_norm": 0.6217625065902543,
      "learning_rate": 0.00017035749900252628,
      "loss": 12.4143,
      "step": 10065
    },
    {
      "epoch": 0.548133929604596,
      "grad_norm": 0.5958225152012777,
      "learning_rate": 0.00017035123228937134,
      "loss": 12.2557,
      "step": 10066
    },
    {
      "epoch": 0.5481883836011789,
      "grad_norm": 0.6090956331991192,
      "learning_rate": 0.0001703449650291534,
      "loss": 12.3631,
      "step": 10067
    },
    {
      "epoch": 0.5482428375977619,
      "grad_norm": 0.6552031333626795,
      "learning_rate": 0.00017033869722192122,
      "loss": 12.4323,
      "step": 10068
    },
    {
      "epoch": 0.5482972915943449,
      "grad_norm": 0.6393619229917588,
      "learning_rate": 0.00017033242886772354,
      "loss": 12.3953,
      "step": 10069
    },
    {
      "epoch": 0.5483517455909279,
      "grad_norm": 0.5917431234863446,
      "learning_rate": 0.00017032615996660905,
      "loss": 12.4018,
      "step": 10070
    },
    {
      "epoch": 0.548406199587511,
      "grad_norm": 0.6115700758668392,
      "learning_rate": 0.0001703198905186266,
      "loss": 12.4389,
      "step": 10071
    },
    {
      "epoch": 0.548460653584094,
      "grad_norm": 0.5654829808206024,
      "learning_rate": 0.0001703136205238249,
      "loss": 12.3738,
      "step": 10072
    },
    {
      "epoch": 0.548515107580677,
      "grad_norm": 0.628121296976426,
      "learning_rate": 0.00017030734998225265,
      "loss": 12.5239,
      "step": 10073
    },
    {
      "epoch": 0.54856956157726,
      "grad_norm": 0.721825207992574,
      "learning_rate": 0.00017030107889395865,
      "loss": 12.4119,
      "step": 10074
    },
    {
      "epoch": 0.548624015573843,
      "grad_norm": 0.5710087081755535,
      "learning_rate": 0.0001702948072589917,
      "loss": 12.3945,
      "step": 10075
    },
    {
      "epoch": 0.548678469570426,
      "grad_norm": 0.5846170878386958,
      "learning_rate": 0.0001702885350774005,
      "loss": 12.3839,
      "step": 10076
    },
    {
      "epoch": 0.5487329235670091,
      "grad_norm": 0.6058575657945217,
      "learning_rate": 0.00017028226234923395,
      "loss": 12.3056,
      "step": 10077
    },
    {
      "epoch": 0.5487873775635921,
      "grad_norm": 0.5312676469096749,
      "learning_rate": 0.00017027598907454067,
      "loss": 12.4184,
      "step": 10078
    },
    {
      "epoch": 0.5488418315601751,
      "grad_norm": 0.6177746184293285,
      "learning_rate": 0.00017026971525336952,
      "loss": 12.4248,
      "step": 10079
    },
    {
      "epoch": 0.5488962855567581,
      "grad_norm": 0.7212320189974394,
      "learning_rate": 0.0001702634408857693,
      "loss": 12.3873,
      "step": 10080
    },
    {
      "epoch": 0.5489507395533411,
      "grad_norm": 0.6057541329016711,
      "learning_rate": 0.00017025716597178877,
      "loss": 12.4775,
      "step": 10081
    },
    {
      "epoch": 0.549005193549924,
      "grad_norm": 0.5892684255084142,
      "learning_rate": 0.00017025089051147675,
      "loss": 12.3358,
      "step": 10082
    },
    {
      "epoch": 0.5490596475465072,
      "grad_norm": 0.5561580135163213,
      "learning_rate": 0.00017024461450488202,
      "loss": 12.3502,
      "step": 10083
    },
    {
      "epoch": 0.5491141015430901,
      "grad_norm": 0.6613960500018959,
      "learning_rate": 0.00017023833795205338,
      "loss": 12.3686,
      "step": 10084
    },
    {
      "epoch": 0.5491685555396731,
      "grad_norm": 0.7155502582204928,
      "learning_rate": 0.00017023206085303965,
      "loss": 12.461,
      "step": 10085
    },
    {
      "epoch": 0.5492230095362561,
      "grad_norm": 0.6151275535683852,
      "learning_rate": 0.00017022578320788963,
      "loss": 12.3676,
      "step": 10086
    },
    {
      "epoch": 0.5492774635328391,
      "grad_norm": 0.6465807638215435,
      "learning_rate": 0.00017021950501665213,
      "loss": 12.3349,
      "step": 10087
    },
    {
      "epoch": 0.5493319175294221,
      "grad_norm": 0.6165783500464499,
      "learning_rate": 0.00017021322627937602,
      "loss": 12.4243,
      "step": 10088
    },
    {
      "epoch": 0.5493863715260052,
      "grad_norm": 0.6234214096780826,
      "learning_rate": 0.00017020694699611006,
      "loss": 12.4481,
      "step": 10089
    },
    {
      "epoch": 0.5494408255225882,
      "grad_norm": 0.6739255625363676,
      "learning_rate": 0.0001702006671669031,
      "loss": 12.4595,
      "step": 10090
    },
    {
      "epoch": 0.5494952795191712,
      "grad_norm": 0.6137959527599097,
      "learning_rate": 0.000170194386791804,
      "loss": 12.4325,
      "step": 10091
    },
    {
      "epoch": 0.5495497335157542,
      "grad_norm": 0.6374066951523512,
      "learning_rate": 0.00017018810587086155,
      "loss": 12.3797,
      "step": 10092
    },
    {
      "epoch": 0.5496041875123372,
      "grad_norm": 0.6517927348980738,
      "learning_rate": 0.00017018182440412468,
      "loss": 12.3723,
      "step": 10093
    },
    {
      "epoch": 0.5496586415089203,
      "grad_norm": 0.6960033350251431,
      "learning_rate": 0.0001701755423916421,
      "loss": 12.429,
      "step": 10094
    },
    {
      "epoch": 0.5497130955055033,
      "grad_norm": 0.6500400376490462,
      "learning_rate": 0.00017016925983346276,
      "loss": 12.4318,
      "step": 10095
    },
    {
      "epoch": 0.5497675495020863,
      "grad_norm": 0.6493288079349868,
      "learning_rate": 0.0001701629767296355,
      "loss": 12.3368,
      "step": 10096
    },
    {
      "epoch": 0.5498220034986693,
      "grad_norm": 0.6466650608087391,
      "learning_rate": 0.00017015669308020917,
      "loss": 12.3259,
      "step": 10097
    },
    {
      "epoch": 0.5498764574952523,
      "grad_norm": 0.620918020415698,
      "learning_rate": 0.00017015040888523263,
      "loss": 12.3392,
      "step": 10098
    },
    {
      "epoch": 0.5499309114918353,
      "grad_norm": 0.603558345767907,
      "learning_rate": 0.00017014412414475473,
      "loss": 12.3562,
      "step": 10099
    },
    {
      "epoch": 0.5499853654884184,
      "grad_norm": 0.688925327145977,
      "learning_rate": 0.00017013783885882434,
      "loss": 12.4577,
      "step": 10100
    },
    {
      "epoch": 0.5500398194850014,
      "grad_norm": 0.6370679160685081,
      "learning_rate": 0.00017013155302749038,
      "loss": 12.3625,
      "step": 10101
    },
    {
      "epoch": 0.5500942734815844,
      "grad_norm": 0.6302329868740851,
      "learning_rate": 0.0001701252666508017,
      "loss": 12.3363,
      "step": 10102
    },
    {
      "epoch": 0.5501487274781673,
      "grad_norm": 0.6859771904180949,
      "learning_rate": 0.0001701189797288072,
      "loss": 12.521,
      "step": 10103
    },
    {
      "epoch": 0.5502031814747503,
      "grad_norm": 0.5725597900784214,
      "learning_rate": 0.00017011269226155574,
      "loss": 12.2956,
      "step": 10104
    },
    {
      "epoch": 0.5502576354713333,
      "grad_norm": 0.6190597219118882,
      "learning_rate": 0.00017010640424909622,
      "loss": 12.4949,
      "step": 10105
    },
    {
      "epoch": 0.5503120894679164,
      "grad_norm": 0.6656002874448648,
      "learning_rate": 0.00017010011569147754,
      "loss": 12.3395,
      "step": 10106
    },
    {
      "epoch": 0.5503665434644994,
      "grad_norm": 0.6106957375460383,
      "learning_rate": 0.0001700938265887486,
      "loss": 12.2956,
      "step": 10107
    },
    {
      "epoch": 0.5504209974610824,
      "grad_norm": 0.6678328208838216,
      "learning_rate": 0.00017008753694095836,
      "loss": 12.2999,
      "step": 10108
    },
    {
      "epoch": 0.5504754514576654,
      "grad_norm": 0.6389549113603158,
      "learning_rate": 0.0001700812467481556,
      "loss": 12.3789,
      "step": 10109
    },
    {
      "epoch": 0.5505299054542484,
      "grad_norm": 0.6767434786700843,
      "learning_rate": 0.00017007495601038938,
      "loss": 12.4813,
      "step": 10110
    },
    {
      "epoch": 0.5505843594508314,
      "grad_norm": 0.6668464446484444,
      "learning_rate": 0.00017006866472770856,
      "loss": 12.3595,
      "step": 10111
    },
    {
      "epoch": 0.5506388134474145,
      "grad_norm": 0.5833039778043949,
      "learning_rate": 0.00017006237290016201,
      "loss": 12.3746,
      "step": 10112
    },
    {
      "epoch": 0.5506932674439975,
      "grad_norm": 0.6248730751987213,
      "learning_rate": 0.00017005608052779868,
      "loss": 12.3588,
      "step": 10113
    },
    {
      "epoch": 0.5507477214405805,
      "grad_norm": 0.6230806633482726,
      "learning_rate": 0.00017004978761066757,
      "loss": 12.4175,
      "step": 10114
    },
    {
      "epoch": 0.5508021754371635,
      "grad_norm": 0.6442464098858275,
      "learning_rate": 0.00017004349414881753,
      "loss": 12.3379,
      "step": 10115
    },
    {
      "epoch": 0.5508566294337465,
      "grad_norm": 0.6266003331634916,
      "learning_rate": 0.00017003720014229754,
      "loss": 12.2762,
      "step": 10116
    },
    {
      "epoch": 0.5509110834303295,
      "grad_norm": 0.552814280799876,
      "learning_rate": 0.00017003090559115656,
      "loss": 12.4019,
      "step": 10117
    },
    {
      "epoch": 0.5509655374269126,
      "grad_norm": 0.5373811486501685,
      "learning_rate": 0.0001700246104954435,
      "loss": 12.3237,
      "step": 10118
    },
    {
      "epoch": 0.5510199914234956,
      "grad_norm": 0.5684336237386249,
      "learning_rate": 0.0001700183148552073,
      "loss": 12.2904,
      "step": 10119
    },
    {
      "epoch": 0.5510744454200786,
      "grad_norm": 0.5888264690377857,
      "learning_rate": 0.00017001201867049696,
      "loss": 12.1247,
      "step": 10120
    },
    {
      "epoch": 0.5511288994166615,
      "grad_norm": 0.5779840814679416,
      "learning_rate": 0.0001700057219413614,
      "loss": 12.3402,
      "step": 10121
    },
    {
      "epoch": 0.5511833534132445,
      "grad_norm": 0.588683708129219,
      "learning_rate": 0.00016999942466784966,
      "loss": 12.4069,
      "step": 10122
    },
    {
      "epoch": 0.5512378074098276,
      "grad_norm": 0.678480509040718,
      "learning_rate": 0.00016999312685001062,
      "loss": 12.2873,
      "step": 10123
    },
    {
      "epoch": 0.5512922614064106,
      "grad_norm": 0.5976064724968893,
      "learning_rate": 0.0001699868284878933,
      "loss": 12.2787,
      "step": 10124
    },
    {
      "epoch": 0.5513467154029936,
      "grad_norm": 0.5916676144071182,
      "learning_rate": 0.00016998052958154666,
      "loss": 12.3731,
      "step": 10125
    },
    {
      "epoch": 0.5514011693995766,
      "grad_norm": 0.6197923034996309,
      "learning_rate": 0.00016997423013101966,
      "loss": 12.39,
      "step": 10126
    },
    {
      "epoch": 0.5514556233961596,
      "grad_norm": 0.6549741234780643,
      "learning_rate": 0.00016996793013636136,
      "loss": 12.4356,
      "step": 10127
    },
    {
      "epoch": 0.5515100773927426,
      "grad_norm": 0.6488966161442786,
      "learning_rate": 0.00016996162959762067,
      "loss": 12.4121,
      "step": 10128
    },
    {
      "epoch": 0.5515645313893257,
      "grad_norm": 0.5858223558061898,
      "learning_rate": 0.00016995532851484663,
      "loss": 12.4635,
      "step": 10129
    },
    {
      "epoch": 0.5516189853859087,
      "grad_norm": 0.6451824416075728,
      "learning_rate": 0.00016994902688808821,
      "loss": 12.5343,
      "step": 10130
    },
    {
      "epoch": 0.5516734393824917,
      "grad_norm": 0.6601972702580284,
      "learning_rate": 0.00016994272471739443,
      "loss": 12.3874,
      "step": 10131
    },
    {
      "epoch": 0.5517278933790747,
      "grad_norm": 0.6182476910168957,
      "learning_rate": 0.00016993642200281432,
      "loss": 12.1093,
      "step": 10132
    },
    {
      "epoch": 0.5517823473756577,
      "grad_norm": 0.6032874164440537,
      "learning_rate": 0.00016993011874439682,
      "loss": 12.5125,
      "step": 10133
    },
    {
      "epoch": 0.5518368013722407,
      "grad_norm": 0.6843307644197107,
      "learning_rate": 0.00016992381494219103,
      "loss": 12.4988,
      "step": 10134
    },
    {
      "epoch": 0.5518912553688238,
      "grad_norm": 0.6136490984502141,
      "learning_rate": 0.0001699175105962459,
      "loss": 12.4561,
      "step": 10135
    },
    {
      "epoch": 0.5519457093654068,
      "grad_norm": 0.5347265528309477,
      "learning_rate": 0.00016991120570661048,
      "loss": 12.365,
      "step": 10136
    },
    {
      "epoch": 0.5520001633619898,
      "grad_norm": 0.8540603107055513,
      "learning_rate": 0.00016990490027333385,
      "loss": 12.4968,
      "step": 10137
    },
    {
      "epoch": 0.5520546173585728,
      "grad_norm": 0.576818131022261,
      "learning_rate": 0.00016989859429646496,
      "loss": 12.3099,
      "step": 10138
    },
    {
      "epoch": 0.5521090713551557,
      "grad_norm": 0.6561478050780211,
      "learning_rate": 0.00016989228777605284,
      "loss": 12.3842,
      "step": 10139
    },
    {
      "epoch": 0.5521635253517387,
      "grad_norm": 0.655042351179101,
      "learning_rate": 0.0001698859807121466,
      "loss": 12.3027,
      "step": 10140
    },
    {
      "epoch": 0.5522179793483218,
      "grad_norm": 0.6798724729283924,
      "learning_rate": 0.00016987967310479527,
      "loss": 12.3973,
      "step": 10141
    },
    {
      "epoch": 0.5522724333449048,
      "grad_norm": 0.6240649516322961,
      "learning_rate": 0.00016987336495404788,
      "loss": 12.2627,
      "step": 10142
    },
    {
      "epoch": 0.5523268873414878,
      "grad_norm": 0.6839594413179446,
      "learning_rate": 0.00016986705625995346,
      "loss": 12.253,
      "step": 10143
    },
    {
      "epoch": 0.5523813413380708,
      "grad_norm": 0.7110332810078518,
      "learning_rate": 0.00016986074702256108,
      "loss": 12.3136,
      "step": 10144
    },
    {
      "epoch": 0.5524357953346538,
      "grad_norm": 0.6476071616334153,
      "learning_rate": 0.00016985443724191988,
      "loss": 12.2806,
      "step": 10145
    },
    {
      "epoch": 0.5524902493312368,
      "grad_norm": 0.6209525263190081,
      "learning_rate": 0.0001698481269180788,
      "loss": 12.3907,
      "step": 10146
    },
    {
      "epoch": 0.5525447033278199,
      "grad_norm": 0.7060291692093761,
      "learning_rate": 0.00016984181605108703,
      "loss": 12.3933,
      "step": 10147
    },
    {
      "epoch": 0.5525991573244029,
      "grad_norm": 0.6365440049787343,
      "learning_rate": 0.00016983550464099353,
      "loss": 12.3701,
      "step": 10148
    },
    {
      "epoch": 0.5526536113209859,
      "grad_norm": 0.6457145627912881,
      "learning_rate": 0.00016982919268784748,
      "loss": 12.3692,
      "step": 10149
    },
    {
      "epoch": 0.5527080653175689,
      "grad_norm": 0.5874194707290666,
      "learning_rate": 0.0001698228801916979,
      "loss": 12.339,
      "step": 10150
    },
    {
      "epoch": 0.5527625193141519,
      "grad_norm": 0.6827235132202331,
      "learning_rate": 0.0001698165671525939,
      "loss": 12.4536,
      "step": 10151
    },
    {
      "epoch": 0.5528169733107349,
      "grad_norm": 0.603995056239589,
      "learning_rate": 0.00016981025357058456,
      "loss": 12.2051,
      "step": 10152
    },
    {
      "epoch": 0.552871427307318,
      "grad_norm": 0.5555255216750008,
      "learning_rate": 0.00016980393944571897,
      "loss": 12.4311,
      "step": 10153
    },
    {
      "epoch": 0.552925881303901,
      "grad_norm": 0.6700662946116988,
      "learning_rate": 0.00016979762477804623,
      "loss": 12.2479,
      "step": 10154
    },
    {
      "epoch": 0.552980335300484,
      "grad_norm": 0.6441931767950304,
      "learning_rate": 0.0001697913095676155,
      "loss": 12.4213,
      "step": 10155
    },
    {
      "epoch": 0.553034789297067,
      "grad_norm": 0.6226291914939351,
      "learning_rate": 0.00016978499381447578,
      "loss": 12.5313,
      "step": 10156
    },
    {
      "epoch": 0.55308924329365,
      "grad_norm": 0.6237855423231016,
      "learning_rate": 0.0001697786775186763,
      "loss": 12.3455,
      "step": 10157
    },
    {
      "epoch": 0.553143697290233,
      "grad_norm": 0.650542570178941,
      "learning_rate": 0.0001697723606802661,
      "loss": 12.3525,
      "step": 10158
    },
    {
      "epoch": 0.553198151286816,
      "grad_norm": 0.650553829713924,
      "learning_rate": 0.00016976604329929434,
      "loss": 12.3732,
      "step": 10159
    },
    {
      "epoch": 0.553252605283399,
      "grad_norm": 0.6459488683075125,
      "learning_rate": 0.00016975972537581008,
      "loss": 12.3562,
      "step": 10160
    },
    {
      "epoch": 0.553307059279982,
      "grad_norm": 0.6128540893647996,
      "learning_rate": 0.00016975340690986252,
      "loss": 12.4002,
      "step": 10161
    },
    {
      "epoch": 0.553361513276565,
      "grad_norm": 0.5602485489371666,
      "learning_rate": 0.0001697470879015008,
      "loss": 12.2805,
      "step": 10162
    },
    {
      "epoch": 0.553415967273148,
      "grad_norm": 0.625146438344186,
      "learning_rate": 0.000169740768350774,
      "loss": 12.349,
      "step": 10163
    },
    {
      "epoch": 0.5534704212697311,
      "grad_norm": 0.6917484957140339,
      "learning_rate": 0.0001697344482577313,
      "loss": 12.3946,
      "step": 10164
    },
    {
      "epoch": 0.5535248752663141,
      "grad_norm": 0.5665765845757563,
      "learning_rate": 0.00016972812762242184,
      "loss": 12.3455,
      "step": 10165
    },
    {
      "epoch": 0.5535793292628971,
      "grad_norm": 0.7588018958176995,
      "learning_rate": 0.00016972180644489476,
      "loss": 12.3631,
      "step": 10166
    },
    {
      "epoch": 0.5536337832594801,
      "grad_norm": 0.6318823657954354,
      "learning_rate": 0.0001697154847251992,
      "loss": 12.3629,
      "step": 10167
    },
    {
      "epoch": 0.5536882372560631,
      "grad_norm": 0.6292295929223287,
      "learning_rate": 0.00016970916246338436,
      "loss": 12.3089,
      "step": 10168
    },
    {
      "epoch": 0.5537426912526461,
      "grad_norm": 0.6476934125782788,
      "learning_rate": 0.00016970283965949938,
      "loss": 12.3641,
      "step": 10169
    },
    {
      "epoch": 0.5537971452492292,
      "grad_norm": 0.5997142351922357,
      "learning_rate": 0.00016969651631359344,
      "loss": 12.3802,
      "step": 10170
    },
    {
      "epoch": 0.5538515992458122,
      "grad_norm": 0.6729543585756947,
      "learning_rate": 0.0001696901924257157,
      "loss": 12.3821,
      "step": 10171
    },
    {
      "epoch": 0.5539060532423952,
      "grad_norm": 0.6407345390836635,
      "learning_rate": 0.0001696838679959153,
      "loss": 12.3207,
      "step": 10172
    },
    {
      "epoch": 0.5539605072389782,
      "grad_norm": 0.5957211650140359,
      "learning_rate": 0.00016967754302424153,
      "loss": 12.3158,
      "step": 10173
    },
    {
      "epoch": 0.5540149612355612,
      "grad_norm": 0.6822451915992077,
      "learning_rate": 0.00016967121751074345,
      "loss": 12.4724,
      "step": 10174
    },
    {
      "epoch": 0.5540694152321441,
      "grad_norm": 0.6309367173751734,
      "learning_rate": 0.0001696648914554703,
      "loss": 12.2831,
      "step": 10175
    },
    {
      "epoch": 0.5541238692287273,
      "grad_norm": 0.6642919979889967,
      "learning_rate": 0.00016965856485847127,
      "loss": 12.4086,
      "step": 10176
    },
    {
      "epoch": 0.5541783232253102,
      "grad_norm": 0.6122523256520281,
      "learning_rate": 0.00016965223771979554,
      "loss": 12.3491,
      "step": 10177
    },
    {
      "epoch": 0.5542327772218932,
      "grad_norm": 0.5613470115381749,
      "learning_rate": 0.0001696459100394924,
      "loss": 12.2773,
      "step": 10178
    },
    {
      "epoch": 0.5542872312184762,
      "grad_norm": 0.667435452999538,
      "learning_rate": 0.0001696395818176109,
      "loss": 12.4032,
      "step": 10179
    },
    {
      "epoch": 0.5543416852150592,
      "grad_norm": 0.5723821218990318,
      "learning_rate": 0.00016963325305420038,
      "loss": 12.1993,
      "step": 10180
    },
    {
      "epoch": 0.5543961392116422,
      "grad_norm": 0.6263000496737242,
      "learning_rate": 0.00016962692374931,
      "loss": 12.4251,
      "step": 10181
    },
    {
      "epoch": 0.5544505932082253,
      "grad_norm": 0.5787263181251557,
      "learning_rate": 0.00016962059390298898,
      "loss": 12.2304,
      "step": 10182
    },
    {
      "epoch": 0.5545050472048083,
      "grad_norm": 0.6270003142827943,
      "learning_rate": 0.00016961426351528656,
      "loss": 12.3031,
      "step": 10183
    },
    {
      "epoch": 0.5545595012013913,
      "grad_norm": 0.654321820603327,
      "learning_rate": 0.00016960793258625193,
      "loss": 12.3821,
      "step": 10184
    },
    {
      "epoch": 0.5546139551979743,
      "grad_norm": 0.5579974955006948,
      "learning_rate": 0.00016960160111593434,
      "loss": 12.2972,
      "step": 10185
    },
    {
      "epoch": 0.5546684091945573,
      "grad_norm": 0.6189528881881449,
      "learning_rate": 0.00016959526910438304,
      "loss": 12.3706,
      "step": 10186
    },
    {
      "epoch": 0.5547228631911403,
      "grad_norm": 0.6198609688250306,
      "learning_rate": 0.00016958893655164725,
      "loss": 12.2323,
      "step": 10187
    },
    {
      "epoch": 0.5547773171877234,
      "grad_norm": 0.6120967966749081,
      "learning_rate": 0.00016958260345777623,
      "loss": 12.3337,
      "step": 10188
    },
    {
      "epoch": 0.5548317711843064,
      "grad_norm": 0.6552890680808466,
      "learning_rate": 0.00016957626982281926,
      "loss": 12.3121,
      "step": 10189
    },
    {
      "epoch": 0.5548862251808894,
      "grad_norm": 0.6545837701604202,
      "learning_rate": 0.00016956993564682548,
      "loss": 12.4834,
      "step": 10190
    },
    {
      "epoch": 0.5549406791774724,
      "grad_norm": 0.5906345025180717,
      "learning_rate": 0.0001695636009298442,
      "loss": 12.3531,
      "step": 10191
    },
    {
      "epoch": 0.5549951331740554,
      "grad_norm": 0.6952886347113761,
      "learning_rate": 0.00016955726567192473,
      "loss": 12.3136,
      "step": 10192
    },
    {
      "epoch": 0.5550495871706385,
      "grad_norm": 0.7026250302498258,
      "learning_rate": 0.0001695509298731163,
      "loss": 12.4446,
      "step": 10193
    },
    {
      "epoch": 0.5551040411672215,
      "grad_norm": 0.5808109078704449,
      "learning_rate": 0.00016954459353346818,
      "loss": 12.408,
      "step": 10194
    },
    {
      "epoch": 0.5551584951638044,
      "grad_norm": 0.5422649522193064,
      "learning_rate": 0.00016953825665302964,
      "loss": 12.4269,
      "step": 10195
    },
    {
      "epoch": 0.5552129491603874,
      "grad_norm": 0.5874935703337653,
      "learning_rate": 0.00016953191923184995,
      "loss": 12.3086,
      "step": 10196
    },
    {
      "epoch": 0.5552674031569704,
      "grad_norm": 0.7222743809111872,
      "learning_rate": 0.00016952558126997837,
      "loss": 12.4497,
      "step": 10197
    },
    {
      "epoch": 0.5553218571535534,
      "grad_norm": 0.61007669020962,
      "learning_rate": 0.00016951924276746425,
      "loss": 12.4958,
      "step": 10198
    },
    {
      "epoch": 0.5553763111501365,
      "grad_norm": 0.6786102403273151,
      "learning_rate": 0.0001695129037243568,
      "loss": 12.3535,
      "step": 10199
    },
    {
      "epoch": 0.5554307651467195,
      "grad_norm": 0.6785427181748215,
      "learning_rate": 0.00016950656414070538,
      "loss": 12.3854,
      "step": 10200
    },
    {
      "epoch": 0.5554852191433025,
      "grad_norm": 0.6132769542909614,
      "learning_rate": 0.00016950022401655926,
      "loss": 12.3423,
      "step": 10201
    },
    {
      "epoch": 0.5555396731398855,
      "grad_norm": 0.6373404879138806,
      "learning_rate": 0.00016949388335196774,
      "loss": 12.3078,
      "step": 10202
    },
    {
      "epoch": 0.5555941271364685,
      "grad_norm": 0.6435052463794952,
      "learning_rate": 0.0001694875421469801,
      "loss": 12.4221,
      "step": 10203
    },
    {
      "epoch": 0.5556485811330515,
      "grad_norm": 0.5621073972746748,
      "learning_rate": 0.00016948120040164572,
      "loss": 12.2795,
      "step": 10204
    },
    {
      "epoch": 0.5557030351296346,
      "grad_norm": 0.7960388659184452,
      "learning_rate": 0.00016947485811601384,
      "loss": 12.5329,
      "step": 10205
    },
    {
      "epoch": 0.5557574891262176,
      "grad_norm": 0.7107737798884084,
      "learning_rate": 0.00016946851529013384,
      "loss": 12.3655,
      "step": 10206
    },
    {
      "epoch": 0.5558119431228006,
      "grad_norm": 0.8029325573958841,
      "learning_rate": 0.00016946217192405501,
      "loss": 12.4019,
      "step": 10207
    },
    {
      "epoch": 0.5558663971193836,
      "grad_norm": 0.7149271685846226,
      "learning_rate": 0.0001694558280178267,
      "loss": 12.5062,
      "step": 10208
    },
    {
      "epoch": 0.5559208511159666,
      "grad_norm": 0.615736615125051,
      "learning_rate": 0.0001694494835714982,
      "loss": 12.2179,
      "step": 10209
    },
    {
      "epoch": 0.5559753051125496,
      "grad_norm": 0.7058656536912001,
      "learning_rate": 0.00016944313858511886,
      "loss": 12.3396,
      "step": 10210
    },
    {
      "epoch": 0.5560297591091327,
      "grad_norm": 0.7412585447567388,
      "learning_rate": 0.00016943679305873803,
      "loss": 12.343,
      "step": 10211
    },
    {
      "epoch": 0.5560842131057157,
      "grad_norm": 0.6208830350613329,
      "learning_rate": 0.00016943044699240507,
      "loss": 12.3917,
      "step": 10212
    },
    {
      "epoch": 0.5561386671022986,
      "grad_norm": 0.7483707455695057,
      "learning_rate": 0.00016942410038616932,
      "loss": 12.3829,
      "step": 10213
    },
    {
      "epoch": 0.5561931210988816,
      "grad_norm": 0.6240932002378885,
      "learning_rate": 0.00016941775324008009,
      "loss": 12.3216,
      "step": 10214
    },
    {
      "epoch": 0.5562475750954646,
      "grad_norm": 0.654778145902865,
      "learning_rate": 0.00016941140555418679,
      "loss": 12.314,
      "step": 10215
    },
    {
      "epoch": 0.5563020290920476,
      "grad_norm": 0.7072083853390658,
      "learning_rate": 0.00016940505732853875,
      "loss": 12.3792,
      "step": 10216
    },
    {
      "epoch": 0.5563564830886307,
      "grad_norm": 0.6776859696620402,
      "learning_rate": 0.00016939870856318533,
      "loss": 12.5402,
      "step": 10217
    },
    {
      "epoch": 0.5564109370852137,
      "grad_norm": 0.7617841105325746,
      "learning_rate": 0.00016939235925817595,
      "loss": 12.5152,
      "step": 10218
    },
    {
      "epoch": 0.5564653910817967,
      "grad_norm": 0.560249323418678,
      "learning_rate": 0.00016938600941355993,
      "loss": 12.3686,
      "step": 10219
    },
    {
      "epoch": 0.5565198450783797,
      "grad_norm": 0.6307457001537511,
      "learning_rate": 0.00016937965902938666,
      "loss": 12.3672,
      "step": 10220
    },
    {
      "epoch": 0.5565742990749627,
      "grad_norm": 0.6787027309352848,
      "learning_rate": 0.00016937330810570553,
      "loss": 12.4397,
      "step": 10221
    },
    {
      "epoch": 0.5566287530715457,
      "grad_norm": 0.5921244360742721,
      "learning_rate": 0.00016936695664256592,
      "loss": 12.3569,
      "step": 10222
    },
    {
      "epoch": 0.5566832070681288,
      "grad_norm": 0.6766229603763056,
      "learning_rate": 0.00016936060464001724,
      "loss": 12.2169,
      "step": 10223
    },
    {
      "epoch": 0.5567376610647118,
      "grad_norm": 0.592326425822779,
      "learning_rate": 0.00016935425209810883,
      "loss": 12.3279,
      "step": 10224
    },
    {
      "epoch": 0.5567921150612948,
      "grad_norm": 0.6994820038562872,
      "learning_rate": 0.00016934789901689018,
      "loss": 12.3537,
      "step": 10225
    },
    {
      "epoch": 0.5568465690578778,
      "grad_norm": 0.6581250385024046,
      "learning_rate": 0.00016934154539641061,
      "loss": 12.1664,
      "step": 10226
    },
    {
      "epoch": 0.5569010230544608,
      "grad_norm": 0.6607535805023356,
      "learning_rate": 0.00016933519123671955,
      "loss": 12.3211,
      "step": 10227
    },
    {
      "epoch": 0.5569554770510439,
      "grad_norm": 0.6100695309818525,
      "learning_rate": 0.00016932883653786643,
      "loss": 12.4688,
      "step": 10228
    },
    {
      "epoch": 0.5570099310476269,
      "grad_norm": 0.6388569267528083,
      "learning_rate": 0.0001693224812999006,
      "loss": 12.3869,
      "step": 10229
    },
    {
      "epoch": 0.5570643850442099,
      "grad_norm": 0.5718089310683809,
      "learning_rate": 0.0001693161255228716,
      "loss": 12.3244,
      "step": 10230
    },
    {
      "epoch": 0.5571188390407928,
      "grad_norm": 0.644449510908533,
      "learning_rate": 0.00016930976920682874,
      "loss": 12.4063,
      "step": 10231
    },
    {
      "epoch": 0.5571732930373758,
      "grad_norm": 0.5847832011670807,
      "learning_rate": 0.0001693034123518215,
      "loss": 12.2008,
      "step": 10232
    },
    {
      "epoch": 0.5572277470339588,
      "grad_norm": 0.6134128450599079,
      "learning_rate": 0.0001692970549578993,
      "loss": 12.2552,
      "step": 10233
    },
    {
      "epoch": 0.5572822010305419,
      "grad_norm": 0.627849807475507,
      "learning_rate": 0.00016929069702511157,
      "loss": 12.488,
      "step": 10234
    },
    {
      "epoch": 0.5573366550271249,
      "grad_norm": 0.6765909156462427,
      "learning_rate": 0.0001692843385535078,
      "loss": 12.4191,
      "step": 10235
    },
    {
      "epoch": 0.5573911090237079,
      "grad_norm": 0.6289812044691674,
      "learning_rate": 0.00016927797954313734,
      "loss": 12.3557,
      "step": 10236
    },
    {
      "epoch": 0.5574455630202909,
      "grad_norm": 0.6431241847241467,
      "learning_rate": 0.00016927161999404975,
      "loss": 12.267,
      "step": 10237
    },
    {
      "epoch": 0.5575000170168739,
      "grad_norm": 0.6980001507327088,
      "learning_rate": 0.00016926525990629442,
      "loss": 12.423,
      "step": 10238
    },
    {
      "epoch": 0.5575544710134569,
      "grad_norm": 0.6329606744797434,
      "learning_rate": 0.00016925889927992075,
      "loss": 12.228,
      "step": 10239
    },
    {
      "epoch": 0.55760892501004,
      "grad_norm": 0.7344798136238765,
      "learning_rate": 0.00016925253811497833,
      "loss": 12.4051,
      "step": 10240
    },
    {
      "epoch": 0.557663379006623,
      "grad_norm": 0.6263600033987291,
      "learning_rate": 0.0001692461764115165,
      "loss": 12.2633,
      "step": 10241
    },
    {
      "epoch": 0.557717833003206,
      "grad_norm": 0.5513225969056479,
      "learning_rate": 0.00016923981416958484,
      "loss": 12.2928,
      "step": 10242
    },
    {
      "epoch": 0.557772286999789,
      "grad_norm": 0.5850290222487247,
      "learning_rate": 0.00016923345138923277,
      "loss": 12.2904,
      "step": 10243
    },
    {
      "epoch": 0.557826740996372,
      "grad_norm": 0.6890330863275668,
      "learning_rate": 0.00016922708807050975,
      "loss": 12.3438,
      "step": 10244
    },
    {
      "epoch": 0.557881194992955,
      "grad_norm": 0.6781093042843492,
      "learning_rate": 0.0001692207242134653,
      "loss": 12.5031,
      "step": 10245
    },
    {
      "epoch": 0.5579356489895381,
      "grad_norm": 0.668948770477723,
      "learning_rate": 0.00016921435981814888,
      "loss": 12.405,
      "step": 10246
    },
    {
      "epoch": 0.5579901029861211,
      "grad_norm": 0.677180028277028,
      "learning_rate": 0.00016920799488461002,
      "loss": 12.4159,
      "step": 10247
    },
    {
      "epoch": 0.558044556982704,
      "grad_norm": 0.6147996023247381,
      "learning_rate": 0.00016920162941289814,
      "loss": 12.4116,
      "step": 10248
    },
    {
      "epoch": 0.558099010979287,
      "grad_norm": 0.6424965673205608,
      "learning_rate": 0.0001691952634030628,
      "loss": 12.4225,
      "step": 10249
    },
    {
      "epoch": 0.55815346497587,
      "grad_norm": 0.5620499261722912,
      "learning_rate": 0.0001691888968551535,
      "loss": 12.2864,
      "step": 10250
    },
    {
      "epoch": 0.558207918972453,
      "grad_norm": 0.6026066590718929,
      "learning_rate": 0.00016918252976921974,
      "loss": 12.2664,
      "step": 10251
    },
    {
      "epoch": 0.5582623729690361,
      "grad_norm": 0.6285570429220287,
      "learning_rate": 0.000169176162145311,
      "loss": 12.395,
      "step": 10252
    },
    {
      "epoch": 0.5583168269656191,
      "grad_norm": 0.5369568931221449,
      "learning_rate": 0.00016916979398347686,
      "loss": 12.2386,
      "step": 10253
    },
    {
      "epoch": 0.5583712809622021,
      "grad_norm": 0.6300205811992972,
      "learning_rate": 0.00016916342528376676,
      "loss": 12.3813,
      "step": 10254
    },
    {
      "epoch": 0.5584257349587851,
      "grad_norm": 0.5810627694012088,
      "learning_rate": 0.00016915705604623029,
      "loss": 12.3711,
      "step": 10255
    },
    {
      "epoch": 0.5584801889553681,
      "grad_norm": 0.6655122858572802,
      "learning_rate": 0.00016915068627091696,
      "loss": 12.2979,
      "step": 10256
    },
    {
      "epoch": 0.5585346429519512,
      "grad_norm": 0.6426723429091911,
      "learning_rate": 0.00016914431595787627,
      "loss": 12.349,
      "step": 10257
    },
    {
      "epoch": 0.5585890969485342,
      "grad_norm": 0.5969986660433423,
      "learning_rate": 0.00016913794510715785,
      "loss": 12.1996,
      "step": 10258
    },
    {
      "epoch": 0.5586435509451172,
      "grad_norm": 0.6376400979426741,
      "learning_rate": 0.0001691315737188111,
      "loss": 12.2946,
      "step": 10259
    },
    {
      "epoch": 0.5586980049417002,
      "grad_norm": 0.6622662726743559,
      "learning_rate": 0.00016912520179288566,
      "loss": 12.4308,
      "step": 10260
    },
    {
      "epoch": 0.5587524589382832,
      "grad_norm": 0.611903090619256,
      "learning_rate": 0.00016911882932943106,
      "loss": 12.3932,
      "step": 10261
    },
    {
      "epoch": 0.5588069129348662,
      "grad_norm": 0.6023841459818996,
      "learning_rate": 0.00016911245632849684,
      "loss": 12.3412,
      "step": 10262
    },
    {
      "epoch": 0.5588613669314493,
      "grad_norm": 0.5844063374490647,
      "learning_rate": 0.0001691060827901326,
      "loss": 12.2498,
      "step": 10263
    },
    {
      "epoch": 0.5589158209280323,
      "grad_norm": 0.599319382728331,
      "learning_rate": 0.00016909970871438788,
      "loss": 12.3185,
      "step": 10264
    },
    {
      "epoch": 0.5589702749246153,
      "grad_norm": 0.5960458771528314,
      "learning_rate": 0.0001690933341013122,
      "loss": 12.4048,
      "step": 10265
    },
    {
      "epoch": 0.5590247289211983,
      "grad_norm": 0.64571759754071,
      "learning_rate": 0.00016908695895095517,
      "loss": 12.4689,
      "step": 10266
    },
    {
      "epoch": 0.5590791829177812,
      "grad_norm": 0.8047723957217814,
      "learning_rate": 0.00016908058326336634,
      "loss": 12.4611,
      "step": 10267
    },
    {
      "epoch": 0.5591336369143642,
      "grad_norm": 0.683586842980497,
      "learning_rate": 0.00016907420703859538,
      "loss": 12.2063,
      "step": 10268
    },
    {
      "epoch": 0.5591880909109473,
      "grad_norm": 0.7147507808213387,
      "learning_rate": 0.00016906783027669176,
      "loss": 12.4311,
      "step": 10269
    },
    {
      "epoch": 0.5592425449075303,
      "grad_norm": 0.5674940897992446,
      "learning_rate": 0.0001690614529777051,
      "loss": 12.4471,
      "step": 10270
    },
    {
      "epoch": 0.5592969989041133,
      "grad_norm": 0.557602325062582,
      "learning_rate": 0.00016905507514168502,
      "loss": 12.4233,
      "step": 10271
    },
    {
      "epoch": 0.5593514529006963,
      "grad_norm": 0.6316895645364624,
      "learning_rate": 0.00016904869676868107,
      "loss": 12.4648,
      "step": 10272
    },
    {
      "epoch": 0.5594059068972793,
      "grad_norm": 0.6515018637341894,
      "learning_rate": 0.0001690423178587429,
      "loss": 12.4008,
      "step": 10273
    },
    {
      "epoch": 0.5594603608938623,
      "grad_norm": 0.5898901503061371,
      "learning_rate": 0.00016903593841192008,
      "loss": 12.3275,
      "step": 10274
    },
    {
      "epoch": 0.5595148148904454,
      "grad_norm": 0.5662931590043602,
      "learning_rate": 0.00016902955842826222,
      "loss": 12.3504,
      "step": 10275
    },
    {
      "epoch": 0.5595692688870284,
      "grad_norm": 0.6778051602663335,
      "learning_rate": 0.00016902317790781895,
      "loss": 12.25,
      "step": 10276
    },
    {
      "epoch": 0.5596237228836114,
      "grad_norm": 0.6092788463144142,
      "learning_rate": 0.00016901679685063986,
      "loss": 12.3418,
      "step": 10277
    },
    {
      "epoch": 0.5596781768801944,
      "grad_norm": 0.618180036489811,
      "learning_rate": 0.0001690104152567746,
      "loss": 12.3607,
      "step": 10278
    },
    {
      "epoch": 0.5597326308767774,
      "grad_norm": 0.5860970509812605,
      "learning_rate": 0.00016900403312627277,
      "loss": 12.3366,
      "step": 10279
    },
    {
      "epoch": 0.5597870848733604,
      "grad_norm": 0.601893208709934,
      "learning_rate": 0.00016899765045918401,
      "loss": 12.458,
      "step": 10280
    },
    {
      "epoch": 0.5598415388699435,
      "grad_norm": 0.5861072547051973,
      "learning_rate": 0.00016899126725555794,
      "loss": 12.4812,
      "step": 10281
    },
    {
      "epoch": 0.5598959928665265,
      "grad_norm": 0.5875897748674739,
      "learning_rate": 0.0001689848835154442,
      "loss": 12.1096,
      "step": 10282
    },
    {
      "epoch": 0.5599504468631095,
      "grad_norm": 0.6160324753910715,
      "learning_rate": 0.00016897849923889246,
      "loss": 12.3096,
      "step": 10283
    },
    {
      "epoch": 0.5600049008596925,
      "grad_norm": 0.6142925610254328,
      "learning_rate": 0.0001689721144259523,
      "loss": 12.3904,
      "step": 10284
    },
    {
      "epoch": 0.5600593548562754,
      "grad_norm": 0.5637627348753075,
      "learning_rate": 0.00016896572907667347,
      "loss": 12.2946,
      "step": 10285
    },
    {
      "epoch": 0.5601138088528584,
      "grad_norm": 0.5518503687523599,
      "learning_rate": 0.00016895934319110555,
      "loss": 12.3553,
      "step": 10286
    },
    {
      "epoch": 0.5601682628494415,
      "grad_norm": 0.6328617298624295,
      "learning_rate": 0.00016895295676929817,
      "loss": 12.4334,
      "step": 10287
    },
    {
      "epoch": 0.5602227168460245,
      "grad_norm": 0.620085531994173,
      "learning_rate": 0.0001689465698113011,
      "loss": 12.3402,
      "step": 10288
    },
    {
      "epoch": 0.5602771708426075,
      "grad_norm": 0.5559505540958086,
      "learning_rate": 0.00016894018231716385,
      "loss": 12.2649,
      "step": 10289
    },
    {
      "epoch": 0.5603316248391905,
      "grad_norm": 0.7008838030079539,
      "learning_rate": 0.00016893379428693626,
      "loss": 12.4094,
      "step": 10290
    },
    {
      "epoch": 0.5603860788357735,
      "grad_norm": 0.605239972915833,
      "learning_rate": 0.0001689274057206679,
      "loss": 12.3578,
      "step": 10291
    },
    {
      "epoch": 0.5604405328323566,
      "grad_norm": 0.6596637595277397,
      "learning_rate": 0.00016892101661840846,
      "loss": 12.2733,
      "step": 10292
    },
    {
      "epoch": 0.5604949868289396,
      "grad_norm": 0.7951158901442011,
      "learning_rate": 0.00016891462698020768,
      "loss": 12.4856,
      "step": 10293
    },
    {
      "epoch": 0.5605494408255226,
      "grad_norm": 0.6574492755937013,
      "learning_rate": 0.00016890823680611517,
      "loss": 12.2087,
      "step": 10294
    },
    {
      "epoch": 0.5606038948221056,
      "grad_norm": 0.653813488593268,
      "learning_rate": 0.00016890184609618064,
      "loss": 12.1862,
      "step": 10295
    },
    {
      "epoch": 0.5606583488186886,
      "grad_norm": 0.7244385612792106,
      "learning_rate": 0.0001688954548504538,
      "loss": 12.3884,
      "step": 10296
    },
    {
      "epoch": 0.5607128028152716,
      "grad_norm": 0.6141729334360738,
      "learning_rate": 0.00016888906306898436,
      "loss": 12.3005,
      "step": 10297
    },
    {
      "epoch": 0.5607672568118547,
      "grad_norm": 0.7246938177635958,
      "learning_rate": 0.00016888267075182206,
      "loss": 12.5213,
      "step": 10298
    },
    {
      "epoch": 0.5608217108084377,
      "grad_norm": 0.7038881657945165,
      "learning_rate": 0.00016887627789901653,
      "loss": 12.5174,
      "step": 10299
    },
    {
      "epoch": 0.5608761648050207,
      "grad_norm": 0.6321702896975392,
      "learning_rate": 0.00016886988451061749,
      "loss": 12.2754,
      "step": 10300
    },
    {
      "epoch": 0.5609306188016037,
      "grad_norm": 0.6103387747847123,
      "learning_rate": 0.0001688634905866747,
      "loss": 12.5386,
      "step": 10301
    },
    {
      "epoch": 0.5609850727981867,
      "grad_norm": 0.732025469317435,
      "learning_rate": 0.00016885709612723783,
      "loss": 12.4871,
      "step": 10302
    },
    {
      "epoch": 0.5610395267947696,
      "grad_norm": 0.6325210977683443,
      "learning_rate": 0.00016885070113235667,
      "loss": 12.4183,
      "step": 10303
    },
    {
      "epoch": 0.5610939807913528,
      "grad_norm": 0.6393350500913199,
      "learning_rate": 0.00016884430560208088,
      "loss": 12.4113,
      "step": 10304
    },
    {
      "epoch": 0.5611484347879357,
      "grad_norm": 0.5909162202743099,
      "learning_rate": 0.00016883790953646025,
      "loss": 12.3757,
      "step": 10305
    },
    {
      "epoch": 0.5612028887845187,
      "grad_norm": 0.7084515897215775,
      "learning_rate": 0.0001688315129355445,
      "loss": 12.3812,
      "step": 10306
    },
    {
      "epoch": 0.5612573427811017,
      "grad_norm": 0.6225796371499448,
      "learning_rate": 0.00016882511579938334,
      "loss": 12.1873,
      "step": 10307
    },
    {
      "epoch": 0.5613117967776847,
      "grad_norm": 0.5897419700844331,
      "learning_rate": 0.00016881871812802652,
      "loss": 12.2328,
      "step": 10308
    },
    {
      "epoch": 0.5613662507742677,
      "grad_norm": 0.626257819909121,
      "learning_rate": 0.00016881231992152385,
      "loss": 12.2687,
      "step": 10309
    },
    {
      "epoch": 0.5614207047708508,
      "grad_norm": 0.6783609424573152,
      "learning_rate": 0.00016880592117992502,
      "loss": 12.4556,
      "step": 10310
    },
    {
      "epoch": 0.5614751587674338,
      "grad_norm": 0.5433012075072357,
      "learning_rate": 0.0001687995219032798,
      "loss": 12.3478,
      "step": 10311
    },
    {
      "epoch": 0.5615296127640168,
      "grad_norm": 0.6673882896234298,
      "learning_rate": 0.00016879312209163797,
      "loss": 12.5079,
      "step": 10312
    },
    {
      "epoch": 0.5615840667605998,
      "grad_norm": 0.636024906690236,
      "learning_rate": 0.00016878672174504926,
      "loss": 12.3692,
      "step": 10313
    },
    {
      "epoch": 0.5616385207571828,
      "grad_norm": 0.5898485469427649,
      "learning_rate": 0.00016878032086356352,
      "loss": 12.4541,
      "step": 10314
    },
    {
      "epoch": 0.5616929747537658,
      "grad_norm": 0.5545109457042696,
      "learning_rate": 0.00016877391944723042,
      "loss": 12.3289,
      "step": 10315
    },
    {
      "epoch": 0.5617474287503489,
      "grad_norm": 0.6048015573465777,
      "learning_rate": 0.00016876751749609982,
      "loss": 12.043,
      "step": 10316
    },
    {
      "epoch": 0.5618018827469319,
      "grad_norm": 0.6191032621169081,
      "learning_rate": 0.00016876111501022147,
      "loss": 12.4121,
      "step": 10317
    },
    {
      "epoch": 0.5618563367435149,
      "grad_norm": 0.6512499912882861,
      "learning_rate": 0.00016875471198964513,
      "loss": 12.4435,
      "step": 10318
    },
    {
      "epoch": 0.5619107907400979,
      "grad_norm": 0.5871615324615288,
      "learning_rate": 0.00016874830843442067,
      "loss": 12.3567,
      "step": 10319
    },
    {
      "epoch": 0.5619652447366809,
      "grad_norm": 0.5919840115805446,
      "learning_rate": 0.00016874190434459777,
      "loss": 12.4784,
      "step": 10320
    },
    {
      "epoch": 0.5620196987332639,
      "grad_norm": 0.5943262467755525,
      "learning_rate": 0.00016873549972022634,
      "loss": 12.2569,
      "step": 10321
    },
    {
      "epoch": 0.562074152729847,
      "grad_norm": 0.6056333912357738,
      "learning_rate": 0.00016872909456135612,
      "loss": 12.2407,
      "step": 10322
    },
    {
      "epoch": 0.56212860672643,
      "grad_norm": 0.6412606462162442,
      "learning_rate": 0.00016872268886803692,
      "loss": 12.4993,
      "step": 10323
    },
    {
      "epoch": 0.5621830607230129,
      "grad_norm": 0.5951322402046102,
      "learning_rate": 0.00016871628264031864,
      "loss": 12.2904,
      "step": 10324
    },
    {
      "epoch": 0.5622375147195959,
      "grad_norm": 0.5957303297392885,
      "learning_rate": 0.00016870987587825094,
      "loss": 12.2501,
      "step": 10325
    },
    {
      "epoch": 0.5622919687161789,
      "grad_norm": 0.6737073150568589,
      "learning_rate": 0.00016870346858188374,
      "loss": 12.3033,
      "step": 10326
    },
    {
      "epoch": 0.562346422712762,
      "grad_norm": 0.5901110815473725,
      "learning_rate": 0.00016869706075126688,
      "loss": 12.3207,
      "step": 10327
    },
    {
      "epoch": 0.562400876709345,
      "grad_norm": 0.6913805237600109,
      "learning_rate": 0.00016869065238645013,
      "loss": 12.2452,
      "step": 10328
    },
    {
      "epoch": 0.562455330705928,
      "grad_norm": 0.597764961757747,
      "learning_rate": 0.00016868424348748335,
      "loss": 12.3216,
      "step": 10329
    },
    {
      "epoch": 0.562509784702511,
      "grad_norm": 0.5970006749445501,
      "learning_rate": 0.00016867783405441638,
      "loss": 12.1318,
      "step": 10330
    },
    {
      "epoch": 0.562564238699094,
      "grad_norm": 0.5930412764100742,
      "learning_rate": 0.00016867142408729904,
      "loss": 12.3634,
      "step": 10331
    },
    {
      "epoch": 0.562618692695677,
      "grad_norm": 0.6212977251363766,
      "learning_rate": 0.0001686650135861812,
      "loss": 12.3131,
      "step": 10332
    },
    {
      "epoch": 0.5626731466922601,
      "grad_norm": 0.5688540545730704,
      "learning_rate": 0.00016865860255111268,
      "loss": 12.3945,
      "step": 10333
    },
    {
      "epoch": 0.5627276006888431,
      "grad_norm": 0.5789013058650888,
      "learning_rate": 0.0001686521909821434,
      "loss": 12.2789,
      "step": 10334
    },
    {
      "epoch": 0.5627820546854261,
      "grad_norm": 0.6335481871931581,
      "learning_rate": 0.00016864577887932313,
      "loss": 12.4302,
      "step": 10335
    },
    {
      "epoch": 0.5628365086820091,
      "grad_norm": 0.5939013797957428,
      "learning_rate": 0.00016863936624270177,
      "loss": 12.3346,
      "step": 10336
    },
    {
      "epoch": 0.5628909626785921,
      "grad_norm": 0.6006353953254291,
      "learning_rate": 0.0001686329530723292,
      "loss": 12.2658,
      "step": 10337
    },
    {
      "epoch": 0.5629454166751751,
      "grad_norm": 0.6047800477910621,
      "learning_rate": 0.0001686265393682553,
      "loss": 12.3286,
      "step": 10338
    },
    {
      "epoch": 0.5629998706717582,
      "grad_norm": 0.6373248187535643,
      "learning_rate": 0.0001686201251305299,
      "loss": 12.3815,
      "step": 10339
    },
    {
      "epoch": 0.5630543246683412,
      "grad_norm": 0.6356603420763361,
      "learning_rate": 0.00016861371035920288,
      "loss": 12.3143,
      "step": 10340
    },
    {
      "epoch": 0.5631087786649241,
      "grad_norm": 0.5582684519094009,
      "learning_rate": 0.0001686072950543242,
      "loss": 12.3037,
      "step": 10341
    },
    {
      "epoch": 0.5631632326615071,
      "grad_norm": 0.5789571181814924,
      "learning_rate": 0.0001686008792159436,
      "loss": 12.317,
      "step": 10342
    },
    {
      "epoch": 0.5632176866580901,
      "grad_norm": 0.6739099289032652,
      "learning_rate": 0.00016859446284411112,
      "loss": 12.3593,
      "step": 10343
    },
    {
      "epoch": 0.5632721406546731,
      "grad_norm": 0.5921743437082129,
      "learning_rate": 0.00016858804593887657,
      "loss": 12.274,
      "step": 10344
    },
    {
      "epoch": 0.5633265946512562,
      "grad_norm": 0.5708854669236476,
      "learning_rate": 0.0001685816285002899,
      "loss": 12.2073,
      "step": 10345
    },
    {
      "epoch": 0.5633810486478392,
      "grad_norm": 0.6308677383448402,
      "learning_rate": 0.00016857521052840096,
      "loss": 12.4676,
      "step": 10346
    },
    {
      "epoch": 0.5634355026444222,
      "grad_norm": 0.6795680317404507,
      "learning_rate": 0.0001685687920232597,
      "loss": 12.4565,
      "step": 10347
    },
    {
      "epoch": 0.5634899566410052,
      "grad_norm": 0.5569416277182532,
      "learning_rate": 0.000168562372984916,
      "loss": 12.347,
      "step": 10348
    },
    {
      "epoch": 0.5635444106375882,
      "grad_norm": 0.562407880823792,
      "learning_rate": 0.00016855595341341977,
      "loss": 12.3733,
      "step": 10349
    },
    {
      "epoch": 0.5635988646341712,
      "grad_norm": 0.5202134266489101,
      "learning_rate": 0.000168549533308821,
      "loss": 12.2996,
      "step": 10350
    },
    {
      "epoch": 0.5636533186307543,
      "grad_norm": 0.6628255911322798,
      "learning_rate": 0.00016854311267116954,
      "loss": 12.4203,
      "step": 10351
    },
    {
      "epoch": 0.5637077726273373,
      "grad_norm": 0.596815972075363,
      "learning_rate": 0.00016853669150051535,
      "loss": 12.3696,
      "step": 10352
    },
    {
      "epoch": 0.5637622266239203,
      "grad_norm": 0.5586849210176513,
      "learning_rate": 0.00016853026979690833,
      "loss": 12.2975,
      "step": 10353
    },
    {
      "epoch": 0.5638166806205033,
      "grad_norm": 0.6823568075738485,
      "learning_rate": 0.00016852384756039848,
      "loss": 12.443,
      "step": 10354
    },
    {
      "epoch": 0.5638711346170863,
      "grad_norm": 0.5933142978854868,
      "learning_rate": 0.00016851742479103563,
      "loss": 12.2016,
      "step": 10355
    },
    {
      "epoch": 0.5639255886136693,
      "grad_norm": 0.597479194997447,
      "learning_rate": 0.00016851100148886986,
      "loss": 12.3549,
      "step": 10356
    },
    {
      "epoch": 0.5639800426102524,
      "grad_norm": 0.6636350663771243,
      "learning_rate": 0.00016850457765395102,
      "loss": 12.4796,
      "step": 10357
    },
    {
      "epoch": 0.5640344966068354,
      "grad_norm": 0.6607380152702863,
      "learning_rate": 0.0001684981532863291,
      "loss": 12.2784,
      "step": 10358
    },
    {
      "epoch": 0.5640889506034183,
      "grad_norm": 0.6877667224482812,
      "learning_rate": 0.00016849172838605408,
      "loss": 12.403,
      "step": 10359
    },
    {
      "epoch": 0.5641434046000013,
      "grad_norm": 0.5659503162516681,
      "learning_rate": 0.00016848530295317585,
      "loss": 12.44,
      "step": 10360
    },
    {
      "epoch": 0.5641978585965843,
      "grad_norm": 0.685369008360853,
      "learning_rate": 0.00016847887698774445,
      "loss": 12.4309,
      "step": 10361
    },
    {
      "epoch": 0.5642523125931674,
      "grad_norm": 0.6320646119966852,
      "learning_rate": 0.0001684724504898098,
      "loss": 12.339,
      "step": 10362
    },
    {
      "epoch": 0.5643067665897504,
      "grad_norm": 0.6212202368714012,
      "learning_rate": 0.00016846602345942191,
      "loss": 12.4303,
      "step": 10363
    },
    {
      "epoch": 0.5643612205863334,
      "grad_norm": 0.6367487206819554,
      "learning_rate": 0.00016845959589663074,
      "loss": 12.421,
      "step": 10364
    },
    {
      "epoch": 0.5644156745829164,
      "grad_norm": 0.681183679084473,
      "learning_rate": 0.00016845316780148627,
      "loss": 12.3895,
      "step": 10365
    },
    {
      "epoch": 0.5644701285794994,
      "grad_norm": 0.6529961642623131,
      "learning_rate": 0.00016844673917403849,
      "loss": 12.4642,
      "step": 10366
    },
    {
      "epoch": 0.5645245825760824,
      "grad_norm": 0.621415406056903,
      "learning_rate": 0.00016844031001433734,
      "loss": 12.2171,
      "step": 10367
    },
    {
      "epoch": 0.5645790365726655,
      "grad_norm": 0.5613170596312487,
      "learning_rate": 0.00016843388032243293,
      "loss": 12.2744,
      "step": 10368
    },
    {
      "epoch": 0.5646334905692485,
      "grad_norm": 0.7287747104215444,
      "learning_rate": 0.00016842745009837515,
      "loss": 12.4659,
      "step": 10369
    },
    {
      "epoch": 0.5646879445658315,
      "grad_norm": 0.544141330334673,
      "learning_rate": 0.00016842101934221406,
      "loss": 12.2132,
      "step": 10370
    },
    {
      "epoch": 0.5647423985624145,
      "grad_norm": 0.6041199910559678,
      "learning_rate": 0.00016841458805399965,
      "loss": 12.3917,
      "step": 10371
    },
    {
      "epoch": 0.5647968525589975,
      "grad_norm": 0.6225477911605025,
      "learning_rate": 0.0001684081562337819,
      "loss": 12.4877,
      "step": 10372
    },
    {
      "epoch": 0.5648513065555805,
      "grad_norm": 0.589768904637465,
      "learning_rate": 0.0001684017238816109,
      "loss": 12.3144,
      "step": 10373
    },
    {
      "epoch": 0.5649057605521636,
      "grad_norm": 0.6221244535619938,
      "learning_rate": 0.0001683952909975366,
      "loss": 12.4835,
      "step": 10374
    },
    {
      "epoch": 0.5649602145487466,
      "grad_norm": 0.6425525815174361,
      "learning_rate": 0.00016838885758160905,
      "loss": 12.4115,
      "step": 10375
    },
    {
      "epoch": 0.5650146685453296,
      "grad_norm": 0.6606842162669248,
      "learning_rate": 0.0001683824236338783,
      "loss": 12.3993,
      "step": 10376
    },
    {
      "epoch": 0.5650691225419125,
      "grad_norm": 0.6084374603777637,
      "learning_rate": 0.0001683759891543943,
      "loss": 12.3677,
      "step": 10377
    },
    {
      "epoch": 0.5651235765384955,
      "grad_norm": 0.6991656978018909,
      "learning_rate": 0.00016836955414320715,
      "loss": 12.3793,
      "step": 10378
    },
    {
      "epoch": 0.5651780305350785,
      "grad_norm": 0.6138426062310326,
      "learning_rate": 0.00016836311860036693,
      "loss": 12.3313,
      "step": 10379
    },
    {
      "epoch": 0.5652324845316616,
      "grad_norm": 0.6275506547891637,
      "learning_rate": 0.0001683566825259236,
      "loss": 12.2722,
      "step": 10380
    },
    {
      "epoch": 0.5652869385282446,
      "grad_norm": 0.9161209021743207,
      "learning_rate": 0.00016835024591992724,
      "loss": 12.3421,
      "step": 10381
    },
    {
      "epoch": 0.5653413925248276,
      "grad_norm": 0.6321254761895913,
      "learning_rate": 0.00016834380878242794,
      "loss": 12.2944,
      "step": 10382
    },
    {
      "epoch": 0.5653958465214106,
      "grad_norm": 0.6345049944779382,
      "learning_rate": 0.00016833737111347568,
      "loss": 12.484,
      "step": 10383
    },
    {
      "epoch": 0.5654503005179936,
      "grad_norm": 0.623691170627115,
      "learning_rate": 0.0001683309329131206,
      "loss": 12.3452,
      "step": 10384
    },
    {
      "epoch": 0.5655047545145766,
      "grad_norm": 0.6169359428690868,
      "learning_rate": 0.0001683244941814127,
      "loss": 12.2734,
      "step": 10385
    },
    {
      "epoch": 0.5655592085111597,
      "grad_norm": 0.5911971693461668,
      "learning_rate": 0.00016831805491840208,
      "loss": 12.4244,
      "step": 10386
    },
    {
      "epoch": 0.5656136625077427,
      "grad_norm": 0.6138794288462133,
      "learning_rate": 0.0001683116151241388,
      "loss": 12.3843,
      "step": 10387
    },
    {
      "epoch": 0.5656681165043257,
      "grad_norm": 0.6087658301053654,
      "learning_rate": 0.00016830517479867298,
      "loss": 12.2409,
      "step": 10388
    },
    {
      "epoch": 0.5657225705009087,
      "grad_norm": 0.5051577573517587,
      "learning_rate": 0.00016829873394205464,
      "loss": 12.2384,
      "step": 10389
    },
    {
      "epoch": 0.5657770244974917,
      "grad_norm": 0.5880845343996741,
      "learning_rate": 0.0001682922925543339,
      "loss": 12.3321,
      "step": 10390
    },
    {
      "epoch": 0.5658314784940748,
      "grad_norm": 0.5928845999091579,
      "learning_rate": 0.00016828585063556083,
      "loss": 12.2001,
      "step": 10391
    },
    {
      "epoch": 0.5658859324906578,
      "grad_norm": 0.6827691259284938,
      "learning_rate": 0.0001682794081857855,
      "loss": 12.5109,
      "step": 10392
    },
    {
      "epoch": 0.5659403864872408,
      "grad_norm": 0.5642687647023201,
      "learning_rate": 0.0001682729652050581,
      "loss": 12.3254,
      "step": 10393
    },
    {
      "epoch": 0.5659948404838238,
      "grad_norm": 0.6148651923790508,
      "learning_rate": 0.00016826652169342867,
      "loss": 12.3495,
      "step": 10394
    },
    {
      "epoch": 0.5660492944804068,
      "grad_norm": 0.6452765348559973,
      "learning_rate": 0.00016826007765094732,
      "loss": 12.3754,
      "step": 10395
    },
    {
      "epoch": 0.5661037484769897,
      "grad_norm": 0.5965890572179519,
      "learning_rate": 0.00016825363307766412,
      "loss": 12.418,
      "step": 10396
    },
    {
      "epoch": 0.5661582024735728,
      "grad_norm": 0.5471346501778647,
      "learning_rate": 0.00016824718797362923,
      "loss": 12.3318,
      "step": 10397
    },
    {
      "epoch": 0.5662126564701558,
      "grad_norm": 0.6751846501375386,
      "learning_rate": 0.00016824074233889278,
      "loss": 12.5202,
      "step": 10398
    },
    {
      "epoch": 0.5662671104667388,
      "grad_norm": 0.6218888957514784,
      "learning_rate": 0.00016823429617350487,
      "loss": 12.276,
      "step": 10399
    },
    {
      "epoch": 0.5663215644633218,
      "grad_norm": 0.6492663612660368,
      "learning_rate": 0.00016822784947751563,
      "loss": 12.391,
      "step": 10400
    },
    {
      "epoch": 0.5663760184599048,
      "grad_norm": 0.6232618935944115,
      "learning_rate": 0.0001682214022509752,
      "loss": 12.3163,
      "step": 10401
    },
    {
      "epoch": 0.5664304724564878,
      "grad_norm": 0.649271965491751,
      "learning_rate": 0.00016821495449393368,
      "loss": 12.4013,
      "step": 10402
    },
    {
      "epoch": 0.5664849264530709,
      "grad_norm": 0.66748998807262,
      "learning_rate": 0.00016820850620644125,
      "loss": 12.2451,
      "step": 10403
    },
    {
      "epoch": 0.5665393804496539,
      "grad_norm": 0.7082320217361083,
      "learning_rate": 0.00016820205738854804,
      "loss": 12.2669,
      "step": 10404
    },
    {
      "epoch": 0.5665938344462369,
      "grad_norm": 0.6083087122633456,
      "learning_rate": 0.0001681956080403042,
      "loss": 12.344,
      "step": 10405
    },
    {
      "epoch": 0.5666482884428199,
      "grad_norm": 0.5845446637283482,
      "learning_rate": 0.00016818915816175985,
      "loss": 12.3627,
      "step": 10406
    },
    {
      "epoch": 0.5667027424394029,
      "grad_norm": 0.6304622759286626,
      "learning_rate": 0.00016818270775296519,
      "loss": 12.3687,
      "step": 10407
    },
    {
      "epoch": 0.5667571964359859,
      "grad_norm": 0.5473269293115471,
      "learning_rate": 0.00016817625681397034,
      "loss": 12.213,
      "step": 10408
    },
    {
      "epoch": 0.566811650432569,
      "grad_norm": 0.6984369382628332,
      "learning_rate": 0.00016816980534482552,
      "loss": 12.2926,
      "step": 10409
    },
    {
      "epoch": 0.566866104429152,
      "grad_norm": 0.6662395723038156,
      "learning_rate": 0.00016816335334558083,
      "loss": 12.4141,
      "step": 10410
    },
    {
      "epoch": 0.566920558425735,
      "grad_norm": 0.6210259725286218,
      "learning_rate": 0.0001681569008162865,
      "loss": 12.3193,
      "step": 10411
    },
    {
      "epoch": 0.566975012422318,
      "grad_norm": 0.5429922620912881,
      "learning_rate": 0.00016815044775699266,
      "loss": 12.2855,
      "step": 10412
    },
    {
      "epoch": 0.567029466418901,
      "grad_norm": 0.5977332368552712,
      "learning_rate": 0.0001681439941677495,
      "loss": 12.2268,
      "step": 10413
    },
    {
      "epoch": 0.5670839204154839,
      "grad_norm": 0.6159050614116562,
      "learning_rate": 0.00016813754004860724,
      "loss": 12.1776,
      "step": 10414
    },
    {
      "epoch": 0.567138374412067,
      "grad_norm": 0.582597585497368,
      "learning_rate": 0.00016813108539961603,
      "loss": 12.3413,
      "step": 10415
    },
    {
      "epoch": 0.56719282840865,
      "grad_norm": 0.6513474497024757,
      "learning_rate": 0.00016812463022082607,
      "loss": 12.394,
      "step": 10416
    },
    {
      "epoch": 0.567247282405233,
      "grad_norm": 0.5895717171760396,
      "learning_rate": 0.00016811817451228757,
      "loss": 12.3976,
      "step": 10417
    },
    {
      "epoch": 0.567301736401816,
      "grad_norm": 0.6145716743427717,
      "learning_rate": 0.00016811171827405073,
      "loss": 12.4249,
      "step": 10418
    },
    {
      "epoch": 0.567356190398399,
      "grad_norm": 0.6429917849423017,
      "learning_rate": 0.00016810526150616572,
      "loss": 12.3861,
      "step": 10419
    },
    {
      "epoch": 0.567410644394982,
      "grad_norm": 0.5341869197548992,
      "learning_rate": 0.0001680988042086828,
      "loss": 12.2854,
      "step": 10420
    },
    {
      "epoch": 0.5674650983915651,
      "grad_norm": 0.6370657440114558,
      "learning_rate": 0.00016809234638165212,
      "loss": 12.3632,
      "step": 10421
    },
    {
      "epoch": 0.5675195523881481,
      "grad_norm": 0.6345636872655567,
      "learning_rate": 0.00016808588802512398,
      "loss": 12.3265,
      "step": 10422
    },
    {
      "epoch": 0.5675740063847311,
      "grad_norm": 0.5308037534905589,
      "learning_rate": 0.00016807942913914855,
      "loss": 12.2471,
      "step": 10423
    },
    {
      "epoch": 0.5676284603813141,
      "grad_norm": 0.7217080141177963,
      "learning_rate": 0.00016807296972377604,
      "loss": 12.3981,
      "step": 10424
    },
    {
      "epoch": 0.5676829143778971,
      "grad_norm": 0.6042083497364779,
      "learning_rate": 0.00016806650977905672,
      "loss": 12.2527,
      "step": 10425
    },
    {
      "epoch": 0.5677373683744802,
      "grad_norm": 0.5745092248979259,
      "learning_rate": 0.00016806004930504078,
      "loss": 12.1766,
      "step": 10426
    },
    {
      "epoch": 0.5677918223710632,
      "grad_norm": 0.5705259495585121,
      "learning_rate": 0.0001680535883017785,
      "loss": 12.2859,
      "step": 10427
    },
    {
      "epoch": 0.5678462763676462,
      "grad_norm": 0.5417193186976931,
      "learning_rate": 0.0001680471267693201,
      "loss": 12.4,
      "step": 10428
    },
    {
      "epoch": 0.5679007303642292,
      "grad_norm": 0.5834722615703474,
      "learning_rate": 0.00016804066470771584,
      "loss": 12.3393,
      "step": 10429
    },
    {
      "epoch": 0.5679551843608122,
      "grad_norm": 0.7036143280210932,
      "learning_rate": 0.00016803420211701598,
      "loss": 12.5283,
      "step": 10430
    },
    {
      "epoch": 0.5680096383573952,
      "grad_norm": 0.5979312547772383,
      "learning_rate": 0.00016802773899727072,
      "loss": 12.4094,
      "step": 10431
    },
    {
      "epoch": 0.5680640923539783,
      "grad_norm": 0.6090176540758172,
      "learning_rate": 0.00016802127534853035,
      "loss": 12.285,
      "step": 10432
    },
    {
      "epoch": 0.5681185463505612,
      "grad_norm": 0.6339843877396358,
      "learning_rate": 0.00016801481117084514,
      "loss": 12.3795,
      "step": 10433
    },
    {
      "epoch": 0.5681730003471442,
      "grad_norm": 0.6647505095307242,
      "learning_rate": 0.0001680083464642654,
      "loss": 12.4322,
      "step": 10434
    },
    {
      "epoch": 0.5682274543437272,
      "grad_norm": 0.5818159565641814,
      "learning_rate": 0.00016800188122884132,
      "loss": 12.4275,
      "step": 10435
    },
    {
      "epoch": 0.5682819083403102,
      "grad_norm": 0.6228052874764259,
      "learning_rate": 0.00016799541546462317,
      "loss": 12.3025,
      "step": 10436
    },
    {
      "epoch": 0.5683363623368932,
      "grad_norm": 0.6335954949628722,
      "learning_rate": 0.0001679889491716613,
      "loss": 12.4529,
      "step": 10437
    },
    {
      "epoch": 0.5683908163334763,
      "grad_norm": 0.6038600388041353,
      "learning_rate": 0.00016798248235000597,
      "loss": 12.3209,
      "step": 10438
    },
    {
      "epoch": 0.5684452703300593,
      "grad_norm": 0.6117645541689094,
      "learning_rate": 0.00016797601499970744,
      "loss": 12.3349,
      "step": 10439
    },
    {
      "epoch": 0.5684997243266423,
      "grad_norm": 0.7230541776819218,
      "learning_rate": 0.00016796954712081604,
      "loss": 12.4097,
      "step": 10440
    },
    {
      "epoch": 0.5685541783232253,
      "grad_norm": 0.6428774400716817,
      "learning_rate": 0.000167963078713382,
      "loss": 12.394,
      "step": 10441
    },
    {
      "epoch": 0.5686086323198083,
      "grad_norm": 0.6505580888411074,
      "learning_rate": 0.00016795660977745572,
      "loss": 12.1937,
      "step": 10442
    },
    {
      "epoch": 0.5686630863163913,
      "grad_norm": 0.6437749858847259,
      "learning_rate": 0.0001679501403130874,
      "loss": 12.2522,
      "step": 10443
    },
    {
      "epoch": 0.5687175403129744,
      "grad_norm": 0.5941359930214059,
      "learning_rate": 0.00016794367032032742,
      "loss": 12.3814,
      "step": 10444
    },
    {
      "epoch": 0.5687719943095574,
      "grad_norm": 0.6410446086869066,
      "learning_rate": 0.00016793719979922605,
      "loss": 12.4891,
      "step": 10445
    },
    {
      "epoch": 0.5688264483061404,
      "grad_norm": 0.5759340350528791,
      "learning_rate": 0.00016793072874983362,
      "loss": 12.3136,
      "step": 10446
    },
    {
      "epoch": 0.5688809023027234,
      "grad_norm": 0.6219826681075947,
      "learning_rate": 0.00016792425717220046,
      "loss": 12.4172,
      "step": 10447
    },
    {
      "epoch": 0.5689353562993064,
      "grad_norm": 0.5902531241211377,
      "learning_rate": 0.00016791778506637688,
      "loss": 12.3866,
      "step": 10448
    },
    {
      "epoch": 0.5689898102958894,
      "grad_norm": 0.5902582942414536,
      "learning_rate": 0.0001679113124324132,
      "loss": 12.3021,
      "step": 10449
    },
    {
      "epoch": 0.5690442642924725,
      "grad_norm": 0.5751312773984891,
      "learning_rate": 0.0001679048392703598,
      "loss": 12.3701,
      "step": 10450
    },
    {
      "epoch": 0.5690987182890554,
      "grad_norm": 0.6790559083330199,
      "learning_rate": 0.00016789836558026697,
      "loss": 12.3923,
      "step": 10451
    },
    {
      "epoch": 0.5691531722856384,
      "grad_norm": 0.6452000054378788,
      "learning_rate": 0.00016789189136218505,
      "loss": 12.3022,
      "step": 10452
    },
    {
      "epoch": 0.5692076262822214,
      "grad_norm": 0.5691936565874609,
      "learning_rate": 0.0001678854166161644,
      "loss": 12.3396,
      "step": 10453
    },
    {
      "epoch": 0.5692620802788044,
      "grad_norm": 0.5693653034216821,
      "learning_rate": 0.00016787894134225536,
      "loss": 12.386,
      "step": 10454
    },
    {
      "epoch": 0.5693165342753874,
      "grad_norm": 0.6508692340480876,
      "learning_rate": 0.0001678724655405083,
      "loss": 12.2845,
      "step": 10455
    },
    {
      "epoch": 0.5693709882719705,
      "grad_norm": 0.6186699806695385,
      "learning_rate": 0.00016786598921097358,
      "loss": 12.3824,
      "step": 10456
    },
    {
      "epoch": 0.5694254422685535,
      "grad_norm": 0.7158082493266653,
      "learning_rate": 0.00016785951235370153,
      "loss": 12.4277,
      "step": 10457
    },
    {
      "epoch": 0.5694798962651365,
      "grad_norm": 0.656346164357275,
      "learning_rate": 0.0001678530349687425,
      "loss": 12.3382,
      "step": 10458
    },
    {
      "epoch": 0.5695343502617195,
      "grad_norm": 0.7203813696149984,
      "learning_rate": 0.00016784655705614693,
      "loss": 12.3475,
      "step": 10459
    },
    {
      "epoch": 0.5695888042583025,
      "grad_norm": 0.6609864561370712,
      "learning_rate": 0.00016784007861596518,
      "loss": 12.4706,
      "step": 10460
    },
    {
      "epoch": 0.5696432582548856,
      "grad_norm": 0.6048227744145742,
      "learning_rate": 0.00016783359964824755,
      "loss": 12.3333,
      "step": 10461
    },
    {
      "epoch": 0.5696977122514686,
      "grad_norm": 0.7298358242850399,
      "learning_rate": 0.0001678271201530445,
      "loss": 12.3152,
      "step": 10462
    },
    {
      "epoch": 0.5697521662480516,
      "grad_norm": 0.6721553190851784,
      "learning_rate": 0.00016782064013040637,
      "loss": 12.3147,
      "step": 10463
    },
    {
      "epoch": 0.5698066202446346,
      "grad_norm": 0.6208164219046698,
      "learning_rate": 0.0001678141595803836,
      "loss": 12.3959,
      "step": 10464
    },
    {
      "epoch": 0.5698610742412176,
      "grad_norm": 0.7267473495148199,
      "learning_rate": 0.00016780767850302654,
      "loss": 12.2043,
      "step": 10465
    },
    {
      "epoch": 0.5699155282378006,
      "grad_norm": 0.5815895018152115,
      "learning_rate": 0.0001678011968983856,
      "loss": 12.3113,
      "step": 10466
    },
    {
      "epoch": 0.5699699822343837,
      "grad_norm": 0.6407164249371863,
      "learning_rate": 0.0001677947147665112,
      "loss": 12.335,
      "step": 10467
    },
    {
      "epoch": 0.5700244362309667,
      "grad_norm": 0.7096471186991803,
      "learning_rate": 0.00016778823210745366,
      "loss": 12.314,
      "step": 10468
    },
    {
      "epoch": 0.5700788902275497,
      "grad_norm": 0.6327973720906753,
      "learning_rate": 0.0001677817489212635,
      "loss": 12.3507,
      "step": 10469
    },
    {
      "epoch": 0.5701333442241326,
      "grad_norm": 0.5987693162255612,
      "learning_rate": 0.00016777526520799115,
      "loss": 12.2983,
      "step": 10470
    },
    {
      "epoch": 0.5701877982207156,
      "grad_norm": 0.6552558142622182,
      "learning_rate": 0.0001677687809676869,
      "loss": 12.3778,
      "step": 10471
    },
    {
      "epoch": 0.5702422522172986,
      "grad_norm": 0.5967141861925086,
      "learning_rate": 0.00016776229620040124,
      "loss": 12.3893,
      "step": 10472
    },
    {
      "epoch": 0.5702967062138817,
      "grad_norm": 0.6176664295765997,
      "learning_rate": 0.00016775581090618463,
      "loss": 12.4305,
      "step": 10473
    },
    {
      "epoch": 0.5703511602104647,
      "grad_norm": 0.6591993654093905,
      "learning_rate": 0.00016774932508508748,
      "loss": 12.3443,
      "step": 10474
    },
    {
      "epoch": 0.5704056142070477,
      "grad_norm": 0.584863376348845,
      "learning_rate": 0.0001677428387371602,
      "loss": 12.3306,
      "step": 10475
    },
    {
      "epoch": 0.5704600682036307,
      "grad_norm": 0.5963330913592126,
      "learning_rate": 0.00016773635186245324,
      "loss": 12.4846,
      "step": 10476
    },
    {
      "epoch": 0.5705145222002137,
      "grad_norm": 0.6438054090971593,
      "learning_rate": 0.00016772986446101707,
      "loss": 12.3858,
      "step": 10477
    },
    {
      "epoch": 0.5705689761967967,
      "grad_norm": 0.6064314578781214,
      "learning_rate": 0.0001677233765329021,
      "loss": 12.3667,
      "step": 10478
    },
    {
      "epoch": 0.5706234301933798,
      "grad_norm": 0.5348696070350466,
      "learning_rate": 0.00016771688807815883,
      "loss": 12.3922,
      "step": 10479
    },
    {
      "epoch": 0.5706778841899628,
      "grad_norm": 0.6136181266951442,
      "learning_rate": 0.00016771039909683767,
      "loss": 12.3923,
      "step": 10480
    },
    {
      "epoch": 0.5707323381865458,
      "grad_norm": 0.6356924851395139,
      "learning_rate": 0.00016770390958898904,
      "loss": 12.4682,
      "step": 10481
    },
    {
      "epoch": 0.5707867921831288,
      "grad_norm": 0.5611718815605147,
      "learning_rate": 0.0001676974195546635,
      "loss": 12.4042,
      "step": 10482
    },
    {
      "epoch": 0.5708412461797118,
      "grad_norm": 0.5599668471880284,
      "learning_rate": 0.00016769092899391146,
      "loss": 12.3262,
      "step": 10483
    },
    {
      "epoch": 0.5708957001762948,
      "grad_norm": 0.6186802999347444,
      "learning_rate": 0.0001676844379067834,
      "loss": 12.3928,
      "step": 10484
    },
    {
      "epoch": 0.5709501541728779,
      "grad_norm": 0.5786533893527239,
      "learning_rate": 0.00016767794629332983,
      "loss": 12.331,
      "step": 10485
    },
    {
      "epoch": 0.5710046081694609,
      "grad_norm": 0.5934158335680023,
      "learning_rate": 0.00016767145415360116,
      "loss": 12.1695,
      "step": 10486
    },
    {
      "epoch": 0.5710590621660439,
      "grad_norm": 0.6016145692911932,
      "learning_rate": 0.00016766496148764792,
      "loss": 12.2989,
      "step": 10487
    },
    {
      "epoch": 0.5711135161626268,
      "grad_norm": 0.6239976307642991,
      "learning_rate": 0.0001676584682955206,
      "loss": 12.3911,
      "step": 10488
    },
    {
      "epoch": 0.5711679701592098,
      "grad_norm": 0.6082768551019266,
      "learning_rate": 0.0001676519745772697,
      "loss": 12.0827,
      "step": 10489
    },
    {
      "epoch": 0.5712224241557928,
      "grad_norm": 0.5871346045320904,
      "learning_rate": 0.00016764548033294568,
      "loss": 12.4464,
      "step": 10490
    },
    {
      "epoch": 0.5712768781523759,
      "grad_norm": 0.613933828774959,
      "learning_rate": 0.00016763898556259907,
      "loss": 12.3568,
      "step": 10491
    },
    {
      "epoch": 0.5713313321489589,
      "grad_norm": 0.5991384426015036,
      "learning_rate": 0.00016763249026628037,
      "loss": 12.4625,
      "step": 10492
    },
    {
      "epoch": 0.5713857861455419,
      "grad_norm": 0.5514872373075452,
      "learning_rate": 0.00016762599444404005,
      "loss": 12.383,
      "step": 10493
    },
    {
      "epoch": 0.5714402401421249,
      "grad_norm": 0.5531371468233106,
      "learning_rate": 0.0001676194980959287,
      "loss": 12.3299,
      "step": 10494
    },
    {
      "epoch": 0.5714946941387079,
      "grad_norm": 0.5941667125298707,
      "learning_rate": 0.00016761300122199678,
      "loss": 12.2687,
      "step": 10495
    },
    {
      "epoch": 0.571549148135291,
      "grad_norm": 0.574816716156372,
      "learning_rate": 0.00016760650382229483,
      "loss": 12.1507,
      "step": 10496
    },
    {
      "epoch": 0.571603602131874,
      "grad_norm": 0.5661023144426395,
      "learning_rate": 0.00016760000589687336,
      "loss": 12.3588,
      "step": 10497
    },
    {
      "epoch": 0.571658056128457,
      "grad_norm": 0.5950806459266346,
      "learning_rate": 0.00016759350744578288,
      "loss": 12.3072,
      "step": 10498
    },
    {
      "epoch": 0.57171251012504,
      "grad_norm": 0.5788836627410617,
      "learning_rate": 0.00016758700846907401,
      "loss": 12.3097,
      "step": 10499
    },
    {
      "epoch": 0.571766964121623,
      "grad_norm": 0.5946412198769391,
      "learning_rate": 0.00016758050896679717,
      "loss": 12.3366,
      "step": 10500
    },
    {
      "epoch": 0.571821418118206,
      "grad_norm": 0.6167651308416774,
      "learning_rate": 0.000167574008939003,
      "loss": 12.3193,
      "step": 10501
    },
    {
      "epoch": 0.5718758721147891,
      "grad_norm": 0.6110161549280384,
      "learning_rate": 0.00016756750838574198,
      "loss": 12.1554,
      "step": 10502
    },
    {
      "epoch": 0.5719303261113721,
      "grad_norm": 0.596606633093422,
      "learning_rate": 0.00016756100730706467,
      "loss": 12.4674,
      "step": 10503
    },
    {
      "epoch": 0.5719847801079551,
      "grad_norm": 0.626567890368138,
      "learning_rate": 0.00016755450570302166,
      "loss": 12.3227,
      "step": 10504
    },
    {
      "epoch": 0.572039234104538,
      "grad_norm": 0.57946182580484,
      "learning_rate": 0.0001675480035736635,
      "loss": 12.2348,
      "step": 10505
    },
    {
      "epoch": 0.572093688101121,
      "grad_norm": 0.5888367222286589,
      "learning_rate": 0.00016754150091904073,
      "loss": 12.3907,
      "step": 10506
    },
    {
      "epoch": 0.572148142097704,
      "grad_norm": 0.5917082862865675,
      "learning_rate": 0.00016753499773920392,
      "loss": 12.4763,
      "step": 10507
    },
    {
      "epoch": 0.5722025960942871,
      "grad_norm": 0.5822949981872415,
      "learning_rate": 0.00016752849403420364,
      "loss": 12.4209,
      "step": 10508
    },
    {
      "epoch": 0.5722570500908701,
      "grad_norm": 0.6384365923747648,
      "learning_rate": 0.00016752198980409045,
      "loss": 12.578,
      "step": 10509
    },
    {
      "epoch": 0.5723115040874531,
      "grad_norm": 0.563623509323039,
      "learning_rate": 0.00016751548504891496,
      "loss": 12.2381,
      "step": 10510
    },
    {
      "epoch": 0.5723659580840361,
      "grad_norm": 0.5695499848134334,
      "learning_rate": 0.00016750897976872773,
      "loss": 12.3816,
      "step": 10511
    },
    {
      "epoch": 0.5724204120806191,
      "grad_norm": 0.5884756822216776,
      "learning_rate": 0.00016750247396357936,
      "loss": 12.4354,
      "step": 10512
    },
    {
      "epoch": 0.5724748660772021,
      "grad_norm": 0.6719139096999517,
      "learning_rate": 0.0001674959676335204,
      "loss": 12.3141,
      "step": 10513
    },
    {
      "epoch": 0.5725293200737852,
      "grad_norm": 0.6136542932337656,
      "learning_rate": 0.0001674894607786015,
      "loss": 12.1397,
      "step": 10514
    },
    {
      "epoch": 0.5725837740703682,
      "grad_norm": 0.6274914061291436,
      "learning_rate": 0.00016748295339887327,
      "loss": 12.3897,
      "step": 10515
    },
    {
      "epoch": 0.5726382280669512,
      "grad_norm": 0.7877708430325404,
      "learning_rate": 0.00016747644549438623,
      "loss": 12.4755,
      "step": 10516
    },
    {
      "epoch": 0.5726926820635342,
      "grad_norm": 0.6202086631901177,
      "learning_rate": 0.00016746993706519105,
      "loss": 12.4147,
      "step": 10517
    },
    {
      "epoch": 0.5727471360601172,
      "grad_norm": 0.5956267922455494,
      "learning_rate": 0.0001674634281113383,
      "loss": 12.3638,
      "step": 10518
    },
    {
      "epoch": 0.5728015900567002,
      "grad_norm": 0.6043432463810966,
      "learning_rate": 0.00016745691863287866,
      "loss": 12.3319,
      "step": 10519
    },
    {
      "epoch": 0.5728560440532833,
      "grad_norm": 0.6873759483732338,
      "learning_rate": 0.0001674504086298627,
      "loss": 12.5279,
      "step": 10520
    },
    {
      "epoch": 0.5729104980498663,
      "grad_norm": 0.6071589858995201,
      "learning_rate": 0.00016744389810234103,
      "loss": 12.1313,
      "step": 10521
    },
    {
      "epoch": 0.5729649520464493,
      "grad_norm": 0.6503637091496456,
      "learning_rate": 0.00016743738705036432,
      "loss": 12.3438,
      "step": 10522
    },
    {
      "epoch": 0.5730194060430323,
      "grad_norm": 0.6086095458823582,
      "learning_rate": 0.00016743087547398315,
      "loss": 12.3626,
      "step": 10523
    },
    {
      "epoch": 0.5730738600396152,
      "grad_norm": 0.5756750286388267,
      "learning_rate": 0.00016742436337324821,
      "loss": 12.4315,
      "step": 10524
    },
    {
      "epoch": 0.5731283140361983,
      "grad_norm": 0.5641999601089314,
      "learning_rate": 0.00016741785074821013,
      "loss": 12.1776,
      "step": 10525
    },
    {
      "epoch": 0.5731827680327813,
      "grad_norm": 0.5547702884056563,
      "learning_rate": 0.00016741133759891948,
      "loss": 12.1544,
      "step": 10526
    },
    {
      "epoch": 0.5732372220293643,
      "grad_norm": 0.542306466498164,
      "learning_rate": 0.00016740482392542703,
      "loss": 12.4514,
      "step": 10527
    },
    {
      "epoch": 0.5732916760259473,
      "grad_norm": 0.6474642132436456,
      "learning_rate": 0.00016739830972778332,
      "loss": 12.4235,
      "step": 10528
    },
    {
      "epoch": 0.5733461300225303,
      "grad_norm": 0.6199669347079713,
      "learning_rate": 0.00016739179500603902,
      "loss": 12.2876,
      "step": 10529
    },
    {
      "epoch": 0.5734005840191133,
      "grad_norm": 0.5828210115421902,
      "learning_rate": 0.0001673852797602449,
      "loss": 12.2469,
      "step": 10530
    },
    {
      "epoch": 0.5734550380156964,
      "grad_norm": 0.6730918250509988,
      "learning_rate": 0.0001673787639904515,
      "loss": 12.4733,
      "step": 10531
    },
    {
      "epoch": 0.5735094920122794,
      "grad_norm": 0.6046063478226098,
      "learning_rate": 0.00016737224769670955,
      "loss": 12.3557,
      "step": 10532
    },
    {
      "epoch": 0.5735639460088624,
      "grad_norm": 0.6125427393316394,
      "learning_rate": 0.0001673657308790697,
      "loss": 12.3181,
      "step": 10533
    },
    {
      "epoch": 0.5736184000054454,
      "grad_norm": 0.6463772232187843,
      "learning_rate": 0.0001673592135375826,
      "loss": 12.4027,
      "step": 10534
    },
    {
      "epoch": 0.5736728540020284,
      "grad_norm": 0.6079419740833918,
      "learning_rate": 0.00016735269567229902,
      "loss": 12.5024,
      "step": 10535
    },
    {
      "epoch": 0.5737273079986114,
      "grad_norm": 0.6303049403018266,
      "learning_rate": 0.00016734617728326952,
      "loss": 12.3737,
      "step": 10536
    },
    {
      "epoch": 0.5737817619951945,
      "grad_norm": 0.6191188706847653,
      "learning_rate": 0.0001673396583705449,
      "loss": 12.4352,
      "step": 10537
    },
    {
      "epoch": 0.5738362159917775,
      "grad_norm": 0.64668220420118,
      "learning_rate": 0.00016733313893417575,
      "loss": 12.3668,
      "step": 10538
    },
    {
      "epoch": 0.5738906699883605,
      "grad_norm": 0.59202145795084,
      "learning_rate": 0.00016732661897421284,
      "loss": 12.4155,
      "step": 10539
    },
    {
      "epoch": 0.5739451239849435,
      "grad_norm": 0.6330526863126181,
      "learning_rate": 0.0001673200984907069,
      "loss": 12.3955,
      "step": 10540
    },
    {
      "epoch": 0.5739995779815265,
      "grad_norm": 0.6282798030727328,
      "learning_rate": 0.00016731357748370852,
      "loss": 12.2411,
      "step": 10541
    },
    {
      "epoch": 0.5740540319781094,
      "grad_norm": 0.5847016640651334,
      "learning_rate": 0.00016730705595326847,
      "loss": 12.2756,
      "step": 10542
    },
    {
      "epoch": 0.5741084859746926,
      "grad_norm": 0.6828398564940577,
      "learning_rate": 0.00016730053389943752,
      "loss": 12.3235,
      "step": 10543
    },
    {
      "epoch": 0.5741629399712755,
      "grad_norm": 0.6756906716831694,
      "learning_rate": 0.0001672940113222663,
      "loss": 12.3097,
      "step": 10544
    },
    {
      "epoch": 0.5742173939678585,
      "grad_norm": 0.5620571029291483,
      "learning_rate": 0.00016728748822180558,
      "loss": 12.4714,
      "step": 10545
    },
    {
      "epoch": 0.5742718479644415,
      "grad_norm": 0.763273675651076,
      "learning_rate": 0.00016728096459810605,
      "loss": 12.3871,
      "step": 10546
    },
    {
      "epoch": 0.5743263019610245,
      "grad_norm": 0.6695252802059574,
      "learning_rate": 0.00016727444045121844,
      "loss": 12.4432,
      "step": 10547
    },
    {
      "epoch": 0.5743807559576075,
      "grad_norm": 0.5481293077792526,
      "learning_rate": 0.00016726791578119352,
      "loss": 12.352,
      "step": 10548
    },
    {
      "epoch": 0.5744352099541906,
      "grad_norm": 0.7292923862086983,
      "learning_rate": 0.00016726139058808204,
      "loss": 12.3642,
      "step": 10549
    },
    {
      "epoch": 0.5744896639507736,
      "grad_norm": 0.6413924900299116,
      "learning_rate": 0.00016725486487193466,
      "loss": 12.4177,
      "step": 10550
    },
    {
      "epoch": 0.5745441179473566,
      "grad_norm": 0.6287623470629832,
      "learning_rate": 0.0001672483386328022,
      "loss": 12.4007,
      "step": 10551
    },
    {
      "epoch": 0.5745985719439396,
      "grad_norm": 0.6708372301718295,
      "learning_rate": 0.00016724181187073532,
      "loss": 12.4079,
      "step": 10552
    },
    {
      "epoch": 0.5746530259405226,
      "grad_norm": 0.6457982109614273,
      "learning_rate": 0.0001672352845857849,
      "loss": 12.4311,
      "step": 10553
    },
    {
      "epoch": 0.5747074799371056,
      "grad_norm": 0.612411900516974,
      "learning_rate": 0.00016722875677800163,
      "loss": 12.4046,
      "step": 10554
    },
    {
      "epoch": 0.5747619339336887,
      "grad_norm": 0.719502918505751,
      "learning_rate": 0.00016722222844743625,
      "loss": 12.3984,
      "step": 10555
    },
    {
      "epoch": 0.5748163879302717,
      "grad_norm": 0.5937007029444668,
      "learning_rate": 0.00016721569959413955,
      "loss": 12.3199,
      "step": 10556
    },
    {
      "epoch": 0.5748708419268547,
      "grad_norm": 0.6120740818602591,
      "learning_rate": 0.00016720917021816232,
      "loss": 12.175,
      "step": 10557
    },
    {
      "epoch": 0.5749252959234377,
      "grad_norm": 0.5793545511529767,
      "learning_rate": 0.00016720264031955528,
      "loss": 12.3855,
      "step": 10558
    },
    {
      "epoch": 0.5749797499200207,
      "grad_norm": 0.611501589018356,
      "learning_rate": 0.00016719610989836926,
      "loss": 12.4145,
      "step": 10559
    },
    {
      "epoch": 0.5750342039166038,
      "grad_norm": 0.5726434094961663,
      "learning_rate": 0.000167189578954655,
      "loss": 12.3545,
      "step": 10560
    },
    {
      "epoch": 0.5750886579131868,
      "grad_norm": 0.6006267957757423,
      "learning_rate": 0.00016718304748846333,
      "loss": 12.3245,
      "step": 10561
    },
    {
      "epoch": 0.5751431119097697,
      "grad_norm": 0.6673176404464749,
      "learning_rate": 0.000167176515499845,
      "loss": 12.3836,
      "step": 10562
    },
    {
      "epoch": 0.5751975659063527,
      "grad_norm": 0.5485759701750896,
      "learning_rate": 0.00016716998298885082,
      "loss": 12.3111,
      "step": 10563
    },
    {
      "epoch": 0.5752520199029357,
      "grad_norm": 0.6530685793868931,
      "learning_rate": 0.00016716344995553159,
      "loss": 12.2943,
      "step": 10564
    },
    {
      "epoch": 0.5753064738995187,
      "grad_norm": 0.6586676988577265,
      "learning_rate": 0.00016715691639993812,
      "loss": 12.2898,
      "step": 10565
    },
    {
      "epoch": 0.5753609278961018,
      "grad_norm": 0.6148634157182052,
      "learning_rate": 0.00016715038232212122,
      "loss": 12.4235,
      "step": 10566
    },
    {
      "epoch": 0.5754153818926848,
      "grad_norm": 0.597136346674799,
      "learning_rate": 0.00016714384772213166,
      "loss": 12.3168,
      "step": 10567
    },
    {
      "epoch": 0.5754698358892678,
      "grad_norm": 0.6297749668051609,
      "learning_rate": 0.0001671373126000203,
      "loss": 12.3847,
      "step": 10568
    },
    {
      "epoch": 0.5755242898858508,
      "grad_norm": 0.5814828821447569,
      "learning_rate": 0.0001671307769558379,
      "loss": 12.3572,
      "step": 10569
    },
    {
      "epoch": 0.5755787438824338,
      "grad_norm": 0.5795019959416221,
      "learning_rate": 0.00016712424078963535,
      "loss": 12.3285,
      "step": 10570
    },
    {
      "epoch": 0.5756331978790168,
      "grad_norm": 0.6514041703969438,
      "learning_rate": 0.00016711770410146343,
      "loss": 12.3869,
      "step": 10571
    },
    {
      "epoch": 0.5756876518755999,
      "grad_norm": 0.5718901998179207,
      "learning_rate": 0.00016711116689137302,
      "loss": 12.3213,
      "step": 10572
    },
    {
      "epoch": 0.5757421058721829,
      "grad_norm": 0.6235036946951503,
      "learning_rate": 0.0001671046291594149,
      "loss": 12.3728,
      "step": 10573
    },
    {
      "epoch": 0.5757965598687659,
      "grad_norm": 0.49537043109391665,
      "learning_rate": 0.00016709809090563991,
      "loss": 12.1873,
      "step": 10574
    },
    {
      "epoch": 0.5758510138653489,
      "grad_norm": 0.5739977173419183,
      "learning_rate": 0.00016709155213009895,
      "loss": 12.4136,
      "step": 10575
    },
    {
      "epoch": 0.5759054678619319,
      "grad_norm": 0.5843541261690129,
      "learning_rate": 0.0001670850128328428,
      "loss": 12.3284,
      "step": 10576
    },
    {
      "epoch": 0.5759599218585149,
      "grad_norm": 0.6166488364641032,
      "learning_rate": 0.00016707847301392236,
      "loss": 12.3386,
      "step": 10577
    },
    {
      "epoch": 0.576014375855098,
      "grad_norm": 0.6460908741074868,
      "learning_rate": 0.00016707193267338844,
      "loss": 12.4182,
      "step": 10578
    },
    {
      "epoch": 0.576068829851681,
      "grad_norm": 0.6129110230694653,
      "learning_rate": 0.00016706539181129195,
      "loss": 12.2851,
      "step": 10579
    },
    {
      "epoch": 0.576123283848264,
      "grad_norm": 0.6352271764388896,
      "learning_rate": 0.00016705885042768372,
      "loss": 12.3625,
      "step": 10580
    },
    {
      "epoch": 0.5761777378448469,
      "grad_norm": 0.7409091782385997,
      "learning_rate": 0.00016705230852261465,
      "loss": 12.4322,
      "step": 10581
    },
    {
      "epoch": 0.5762321918414299,
      "grad_norm": 0.5790484078560002,
      "learning_rate": 0.00016704576609613553,
      "loss": 12.3396,
      "step": 10582
    },
    {
      "epoch": 0.5762866458380129,
      "grad_norm": 0.5480996478430834,
      "learning_rate": 0.0001670392231482973,
      "loss": 12.2522,
      "step": 10583
    },
    {
      "epoch": 0.576341099834596,
      "grad_norm": 0.70971966257255,
      "learning_rate": 0.00016703267967915086,
      "loss": 12.3536,
      "step": 10584
    },
    {
      "epoch": 0.576395553831179,
      "grad_norm": 0.6783815309016988,
      "learning_rate": 0.00016702613568874702,
      "loss": 12.3734,
      "step": 10585
    },
    {
      "epoch": 0.576450007827762,
      "grad_norm": 0.6505384157020722,
      "learning_rate": 0.00016701959117713675,
      "loss": 12.4639,
      "step": 10586
    },
    {
      "epoch": 0.576504461824345,
      "grad_norm": 0.5584833546030031,
      "learning_rate": 0.0001670130461443709,
      "loss": 12.245,
      "step": 10587
    },
    {
      "epoch": 0.576558915820928,
      "grad_norm": 0.6178251227046433,
      "learning_rate": 0.00016700650059050035,
      "loss": 12.3666,
      "step": 10588
    },
    {
      "epoch": 0.576613369817511,
      "grad_norm": 0.6046084629237625,
      "learning_rate": 0.00016699995451557599,
      "loss": 12.3203,
      "step": 10589
    },
    {
      "epoch": 0.5766678238140941,
      "grad_norm": 0.6602116593538877,
      "learning_rate": 0.00016699340791964876,
      "loss": 12.3469,
      "step": 10590
    },
    {
      "epoch": 0.5767222778106771,
      "grad_norm": 0.5750978231757151,
      "learning_rate": 0.0001669868608027696,
      "loss": 12.2704,
      "step": 10591
    },
    {
      "epoch": 0.5767767318072601,
      "grad_norm": 0.713198247391486,
      "learning_rate": 0.00016698031316498933,
      "loss": 12.4092,
      "step": 10592
    },
    {
      "epoch": 0.5768311858038431,
      "grad_norm": 0.5734036929963306,
      "learning_rate": 0.00016697376500635894,
      "loss": 12.3519,
      "step": 10593
    },
    {
      "epoch": 0.5768856398004261,
      "grad_norm": 0.6086445680995939,
      "learning_rate": 0.00016696721632692926,
      "loss": 12.3711,
      "step": 10594
    },
    {
      "epoch": 0.5769400937970092,
      "grad_norm": 0.5706324360010867,
      "learning_rate": 0.00016696066712675134,
      "loss": 12.3824,
      "step": 10595
    },
    {
      "epoch": 0.5769945477935922,
      "grad_norm": 0.5712190011034245,
      "learning_rate": 0.000166954117405876,
      "loss": 12.1989,
      "step": 10596
    },
    {
      "epoch": 0.5770490017901752,
      "grad_norm": 0.597188553842159,
      "learning_rate": 0.00016694756716435427,
      "loss": 12.224,
      "step": 10597
    },
    {
      "epoch": 0.5771034557867581,
      "grad_norm": 0.5742012347288625,
      "learning_rate": 0.000166941016402237,
      "loss": 12.2663,
      "step": 10598
    },
    {
      "epoch": 0.5771579097833411,
      "grad_norm": 0.5841098572750677,
      "learning_rate": 0.00016693446511957514,
      "loss": 12.3919,
      "step": 10599
    },
    {
      "epoch": 0.5772123637799241,
      "grad_norm": 0.5455079817642827,
      "learning_rate": 0.00016692791331641968,
      "loss": 12.1867,
      "step": 10600
    },
    {
      "epoch": 0.5772668177765072,
      "grad_norm": 0.6024082007021208,
      "learning_rate": 0.00016692136099282154,
      "loss": 12.2927,
      "step": 10601
    },
    {
      "epoch": 0.5773212717730902,
      "grad_norm": 0.5426123737996211,
      "learning_rate": 0.0001669148081488317,
      "loss": 12.3577,
      "step": 10602
    },
    {
      "epoch": 0.5773757257696732,
      "grad_norm": 0.6663180089415239,
      "learning_rate": 0.00016690825478450104,
      "loss": 12.3949,
      "step": 10603
    },
    {
      "epoch": 0.5774301797662562,
      "grad_norm": 0.6024385898428207,
      "learning_rate": 0.0001669017008998806,
      "loss": 12.2587,
      "step": 10604
    },
    {
      "epoch": 0.5774846337628392,
      "grad_norm": 0.5818281069112449,
      "learning_rate": 0.0001668951464950213,
      "loss": 12.261,
      "step": 10605
    },
    {
      "epoch": 0.5775390877594222,
      "grad_norm": 0.5903345499037228,
      "learning_rate": 0.00016688859156997415,
      "loss": 12.2928,
      "step": 10606
    },
    {
      "epoch": 0.5775935417560053,
      "grad_norm": 0.5794801873580113,
      "learning_rate": 0.00016688203612479004,
      "loss": 12.4412,
      "step": 10607
    },
    {
      "epoch": 0.5776479957525883,
      "grad_norm": 0.6805587694466598,
      "learning_rate": 0.00016687548015952003,
      "loss": 12.3809,
      "step": 10608
    },
    {
      "epoch": 0.5777024497491713,
      "grad_norm": 0.6843568923556025,
      "learning_rate": 0.0001668689236742151,
      "loss": 12.3297,
      "step": 10609
    },
    {
      "epoch": 0.5777569037457543,
      "grad_norm": 0.6010521056935886,
      "learning_rate": 0.00016686236666892617,
      "loss": 12.3508,
      "step": 10610
    },
    {
      "epoch": 0.5778113577423373,
      "grad_norm": 0.5713679177009463,
      "learning_rate": 0.00016685580914370428,
      "loss": 12.4091,
      "step": 10611
    },
    {
      "epoch": 0.5778658117389203,
      "grad_norm": 0.6063980630363172,
      "learning_rate": 0.00016684925109860038,
      "loss": 12.3177,
      "step": 10612
    },
    {
      "epoch": 0.5779202657355034,
      "grad_norm": 0.613071108821596,
      "learning_rate": 0.0001668426925336655,
      "loss": 12.3598,
      "step": 10613
    },
    {
      "epoch": 0.5779747197320864,
      "grad_norm": 0.6988243533468197,
      "learning_rate": 0.0001668361334489507,
      "loss": 12.3623,
      "step": 10614
    },
    {
      "epoch": 0.5780291737286694,
      "grad_norm": 0.6237317929685072,
      "learning_rate": 0.00016682957384450684,
      "loss": 12.3544,
      "step": 10615
    },
    {
      "epoch": 0.5780836277252523,
      "grad_norm": 0.5961809283619698,
      "learning_rate": 0.00016682301372038504,
      "loss": 12.4062,
      "step": 10616
    },
    {
      "epoch": 0.5781380817218353,
      "grad_norm": 0.6010875834008084,
      "learning_rate": 0.00016681645307663626,
      "loss": 12.2882,
      "step": 10617
    },
    {
      "epoch": 0.5781925357184183,
      "grad_norm": 0.5755892550319603,
      "learning_rate": 0.00016680989191331157,
      "loss": 12.2412,
      "step": 10618
    },
    {
      "epoch": 0.5782469897150014,
      "grad_norm": 0.6412664803443131,
      "learning_rate": 0.0001668033302304619,
      "loss": 12.2868,
      "step": 10619
    },
    {
      "epoch": 0.5783014437115844,
      "grad_norm": 0.5556221992873542,
      "learning_rate": 0.00016679676802813838,
      "loss": 12.3292,
      "step": 10620
    },
    {
      "epoch": 0.5783558977081674,
      "grad_norm": 0.6912225305689517,
      "learning_rate": 0.00016679020530639197,
      "loss": 12.4867,
      "step": 10621
    },
    {
      "epoch": 0.5784103517047504,
      "grad_norm": 0.5984882630019411,
      "learning_rate": 0.00016678364206527372,
      "loss": 12.3977,
      "step": 10622
    },
    {
      "epoch": 0.5784648057013334,
      "grad_norm": 0.561838444976537,
      "learning_rate": 0.00016677707830483468,
      "loss": 12.3208,
      "step": 10623
    },
    {
      "epoch": 0.5785192596979164,
      "grad_norm": 0.6207495383056233,
      "learning_rate": 0.00016677051402512588,
      "loss": 12.3599,
      "step": 10624
    },
    {
      "epoch": 0.5785737136944995,
      "grad_norm": 0.5758976293841622,
      "learning_rate": 0.00016676394922619835,
      "loss": 12.2409,
      "step": 10625
    },
    {
      "epoch": 0.5786281676910825,
      "grad_norm": 0.5909502850088595,
      "learning_rate": 0.00016675738390810319,
      "loss": 12.3915,
      "step": 10626
    },
    {
      "epoch": 0.5786826216876655,
      "grad_norm": 0.6320800577473589,
      "learning_rate": 0.0001667508180708914,
      "loss": 12.3689,
      "step": 10627
    },
    {
      "epoch": 0.5787370756842485,
      "grad_norm": 0.5729497199492393,
      "learning_rate": 0.00016674425171461402,
      "loss": 12.2811,
      "step": 10628
    },
    {
      "epoch": 0.5787915296808315,
      "grad_norm": 0.6089854746204572,
      "learning_rate": 0.0001667376848393222,
      "loss": 12.3424,
      "step": 10629
    },
    {
      "epoch": 0.5788459836774146,
      "grad_norm": 0.6253594955335714,
      "learning_rate": 0.0001667311174450669,
      "loss": 12.3144,
      "step": 10630
    },
    {
      "epoch": 0.5789004376739976,
      "grad_norm": 0.5667608081147807,
      "learning_rate": 0.0001667245495318993,
      "loss": 12.3947,
      "step": 10631
    },
    {
      "epoch": 0.5789548916705806,
      "grad_norm": 0.6135287869538242,
      "learning_rate": 0.00016671798109987035,
      "loss": 12.3181,
      "step": 10632
    },
    {
      "epoch": 0.5790093456671636,
      "grad_norm": 0.5553515736538324,
      "learning_rate": 0.00016671141214903124,
      "loss": 12.2642,
      "step": 10633
    },
    {
      "epoch": 0.5790637996637465,
      "grad_norm": 0.5392317028369967,
      "learning_rate": 0.00016670484267943296,
      "loss": 12.43,
      "step": 10634
    },
    {
      "epoch": 0.5791182536603295,
      "grad_norm": 0.6236632127739922,
      "learning_rate": 0.00016669827269112666,
      "loss": 12.3473,
      "step": 10635
    },
    {
      "epoch": 0.5791727076569126,
      "grad_norm": 0.5955810877538578,
      "learning_rate": 0.00016669170218416342,
      "loss": 12.1647,
      "step": 10636
    },
    {
      "epoch": 0.5792271616534956,
      "grad_norm": 0.683404071019631,
      "learning_rate": 0.0001666851311585943,
      "loss": 12.4758,
      "step": 10637
    },
    {
      "epoch": 0.5792816156500786,
      "grad_norm": 0.5430831866083706,
      "learning_rate": 0.0001666785596144704,
      "loss": 12.1772,
      "step": 10638
    },
    {
      "epoch": 0.5793360696466616,
      "grad_norm": 0.6560979890502256,
      "learning_rate": 0.00016667198755184286,
      "loss": 12.3382,
      "step": 10639
    },
    {
      "epoch": 0.5793905236432446,
      "grad_norm": 0.5806297471494171,
      "learning_rate": 0.00016666541497076278,
      "loss": 12.28,
      "step": 10640
    },
    {
      "epoch": 0.5794449776398276,
      "grad_norm": 0.613292184104423,
      "learning_rate": 0.00016665884187128124,
      "loss": 12.4239,
      "step": 10641
    },
    {
      "epoch": 0.5794994316364107,
      "grad_norm": 0.7204491078949676,
      "learning_rate": 0.00016665226825344936,
      "loss": 12.3856,
      "step": 10642
    },
    {
      "epoch": 0.5795538856329937,
      "grad_norm": 0.5956727900025345,
      "learning_rate": 0.00016664569411731827,
      "loss": 12.248,
      "step": 10643
    },
    {
      "epoch": 0.5796083396295767,
      "grad_norm": 0.6538876553886454,
      "learning_rate": 0.00016663911946293908,
      "loss": 12.261,
      "step": 10644
    },
    {
      "epoch": 0.5796627936261597,
      "grad_norm": 0.6123826693547904,
      "learning_rate": 0.00016663254429036292,
      "loss": 12.3832,
      "step": 10645
    },
    {
      "epoch": 0.5797172476227427,
      "grad_norm": 0.6421627005585591,
      "learning_rate": 0.00016662596859964092,
      "loss": 12.3551,
      "step": 10646
    },
    {
      "epoch": 0.5797717016193257,
      "grad_norm": 0.6798698187958897,
      "learning_rate": 0.00016661939239082422,
      "loss": 12.4644,
      "step": 10647
    },
    {
      "epoch": 0.5798261556159088,
      "grad_norm": 0.620210117669651,
      "learning_rate": 0.00016661281566396395,
      "loss": 12.2705,
      "step": 10648
    },
    {
      "epoch": 0.5798806096124918,
      "grad_norm": 0.6167602633822272,
      "learning_rate": 0.00016660623841911127,
      "loss": 12.3554,
      "step": 10649
    },
    {
      "epoch": 0.5799350636090748,
      "grad_norm": 0.7079619021600241,
      "learning_rate": 0.0001665996606563173,
      "loss": 12.37,
      "step": 10650
    },
    {
      "epoch": 0.5799895176056578,
      "grad_norm": 0.6471379288128399,
      "learning_rate": 0.0001665930823756332,
      "loss": 12.2851,
      "step": 10651
    },
    {
      "epoch": 0.5800439716022407,
      "grad_norm": 0.6250926694948886,
      "learning_rate": 0.00016658650357711014,
      "loss": 12.5016,
      "step": 10652
    },
    {
      "epoch": 0.5800984255988237,
      "grad_norm": 0.7285422010970602,
      "learning_rate": 0.00016657992426079922,
      "loss": 12.3769,
      "step": 10653
    },
    {
      "epoch": 0.5801528795954068,
      "grad_norm": 0.5863535987053579,
      "learning_rate": 0.00016657334442675168,
      "loss": 12.2994,
      "step": 10654
    },
    {
      "epoch": 0.5802073335919898,
      "grad_norm": 1.0460052189248172,
      "learning_rate": 0.00016656676407501863,
      "loss": 12.3731,
      "step": 10655
    },
    {
      "epoch": 0.5802617875885728,
      "grad_norm": 0.6388253747460793,
      "learning_rate": 0.00016656018320565128,
      "loss": 12.3517,
      "step": 10656
    },
    {
      "epoch": 0.5803162415851558,
      "grad_norm": 0.6482402051013342,
      "learning_rate": 0.00016655360181870078,
      "loss": 12.1586,
      "step": 10657
    },
    {
      "epoch": 0.5803706955817388,
      "grad_norm": 0.6075242808500345,
      "learning_rate": 0.0001665470199142183,
      "loss": 12.2407,
      "step": 10658
    },
    {
      "epoch": 0.5804251495783219,
      "grad_norm": 0.5845619181963486,
      "learning_rate": 0.000166540437492255,
      "loss": 12.2608,
      "step": 10659
    },
    {
      "epoch": 0.5804796035749049,
      "grad_norm": 0.6034154716906985,
      "learning_rate": 0.00016653385455286213,
      "loss": 12.3674,
      "step": 10660
    },
    {
      "epoch": 0.5805340575714879,
      "grad_norm": 0.5829797356728805,
      "learning_rate": 0.0001665272710960909,
      "loss": 12.3774,
      "step": 10661
    },
    {
      "epoch": 0.5805885115680709,
      "grad_norm": 0.6855172700573546,
      "learning_rate": 0.00016652068712199239,
      "loss": 12.4455,
      "step": 10662
    },
    {
      "epoch": 0.5806429655646539,
      "grad_norm": 0.6075926608862877,
      "learning_rate": 0.00016651410263061786,
      "loss": 12.4299,
      "step": 10663
    },
    {
      "epoch": 0.5806974195612369,
      "grad_norm": 0.5648326355226941,
      "learning_rate": 0.00016650751762201855,
      "loss": 12.3492,
      "step": 10664
    },
    {
      "epoch": 0.58075187355782,
      "grad_norm": 0.6824578992264828,
      "learning_rate": 0.00016650093209624557,
      "loss": 12.3532,
      "step": 10665
    },
    {
      "epoch": 0.580806327554403,
      "grad_norm": 0.5804425579415122,
      "learning_rate": 0.00016649434605335025,
      "loss": 12.2374,
      "step": 10666
    },
    {
      "epoch": 0.580860781550986,
      "grad_norm": 0.6508938636771904,
      "learning_rate": 0.00016648775949338373,
      "loss": 12.4581,
      "step": 10667
    },
    {
      "epoch": 0.580915235547569,
      "grad_norm": 0.6252040919432668,
      "learning_rate": 0.00016648117241639722,
      "loss": 12.3138,
      "step": 10668
    },
    {
      "epoch": 0.580969689544152,
      "grad_norm": 0.6052652126208161,
      "learning_rate": 0.000166474584822442,
      "loss": 12.4502,
      "step": 10669
    },
    {
      "epoch": 0.581024143540735,
      "grad_norm": 0.6035779320123137,
      "learning_rate": 0.00016646799671156924,
      "loss": 12.2472,
      "step": 10670
    },
    {
      "epoch": 0.581078597537318,
      "grad_norm": 0.6325234440978983,
      "learning_rate": 0.00016646140808383023,
      "loss": 12.2745,
      "step": 10671
    },
    {
      "epoch": 0.581133051533901,
      "grad_norm": 0.5833686290661749,
      "learning_rate": 0.0001664548189392761,
      "loss": 12.3776,
      "step": 10672
    },
    {
      "epoch": 0.581187505530484,
      "grad_norm": 0.5924299333881107,
      "learning_rate": 0.00016644822927795817,
      "loss": 12.3912,
      "step": 10673
    },
    {
      "epoch": 0.581241959527067,
      "grad_norm": 0.5966865121409107,
      "learning_rate": 0.00016644163909992768,
      "loss": 12.2941,
      "step": 10674
    },
    {
      "epoch": 0.58129641352365,
      "grad_norm": 0.571458474944623,
      "learning_rate": 0.00016643504840523586,
      "loss": 12.2669,
      "step": 10675
    },
    {
      "epoch": 0.581350867520233,
      "grad_norm": 0.6325490664211092,
      "learning_rate": 0.00016642845719393398,
      "loss": 12.2097,
      "step": 10676
    },
    {
      "epoch": 0.5814053215168161,
      "grad_norm": 0.5714605500093208,
      "learning_rate": 0.00016642186546607322,
      "loss": 12.2973,
      "step": 10677
    },
    {
      "epoch": 0.5814597755133991,
      "grad_norm": 0.5435175392981482,
      "learning_rate": 0.00016641527322170494,
      "loss": 12.315,
      "step": 10678
    },
    {
      "epoch": 0.5815142295099821,
      "grad_norm": 0.5737258722843944,
      "learning_rate": 0.00016640868046088037,
      "loss": 12.325,
      "step": 10679
    },
    {
      "epoch": 0.5815686835065651,
      "grad_norm": 0.7006617008446661,
      "learning_rate": 0.00016640208718365074,
      "loss": 12.5486,
      "step": 10680
    },
    {
      "epoch": 0.5816231375031481,
      "grad_norm": 0.6174886267902199,
      "learning_rate": 0.00016639549339006736,
      "loss": 12.3535,
      "step": 10681
    },
    {
      "epoch": 0.5816775914997311,
      "grad_norm": 0.6928596384395789,
      "learning_rate": 0.00016638889908018146,
      "loss": 12.2791,
      "step": 10682
    },
    {
      "epoch": 0.5817320454963142,
      "grad_norm": 0.6084022066172773,
      "learning_rate": 0.00016638230425404437,
      "loss": 12.3526,
      "step": 10683
    },
    {
      "epoch": 0.5817864994928972,
      "grad_norm": 0.5687893948538565,
      "learning_rate": 0.00016637570891170732,
      "loss": 12.4278,
      "step": 10684
    },
    {
      "epoch": 0.5818409534894802,
      "grad_norm": 0.6759339144459159,
      "learning_rate": 0.00016636911305322168,
      "loss": 12.366,
      "step": 10685
    },
    {
      "epoch": 0.5818954074860632,
      "grad_norm": 0.6087122523927804,
      "learning_rate": 0.00016636251667863868,
      "loss": 12.3715,
      "step": 10686
    },
    {
      "epoch": 0.5819498614826462,
      "grad_norm": 0.6395203659277144,
      "learning_rate": 0.00016635591978800957,
      "loss": 12.2675,
      "step": 10687
    },
    {
      "epoch": 0.5820043154792292,
      "grad_norm": 0.7568721198329359,
      "learning_rate": 0.00016634932238138574,
      "loss": 12.435,
      "step": 10688
    },
    {
      "epoch": 0.5820587694758123,
      "grad_norm": 0.6359199886095321,
      "learning_rate": 0.00016634272445881844,
      "loss": 12.3296,
      "step": 10689
    },
    {
      "epoch": 0.5821132234723952,
      "grad_norm": 0.6054270859888692,
      "learning_rate": 0.000166336126020359,
      "loss": 12.3969,
      "step": 10690
    },
    {
      "epoch": 0.5821676774689782,
      "grad_norm": 0.6476557577065587,
      "learning_rate": 0.0001663295270660587,
      "loss": 12.3727,
      "step": 10691
    },
    {
      "epoch": 0.5822221314655612,
      "grad_norm": 0.548446163354268,
      "learning_rate": 0.0001663229275959689,
      "loss": 12.1942,
      "step": 10692
    },
    {
      "epoch": 0.5822765854621442,
      "grad_norm": 0.5874516079699055,
      "learning_rate": 0.00016631632761014088,
      "loss": 12.294,
      "step": 10693
    },
    {
      "epoch": 0.5823310394587273,
      "grad_norm": 0.7223823835601816,
      "learning_rate": 0.00016630972710862595,
      "loss": 12.3554,
      "step": 10694
    },
    {
      "epoch": 0.5823854934553103,
      "grad_norm": 0.6923158838651609,
      "learning_rate": 0.0001663031260914755,
      "loss": 12.4702,
      "step": 10695
    },
    {
      "epoch": 0.5824399474518933,
      "grad_norm": 0.6265117547125986,
      "learning_rate": 0.0001662965245587408,
      "loss": 12.266,
      "step": 10696
    },
    {
      "epoch": 0.5824944014484763,
      "grad_norm": 0.6029729478880793,
      "learning_rate": 0.00016628992251047322,
      "loss": 12.3511,
      "step": 10697
    },
    {
      "epoch": 0.5825488554450593,
      "grad_norm": 0.5424235117126971,
      "learning_rate": 0.00016628331994672407,
      "loss": 12.2765,
      "step": 10698
    },
    {
      "epoch": 0.5826033094416423,
      "grad_norm": 0.6365971163700855,
      "learning_rate": 0.00016627671686754471,
      "loss": 12.447,
      "step": 10699
    },
    {
      "epoch": 0.5826577634382254,
      "grad_norm": 0.6047554713915011,
      "learning_rate": 0.00016627011327298652,
      "loss": 12.2762,
      "step": 10700
    },
    {
      "epoch": 0.5827122174348084,
      "grad_norm": 0.635240192156098,
      "learning_rate": 0.00016626350916310078,
      "loss": 12.3235,
      "step": 10701
    },
    {
      "epoch": 0.5827666714313914,
      "grad_norm": 0.6115231624219086,
      "learning_rate": 0.00016625690453793887,
      "loss": 12.3012,
      "step": 10702
    },
    {
      "epoch": 0.5828211254279744,
      "grad_norm": 0.6807057914114546,
      "learning_rate": 0.0001662502993975522,
      "loss": 12.2399,
      "step": 10703
    },
    {
      "epoch": 0.5828755794245574,
      "grad_norm": 0.6231743295966256,
      "learning_rate": 0.00016624369374199205,
      "loss": 12.2065,
      "step": 10704
    },
    {
      "epoch": 0.5829300334211404,
      "grad_norm": 0.620544586110963,
      "learning_rate": 0.00016623708757130986,
      "loss": 12.3695,
      "step": 10705
    },
    {
      "epoch": 0.5829844874177235,
      "grad_norm": 0.5689139742984867,
      "learning_rate": 0.00016623048088555695,
      "loss": 12.3346,
      "step": 10706
    },
    {
      "epoch": 0.5830389414143065,
      "grad_norm": 0.5860319651455662,
      "learning_rate": 0.0001662238736847847,
      "loss": 12.4899,
      "step": 10707
    },
    {
      "epoch": 0.5830933954108894,
      "grad_norm": 0.6495902268166277,
      "learning_rate": 0.00016621726596904456,
      "loss": 12.3605,
      "step": 10708
    },
    {
      "epoch": 0.5831478494074724,
      "grad_norm": 0.6258170212278469,
      "learning_rate": 0.00016621065773838779,
      "loss": 12.1947,
      "step": 10709
    },
    {
      "epoch": 0.5832023034040554,
      "grad_norm": 0.6346930520938242,
      "learning_rate": 0.00016620404899286587,
      "loss": 12.4736,
      "step": 10710
    },
    {
      "epoch": 0.5832567574006384,
      "grad_norm": 0.6556064557353416,
      "learning_rate": 0.00016619743973253018,
      "loss": 12.3694,
      "step": 10711
    },
    {
      "epoch": 0.5833112113972215,
      "grad_norm": 0.6348183258851631,
      "learning_rate": 0.0001661908299574321,
      "loss": 12.3759,
      "step": 10712
    },
    {
      "epoch": 0.5833656653938045,
      "grad_norm": 0.5517331420369923,
      "learning_rate": 0.00016618421966762298,
      "loss": 12.2347,
      "step": 10713
    },
    {
      "epoch": 0.5834201193903875,
      "grad_norm": 0.6066742407789687,
      "learning_rate": 0.0001661776088631543,
      "loss": 12.5373,
      "step": 10714
    },
    {
      "epoch": 0.5834745733869705,
      "grad_norm": 0.5705601878893244,
      "learning_rate": 0.00016617099754407744,
      "loss": 12.3859,
      "step": 10715
    },
    {
      "epoch": 0.5835290273835535,
      "grad_norm": 0.6543925268915343,
      "learning_rate": 0.00016616438571044379,
      "loss": 12.2492,
      "step": 10716
    },
    {
      "epoch": 0.5835834813801365,
      "grad_norm": 0.559740718934576,
      "learning_rate": 0.0001661577733623048,
      "loss": 12.3315,
      "step": 10717
    },
    {
      "epoch": 0.5836379353767196,
      "grad_norm": 0.7204271917294793,
      "learning_rate": 0.00016615116049971184,
      "loss": 12.3104,
      "step": 10718
    },
    {
      "epoch": 0.5836923893733026,
      "grad_norm": 0.5570670501316027,
      "learning_rate": 0.0001661445471227164,
      "loss": 12.2308,
      "step": 10719
    },
    {
      "epoch": 0.5837468433698856,
      "grad_norm": 0.9876212452291068,
      "learning_rate": 0.00016613793323136983,
      "loss": 12.3422,
      "step": 10720
    },
    {
      "epoch": 0.5838012973664686,
      "grad_norm": 0.5470216742140983,
      "learning_rate": 0.00016613131882572363,
      "loss": 12.2743,
      "step": 10721
    },
    {
      "epoch": 0.5838557513630516,
      "grad_norm": 0.6491742554543132,
      "learning_rate": 0.00016612470390582918,
      "loss": 12.4555,
      "step": 10722
    },
    {
      "epoch": 0.5839102053596346,
      "grad_norm": 0.5823548501056763,
      "learning_rate": 0.00016611808847173798,
      "loss": 12.31,
      "step": 10723
    },
    {
      "epoch": 0.5839646593562177,
      "grad_norm": 0.6165231576326428,
      "learning_rate": 0.00016611147252350137,
      "loss": 12.2579,
      "step": 10724
    },
    {
      "epoch": 0.5840191133528007,
      "grad_norm": 0.6820426398894804,
      "learning_rate": 0.00016610485606117093,
      "loss": 12.3685,
      "step": 10725
    },
    {
      "epoch": 0.5840735673493836,
      "grad_norm": 0.5495775000325852,
      "learning_rate": 0.00016609823908479804,
      "loss": 12.3707,
      "step": 10726
    },
    {
      "epoch": 0.5841280213459666,
      "grad_norm": 0.7082474221272369,
      "learning_rate": 0.00016609162159443412,
      "loss": 12.3602,
      "step": 10727
    },
    {
      "epoch": 0.5841824753425496,
      "grad_norm": 0.5726077459516642,
      "learning_rate": 0.0001660850035901307,
      "loss": 12.3536,
      "step": 10728
    },
    {
      "epoch": 0.5842369293391327,
      "grad_norm": 0.6416664026749477,
      "learning_rate": 0.00016607838507193918,
      "loss": 12.3802,
      "step": 10729
    },
    {
      "epoch": 0.5842913833357157,
      "grad_norm": 0.5706655013876935,
      "learning_rate": 0.00016607176603991106,
      "loss": 12.3543,
      "step": 10730
    },
    {
      "epoch": 0.5843458373322987,
      "grad_norm": 0.6420489842793392,
      "learning_rate": 0.00016606514649409782,
      "loss": 12.3892,
      "step": 10731
    },
    {
      "epoch": 0.5844002913288817,
      "grad_norm": 0.5834620300852785,
      "learning_rate": 0.00016605852643455094,
      "loss": 12.275,
      "step": 10732
    },
    {
      "epoch": 0.5844547453254647,
      "grad_norm": 0.5570349802673065,
      "learning_rate": 0.00016605190586132184,
      "loss": 12.4122,
      "step": 10733
    },
    {
      "epoch": 0.5845091993220477,
      "grad_norm": 0.6062036403931096,
      "learning_rate": 0.00016604528477446207,
      "loss": 12.2577,
      "step": 10734
    },
    {
      "epoch": 0.5845636533186308,
      "grad_norm": 0.6008188923462245,
      "learning_rate": 0.00016603866317402307,
      "loss": 12.4129,
      "step": 10735
    },
    {
      "epoch": 0.5846181073152138,
      "grad_norm": 0.5870953040476993,
      "learning_rate": 0.00016603204106005638,
      "loss": 12.3131,
      "step": 10736
    },
    {
      "epoch": 0.5846725613117968,
      "grad_norm": 0.5775654571884085,
      "learning_rate": 0.00016602541843261347,
      "loss": 12.1748,
      "step": 10737
    },
    {
      "epoch": 0.5847270153083798,
      "grad_norm": 0.6057726613370205,
      "learning_rate": 0.0001660187952917458,
      "loss": 12.3621,
      "step": 10738
    },
    {
      "epoch": 0.5847814693049628,
      "grad_norm": 0.6902081706570492,
      "learning_rate": 0.00016601217163750488,
      "loss": 12.5363,
      "step": 10739
    },
    {
      "epoch": 0.5848359233015458,
      "grad_norm": 0.577736725073556,
      "learning_rate": 0.0001660055474699423,
      "loss": 12.3637,
      "step": 10740
    },
    {
      "epoch": 0.5848903772981289,
      "grad_norm": 0.5566594113188544,
      "learning_rate": 0.0001659989227891095,
      "loss": 12.3016,
      "step": 10741
    },
    {
      "epoch": 0.5849448312947119,
      "grad_norm": 0.5903814894869801,
      "learning_rate": 0.000165992297595058,
      "loss": 12.2161,
      "step": 10742
    },
    {
      "epoch": 0.5849992852912949,
      "grad_norm": 0.6212620436447776,
      "learning_rate": 0.00016598567188783934,
      "loss": 12.2501,
      "step": 10743
    },
    {
      "epoch": 0.5850537392878778,
      "grad_norm": 0.6641309249568489,
      "learning_rate": 0.000165979045667505,
      "loss": 12.3462,
      "step": 10744
    },
    {
      "epoch": 0.5851081932844608,
      "grad_norm": 0.6597269040271564,
      "learning_rate": 0.00016597241893410658,
      "loss": 12.4123,
      "step": 10745
    },
    {
      "epoch": 0.5851626472810438,
      "grad_norm": 0.6040301077382447,
      "learning_rate": 0.00016596579168769553,
      "loss": 12.328,
      "step": 10746
    },
    {
      "epoch": 0.5852171012776269,
      "grad_norm": 0.6346788135306696,
      "learning_rate": 0.00016595916392832342,
      "loss": 12.4054,
      "step": 10747
    },
    {
      "epoch": 0.5852715552742099,
      "grad_norm": 0.6213996954737067,
      "learning_rate": 0.00016595253565604182,
      "loss": 12.2539,
      "step": 10748
    },
    {
      "epoch": 0.5853260092707929,
      "grad_norm": 0.7113686564804004,
      "learning_rate": 0.00016594590687090224,
      "loss": 12.3629,
      "step": 10749
    },
    {
      "epoch": 0.5853804632673759,
      "grad_norm": 0.6576936109864955,
      "learning_rate": 0.0001659392775729562,
      "loss": 12.3767,
      "step": 10750
    },
    {
      "epoch": 0.5854349172639589,
      "grad_norm": 0.6351268245810511,
      "learning_rate": 0.00016593264776225528,
      "loss": 12.4524,
      "step": 10751
    },
    {
      "epoch": 0.5854893712605419,
      "grad_norm": 0.7243956423960791,
      "learning_rate": 0.00016592601743885106,
      "loss": 12.2262,
      "step": 10752
    },
    {
      "epoch": 0.585543825257125,
      "grad_norm": 0.5824637034568017,
      "learning_rate": 0.00016591938660279507,
      "loss": 12.1616,
      "step": 10753
    },
    {
      "epoch": 0.585598279253708,
      "grad_norm": 0.6756771306162614,
      "learning_rate": 0.00016591275525413887,
      "loss": 12.3993,
      "step": 10754
    },
    {
      "epoch": 0.585652733250291,
      "grad_norm": 0.6143731101255684,
      "learning_rate": 0.00016590612339293403,
      "loss": 12.2452,
      "step": 10755
    },
    {
      "epoch": 0.585707187246874,
      "grad_norm": 0.6759993024361596,
      "learning_rate": 0.0001658994910192321,
      "loss": 12.3465,
      "step": 10756
    },
    {
      "epoch": 0.585761641243457,
      "grad_norm": 0.5906562704215874,
      "learning_rate": 0.0001658928581330847,
      "loss": 12.3051,
      "step": 10757
    },
    {
      "epoch": 0.58581609524004,
      "grad_norm": 0.7311117839177587,
      "learning_rate": 0.0001658862247345434,
      "loss": 12.3553,
      "step": 10758
    },
    {
      "epoch": 0.5858705492366231,
      "grad_norm": 0.5858251134557313,
      "learning_rate": 0.00016587959082365976,
      "loss": 12.4093,
      "step": 10759
    },
    {
      "epoch": 0.5859250032332061,
      "grad_norm": 0.6897485784356513,
      "learning_rate": 0.00016587295640048535,
      "loss": 12.2952,
      "step": 10760
    },
    {
      "epoch": 0.5859794572297891,
      "grad_norm": 0.6207767171834111,
      "learning_rate": 0.0001658663214650718,
      "loss": 12.1458,
      "step": 10761
    },
    {
      "epoch": 0.586033911226372,
      "grad_norm": 0.6302702106318713,
      "learning_rate": 0.0001658596860174707,
      "loss": 12.352,
      "step": 10762
    },
    {
      "epoch": 0.586088365222955,
      "grad_norm": 0.6097753042029566,
      "learning_rate": 0.0001658530500577336,
      "loss": 12.4807,
      "step": 10763
    },
    {
      "epoch": 0.5861428192195381,
      "grad_norm": 0.691173923513651,
      "learning_rate": 0.00016584641358591217,
      "loss": 12.3892,
      "step": 10764
    },
    {
      "epoch": 0.5861972732161211,
      "grad_norm": 0.6443847707271058,
      "learning_rate": 0.00016583977660205798,
      "loss": 12.4726,
      "step": 10765
    },
    {
      "epoch": 0.5862517272127041,
      "grad_norm": 0.5835580112949047,
      "learning_rate": 0.00016583313910622263,
      "loss": 12.3632,
      "step": 10766
    },
    {
      "epoch": 0.5863061812092871,
      "grad_norm": 0.5751536494120262,
      "learning_rate": 0.00016582650109845778,
      "loss": 12.5108,
      "step": 10767
    },
    {
      "epoch": 0.5863606352058701,
      "grad_norm": 0.6089091834016841,
      "learning_rate": 0.00016581986257881498,
      "loss": 12.2119,
      "step": 10768
    },
    {
      "epoch": 0.5864150892024531,
      "grad_norm": 0.5850101068908325,
      "learning_rate": 0.00016581322354734592,
      "loss": 12.4388,
      "step": 10769
    },
    {
      "epoch": 0.5864695431990362,
      "grad_norm": 0.6138029272911438,
      "learning_rate": 0.0001658065840041022,
      "loss": 12.3174,
      "step": 10770
    },
    {
      "epoch": 0.5865239971956192,
      "grad_norm": 0.5908410281167666,
      "learning_rate": 0.00016579994394913544,
      "loss": 12.4531,
      "step": 10771
    },
    {
      "epoch": 0.5865784511922022,
      "grad_norm": 0.5352941946472402,
      "learning_rate": 0.00016579330338249728,
      "loss": 12.3342,
      "step": 10772
    },
    {
      "epoch": 0.5866329051887852,
      "grad_norm": 0.5648190561349754,
      "learning_rate": 0.00016578666230423935,
      "loss": 12.2958,
      "step": 10773
    },
    {
      "epoch": 0.5866873591853682,
      "grad_norm": 0.6392221609775258,
      "learning_rate": 0.0001657800207144133,
      "loss": 12.4991,
      "step": 10774
    },
    {
      "epoch": 0.5867418131819512,
      "grad_norm": 0.6129857611687894,
      "learning_rate": 0.00016577337861307077,
      "loss": 12.4036,
      "step": 10775
    },
    {
      "epoch": 0.5867962671785343,
      "grad_norm": 0.627119067624158,
      "learning_rate": 0.00016576673600026345,
      "loss": 12.3447,
      "step": 10776
    },
    {
      "epoch": 0.5868507211751173,
      "grad_norm": 0.5847458260011068,
      "learning_rate": 0.00016576009287604294,
      "loss": 12.3191,
      "step": 10777
    },
    {
      "epoch": 0.5869051751717003,
      "grad_norm": 0.6644982665189451,
      "learning_rate": 0.00016575344924046092,
      "loss": 12.3171,
      "step": 10778
    },
    {
      "epoch": 0.5869596291682833,
      "grad_norm": 0.6030409735792923,
      "learning_rate": 0.00016574680509356908,
      "loss": 12.3177,
      "step": 10779
    },
    {
      "epoch": 0.5870140831648663,
      "grad_norm": 0.639646300876133,
      "learning_rate": 0.00016574016043541901,
      "loss": 12.3706,
      "step": 10780
    },
    {
      "epoch": 0.5870685371614492,
      "grad_norm": 0.6364669286753462,
      "learning_rate": 0.00016573351526606248,
      "loss": 12.3207,
      "step": 10781
    },
    {
      "epoch": 0.5871229911580323,
      "grad_norm": 0.6027203188028022,
      "learning_rate": 0.00016572686958555107,
      "loss": 12.3687,
      "step": 10782
    },
    {
      "epoch": 0.5871774451546153,
      "grad_norm": 0.586665394642215,
      "learning_rate": 0.00016572022339393652,
      "loss": 12.1757,
      "step": 10783
    },
    {
      "epoch": 0.5872318991511983,
      "grad_norm": 0.5734703142469153,
      "learning_rate": 0.00016571357669127048,
      "loss": 12.3066,
      "step": 10784
    },
    {
      "epoch": 0.5872863531477813,
      "grad_norm": 0.6073111293302085,
      "learning_rate": 0.00016570692947760464,
      "loss": 12.3468,
      "step": 10785
    },
    {
      "epoch": 0.5873408071443643,
      "grad_norm": 0.6017266465861099,
      "learning_rate": 0.00016570028175299072,
      "loss": 12.0879,
      "step": 10786
    },
    {
      "epoch": 0.5873952611409473,
      "grad_norm": 0.6343969140689849,
      "learning_rate": 0.0001656936335174804,
      "loss": 12.3714,
      "step": 10787
    },
    {
      "epoch": 0.5874497151375304,
      "grad_norm": 0.6209890963122027,
      "learning_rate": 0.00016568698477112533,
      "loss": 12.4377,
      "step": 10788
    },
    {
      "epoch": 0.5875041691341134,
      "grad_norm": 0.5831684736009015,
      "learning_rate": 0.00016568033551397728,
      "loss": 12.3306,
      "step": 10789
    },
    {
      "epoch": 0.5875586231306964,
      "grad_norm": 0.6591646439527126,
      "learning_rate": 0.00016567368574608792,
      "loss": 12.4348,
      "step": 10790
    },
    {
      "epoch": 0.5876130771272794,
      "grad_norm": 0.6084255226376905,
      "learning_rate": 0.00016566703546750896,
      "loss": 12.2603,
      "step": 10791
    },
    {
      "epoch": 0.5876675311238624,
      "grad_norm": 0.5709180561818189,
      "learning_rate": 0.00016566038467829213,
      "loss": 12.384,
      "step": 10792
    },
    {
      "epoch": 0.5877219851204454,
      "grad_norm": 0.6491470129190081,
      "learning_rate": 0.00016565373337848913,
      "loss": 12.4071,
      "step": 10793
    },
    {
      "epoch": 0.5877764391170285,
      "grad_norm": 0.6136482109577118,
      "learning_rate": 0.00016564708156815167,
      "loss": 12.2885,
      "step": 10794
    },
    {
      "epoch": 0.5878308931136115,
      "grad_norm": 0.5805410890498897,
      "learning_rate": 0.00016564042924733152,
      "loss": 12.3627,
      "step": 10795
    },
    {
      "epoch": 0.5878853471101945,
      "grad_norm": 0.5737420997967682,
      "learning_rate": 0.00016563377641608037,
      "loss": 12.336,
      "step": 10796
    },
    {
      "epoch": 0.5879398011067775,
      "grad_norm": 0.6520150115053743,
      "learning_rate": 0.00016562712307445,
      "loss": 12.3845,
      "step": 10797
    },
    {
      "epoch": 0.5879942551033605,
      "grad_norm": 0.5972841491680021,
      "learning_rate": 0.0001656204692224921,
      "loss": 12.3759,
      "step": 10798
    },
    {
      "epoch": 0.5880487090999436,
      "grad_norm": 0.5493871853747994,
      "learning_rate": 0.0001656138148602584,
      "loss": 12.1783,
      "step": 10799
    },
    {
      "epoch": 0.5881031630965265,
      "grad_norm": 0.5822890201774393,
      "learning_rate": 0.0001656071599878007,
      "loss": 12.2892,
      "step": 10800
    },
    {
      "epoch": 0.5881576170931095,
      "grad_norm": 0.6324818215286,
      "learning_rate": 0.0001656005046051707,
      "loss": 12.3577,
      "step": 10801
    },
    {
      "epoch": 0.5882120710896925,
      "grad_norm": 0.5642795838039929,
      "learning_rate": 0.0001655938487124202,
      "loss": 12.3569,
      "step": 10802
    },
    {
      "epoch": 0.5882665250862755,
      "grad_norm": 0.6910574345507834,
      "learning_rate": 0.00016558719230960094,
      "loss": 12.4795,
      "step": 10803
    },
    {
      "epoch": 0.5883209790828585,
      "grad_norm": 0.542802476435009,
      "learning_rate": 0.00016558053539676463,
      "loss": 12.2123,
      "step": 10804
    },
    {
      "epoch": 0.5883754330794416,
      "grad_norm": 0.6043617230353788,
      "learning_rate": 0.0001655738779739631,
      "loss": 12.3511,
      "step": 10805
    },
    {
      "epoch": 0.5884298870760246,
      "grad_norm": 0.5661938081811106,
      "learning_rate": 0.0001655672200412481,
      "loss": 12.3841,
      "step": 10806
    },
    {
      "epoch": 0.5884843410726076,
      "grad_norm": 0.5513840276449579,
      "learning_rate": 0.0001655605615986714,
      "loss": 12.3578,
      "step": 10807
    },
    {
      "epoch": 0.5885387950691906,
      "grad_norm": 0.5935567757194304,
      "learning_rate": 0.00016555390264628482,
      "loss": 12.2641,
      "step": 10808
    },
    {
      "epoch": 0.5885932490657736,
      "grad_norm": 0.6094467186140181,
      "learning_rate": 0.00016554724318414005,
      "loss": 12.3186,
      "step": 10809
    },
    {
      "epoch": 0.5886477030623566,
      "grad_norm": 0.587326969918712,
      "learning_rate": 0.00016554058321228892,
      "loss": 12.3582,
      "step": 10810
    },
    {
      "epoch": 0.5887021570589397,
      "grad_norm": 0.605656258922736,
      "learning_rate": 0.00016553392273078324,
      "loss": 12.3631,
      "step": 10811
    },
    {
      "epoch": 0.5887566110555227,
      "grad_norm": 0.5416599548400733,
      "learning_rate": 0.00016552726173967478,
      "loss": 12.1228,
      "step": 10812
    },
    {
      "epoch": 0.5888110650521057,
      "grad_norm": 0.6324066438312699,
      "learning_rate": 0.00016552060023901537,
      "loss": 12.368,
      "step": 10813
    },
    {
      "epoch": 0.5888655190486887,
      "grad_norm": 0.5809986823631331,
      "learning_rate": 0.00016551393822885675,
      "loss": 12.251,
      "step": 10814
    },
    {
      "epoch": 0.5889199730452717,
      "grad_norm": 0.5715275065882698,
      "learning_rate": 0.00016550727570925077,
      "loss": 12.3118,
      "step": 10815
    },
    {
      "epoch": 0.5889744270418547,
      "grad_norm": 0.5556233171285592,
      "learning_rate": 0.00016550061268024924,
      "loss": 12.3867,
      "step": 10816
    },
    {
      "epoch": 0.5890288810384378,
      "grad_norm": 0.551949562675175,
      "learning_rate": 0.00016549394914190394,
      "loss": 12.2687,
      "step": 10817
    },
    {
      "epoch": 0.5890833350350207,
      "grad_norm": 0.6604225135995097,
      "learning_rate": 0.0001654872850942667,
      "loss": 12.2866,
      "step": 10818
    },
    {
      "epoch": 0.5891377890316037,
      "grad_norm": 0.6122826288090935,
      "learning_rate": 0.00016548062053738943,
      "loss": 12.3139,
      "step": 10819
    },
    {
      "epoch": 0.5891922430281867,
      "grad_norm": 0.6336015057735573,
      "learning_rate": 0.00016547395547132377,
      "loss": 12.3545,
      "step": 10820
    },
    {
      "epoch": 0.5892466970247697,
      "grad_norm": 0.6077660242831809,
      "learning_rate": 0.0001654672898961217,
      "loss": 12.4284,
      "step": 10821
    },
    {
      "epoch": 0.5893011510213527,
      "grad_norm": 0.5933924734135819,
      "learning_rate": 0.00016546062381183504,
      "loss": 12.3477,
      "step": 10822
    },
    {
      "epoch": 0.5893556050179358,
      "grad_norm": 0.5704821578368983,
      "learning_rate": 0.00016545395721851553,
      "loss": 12.3658,
      "step": 10823
    },
    {
      "epoch": 0.5894100590145188,
      "grad_norm": 0.5531890981277237,
      "learning_rate": 0.00016544729011621509,
      "loss": 12.1956,
      "step": 10824
    },
    {
      "epoch": 0.5894645130111018,
      "grad_norm": 0.6078577362006647,
      "learning_rate": 0.00016544062250498556,
      "loss": 12.1014,
      "step": 10825
    },
    {
      "epoch": 0.5895189670076848,
      "grad_norm": 0.6041162891948126,
      "learning_rate": 0.00016543395438487876,
      "loss": 12.3498,
      "step": 10826
    },
    {
      "epoch": 0.5895734210042678,
      "grad_norm": 0.59779517633144,
      "learning_rate": 0.00016542728575594657,
      "loss": 12.3556,
      "step": 10827
    },
    {
      "epoch": 0.5896278750008509,
      "grad_norm": 0.6802641767919028,
      "learning_rate": 0.00016542061661824081,
      "loss": 12.3733,
      "step": 10828
    },
    {
      "epoch": 0.5896823289974339,
      "grad_norm": 0.6133795427116091,
      "learning_rate": 0.0001654139469718134,
      "loss": 12.3098,
      "step": 10829
    },
    {
      "epoch": 0.5897367829940169,
      "grad_norm": 0.625457286881729,
      "learning_rate": 0.00016540727681671613,
      "loss": 12.326,
      "step": 10830
    },
    {
      "epoch": 0.5897912369905999,
      "grad_norm": 0.5949393486441156,
      "learning_rate": 0.00016540060615300096,
      "loss": 12.2548,
      "step": 10831
    },
    {
      "epoch": 0.5898456909871829,
      "grad_norm": 0.5972619721684153,
      "learning_rate": 0.00016539393498071967,
      "loss": 12.2434,
      "step": 10832
    },
    {
      "epoch": 0.5899001449837659,
      "grad_norm": 0.6253120159387654,
      "learning_rate": 0.00016538726329992418,
      "loss": 12.392,
      "step": 10833
    },
    {
      "epoch": 0.589954598980349,
      "grad_norm": 0.6160790096062935,
      "learning_rate": 0.00016538059111066635,
      "loss": 12.4191,
      "step": 10834
    },
    {
      "epoch": 0.590009052976932,
      "grad_norm": 0.6596655883939287,
      "learning_rate": 0.0001653739184129981,
      "loss": 12.4099,
      "step": 10835
    },
    {
      "epoch": 0.590063506973515,
      "grad_norm": 0.5650329151645308,
      "learning_rate": 0.0001653672452069713,
      "loss": 12.2878,
      "step": 10836
    },
    {
      "epoch": 0.5901179609700979,
      "grad_norm": 0.582767331234623,
      "learning_rate": 0.00016536057149263783,
      "loss": 12.3784,
      "step": 10837
    },
    {
      "epoch": 0.5901724149666809,
      "grad_norm": 0.6578672810819257,
      "learning_rate": 0.0001653538972700496,
      "loss": 12.3418,
      "step": 10838
    },
    {
      "epoch": 0.5902268689632639,
      "grad_norm": 0.5438566298833007,
      "learning_rate": 0.0001653472225392585,
      "loss": 12.3361,
      "step": 10839
    },
    {
      "epoch": 0.590281322959847,
      "grad_norm": 0.5609747831008461,
      "learning_rate": 0.00016534054730031643,
      "loss": 12.4159,
      "step": 10840
    },
    {
      "epoch": 0.59033577695643,
      "grad_norm": 0.5772256672982281,
      "learning_rate": 0.00016533387155327533,
      "loss": 12.3209,
      "step": 10841
    },
    {
      "epoch": 0.590390230953013,
      "grad_norm": 0.564930919863431,
      "learning_rate": 0.00016532719529818704,
      "loss": 12.1613,
      "step": 10842
    },
    {
      "epoch": 0.590444684949596,
      "grad_norm": 0.5629067263774307,
      "learning_rate": 0.00016532051853510358,
      "loss": 12.4296,
      "step": 10843
    },
    {
      "epoch": 0.590499138946179,
      "grad_norm": 0.5442203642637816,
      "learning_rate": 0.00016531384126407681,
      "loss": 12.3441,
      "step": 10844
    },
    {
      "epoch": 0.590553592942762,
      "grad_norm": 0.6046789600876962,
      "learning_rate": 0.00016530716348515863,
      "loss": 12.3517,
      "step": 10845
    },
    {
      "epoch": 0.5906080469393451,
      "grad_norm": 0.580830860122665,
      "learning_rate": 0.000165300485198401,
      "loss": 12.2095,
      "step": 10846
    },
    {
      "epoch": 0.5906625009359281,
      "grad_norm": 0.6466629974220863,
      "learning_rate": 0.00016529380640385589,
      "loss": 12.4153,
      "step": 10847
    },
    {
      "epoch": 0.5907169549325111,
      "grad_norm": 0.6339639913605484,
      "learning_rate": 0.00016528712710157513,
      "loss": 12.4205,
      "step": 10848
    },
    {
      "epoch": 0.5907714089290941,
      "grad_norm": 0.6195441409327394,
      "learning_rate": 0.00016528044729161075,
      "loss": 12.2604,
      "step": 10849
    },
    {
      "epoch": 0.5908258629256771,
      "grad_norm": 0.6118504980387967,
      "learning_rate": 0.00016527376697401464,
      "loss": 12.3924,
      "step": 10850
    },
    {
      "epoch": 0.5908803169222601,
      "grad_norm": 0.5882431282833214,
      "learning_rate": 0.0001652670861488388,
      "loss": 12.2959,
      "step": 10851
    },
    {
      "epoch": 0.5909347709188432,
      "grad_norm": 0.6832713314144646,
      "learning_rate": 0.00016526040481613515,
      "loss": 12.3463,
      "step": 10852
    },
    {
      "epoch": 0.5909892249154262,
      "grad_norm": 0.6671222847117588,
      "learning_rate": 0.00016525372297595563,
      "loss": 12.3169,
      "step": 10853
    },
    {
      "epoch": 0.5910436789120092,
      "grad_norm": 0.5630347120264059,
      "learning_rate": 0.0001652470406283522,
      "loss": 12.3274,
      "step": 10854
    },
    {
      "epoch": 0.5910981329085921,
      "grad_norm": 0.69419863059305,
      "learning_rate": 0.0001652403577733769,
      "loss": 12.4147,
      "step": 10855
    },
    {
      "epoch": 0.5911525869051751,
      "grad_norm": 0.6071255184021517,
      "learning_rate": 0.0001652336744110816,
      "loss": 12.2184,
      "step": 10856
    },
    {
      "epoch": 0.5912070409017581,
      "grad_norm": 0.6381170907717836,
      "learning_rate": 0.0001652269905415183,
      "loss": 12.4493,
      "step": 10857
    },
    {
      "epoch": 0.5912614948983412,
      "grad_norm": 0.6156674336316467,
      "learning_rate": 0.000165220306164739,
      "loss": 12.4876,
      "step": 10858
    },
    {
      "epoch": 0.5913159488949242,
      "grad_norm": 0.6300556219877623,
      "learning_rate": 0.00016521362128079568,
      "loss": 12.4234,
      "step": 10859
    },
    {
      "epoch": 0.5913704028915072,
      "grad_norm": 0.5903640259607501,
      "learning_rate": 0.00016520693588974026,
      "loss": 12.3806,
      "step": 10860
    },
    {
      "epoch": 0.5914248568880902,
      "grad_norm": 0.6030768851262162,
      "learning_rate": 0.0001652002499916248,
      "loss": 12.2749,
      "step": 10861
    },
    {
      "epoch": 0.5914793108846732,
      "grad_norm": 0.5589972880187812,
      "learning_rate": 0.00016519356358650125,
      "loss": 12.2358,
      "step": 10862
    },
    {
      "epoch": 0.5915337648812563,
      "grad_norm": 0.6032709667447022,
      "learning_rate": 0.00016518687667442164,
      "loss": 12.352,
      "step": 10863
    },
    {
      "epoch": 0.5915882188778393,
      "grad_norm": 0.5754656477262209,
      "learning_rate": 0.00016518018925543791,
      "loss": 12.3834,
      "step": 10864
    },
    {
      "epoch": 0.5916426728744223,
      "grad_norm": 0.7367348500947619,
      "learning_rate": 0.0001651735013296021,
      "loss": 12.4114,
      "step": 10865
    },
    {
      "epoch": 0.5916971268710053,
      "grad_norm": 0.5771815214137505,
      "learning_rate": 0.00016516681289696625,
      "loss": 12.2369,
      "step": 10866
    },
    {
      "epoch": 0.5917515808675883,
      "grad_norm": 0.5986111921661764,
      "learning_rate": 0.00016516012395758231,
      "loss": 12.2583,
      "step": 10867
    },
    {
      "epoch": 0.5918060348641713,
      "grad_norm": 0.7461476212901754,
      "learning_rate": 0.00016515343451150232,
      "loss": 12.3905,
      "step": 10868
    },
    {
      "epoch": 0.5918604888607544,
      "grad_norm": 0.6240151613156143,
      "learning_rate": 0.0001651467445587783,
      "loss": 12.3152,
      "step": 10869
    },
    {
      "epoch": 0.5919149428573374,
      "grad_norm": 0.6227323246109967,
      "learning_rate": 0.00016514005409946228,
      "loss": 12.2702,
      "step": 10870
    },
    {
      "epoch": 0.5919693968539204,
      "grad_norm": 0.6066607791261394,
      "learning_rate": 0.00016513336313360623,
      "loss": 12.4799,
      "step": 10871
    },
    {
      "epoch": 0.5920238508505034,
      "grad_norm": 0.621186986997314,
      "learning_rate": 0.00016512667166126228,
      "loss": 12.3654,
      "step": 10872
    },
    {
      "epoch": 0.5920783048470863,
      "grad_norm": 0.6127640166956305,
      "learning_rate": 0.0001651199796824824,
      "loss": 12.3064,
      "step": 10873
    },
    {
      "epoch": 0.5921327588436693,
      "grad_norm": 0.5862571393568207,
      "learning_rate": 0.00016511328719731862,
      "loss": 12.268,
      "step": 10874
    },
    {
      "epoch": 0.5921872128402524,
      "grad_norm": 0.5395592050833669,
      "learning_rate": 0.00016510659420582302,
      "loss": 12.2713,
      "step": 10875
    },
    {
      "epoch": 0.5922416668368354,
      "grad_norm": 0.540435062260064,
      "learning_rate": 0.0001650999007080476,
      "loss": 12.3872,
      "step": 10876
    },
    {
      "epoch": 0.5922961208334184,
      "grad_norm": 0.5877623442475959,
      "learning_rate": 0.00016509320670404445,
      "loss": 12.2217,
      "step": 10877
    },
    {
      "epoch": 0.5923505748300014,
      "grad_norm": 0.5951671429030703,
      "learning_rate": 0.0001650865121938656,
      "loss": 12.4027,
      "step": 10878
    },
    {
      "epoch": 0.5924050288265844,
      "grad_norm": 0.6337323510468316,
      "learning_rate": 0.0001650798171775631,
      "loss": 12.3653,
      "step": 10879
    },
    {
      "epoch": 0.5924594828231674,
      "grad_norm": 0.6740988540522351,
      "learning_rate": 0.00016507312165518908,
      "loss": 12.3495,
      "step": 10880
    },
    {
      "epoch": 0.5925139368197505,
      "grad_norm": 0.5834673840846617,
      "learning_rate": 0.00016506642562679548,
      "loss": 12.0696,
      "step": 10881
    },
    {
      "epoch": 0.5925683908163335,
      "grad_norm": 0.595216213228313,
      "learning_rate": 0.0001650597290924345,
      "loss": 12.3565,
      "step": 10882
    },
    {
      "epoch": 0.5926228448129165,
      "grad_norm": 0.6432560364263551,
      "learning_rate": 0.00016505303205215815,
      "loss": 12.3607,
      "step": 10883
    },
    {
      "epoch": 0.5926772988094995,
      "grad_norm": 0.586856815044308,
      "learning_rate": 0.0001650463345060185,
      "loss": 12.312,
      "step": 10884
    },
    {
      "epoch": 0.5927317528060825,
      "grad_norm": 0.5832360201559386,
      "learning_rate": 0.00016503963645406763,
      "loss": 12.2108,
      "step": 10885
    },
    {
      "epoch": 0.5927862068026655,
      "grad_norm": 0.6363677337191588,
      "learning_rate": 0.00016503293789635768,
      "loss": 12.2894,
      "step": 10886
    },
    {
      "epoch": 0.5928406607992486,
      "grad_norm": 0.6366862094278207,
      "learning_rate": 0.00016502623883294065,
      "loss": 12.2471,
      "step": 10887
    },
    {
      "epoch": 0.5928951147958316,
      "grad_norm": 0.6290404501627677,
      "learning_rate": 0.00016501953926386872,
      "loss": 12.3175,
      "step": 10888
    },
    {
      "epoch": 0.5929495687924146,
      "grad_norm": 0.6032367742735714,
      "learning_rate": 0.00016501283918919394,
      "loss": 12.303,
      "step": 10889
    },
    {
      "epoch": 0.5930040227889976,
      "grad_norm": 0.5908359572494107,
      "learning_rate": 0.00016500613860896842,
      "loss": 12.3907,
      "step": 10890
    },
    {
      "epoch": 0.5930584767855805,
      "grad_norm": 0.6263770904868716,
      "learning_rate": 0.00016499943752324426,
      "loss": 12.3837,
      "step": 10891
    },
    {
      "epoch": 0.5931129307821635,
      "grad_norm": 0.5627338226718853,
      "learning_rate": 0.0001649927359320736,
      "loss": 12.3537,
      "step": 10892
    },
    {
      "epoch": 0.5931673847787466,
      "grad_norm": 0.5863381767298946,
      "learning_rate": 0.00016498603383550847,
      "loss": 12.311,
      "step": 10893
    },
    {
      "epoch": 0.5932218387753296,
      "grad_norm": 0.5987115449104812,
      "learning_rate": 0.0001649793312336011,
      "loss": 12.4243,
      "step": 10894
    },
    {
      "epoch": 0.5932762927719126,
      "grad_norm": 0.6024755309453307,
      "learning_rate": 0.0001649726281264035,
      "loss": 12.3362,
      "step": 10895
    },
    {
      "epoch": 0.5933307467684956,
      "grad_norm": 0.5197680717807689,
      "learning_rate": 0.00016496592451396787,
      "loss": 12.3468,
      "step": 10896
    },
    {
      "epoch": 0.5933852007650786,
      "grad_norm": 0.6111589305287626,
      "learning_rate": 0.00016495922039634633,
      "loss": 12.2591,
      "step": 10897
    },
    {
      "epoch": 0.5934396547616617,
      "grad_norm": 0.6600374756617288,
      "learning_rate": 0.000164952515773591,
      "loss": 12.4693,
      "step": 10898
    },
    {
      "epoch": 0.5934941087582447,
      "grad_norm": 0.6008571027135967,
      "learning_rate": 0.00016494581064575397,
      "loss": 12.2133,
      "step": 10899
    },
    {
      "epoch": 0.5935485627548277,
      "grad_norm": 0.6174842596701852,
      "learning_rate": 0.0001649391050128875,
      "loss": 12.3995,
      "step": 10900
    },
    {
      "epoch": 0.5936030167514107,
      "grad_norm": 0.5966629019054353,
      "learning_rate": 0.0001649323988750436,
      "loss": 12.3835,
      "step": 10901
    },
    {
      "epoch": 0.5936574707479937,
      "grad_norm": 0.6241955185191029,
      "learning_rate": 0.0001649256922322745,
      "loss": 12.3144,
      "step": 10902
    },
    {
      "epoch": 0.5937119247445767,
      "grad_norm": 0.6195522331438866,
      "learning_rate": 0.00016491898508463234,
      "loss": 12.3529,
      "step": 10903
    },
    {
      "epoch": 0.5937663787411598,
      "grad_norm": 0.6498619538274419,
      "learning_rate": 0.00016491227743216925,
      "loss": 12.346,
      "step": 10904
    },
    {
      "epoch": 0.5938208327377428,
      "grad_norm": 0.6874857310664496,
      "learning_rate": 0.00016490556927493738,
      "loss": 12.3812,
      "step": 10905
    },
    {
      "epoch": 0.5938752867343258,
      "grad_norm": 0.5901940883933078,
      "learning_rate": 0.00016489886061298896,
      "loss": 12.3213,
      "step": 10906
    },
    {
      "epoch": 0.5939297407309088,
      "grad_norm": 0.686173498986228,
      "learning_rate": 0.0001648921514463761,
      "loss": 12.3013,
      "step": 10907
    },
    {
      "epoch": 0.5939841947274918,
      "grad_norm": 0.6804757445573253,
      "learning_rate": 0.000164885441775151,
      "loss": 12.4066,
      "step": 10908
    },
    {
      "epoch": 0.5940386487240747,
      "grad_norm": 0.5947407757059022,
      "learning_rate": 0.0001648787315993658,
      "loss": 12.3664,
      "step": 10909
    },
    {
      "epoch": 0.5940931027206579,
      "grad_norm": 0.5568712242564401,
      "learning_rate": 0.0001648720209190727,
      "loss": 12.3068,
      "step": 10910
    },
    {
      "epoch": 0.5941475567172408,
      "grad_norm": 0.6673244286932989,
      "learning_rate": 0.00016486530973432387,
      "loss": 12.4117,
      "step": 10911
    },
    {
      "epoch": 0.5942020107138238,
      "grad_norm": 0.5962148224888308,
      "learning_rate": 0.00016485859804517156,
      "loss": 12.3541,
      "step": 10912
    },
    {
      "epoch": 0.5942564647104068,
      "grad_norm": 0.7011669751181724,
      "learning_rate": 0.00016485188585166794,
      "loss": 12.3054,
      "step": 10913
    },
    {
      "epoch": 0.5943109187069898,
      "grad_norm": 0.5983241513440197,
      "learning_rate": 0.0001648451731538651,
      "loss": 12.3304,
      "step": 10914
    },
    {
      "epoch": 0.5943653727035728,
      "grad_norm": 0.6500449585705957,
      "learning_rate": 0.00016483845995181537,
      "loss": 12.3092,
      "step": 10915
    },
    {
      "epoch": 0.5944198267001559,
      "grad_norm": 0.7336462821120124,
      "learning_rate": 0.0001648317462455709,
      "loss": 12.2935,
      "step": 10916
    },
    {
      "epoch": 0.5944742806967389,
      "grad_norm": 0.6561148282831947,
      "learning_rate": 0.00016482503203518387,
      "loss": 12.3112,
      "step": 10917
    },
    {
      "epoch": 0.5945287346933219,
      "grad_norm": 0.5695500651013421,
      "learning_rate": 0.00016481831732070656,
      "loss": 12.4201,
      "step": 10918
    },
    {
      "epoch": 0.5945831886899049,
      "grad_norm": 0.5918640574477187,
      "learning_rate": 0.0001648116021021911,
      "loss": 12.185,
      "step": 10919
    },
    {
      "epoch": 0.5946376426864879,
      "grad_norm": 0.5833610791625058,
      "learning_rate": 0.00016480488637968978,
      "loss": 12.256,
      "step": 10920
    },
    {
      "epoch": 0.5946920966830709,
      "grad_norm": 0.6358894046455806,
      "learning_rate": 0.00016479817015325478,
      "loss": 12.2758,
      "step": 10921
    },
    {
      "epoch": 0.594746550679654,
      "grad_norm": 0.6786451896587979,
      "learning_rate": 0.00016479145342293837,
      "loss": 12.3409,
      "step": 10922
    },
    {
      "epoch": 0.594801004676237,
      "grad_norm": 0.6305781227330443,
      "learning_rate": 0.00016478473618879272,
      "loss": 12.2793,
      "step": 10923
    },
    {
      "epoch": 0.59485545867282,
      "grad_norm": 0.6153787188546451,
      "learning_rate": 0.00016477801845087012,
      "loss": 12.2978,
      "step": 10924
    },
    {
      "epoch": 0.594909912669403,
      "grad_norm": 0.6843363709563547,
      "learning_rate": 0.00016477130020922277,
      "loss": 12.4037,
      "step": 10925
    },
    {
      "epoch": 0.594964366665986,
      "grad_norm": 0.6054378314241865,
      "learning_rate": 0.00016476458146390296,
      "loss": 12.3749,
      "step": 10926
    },
    {
      "epoch": 0.595018820662569,
      "grad_norm": 0.6777685328611128,
      "learning_rate": 0.0001647578622149629,
      "loss": 12.3875,
      "step": 10927
    },
    {
      "epoch": 0.595073274659152,
      "grad_norm": 0.6999698137431501,
      "learning_rate": 0.00016475114246245482,
      "loss": 12.2763,
      "step": 10928
    },
    {
      "epoch": 0.595127728655735,
      "grad_norm": 0.6153828446300298,
      "learning_rate": 0.00016474442220643098,
      "loss": 12.386,
      "step": 10929
    },
    {
      "epoch": 0.595182182652318,
      "grad_norm": 0.6144697499687396,
      "learning_rate": 0.00016473770144694367,
      "loss": 12.4144,
      "step": 10930
    },
    {
      "epoch": 0.595236636648901,
      "grad_norm": 0.6330746485334776,
      "learning_rate": 0.00016473098018404513,
      "loss": 12.3863,
      "step": 10931
    },
    {
      "epoch": 0.595291090645484,
      "grad_norm": 0.5948147818482774,
      "learning_rate": 0.00016472425841778767,
      "loss": 12.3491,
      "step": 10932
    },
    {
      "epoch": 0.5953455446420671,
      "grad_norm": 0.5672408621231471,
      "learning_rate": 0.0001647175361482235,
      "loss": 12.1104,
      "step": 10933
    },
    {
      "epoch": 0.5953999986386501,
      "grad_norm": 0.6833874407205437,
      "learning_rate": 0.0001647108133754049,
      "loss": 12.4,
      "step": 10934
    },
    {
      "epoch": 0.5954544526352331,
      "grad_norm": 0.6494403309332063,
      "learning_rate": 0.0001647040900993842,
      "loss": 12.31,
      "step": 10935
    },
    {
      "epoch": 0.5955089066318161,
      "grad_norm": 0.6642420865825933,
      "learning_rate": 0.0001646973663202136,
      "loss": 12.435,
      "step": 10936
    },
    {
      "epoch": 0.5955633606283991,
      "grad_norm": 0.7412025529100646,
      "learning_rate": 0.00016469064203794543,
      "loss": 12.5043,
      "step": 10937
    },
    {
      "epoch": 0.5956178146249821,
      "grad_norm": 0.6756310812150244,
      "learning_rate": 0.00016468391725263203,
      "loss": 12.4486,
      "step": 10938
    },
    {
      "epoch": 0.5956722686215652,
      "grad_norm": 0.6006988925370298,
      "learning_rate": 0.0001646771919643256,
      "loss": 12.2226,
      "step": 10939
    },
    {
      "epoch": 0.5957267226181482,
      "grad_norm": 0.7039158770908783,
      "learning_rate": 0.0001646704661730785,
      "loss": 12.4274,
      "step": 10940
    },
    {
      "epoch": 0.5957811766147312,
      "grad_norm": 0.6309968178306992,
      "learning_rate": 0.000164663739878943,
      "loss": 12.4284,
      "step": 10941
    },
    {
      "epoch": 0.5958356306113142,
      "grad_norm": 0.5875586723598385,
      "learning_rate": 0.00016465701308197143,
      "loss": 12.3632,
      "step": 10942
    },
    {
      "epoch": 0.5958900846078972,
      "grad_norm": 0.6376103253963531,
      "learning_rate": 0.00016465028578221605,
      "loss": 12.3814,
      "step": 10943
    },
    {
      "epoch": 0.5959445386044802,
      "grad_norm": 0.5810749677416577,
      "learning_rate": 0.00016464355797972922,
      "loss": 12.3421,
      "step": 10944
    },
    {
      "epoch": 0.5959989926010633,
      "grad_norm": 0.6630727468617987,
      "learning_rate": 0.00016463682967456325,
      "loss": 12.3919,
      "step": 10945
    },
    {
      "epoch": 0.5960534465976463,
      "grad_norm": 0.633931614301666,
      "learning_rate": 0.00016463010086677048,
      "loss": 12.36,
      "step": 10946
    },
    {
      "epoch": 0.5961079005942292,
      "grad_norm": 0.58915438361807,
      "learning_rate": 0.00016462337155640316,
      "loss": 12.3195,
      "step": 10947
    },
    {
      "epoch": 0.5961623545908122,
      "grad_norm": 0.5674771043498464,
      "learning_rate": 0.0001646166417435137,
      "loss": 12.3314,
      "step": 10948
    },
    {
      "epoch": 0.5962168085873952,
      "grad_norm": 0.6278043046671282,
      "learning_rate": 0.00016460991142815435,
      "loss": 12.4202,
      "step": 10949
    },
    {
      "epoch": 0.5962712625839782,
      "grad_norm": 0.6091194463995885,
      "learning_rate": 0.00016460318061037752,
      "loss": 12.2741,
      "step": 10950
    },
    {
      "epoch": 0.5963257165805613,
      "grad_norm": 0.5929091072302531,
      "learning_rate": 0.00016459644929023553,
      "loss": 12.215,
      "step": 10951
    },
    {
      "epoch": 0.5963801705771443,
      "grad_norm": 0.539969046068889,
      "learning_rate": 0.00016458971746778072,
      "loss": 12.2799,
      "step": 10952
    },
    {
      "epoch": 0.5964346245737273,
      "grad_norm": 0.5477239819241904,
      "learning_rate": 0.00016458298514306546,
      "loss": 12.2254,
      "step": 10953
    },
    {
      "epoch": 0.5964890785703103,
      "grad_norm": 0.5945626403678087,
      "learning_rate": 0.000164576252316142,
      "loss": 12.2452,
      "step": 10954
    },
    {
      "epoch": 0.5965435325668933,
      "grad_norm": 0.6720891255289737,
      "learning_rate": 0.00016456951898706284,
      "loss": 12.3979,
      "step": 10955
    },
    {
      "epoch": 0.5965979865634763,
      "grad_norm": 0.6707997944507423,
      "learning_rate": 0.00016456278515588024,
      "loss": 12.4544,
      "step": 10956
    },
    {
      "epoch": 0.5966524405600594,
      "grad_norm": 0.6431423071095793,
      "learning_rate": 0.0001645560508226466,
      "loss": 12.301,
      "step": 10957
    },
    {
      "epoch": 0.5967068945566424,
      "grad_norm": 0.6959504961207419,
      "learning_rate": 0.0001645493159874143,
      "loss": 12.3559,
      "step": 10958
    },
    {
      "epoch": 0.5967613485532254,
      "grad_norm": 0.5793628168767085,
      "learning_rate": 0.00016454258065023568,
      "loss": 12.3758,
      "step": 10959
    },
    {
      "epoch": 0.5968158025498084,
      "grad_norm": 0.6424349375411575,
      "learning_rate": 0.00016453584481116313,
      "loss": 12.3487,
      "step": 10960
    },
    {
      "epoch": 0.5968702565463914,
      "grad_norm": 0.6275970493399082,
      "learning_rate": 0.00016452910847024901,
      "loss": 12.4393,
      "step": 10961
    },
    {
      "epoch": 0.5969247105429745,
      "grad_norm": 0.6446562906094953,
      "learning_rate": 0.00016452237162754577,
      "loss": 12.4605,
      "step": 10962
    },
    {
      "epoch": 0.5969791645395575,
      "grad_norm": 0.5735543134471323,
      "learning_rate": 0.00016451563428310575,
      "loss": 12.2989,
      "step": 10963
    },
    {
      "epoch": 0.5970336185361405,
      "grad_norm": 0.6222944111973252,
      "learning_rate": 0.0001645088964369813,
      "loss": 12.2292,
      "step": 10964
    },
    {
      "epoch": 0.5970880725327234,
      "grad_norm": 0.5985903452256347,
      "learning_rate": 0.00016450215808922483,
      "loss": 12.3215,
      "step": 10965
    },
    {
      "epoch": 0.5971425265293064,
      "grad_norm": 0.7412074618453982,
      "learning_rate": 0.0001644954192398888,
      "loss": 12.264,
      "step": 10966
    },
    {
      "epoch": 0.5971969805258894,
      "grad_norm": 0.6439435344918585,
      "learning_rate": 0.00016448867988902557,
      "loss": 12.3645,
      "step": 10967
    },
    {
      "epoch": 0.5972514345224725,
      "grad_norm": 0.6429930022822522,
      "learning_rate": 0.00016448194003668756,
      "loss": 12.1924,
      "step": 10968
    },
    {
      "epoch": 0.5973058885190555,
      "grad_norm": 0.6067151013037995,
      "learning_rate": 0.00016447519968292715,
      "loss": 12.2995,
      "step": 10969
    },
    {
      "epoch": 0.5973603425156385,
      "grad_norm": 0.6413785659924183,
      "learning_rate": 0.0001644684588277968,
      "loss": 12.2482,
      "step": 10970
    },
    {
      "epoch": 0.5974147965122215,
      "grad_norm": 0.6230928886773899,
      "learning_rate": 0.00016446171747134888,
      "loss": 12.3143,
      "step": 10971
    },
    {
      "epoch": 0.5974692505088045,
      "grad_norm": 0.6399022650982504,
      "learning_rate": 0.0001644549756136358,
      "loss": 12.2518,
      "step": 10972
    },
    {
      "epoch": 0.5975237045053875,
      "grad_norm": 0.6035692650993121,
      "learning_rate": 0.0001644482332547101,
      "loss": 12.3228,
      "step": 10973
    },
    {
      "epoch": 0.5975781585019706,
      "grad_norm": 0.5990368630408056,
      "learning_rate": 0.00016444149039462409,
      "loss": 12.218,
      "step": 10974
    },
    {
      "epoch": 0.5976326124985536,
      "grad_norm": 0.6111786788841953,
      "learning_rate": 0.00016443474703343024,
      "loss": 12.3642,
      "step": 10975
    },
    {
      "epoch": 0.5976870664951366,
      "grad_norm": 0.6189012848292096,
      "learning_rate": 0.000164428003171181,
      "loss": 12.3183,
      "step": 10976
    },
    {
      "epoch": 0.5977415204917196,
      "grad_norm": 0.5916709873890514,
      "learning_rate": 0.0001644212588079288,
      "loss": 12.3817,
      "step": 10977
    },
    {
      "epoch": 0.5977959744883026,
      "grad_norm": 0.7836661075155655,
      "learning_rate": 0.00016441451394372608,
      "loss": 12.2071,
      "step": 10978
    },
    {
      "epoch": 0.5978504284848856,
      "grad_norm": 0.5987173941667159,
      "learning_rate": 0.0001644077685786253,
      "loss": 12.5297,
      "step": 10979
    },
    {
      "epoch": 0.5979048824814687,
      "grad_norm": 0.5824740328749117,
      "learning_rate": 0.00016440102271267893,
      "loss": 12.3024,
      "step": 10980
    },
    {
      "epoch": 0.5979593364780517,
      "grad_norm": 0.5981170336986796,
      "learning_rate": 0.00016439427634593938,
      "loss": 12.1627,
      "step": 10981
    },
    {
      "epoch": 0.5980137904746347,
      "grad_norm": 0.6307947757447596,
      "learning_rate": 0.00016438752947845915,
      "loss": 12.3911,
      "step": 10982
    },
    {
      "epoch": 0.5980682444712176,
      "grad_norm": 0.6119378461428637,
      "learning_rate": 0.0001643807821102907,
      "loss": 12.4517,
      "step": 10983
    },
    {
      "epoch": 0.5981226984678006,
      "grad_norm": 0.5535905143981861,
      "learning_rate": 0.00016437403424148647,
      "loss": 12.1956,
      "step": 10984
    },
    {
      "epoch": 0.5981771524643836,
      "grad_norm": 0.6197162454189239,
      "learning_rate": 0.00016436728587209898,
      "loss": 12.4103,
      "step": 10985
    },
    {
      "epoch": 0.5982316064609667,
      "grad_norm": 0.6043613316867456,
      "learning_rate": 0.00016436053700218066,
      "loss": 12.3657,
      "step": 10986
    },
    {
      "epoch": 0.5982860604575497,
      "grad_norm": 0.6080395119756771,
      "learning_rate": 0.00016435378763178404,
      "loss": 12.3862,
      "step": 10987
    },
    {
      "epoch": 0.5983405144541327,
      "grad_norm": 0.6295366320303318,
      "learning_rate": 0.00016434703776096153,
      "loss": 12.3586,
      "step": 10988
    },
    {
      "epoch": 0.5983949684507157,
      "grad_norm": 0.6145068345615352,
      "learning_rate": 0.0001643402873897657,
      "loss": 12.3016,
      "step": 10989
    },
    {
      "epoch": 0.5984494224472987,
      "grad_norm": 0.5839673827473969,
      "learning_rate": 0.000164333536518249,
      "loss": 12.4692,
      "step": 10990
    },
    {
      "epoch": 0.5985038764438817,
      "grad_norm": 0.6179508354992466,
      "learning_rate": 0.00016432678514646392,
      "loss": 12.321,
      "step": 10991
    },
    {
      "epoch": 0.5985583304404648,
      "grad_norm": 0.5806189956726392,
      "learning_rate": 0.000164320033274463,
      "loss": 12.4132,
      "step": 10992
    },
    {
      "epoch": 0.5986127844370478,
      "grad_norm": 0.6172660230088375,
      "learning_rate": 0.0001643132809022987,
      "loss": 12.3561,
      "step": 10993
    },
    {
      "epoch": 0.5986672384336308,
      "grad_norm": 0.574460998452746,
      "learning_rate": 0.0001643065280300235,
      "loss": 12.3413,
      "step": 10994
    },
    {
      "epoch": 0.5987216924302138,
      "grad_norm": 0.5811058668708309,
      "learning_rate": 0.00016429977465769,
      "loss": 12.3356,
      "step": 10995
    },
    {
      "epoch": 0.5987761464267968,
      "grad_norm": 0.61660924334426,
      "learning_rate": 0.00016429302078535066,
      "loss": 12.1396,
      "step": 10996
    },
    {
      "epoch": 0.5988306004233799,
      "grad_norm": 0.5950546465610835,
      "learning_rate": 0.000164286266413058,
      "loss": 12.2852,
      "step": 10997
    },
    {
      "epoch": 0.5988850544199629,
      "grad_norm": 0.576631418146556,
      "learning_rate": 0.00016427951154086455,
      "loss": 12.3239,
      "step": 10998
    },
    {
      "epoch": 0.5989395084165459,
      "grad_norm": 0.5844309618486992,
      "learning_rate": 0.00016427275616882287,
      "loss": 12.3428,
      "step": 10999
    },
    {
      "epoch": 0.5989939624131289,
      "grad_norm": 0.6278048459137753,
      "learning_rate": 0.00016426600029698543,
      "loss": 12.3291,
      "step": 11000
    },
    {
      "epoch": 0.5990484164097118,
      "grad_norm": 0.5475127016067086,
      "learning_rate": 0.0001642592439254048,
      "loss": 12.2299,
      "step": 11001
    },
    {
      "epoch": 0.5991028704062948,
      "grad_norm": 0.6725176166321495,
      "learning_rate": 0.0001642524870541335,
      "loss": 12.2709,
      "step": 11002
    },
    {
      "epoch": 0.5991573244028779,
      "grad_norm": 0.6494389238030606,
      "learning_rate": 0.00016424572968322412,
      "loss": 12.3964,
      "step": 11003
    },
    {
      "epoch": 0.5992117783994609,
      "grad_norm": 0.6780755193103549,
      "learning_rate": 0.00016423897181272915,
      "loss": 12.3481,
      "step": 11004
    },
    {
      "epoch": 0.5992662323960439,
      "grad_norm": 0.5891529447522418,
      "learning_rate": 0.00016423221344270114,
      "loss": 12.4465,
      "step": 11005
    },
    {
      "epoch": 0.5993206863926269,
      "grad_norm": 0.6440257797821093,
      "learning_rate": 0.0001642254545731927,
      "loss": 12.4303,
      "step": 11006
    },
    {
      "epoch": 0.5993751403892099,
      "grad_norm": 0.6291431127412522,
      "learning_rate": 0.00016421869520425632,
      "loss": 12.3692,
      "step": 11007
    },
    {
      "epoch": 0.5994295943857929,
      "grad_norm": 0.5696472824113517,
      "learning_rate": 0.00016421193533594465,
      "loss": 12.2237,
      "step": 11008
    },
    {
      "epoch": 0.599484048382376,
      "grad_norm": 0.6722822267146954,
      "learning_rate": 0.00016420517496831016,
      "loss": 12.3112,
      "step": 11009
    },
    {
      "epoch": 0.599538502378959,
      "grad_norm": 0.6023223915088669,
      "learning_rate": 0.00016419841410140546,
      "loss": 12.2093,
      "step": 11010
    },
    {
      "epoch": 0.599592956375542,
      "grad_norm": 0.6379524481212994,
      "learning_rate": 0.00016419165273528317,
      "loss": 12.233,
      "step": 11011
    },
    {
      "epoch": 0.599647410372125,
      "grad_norm": 0.5829972476929912,
      "learning_rate": 0.00016418489086999577,
      "loss": 12.3742,
      "step": 11012
    },
    {
      "epoch": 0.599701864368708,
      "grad_norm": 0.6268520971080569,
      "learning_rate": 0.00016417812850559593,
      "loss": 12.3992,
      "step": 11013
    },
    {
      "epoch": 0.599756318365291,
      "grad_norm": 0.6059742211114041,
      "learning_rate": 0.0001641713656421362,
      "loss": 12.2391,
      "step": 11014
    },
    {
      "epoch": 0.5998107723618741,
      "grad_norm": 0.5977241627542447,
      "learning_rate": 0.00016416460227966915,
      "loss": 12.3261,
      "step": 11015
    },
    {
      "epoch": 0.5998652263584571,
      "grad_norm": 0.6257018673855346,
      "learning_rate": 0.00016415783841824738,
      "loss": 12.3806,
      "step": 11016
    },
    {
      "epoch": 0.5999196803550401,
      "grad_norm": 0.7731443352708557,
      "learning_rate": 0.00016415107405792352,
      "loss": 12.4871,
      "step": 11017
    },
    {
      "epoch": 0.5999741343516231,
      "grad_norm": 0.5933405099016545,
      "learning_rate": 0.0001641443091987502,
      "loss": 12.2664,
      "step": 11018
    },
    {
      "epoch": 0.600028588348206,
      "grad_norm": 0.6599217013753389,
      "learning_rate": 0.0001641375438407799,
      "loss": 12.399,
      "step": 11019
    },
    {
      "epoch": 0.600083042344789,
      "grad_norm": 0.732712892593169,
      "learning_rate": 0.00016413077798406534,
      "loss": 12.4574,
      "step": 11020
    },
    {
      "epoch": 0.6001374963413721,
      "grad_norm": 0.618903863653971,
      "learning_rate": 0.00016412401162865906,
      "loss": 12.3222,
      "step": 11021
    },
    {
      "epoch": 0.6001919503379551,
      "grad_norm": 0.5803114091083403,
      "learning_rate": 0.00016411724477461372,
      "loss": 12.2771,
      "step": 11022
    },
    {
      "epoch": 0.6002464043345381,
      "grad_norm": 0.6900362551315334,
      "learning_rate": 0.00016411047742198197,
      "loss": 12.3792,
      "step": 11023
    },
    {
      "epoch": 0.6003008583311211,
      "grad_norm": 0.6273452602392071,
      "learning_rate": 0.00016410370957081638,
      "loss": 12.1368,
      "step": 11024
    },
    {
      "epoch": 0.6003553123277041,
      "grad_norm": 0.6024275706084519,
      "learning_rate": 0.00016409694122116958,
      "loss": 12.4001,
      "step": 11025
    },
    {
      "epoch": 0.6004097663242871,
      "grad_norm": 0.5853978343141101,
      "learning_rate": 0.00016409017237309424,
      "loss": 12.2973,
      "step": 11026
    },
    {
      "epoch": 0.6004642203208702,
      "grad_norm": 0.6306740379428283,
      "learning_rate": 0.00016408340302664294,
      "loss": 12.4726,
      "step": 11027
    },
    {
      "epoch": 0.6005186743174532,
      "grad_norm": 0.582678644761307,
      "learning_rate": 0.00016407663318186838,
      "loss": 12.3409,
      "step": 11028
    },
    {
      "epoch": 0.6005731283140362,
      "grad_norm": 0.6438980879627172,
      "learning_rate": 0.00016406986283882315,
      "loss": 12.4626,
      "step": 11029
    },
    {
      "epoch": 0.6006275823106192,
      "grad_norm": 0.6516172706708898,
      "learning_rate": 0.00016406309199755992,
      "loss": 12.3803,
      "step": 11030
    },
    {
      "epoch": 0.6006820363072022,
      "grad_norm": 0.6589363085519371,
      "learning_rate": 0.00016405632065813136,
      "loss": 12.3313,
      "step": 11031
    },
    {
      "epoch": 0.6007364903037853,
      "grad_norm": 0.6892402550538832,
      "learning_rate": 0.00016404954882059012,
      "loss": 12.4779,
      "step": 11032
    },
    {
      "epoch": 0.6007909443003683,
      "grad_norm": 0.5614826899245277,
      "learning_rate": 0.00016404277648498881,
      "loss": 12.2103,
      "step": 11033
    },
    {
      "epoch": 0.6008453982969513,
      "grad_norm": 0.5673320871722737,
      "learning_rate": 0.00016403600365138017,
      "loss": 12.3491,
      "step": 11034
    },
    {
      "epoch": 0.6008998522935343,
      "grad_norm": 0.6451029222343776,
      "learning_rate": 0.00016402923031981682,
      "loss": 12.3691,
      "step": 11035
    },
    {
      "epoch": 0.6009543062901173,
      "grad_norm": 0.5993129509634276,
      "learning_rate": 0.00016402245649035143,
      "loss": 12.3083,
      "step": 11036
    },
    {
      "epoch": 0.6010087602867002,
      "grad_norm": 0.8425288989468626,
      "learning_rate": 0.00016401568216303666,
      "loss": 12.2698,
      "step": 11037
    },
    {
      "epoch": 0.6010632142832834,
      "grad_norm": 0.6080488327074888,
      "learning_rate": 0.00016400890733792522,
      "loss": 12.3828,
      "step": 11038
    },
    {
      "epoch": 0.6011176682798663,
      "grad_norm": 0.6149323405989047,
      "learning_rate": 0.0001640021320150698,
      "loss": 12.29,
      "step": 11039
    },
    {
      "epoch": 0.6011721222764493,
      "grad_norm": 0.6503595167012923,
      "learning_rate": 0.00016399535619452307,
      "loss": 12.4554,
      "step": 11040
    },
    {
      "epoch": 0.6012265762730323,
      "grad_norm": 0.6144374874038958,
      "learning_rate": 0.00016398857987633768,
      "loss": 12.3855,
      "step": 11041
    },
    {
      "epoch": 0.6012810302696153,
      "grad_norm": 0.5659336071137602,
      "learning_rate": 0.0001639818030605664,
      "loss": 12.3596,
      "step": 11042
    },
    {
      "epoch": 0.6013354842661983,
      "grad_norm": 0.5376304352006531,
      "learning_rate": 0.00016397502574726187,
      "loss": 12.2335,
      "step": 11043
    },
    {
      "epoch": 0.6013899382627814,
      "grad_norm": 0.5924159115387776,
      "learning_rate": 0.00016396824793647684,
      "loss": 12.2906,
      "step": 11044
    },
    {
      "epoch": 0.6014443922593644,
      "grad_norm": 0.6760554560177521,
      "learning_rate": 0.00016396146962826394,
      "loss": 12.3811,
      "step": 11045
    },
    {
      "epoch": 0.6014988462559474,
      "grad_norm": 0.6041717706336399,
      "learning_rate": 0.00016395469082267598,
      "loss": 12.4151,
      "step": 11046
    },
    {
      "epoch": 0.6015533002525304,
      "grad_norm": 0.6416209500189765,
      "learning_rate": 0.00016394791151976557,
      "loss": 12.458,
      "step": 11047
    },
    {
      "epoch": 0.6016077542491134,
      "grad_norm": 0.622185400302005,
      "learning_rate": 0.0001639411317195855,
      "loss": 12.3202,
      "step": 11048
    },
    {
      "epoch": 0.6016622082456964,
      "grad_norm": 0.6290605003665506,
      "learning_rate": 0.0001639343514221885,
      "loss": 12.4226,
      "step": 11049
    },
    {
      "epoch": 0.6017166622422795,
      "grad_norm": 0.6200775111861807,
      "learning_rate": 0.00016392757062762723,
      "loss": 12.3675,
      "step": 11050
    },
    {
      "epoch": 0.6017711162388625,
      "grad_norm": 0.6044956149656243,
      "learning_rate": 0.00016392078933595442,
      "loss": 12.4052,
      "step": 11051
    },
    {
      "epoch": 0.6018255702354455,
      "grad_norm": 0.5622836150915892,
      "learning_rate": 0.00016391400754722286,
      "loss": 12.2558,
      "step": 11052
    },
    {
      "epoch": 0.6018800242320285,
      "grad_norm": 0.7140968733839776,
      "learning_rate": 0.00016390722526148525,
      "loss": 12.4013,
      "step": 11053
    },
    {
      "epoch": 0.6019344782286115,
      "grad_norm": 0.6898733093538983,
      "learning_rate": 0.00016390044247879435,
      "loss": 12.3683,
      "step": 11054
    },
    {
      "epoch": 0.6019889322251945,
      "grad_norm": 0.6503239657646119,
      "learning_rate": 0.0001638936591992029,
      "loss": 12.3846,
      "step": 11055
    },
    {
      "epoch": 0.6020433862217776,
      "grad_norm": 0.5894666821242075,
      "learning_rate": 0.00016388687542276363,
      "loss": 12.3637,
      "step": 11056
    },
    {
      "epoch": 0.6020978402183605,
      "grad_norm": 0.5934108932985701,
      "learning_rate": 0.00016388009114952929,
      "loss": 12.3284,
      "step": 11057
    },
    {
      "epoch": 0.6021522942149435,
      "grad_norm": 0.7523011319008442,
      "learning_rate": 0.00016387330637955265,
      "loss": 12.2661,
      "step": 11058
    },
    {
      "epoch": 0.6022067482115265,
      "grad_norm": 0.647101465243184,
      "learning_rate": 0.00016386652111288652,
      "loss": 12.4377,
      "step": 11059
    },
    {
      "epoch": 0.6022612022081095,
      "grad_norm": 0.5512134794367408,
      "learning_rate": 0.00016385973534958356,
      "loss": 12.3801,
      "step": 11060
    },
    {
      "epoch": 0.6023156562046925,
      "grad_norm": 0.5902503412478513,
      "learning_rate": 0.0001638529490896966,
      "loss": 12.3326,
      "step": 11061
    },
    {
      "epoch": 0.6023701102012756,
      "grad_norm": 0.633910662052685,
      "learning_rate": 0.00016384616233327837,
      "loss": 12.4507,
      "step": 11062
    },
    {
      "epoch": 0.6024245641978586,
      "grad_norm": 0.5754040976327168,
      "learning_rate": 0.00016383937508038173,
      "loss": 12.3825,
      "step": 11063
    },
    {
      "epoch": 0.6024790181944416,
      "grad_norm": 0.6043470862631053,
      "learning_rate": 0.00016383258733105937,
      "loss": 12.3094,
      "step": 11064
    },
    {
      "epoch": 0.6025334721910246,
      "grad_norm": 0.6048248874531785,
      "learning_rate": 0.00016382579908536413,
      "loss": 12.2771,
      "step": 11065
    },
    {
      "epoch": 0.6025879261876076,
      "grad_norm": 0.5452843984000415,
      "learning_rate": 0.00016381901034334873,
      "loss": 12.2863,
      "step": 11066
    },
    {
      "epoch": 0.6026423801841907,
      "grad_norm": 0.6076571075716745,
      "learning_rate": 0.00016381222110506603,
      "loss": 12.3473,
      "step": 11067
    },
    {
      "epoch": 0.6026968341807737,
      "grad_norm": 0.5652508244979979,
      "learning_rate": 0.00016380543137056882,
      "loss": 12.4141,
      "step": 11068
    },
    {
      "epoch": 0.6027512881773567,
      "grad_norm": 0.585225819813063,
      "learning_rate": 0.00016379864113990985,
      "loss": 12.305,
      "step": 11069
    },
    {
      "epoch": 0.6028057421739397,
      "grad_norm": 0.5795408250187182,
      "learning_rate": 0.00016379185041314194,
      "loss": 12.2756,
      "step": 11070
    },
    {
      "epoch": 0.6028601961705227,
      "grad_norm": 0.5995362149946898,
      "learning_rate": 0.00016378505919031794,
      "loss": 12.3073,
      "step": 11071
    },
    {
      "epoch": 0.6029146501671057,
      "grad_norm": 0.5508610806139238,
      "learning_rate": 0.0001637782674714906,
      "loss": 12.174,
      "step": 11072
    },
    {
      "epoch": 0.6029691041636888,
      "grad_norm": 0.5783209754285752,
      "learning_rate": 0.00016377147525671273,
      "loss": 12.3101,
      "step": 11073
    },
    {
      "epoch": 0.6030235581602718,
      "grad_norm": 0.5626050721429561,
      "learning_rate": 0.0001637646825460372,
      "loss": 12.2853,
      "step": 11074
    },
    {
      "epoch": 0.6030780121568547,
      "grad_norm": 0.6027064270819751,
      "learning_rate": 0.00016375788933951682,
      "loss": 12.3111,
      "step": 11075
    },
    {
      "epoch": 0.6031324661534377,
      "grad_norm": 0.6690295265995301,
      "learning_rate": 0.00016375109563720436,
      "loss": 12.3696,
      "step": 11076
    },
    {
      "epoch": 0.6031869201500207,
      "grad_norm": 0.5837871491600592,
      "learning_rate": 0.0001637443014391527,
      "loss": 12.2745,
      "step": 11077
    },
    {
      "epoch": 0.6032413741466037,
      "grad_norm": 0.5733628207555487,
      "learning_rate": 0.00016373750674541468,
      "loss": 12.2662,
      "step": 11078
    },
    {
      "epoch": 0.6032958281431868,
      "grad_norm": 0.6237040428548752,
      "learning_rate": 0.00016373071155604312,
      "loss": 12.3948,
      "step": 11079
    },
    {
      "epoch": 0.6033502821397698,
      "grad_norm": 0.6324278402150281,
      "learning_rate": 0.00016372391587109086,
      "loss": 12.4646,
      "step": 11080
    },
    {
      "epoch": 0.6034047361363528,
      "grad_norm": 0.5900512368765062,
      "learning_rate": 0.0001637171196906107,
      "loss": 12.293,
      "step": 11081
    },
    {
      "epoch": 0.6034591901329358,
      "grad_norm": 0.6510319242318467,
      "learning_rate": 0.0001637103230146556,
      "loss": 12.3156,
      "step": 11082
    },
    {
      "epoch": 0.6035136441295188,
      "grad_norm": 0.603809943286475,
      "learning_rate": 0.0001637035258432783,
      "loss": 12.425,
      "step": 11083
    },
    {
      "epoch": 0.6035680981261018,
      "grad_norm": 0.5907690636784768,
      "learning_rate": 0.00016369672817653173,
      "loss": 12.3971,
      "step": 11084
    },
    {
      "epoch": 0.6036225521226849,
      "grad_norm": 0.6478519079380884,
      "learning_rate": 0.0001636899300144687,
      "loss": 12.3261,
      "step": 11085
    },
    {
      "epoch": 0.6036770061192679,
      "grad_norm": 0.5575314006519871,
      "learning_rate": 0.0001636831313571421,
      "loss": 12.3727,
      "step": 11086
    },
    {
      "epoch": 0.6037314601158509,
      "grad_norm": 0.6231567452468517,
      "learning_rate": 0.00016367633220460478,
      "loss": 12.4183,
      "step": 11087
    },
    {
      "epoch": 0.6037859141124339,
      "grad_norm": 0.5987386128539546,
      "learning_rate": 0.00016366953255690962,
      "loss": 12.3108,
      "step": 11088
    },
    {
      "epoch": 0.6038403681090169,
      "grad_norm": 0.5557211914285582,
      "learning_rate": 0.00016366273241410952,
      "loss": 12.1499,
      "step": 11089
    },
    {
      "epoch": 0.6038948221055999,
      "grad_norm": 0.676169499379147,
      "learning_rate": 0.0001636559317762573,
      "loss": 12.4032,
      "step": 11090
    },
    {
      "epoch": 0.603949276102183,
      "grad_norm": 0.5760384995265102,
      "learning_rate": 0.0001636491306434059,
      "loss": 12.2913,
      "step": 11091
    },
    {
      "epoch": 0.604003730098766,
      "grad_norm": 0.5933103287988173,
      "learning_rate": 0.0001636423290156082,
      "loss": 12.428,
      "step": 11092
    },
    {
      "epoch": 0.604058184095349,
      "grad_norm": 0.7065631146403468,
      "learning_rate": 0.00016363552689291705,
      "loss": 12.2464,
      "step": 11093
    },
    {
      "epoch": 0.6041126380919319,
      "grad_norm": 0.610605509954838,
      "learning_rate": 0.0001636287242753854,
      "loss": 12.333,
      "step": 11094
    },
    {
      "epoch": 0.6041670920885149,
      "grad_norm": 0.5975864259914512,
      "learning_rate": 0.00016362192116306612,
      "loss": 12.2726,
      "step": 11095
    },
    {
      "epoch": 0.604221546085098,
      "grad_norm": 0.6099596482197528,
      "learning_rate": 0.00016361511755601206,
      "loss": 12.3904,
      "step": 11096
    },
    {
      "epoch": 0.604276000081681,
      "grad_norm": 0.5486189081414413,
      "learning_rate": 0.00016360831345427622,
      "loss": 12.3003,
      "step": 11097
    },
    {
      "epoch": 0.604330454078264,
      "grad_norm": 0.6590723996025494,
      "learning_rate": 0.0001636015088579115,
      "loss": 12.3686,
      "step": 11098
    },
    {
      "epoch": 0.604384908074847,
      "grad_norm": 0.5965197006105573,
      "learning_rate": 0.00016359470376697073,
      "loss": 12.3768,
      "step": 11099
    },
    {
      "epoch": 0.60443936207143,
      "grad_norm": 0.6000795522191682,
      "learning_rate": 0.00016358789818150688,
      "loss": 12.3006,
      "step": 11100
    },
    {
      "epoch": 0.604493816068013,
      "grad_norm": 0.5966457566072743,
      "learning_rate": 0.0001635810921015729,
      "loss": 12.3902,
      "step": 11101
    },
    {
      "epoch": 0.6045482700645961,
      "grad_norm": 0.5723338160767066,
      "learning_rate": 0.00016357428552722165,
      "loss": 12.2545,
      "step": 11102
    },
    {
      "epoch": 0.6046027240611791,
      "grad_norm": 0.5917006614386284,
      "learning_rate": 0.0001635674784585061,
      "loss": 12.3173,
      "step": 11103
    },
    {
      "epoch": 0.6046571780577621,
      "grad_norm": 0.6300577844312848,
      "learning_rate": 0.00016356067089547919,
      "loss": 12.4326,
      "step": 11104
    },
    {
      "epoch": 0.6047116320543451,
      "grad_norm": 0.5941836003594154,
      "learning_rate": 0.00016355386283819386,
      "loss": 12.3251,
      "step": 11105
    },
    {
      "epoch": 0.6047660860509281,
      "grad_norm": 0.6859987389573252,
      "learning_rate": 0.000163547054286703,
      "loss": 12.3612,
      "step": 11106
    },
    {
      "epoch": 0.6048205400475111,
      "grad_norm": 0.7555113851041918,
      "learning_rate": 0.0001635402452410596,
      "loss": 12.3068,
      "step": 11107
    },
    {
      "epoch": 0.6048749940440942,
      "grad_norm": 0.7143705006050696,
      "learning_rate": 0.0001635334357013166,
      "loss": 12.3739,
      "step": 11108
    },
    {
      "epoch": 0.6049294480406772,
      "grad_norm": 0.6897982973385011,
      "learning_rate": 0.00016352662566752698,
      "loss": 12.384,
      "step": 11109
    },
    {
      "epoch": 0.6049839020372602,
      "grad_norm": 0.6588045553449117,
      "learning_rate": 0.00016351981513974365,
      "loss": 12.4431,
      "step": 11110
    },
    {
      "epoch": 0.6050383560338431,
      "grad_norm": 0.5990230565766719,
      "learning_rate": 0.00016351300411801954,
      "loss": 12.3734,
      "step": 11111
    },
    {
      "epoch": 0.6050928100304261,
      "grad_norm": 0.6978925280375029,
      "learning_rate": 0.00016350619260240769,
      "loss": 12.3558,
      "step": 11112
    },
    {
      "epoch": 0.6051472640270091,
      "grad_norm": 0.5959688891627724,
      "learning_rate": 0.00016349938059296104,
      "loss": 12.2814,
      "step": 11113
    },
    {
      "epoch": 0.6052017180235922,
      "grad_norm": 0.6048401762930573,
      "learning_rate": 0.00016349256808973256,
      "loss": 12.2942,
      "step": 11114
    },
    {
      "epoch": 0.6052561720201752,
      "grad_norm": 0.6237154293696561,
      "learning_rate": 0.00016348575509277522,
      "loss": 12.3011,
      "step": 11115
    },
    {
      "epoch": 0.6053106260167582,
      "grad_norm": 0.5894504081881295,
      "learning_rate": 0.000163478941602142,
      "loss": 12.4032,
      "step": 11116
    },
    {
      "epoch": 0.6053650800133412,
      "grad_norm": 0.6673670271952441,
      "learning_rate": 0.0001634721276178859,
      "loss": 12.3395,
      "step": 11117
    },
    {
      "epoch": 0.6054195340099242,
      "grad_norm": 0.5951869230383666,
      "learning_rate": 0.00016346531314005987,
      "loss": 12.3577,
      "step": 11118
    },
    {
      "epoch": 0.6054739880065072,
      "grad_norm": 0.6334316531072363,
      "learning_rate": 0.00016345849816871692,
      "loss": 12.3382,
      "step": 11119
    },
    {
      "epoch": 0.6055284420030903,
      "grad_norm": 0.6162645161211552,
      "learning_rate": 0.00016345168270391004,
      "loss": 12.2919,
      "step": 11120
    },
    {
      "epoch": 0.6055828959996733,
      "grad_norm": 0.5205672866936765,
      "learning_rate": 0.00016344486674569226,
      "loss": 12.327,
      "step": 11121
    },
    {
      "epoch": 0.6056373499962563,
      "grad_norm": 0.5930119109328588,
      "learning_rate": 0.00016343805029411654,
      "loss": 12.3136,
      "step": 11122
    },
    {
      "epoch": 0.6056918039928393,
      "grad_norm": 0.5917678881795895,
      "learning_rate": 0.00016343123334923592,
      "loss": 12.3358,
      "step": 11123
    },
    {
      "epoch": 0.6057462579894223,
      "grad_norm": 0.6278136764825851,
      "learning_rate": 0.0001634244159111034,
      "loss": 12.4578,
      "step": 11124
    },
    {
      "epoch": 0.6058007119860053,
      "grad_norm": 0.5941650915734099,
      "learning_rate": 0.00016341759797977194,
      "loss": 12.4215,
      "step": 11125
    },
    {
      "epoch": 0.6058551659825884,
      "grad_norm": 0.6487586325937854,
      "learning_rate": 0.00016341077955529465,
      "loss": 12.4542,
      "step": 11126
    },
    {
      "epoch": 0.6059096199791714,
      "grad_norm": 0.6134015359441436,
      "learning_rate": 0.00016340396063772449,
      "loss": 12.3903,
      "step": 11127
    },
    {
      "epoch": 0.6059640739757544,
      "grad_norm": 0.6836090618022848,
      "learning_rate": 0.00016339714122711447,
      "loss": 12.36,
      "step": 11128
    },
    {
      "epoch": 0.6060185279723374,
      "grad_norm": 0.6845856838845686,
      "learning_rate": 0.00016339032132351767,
      "loss": 12.2901,
      "step": 11129
    },
    {
      "epoch": 0.6060729819689203,
      "grad_norm": 0.5640102141813625,
      "learning_rate": 0.0001633835009269871,
      "loss": 12.3379,
      "step": 11130
    },
    {
      "epoch": 0.6061274359655034,
      "grad_norm": 0.5737426198031063,
      "learning_rate": 0.0001633766800375758,
      "loss": 12.3716,
      "step": 11131
    },
    {
      "epoch": 0.6061818899620864,
      "grad_norm": 0.5824786880831666,
      "learning_rate": 0.00016336985865533682,
      "loss": 12.4523,
      "step": 11132
    },
    {
      "epoch": 0.6062363439586694,
      "grad_norm": 0.5329751071521219,
      "learning_rate": 0.00016336303678032317,
      "loss": 12.1134,
      "step": 11133
    },
    {
      "epoch": 0.6062907979552524,
      "grad_norm": 0.6714939419589115,
      "learning_rate": 0.00016335621441258792,
      "loss": 12.3629,
      "step": 11134
    },
    {
      "epoch": 0.6063452519518354,
      "grad_norm": 0.7119477262638965,
      "learning_rate": 0.00016334939155218415,
      "loss": 12.4187,
      "step": 11135
    },
    {
      "epoch": 0.6063997059484184,
      "grad_norm": 0.6086057695680839,
      "learning_rate": 0.00016334256819916485,
      "loss": 12.368,
      "step": 11136
    },
    {
      "epoch": 0.6064541599450015,
      "grad_norm": 0.6078255268695479,
      "learning_rate": 0.00016333574435358313,
      "loss": 12.4414,
      "step": 11137
    },
    {
      "epoch": 0.6065086139415845,
      "grad_norm": 0.6020225475285034,
      "learning_rate": 0.00016332892001549206,
      "loss": 12.2573,
      "step": 11138
    },
    {
      "epoch": 0.6065630679381675,
      "grad_norm": 0.7082157627835164,
      "learning_rate": 0.00016332209518494468,
      "loss": 12.4737,
      "step": 11139
    },
    {
      "epoch": 0.6066175219347505,
      "grad_norm": 0.5873016676995043,
      "learning_rate": 0.00016331526986199406,
      "loss": 12.4236,
      "step": 11140
    },
    {
      "epoch": 0.6066719759313335,
      "grad_norm": 0.631074673123381,
      "learning_rate": 0.00016330844404669327,
      "loss": 12.4155,
      "step": 11141
    },
    {
      "epoch": 0.6067264299279165,
      "grad_norm": 0.5926052711170497,
      "learning_rate": 0.00016330161773909542,
      "loss": 12.3943,
      "step": 11142
    },
    {
      "epoch": 0.6067808839244996,
      "grad_norm": 0.7595425087844859,
      "learning_rate": 0.00016329479093925357,
      "loss": 12.4255,
      "step": 11143
    },
    {
      "epoch": 0.6068353379210826,
      "grad_norm": 0.6165882148007943,
      "learning_rate": 0.00016328796364722083,
      "loss": 12.3621,
      "step": 11144
    },
    {
      "epoch": 0.6068897919176656,
      "grad_norm": 0.5348979092332994,
      "learning_rate": 0.00016328113586305026,
      "loss": 12.3285,
      "step": 11145
    },
    {
      "epoch": 0.6069442459142486,
      "grad_norm": 0.5812507187291274,
      "learning_rate": 0.00016327430758679494,
      "loss": 12.3238,
      "step": 11146
    },
    {
      "epoch": 0.6069986999108316,
      "grad_norm": 0.57940112965112,
      "learning_rate": 0.00016326747881850802,
      "loss": 12.3121,
      "step": 11147
    },
    {
      "epoch": 0.6070531539074145,
      "grad_norm": 0.5617843893592939,
      "learning_rate": 0.00016326064955824257,
      "loss": 12.3478,
      "step": 11148
    },
    {
      "epoch": 0.6071076079039976,
      "grad_norm": 0.5391895305756338,
      "learning_rate": 0.0001632538198060517,
      "loss": 12.3156,
      "step": 11149
    },
    {
      "epoch": 0.6071620619005806,
      "grad_norm": 0.5712370867105543,
      "learning_rate": 0.0001632469895619885,
      "loss": 12.2493,
      "step": 11150
    },
    {
      "epoch": 0.6072165158971636,
      "grad_norm": 0.613671361377228,
      "learning_rate": 0.00016324015882610615,
      "loss": 12.3102,
      "step": 11151
    },
    {
      "epoch": 0.6072709698937466,
      "grad_norm": 0.6108727049954086,
      "learning_rate": 0.00016323332759845765,
      "loss": 12.3195,
      "step": 11152
    },
    {
      "epoch": 0.6073254238903296,
      "grad_norm": 0.5904626345424183,
      "learning_rate": 0.00016322649587909623,
      "loss": 12.3891,
      "step": 11153
    },
    {
      "epoch": 0.6073798778869126,
      "grad_norm": 0.5562410258535025,
      "learning_rate": 0.000163219663668075,
      "loss": 12.258,
      "step": 11154
    },
    {
      "epoch": 0.6074343318834957,
      "grad_norm": 0.5573520133764401,
      "learning_rate": 0.00016321283096544704,
      "loss": 12.3414,
      "step": 11155
    },
    {
      "epoch": 0.6074887858800787,
      "grad_norm": 0.6867050860705599,
      "learning_rate": 0.00016320599777126552,
      "loss": 12.3879,
      "step": 11156
    },
    {
      "epoch": 0.6075432398766617,
      "grad_norm": 0.6148388888394638,
      "learning_rate": 0.00016319916408558352,
      "loss": 12.2203,
      "step": 11157
    },
    {
      "epoch": 0.6075976938732447,
      "grad_norm": 0.6316122300461814,
      "learning_rate": 0.00016319232990845425,
      "loss": 12.3192,
      "step": 11158
    },
    {
      "epoch": 0.6076521478698277,
      "grad_norm": 0.5381826224839812,
      "learning_rate": 0.00016318549523993084,
      "loss": 12.1295,
      "step": 11159
    },
    {
      "epoch": 0.6077066018664107,
      "grad_norm": 0.639597752296276,
      "learning_rate": 0.00016317866008006639,
      "loss": 12.3341,
      "step": 11160
    },
    {
      "epoch": 0.6077610558629938,
      "grad_norm": 0.5850046410800002,
      "learning_rate": 0.0001631718244289141,
      "loss": 12.3696,
      "step": 11161
    },
    {
      "epoch": 0.6078155098595768,
      "grad_norm": 0.5409904898717542,
      "learning_rate": 0.00016316498828652712,
      "loss": 12.2983,
      "step": 11162
    },
    {
      "epoch": 0.6078699638561598,
      "grad_norm": 0.6560920448327437,
      "learning_rate": 0.00016315815165295855,
      "loss": 12.3445,
      "step": 11163
    },
    {
      "epoch": 0.6079244178527428,
      "grad_norm": 0.5881682714143308,
      "learning_rate": 0.00016315131452826167,
      "loss": 12.2529,
      "step": 11164
    },
    {
      "epoch": 0.6079788718493258,
      "grad_norm": 0.6556078342042584,
      "learning_rate": 0.00016314447691248956,
      "loss": 12.4407,
      "step": 11165
    },
    {
      "epoch": 0.6080333258459089,
      "grad_norm": 0.5998152193241213,
      "learning_rate": 0.00016313763880569537,
      "loss": 12.1111,
      "step": 11166
    },
    {
      "epoch": 0.6080877798424918,
      "grad_norm": 0.6040623193833304,
      "learning_rate": 0.00016313080020793235,
      "loss": 12.2528,
      "step": 11167
    },
    {
      "epoch": 0.6081422338390748,
      "grad_norm": 0.565811403827649,
      "learning_rate": 0.00016312396111925362,
      "loss": 12.3451,
      "step": 11168
    },
    {
      "epoch": 0.6081966878356578,
      "grad_norm": 0.6303795224857877,
      "learning_rate": 0.00016311712153971238,
      "loss": 12.3896,
      "step": 11169
    },
    {
      "epoch": 0.6082511418322408,
      "grad_norm": 0.6103204271982208,
      "learning_rate": 0.00016311028146936184,
      "loss": 12.3071,
      "step": 11170
    },
    {
      "epoch": 0.6083055958288238,
      "grad_norm": 0.5907392547073242,
      "learning_rate": 0.00016310344090825516,
      "loss": 12.3801,
      "step": 11171
    },
    {
      "epoch": 0.6083600498254069,
      "grad_norm": 0.5898616925852171,
      "learning_rate": 0.00016309659985644555,
      "loss": 12.2173,
      "step": 11172
    },
    {
      "epoch": 0.6084145038219899,
      "grad_norm": 0.6721267658389909,
      "learning_rate": 0.00016308975831398617,
      "loss": 12.4195,
      "step": 11173
    },
    {
      "epoch": 0.6084689578185729,
      "grad_norm": 0.5737286379187838,
      "learning_rate": 0.00016308291628093025,
      "loss": 12.2979,
      "step": 11174
    },
    {
      "epoch": 0.6085234118151559,
      "grad_norm": 0.5948908660886684,
      "learning_rate": 0.00016307607375733103,
      "loss": 12.4614,
      "step": 11175
    },
    {
      "epoch": 0.6085778658117389,
      "grad_norm": 0.5847183241982178,
      "learning_rate": 0.00016306923074324166,
      "loss": 12.1829,
      "step": 11176
    },
    {
      "epoch": 0.6086323198083219,
      "grad_norm": 0.5348038205117897,
      "learning_rate": 0.00016306238723871536,
      "loss": 12.2261,
      "step": 11177
    },
    {
      "epoch": 0.608686773804905,
      "grad_norm": 0.7186846714099959,
      "learning_rate": 0.00016305554324380536,
      "loss": 12.3023,
      "step": 11178
    },
    {
      "epoch": 0.608741227801488,
      "grad_norm": 0.656779753977201,
      "learning_rate": 0.00016304869875856493,
      "loss": 12.4631,
      "step": 11179
    },
    {
      "epoch": 0.608795681798071,
      "grad_norm": 0.6612372166326377,
      "learning_rate": 0.0001630418537830472,
      "loss": 12.3458,
      "step": 11180
    },
    {
      "epoch": 0.608850135794654,
      "grad_norm": 0.6116698754329289,
      "learning_rate": 0.00016303500831730546,
      "loss": 12.2474,
      "step": 11181
    },
    {
      "epoch": 0.608904589791237,
      "grad_norm": 0.6175693550944324,
      "learning_rate": 0.00016302816236139292,
      "loss": 12.3383,
      "step": 11182
    },
    {
      "epoch": 0.60895904378782,
      "grad_norm": 0.6838026604965294,
      "learning_rate": 0.0001630213159153628,
      "loss": 12.3036,
      "step": 11183
    },
    {
      "epoch": 0.6090134977844031,
      "grad_norm": 0.6209752444596405,
      "learning_rate": 0.0001630144689792684,
      "loss": 12.2806,
      "step": 11184
    },
    {
      "epoch": 0.609067951780986,
      "grad_norm": 0.6992091942059789,
      "learning_rate": 0.0001630076215531629,
      "loss": 12.3381,
      "step": 11185
    },
    {
      "epoch": 0.609122405777569,
      "grad_norm": 0.6464572432173553,
      "learning_rate": 0.0001630007736370996,
      "loss": 12.3463,
      "step": 11186
    },
    {
      "epoch": 0.609176859774152,
      "grad_norm": 0.6038931960005121,
      "learning_rate": 0.00016299392523113165,
      "loss": 12.2283,
      "step": 11187
    },
    {
      "epoch": 0.609231313770735,
      "grad_norm": 0.6692542463517684,
      "learning_rate": 0.00016298707633531244,
      "loss": 12.1545,
      "step": 11188
    },
    {
      "epoch": 0.609285767767318,
      "grad_norm": 0.5934411500079831,
      "learning_rate": 0.0001629802269496951,
      "loss": 12.279,
      "step": 11189
    },
    {
      "epoch": 0.6093402217639011,
      "grad_norm": 0.6520832855912216,
      "learning_rate": 0.000162973377074333,
      "loss": 12.4605,
      "step": 11190
    },
    {
      "epoch": 0.6093946757604841,
      "grad_norm": 0.6356145983590724,
      "learning_rate": 0.00016296652670927934,
      "loss": 12.3357,
      "step": 11191
    },
    {
      "epoch": 0.6094491297570671,
      "grad_norm": 0.6684675532985174,
      "learning_rate": 0.00016295967585458742,
      "loss": 12.3771,
      "step": 11192
    },
    {
      "epoch": 0.6095035837536501,
      "grad_norm": 0.5676216790776536,
      "learning_rate": 0.00016295282451031048,
      "loss": 12.313,
      "step": 11193
    },
    {
      "epoch": 0.6095580377502331,
      "grad_norm": 0.6318426817865253,
      "learning_rate": 0.00016294597267650185,
      "loss": 12.2556,
      "step": 11194
    },
    {
      "epoch": 0.6096124917468161,
      "grad_norm": 0.546346474512078,
      "learning_rate": 0.00016293912035321477,
      "loss": 12.3492,
      "step": 11195
    },
    {
      "epoch": 0.6096669457433992,
      "grad_norm": 0.5597620710316814,
      "learning_rate": 0.00016293226754050252,
      "loss": 12.2942,
      "step": 11196
    },
    {
      "epoch": 0.6097213997399822,
      "grad_norm": 0.6019576331614028,
      "learning_rate": 0.00016292541423841843,
      "loss": 12.4731,
      "step": 11197
    },
    {
      "epoch": 0.6097758537365652,
      "grad_norm": 0.6087530645396827,
      "learning_rate": 0.00016291856044701574,
      "loss": 12.414,
      "step": 11198
    },
    {
      "epoch": 0.6098303077331482,
      "grad_norm": 0.5511451580189743,
      "learning_rate": 0.0001629117061663478,
      "loss": 12.3977,
      "step": 11199
    },
    {
      "epoch": 0.6098847617297312,
      "grad_norm": 0.5673407108230473,
      "learning_rate": 0.00016290485139646788,
      "loss": 12.3911,
      "step": 11200
    },
    {
      "epoch": 0.6099392157263143,
      "grad_norm": 0.568847143885813,
      "learning_rate": 0.00016289799613742925,
      "loss": 12.2846,
      "step": 11201
    },
    {
      "epoch": 0.6099936697228973,
      "grad_norm": 0.7944651261831513,
      "learning_rate": 0.0001628911403892853,
      "loss": 12.4732,
      "step": 11202
    },
    {
      "epoch": 0.6100481237194803,
      "grad_norm": 0.646189949260018,
      "learning_rate": 0.00016288428415208925,
      "loss": 12.3322,
      "step": 11203
    },
    {
      "epoch": 0.6101025777160632,
      "grad_norm": 0.5732514537043685,
      "learning_rate": 0.0001628774274258945,
      "loss": 12.3178,
      "step": 11204
    },
    {
      "epoch": 0.6101570317126462,
      "grad_norm": 0.654862812790952,
      "learning_rate": 0.00016287057021075428,
      "loss": 12.3292,
      "step": 11205
    },
    {
      "epoch": 0.6102114857092292,
      "grad_norm": 0.6568052904002022,
      "learning_rate": 0.00016286371250672201,
      "loss": 12.4505,
      "step": 11206
    },
    {
      "epoch": 0.6102659397058123,
      "grad_norm": 0.6089929626588377,
      "learning_rate": 0.00016285685431385096,
      "loss": 12.2601,
      "step": 11207
    },
    {
      "epoch": 0.6103203937023953,
      "grad_norm": 0.6441744367395216,
      "learning_rate": 0.00016284999563219446,
      "loss": 12.2596,
      "step": 11208
    },
    {
      "epoch": 0.6103748476989783,
      "grad_norm": 0.5886980647496863,
      "learning_rate": 0.00016284313646180586,
      "loss": 12.4004,
      "step": 11209
    },
    {
      "epoch": 0.6104293016955613,
      "grad_norm": 0.6567455919150536,
      "learning_rate": 0.0001628362768027385,
      "loss": 12.4423,
      "step": 11210
    },
    {
      "epoch": 0.6104837556921443,
      "grad_norm": 0.6434126784891697,
      "learning_rate": 0.0001628294166550457,
      "loss": 12.3281,
      "step": 11211
    },
    {
      "epoch": 0.6105382096887273,
      "grad_norm": 0.5550889914663049,
      "learning_rate": 0.00016282255601878082,
      "loss": 12.2448,
      "step": 11212
    },
    {
      "epoch": 0.6105926636853104,
      "grad_norm": 0.658643938541344,
      "learning_rate": 0.0001628156948939972,
      "loss": 12.1808,
      "step": 11213
    },
    {
      "epoch": 0.6106471176818934,
      "grad_norm": 0.5953183504399878,
      "learning_rate": 0.00016280883328074824,
      "loss": 12.3536,
      "step": 11214
    },
    {
      "epoch": 0.6107015716784764,
      "grad_norm": 0.6197836107074511,
      "learning_rate": 0.00016280197117908723,
      "loss": 12.3475,
      "step": 11215
    },
    {
      "epoch": 0.6107560256750594,
      "grad_norm": 0.6536735721454394,
      "learning_rate": 0.00016279510858906756,
      "loss": 12.3812,
      "step": 11216
    },
    {
      "epoch": 0.6108104796716424,
      "grad_norm": 0.5736370991155088,
      "learning_rate": 0.00016278824551074262,
      "loss": 12.3155,
      "step": 11217
    },
    {
      "epoch": 0.6108649336682254,
      "grad_norm": 0.650879421311335,
      "learning_rate": 0.0001627813819441657,
      "loss": 12.3652,
      "step": 11218
    },
    {
      "epoch": 0.6109193876648085,
      "grad_norm": 0.640251354561717,
      "learning_rate": 0.00016277451788939028,
      "loss": 12.1254,
      "step": 11219
    },
    {
      "epoch": 0.6109738416613915,
      "grad_norm": 0.5670359796552973,
      "learning_rate": 0.00016276765334646967,
      "loss": 12.3489,
      "step": 11220
    },
    {
      "epoch": 0.6110282956579745,
      "grad_norm": 0.6384343722609144,
      "learning_rate": 0.00016276078831545726,
      "loss": 12.2302,
      "step": 11221
    },
    {
      "epoch": 0.6110827496545574,
      "grad_norm": 0.6596286383025379,
      "learning_rate": 0.00016275392279640642,
      "loss": 12.3865,
      "step": 11222
    },
    {
      "epoch": 0.6111372036511404,
      "grad_norm": 0.5891279826921046,
      "learning_rate": 0.00016274705678937057,
      "loss": 12.2326,
      "step": 11223
    },
    {
      "epoch": 0.6111916576477234,
      "grad_norm": 0.6565771235725542,
      "learning_rate": 0.00016274019029440307,
      "loss": 12.3797,
      "step": 11224
    },
    {
      "epoch": 0.6112461116443065,
      "grad_norm": 0.5606421552391951,
      "learning_rate": 0.00016273332331155732,
      "loss": 12.2621,
      "step": 11225
    },
    {
      "epoch": 0.6113005656408895,
      "grad_norm": 0.5888887080072942,
      "learning_rate": 0.00016272645584088674,
      "loss": 12.3646,
      "step": 11226
    },
    {
      "epoch": 0.6113550196374725,
      "grad_norm": 0.5660003854592005,
      "learning_rate": 0.00016271958788244474,
      "loss": 12.3718,
      "step": 11227
    },
    {
      "epoch": 0.6114094736340555,
      "grad_norm": 0.5565818029336428,
      "learning_rate": 0.00016271271943628466,
      "loss": 12.3108,
      "step": 11228
    },
    {
      "epoch": 0.6114639276306385,
      "grad_norm": 0.6250573083091369,
      "learning_rate": 0.00016270585050245999,
      "loss": 12.414,
      "step": 11229
    },
    {
      "epoch": 0.6115183816272216,
      "grad_norm": 0.5887684866794202,
      "learning_rate": 0.00016269898108102414,
      "loss": 12.4102,
      "step": 11230
    },
    {
      "epoch": 0.6115728356238046,
      "grad_norm": 0.6040457465405051,
      "learning_rate": 0.00016269211117203044,
      "loss": 12.4095,
      "step": 11231
    },
    {
      "epoch": 0.6116272896203876,
      "grad_norm": 0.6114239519382482,
      "learning_rate": 0.00016268524077553238,
      "loss": 12.2274,
      "step": 11232
    },
    {
      "epoch": 0.6116817436169706,
      "grad_norm": 0.5965745055703814,
      "learning_rate": 0.00016267836989158338,
      "loss": 12.2036,
      "step": 11233
    },
    {
      "epoch": 0.6117361976135536,
      "grad_norm": 0.6143799913864648,
      "learning_rate": 0.0001626714985202369,
      "loss": 12.4022,
      "step": 11234
    },
    {
      "epoch": 0.6117906516101366,
      "grad_norm": 0.6054444200965224,
      "learning_rate": 0.0001626646266615463,
      "loss": 12.2948,
      "step": 11235
    },
    {
      "epoch": 0.6118451056067197,
      "grad_norm": 0.5873370076509535,
      "learning_rate": 0.00016265775431556506,
      "loss": 12.3566,
      "step": 11236
    },
    {
      "epoch": 0.6118995596033027,
      "grad_norm": 0.6543434983368583,
      "learning_rate": 0.0001626508814823466,
      "loss": 12.2912,
      "step": 11237
    },
    {
      "epoch": 0.6119540135998857,
      "grad_norm": 0.622245962456289,
      "learning_rate": 0.0001626440081619444,
      "loss": 12.3253,
      "step": 11238
    },
    {
      "epoch": 0.6120084675964687,
      "grad_norm": 0.5760824372813074,
      "learning_rate": 0.00016263713435441188,
      "loss": 12.3145,
      "step": 11239
    },
    {
      "epoch": 0.6120629215930516,
      "grad_norm": 0.6597670651926038,
      "learning_rate": 0.00016263026005980253,
      "loss": 12.3367,
      "step": 11240
    },
    {
      "epoch": 0.6121173755896346,
      "grad_norm": 0.67169983719772,
      "learning_rate": 0.00016262338527816972,
      "loss": 12.0397,
      "step": 11241
    },
    {
      "epoch": 0.6121718295862177,
      "grad_norm": 0.6221931850437608,
      "learning_rate": 0.00016261651000956703,
      "loss": 12.2584,
      "step": 11242
    },
    {
      "epoch": 0.6122262835828007,
      "grad_norm": 0.6342830763189989,
      "learning_rate": 0.0001626096342540478,
      "loss": 12.3914,
      "step": 11243
    },
    {
      "epoch": 0.6122807375793837,
      "grad_norm": 0.567870951485226,
      "learning_rate": 0.0001626027580116656,
      "loss": 12.2923,
      "step": 11244
    },
    {
      "epoch": 0.6123351915759667,
      "grad_norm": 0.5671950134759478,
      "learning_rate": 0.0001625958812824738,
      "loss": 12.3838,
      "step": 11245
    },
    {
      "epoch": 0.6123896455725497,
      "grad_norm": 0.6211012286514693,
      "learning_rate": 0.00016258900406652596,
      "loss": 12.4002,
      "step": 11246
    },
    {
      "epoch": 0.6124440995691327,
      "grad_norm": 0.619630309091858,
      "learning_rate": 0.00016258212636387556,
      "loss": 12.3923,
      "step": 11247
    },
    {
      "epoch": 0.6124985535657158,
      "grad_norm": 0.6709943689223554,
      "learning_rate": 0.00016257524817457598,
      "loss": 12.1034,
      "step": 11248
    },
    {
      "epoch": 0.6125530075622988,
      "grad_norm": 0.5728013482326856,
      "learning_rate": 0.00016256836949868082,
      "loss": 12.2234,
      "step": 11249
    },
    {
      "epoch": 0.6126074615588818,
      "grad_norm": 0.5378034627203535,
      "learning_rate": 0.00016256149033624354,
      "loss": 12.3361,
      "step": 11250
    },
    {
      "epoch": 0.6126619155554648,
      "grad_norm": 0.5506895229670732,
      "learning_rate": 0.0001625546106873176,
      "loss": 12.1849,
      "step": 11251
    },
    {
      "epoch": 0.6127163695520478,
      "grad_norm": 0.6064870841906332,
      "learning_rate": 0.00016254773055195652,
      "loss": 12.3334,
      "step": 11252
    },
    {
      "epoch": 0.6127708235486308,
      "grad_norm": 0.6667096903245954,
      "learning_rate": 0.0001625408499302138,
      "loss": 12.4349,
      "step": 11253
    },
    {
      "epoch": 0.6128252775452139,
      "grad_norm": 0.6123349006803258,
      "learning_rate": 0.00016253396882214292,
      "loss": 12.415,
      "step": 11254
    },
    {
      "epoch": 0.6128797315417969,
      "grad_norm": 0.5828800827037145,
      "learning_rate": 0.00016252708722779742,
      "loss": 12.2713,
      "step": 11255
    },
    {
      "epoch": 0.6129341855383799,
      "grad_norm": 0.5810626939893756,
      "learning_rate": 0.00016252020514723084,
      "loss": 12.2411,
      "step": 11256
    },
    {
      "epoch": 0.6129886395349629,
      "grad_norm": 0.610484695656065,
      "learning_rate": 0.00016251332258049664,
      "loss": 12.3313,
      "step": 11257
    },
    {
      "epoch": 0.6130430935315458,
      "grad_norm": 0.5960070240782448,
      "learning_rate": 0.00016250643952764833,
      "loss": 12.379,
      "step": 11258
    },
    {
      "epoch": 0.6130975475281288,
      "grad_norm": 0.6448646995063299,
      "learning_rate": 0.00016249955598873948,
      "loss": 12.3597,
      "step": 11259
    },
    {
      "epoch": 0.6131520015247119,
      "grad_norm": 0.5791912269801887,
      "learning_rate": 0.00016249267196382362,
      "loss": 12.2723,
      "step": 11260
    },
    {
      "epoch": 0.6132064555212949,
      "grad_norm": 0.5642981121650137,
      "learning_rate": 0.00016248578745295427,
      "loss": 12.3564,
      "step": 11261
    },
    {
      "epoch": 0.6132609095178779,
      "grad_norm": 0.6165409779113825,
      "learning_rate": 0.00016247890245618492,
      "loss": 12.2522,
      "step": 11262
    },
    {
      "epoch": 0.6133153635144609,
      "grad_norm": 0.627470761262432,
      "learning_rate": 0.00016247201697356918,
      "loss": 12.3381,
      "step": 11263
    },
    {
      "epoch": 0.6133698175110439,
      "grad_norm": 0.6276131776598461,
      "learning_rate": 0.00016246513100516053,
      "loss": 12.2903,
      "step": 11264
    },
    {
      "epoch": 0.613424271507627,
      "grad_norm": 0.6311633542362459,
      "learning_rate": 0.00016245824455101256,
      "loss": 12.385,
      "step": 11265
    },
    {
      "epoch": 0.61347872550421,
      "grad_norm": 0.6036525845880586,
      "learning_rate": 0.00016245135761117882,
      "loss": 12.3985,
      "step": 11266
    },
    {
      "epoch": 0.613533179500793,
      "grad_norm": 0.6372653976777525,
      "learning_rate": 0.00016244447018571285,
      "loss": 12.3874,
      "step": 11267
    },
    {
      "epoch": 0.613587633497376,
      "grad_norm": 0.6120876586946324,
      "learning_rate": 0.00016243758227466816,
      "loss": 12.3954,
      "step": 11268
    },
    {
      "epoch": 0.613642087493959,
      "grad_norm": 0.5670362688938073,
      "learning_rate": 0.00016243069387809843,
      "loss": 12.2247,
      "step": 11269
    },
    {
      "epoch": 0.613696541490542,
      "grad_norm": 0.6061437844694163,
      "learning_rate": 0.00016242380499605712,
      "loss": 12.3352,
      "step": 11270
    },
    {
      "epoch": 0.6137509954871251,
      "grad_norm": 0.6213968616718148,
      "learning_rate": 0.00016241691562859782,
      "loss": 12.2934,
      "step": 11271
    },
    {
      "epoch": 0.6138054494837081,
      "grad_norm": 0.572983281791859,
      "learning_rate": 0.00016241002577577413,
      "loss": 12.296,
      "step": 11272
    },
    {
      "epoch": 0.6138599034802911,
      "grad_norm": 0.565001252538547,
      "learning_rate": 0.0001624031354376396,
      "loss": 12.3484,
      "step": 11273
    },
    {
      "epoch": 0.6139143574768741,
      "grad_norm": 0.6102422361467659,
      "learning_rate": 0.00016239624461424784,
      "loss": 12.5153,
      "step": 11274
    },
    {
      "epoch": 0.613968811473457,
      "grad_norm": 0.791363759356964,
      "learning_rate": 0.00016238935330565243,
      "loss": 12.4972,
      "step": 11275
    },
    {
      "epoch": 0.61402326547004,
      "grad_norm": 0.5937446958082347,
      "learning_rate": 0.00016238246151190696,
      "loss": 12.1792,
      "step": 11276
    },
    {
      "epoch": 0.6140777194666232,
      "grad_norm": 0.6128242185531629,
      "learning_rate": 0.00016237556923306496,
      "loss": 12.4127,
      "step": 11277
    },
    {
      "epoch": 0.6141321734632061,
      "grad_norm": 0.5133072704141306,
      "learning_rate": 0.00016236867646918007,
      "loss": 12.2986,
      "step": 11278
    },
    {
      "epoch": 0.6141866274597891,
      "grad_norm": 0.616740205009282,
      "learning_rate": 0.00016236178322030594,
      "loss": 12.4195,
      "step": 11279
    },
    {
      "epoch": 0.6142410814563721,
      "grad_norm": 0.6902749010715484,
      "learning_rate": 0.0001623548894864961,
      "loss": 12.2448,
      "step": 11280
    },
    {
      "epoch": 0.6142955354529551,
      "grad_norm": 0.5800100682633578,
      "learning_rate": 0.00016234799526780418,
      "loss": 12.3209,
      "step": 11281
    },
    {
      "epoch": 0.6143499894495381,
      "grad_norm": 0.5863122766710325,
      "learning_rate": 0.0001623411005642838,
      "loss": 12.2777,
      "step": 11282
    },
    {
      "epoch": 0.6144044434461212,
      "grad_norm": 0.5681102531243815,
      "learning_rate": 0.00016233420537598855,
      "loss": 12.4094,
      "step": 11283
    },
    {
      "epoch": 0.6144588974427042,
      "grad_norm": 0.5431159450349899,
      "learning_rate": 0.0001623273097029721,
      "loss": 12.3014,
      "step": 11284
    },
    {
      "epoch": 0.6145133514392872,
      "grad_norm": 0.5311266917053576,
      "learning_rate": 0.00016232041354528802,
      "loss": 12.2448,
      "step": 11285
    },
    {
      "epoch": 0.6145678054358702,
      "grad_norm": 0.610874460094281,
      "learning_rate": 0.00016231351690298995,
      "loss": 12.4282,
      "step": 11286
    },
    {
      "epoch": 0.6146222594324532,
      "grad_norm": 0.6480890556816988,
      "learning_rate": 0.0001623066197761315,
      "loss": 12.4236,
      "step": 11287
    },
    {
      "epoch": 0.6146767134290362,
      "grad_norm": 0.6069702570834702,
      "learning_rate": 0.00016229972216476635,
      "loss": 12.4945,
      "step": 11288
    },
    {
      "epoch": 0.6147311674256193,
      "grad_norm": 0.5934019147160708,
      "learning_rate": 0.00016229282406894811,
      "loss": 12.3161,
      "step": 11289
    },
    {
      "epoch": 0.6147856214222023,
      "grad_norm": 0.5672283273861385,
      "learning_rate": 0.00016228592548873043,
      "loss": 12.2918,
      "step": 11290
    },
    {
      "epoch": 0.6148400754187853,
      "grad_norm": 0.7098678134884511,
      "learning_rate": 0.00016227902642416695,
      "loss": 12.3796,
      "step": 11291
    },
    {
      "epoch": 0.6148945294153683,
      "grad_norm": 0.5864296809404334,
      "learning_rate": 0.0001622721268753113,
      "loss": 12.2628,
      "step": 11292
    },
    {
      "epoch": 0.6149489834119513,
      "grad_norm": 0.5438277220093632,
      "learning_rate": 0.00016226522684221716,
      "loss": 12.2331,
      "step": 11293
    },
    {
      "epoch": 0.6150034374085342,
      "grad_norm": 0.6313422982743069,
      "learning_rate": 0.00016225832632493819,
      "loss": 12.194,
      "step": 11294
    },
    {
      "epoch": 0.6150578914051174,
      "grad_norm": 0.5614566345902298,
      "learning_rate": 0.00016225142532352803,
      "loss": 12.4039,
      "step": 11295
    },
    {
      "epoch": 0.6151123454017003,
      "grad_norm": 0.5955111725120733,
      "learning_rate": 0.0001622445238380403,
      "loss": 12.3284,
      "step": 11296
    },
    {
      "epoch": 0.6151667993982833,
      "grad_norm": 0.5981568208880793,
      "learning_rate": 0.00016223762186852876,
      "loss": 12.386,
      "step": 11297
    },
    {
      "epoch": 0.6152212533948663,
      "grad_norm": 0.5779828111185735,
      "learning_rate": 0.000162230719415047,
      "loss": 12.3549,
      "step": 11298
    },
    {
      "epoch": 0.6152757073914493,
      "grad_norm": 0.5988882027190688,
      "learning_rate": 0.00016222381647764875,
      "loss": 12.2865,
      "step": 11299
    },
    {
      "epoch": 0.6153301613880324,
      "grad_norm": 0.5598859235243391,
      "learning_rate": 0.0001622169130563877,
      "loss": 12.3623,
      "step": 11300
    },
    {
      "epoch": 0.6153846153846154,
      "grad_norm": 0.6025884901045108,
      "learning_rate": 0.00016221000915131746,
      "loss": 12.3875,
      "step": 11301
    },
    {
      "epoch": 0.6154390693811984,
      "grad_norm": 0.5788006219062364,
      "learning_rate": 0.00016220310476249176,
      "loss": 12.275,
      "step": 11302
    },
    {
      "epoch": 0.6154935233777814,
      "grad_norm": 0.6248198295930657,
      "learning_rate": 0.00016219619988996428,
      "loss": 12.1747,
      "step": 11303
    },
    {
      "epoch": 0.6155479773743644,
      "grad_norm": 0.6097860819021905,
      "learning_rate": 0.00016218929453378874,
      "loss": 12.3079,
      "step": 11304
    },
    {
      "epoch": 0.6156024313709474,
      "grad_norm": 0.6942998620702879,
      "learning_rate": 0.0001621823886940188,
      "loss": 12.4143,
      "step": 11305
    },
    {
      "epoch": 0.6156568853675305,
      "grad_norm": 0.6263141335439183,
      "learning_rate": 0.0001621754823707082,
      "loss": 12.361,
      "step": 11306
    },
    {
      "epoch": 0.6157113393641135,
      "grad_norm": 0.6863996262057079,
      "learning_rate": 0.00016216857556391063,
      "loss": 12.3674,
      "step": 11307
    },
    {
      "epoch": 0.6157657933606965,
      "grad_norm": 0.616314884930882,
      "learning_rate": 0.00016216166827367974,
      "loss": 12.2604,
      "step": 11308
    },
    {
      "epoch": 0.6158202473572795,
      "grad_norm": 0.5983392937109377,
      "learning_rate": 0.00016215476050006932,
      "loss": 12.2909,
      "step": 11309
    },
    {
      "epoch": 0.6158747013538625,
      "grad_norm": 0.6468327132530235,
      "learning_rate": 0.0001621478522431331,
      "loss": 12.3449,
      "step": 11310
    },
    {
      "epoch": 0.6159291553504455,
      "grad_norm": 0.6584847095297925,
      "learning_rate": 0.00016214094350292468,
      "loss": 12.4412,
      "step": 11311
    },
    {
      "epoch": 0.6159836093470286,
      "grad_norm": 0.5925212894375936,
      "learning_rate": 0.00016213403427949792,
      "loss": 12.3087,
      "step": 11312
    },
    {
      "epoch": 0.6160380633436116,
      "grad_norm": 0.6141912148999293,
      "learning_rate": 0.00016212712457290646,
      "loss": 12.2135,
      "step": 11313
    },
    {
      "epoch": 0.6160925173401945,
      "grad_norm": 0.5730557449185941,
      "learning_rate": 0.0001621202143832041,
      "loss": 12.2448,
      "step": 11314
    },
    {
      "epoch": 0.6161469713367775,
      "grad_norm": 0.6319410511618868,
      "learning_rate": 0.0001621133037104445,
      "loss": 12.2875,
      "step": 11315
    },
    {
      "epoch": 0.6162014253333605,
      "grad_norm": 0.566548677650518,
      "learning_rate": 0.00016210639255468144,
      "loss": 12.3506,
      "step": 11316
    },
    {
      "epoch": 0.6162558793299435,
      "grad_norm": 0.571999101358985,
      "learning_rate": 0.00016209948091596864,
      "loss": 12.3217,
      "step": 11317
    },
    {
      "epoch": 0.6163103333265266,
      "grad_norm": 0.5840680525297499,
      "learning_rate": 0.00016209256879435988,
      "loss": 12.2879,
      "step": 11318
    },
    {
      "epoch": 0.6163647873231096,
      "grad_norm": 0.6271115358651692,
      "learning_rate": 0.0001620856561899089,
      "loss": 12.3066,
      "step": 11319
    },
    {
      "epoch": 0.6164192413196926,
      "grad_norm": 0.5674867522352245,
      "learning_rate": 0.00016207874310266945,
      "loss": 12.404,
      "step": 11320
    },
    {
      "epoch": 0.6164736953162756,
      "grad_norm": 0.5963678282919692,
      "learning_rate": 0.00016207182953269523,
      "loss": 12.3059,
      "step": 11321
    },
    {
      "epoch": 0.6165281493128586,
      "grad_norm": 0.6584712900680062,
      "learning_rate": 0.00016206491548004012,
      "loss": 12.1358,
      "step": 11322
    },
    {
      "epoch": 0.6165826033094416,
      "grad_norm": 0.6232001423449859,
      "learning_rate": 0.00016205800094475778,
      "loss": 12.2733,
      "step": 11323
    },
    {
      "epoch": 0.6166370573060247,
      "grad_norm": 0.5961936391155271,
      "learning_rate": 0.00016205108592690204,
      "loss": 12.267,
      "step": 11324
    },
    {
      "epoch": 0.6166915113026077,
      "grad_norm": 0.6233694372694509,
      "learning_rate": 0.00016204417042652665,
      "loss": 12.3168,
      "step": 11325
    },
    {
      "epoch": 0.6167459652991907,
      "grad_norm": 0.5967553232670233,
      "learning_rate": 0.00016203725444368538,
      "loss": 12.2894,
      "step": 11326
    },
    {
      "epoch": 0.6168004192957737,
      "grad_norm": 0.6328880831888363,
      "learning_rate": 0.000162030337978432,
      "loss": 12.4589,
      "step": 11327
    },
    {
      "epoch": 0.6168548732923567,
      "grad_norm": 0.6599660413066143,
      "learning_rate": 0.00016202342103082033,
      "loss": 12.133,
      "step": 11328
    },
    {
      "epoch": 0.6169093272889397,
      "grad_norm": 0.6050448734616868,
      "learning_rate": 0.00016201650360090413,
      "loss": 12.3368,
      "step": 11329
    },
    {
      "epoch": 0.6169637812855228,
      "grad_norm": 0.616854598836734,
      "learning_rate": 0.0001620095856887372,
      "loss": 12.2232,
      "step": 11330
    },
    {
      "epoch": 0.6170182352821058,
      "grad_norm": 0.6680516335137024,
      "learning_rate": 0.00016200266729437333,
      "loss": 12.3242,
      "step": 11331
    },
    {
      "epoch": 0.6170726892786887,
      "grad_norm": 0.5497639299706962,
      "learning_rate": 0.00016199574841786635,
      "loss": 12.3369,
      "step": 11332
    },
    {
      "epoch": 0.6171271432752717,
      "grad_norm": 0.6708234481169313,
      "learning_rate": 0.00016198882905926997,
      "loss": 12.3544,
      "step": 11333
    },
    {
      "epoch": 0.6171815972718547,
      "grad_norm": 0.6315538192135355,
      "learning_rate": 0.0001619819092186381,
      "loss": 12.1928,
      "step": 11334
    },
    {
      "epoch": 0.6172360512684378,
      "grad_norm": 0.6490092063174805,
      "learning_rate": 0.00016197498889602448,
      "loss": 12.1877,
      "step": 11335
    },
    {
      "epoch": 0.6172905052650208,
      "grad_norm": 0.5577690708953591,
      "learning_rate": 0.00016196806809148302,
      "loss": 12.4101,
      "step": 11336
    },
    {
      "epoch": 0.6173449592616038,
      "grad_norm": 0.6033656557328602,
      "learning_rate": 0.00016196114680506741,
      "loss": 12.3735,
      "step": 11337
    },
    {
      "epoch": 0.6173994132581868,
      "grad_norm": 0.6226698585246981,
      "learning_rate": 0.00016195422503683155,
      "loss": 12.3514,
      "step": 11338
    },
    {
      "epoch": 0.6174538672547698,
      "grad_norm": 0.6146688034799889,
      "learning_rate": 0.00016194730278682923,
      "loss": 12.3374,
      "step": 11339
    },
    {
      "epoch": 0.6175083212513528,
      "grad_norm": 0.6202871794806031,
      "learning_rate": 0.00016194038005511432,
      "loss": 12.3491,
      "step": 11340
    },
    {
      "epoch": 0.6175627752479359,
      "grad_norm": 0.654543386433988,
      "learning_rate": 0.0001619334568417406,
      "loss": 12.4703,
      "step": 11341
    },
    {
      "epoch": 0.6176172292445189,
      "grad_norm": 0.6079034969424639,
      "learning_rate": 0.00016192653314676196,
      "loss": 12.2333,
      "step": 11342
    },
    {
      "epoch": 0.6176716832411019,
      "grad_norm": 0.666143872758608,
      "learning_rate": 0.0001619196089702322,
      "loss": 12.3581,
      "step": 11343
    },
    {
      "epoch": 0.6177261372376849,
      "grad_norm": 0.5102477274485845,
      "learning_rate": 0.00016191268431220519,
      "loss": 12.3265,
      "step": 11344
    },
    {
      "epoch": 0.6177805912342679,
      "grad_norm": 0.5784737794499895,
      "learning_rate": 0.00016190575917273474,
      "loss": 12.3641,
      "step": 11345
    },
    {
      "epoch": 0.6178350452308509,
      "grad_norm": 0.684633075833952,
      "learning_rate": 0.00016189883355187477,
      "loss": 12.3834,
      "step": 11346
    },
    {
      "epoch": 0.617889499227434,
      "grad_norm": 0.6336297914409275,
      "learning_rate": 0.00016189190744967906,
      "loss": 12.3893,
      "step": 11347
    },
    {
      "epoch": 0.617943953224017,
      "grad_norm": 0.6382634194907015,
      "learning_rate": 0.00016188498086620146,
      "loss": 12.4647,
      "step": 11348
    },
    {
      "epoch": 0.6179984072206,
      "grad_norm": 0.6638108620704917,
      "learning_rate": 0.00016187805380149596,
      "loss": 12.4085,
      "step": 11349
    },
    {
      "epoch": 0.618052861217183,
      "grad_norm": 0.5642107833644697,
      "learning_rate": 0.00016187112625561624,
      "loss": 12.2775,
      "step": 11350
    },
    {
      "epoch": 0.6181073152137659,
      "grad_norm": 0.618448690303876,
      "learning_rate": 0.00016186419822861634,
      "loss": 12.323,
      "step": 11351
    },
    {
      "epoch": 0.6181617692103489,
      "grad_norm": 0.6144504495899537,
      "learning_rate": 0.00016185726972055002,
      "loss": 12.2607,
      "step": 11352
    },
    {
      "epoch": 0.618216223206932,
      "grad_norm": 0.5653022098286621,
      "learning_rate": 0.0001618503407314712,
      "loss": 12.2345,
      "step": 11353
    },
    {
      "epoch": 0.618270677203515,
      "grad_norm": 0.5923636217015734,
      "learning_rate": 0.00016184341126143376,
      "loss": 12.3753,
      "step": 11354
    },
    {
      "epoch": 0.618325131200098,
      "grad_norm": 0.6090629767174827,
      "learning_rate": 0.0001618364813104916,
      "loss": 12.3978,
      "step": 11355
    },
    {
      "epoch": 0.618379585196681,
      "grad_norm": 0.760239146374173,
      "learning_rate": 0.00016182955087869859,
      "loss": 12.3513,
      "step": 11356
    },
    {
      "epoch": 0.618434039193264,
      "grad_norm": 0.5499640804080805,
      "learning_rate": 0.0001618226199661086,
      "loss": 12.2744,
      "step": 11357
    },
    {
      "epoch": 0.618488493189847,
      "grad_norm": 0.5413087096896169,
      "learning_rate": 0.0001618156885727756,
      "loss": 12.2719,
      "step": 11358
    },
    {
      "epoch": 0.6185429471864301,
      "grad_norm": 0.6448601280327934,
      "learning_rate": 0.0001618087566987534,
      "loss": 12.3079,
      "step": 11359
    },
    {
      "epoch": 0.6185974011830131,
      "grad_norm": 0.6204740814101944,
      "learning_rate": 0.00016180182434409593,
      "loss": 12.3334,
      "step": 11360
    },
    {
      "epoch": 0.6186518551795961,
      "grad_norm": 0.6521050686806655,
      "learning_rate": 0.00016179489150885715,
      "loss": 12.1476,
      "step": 11361
    },
    {
      "epoch": 0.6187063091761791,
      "grad_norm": 0.5789250892511003,
      "learning_rate": 0.00016178795819309086,
      "loss": 12.3211,
      "step": 11362
    },
    {
      "epoch": 0.6187607631727621,
      "grad_norm": 0.6427843582992568,
      "learning_rate": 0.00016178102439685113,
      "loss": 12.4807,
      "step": 11363
    },
    {
      "epoch": 0.6188152171693452,
      "grad_norm": 0.7076880164255421,
      "learning_rate": 0.00016177409012019175,
      "loss": 12.3935,
      "step": 11364
    },
    {
      "epoch": 0.6188696711659282,
      "grad_norm": 0.5605514128844377,
      "learning_rate": 0.0001617671553631667,
      "loss": 12.2769,
      "step": 11365
    },
    {
      "epoch": 0.6189241251625112,
      "grad_norm": 0.6799978308680839,
      "learning_rate": 0.0001617602201258299,
      "loss": 12.2532,
      "step": 11366
    },
    {
      "epoch": 0.6189785791590942,
      "grad_norm": 0.7108657629179751,
      "learning_rate": 0.00016175328440823524,
      "loss": 12.2893,
      "step": 11367
    },
    {
      "epoch": 0.6190330331556771,
      "grad_norm": 0.6701568668267008,
      "learning_rate": 0.00016174634821043666,
      "loss": 12.331,
      "step": 11368
    },
    {
      "epoch": 0.6190874871522601,
      "grad_norm": 0.6167861448581061,
      "learning_rate": 0.00016173941153248817,
      "loss": 12.1841,
      "step": 11369
    },
    {
      "epoch": 0.6191419411488432,
      "grad_norm": 0.5542796556543904,
      "learning_rate": 0.00016173247437444366,
      "loss": 12.2144,
      "step": 11370
    },
    {
      "epoch": 0.6191963951454262,
      "grad_norm": 0.5561406263549451,
      "learning_rate": 0.00016172553673635706,
      "loss": 12.3912,
      "step": 11371
    },
    {
      "epoch": 0.6192508491420092,
      "grad_norm": 0.6768993477226342,
      "learning_rate": 0.00016171859861828237,
      "loss": 12.3638,
      "step": 11372
    },
    {
      "epoch": 0.6193053031385922,
      "grad_norm": 0.5643467859039376,
      "learning_rate": 0.00016171166002027344,
      "loss": 12.2933,
      "step": 11373
    },
    {
      "epoch": 0.6193597571351752,
      "grad_norm": 0.6870602034201745,
      "learning_rate": 0.00016170472094238436,
      "loss": 12.4109,
      "step": 11374
    },
    {
      "epoch": 0.6194142111317582,
      "grad_norm": 0.6432492313425489,
      "learning_rate": 0.00016169778138466897,
      "loss": 12.3329,
      "step": 11375
    },
    {
      "epoch": 0.6194686651283413,
      "grad_norm": 0.5871531262125809,
      "learning_rate": 0.00016169084134718133,
      "loss": 12.3689,
      "step": 11376
    },
    {
      "epoch": 0.6195231191249243,
      "grad_norm": 0.581973602589214,
      "learning_rate": 0.00016168390082997534,
      "loss": 12.3192,
      "step": 11377
    },
    {
      "epoch": 0.6195775731215073,
      "grad_norm": 0.7273779095746546,
      "learning_rate": 0.000161676959833105,
      "loss": 12.4788,
      "step": 11378
    },
    {
      "epoch": 0.6196320271180903,
      "grad_norm": 0.6365631002352293,
      "learning_rate": 0.0001616700183566243,
      "loss": 12.4043,
      "step": 11379
    },
    {
      "epoch": 0.6196864811146733,
      "grad_norm": 0.5885970626918695,
      "learning_rate": 0.00016166307640058712,
      "loss": 12.3859,
      "step": 11380
    },
    {
      "epoch": 0.6197409351112563,
      "grad_norm": 0.6437308372453692,
      "learning_rate": 0.0001616561339650476,
      "loss": 12.3077,
      "step": 11381
    },
    {
      "epoch": 0.6197953891078394,
      "grad_norm": 0.7012567621229076,
      "learning_rate": 0.00016164919105005957,
      "loss": 12.2332,
      "step": 11382
    },
    {
      "epoch": 0.6198498431044224,
      "grad_norm": 0.5809889464584606,
      "learning_rate": 0.00016164224765567714,
      "loss": 12.2526,
      "step": 11383
    },
    {
      "epoch": 0.6199042971010054,
      "grad_norm": 0.6078813911317118,
      "learning_rate": 0.00016163530378195424,
      "loss": 12.47,
      "step": 11384
    },
    {
      "epoch": 0.6199587510975884,
      "grad_norm": 0.5687762963315507,
      "learning_rate": 0.0001616283594289449,
      "loss": 12.4204,
      "step": 11385
    },
    {
      "epoch": 0.6200132050941713,
      "grad_norm": 0.6699596697271144,
      "learning_rate": 0.00016162141459670308,
      "loss": 12.3298,
      "step": 11386
    },
    {
      "epoch": 0.6200676590907543,
      "grad_norm": 0.688363027506524,
      "learning_rate": 0.00016161446928528284,
      "loss": 12.3479,
      "step": 11387
    },
    {
      "epoch": 0.6201221130873374,
      "grad_norm": 0.5737921994052604,
      "learning_rate": 0.00016160752349473812,
      "loss": 12.3075,
      "step": 11388
    },
    {
      "epoch": 0.6201765670839204,
      "grad_norm": 0.5286850608693549,
      "learning_rate": 0.00016160057722512295,
      "loss": 12.1068,
      "step": 11389
    },
    {
      "epoch": 0.6202310210805034,
      "grad_norm": 0.5964030221046523,
      "learning_rate": 0.00016159363047649138,
      "loss": 12.3793,
      "step": 11390
    },
    {
      "epoch": 0.6202854750770864,
      "grad_norm": 0.6197530079107472,
      "learning_rate": 0.00016158668324889742,
      "loss": 12.4887,
      "step": 11391
    },
    {
      "epoch": 0.6203399290736694,
      "grad_norm": 0.5881884732597834,
      "learning_rate": 0.0001615797355423951,
      "loss": 12.336,
      "step": 11392
    },
    {
      "epoch": 0.6203943830702524,
      "grad_norm": 0.6180850844308952,
      "learning_rate": 0.0001615727873570384,
      "loss": 12.3021,
      "step": 11393
    },
    {
      "epoch": 0.6204488370668355,
      "grad_norm": 0.5862481019080754,
      "learning_rate": 0.00016156583869288138,
      "loss": 12.4791,
      "step": 11394
    },
    {
      "epoch": 0.6205032910634185,
      "grad_norm": 0.5721204858048597,
      "learning_rate": 0.0001615588895499781,
      "loss": 12.3929,
      "step": 11395
    },
    {
      "epoch": 0.6205577450600015,
      "grad_norm": 0.6527465480459093,
      "learning_rate": 0.00016155193992838253,
      "loss": 12.4634,
      "step": 11396
    },
    {
      "epoch": 0.6206121990565845,
      "grad_norm": 0.5626923759738245,
      "learning_rate": 0.0001615449898281488,
      "loss": 12.2249,
      "step": 11397
    },
    {
      "epoch": 0.6206666530531675,
      "grad_norm": 0.5240694474890504,
      "learning_rate": 0.00016153803924933086,
      "loss": 12.268,
      "step": 11398
    },
    {
      "epoch": 0.6207211070497506,
      "grad_norm": 0.6662649597332017,
      "learning_rate": 0.00016153108819198285,
      "loss": 12.3597,
      "step": 11399
    },
    {
      "epoch": 0.6207755610463336,
      "grad_norm": 0.6755992426640863,
      "learning_rate": 0.00016152413665615874,
      "loss": 12.333,
      "step": 11400
    },
    {
      "epoch": 0.6208300150429166,
      "grad_norm": 0.5522321110889248,
      "learning_rate": 0.00016151718464191265,
      "loss": 12.2361,
      "step": 11401
    },
    {
      "epoch": 0.6208844690394996,
      "grad_norm": 0.6126956189989169,
      "learning_rate": 0.0001615102321492986,
      "loss": 12.4911,
      "step": 11402
    },
    {
      "epoch": 0.6209389230360826,
      "grad_norm": 0.6273283474136503,
      "learning_rate": 0.0001615032791783707,
      "loss": 12.4404,
      "step": 11403
    },
    {
      "epoch": 0.6209933770326655,
      "grad_norm": 0.5857263956973275,
      "learning_rate": 0.00016149632572918295,
      "loss": 12.3055,
      "step": 11404
    },
    {
      "epoch": 0.6210478310292487,
      "grad_norm": 0.6970385254814665,
      "learning_rate": 0.00016148937180178948,
      "loss": 12.4064,
      "step": 11405
    },
    {
      "epoch": 0.6211022850258316,
      "grad_norm": 0.6333024467892219,
      "learning_rate": 0.00016148241739624431,
      "loss": 12.2791,
      "step": 11406
    },
    {
      "epoch": 0.6211567390224146,
      "grad_norm": 0.6105156550817371,
      "learning_rate": 0.0001614754625126016,
      "loss": 12.2001,
      "step": 11407
    },
    {
      "epoch": 0.6212111930189976,
      "grad_norm": 0.5839111937695641,
      "learning_rate": 0.00016146850715091537,
      "loss": 12.3525,
      "step": 11408
    },
    {
      "epoch": 0.6212656470155806,
      "grad_norm": 0.5676464285172782,
      "learning_rate": 0.00016146155131123972,
      "loss": 12.2968,
      "step": 11409
    },
    {
      "epoch": 0.6213201010121636,
      "grad_norm": 0.6103215468950094,
      "learning_rate": 0.00016145459499362872,
      "loss": 12.3521,
      "step": 11410
    },
    {
      "epoch": 0.6213745550087467,
      "grad_norm": 0.5476910263333569,
      "learning_rate": 0.0001614476381981365,
      "loss": 12.3984,
      "step": 11411
    },
    {
      "epoch": 0.6214290090053297,
      "grad_norm": 0.6216564371614643,
      "learning_rate": 0.00016144068092481715,
      "loss": 12.4056,
      "step": 11412
    },
    {
      "epoch": 0.6214834630019127,
      "grad_norm": 0.5650350514742664,
      "learning_rate": 0.00016143372317372476,
      "loss": 12.3514,
      "step": 11413
    },
    {
      "epoch": 0.6215379169984957,
      "grad_norm": 0.5629443200885589,
      "learning_rate": 0.0001614267649449134,
      "loss": 12.2287,
      "step": 11414
    },
    {
      "epoch": 0.6215923709950787,
      "grad_norm": 0.6506262615831745,
      "learning_rate": 0.00016141980623843725,
      "loss": 12.3663,
      "step": 11415
    },
    {
      "epoch": 0.6216468249916617,
      "grad_norm": 0.5682975024890445,
      "learning_rate": 0.00016141284705435037,
      "loss": 12.3697,
      "step": 11416
    },
    {
      "epoch": 0.6217012789882448,
      "grad_norm": 0.7033118716417303,
      "learning_rate": 0.0001614058873927069,
      "loss": 12.4198,
      "step": 11417
    },
    {
      "epoch": 0.6217557329848278,
      "grad_norm": 0.5432031850109923,
      "learning_rate": 0.00016139892725356095,
      "loss": 12.1988,
      "step": 11418
    },
    {
      "epoch": 0.6218101869814108,
      "grad_norm": 0.6199558312329202,
      "learning_rate": 0.00016139196663696666,
      "loss": 12.3618,
      "step": 11419
    },
    {
      "epoch": 0.6218646409779938,
      "grad_norm": 0.6876465073353317,
      "learning_rate": 0.0001613850055429781,
      "loss": 12.3008,
      "step": 11420
    },
    {
      "epoch": 0.6219190949745768,
      "grad_norm": 0.5512823343568609,
      "learning_rate": 0.0001613780439716495,
      "loss": 12.3898,
      "step": 11421
    },
    {
      "epoch": 0.6219735489711598,
      "grad_norm": 0.5895349045147323,
      "learning_rate": 0.00016137108192303492,
      "loss": 12.3176,
      "step": 11422
    },
    {
      "epoch": 0.6220280029677429,
      "grad_norm": 0.629941616548519,
      "learning_rate": 0.00016136411939718847,
      "loss": 12.1728,
      "step": 11423
    },
    {
      "epoch": 0.6220824569643258,
      "grad_norm": 0.610110213177622,
      "learning_rate": 0.00016135715639416438,
      "loss": 12.3168,
      "step": 11424
    },
    {
      "epoch": 0.6221369109609088,
      "grad_norm": 0.6671360522586627,
      "learning_rate": 0.00016135019291401673,
      "loss": 12.3129,
      "step": 11425
    },
    {
      "epoch": 0.6221913649574918,
      "grad_norm": 0.630821330701,
      "learning_rate": 0.00016134322895679972,
      "loss": 12.3731,
      "step": 11426
    },
    {
      "epoch": 0.6222458189540748,
      "grad_norm": 0.6827847510004765,
      "learning_rate": 0.00016133626452256747,
      "loss": 12.3675,
      "step": 11427
    },
    {
      "epoch": 0.6223002729506578,
      "grad_norm": 0.638241364269164,
      "learning_rate": 0.00016132929961137414,
      "loss": 12.3356,
      "step": 11428
    },
    {
      "epoch": 0.6223547269472409,
      "grad_norm": 0.7731307743750341,
      "learning_rate": 0.00016132233422327385,
      "loss": 12.44,
      "step": 11429
    },
    {
      "epoch": 0.6224091809438239,
      "grad_norm": 0.6614816238411625,
      "learning_rate": 0.00016131536835832085,
      "loss": 12.2131,
      "step": 11430
    },
    {
      "epoch": 0.6224636349404069,
      "grad_norm": 0.6298317845000152,
      "learning_rate": 0.00016130840201656924,
      "loss": 12.3037,
      "step": 11431
    },
    {
      "epoch": 0.6225180889369899,
      "grad_norm": 0.6875551830205758,
      "learning_rate": 0.00016130143519807322,
      "loss": 12.3649,
      "step": 11432
    },
    {
      "epoch": 0.6225725429335729,
      "grad_norm": 0.6613920426623728,
      "learning_rate": 0.00016129446790288699,
      "loss": 12.231,
      "step": 11433
    },
    {
      "epoch": 0.622626996930156,
      "grad_norm": 0.5661040079677665,
      "learning_rate": 0.00016128750013106463,
      "loss": 12.3363,
      "step": 11434
    },
    {
      "epoch": 0.622681450926739,
      "grad_norm": 0.6117659979479873,
      "learning_rate": 0.00016128053188266045,
      "loss": 12.3404,
      "step": 11435
    },
    {
      "epoch": 0.622735904923322,
      "grad_norm": 0.6470145341604084,
      "learning_rate": 0.00016127356315772857,
      "loss": 12.4804,
      "step": 11436
    },
    {
      "epoch": 0.622790358919905,
      "grad_norm": 0.5893466214077081,
      "learning_rate": 0.00016126659395632317,
      "loss": 12.3519,
      "step": 11437
    },
    {
      "epoch": 0.622844812916488,
      "grad_norm": 0.6130572078926236,
      "learning_rate": 0.00016125962427849847,
      "loss": 12.2067,
      "step": 11438
    },
    {
      "epoch": 0.622899266913071,
      "grad_norm": 0.649299424382659,
      "learning_rate": 0.00016125265412430867,
      "loss": 12.3978,
      "step": 11439
    },
    {
      "epoch": 0.6229537209096541,
      "grad_norm": 0.6643560544717511,
      "learning_rate": 0.0001612456834938079,
      "loss": 12.2235,
      "step": 11440
    },
    {
      "epoch": 0.6230081749062371,
      "grad_norm": 0.6577597286908099,
      "learning_rate": 0.00016123871238705052,
      "loss": 12.2824,
      "step": 11441
    },
    {
      "epoch": 0.62306262890282,
      "grad_norm": 0.6010533868191522,
      "learning_rate": 0.00016123174080409056,
      "loss": 12.2624,
      "step": 11442
    },
    {
      "epoch": 0.623117082899403,
      "grad_norm": 0.5642128036208448,
      "learning_rate": 0.00016122476874498234,
      "loss": 12.2395,
      "step": 11443
    },
    {
      "epoch": 0.623171536895986,
      "grad_norm": 0.820910878340875,
      "learning_rate": 0.00016121779620978009,
      "loss": 12.3242,
      "step": 11444
    },
    {
      "epoch": 0.623225990892569,
      "grad_norm": 0.5726982445133031,
      "learning_rate": 0.00016121082319853796,
      "loss": 12.1313,
      "step": 11445
    },
    {
      "epoch": 0.6232804448891521,
      "grad_norm": 0.684093458108693,
      "learning_rate": 0.0001612038497113102,
      "loss": 12.177,
      "step": 11446
    },
    {
      "epoch": 0.6233348988857351,
      "grad_norm": 0.7989837270669965,
      "learning_rate": 0.00016119687574815103,
      "loss": 12.4761,
      "step": 11447
    },
    {
      "epoch": 0.6233893528823181,
      "grad_norm": 0.6048684246884773,
      "learning_rate": 0.00016118990130911472,
      "loss": 12.3019,
      "step": 11448
    },
    {
      "epoch": 0.6234438068789011,
      "grad_norm": 0.6627434098596703,
      "learning_rate": 0.00016118292639425545,
      "loss": 12.3217,
      "step": 11449
    },
    {
      "epoch": 0.6234982608754841,
      "grad_norm": 0.5933997333423316,
      "learning_rate": 0.0001611759510036275,
      "loss": 12.346,
      "step": 11450
    },
    {
      "epoch": 0.6235527148720671,
      "grad_norm": 0.6311707370727244,
      "learning_rate": 0.00016116897513728507,
      "loss": 12.2266,
      "step": 11451
    },
    {
      "epoch": 0.6236071688686502,
      "grad_norm": 0.5682200494836679,
      "learning_rate": 0.00016116199879528245,
      "loss": 12.2795,
      "step": 11452
    },
    {
      "epoch": 0.6236616228652332,
      "grad_norm": 0.5491493618190088,
      "learning_rate": 0.0001611550219776739,
      "loss": 12.3042,
      "step": 11453
    },
    {
      "epoch": 0.6237160768618162,
      "grad_norm": 0.5664628758150743,
      "learning_rate": 0.00016114804468451359,
      "loss": 12.3853,
      "step": 11454
    },
    {
      "epoch": 0.6237705308583992,
      "grad_norm": 0.5770704099528122,
      "learning_rate": 0.00016114106691585587,
      "loss": 12.3271,
      "step": 11455
    },
    {
      "epoch": 0.6238249848549822,
      "grad_norm": 0.5652321308945724,
      "learning_rate": 0.00016113408867175495,
      "loss": 12.2696,
      "step": 11456
    },
    {
      "epoch": 0.6238794388515652,
      "grad_norm": 0.5560296538187035,
      "learning_rate": 0.0001611271099522651,
      "loss": 12.2529,
      "step": 11457
    },
    {
      "epoch": 0.6239338928481483,
      "grad_norm": 0.530748673775166,
      "learning_rate": 0.0001611201307574406,
      "loss": 12.3456,
      "step": 11458
    },
    {
      "epoch": 0.6239883468447313,
      "grad_norm": 0.5964895425684653,
      "learning_rate": 0.00016111315108733568,
      "loss": 12.2588,
      "step": 11459
    },
    {
      "epoch": 0.6240428008413142,
      "grad_norm": 0.5661140090484537,
      "learning_rate": 0.0001611061709420047,
      "loss": 12.3302,
      "step": 11460
    },
    {
      "epoch": 0.6240972548378972,
      "grad_norm": 0.5239071385889489,
      "learning_rate": 0.00016109919032150186,
      "loss": 12.2311,
      "step": 11461
    },
    {
      "epoch": 0.6241517088344802,
      "grad_norm": 0.6031453262958488,
      "learning_rate": 0.00016109220922588146,
      "loss": 12.3783,
      "step": 11462
    },
    {
      "epoch": 0.6242061628310632,
      "grad_norm": 0.5879324423494298,
      "learning_rate": 0.00016108522765519783,
      "loss": 12.2452,
      "step": 11463
    },
    {
      "epoch": 0.6242606168276463,
      "grad_norm": 0.5270952355855081,
      "learning_rate": 0.00016107824560950516,
      "loss": 12.1846,
      "step": 11464
    },
    {
      "epoch": 0.6243150708242293,
      "grad_norm": 0.5950726560960155,
      "learning_rate": 0.00016107126308885787,
      "loss": 12.2707,
      "step": 11465
    },
    {
      "epoch": 0.6243695248208123,
      "grad_norm": 0.6100586106298427,
      "learning_rate": 0.00016106428009331016,
      "loss": 12.3043,
      "step": 11466
    },
    {
      "epoch": 0.6244239788173953,
      "grad_norm": 0.6366884073275398,
      "learning_rate": 0.00016105729662291643,
      "loss": 12.4719,
      "step": 11467
    },
    {
      "epoch": 0.6244784328139783,
      "grad_norm": 0.5748578837822204,
      "learning_rate": 0.00016105031267773086,
      "loss": 12.313,
      "step": 11468
    },
    {
      "epoch": 0.6245328868105614,
      "grad_norm": 0.6744817511043489,
      "learning_rate": 0.00016104332825780783,
      "loss": 12.4107,
      "step": 11469
    },
    {
      "epoch": 0.6245873408071444,
      "grad_norm": 0.5967398349761409,
      "learning_rate": 0.00016103634336320165,
      "loss": 12.3625,
      "step": 11470
    },
    {
      "epoch": 0.6246417948037274,
      "grad_norm": 0.6167695078489145,
      "learning_rate": 0.00016102935799396662,
      "loss": 12.3634,
      "step": 11471
    },
    {
      "epoch": 0.6246962488003104,
      "grad_norm": 0.5803870896505192,
      "learning_rate": 0.0001610223721501571,
      "loss": 12.2731,
      "step": 11472
    },
    {
      "epoch": 0.6247507027968934,
      "grad_norm": 0.6282871396517364,
      "learning_rate": 0.00016101538583182735,
      "loss": 12.4319,
      "step": 11473
    },
    {
      "epoch": 0.6248051567934764,
      "grad_norm": 0.5646362615407534,
      "learning_rate": 0.00016100839903903174,
      "loss": 12.299,
      "step": 11474
    },
    {
      "epoch": 0.6248596107900595,
      "grad_norm": 0.5939754539784448,
      "learning_rate": 0.00016100141177182456,
      "loss": 12.3805,
      "step": 11475
    },
    {
      "epoch": 0.6249140647866425,
      "grad_norm": 0.5399769811511669,
      "learning_rate": 0.0001609944240302602,
      "loss": 12.3223,
      "step": 11476
    },
    {
      "epoch": 0.6249685187832255,
      "grad_norm": 0.5462233913359226,
      "learning_rate": 0.00016098743581439298,
      "loss": 12.2263,
      "step": 11477
    },
    {
      "epoch": 0.6250229727798085,
      "grad_norm": 0.5445448329430717,
      "learning_rate": 0.0001609804471242772,
      "loss": 12.3775,
      "step": 11478
    },
    {
      "epoch": 0.6250774267763914,
      "grad_norm": 0.5906469011451747,
      "learning_rate": 0.00016097345795996728,
      "loss": 12.4571,
      "step": 11479
    },
    {
      "epoch": 0.6251318807729744,
      "grad_norm": 0.569176070603741,
      "learning_rate": 0.00016096646832151746,
      "loss": 12.3415,
      "step": 11480
    },
    {
      "epoch": 0.6251863347695575,
      "grad_norm": 0.6055177138206724,
      "learning_rate": 0.00016095947820898222,
      "loss": 12.3101,
      "step": 11481
    },
    {
      "epoch": 0.6252407887661405,
      "grad_norm": 0.5858300768306324,
      "learning_rate": 0.00016095248762241585,
      "loss": 12.2721,
      "step": 11482
    },
    {
      "epoch": 0.6252952427627235,
      "grad_norm": 0.6811483796041748,
      "learning_rate": 0.0001609454965618727,
      "loss": 12.3685,
      "step": 11483
    },
    {
      "epoch": 0.6253496967593065,
      "grad_norm": 0.5647578694435895,
      "learning_rate": 0.00016093850502740714,
      "loss": 12.254,
      "step": 11484
    },
    {
      "epoch": 0.6254041507558895,
      "grad_norm": 0.7362467140997495,
      "learning_rate": 0.00016093151301907352,
      "loss": 12.3546,
      "step": 11485
    },
    {
      "epoch": 0.6254586047524725,
      "grad_norm": 0.6211658326926465,
      "learning_rate": 0.00016092452053692629,
      "loss": 12.3764,
      "step": 11486
    },
    {
      "epoch": 0.6255130587490556,
      "grad_norm": 0.5776185249749698,
      "learning_rate": 0.00016091752758101976,
      "loss": 12.3521,
      "step": 11487
    },
    {
      "epoch": 0.6255675127456386,
      "grad_norm": 0.6806993333104564,
      "learning_rate": 0.00016091053415140827,
      "loss": 12.2411,
      "step": 11488
    },
    {
      "epoch": 0.6256219667422216,
      "grad_norm": 0.5323216271189978,
      "learning_rate": 0.00016090354024814632,
      "loss": 12.3461,
      "step": 11489
    },
    {
      "epoch": 0.6256764207388046,
      "grad_norm": 0.5230959798871765,
      "learning_rate": 0.0001608965458712882,
      "loss": 12.2264,
      "step": 11490
    },
    {
      "epoch": 0.6257308747353876,
      "grad_norm": 0.6505345111176285,
      "learning_rate": 0.0001608895510208883,
      "loss": 12.3135,
      "step": 11491
    },
    {
      "epoch": 0.6257853287319706,
      "grad_norm": 0.6001563708139264,
      "learning_rate": 0.00016088255569700108,
      "loss": 12.2745,
      "step": 11492
    },
    {
      "epoch": 0.6258397827285537,
      "grad_norm": 0.5530352308068975,
      "learning_rate": 0.0001608755598996809,
      "loss": 12.3888,
      "step": 11493
    },
    {
      "epoch": 0.6258942367251367,
      "grad_norm": 0.5921117482992737,
      "learning_rate": 0.0001608685636289821,
      "loss": 12.3241,
      "step": 11494
    },
    {
      "epoch": 0.6259486907217197,
      "grad_norm": 0.5998253973463721,
      "learning_rate": 0.00016086156688495918,
      "loss": 12.3406,
      "step": 11495
    },
    {
      "epoch": 0.6260031447183027,
      "grad_norm": 0.6089749393227019,
      "learning_rate": 0.00016085456966766652,
      "loss": 12.4072,
      "step": 11496
    },
    {
      "epoch": 0.6260575987148856,
      "grad_norm": 0.6446561830580213,
      "learning_rate": 0.00016084757197715852,
      "loss": 12.3605,
      "step": 11497
    },
    {
      "epoch": 0.6261120527114687,
      "grad_norm": 0.6811082040294554,
      "learning_rate": 0.00016084057381348957,
      "loss": 12.1462,
      "step": 11498
    },
    {
      "epoch": 0.6261665067080517,
      "grad_norm": 0.6105997784239102,
      "learning_rate": 0.00016083357517671413,
      "loss": 12.3169,
      "step": 11499
    },
    {
      "epoch": 0.6262209607046347,
      "grad_norm": 0.5720422689859095,
      "learning_rate": 0.0001608265760668866,
      "loss": 12.3158,
      "step": 11500
    },
    {
      "epoch": 0.6262754147012177,
      "grad_norm": 0.5999332881450222,
      "learning_rate": 0.00016081957648406142,
      "loss": 12.3959,
      "step": 11501
    },
    {
      "epoch": 0.6263298686978007,
      "grad_norm": 0.6285254125685592,
      "learning_rate": 0.00016081257642829304,
      "loss": 12.1985,
      "step": 11502
    },
    {
      "epoch": 0.6263843226943837,
      "grad_norm": 0.5574983213776983,
      "learning_rate": 0.00016080557589963584,
      "loss": 12.4112,
      "step": 11503
    },
    {
      "epoch": 0.6264387766909668,
      "grad_norm": 0.5640310694002958,
      "learning_rate": 0.00016079857489814428,
      "loss": 12.1416,
      "step": 11504
    },
    {
      "epoch": 0.6264932306875498,
      "grad_norm": 0.5408204607168074,
      "learning_rate": 0.00016079157342387284,
      "loss": 12.3197,
      "step": 11505
    },
    {
      "epoch": 0.6265476846841328,
      "grad_norm": 0.7008687453180122,
      "learning_rate": 0.00016078457147687588,
      "loss": 12.3588,
      "step": 11506
    },
    {
      "epoch": 0.6266021386807158,
      "grad_norm": 0.6096810692741483,
      "learning_rate": 0.00016077756905720793,
      "loss": 12.3563,
      "step": 11507
    },
    {
      "epoch": 0.6266565926772988,
      "grad_norm": 0.6437700499832963,
      "learning_rate": 0.0001607705661649234,
      "loss": 12.4718,
      "step": 11508
    },
    {
      "epoch": 0.6267110466738818,
      "grad_norm": 0.6347956814809077,
      "learning_rate": 0.00016076356280007677,
      "loss": 12.4226,
      "step": 11509
    },
    {
      "epoch": 0.6267655006704649,
      "grad_norm": 0.5734954095097526,
      "learning_rate": 0.00016075655896272248,
      "loss": 12.2312,
      "step": 11510
    },
    {
      "epoch": 0.6268199546670479,
      "grad_norm": 0.6054178017659978,
      "learning_rate": 0.00016074955465291498,
      "loss": 12.2454,
      "step": 11511
    },
    {
      "epoch": 0.6268744086636309,
      "grad_norm": 0.5455242527056521,
      "learning_rate": 0.0001607425498707088,
      "loss": 12.2335,
      "step": 11512
    },
    {
      "epoch": 0.6269288626602139,
      "grad_norm": 0.6139827424512838,
      "learning_rate": 0.0001607355446161583,
      "loss": 12.2324,
      "step": 11513
    },
    {
      "epoch": 0.6269833166567969,
      "grad_norm": 0.604926511533672,
      "learning_rate": 0.00016072853888931808,
      "loss": 12.3358,
      "step": 11514
    },
    {
      "epoch": 0.6270377706533798,
      "grad_norm": 0.6322046476901865,
      "learning_rate": 0.00016072153269024254,
      "loss": 12.3039,
      "step": 11515
    },
    {
      "epoch": 0.627092224649963,
      "grad_norm": 0.5496567982649935,
      "learning_rate": 0.00016071452601898616,
      "loss": 12.3524,
      "step": 11516
    },
    {
      "epoch": 0.6271466786465459,
      "grad_norm": 0.6463799631048145,
      "learning_rate": 0.00016070751887560346,
      "loss": 12.3817,
      "step": 11517
    },
    {
      "epoch": 0.6272011326431289,
      "grad_norm": 0.554577880732961,
      "learning_rate": 0.0001607005112601489,
      "loss": 12.2941,
      "step": 11518
    },
    {
      "epoch": 0.6272555866397119,
      "grad_norm": 0.6290277561324147,
      "learning_rate": 0.00016069350317267697,
      "loss": 12.4027,
      "step": 11519
    },
    {
      "epoch": 0.6273100406362949,
      "grad_norm": 0.6559922146220869,
      "learning_rate": 0.0001606864946132422,
      "loss": 12.1408,
      "step": 11520
    },
    {
      "epoch": 0.6273644946328779,
      "grad_norm": 0.565898840105582,
      "learning_rate": 0.0001606794855818991,
      "loss": 12.313,
      "step": 11521
    },
    {
      "epoch": 0.627418948629461,
      "grad_norm": 0.6088149405508227,
      "learning_rate": 0.00016067247607870212,
      "loss": 12.4226,
      "step": 11522
    },
    {
      "epoch": 0.627473402626044,
      "grad_norm": 0.6121159921496703,
      "learning_rate": 0.00016066546610370578,
      "loss": 12.4291,
      "step": 11523
    },
    {
      "epoch": 0.627527856622627,
      "grad_norm": 0.6382194354578349,
      "learning_rate": 0.00016065845565696463,
      "loss": 12.3131,
      "step": 11524
    },
    {
      "epoch": 0.62758231061921,
      "grad_norm": 0.5665618795827646,
      "learning_rate": 0.00016065144473853313,
      "loss": 12.298,
      "step": 11525
    },
    {
      "epoch": 0.627636764615793,
      "grad_norm": 0.6252068636147795,
      "learning_rate": 0.00016064443334846585,
      "loss": 12.3406,
      "step": 11526
    },
    {
      "epoch": 0.627691218612376,
      "grad_norm": 0.6200796722836409,
      "learning_rate": 0.00016063742148681725,
      "loss": 12.2552,
      "step": 11527
    },
    {
      "epoch": 0.6277456726089591,
      "grad_norm": 0.6278916891273529,
      "learning_rate": 0.00016063040915364191,
      "loss": 12.3177,
      "step": 11528
    },
    {
      "epoch": 0.6278001266055421,
      "grad_norm": 0.7228648431792107,
      "learning_rate": 0.00016062339634899435,
      "loss": 12.28,
      "step": 11529
    },
    {
      "epoch": 0.6278545806021251,
      "grad_norm": 0.6419605641087148,
      "learning_rate": 0.00016061638307292907,
      "loss": 12.3749,
      "step": 11530
    },
    {
      "epoch": 0.6279090345987081,
      "grad_norm": 0.654333567110793,
      "learning_rate": 0.00016060936932550064,
      "loss": 12.4443,
      "step": 11531
    },
    {
      "epoch": 0.627963488595291,
      "grad_norm": 0.6178585671956981,
      "learning_rate": 0.0001606023551067636,
      "loss": 12.4102,
      "step": 11532
    },
    {
      "epoch": 0.6280179425918742,
      "grad_norm": 0.6668895511152603,
      "learning_rate": 0.0001605953404167725,
      "loss": 12.3039,
      "step": 11533
    },
    {
      "epoch": 0.6280723965884571,
      "grad_norm": 0.6314002451567844,
      "learning_rate": 0.00016058832525558186,
      "loss": 12.3504,
      "step": 11534
    },
    {
      "epoch": 0.6281268505850401,
      "grad_norm": 0.6920156557036148,
      "learning_rate": 0.0001605813096232462,
      "loss": 12.2574,
      "step": 11535
    },
    {
      "epoch": 0.6281813045816231,
      "grad_norm": 0.6072329811421321,
      "learning_rate": 0.00016057429351982013,
      "loss": 12.3304,
      "step": 11536
    },
    {
      "epoch": 0.6282357585782061,
      "grad_norm": 0.6583826115350645,
      "learning_rate": 0.00016056727694535824,
      "loss": 12.3934,
      "step": 11537
    },
    {
      "epoch": 0.6282902125747891,
      "grad_norm": 0.6324447219536025,
      "learning_rate": 0.000160560259899915,
      "loss": 12.3431,
      "step": 11538
    },
    {
      "epoch": 0.6283446665713722,
      "grad_norm": 0.6568949683285629,
      "learning_rate": 0.00016055324238354506,
      "loss": 12.2794,
      "step": 11539
    },
    {
      "epoch": 0.6283991205679552,
      "grad_norm": 0.6226100446959809,
      "learning_rate": 0.00016054622439630293,
      "loss": 12.3236,
      "step": 11540
    },
    {
      "epoch": 0.6284535745645382,
      "grad_norm": 0.5624371013426458,
      "learning_rate": 0.0001605392059382432,
      "loss": 12.3061,
      "step": 11541
    },
    {
      "epoch": 0.6285080285611212,
      "grad_norm": 0.7739453888229896,
      "learning_rate": 0.0001605321870094205,
      "loss": 12.2432,
      "step": 11542
    },
    {
      "epoch": 0.6285624825577042,
      "grad_norm": 0.6571712760851335,
      "learning_rate": 0.0001605251676098893,
      "loss": 12.3277,
      "step": 11543
    },
    {
      "epoch": 0.6286169365542872,
      "grad_norm": 0.659169789649659,
      "learning_rate": 0.00016051814773970427,
      "loss": 12.3759,
      "step": 11544
    },
    {
      "epoch": 0.6286713905508703,
      "grad_norm": 0.567263752452329,
      "learning_rate": 0.00016051112739891998,
      "loss": 12.1465,
      "step": 11545
    },
    {
      "epoch": 0.6287258445474533,
      "grad_norm": 0.584308563991564,
      "learning_rate": 0.00016050410658759103,
      "loss": 12.2037,
      "step": 11546
    },
    {
      "epoch": 0.6287802985440363,
      "grad_norm": 0.6845359198235154,
      "learning_rate": 0.000160497085305772,
      "loss": 12.4059,
      "step": 11547
    },
    {
      "epoch": 0.6288347525406193,
      "grad_norm": 0.5190582578308734,
      "learning_rate": 0.00016049006355351746,
      "loss": 12.2964,
      "step": 11548
    },
    {
      "epoch": 0.6288892065372023,
      "grad_norm": 0.6111561586009826,
      "learning_rate": 0.00016048304133088202,
      "loss": 12.2567,
      "step": 11549
    },
    {
      "epoch": 0.6289436605337853,
      "grad_norm": 0.6995756765034283,
      "learning_rate": 0.00016047601863792036,
      "loss": 12.0985,
      "step": 11550
    },
    {
      "epoch": 0.6289981145303684,
      "grad_norm": 0.6538510687733482,
      "learning_rate": 0.000160468995474687,
      "loss": 12.2661,
      "step": 11551
    },
    {
      "epoch": 0.6290525685269514,
      "grad_norm": 0.6233392511529413,
      "learning_rate": 0.00016046197184123667,
      "loss": 12.3554,
      "step": 11552
    },
    {
      "epoch": 0.6291070225235343,
      "grad_norm": 0.6012815233663004,
      "learning_rate": 0.00016045494773762382,
      "loss": 12.2265,
      "step": 11553
    },
    {
      "epoch": 0.6291614765201173,
      "grad_norm": 0.5931580882714241,
      "learning_rate": 0.0001604479231639032,
      "loss": 12.396,
      "step": 11554
    },
    {
      "epoch": 0.6292159305167003,
      "grad_norm": 0.6316027311980901,
      "learning_rate": 0.00016044089812012935,
      "loss": 12.2889,
      "step": 11555
    },
    {
      "epoch": 0.6292703845132833,
      "grad_norm": 0.7595009750544222,
      "learning_rate": 0.00016043387260635696,
      "loss": 12.5283,
      "step": 11556
    },
    {
      "epoch": 0.6293248385098664,
      "grad_norm": 0.5545253224480045,
      "learning_rate": 0.00016042684662264067,
      "loss": 12.2653,
      "step": 11557
    },
    {
      "epoch": 0.6293792925064494,
      "grad_norm": 0.598868192716635,
      "learning_rate": 0.00016041982016903506,
      "loss": 12.2521,
      "step": 11558
    },
    {
      "epoch": 0.6294337465030324,
      "grad_norm": 0.6504218467817201,
      "learning_rate": 0.00016041279324559483,
      "loss": 12.3431,
      "step": 11559
    },
    {
      "epoch": 0.6294882004996154,
      "grad_norm": 0.5507058063814501,
      "learning_rate": 0.00016040576585237454,
      "loss": 12.3622,
      "step": 11560
    },
    {
      "epoch": 0.6295426544961984,
      "grad_norm": 0.5483976461810933,
      "learning_rate": 0.00016039873798942892,
      "loss": 12.2512,
      "step": 11561
    },
    {
      "epoch": 0.6295971084927814,
      "grad_norm": 0.5704138722303417,
      "learning_rate": 0.00016039170965681255,
      "loss": 12.3039,
      "step": 11562
    },
    {
      "epoch": 0.6296515624893645,
      "grad_norm": 0.5907463783061986,
      "learning_rate": 0.00016038468085458014,
      "loss": 12.262,
      "step": 11563
    },
    {
      "epoch": 0.6297060164859475,
      "grad_norm": 0.6172033679213188,
      "learning_rate": 0.00016037765158278636,
      "loss": 12.3686,
      "step": 11564
    },
    {
      "epoch": 0.6297604704825305,
      "grad_norm": 0.6291212721767191,
      "learning_rate": 0.00016037062184148576,
      "loss": 12.2868,
      "step": 11565
    },
    {
      "epoch": 0.6298149244791135,
      "grad_norm": 0.5517508471945034,
      "learning_rate": 0.0001603635916307331,
      "loss": 12.1767,
      "step": 11566
    },
    {
      "epoch": 0.6298693784756965,
      "grad_norm": 0.5935989273602583,
      "learning_rate": 0.00016035656095058308,
      "loss": 12.4345,
      "step": 11567
    },
    {
      "epoch": 0.6299238324722796,
      "grad_norm": 0.6190845781659723,
      "learning_rate": 0.0001603495298010903,
      "loss": 12.3404,
      "step": 11568
    },
    {
      "epoch": 0.6299782864688626,
      "grad_norm": 0.5894594749346714,
      "learning_rate": 0.00016034249818230943,
      "loss": 12.2778,
      "step": 11569
    },
    {
      "epoch": 0.6300327404654456,
      "grad_norm": 0.70588711193054,
      "learning_rate": 0.0001603354660942952,
      "loss": 12.291,
      "step": 11570
    },
    {
      "epoch": 0.6300871944620285,
      "grad_norm": 0.6115013551554997,
      "learning_rate": 0.00016032843353710224,
      "loss": 12.2606,
      "step": 11571
    },
    {
      "epoch": 0.6301416484586115,
      "grad_norm": 0.681756775481514,
      "learning_rate": 0.00016032140051078532,
      "loss": 12.402,
      "step": 11572
    },
    {
      "epoch": 0.6301961024551945,
      "grad_norm": 0.7152374060110404,
      "learning_rate": 0.000160314367015399,
      "loss": 12.3076,
      "step": 11573
    },
    {
      "epoch": 0.6302505564517776,
      "grad_norm": 0.6401385865783593,
      "learning_rate": 0.0001603073330509981,
      "loss": 12.2946,
      "step": 11574
    },
    {
      "epoch": 0.6303050104483606,
      "grad_norm": 0.7067366584131319,
      "learning_rate": 0.00016030029861763723,
      "loss": 12.3269,
      "step": 11575
    },
    {
      "epoch": 0.6303594644449436,
      "grad_norm": 0.5837445863525274,
      "learning_rate": 0.00016029326371537115,
      "loss": 12.2874,
      "step": 11576
    },
    {
      "epoch": 0.6304139184415266,
      "grad_norm": 0.6664881012459636,
      "learning_rate": 0.00016028622834425455,
      "loss": 12.4066,
      "step": 11577
    },
    {
      "epoch": 0.6304683724381096,
      "grad_norm": 0.5732167094095257,
      "learning_rate": 0.0001602791925043421,
      "loss": 12.2736,
      "step": 11578
    },
    {
      "epoch": 0.6305228264346926,
      "grad_norm": 0.5701728735843571,
      "learning_rate": 0.00016027215619568853,
      "loss": 12.3236,
      "step": 11579
    },
    {
      "epoch": 0.6305772804312757,
      "grad_norm": 0.6263392417410668,
      "learning_rate": 0.00016026511941834862,
      "loss": 12.3113,
      "step": 11580
    },
    {
      "epoch": 0.6306317344278587,
      "grad_norm": 0.5986858779759879,
      "learning_rate": 0.00016025808217237696,
      "loss": 12.4141,
      "step": 11581
    },
    {
      "epoch": 0.6306861884244417,
      "grad_norm": 0.6120558822405074,
      "learning_rate": 0.0001602510444578284,
      "loss": 12.3433,
      "step": 11582
    },
    {
      "epoch": 0.6307406424210247,
      "grad_norm": 0.5997141772225506,
      "learning_rate": 0.00016024400627475763,
      "loss": 12.2342,
      "step": 11583
    },
    {
      "epoch": 0.6307950964176077,
      "grad_norm": 0.6046918603331474,
      "learning_rate": 0.00016023696762321933,
      "loss": 12.4326,
      "step": 11584
    },
    {
      "epoch": 0.6308495504141907,
      "grad_norm": 0.7145511224984784,
      "learning_rate": 0.0001602299285032683,
      "loss": 12.4418,
      "step": 11585
    },
    {
      "epoch": 0.6309040044107738,
      "grad_norm": 0.5674460638373898,
      "learning_rate": 0.00016022288891495918,
      "loss": 12.3899,
      "step": 11586
    },
    {
      "epoch": 0.6309584584073568,
      "grad_norm": 0.6078663393707878,
      "learning_rate": 0.00016021584885834682,
      "loss": 12.3836,
      "step": 11587
    },
    {
      "epoch": 0.6310129124039398,
      "grad_norm": 0.6731437267177017,
      "learning_rate": 0.00016020880833348593,
      "loss": 12.4028,
      "step": 11588
    },
    {
      "epoch": 0.6310673664005227,
      "grad_norm": 0.7464619177043753,
      "learning_rate": 0.00016020176734043125,
      "loss": 12.3038,
      "step": 11589
    },
    {
      "epoch": 0.6311218203971057,
      "grad_norm": 0.6604754928634954,
      "learning_rate": 0.0001601947258792375,
      "loss": 12.3084,
      "step": 11590
    },
    {
      "epoch": 0.6311762743936887,
      "grad_norm": 0.625782172031097,
      "learning_rate": 0.00016018768394995947,
      "loss": 12.319,
      "step": 11591
    },
    {
      "epoch": 0.6312307283902718,
      "grad_norm": 0.6305290661457754,
      "learning_rate": 0.00016018064155265196,
      "loss": 12.3992,
      "step": 11592
    },
    {
      "epoch": 0.6312851823868548,
      "grad_norm": 0.6876046735716144,
      "learning_rate": 0.00016017359868736964,
      "loss": 12.3389,
      "step": 11593
    },
    {
      "epoch": 0.6313396363834378,
      "grad_norm": 0.6017411391397692,
      "learning_rate": 0.00016016655535416735,
      "loss": 12.3124,
      "step": 11594
    },
    {
      "epoch": 0.6313940903800208,
      "grad_norm": 0.5489052272328483,
      "learning_rate": 0.00016015951155309982,
      "loss": 12.4313,
      "step": 11595
    },
    {
      "epoch": 0.6314485443766038,
      "grad_norm": 0.5643980944234432,
      "learning_rate": 0.00016015246728422186,
      "loss": 12.3021,
      "step": 11596
    },
    {
      "epoch": 0.6315029983731868,
      "grad_norm": 0.6283739932043341,
      "learning_rate": 0.00016014542254758825,
      "loss": 12.2458,
      "step": 11597
    },
    {
      "epoch": 0.6315574523697699,
      "grad_norm": 0.6559159408597222,
      "learning_rate": 0.0001601383773432537,
      "loss": 12.3632,
      "step": 11598
    },
    {
      "epoch": 0.6316119063663529,
      "grad_norm": 0.5743836786873061,
      "learning_rate": 0.00016013133167127306,
      "loss": 12.28,
      "step": 11599
    },
    {
      "epoch": 0.6316663603629359,
      "grad_norm": 0.6037395609902132,
      "learning_rate": 0.0001601242855317011,
      "loss": 12.2813,
      "step": 11600
    },
    {
      "epoch": 0.6317208143595189,
      "grad_norm": 0.8138665820981366,
      "learning_rate": 0.0001601172389245926,
      "loss": 12.3117,
      "step": 11601
    },
    {
      "epoch": 0.6317752683561019,
      "grad_norm": 0.6117045153426184,
      "learning_rate": 0.00016011019185000237,
      "loss": 12.277,
      "step": 11602
    },
    {
      "epoch": 0.631829722352685,
      "grad_norm": 0.6571968896939481,
      "learning_rate": 0.00016010314430798522,
      "loss": 12.3114,
      "step": 11603
    },
    {
      "epoch": 0.631884176349268,
      "grad_norm": 0.5819354329428992,
      "learning_rate": 0.00016009609629859598,
      "loss": 12.3988,
      "step": 11604
    },
    {
      "epoch": 0.631938630345851,
      "grad_norm": 0.6130792154075189,
      "learning_rate": 0.00016008904782188936,
      "loss": 12.3442,
      "step": 11605
    },
    {
      "epoch": 0.631993084342434,
      "grad_norm": 0.6161087981605953,
      "learning_rate": 0.00016008199887792022,
      "loss": 12.1447,
      "step": 11606
    },
    {
      "epoch": 0.632047538339017,
      "grad_norm": 0.59213108698951,
      "learning_rate": 0.0001600749494667434,
      "loss": 12.4355,
      "step": 11607
    },
    {
      "epoch": 0.6321019923355999,
      "grad_norm": 0.5643485490047897,
      "learning_rate": 0.00016006789958841373,
      "loss": 12.3072,
      "step": 11608
    },
    {
      "epoch": 0.632156446332183,
      "grad_norm": 0.6552253773174634,
      "learning_rate": 0.00016006084924298597,
      "loss": 12.29,
      "step": 11609
    },
    {
      "epoch": 0.632210900328766,
      "grad_norm": 0.5629019501817991,
      "learning_rate": 0.000160053798430515,
      "loss": 12.3558,
      "step": 11610
    },
    {
      "epoch": 0.632265354325349,
      "grad_norm": 0.6248217978673832,
      "learning_rate": 0.00016004674715105558,
      "loss": 12.1634,
      "step": 11611
    },
    {
      "epoch": 0.632319808321932,
      "grad_norm": 0.7052249240995686,
      "learning_rate": 0.0001600396954046626,
      "loss": 12.4626,
      "step": 11612
    },
    {
      "epoch": 0.632374262318515,
      "grad_norm": 0.6229537066548735,
      "learning_rate": 0.00016003264319139088,
      "loss": 12.2556,
      "step": 11613
    },
    {
      "epoch": 0.632428716315098,
      "grad_norm": 0.6232155182387574,
      "learning_rate": 0.0001600255905112953,
      "loss": 12.3656,
      "step": 11614
    },
    {
      "epoch": 0.6324831703116811,
      "grad_norm": 0.5985385901002135,
      "learning_rate": 0.0001600185373644306,
      "loss": 12.2761,
      "step": 11615
    },
    {
      "epoch": 0.6325376243082641,
      "grad_norm": 0.5804199678921629,
      "learning_rate": 0.00016001148375085168,
      "loss": 12.335,
      "step": 11616
    },
    {
      "epoch": 0.6325920783048471,
      "grad_norm": 0.6273430482799724,
      "learning_rate": 0.00016000442967061346,
      "loss": 12.2817,
      "step": 11617
    },
    {
      "epoch": 0.6326465323014301,
      "grad_norm": 0.62391369595692,
      "learning_rate": 0.00015999737512377072,
      "loss": 12.1001,
      "step": 11618
    },
    {
      "epoch": 0.6327009862980131,
      "grad_norm": 0.6135274564031256,
      "learning_rate": 0.0001599903201103783,
      "loss": 12.3653,
      "step": 11619
    },
    {
      "epoch": 0.6327554402945961,
      "grad_norm": 0.5769543408166947,
      "learning_rate": 0.0001599832646304911,
      "loss": 12.2763,
      "step": 11620
    },
    {
      "epoch": 0.6328098942911792,
      "grad_norm": 0.6553316798350617,
      "learning_rate": 0.00015997620868416396,
      "loss": 12.3955,
      "step": 11621
    },
    {
      "epoch": 0.6328643482877622,
      "grad_norm": 0.6165252378482168,
      "learning_rate": 0.00015996915227145178,
      "loss": 12.3882,
      "step": 11622
    },
    {
      "epoch": 0.6329188022843452,
      "grad_norm": 0.578197074875526,
      "learning_rate": 0.00015996209539240942,
      "loss": 12.2141,
      "step": 11623
    },
    {
      "epoch": 0.6329732562809282,
      "grad_norm": 0.6546997218876552,
      "learning_rate": 0.00015995503804709175,
      "loss": 12.2659,
      "step": 11624
    },
    {
      "epoch": 0.6330277102775111,
      "grad_norm": 0.5346845777961047,
      "learning_rate": 0.00015994798023555363,
      "loss": 12.4347,
      "step": 11625
    },
    {
      "epoch": 0.6330821642740941,
      "grad_norm": 0.7272621472840548,
      "learning_rate": 0.00015994092195785,
      "loss": 12.3033,
      "step": 11626
    },
    {
      "epoch": 0.6331366182706772,
      "grad_norm": 0.5912965473134938,
      "learning_rate": 0.0001599338632140357,
      "loss": 12.3621,
      "step": 11627
    },
    {
      "epoch": 0.6331910722672602,
      "grad_norm": 0.6084544466595607,
      "learning_rate": 0.0001599268040041656,
      "loss": 12.3431,
      "step": 11628
    },
    {
      "epoch": 0.6332455262638432,
      "grad_norm": 0.6195207017613764,
      "learning_rate": 0.00015991974432829468,
      "loss": 12.2912,
      "step": 11629
    },
    {
      "epoch": 0.6332999802604262,
      "grad_norm": 0.6689588330255158,
      "learning_rate": 0.00015991268418647772,
      "loss": 12.2959,
      "step": 11630
    },
    {
      "epoch": 0.6333544342570092,
      "grad_norm": 0.566247369536983,
      "learning_rate": 0.00015990562357876968,
      "loss": 12.2669,
      "step": 11631
    },
    {
      "epoch": 0.6334088882535923,
      "grad_norm": 0.7277889233888325,
      "learning_rate": 0.00015989856250522548,
      "loss": 12.4804,
      "step": 11632
    },
    {
      "epoch": 0.6334633422501753,
      "grad_norm": 0.6384352082662375,
      "learning_rate": 0.00015989150096590003,
      "loss": 12.3755,
      "step": 11633
    },
    {
      "epoch": 0.6335177962467583,
      "grad_norm": 0.7476197942467356,
      "learning_rate": 0.00015988443896084822,
      "loss": 12.3098,
      "step": 11634
    },
    {
      "epoch": 0.6335722502433413,
      "grad_norm": 0.6117565640338275,
      "learning_rate": 0.00015987737649012497,
      "loss": 12.2888,
      "step": 11635
    },
    {
      "epoch": 0.6336267042399243,
      "grad_norm": 0.5877500335920101,
      "learning_rate": 0.00015987031355378518,
      "loss": 12.3271,
      "step": 11636
    },
    {
      "epoch": 0.6336811582365073,
      "grad_norm": 0.7537103433883942,
      "learning_rate": 0.0001598632501518838,
      "loss": 12.3379,
      "step": 11637
    },
    {
      "epoch": 0.6337356122330904,
      "grad_norm": 0.562326931604982,
      "learning_rate": 0.00015985618628447577,
      "loss": 12.1521,
      "step": 11638
    },
    {
      "epoch": 0.6337900662296734,
      "grad_norm": 0.6720970545937142,
      "learning_rate": 0.00015984912195161595,
      "loss": 12.2289,
      "step": 11639
    },
    {
      "epoch": 0.6338445202262564,
      "grad_norm": 0.7437459848399414,
      "learning_rate": 0.00015984205715335935,
      "loss": 12.2812,
      "step": 11640
    },
    {
      "epoch": 0.6338989742228394,
      "grad_norm": 0.5977958341585544,
      "learning_rate": 0.00015983499188976087,
      "loss": 12.3845,
      "step": 11641
    },
    {
      "epoch": 0.6339534282194224,
      "grad_norm": 0.6642266351088731,
      "learning_rate": 0.00015982792616087545,
      "loss": 12.3436,
      "step": 11642
    },
    {
      "epoch": 0.6340078822160053,
      "grad_norm": 0.6253891353430321,
      "learning_rate": 0.0001598208599667581,
      "loss": 12.271,
      "step": 11643
    },
    {
      "epoch": 0.6340623362125885,
      "grad_norm": 0.6099370956862907,
      "learning_rate": 0.00015981379330746363,
      "loss": 12.3552,
      "step": 11644
    },
    {
      "epoch": 0.6341167902091714,
      "grad_norm": 0.6318726619121243,
      "learning_rate": 0.0001598067261830471,
      "loss": 12.2443,
      "step": 11645
    },
    {
      "epoch": 0.6341712442057544,
      "grad_norm": 0.5609364407721689,
      "learning_rate": 0.00015979965859356347,
      "loss": 12.4605,
      "step": 11646
    },
    {
      "epoch": 0.6342256982023374,
      "grad_norm": 0.702645292105402,
      "learning_rate": 0.00015979259053906764,
      "loss": 12.3523,
      "step": 11647
    },
    {
      "epoch": 0.6342801521989204,
      "grad_norm": 0.6312822853043644,
      "learning_rate": 0.00015978552201961464,
      "loss": 12.3942,
      "step": 11648
    },
    {
      "epoch": 0.6343346061955034,
      "grad_norm": 0.6827895428823186,
      "learning_rate": 0.00015977845303525933,
      "loss": 12.3318,
      "step": 11649
    },
    {
      "epoch": 0.6343890601920865,
      "grad_norm": 0.5633296696619046,
      "learning_rate": 0.00015977138358605676,
      "loss": 12.3435,
      "step": 11650
    },
    {
      "epoch": 0.6344435141886695,
      "grad_norm": 0.6803882086394497,
      "learning_rate": 0.00015976431367206191,
      "loss": 12.2223,
      "step": 11651
    },
    {
      "epoch": 0.6344979681852525,
      "grad_norm": 0.5938162328387466,
      "learning_rate": 0.00015975724329332972,
      "loss": 12.2128,
      "step": 11652
    },
    {
      "epoch": 0.6345524221818355,
      "grad_norm": 0.6310709849247171,
      "learning_rate": 0.00015975017244991522,
      "loss": 12.3975,
      "step": 11653
    },
    {
      "epoch": 0.6346068761784185,
      "grad_norm": 0.6423715628698327,
      "learning_rate": 0.0001597431011418733,
      "loss": 12.3256,
      "step": 11654
    },
    {
      "epoch": 0.6346613301750015,
      "grad_norm": 0.6112767294965891,
      "learning_rate": 0.00015973602936925904,
      "loss": 12.2323,
      "step": 11655
    },
    {
      "epoch": 0.6347157841715846,
      "grad_norm": 0.63268138600798,
      "learning_rate": 0.0001597289571321274,
      "loss": 12.2258,
      "step": 11656
    },
    {
      "epoch": 0.6347702381681676,
      "grad_norm": 0.6132891422810992,
      "learning_rate": 0.00015972188443053335,
      "loss": 12.2712,
      "step": 11657
    },
    {
      "epoch": 0.6348246921647506,
      "grad_norm": 0.6007827090687096,
      "learning_rate": 0.00015971481126453196,
      "loss": 12.3272,
      "step": 11658
    },
    {
      "epoch": 0.6348791461613336,
      "grad_norm": 0.5534845093934893,
      "learning_rate": 0.00015970773763417814,
      "loss": 12.3964,
      "step": 11659
    },
    {
      "epoch": 0.6349336001579166,
      "grad_norm": 0.5471382906969506,
      "learning_rate": 0.00015970066353952696,
      "loss": 12.2745,
      "step": 11660
    },
    {
      "epoch": 0.6349880541544995,
      "grad_norm": 0.5633804601387051,
      "learning_rate": 0.0001596935889806334,
      "loss": 12.3712,
      "step": 11661
    },
    {
      "epoch": 0.6350425081510827,
      "grad_norm": 0.5419783900118207,
      "learning_rate": 0.00015968651395755246,
      "loss": 12.2674,
      "step": 11662
    },
    {
      "epoch": 0.6350969621476656,
      "grad_norm": 0.6110092220859774,
      "learning_rate": 0.00015967943847033922,
      "loss": 12.2362,
      "step": 11663
    },
    {
      "epoch": 0.6351514161442486,
      "grad_norm": 0.5929826979767274,
      "learning_rate": 0.00015967236251904863,
      "loss": 12.3111,
      "step": 11664
    },
    {
      "epoch": 0.6352058701408316,
      "grad_norm": 0.6319140215443649,
      "learning_rate": 0.00015966528610373572,
      "loss": 12.2552,
      "step": 11665
    },
    {
      "epoch": 0.6352603241374146,
      "grad_norm": 0.6250575551924299,
      "learning_rate": 0.00015965820922445557,
      "loss": 12.1944,
      "step": 11666
    },
    {
      "epoch": 0.6353147781339977,
      "grad_norm": 0.634596741902867,
      "learning_rate": 0.0001596511318812632,
      "loss": 12.4088,
      "step": 11667
    },
    {
      "epoch": 0.6353692321305807,
      "grad_norm": 0.5708759498126883,
      "learning_rate": 0.00015964405407421358,
      "loss": 12.4004,
      "step": 11668
    },
    {
      "epoch": 0.6354236861271637,
      "grad_norm": 0.6274291085104535,
      "learning_rate": 0.00015963697580336183,
      "loss": 12.3656,
      "step": 11669
    },
    {
      "epoch": 0.6354781401237467,
      "grad_norm": 0.6381275977567078,
      "learning_rate": 0.00015962989706876294,
      "loss": 12.282,
      "step": 11670
    },
    {
      "epoch": 0.6355325941203297,
      "grad_norm": 0.5877263890706275,
      "learning_rate": 0.00015962281787047197,
      "loss": 12.4272,
      "step": 11671
    },
    {
      "epoch": 0.6355870481169127,
      "grad_norm": 0.5476695346740104,
      "learning_rate": 0.00015961573820854396,
      "loss": 12.2424,
      "step": 11672
    },
    {
      "epoch": 0.6356415021134958,
      "grad_norm": 0.5647625962262167,
      "learning_rate": 0.000159608658083034,
      "loss": 12.2057,
      "step": 11673
    },
    {
      "epoch": 0.6356959561100788,
      "grad_norm": 0.6079880681478522,
      "learning_rate": 0.0001596015774939971,
      "loss": 12.4301,
      "step": 11674
    },
    {
      "epoch": 0.6357504101066618,
      "grad_norm": 0.627898353169781,
      "learning_rate": 0.00015959449644148833,
      "loss": 12.2848,
      "step": 11675
    },
    {
      "epoch": 0.6358048641032448,
      "grad_norm": 0.6000947735191888,
      "learning_rate": 0.00015958741492556278,
      "loss": 12.5301,
      "step": 11676
    },
    {
      "epoch": 0.6358593180998278,
      "grad_norm": 0.6383110683869572,
      "learning_rate": 0.0001595803329462755,
      "loss": 12.3761,
      "step": 11677
    },
    {
      "epoch": 0.6359137720964108,
      "grad_norm": 0.6226714343986417,
      "learning_rate": 0.00015957325050368156,
      "loss": 12.2989,
      "step": 11678
    },
    {
      "epoch": 0.6359682260929939,
      "grad_norm": 0.5994882158830673,
      "learning_rate": 0.00015956616759783604,
      "loss": 12.4552,
      "step": 11679
    },
    {
      "epoch": 0.6360226800895769,
      "grad_norm": 0.6428334737207511,
      "learning_rate": 0.00015955908422879395,
      "loss": 12.2438,
      "step": 11680
    },
    {
      "epoch": 0.6360771340861598,
      "grad_norm": 0.6423154220996011,
      "learning_rate": 0.00015955200039661049,
      "loss": 12.4005,
      "step": 11681
    },
    {
      "epoch": 0.6361315880827428,
      "grad_norm": 0.5486200532788184,
      "learning_rate": 0.00015954491610134066,
      "loss": 12.3216,
      "step": 11682
    },
    {
      "epoch": 0.6361860420793258,
      "grad_norm": 0.6309935704318574,
      "learning_rate": 0.00015953783134303962,
      "loss": 12.2927,
      "step": 11683
    },
    {
      "epoch": 0.6362404960759088,
      "grad_norm": 0.5824870916214382,
      "learning_rate": 0.00015953074612176237,
      "loss": 12.3833,
      "step": 11684
    },
    {
      "epoch": 0.6362949500724919,
      "grad_norm": 0.6877518652979272,
      "learning_rate": 0.0001595236604375641,
      "loss": 12.3747,
      "step": 11685
    },
    {
      "epoch": 0.6363494040690749,
      "grad_norm": 0.6248540007950087,
      "learning_rate": 0.00015951657429049982,
      "loss": 12.4178,
      "step": 11686
    },
    {
      "epoch": 0.6364038580656579,
      "grad_norm": 0.5595550579826128,
      "learning_rate": 0.0001595094876806247,
      "loss": 12.3002,
      "step": 11687
    },
    {
      "epoch": 0.6364583120622409,
      "grad_norm": 0.5609830813350578,
      "learning_rate": 0.00015950240060799383,
      "loss": 12.1729,
      "step": 11688
    },
    {
      "epoch": 0.6365127660588239,
      "grad_norm": 0.6840906942224464,
      "learning_rate": 0.00015949531307266233,
      "loss": 12.2164,
      "step": 11689
    },
    {
      "epoch": 0.6365672200554069,
      "grad_norm": 0.6102936461713946,
      "learning_rate": 0.00015948822507468526,
      "loss": 12.2205,
      "step": 11690
    },
    {
      "epoch": 0.63662167405199,
      "grad_norm": 0.7109447118428419,
      "learning_rate": 0.00015948113661411778,
      "loss": 12.2895,
      "step": 11691
    },
    {
      "epoch": 0.636676128048573,
      "grad_norm": 0.5232082210855122,
      "learning_rate": 0.000159474047691015,
      "loss": 12.3388,
      "step": 11692
    },
    {
      "epoch": 0.636730582045156,
      "grad_norm": 0.6181267829552438,
      "learning_rate": 0.0001594669583054321,
      "loss": 12.4013,
      "step": 11693
    },
    {
      "epoch": 0.636785036041739,
      "grad_norm": 0.5853241566666262,
      "learning_rate": 0.00015945986845742414,
      "loss": 12.2416,
      "step": 11694
    },
    {
      "epoch": 0.636839490038322,
      "grad_norm": 0.6440508892552266,
      "learning_rate": 0.00015945277814704623,
      "loss": 12.3312,
      "step": 11695
    },
    {
      "epoch": 0.636893944034905,
      "grad_norm": 0.5764079526694624,
      "learning_rate": 0.0001594456873743536,
      "loss": 12.3378,
      "step": 11696
    },
    {
      "epoch": 0.6369483980314881,
      "grad_norm": 0.5912901229404963,
      "learning_rate": 0.0001594385961394013,
      "loss": 12.2194,
      "step": 11697
    },
    {
      "epoch": 0.637002852028071,
      "grad_norm": 0.6094323269680411,
      "learning_rate": 0.00015943150444224454,
      "loss": 12.319,
      "step": 11698
    },
    {
      "epoch": 0.637057306024654,
      "grad_norm": 0.5635457642014866,
      "learning_rate": 0.0001594244122829384,
      "loss": 12.3567,
      "step": 11699
    },
    {
      "epoch": 0.637111760021237,
      "grad_norm": 0.587726048800709,
      "learning_rate": 0.00015941731966153807,
      "loss": 12.3092,
      "step": 11700
    },
    {
      "epoch": 0.63716621401782,
      "grad_norm": 0.6048198304309254,
      "learning_rate": 0.0001594102265780987,
      "loss": 12.3443,
      "step": 11701
    },
    {
      "epoch": 0.6372206680144031,
      "grad_norm": 0.5849478044884568,
      "learning_rate": 0.00015940313303267547,
      "loss": 12.3281,
      "step": 11702
    },
    {
      "epoch": 0.6372751220109861,
      "grad_norm": 0.6299700554928337,
      "learning_rate": 0.0001593960390253235,
      "loss": 12.3386,
      "step": 11703
    },
    {
      "epoch": 0.6373295760075691,
      "grad_norm": 0.7091301677395995,
      "learning_rate": 0.00015938894455609797,
      "loss": 12.2443,
      "step": 11704
    },
    {
      "epoch": 0.6373840300041521,
      "grad_norm": 0.6079507577811608,
      "learning_rate": 0.00015938184962505404,
      "loss": 12.2513,
      "step": 11705
    },
    {
      "epoch": 0.6374384840007351,
      "grad_norm": 0.6279045948807624,
      "learning_rate": 0.00015937475423224688,
      "loss": 12.3791,
      "step": 11706
    },
    {
      "epoch": 0.6374929379973181,
      "grad_norm": 0.5422081747141756,
      "learning_rate": 0.00015936765837773172,
      "loss": 12.2061,
      "step": 11707
    },
    {
      "epoch": 0.6375473919939012,
      "grad_norm": 0.5599573109726491,
      "learning_rate": 0.00015936056206156365,
      "loss": 12.1982,
      "step": 11708
    },
    {
      "epoch": 0.6376018459904842,
      "grad_norm": 0.6439053391485484,
      "learning_rate": 0.00015935346528379792,
      "loss": 12.3223,
      "step": 11709
    },
    {
      "epoch": 0.6376562999870672,
      "grad_norm": 0.6076777657170431,
      "learning_rate": 0.00015934636804448968,
      "loss": 12.2359,
      "step": 11710
    },
    {
      "epoch": 0.6377107539836502,
      "grad_norm": 0.6062940962482165,
      "learning_rate": 0.0001593392703436941,
      "loss": 12.2789,
      "step": 11711
    },
    {
      "epoch": 0.6377652079802332,
      "grad_norm": 0.6090446598951548,
      "learning_rate": 0.00015933217218146644,
      "loss": 12.3359,
      "step": 11712
    },
    {
      "epoch": 0.6378196619768162,
      "grad_norm": 0.5609615308756994,
      "learning_rate": 0.00015932507355786182,
      "loss": 12.3445,
      "step": 11713
    },
    {
      "epoch": 0.6378741159733993,
      "grad_norm": 0.6507111609328309,
      "learning_rate": 0.00015931797447293552,
      "loss": 12.255,
      "step": 11714
    },
    {
      "epoch": 0.6379285699699823,
      "grad_norm": 0.683631714011919,
      "learning_rate": 0.0001593108749267427,
      "loss": 12.21,
      "step": 11715
    },
    {
      "epoch": 0.6379830239665653,
      "grad_norm": 0.5426813755368318,
      "learning_rate": 0.00015930377491933854,
      "loss": 12.2444,
      "step": 11716
    },
    {
      "epoch": 0.6380374779631482,
      "grad_norm": 0.6285851116986505,
      "learning_rate": 0.0001592966744507783,
      "loss": 12.4862,
      "step": 11717
    },
    {
      "epoch": 0.6380919319597312,
      "grad_norm": 0.6077576905377188,
      "learning_rate": 0.0001592895735211172,
      "loss": 12.3795,
      "step": 11718
    },
    {
      "epoch": 0.6381463859563142,
      "grad_norm": 0.641708820048512,
      "learning_rate": 0.0001592824721304104,
      "loss": 12.2624,
      "step": 11719
    },
    {
      "epoch": 0.6382008399528973,
      "grad_norm": 0.5832512250990588,
      "learning_rate": 0.00015927537027871316,
      "loss": 12.4107,
      "step": 11720
    },
    {
      "epoch": 0.6382552939494803,
      "grad_norm": 0.5560870907638484,
      "learning_rate": 0.00015926826796608073,
      "loss": 12.2068,
      "step": 11721
    },
    {
      "epoch": 0.6383097479460633,
      "grad_norm": 0.621569606432972,
      "learning_rate": 0.00015926116519256826,
      "loss": 12.1528,
      "step": 11722
    },
    {
      "epoch": 0.6383642019426463,
      "grad_norm": 0.6470496143775933,
      "learning_rate": 0.00015925406195823108,
      "loss": 12.2376,
      "step": 11723
    },
    {
      "epoch": 0.6384186559392293,
      "grad_norm": 0.5339169789036717,
      "learning_rate": 0.00015924695826312435,
      "loss": 12.3241,
      "step": 11724
    },
    {
      "epoch": 0.6384731099358123,
      "grad_norm": 0.5973303793387309,
      "learning_rate": 0.00015923985410730334,
      "loss": 12.3649,
      "step": 11725
    },
    {
      "epoch": 0.6385275639323954,
      "grad_norm": 0.6828561520479584,
      "learning_rate": 0.00015923274949082328,
      "loss": 12.4293,
      "step": 11726
    },
    {
      "epoch": 0.6385820179289784,
      "grad_norm": 0.6080382653576513,
      "learning_rate": 0.00015922564441373945,
      "loss": 12.3078,
      "step": 11727
    },
    {
      "epoch": 0.6386364719255614,
      "grad_norm": 0.6970886252419852,
      "learning_rate": 0.00015921853887610707,
      "loss": 12.3816,
      "step": 11728
    },
    {
      "epoch": 0.6386909259221444,
      "grad_norm": 0.5930463408717045,
      "learning_rate": 0.00015921143287798138,
      "loss": 12.4402,
      "step": 11729
    },
    {
      "epoch": 0.6387453799187274,
      "grad_norm": 0.5680002960969729,
      "learning_rate": 0.00015920432641941768,
      "loss": 12.455,
      "step": 11730
    },
    {
      "epoch": 0.6387998339153104,
      "grad_norm": 0.5601685294026225,
      "learning_rate": 0.00015919721950047119,
      "loss": 12.3696,
      "step": 11731
    },
    {
      "epoch": 0.6388542879118935,
      "grad_norm": 0.5913415111135867,
      "learning_rate": 0.0001591901121211972,
      "loss": 12.2751,
      "step": 11732
    },
    {
      "epoch": 0.6389087419084765,
      "grad_norm": 0.559035357315235,
      "learning_rate": 0.00015918300428165099,
      "loss": 12.3311,
      "step": 11733
    },
    {
      "epoch": 0.6389631959050595,
      "grad_norm": 0.7960160680407351,
      "learning_rate": 0.0001591758959818878,
      "loss": 12.561,
      "step": 11734
    },
    {
      "epoch": 0.6390176499016424,
      "grad_norm": 0.5580073249870375,
      "learning_rate": 0.00015916878722196291,
      "loss": 12.3125,
      "step": 11735
    },
    {
      "epoch": 0.6390721038982254,
      "grad_norm": 0.5980153973018723,
      "learning_rate": 0.00015916167800193162,
      "loss": 12.2887,
      "step": 11736
    },
    {
      "epoch": 0.6391265578948085,
      "grad_norm": 0.6145853706411524,
      "learning_rate": 0.00015915456832184922,
      "loss": 12.3097,
      "step": 11737
    },
    {
      "epoch": 0.6391810118913915,
      "grad_norm": 0.5761368251670567,
      "learning_rate": 0.00015914745818177095,
      "loss": 12.1488,
      "step": 11738
    },
    {
      "epoch": 0.6392354658879745,
      "grad_norm": 0.5720363711923273,
      "learning_rate": 0.00015914034758175211,
      "loss": 12.3468,
      "step": 11739
    },
    {
      "epoch": 0.6392899198845575,
      "grad_norm": 0.5730666501611782,
      "learning_rate": 0.00015913323652184803,
      "loss": 12.2901,
      "step": 11740
    },
    {
      "epoch": 0.6393443738811405,
      "grad_norm": 0.5967192597957882,
      "learning_rate": 0.000159126125002114,
      "loss": 12.133,
      "step": 11741
    },
    {
      "epoch": 0.6393988278777235,
      "grad_norm": 0.5847137465942532,
      "learning_rate": 0.00015911901302260528,
      "loss": 12.2619,
      "step": 11742
    },
    {
      "epoch": 0.6394532818743066,
      "grad_norm": 0.6527293239756213,
      "learning_rate": 0.00015911190058337722,
      "loss": 12.3478,
      "step": 11743
    },
    {
      "epoch": 0.6395077358708896,
      "grad_norm": 0.6833711071046793,
      "learning_rate": 0.0001591047876844851,
      "loss": 12.3742,
      "step": 11744
    },
    {
      "epoch": 0.6395621898674726,
      "grad_norm": 0.6587852445363322,
      "learning_rate": 0.00015909767432598426,
      "loss": 12.274,
      "step": 11745
    },
    {
      "epoch": 0.6396166438640556,
      "grad_norm": 0.6486099259602204,
      "learning_rate": 0.00015909056050792994,
      "loss": 12.3702,
      "step": 11746
    },
    {
      "epoch": 0.6396710978606386,
      "grad_norm": 0.7265266626848722,
      "learning_rate": 0.00015908344623037757,
      "loss": 12.1387,
      "step": 11747
    },
    {
      "epoch": 0.6397255518572216,
      "grad_norm": 0.6899028392303771,
      "learning_rate": 0.00015907633149338238,
      "loss": 12.334,
      "step": 11748
    },
    {
      "epoch": 0.6397800058538047,
      "grad_norm": 0.5360899758896915,
      "learning_rate": 0.00015906921629699975,
      "loss": 12.3807,
      "step": 11749
    },
    {
      "epoch": 0.6398344598503877,
      "grad_norm": 0.5463456124183731,
      "learning_rate": 0.00015906210064128498,
      "loss": 12.3499,
      "step": 11750
    },
    {
      "epoch": 0.6398889138469707,
      "grad_norm": 0.6538197908044291,
      "learning_rate": 0.00015905498452629342,
      "loss": 12.2644,
      "step": 11751
    },
    {
      "epoch": 0.6399433678435537,
      "grad_norm": 0.6847980090091363,
      "learning_rate": 0.00015904786795208038,
      "loss": 12.4953,
      "step": 11752
    },
    {
      "epoch": 0.6399978218401366,
      "grad_norm": 0.5717753214335817,
      "learning_rate": 0.0001590407509187012,
      "loss": 12.3944,
      "step": 11753
    },
    {
      "epoch": 0.6400522758367196,
      "grad_norm": 0.5897375896614131,
      "learning_rate": 0.0001590336334262113,
      "loss": 12.3208,
      "step": 11754
    },
    {
      "epoch": 0.6401067298333027,
      "grad_norm": 0.6509711952414429,
      "learning_rate": 0.00015902651547466593,
      "loss": 12.3059,
      "step": 11755
    },
    {
      "epoch": 0.6401611838298857,
      "grad_norm": 0.5686185105629006,
      "learning_rate": 0.00015901939706412045,
      "loss": 12.3686,
      "step": 11756
    },
    {
      "epoch": 0.6402156378264687,
      "grad_norm": 0.5761598035037799,
      "learning_rate": 0.00015901227819463024,
      "loss": 12.292,
      "step": 11757
    },
    {
      "epoch": 0.6402700918230517,
      "grad_norm": 0.6233562948043391,
      "learning_rate": 0.0001590051588662507,
      "loss": 12.2645,
      "step": 11758
    },
    {
      "epoch": 0.6403245458196347,
      "grad_norm": 0.5431735823508531,
      "learning_rate": 0.00015899803907903714,
      "loss": 12.3071,
      "step": 11759
    },
    {
      "epoch": 0.6403789998162177,
      "grad_norm": 0.5844067916605888,
      "learning_rate": 0.00015899091883304494,
      "loss": 12.3562,
      "step": 11760
    },
    {
      "epoch": 0.6404334538128008,
      "grad_norm": 0.5775938645263473,
      "learning_rate": 0.00015898379812832945,
      "loss": 12.2704,
      "step": 11761
    },
    {
      "epoch": 0.6404879078093838,
      "grad_norm": 0.6114526535574534,
      "learning_rate": 0.00015897667696494606,
      "loss": 12.2708,
      "step": 11762
    },
    {
      "epoch": 0.6405423618059668,
      "grad_norm": 0.543502985810955,
      "learning_rate": 0.0001589695553429501,
      "loss": 12.2512,
      "step": 11763
    },
    {
      "epoch": 0.6405968158025498,
      "grad_norm": 0.5491499746554386,
      "learning_rate": 0.00015896243326239703,
      "loss": 12.1462,
      "step": 11764
    },
    {
      "epoch": 0.6406512697991328,
      "grad_norm": 0.659161048417712,
      "learning_rate": 0.00015895531072334217,
      "loss": 12.2656,
      "step": 11765
    },
    {
      "epoch": 0.6407057237957159,
      "grad_norm": 0.5578738711061708,
      "learning_rate": 0.00015894818772584096,
      "loss": 12.2665,
      "step": 11766
    },
    {
      "epoch": 0.6407601777922989,
      "grad_norm": 0.5485848619593658,
      "learning_rate": 0.00015894106426994875,
      "loss": 12.3885,
      "step": 11767
    },
    {
      "epoch": 0.6408146317888819,
      "grad_norm": 0.6042684554813096,
      "learning_rate": 0.0001589339403557209,
      "loss": 12.268,
      "step": 11768
    },
    {
      "epoch": 0.6408690857854649,
      "grad_norm": 0.5603475640961582,
      "learning_rate": 0.0001589268159832129,
      "loss": 12.2199,
      "step": 11769
    },
    {
      "epoch": 0.6409235397820479,
      "grad_norm": 0.6086400504391121,
      "learning_rate": 0.00015891969115248007,
      "loss": 12.3038,
      "step": 11770
    },
    {
      "epoch": 0.6409779937786309,
      "grad_norm": 0.7496642611061374,
      "learning_rate": 0.00015891256586357782,
      "loss": 12.3797,
      "step": 11771
    },
    {
      "epoch": 0.641032447775214,
      "grad_norm": 0.6043326573937473,
      "learning_rate": 0.00015890544011656161,
      "loss": 12.3197,
      "step": 11772
    },
    {
      "epoch": 0.641086901771797,
      "grad_norm": 0.5879429958227689,
      "learning_rate": 0.0001588983139114868,
      "loss": 12.3328,
      "step": 11773
    },
    {
      "epoch": 0.6411413557683799,
      "grad_norm": 0.6431041182157121,
      "learning_rate": 0.00015889118724840887,
      "loss": 12.5443,
      "step": 11774
    },
    {
      "epoch": 0.6411958097649629,
      "grad_norm": 0.6387769615888651,
      "learning_rate": 0.00015888406012738314,
      "loss": 12.2873,
      "step": 11775
    },
    {
      "epoch": 0.6412502637615459,
      "grad_norm": 0.5907844038891188,
      "learning_rate": 0.00015887693254846509,
      "loss": 12.1766,
      "step": 11776
    },
    {
      "epoch": 0.6413047177581289,
      "grad_norm": 0.8089658170344911,
      "learning_rate": 0.0001588698045117101,
      "loss": 12.319,
      "step": 11777
    },
    {
      "epoch": 0.641359171754712,
      "grad_norm": 0.5795627418817415,
      "learning_rate": 0.00015886267601717373,
      "loss": 12.1944,
      "step": 11778
    },
    {
      "epoch": 0.641413625751295,
      "grad_norm": 0.6330996063985966,
      "learning_rate": 0.00015885554706491126,
      "loss": 12.3079,
      "step": 11779
    },
    {
      "epoch": 0.641468079747878,
      "grad_norm": 0.6312142988227856,
      "learning_rate": 0.0001588484176549782,
      "loss": 12.2919,
      "step": 11780
    },
    {
      "epoch": 0.641522533744461,
      "grad_norm": 0.7992335536619127,
      "learning_rate": 0.00015884128778743,
      "loss": 12.3403,
      "step": 11781
    },
    {
      "epoch": 0.641576987741044,
      "grad_norm": 0.6054780153306412,
      "learning_rate": 0.00015883415746232205,
      "loss": 12.1981,
      "step": 11782
    },
    {
      "epoch": 0.641631441737627,
      "grad_norm": 0.6121608821672163,
      "learning_rate": 0.00015882702667970982,
      "loss": 12.237,
      "step": 11783
    },
    {
      "epoch": 0.6416858957342101,
      "grad_norm": 0.6458800184805136,
      "learning_rate": 0.00015881989543964877,
      "loss": 12.3485,
      "step": 11784
    },
    {
      "epoch": 0.6417403497307931,
      "grad_norm": 0.610389351156267,
      "learning_rate": 0.00015881276374219436,
      "loss": 12.2533,
      "step": 11785
    },
    {
      "epoch": 0.6417948037273761,
      "grad_norm": 0.6011654744945086,
      "learning_rate": 0.00015880563158740202,
      "loss": 12.2968,
      "step": 11786
    },
    {
      "epoch": 0.6418492577239591,
      "grad_norm": 0.6489037502758117,
      "learning_rate": 0.00015879849897532724,
      "loss": 12.4367,
      "step": 11787
    },
    {
      "epoch": 0.6419037117205421,
      "grad_norm": 0.6295543987341068,
      "learning_rate": 0.00015879136590602545,
      "loss": 12.2218,
      "step": 11788
    },
    {
      "epoch": 0.641958165717125,
      "grad_norm": 0.7691534351517384,
      "learning_rate": 0.00015878423237955218,
      "loss": 12.3219,
      "step": 11789
    },
    {
      "epoch": 0.6420126197137082,
      "grad_norm": 0.7059532026340123,
      "learning_rate": 0.00015877709839596285,
      "loss": 12.2999,
      "step": 11790
    },
    {
      "epoch": 0.6420670737102911,
      "grad_norm": 0.6641858468829318,
      "learning_rate": 0.0001587699639553129,
      "loss": 12.3139,
      "step": 11791
    },
    {
      "epoch": 0.6421215277068741,
      "grad_norm": 0.6673144977966753,
      "learning_rate": 0.0001587628290576579,
      "loss": 12.4529,
      "step": 11792
    },
    {
      "epoch": 0.6421759817034571,
      "grad_norm": 0.7522695067144315,
      "learning_rate": 0.0001587556937030533,
      "loss": 12.2677,
      "step": 11793
    },
    {
      "epoch": 0.6422304357000401,
      "grad_norm": 0.533954295083735,
      "learning_rate": 0.00015874855789155455,
      "loss": 12.2353,
      "step": 11794
    },
    {
      "epoch": 0.6422848896966231,
      "grad_norm": 0.6863388740679423,
      "learning_rate": 0.0001587414216232171,
      "loss": 12.3066,
      "step": 11795
    },
    {
      "epoch": 0.6423393436932062,
      "grad_norm": 0.7028548606498654,
      "learning_rate": 0.0001587342848980966,
      "loss": 12.3073,
      "step": 11796
    },
    {
      "epoch": 0.6423937976897892,
      "grad_norm": 0.5838235714771344,
      "learning_rate": 0.0001587271477162484,
      "loss": 12.3099,
      "step": 11797
    },
    {
      "epoch": 0.6424482516863722,
      "grad_norm": 0.5491865981727515,
      "learning_rate": 0.0001587200100777281,
      "loss": 12.3046,
      "step": 11798
    },
    {
      "epoch": 0.6425027056829552,
      "grad_norm": 0.6396741653719359,
      "learning_rate": 0.00015871287198259112,
      "loss": 12.4255,
      "step": 11799
    },
    {
      "epoch": 0.6425571596795382,
      "grad_norm": 0.5941077393260389,
      "learning_rate": 0.00015870573343089298,
      "loss": 12.3175,
      "step": 11800
    },
    {
      "epoch": 0.6426116136761213,
      "grad_norm": 0.5365098578871422,
      "learning_rate": 0.00015869859442268924,
      "loss": 12.2355,
      "step": 11801
    },
    {
      "epoch": 0.6426660676727043,
      "grad_norm": 0.629599542239981,
      "learning_rate": 0.00015869145495803539,
      "loss": 12.2843,
      "step": 11802
    },
    {
      "epoch": 0.6427205216692873,
      "grad_norm": 0.6976645290629956,
      "learning_rate": 0.0001586843150369869,
      "loss": 12.533,
      "step": 11803
    },
    {
      "epoch": 0.6427749756658703,
      "grad_norm": 0.6257102830411845,
      "learning_rate": 0.0001586771746595994,
      "loss": 12.4957,
      "step": 11804
    },
    {
      "epoch": 0.6428294296624533,
      "grad_norm": 0.5139680368557052,
      "learning_rate": 0.0001586700338259283,
      "loss": 12.3099,
      "step": 11805
    },
    {
      "epoch": 0.6428838836590363,
      "grad_norm": 0.6795158134138467,
      "learning_rate": 0.0001586628925360292,
      "loss": 12.3531,
      "step": 11806
    },
    {
      "epoch": 0.6429383376556194,
      "grad_norm": 0.6336616387516697,
      "learning_rate": 0.0001586557507899576,
      "loss": 12.2245,
      "step": 11807
    },
    {
      "epoch": 0.6429927916522024,
      "grad_norm": 0.5827707567392567,
      "learning_rate": 0.00015864860858776908,
      "loss": 12.2562,
      "step": 11808
    },
    {
      "epoch": 0.6430472456487853,
      "grad_norm": 0.6793117933085399,
      "learning_rate": 0.0001586414659295191,
      "loss": 12.4026,
      "step": 11809
    },
    {
      "epoch": 0.6431016996453683,
      "grad_norm": 0.7724808780527022,
      "learning_rate": 0.00015863432281526326,
      "loss": 12.2154,
      "step": 11810
    },
    {
      "epoch": 0.6431561536419513,
      "grad_norm": 0.669858522966245,
      "learning_rate": 0.0001586271792450571,
      "loss": 12.2635,
      "step": 11811
    },
    {
      "epoch": 0.6432106076385343,
      "grad_norm": 0.602202572719811,
      "learning_rate": 0.00015862003521895614,
      "loss": 12.1333,
      "step": 11812
    },
    {
      "epoch": 0.6432650616351174,
      "grad_norm": 0.5983408313173444,
      "learning_rate": 0.00015861289073701597,
      "loss": 12.3779,
      "step": 11813
    },
    {
      "epoch": 0.6433195156317004,
      "grad_norm": 0.6113563849314245,
      "learning_rate": 0.00015860574579929215,
      "loss": 12.3404,
      "step": 11814
    },
    {
      "epoch": 0.6433739696282834,
      "grad_norm": 0.6883609075478769,
      "learning_rate": 0.0001585986004058402,
      "loss": 12.4151,
      "step": 11815
    },
    {
      "epoch": 0.6434284236248664,
      "grad_norm": 0.6060410474299993,
      "learning_rate": 0.0001585914545567157,
      "loss": 12.3261,
      "step": 11816
    },
    {
      "epoch": 0.6434828776214494,
      "grad_norm": 0.549030679214863,
      "learning_rate": 0.00015858430825197426,
      "loss": 12.3239,
      "step": 11817
    },
    {
      "epoch": 0.6435373316180324,
      "grad_norm": 0.5995797386872938,
      "learning_rate": 0.00015857716149167138,
      "loss": 12.2988,
      "step": 11818
    },
    {
      "epoch": 0.6435917856146155,
      "grad_norm": 0.6321642957582062,
      "learning_rate": 0.00015857001427586269,
      "loss": 12.3957,
      "step": 11819
    },
    {
      "epoch": 0.6436462396111985,
      "grad_norm": 0.551558625463365,
      "learning_rate": 0.00015856286660460373,
      "loss": 12.3422,
      "step": 11820
    },
    {
      "epoch": 0.6437006936077815,
      "grad_norm": 0.5490497130441678,
      "learning_rate": 0.00015855571847795012,
      "loss": 12.2613,
      "step": 11821
    },
    {
      "epoch": 0.6437551476043645,
      "grad_norm": 0.5985946306848401,
      "learning_rate": 0.0001585485698959574,
      "loss": 12.3177,
      "step": 11822
    },
    {
      "epoch": 0.6438096016009475,
      "grad_norm": 0.5911531978371243,
      "learning_rate": 0.00015854142085868118,
      "loss": 12.2837,
      "step": 11823
    },
    {
      "epoch": 0.6438640555975305,
      "grad_norm": 0.5405399476121963,
      "learning_rate": 0.00015853427136617708,
      "loss": 12.3125,
      "step": 11824
    },
    {
      "epoch": 0.6439185095941136,
      "grad_norm": 0.6308899194181132,
      "learning_rate": 0.00015852712141850063,
      "loss": 12.32,
      "step": 11825
    },
    {
      "epoch": 0.6439729635906966,
      "grad_norm": 0.5272000889908651,
      "learning_rate": 0.00015851997101570752,
      "loss": 12.2391,
      "step": 11826
    },
    {
      "epoch": 0.6440274175872795,
      "grad_norm": 0.6978034099378577,
      "learning_rate": 0.00015851282015785328,
      "loss": 12.2693,
      "step": 11827
    },
    {
      "epoch": 0.6440818715838625,
      "grad_norm": 0.6859417143713247,
      "learning_rate": 0.00015850566884499352,
      "loss": 12.1768,
      "step": 11828
    },
    {
      "epoch": 0.6441363255804455,
      "grad_norm": 0.6287347167833705,
      "learning_rate": 0.00015849851707718389,
      "loss": 12.2717,
      "step": 11829
    },
    {
      "epoch": 0.6441907795770285,
      "grad_norm": 0.6238801089136621,
      "learning_rate": 0.00015849136485447996,
      "loss": 12.3766,
      "step": 11830
    },
    {
      "epoch": 0.6442452335736116,
      "grad_norm": 0.5472991333501801,
      "learning_rate": 0.0001584842121769374,
      "loss": 12.2784,
      "step": 11831
    },
    {
      "epoch": 0.6442996875701946,
      "grad_norm": 0.6514650993952252,
      "learning_rate": 0.00015847705904461178,
      "loss": 12.3264,
      "step": 11832
    },
    {
      "epoch": 0.6443541415667776,
      "grad_norm": 0.5794851645266091,
      "learning_rate": 0.0001584699054575587,
      "loss": 12.4061,
      "step": 11833
    },
    {
      "epoch": 0.6444085955633606,
      "grad_norm": 0.6213116335233053,
      "learning_rate": 0.00015846275141583388,
      "loss": 12.3477,
      "step": 11834
    },
    {
      "epoch": 0.6444630495599436,
      "grad_norm": 0.5722335188355683,
      "learning_rate": 0.0001584555969194929,
      "loss": 12.306,
      "step": 11835
    },
    {
      "epoch": 0.6445175035565267,
      "grad_norm": 0.6736011912847758,
      "learning_rate": 0.0001584484419685914,
      "loss": 12.1462,
      "step": 11836
    },
    {
      "epoch": 0.6445719575531097,
      "grad_norm": 0.5385332998875926,
      "learning_rate": 0.000158441286563185,
      "loss": 12.3111,
      "step": 11837
    },
    {
      "epoch": 0.6446264115496927,
      "grad_norm": 0.6159916819535322,
      "learning_rate": 0.00015843413070332934,
      "loss": 12.4208,
      "step": 11838
    },
    {
      "epoch": 0.6446808655462757,
      "grad_norm": 0.5688143212500484,
      "learning_rate": 0.0001584269743890801,
      "loss": 12.0938,
      "step": 11839
    },
    {
      "epoch": 0.6447353195428587,
      "grad_norm": 0.5963157827362754,
      "learning_rate": 0.0001584198176204929,
      "loss": 12.339,
      "step": 11840
    },
    {
      "epoch": 0.6447897735394417,
      "grad_norm": 0.6757611489483027,
      "learning_rate": 0.0001584126603976234,
      "loss": 12.3657,
      "step": 11841
    },
    {
      "epoch": 0.6448442275360248,
      "grad_norm": 0.5805181095282763,
      "learning_rate": 0.00015840550272052726,
      "loss": 12.3187,
      "step": 11842
    },
    {
      "epoch": 0.6448986815326078,
      "grad_norm": 0.635858992716586,
      "learning_rate": 0.00015839834458926012,
      "loss": 12.4277,
      "step": 11843
    },
    {
      "epoch": 0.6449531355291908,
      "grad_norm": 0.68439571005598,
      "learning_rate": 0.00015839118600387771,
      "loss": 12.373,
      "step": 11844
    },
    {
      "epoch": 0.6450075895257738,
      "grad_norm": 0.6642118909612381,
      "learning_rate": 0.0001583840269644356,
      "loss": 12.2769,
      "step": 11845
    },
    {
      "epoch": 0.6450620435223567,
      "grad_norm": 0.7855157673961042,
      "learning_rate": 0.00015837686747098952,
      "loss": 12.4094,
      "step": 11846
    },
    {
      "epoch": 0.6451164975189397,
      "grad_norm": 0.7055329344152202,
      "learning_rate": 0.00015836970752359513,
      "loss": 12.4626,
      "step": 11847
    },
    {
      "epoch": 0.6451709515155228,
      "grad_norm": 0.6876268336548789,
      "learning_rate": 0.00015836254712230807,
      "loss": 12.4605,
      "step": 11848
    },
    {
      "epoch": 0.6452254055121058,
      "grad_norm": 0.6088296851825628,
      "learning_rate": 0.00015835538626718412,
      "loss": 12.3367,
      "step": 11849
    },
    {
      "epoch": 0.6452798595086888,
      "grad_norm": 0.6061418913585683,
      "learning_rate": 0.00015834822495827886,
      "loss": 12.2388,
      "step": 11850
    },
    {
      "epoch": 0.6453343135052718,
      "grad_norm": 0.5872816859150357,
      "learning_rate": 0.00015834106319564804,
      "loss": 12.1829,
      "step": 11851
    },
    {
      "epoch": 0.6453887675018548,
      "grad_norm": 0.5548108060461536,
      "learning_rate": 0.00015833390097934728,
      "loss": 12.1922,
      "step": 11852
    },
    {
      "epoch": 0.6454432214984378,
      "grad_norm": 0.62230989731747,
      "learning_rate": 0.00015832673830943236,
      "loss": 12.3993,
      "step": 11853
    },
    {
      "epoch": 0.6454976754950209,
      "grad_norm": 0.6812107135336923,
      "learning_rate": 0.00015831957518595897,
      "loss": 12.3115,
      "step": 11854
    },
    {
      "epoch": 0.6455521294916039,
      "grad_norm": 0.6203214207289377,
      "learning_rate": 0.00015831241160898274,
      "loss": 12.3399,
      "step": 11855
    },
    {
      "epoch": 0.6456065834881869,
      "grad_norm": 0.6639827570460414,
      "learning_rate": 0.00015830524757855943,
      "loss": 12.2929,
      "step": 11856
    },
    {
      "epoch": 0.6456610374847699,
      "grad_norm": 0.6651863119017111,
      "learning_rate": 0.00015829808309474473,
      "loss": 12.3096,
      "step": 11857
    },
    {
      "epoch": 0.6457154914813529,
      "grad_norm": 0.5703054855457829,
      "learning_rate": 0.00015829091815759436,
      "loss": 12.2785,
      "step": 11858
    },
    {
      "epoch": 0.6457699454779359,
      "grad_norm": 0.6143567571433828,
      "learning_rate": 0.00015828375276716406,
      "loss": 12.3569,
      "step": 11859
    },
    {
      "epoch": 0.645824399474519,
      "grad_norm": 0.6088561935498821,
      "learning_rate": 0.0001582765869235095,
      "loss": 12.3654,
      "step": 11860
    },
    {
      "epoch": 0.645878853471102,
      "grad_norm": 0.6353000312500281,
      "learning_rate": 0.00015826942062668645,
      "loss": 12.2322,
      "step": 11861
    },
    {
      "epoch": 0.645933307467685,
      "grad_norm": 0.6765711476284719,
      "learning_rate": 0.0001582622538767506,
      "loss": 12.3561,
      "step": 11862
    },
    {
      "epoch": 0.645987761464268,
      "grad_norm": 0.585489756512397,
      "learning_rate": 0.00015825508667375768,
      "loss": 12.1992,
      "step": 11863
    },
    {
      "epoch": 0.6460422154608509,
      "grad_norm": 0.6253853276677027,
      "learning_rate": 0.00015824791901776342,
      "loss": 12.3809,
      "step": 11864
    },
    {
      "epoch": 0.6460966694574339,
      "grad_norm": 0.6279991291887276,
      "learning_rate": 0.00015824075090882365,
      "loss": 12.3028,
      "step": 11865
    },
    {
      "epoch": 0.646151123454017,
      "grad_norm": 0.6155440485579453,
      "learning_rate": 0.00015823358234699398,
      "loss": 12.3539,
      "step": 11866
    },
    {
      "epoch": 0.6462055774506,
      "grad_norm": 0.6762972682778122,
      "learning_rate": 0.0001582264133323302,
      "loss": 12.3892,
      "step": 11867
    },
    {
      "epoch": 0.646260031447183,
      "grad_norm": 0.5950789371522833,
      "learning_rate": 0.00015821924386488808,
      "loss": 12.2143,
      "step": 11868
    },
    {
      "epoch": 0.646314485443766,
      "grad_norm": 0.606593652587477,
      "learning_rate": 0.00015821207394472336,
      "loss": 12.2222,
      "step": 11869
    },
    {
      "epoch": 0.646368939440349,
      "grad_norm": 0.5798466789294091,
      "learning_rate": 0.00015820490357189177,
      "loss": 12.2193,
      "step": 11870
    },
    {
      "epoch": 0.6464233934369321,
      "grad_norm": 0.598060235563983,
      "learning_rate": 0.0001581977327464491,
      "loss": 12.3152,
      "step": 11871
    },
    {
      "epoch": 0.6464778474335151,
      "grad_norm": 0.6380850115231458,
      "learning_rate": 0.0001581905614684511,
      "loss": 12.3271,
      "step": 11872
    },
    {
      "epoch": 0.6465323014300981,
      "grad_norm": 0.5917032614504739,
      "learning_rate": 0.00015818338973795355,
      "loss": 12.3361,
      "step": 11873
    },
    {
      "epoch": 0.6465867554266811,
      "grad_norm": 0.6309372172799924,
      "learning_rate": 0.0001581762175550122,
      "loss": 12.343,
      "step": 11874
    },
    {
      "epoch": 0.6466412094232641,
      "grad_norm": 0.6536200778124828,
      "learning_rate": 0.00015816904491968282,
      "loss": 12.4174,
      "step": 11875
    },
    {
      "epoch": 0.6466956634198471,
      "grad_norm": 0.6009748805400857,
      "learning_rate": 0.00015816187183202121,
      "loss": 12.4067,
      "step": 11876
    },
    {
      "epoch": 0.6467501174164302,
      "grad_norm": 0.5543701890670989,
      "learning_rate": 0.0001581546982920831,
      "loss": 12.3837,
      "step": 11877
    },
    {
      "epoch": 0.6468045714130132,
      "grad_norm": 0.5888343705499574,
      "learning_rate": 0.0001581475242999243,
      "loss": 12.2352,
      "step": 11878
    },
    {
      "epoch": 0.6468590254095962,
      "grad_norm": 0.6293347929653329,
      "learning_rate": 0.00015814034985560063,
      "loss": 12.1845,
      "step": 11879
    },
    {
      "epoch": 0.6469134794061792,
      "grad_norm": 0.6254066864745504,
      "learning_rate": 0.0001581331749591678,
      "loss": 12.1629,
      "step": 11880
    },
    {
      "epoch": 0.6469679334027622,
      "grad_norm": 0.5798924916404218,
      "learning_rate": 0.0001581259996106817,
      "loss": 12.2957,
      "step": 11881
    },
    {
      "epoch": 0.6470223873993451,
      "grad_norm": 0.6548652982042967,
      "learning_rate": 0.0001581188238101981,
      "loss": 12.2919,
      "step": 11882
    },
    {
      "epoch": 0.6470768413959282,
      "grad_norm": 0.6470119826591115,
      "learning_rate": 0.00015811164755777274,
      "loss": 12.4236,
      "step": 11883
    },
    {
      "epoch": 0.6471312953925112,
      "grad_norm": 0.7396324813529676,
      "learning_rate": 0.00015810447085346145,
      "loss": 12.3413,
      "step": 11884
    },
    {
      "epoch": 0.6471857493890942,
      "grad_norm": 0.659380387570783,
      "learning_rate": 0.0001580972936973201,
      "loss": 12.2496,
      "step": 11885
    },
    {
      "epoch": 0.6472402033856772,
      "grad_norm": 0.6407359324597263,
      "learning_rate": 0.0001580901160894044,
      "loss": 12.348,
      "step": 11886
    },
    {
      "epoch": 0.6472946573822602,
      "grad_norm": 0.5541698793238401,
      "learning_rate": 0.00015808293802977024,
      "loss": 12.1559,
      "step": 11887
    },
    {
      "epoch": 0.6473491113788432,
      "grad_norm": 0.6298551123169205,
      "learning_rate": 0.00015807575951847343,
      "loss": 12.4117,
      "step": 11888
    },
    {
      "epoch": 0.6474035653754263,
      "grad_norm": 0.696425727106263,
      "learning_rate": 0.00015806858055556971,
      "loss": 12.3259,
      "step": 11889
    },
    {
      "epoch": 0.6474580193720093,
      "grad_norm": 0.6215328446217157,
      "learning_rate": 0.00015806140114111504,
      "loss": 12.298,
      "step": 11890
    },
    {
      "epoch": 0.6475124733685923,
      "grad_norm": 0.6811713311025708,
      "learning_rate": 0.00015805422127516513,
      "loss": 12.416,
      "step": 11891
    },
    {
      "epoch": 0.6475669273651753,
      "grad_norm": 0.5875299931350033,
      "learning_rate": 0.00015804704095777588,
      "loss": 12.2622,
      "step": 11892
    },
    {
      "epoch": 0.6476213813617583,
      "grad_norm": 0.6182923010880244,
      "learning_rate": 0.00015803986018900315,
      "loss": 12.4197,
      "step": 11893
    },
    {
      "epoch": 0.6476758353583413,
      "grad_norm": 0.7301611412615433,
      "learning_rate": 0.00015803267896890265,
      "loss": 12.3345,
      "step": 11894
    },
    {
      "epoch": 0.6477302893549244,
      "grad_norm": 0.59758392143012,
      "learning_rate": 0.0001580254972975304,
      "loss": 12.3508,
      "step": 11895
    },
    {
      "epoch": 0.6477847433515074,
      "grad_norm": 0.7039951589381491,
      "learning_rate": 0.00015801831517494208,
      "loss": 12.3217,
      "step": 11896
    },
    {
      "epoch": 0.6478391973480904,
      "grad_norm": 0.6196817665670296,
      "learning_rate": 0.00015801113260119362,
      "loss": 12.138,
      "step": 11897
    },
    {
      "epoch": 0.6478936513446734,
      "grad_norm": 0.6676907346235882,
      "learning_rate": 0.00015800394957634088,
      "loss": 12.3381,
      "step": 11898
    },
    {
      "epoch": 0.6479481053412564,
      "grad_norm": 0.7448106654204392,
      "learning_rate": 0.0001579967661004397,
      "loss": 12.3667,
      "step": 11899
    },
    {
      "epoch": 0.6480025593378395,
      "grad_norm": 0.6139973688894776,
      "learning_rate": 0.00015798958217354592,
      "loss": 12.2898,
      "step": 11900
    },
    {
      "epoch": 0.6480570133344224,
      "grad_norm": 0.6219380327775433,
      "learning_rate": 0.00015798239779571546,
      "loss": 12.3213,
      "step": 11901
    },
    {
      "epoch": 0.6481114673310054,
      "grad_norm": 0.624283925317137,
      "learning_rate": 0.00015797521296700412,
      "loss": 12.3334,
      "step": 11902
    },
    {
      "epoch": 0.6481659213275884,
      "grad_norm": 0.6852416421453453,
      "learning_rate": 0.0001579680276874678,
      "loss": 12.4149,
      "step": 11903
    },
    {
      "epoch": 0.6482203753241714,
      "grad_norm": 0.588462029793721,
      "learning_rate": 0.00015796084195716242,
      "loss": 12.3818,
      "step": 11904
    },
    {
      "epoch": 0.6482748293207544,
      "grad_norm": 0.622781245012364,
      "learning_rate": 0.00015795365577614377,
      "loss": 12.3759,
      "step": 11905
    },
    {
      "epoch": 0.6483292833173375,
      "grad_norm": 0.7072158092814134,
      "learning_rate": 0.00015794646914446778,
      "loss": 12.4398,
      "step": 11906
    },
    {
      "epoch": 0.6483837373139205,
      "grad_norm": 0.610219828969978,
      "learning_rate": 0.00015793928206219034,
      "loss": 12.3247,
      "step": 11907
    },
    {
      "epoch": 0.6484381913105035,
      "grad_norm": 0.592606307805623,
      "learning_rate": 0.00015793209452936733,
      "loss": 12.371,
      "step": 11908
    },
    {
      "epoch": 0.6484926453070865,
      "grad_norm": 0.6109845084666096,
      "learning_rate": 0.0001579249065460546,
      "loss": 12.3556,
      "step": 11909
    },
    {
      "epoch": 0.6485470993036695,
      "grad_norm": 0.5691135905588112,
      "learning_rate": 0.00015791771811230813,
      "loss": 12.3765,
      "step": 11910
    },
    {
      "epoch": 0.6486015533002525,
      "grad_norm": 0.627849729098813,
      "learning_rate": 0.00015791052922818375,
      "loss": 12.4136,
      "step": 11911
    },
    {
      "epoch": 0.6486560072968356,
      "grad_norm": 0.5720576719145197,
      "learning_rate": 0.00015790333989373738,
      "loss": 12.187,
      "step": 11912
    },
    {
      "epoch": 0.6487104612934186,
      "grad_norm": 0.5913965852750814,
      "learning_rate": 0.00015789615010902494,
      "loss": 12.22,
      "step": 11913
    },
    {
      "epoch": 0.6487649152900016,
      "grad_norm": 0.6458783596766352,
      "learning_rate": 0.00015788895987410234,
      "loss": 12.4376,
      "step": 11914
    },
    {
      "epoch": 0.6488193692865846,
      "grad_norm": 0.5357031183400054,
      "learning_rate": 0.00015788176918902545,
      "loss": 12.3014,
      "step": 11915
    },
    {
      "epoch": 0.6488738232831676,
      "grad_norm": 0.5878185119517014,
      "learning_rate": 0.00015787457805385022,
      "loss": 12.2597,
      "step": 11916
    },
    {
      "epoch": 0.6489282772797506,
      "grad_norm": 0.5408649338255814,
      "learning_rate": 0.00015786738646863258,
      "loss": 12.1758,
      "step": 11917
    },
    {
      "epoch": 0.6489827312763337,
      "grad_norm": 0.5962386022181083,
      "learning_rate": 0.0001578601944334284,
      "loss": 12.3954,
      "step": 11918
    },
    {
      "epoch": 0.6490371852729167,
      "grad_norm": 0.6894379242956311,
      "learning_rate": 0.0001578530019482937,
      "loss": 12.4568,
      "step": 11919
    },
    {
      "epoch": 0.6490916392694996,
      "grad_norm": 0.6447570054291695,
      "learning_rate": 0.00015784580901328433,
      "loss": 12.3826,
      "step": 11920
    },
    {
      "epoch": 0.6491460932660826,
      "grad_norm": 0.6538136842300999,
      "learning_rate": 0.00015783861562845624,
      "loss": 12.1867,
      "step": 11921
    },
    {
      "epoch": 0.6492005472626656,
      "grad_norm": 0.6588957923748714,
      "learning_rate": 0.00015783142179386542,
      "loss": 12.3532,
      "step": 11922
    },
    {
      "epoch": 0.6492550012592486,
      "grad_norm": 0.609883320307594,
      "learning_rate": 0.0001578242275095677,
      "loss": 12.2295,
      "step": 11923
    },
    {
      "epoch": 0.6493094552558317,
      "grad_norm": 0.5832732508989199,
      "learning_rate": 0.00015781703277561912,
      "loss": 12.3784,
      "step": 11924
    },
    {
      "epoch": 0.6493639092524147,
      "grad_norm": 0.581475044581891,
      "learning_rate": 0.0001578098375920756,
      "loss": 12.1825,
      "step": 11925
    },
    {
      "epoch": 0.6494183632489977,
      "grad_norm": 0.5271469409272188,
      "learning_rate": 0.0001578026419589931,
      "loss": 12.2623,
      "step": 11926
    },
    {
      "epoch": 0.6494728172455807,
      "grad_norm": 0.5873360893022631,
      "learning_rate": 0.00015779544587642754,
      "loss": 12.3341,
      "step": 11927
    },
    {
      "epoch": 0.6495272712421637,
      "grad_norm": 0.7344385378568903,
      "learning_rate": 0.0001577882493444349,
      "loss": 12.0677,
      "step": 11928
    },
    {
      "epoch": 0.6495817252387467,
      "grad_norm": 0.5793826632935817,
      "learning_rate": 0.00015778105236307117,
      "loss": 12.444,
      "step": 11929
    },
    {
      "epoch": 0.6496361792353298,
      "grad_norm": 0.5508804313879363,
      "learning_rate": 0.00015777385493239226,
      "loss": 12.343,
      "step": 11930
    },
    {
      "epoch": 0.6496906332319128,
      "grad_norm": 0.5574879850197885,
      "learning_rate": 0.00015776665705245416,
      "loss": 12.3612,
      "step": 11931
    },
    {
      "epoch": 0.6497450872284958,
      "grad_norm": 0.6029465954215187,
      "learning_rate": 0.00015775945872331288,
      "loss": 12.3669,
      "step": 11932
    },
    {
      "epoch": 0.6497995412250788,
      "grad_norm": 0.621974590243019,
      "learning_rate": 0.00015775225994502434,
      "loss": 12.2378,
      "step": 11933
    },
    {
      "epoch": 0.6498539952216618,
      "grad_norm": 0.5669301302625296,
      "learning_rate": 0.00015774506071764455,
      "loss": 12.3814,
      "step": 11934
    },
    {
      "epoch": 0.6499084492182449,
      "grad_norm": 0.6069993807753911,
      "learning_rate": 0.00015773786104122947,
      "loss": 12.31,
      "step": 11935
    },
    {
      "epoch": 0.6499629032148279,
      "grad_norm": 0.6790511836398206,
      "learning_rate": 0.0001577306609158351,
      "loss": 12.4608,
      "step": 11936
    },
    {
      "epoch": 0.6500173572114109,
      "grad_norm": 0.5829094050295218,
      "learning_rate": 0.00015772346034151745,
      "loss": 12.3586,
      "step": 11937
    },
    {
      "epoch": 0.6500718112079938,
      "grad_norm": 0.6722397218659423,
      "learning_rate": 0.00015771625931833248,
      "loss": 12.2938,
      "step": 11938
    },
    {
      "epoch": 0.6501262652045768,
      "grad_norm": 0.5748677087191401,
      "learning_rate": 0.0001577090578463362,
      "loss": 12.3666,
      "step": 11939
    },
    {
      "epoch": 0.6501807192011598,
      "grad_norm": 0.5524163198640157,
      "learning_rate": 0.00015770185592558459,
      "loss": 12.1791,
      "step": 11940
    },
    {
      "epoch": 0.6502351731977429,
      "grad_norm": 0.64979857567962,
      "learning_rate": 0.00015769465355613372,
      "loss": 12.2564,
      "step": 11941
    },
    {
      "epoch": 0.6502896271943259,
      "grad_norm": 0.6487747620285252,
      "learning_rate": 0.0001576874507380395,
      "loss": 12.2984,
      "step": 11942
    },
    {
      "epoch": 0.6503440811909089,
      "grad_norm": 0.5831819737136476,
      "learning_rate": 0.00015768024747135802,
      "loss": 12.3315,
      "step": 11943
    },
    {
      "epoch": 0.6503985351874919,
      "grad_norm": 0.726194846556106,
      "learning_rate": 0.00015767304375614524,
      "loss": 12.373,
      "step": 11944
    },
    {
      "epoch": 0.6504529891840749,
      "grad_norm": 0.7572436990405581,
      "learning_rate": 0.00015766583959245722,
      "loss": 12.3561,
      "step": 11945
    },
    {
      "epoch": 0.6505074431806579,
      "grad_norm": 0.8388816417572426,
      "learning_rate": 0.00015765863498034993,
      "loss": 12.3658,
      "step": 11946
    },
    {
      "epoch": 0.650561897177241,
      "grad_norm": 0.6220669970386754,
      "learning_rate": 0.00015765142991987948,
      "loss": 12.3762,
      "step": 11947
    },
    {
      "epoch": 0.650616351173824,
      "grad_norm": 0.5850096096207479,
      "learning_rate": 0.00015764422441110179,
      "loss": 12.3122,
      "step": 11948
    },
    {
      "epoch": 0.650670805170407,
      "grad_norm": 0.5960887585380966,
      "learning_rate": 0.00015763701845407293,
      "loss": 12.2445,
      "step": 11949
    },
    {
      "epoch": 0.65072525916699,
      "grad_norm": 0.6286614967255224,
      "learning_rate": 0.000157629812048849,
      "loss": 12.3693,
      "step": 11950
    },
    {
      "epoch": 0.650779713163573,
      "grad_norm": 0.5755252327414278,
      "learning_rate": 0.00015762260519548596,
      "loss": 12.3737,
      "step": 11951
    },
    {
      "epoch": 0.650834167160156,
      "grad_norm": 0.6015708782836013,
      "learning_rate": 0.0001576153978940399,
      "loss": 12.3598,
      "step": 11952
    },
    {
      "epoch": 0.6508886211567391,
      "grad_norm": 0.5503492430025636,
      "learning_rate": 0.0001576081901445668,
      "loss": 12.147,
      "step": 11953
    },
    {
      "epoch": 0.6509430751533221,
      "grad_norm": 0.6174251701802627,
      "learning_rate": 0.0001576009819471228,
      "loss": 12.3472,
      "step": 11954
    },
    {
      "epoch": 0.650997529149905,
      "grad_norm": 0.6617496217312884,
      "learning_rate": 0.0001575937733017639,
      "loss": 12.335,
      "step": 11955
    },
    {
      "epoch": 0.651051983146488,
      "grad_norm": 0.6619382754910553,
      "learning_rate": 0.00015758656420854615,
      "loss": 12.297,
      "step": 11956
    },
    {
      "epoch": 0.651106437143071,
      "grad_norm": 0.6114447904475865,
      "learning_rate": 0.00015757935466752563,
      "loss": 12.2679,
      "step": 11957
    },
    {
      "epoch": 0.651160891139654,
      "grad_norm": 0.5867492114594228,
      "learning_rate": 0.00015757214467875837,
      "loss": 12.3383,
      "step": 11958
    },
    {
      "epoch": 0.6512153451362371,
      "grad_norm": 0.5545629658830868,
      "learning_rate": 0.00015756493424230045,
      "loss": 12.3276,
      "step": 11959
    },
    {
      "epoch": 0.6512697991328201,
      "grad_norm": 0.6228513492480882,
      "learning_rate": 0.00015755772335820798,
      "loss": 12.2693,
      "step": 11960
    },
    {
      "epoch": 0.6513242531294031,
      "grad_norm": 0.5538829374813401,
      "learning_rate": 0.00015755051202653698,
      "loss": 12.1914,
      "step": 11961
    },
    {
      "epoch": 0.6513787071259861,
      "grad_norm": 0.6761415228572771,
      "learning_rate": 0.00015754330024734357,
      "loss": 12.2367,
      "step": 11962
    },
    {
      "epoch": 0.6514331611225691,
      "grad_norm": 0.5784702625909043,
      "learning_rate": 0.00015753608802068376,
      "loss": 12.3423,
      "step": 11963
    },
    {
      "epoch": 0.6514876151191521,
      "grad_norm": 0.5735770424793095,
      "learning_rate": 0.0001575288753466137,
      "loss": 12.2221,
      "step": 11964
    },
    {
      "epoch": 0.6515420691157352,
      "grad_norm": 0.7926214136031734,
      "learning_rate": 0.0001575216622251895,
      "loss": 12.3241,
      "step": 11965
    },
    {
      "epoch": 0.6515965231123182,
      "grad_norm": 0.48699907812999516,
      "learning_rate": 0.00015751444865646716,
      "loss": 12.2729,
      "step": 11966
    },
    {
      "epoch": 0.6516509771089012,
      "grad_norm": 0.64560719180413,
      "learning_rate": 0.00015750723464050286,
      "loss": 12.3966,
      "step": 11967
    },
    {
      "epoch": 0.6517054311054842,
      "grad_norm": 0.6850859013375421,
      "learning_rate": 0.0001575000201773526,
      "loss": 12.3712,
      "step": 11968
    },
    {
      "epoch": 0.6517598851020672,
      "grad_norm": 0.6505676416890495,
      "learning_rate": 0.0001574928052670726,
      "loss": 12.352,
      "step": 11969
    },
    {
      "epoch": 0.6518143390986503,
      "grad_norm": 0.619633594619558,
      "learning_rate": 0.00015748558990971888,
      "loss": 12.4747,
      "step": 11970
    },
    {
      "epoch": 0.6518687930952333,
      "grad_norm": 0.6128752373979734,
      "learning_rate": 0.00015747837410534757,
      "loss": 12.3141,
      "step": 11971
    },
    {
      "epoch": 0.6519232470918163,
      "grad_norm": 0.5500144202818116,
      "learning_rate": 0.0001574711578540148,
      "loss": 12.3213,
      "step": 11972
    },
    {
      "epoch": 0.6519777010883993,
      "grad_norm": 0.5148268142700347,
      "learning_rate": 0.00015746394115577665,
      "loss": 12.3108,
      "step": 11973
    },
    {
      "epoch": 0.6520321550849822,
      "grad_norm": 0.6139723632259834,
      "learning_rate": 0.00015745672401068928,
      "loss": 12.2607,
      "step": 11974
    },
    {
      "epoch": 0.6520866090815652,
      "grad_norm": 0.5864930516835105,
      "learning_rate": 0.0001574495064188088,
      "loss": 12.0589,
      "step": 11975
    },
    {
      "epoch": 0.6521410630781483,
      "grad_norm": 0.5365876166550633,
      "learning_rate": 0.00015744228838019127,
      "loss": 12.3051,
      "step": 11976
    },
    {
      "epoch": 0.6521955170747313,
      "grad_norm": 0.5499501950636869,
      "learning_rate": 0.0001574350698948929,
      "loss": 12.3038,
      "step": 11977
    },
    {
      "epoch": 0.6522499710713143,
      "grad_norm": 0.6472124705907247,
      "learning_rate": 0.0001574278509629698,
      "loss": 12.3738,
      "step": 11978
    },
    {
      "epoch": 0.6523044250678973,
      "grad_norm": 0.5453405892615903,
      "learning_rate": 0.0001574206315844781,
      "loss": 12.3297,
      "step": 11979
    },
    {
      "epoch": 0.6523588790644803,
      "grad_norm": 0.6329484377503383,
      "learning_rate": 0.00015741341175947392,
      "loss": 12.2167,
      "step": 11980
    },
    {
      "epoch": 0.6524133330610633,
      "grad_norm": 0.5707382286572685,
      "learning_rate": 0.00015740619148801342,
      "loss": 12.1752,
      "step": 11981
    },
    {
      "epoch": 0.6524677870576464,
      "grad_norm": 0.5732738756007905,
      "learning_rate": 0.00015739897077015277,
      "loss": 12.3096,
      "step": 11982
    },
    {
      "epoch": 0.6525222410542294,
      "grad_norm": 0.7108703247791771,
      "learning_rate": 0.00015739174960594809,
      "loss": 12.3917,
      "step": 11983
    },
    {
      "epoch": 0.6525766950508124,
      "grad_norm": 0.5927231949777018,
      "learning_rate": 0.00015738452799545557,
      "loss": 12.4086,
      "step": 11984
    },
    {
      "epoch": 0.6526311490473954,
      "grad_norm": 0.5632680185066232,
      "learning_rate": 0.0001573773059387313,
      "loss": 12.3268,
      "step": 11985
    },
    {
      "epoch": 0.6526856030439784,
      "grad_norm": 0.5575083758906307,
      "learning_rate": 0.00015737008343583148,
      "loss": 12.3571,
      "step": 11986
    },
    {
      "epoch": 0.6527400570405614,
      "grad_norm": 0.5772371140502875,
      "learning_rate": 0.00015736286048681229,
      "loss": 12.3336,
      "step": 11987
    },
    {
      "epoch": 0.6527945110371445,
      "grad_norm": 0.6479961373639044,
      "learning_rate": 0.00015735563709172985,
      "loss": 12.2698,
      "step": 11988
    },
    {
      "epoch": 0.6528489650337275,
      "grad_norm": 0.6505984500018023,
      "learning_rate": 0.00015734841325064038,
      "loss": 12.3785,
      "step": 11989
    },
    {
      "epoch": 0.6529034190303105,
      "grad_norm": 0.5773942885565875,
      "learning_rate": 0.00015734118896360003,
      "loss": 12.3468,
      "step": 11990
    },
    {
      "epoch": 0.6529578730268935,
      "grad_norm": 0.5825216433858514,
      "learning_rate": 0.00015733396423066496,
      "loss": 12.3984,
      "step": 11991
    },
    {
      "epoch": 0.6530123270234764,
      "grad_norm": 0.6430205692159725,
      "learning_rate": 0.00015732673905189136,
      "loss": 12.3095,
      "step": 11992
    },
    {
      "epoch": 0.6530667810200594,
      "grad_norm": 0.6068657586227032,
      "learning_rate": 0.00015731951342733545,
      "loss": 12.2339,
      "step": 11993
    },
    {
      "epoch": 0.6531212350166425,
      "grad_norm": 0.5679813776155134,
      "learning_rate": 0.00015731228735705338,
      "loss": 12.3382,
      "step": 11994
    },
    {
      "epoch": 0.6531756890132255,
      "grad_norm": 0.611364306339243,
      "learning_rate": 0.00015730506084110136,
      "loss": 12.2155,
      "step": 11995
    },
    {
      "epoch": 0.6532301430098085,
      "grad_norm": 0.5722300818041792,
      "learning_rate": 0.00015729783387953558,
      "loss": 12.3042,
      "step": 11996
    },
    {
      "epoch": 0.6532845970063915,
      "grad_norm": 0.6272152796074103,
      "learning_rate": 0.00015729060647241223,
      "loss": 12.3985,
      "step": 11997
    },
    {
      "epoch": 0.6533390510029745,
      "grad_norm": 0.5878780174505018,
      "learning_rate": 0.0001572833786197875,
      "loss": 12.2693,
      "step": 11998
    },
    {
      "epoch": 0.6533935049995575,
      "grad_norm": 0.5765005193129974,
      "learning_rate": 0.00015727615032171764,
      "loss": 12.2288,
      "step": 11999
    },
    {
      "epoch": 0.6534479589961406,
      "grad_norm": 0.5794060488531453,
      "learning_rate": 0.00015726892157825884,
      "loss": 12.2515,
      "step": 12000
    },
    {
      "epoch": 0.6535024129927236,
      "grad_norm": 0.5959742103943618,
      "learning_rate": 0.00015726169238946725,
      "loss": 12.3568,
      "step": 12001
    },
    {
      "epoch": 0.6535568669893066,
      "grad_norm": 0.5656155306898348,
      "learning_rate": 0.00015725446275539917,
      "loss": 12.3548,
      "step": 12002
    },
    {
      "epoch": 0.6536113209858896,
      "grad_norm": 0.5809509610831306,
      "learning_rate": 0.00015724723267611078,
      "loss": 12.2434,
      "step": 12003
    },
    {
      "epoch": 0.6536657749824726,
      "grad_norm": 0.5723505321533915,
      "learning_rate": 0.0001572400021516583,
      "loss": 12.3567,
      "step": 12004
    },
    {
      "epoch": 0.6537202289790557,
      "grad_norm": 0.5640139894343855,
      "learning_rate": 0.000157232771182098,
      "loss": 12.4136,
      "step": 12005
    },
    {
      "epoch": 0.6537746829756387,
      "grad_norm": 0.6097832305343643,
      "learning_rate": 0.00015722553976748604,
      "loss": 12.316,
      "step": 12006
    },
    {
      "epoch": 0.6538291369722217,
      "grad_norm": 0.6075626242319075,
      "learning_rate": 0.00015721830790787868,
      "loss": 12.448,
      "step": 12007
    },
    {
      "epoch": 0.6538835909688047,
      "grad_norm": 0.6135216935236425,
      "learning_rate": 0.00015721107560333217,
      "loss": 12.2811,
      "step": 12008
    },
    {
      "epoch": 0.6539380449653877,
      "grad_norm": 0.5900328630403623,
      "learning_rate": 0.00015720384285390274,
      "loss": 12.3143,
      "step": 12009
    },
    {
      "epoch": 0.6539924989619706,
      "grad_norm": 0.5593576082224303,
      "learning_rate": 0.00015719660965964668,
      "loss": 12.2963,
      "step": 12010
    },
    {
      "epoch": 0.6540469529585538,
      "grad_norm": 0.6323270839974057,
      "learning_rate": 0.00015718937602062015,
      "loss": 12.3916,
      "step": 12011
    },
    {
      "epoch": 0.6541014069551367,
      "grad_norm": 0.634713507031242,
      "learning_rate": 0.00015718214193687945,
      "loss": 12.4702,
      "step": 12012
    },
    {
      "epoch": 0.6541558609517197,
      "grad_norm": 0.5918466121827748,
      "learning_rate": 0.0001571749074084808,
      "loss": 12.3848,
      "step": 12013
    },
    {
      "epoch": 0.6542103149483027,
      "grad_norm": 0.5930684196620623,
      "learning_rate": 0.0001571676724354805,
      "loss": 12.4133,
      "step": 12014
    },
    {
      "epoch": 0.6542647689448857,
      "grad_norm": 0.5565132695659449,
      "learning_rate": 0.0001571604370179348,
      "loss": 12.2225,
      "step": 12015
    },
    {
      "epoch": 0.6543192229414687,
      "grad_norm": 0.571361595467882,
      "learning_rate": 0.00015715320115589995,
      "loss": 12.248,
      "step": 12016
    },
    {
      "epoch": 0.6543736769380518,
      "grad_norm": 0.5981817106947294,
      "learning_rate": 0.0001571459648494322,
      "loss": 12.407,
      "step": 12017
    },
    {
      "epoch": 0.6544281309346348,
      "grad_norm": 0.5843807488937806,
      "learning_rate": 0.00015713872809858788,
      "loss": 12.299,
      "step": 12018
    },
    {
      "epoch": 0.6544825849312178,
      "grad_norm": 0.5713110865905602,
      "learning_rate": 0.00015713149090342321,
      "loss": 12.2149,
      "step": 12019
    },
    {
      "epoch": 0.6545370389278008,
      "grad_norm": 0.6267794263402403,
      "learning_rate": 0.0001571242532639945,
      "loss": 12.4155,
      "step": 12020
    },
    {
      "epoch": 0.6545914929243838,
      "grad_norm": 0.6539645542446826,
      "learning_rate": 0.000157117015180358,
      "loss": 12.3244,
      "step": 12021
    },
    {
      "epoch": 0.6546459469209668,
      "grad_norm": 0.6833958466531187,
      "learning_rate": 0.00015710977665257003,
      "loss": 12.4647,
      "step": 12022
    },
    {
      "epoch": 0.6547004009175499,
      "grad_norm": 0.5835734191459407,
      "learning_rate": 0.0001571025376806868,
      "loss": 12.4632,
      "step": 12023
    },
    {
      "epoch": 0.6547548549141329,
      "grad_norm": 0.5560923116448497,
      "learning_rate": 0.00015709529826476475,
      "loss": 12.2651,
      "step": 12024
    },
    {
      "epoch": 0.6548093089107159,
      "grad_norm": 0.6115189866766454,
      "learning_rate": 0.00015708805840486005,
      "loss": 12.4087,
      "step": 12025
    },
    {
      "epoch": 0.6548637629072989,
      "grad_norm": 0.5607442964773205,
      "learning_rate": 0.000157080818101029,
      "loss": 12.2975,
      "step": 12026
    },
    {
      "epoch": 0.6549182169038819,
      "grad_norm": 0.5778224307829855,
      "learning_rate": 0.00015707357735332797,
      "loss": 12.4361,
      "step": 12027
    },
    {
      "epoch": 0.6549726709004648,
      "grad_norm": 0.6330981188570893,
      "learning_rate": 0.00015706633616181323,
      "loss": 12.3908,
      "step": 12028
    },
    {
      "epoch": 0.655027124897048,
      "grad_norm": 0.6278392162178607,
      "learning_rate": 0.00015705909452654108,
      "loss": 12.3544,
      "step": 12029
    },
    {
      "epoch": 0.655081578893631,
      "grad_norm": 0.5846662559335308,
      "learning_rate": 0.00015705185244756787,
      "loss": 12.2941,
      "step": 12030
    },
    {
      "epoch": 0.6551360328902139,
      "grad_norm": 0.6050336247846226,
      "learning_rate": 0.00015704460992494986,
      "loss": 12.2861,
      "step": 12031
    },
    {
      "epoch": 0.6551904868867969,
      "grad_norm": 0.6321047953668683,
      "learning_rate": 0.0001570373669587434,
      "loss": 12.3586,
      "step": 12032
    },
    {
      "epoch": 0.6552449408833799,
      "grad_norm": 0.608849734510545,
      "learning_rate": 0.00015703012354900483,
      "loss": 12.3521,
      "step": 12033
    },
    {
      "epoch": 0.655299394879963,
      "grad_norm": 0.617231457089484,
      "learning_rate": 0.0001570228796957904,
      "loss": 12.2676,
      "step": 12034
    },
    {
      "epoch": 0.655353848876546,
      "grad_norm": 0.6180049647494912,
      "learning_rate": 0.00015701563539915656,
      "loss": 12.2132,
      "step": 12035
    },
    {
      "epoch": 0.655408302873129,
      "grad_norm": 0.6431172833908424,
      "learning_rate": 0.00015700839065915955,
      "loss": 12.2558,
      "step": 12036
    },
    {
      "epoch": 0.655462756869712,
      "grad_norm": 0.6019823191469429,
      "learning_rate": 0.0001570011454758557,
      "loss": 12.2704,
      "step": 12037
    },
    {
      "epoch": 0.655517210866295,
      "grad_norm": 0.5671224077287408,
      "learning_rate": 0.00015699389984930143,
      "loss": 12.3156,
      "step": 12038
    },
    {
      "epoch": 0.655571664862878,
      "grad_norm": 0.7012576025712719,
      "learning_rate": 0.00015698665377955303,
      "loss": 12.43,
      "step": 12039
    },
    {
      "epoch": 0.6556261188594611,
      "grad_norm": 0.6061221280777732,
      "learning_rate": 0.00015697940726666683,
      "loss": 12.2937,
      "step": 12040
    },
    {
      "epoch": 0.6556805728560441,
      "grad_norm": 0.7156210816199511,
      "learning_rate": 0.00015697216031069925,
      "loss": 12.3705,
      "step": 12041
    },
    {
      "epoch": 0.6557350268526271,
      "grad_norm": 0.6125638351685208,
      "learning_rate": 0.00015696491291170657,
      "loss": 12.3893,
      "step": 12042
    },
    {
      "epoch": 0.6557894808492101,
      "grad_norm": 0.6318843500494461,
      "learning_rate": 0.00015695766506974517,
      "loss": 12.4039,
      "step": 12043
    },
    {
      "epoch": 0.6558439348457931,
      "grad_norm": 0.6600868298812459,
      "learning_rate": 0.00015695041678487142,
      "loss": 12.2972,
      "step": 12044
    },
    {
      "epoch": 0.6558983888423761,
      "grad_norm": 0.542910202250795,
      "learning_rate": 0.00015694316805714165,
      "loss": 12.2664,
      "step": 12045
    },
    {
      "epoch": 0.6559528428389592,
      "grad_norm": 0.5845324019516077,
      "learning_rate": 0.0001569359188866123,
      "loss": 12.3188,
      "step": 12046
    },
    {
      "epoch": 0.6560072968355422,
      "grad_norm": 0.7004534797723055,
      "learning_rate": 0.00015692866927333968,
      "loss": 12.41,
      "step": 12047
    },
    {
      "epoch": 0.6560617508321251,
      "grad_norm": 0.6112982977934652,
      "learning_rate": 0.00015692141921738018,
      "loss": 12.3632,
      "step": 12048
    },
    {
      "epoch": 0.6561162048287081,
      "grad_norm": 0.6132310081784185,
      "learning_rate": 0.00015691416871879018,
      "loss": 12.3798,
      "step": 12049
    },
    {
      "epoch": 0.6561706588252911,
      "grad_norm": 0.6721167651882155,
      "learning_rate": 0.00015690691777762604,
      "loss": 12.2113,
      "step": 12050
    },
    {
      "epoch": 0.6562251128218741,
      "grad_norm": 0.6058156618250249,
      "learning_rate": 0.0001568996663939442,
      "loss": 12.4447,
      "step": 12051
    },
    {
      "epoch": 0.6562795668184572,
      "grad_norm": 0.6217678395993428,
      "learning_rate": 0.000156892414567801,
      "loss": 12.4583,
      "step": 12052
    },
    {
      "epoch": 0.6563340208150402,
      "grad_norm": 0.8360757048261092,
      "learning_rate": 0.00015688516229925284,
      "loss": 12.3777,
      "step": 12053
    },
    {
      "epoch": 0.6563884748116232,
      "grad_norm": 0.5595752466357918,
      "learning_rate": 0.0001568779095883561,
      "loss": 12.2718,
      "step": 12054
    },
    {
      "epoch": 0.6564429288082062,
      "grad_norm": 0.5796818253610839,
      "learning_rate": 0.00015687065643516722,
      "loss": 12.1915,
      "step": 12055
    },
    {
      "epoch": 0.6564973828047892,
      "grad_norm": 0.6068369376036751,
      "learning_rate": 0.00015686340283974258,
      "loss": 12.1375,
      "step": 12056
    },
    {
      "epoch": 0.6565518368013722,
      "grad_norm": 0.5964555552588784,
      "learning_rate": 0.00015685614880213856,
      "loss": 12.385,
      "step": 12057
    },
    {
      "epoch": 0.6566062907979553,
      "grad_norm": 0.5893076488767527,
      "learning_rate": 0.0001568488943224116,
      "loss": 12.3554,
      "step": 12058
    },
    {
      "epoch": 0.6566607447945383,
      "grad_norm": 0.6133767840778955,
      "learning_rate": 0.0001568416394006181,
      "loss": 12.3736,
      "step": 12059
    },
    {
      "epoch": 0.6567151987911213,
      "grad_norm": 0.6707278629224666,
      "learning_rate": 0.00015683438403681448,
      "loss": 12.4436,
      "step": 12060
    },
    {
      "epoch": 0.6567696527877043,
      "grad_norm": 0.5560126577532122,
      "learning_rate": 0.0001568271282310572,
      "loss": 12.1057,
      "step": 12061
    },
    {
      "epoch": 0.6568241067842873,
      "grad_norm": 0.6215607526503611,
      "learning_rate": 0.0001568198719834026,
      "loss": 12.1294,
      "step": 12062
    },
    {
      "epoch": 0.6568785607808703,
      "grad_norm": 0.6286325445732135,
      "learning_rate": 0.00015681261529390715,
      "loss": 12.2769,
      "step": 12063
    },
    {
      "epoch": 0.6569330147774534,
      "grad_norm": 0.5312290249893128,
      "learning_rate": 0.00015680535816262728,
      "loss": 12.2747,
      "step": 12064
    },
    {
      "epoch": 0.6569874687740364,
      "grad_norm": 0.5738259013012558,
      "learning_rate": 0.00015679810058961944,
      "loss": 12.3717,
      "step": 12065
    },
    {
      "epoch": 0.6570419227706193,
      "grad_norm": 0.5940529983518259,
      "learning_rate": 0.00015679084257494002,
      "loss": 12.4212,
      "step": 12066
    },
    {
      "epoch": 0.6570963767672023,
      "grad_norm": 0.583664293392366,
      "learning_rate": 0.00015678358411864548,
      "loss": 12.2232,
      "step": 12067
    },
    {
      "epoch": 0.6571508307637853,
      "grad_norm": 0.5896927482492248,
      "learning_rate": 0.0001567763252207923,
      "loss": 12.4077,
      "step": 12068
    },
    {
      "epoch": 0.6572052847603684,
      "grad_norm": 0.6049544554721654,
      "learning_rate": 0.00015676906588143685,
      "loss": 12.4348,
      "step": 12069
    },
    {
      "epoch": 0.6572597387569514,
      "grad_norm": 0.5725497093698602,
      "learning_rate": 0.00015676180610063565,
      "loss": 12.4809,
      "step": 12070
    },
    {
      "epoch": 0.6573141927535344,
      "grad_norm": 0.7405786368932967,
      "learning_rate": 0.00015675454587844515,
      "loss": 12.2816,
      "step": 12071
    },
    {
      "epoch": 0.6573686467501174,
      "grad_norm": 0.600552284188916,
      "learning_rate": 0.00015674728521492172,
      "loss": 12.2328,
      "step": 12072
    },
    {
      "epoch": 0.6574231007467004,
      "grad_norm": 0.5390920058221059,
      "learning_rate": 0.0001567400241101219,
      "loss": 12.2939,
      "step": 12073
    },
    {
      "epoch": 0.6574775547432834,
      "grad_norm": 0.5996252240800342,
      "learning_rate": 0.0001567327625641022,
      "loss": 12.2745,
      "step": 12074
    },
    {
      "epoch": 0.6575320087398665,
      "grad_norm": 0.6203639181018388,
      "learning_rate": 0.00015672550057691895,
      "loss": 12.2983,
      "step": 12075
    },
    {
      "epoch": 0.6575864627364495,
      "grad_norm": 0.5831791648502749,
      "learning_rate": 0.00015671823814862875,
      "loss": 12.2538,
      "step": 12076
    },
    {
      "epoch": 0.6576409167330325,
      "grad_norm": 0.6456117615006848,
      "learning_rate": 0.00015671097527928795,
      "loss": 12.2313,
      "step": 12077
    },
    {
      "epoch": 0.6576953707296155,
      "grad_norm": 0.6120533649848845,
      "learning_rate": 0.00015670371196895317,
      "loss": 12.3152,
      "step": 12078
    },
    {
      "epoch": 0.6577498247261985,
      "grad_norm": 0.5909407598412671,
      "learning_rate": 0.00015669644821768078,
      "loss": 12.4194,
      "step": 12079
    },
    {
      "epoch": 0.6578042787227815,
      "grad_norm": 0.5897918822232837,
      "learning_rate": 0.0001566891840255273,
      "loss": 12.3486,
      "step": 12080
    },
    {
      "epoch": 0.6578587327193646,
      "grad_norm": 0.5691125296830504,
      "learning_rate": 0.00015668191939254925,
      "loss": 12.2627,
      "step": 12081
    },
    {
      "epoch": 0.6579131867159476,
      "grad_norm": 0.6035693166090563,
      "learning_rate": 0.00015667465431880304,
      "loss": 12.2438,
      "step": 12082
    },
    {
      "epoch": 0.6579676407125306,
      "grad_norm": 0.6274229351147108,
      "learning_rate": 0.00015666738880434523,
      "loss": 12.2951,
      "step": 12083
    },
    {
      "epoch": 0.6580220947091135,
      "grad_norm": 0.5970422016358655,
      "learning_rate": 0.00015666012284923231,
      "loss": 12.1996,
      "step": 12084
    },
    {
      "epoch": 0.6580765487056965,
      "grad_norm": 0.613287883262713,
      "learning_rate": 0.00015665285645352078,
      "loss": 12.2262,
      "step": 12085
    },
    {
      "epoch": 0.6581310027022795,
      "grad_norm": 0.8218009396210703,
      "learning_rate": 0.0001566455896172671,
      "loss": 12.6371,
      "step": 12086
    },
    {
      "epoch": 0.6581854566988626,
      "grad_norm": 0.6123225370787431,
      "learning_rate": 0.00015663832234052787,
      "loss": 12.2022,
      "step": 12087
    },
    {
      "epoch": 0.6582399106954456,
      "grad_norm": 0.5879275100204007,
      "learning_rate": 0.0001566310546233595,
      "loss": 12.4095,
      "step": 12088
    },
    {
      "epoch": 0.6582943646920286,
      "grad_norm": 0.5717092565134987,
      "learning_rate": 0.0001566237864658186,
      "loss": 12.3095,
      "step": 12089
    },
    {
      "epoch": 0.6583488186886116,
      "grad_norm": 0.5784460652054834,
      "learning_rate": 0.0001566165178679616,
      "loss": 12.219,
      "step": 12090
    },
    {
      "epoch": 0.6584032726851946,
      "grad_norm": 0.643811496962863,
      "learning_rate": 0.0001566092488298451,
      "loss": 12.3034,
      "step": 12091
    },
    {
      "epoch": 0.6584577266817776,
      "grad_norm": 0.5680520389238057,
      "learning_rate": 0.00015660197935152555,
      "loss": 12.2748,
      "step": 12092
    },
    {
      "epoch": 0.6585121806783607,
      "grad_norm": 0.5790373555853809,
      "learning_rate": 0.00015659470943305955,
      "loss": 12.3464,
      "step": 12093
    },
    {
      "epoch": 0.6585666346749437,
      "grad_norm": 0.6087938649370235,
      "learning_rate": 0.00015658743907450356,
      "loss": 12.3078,
      "step": 12094
    },
    {
      "epoch": 0.6586210886715267,
      "grad_norm": 0.6270072053969875,
      "learning_rate": 0.00015658016827591417,
      "loss": 12.1714,
      "step": 12095
    },
    {
      "epoch": 0.6586755426681097,
      "grad_norm": 0.5861641476491468,
      "learning_rate": 0.0001565728970373479,
      "loss": 12.3368,
      "step": 12096
    },
    {
      "epoch": 0.6587299966646927,
      "grad_norm": 0.5804427884989929,
      "learning_rate": 0.0001565656253588613,
      "loss": 12.1728,
      "step": 12097
    },
    {
      "epoch": 0.6587844506612757,
      "grad_norm": 0.5704907114632229,
      "learning_rate": 0.00015655835324051093,
      "loss": 12.053,
      "step": 12098
    },
    {
      "epoch": 0.6588389046578588,
      "grad_norm": 0.591580493014265,
      "learning_rate": 0.0001565510806823533,
      "loss": 12.3192,
      "step": 12099
    },
    {
      "epoch": 0.6588933586544418,
      "grad_norm": 0.5779853695319296,
      "learning_rate": 0.000156543807684445,
      "loss": 12.2956,
      "step": 12100
    },
    {
      "epoch": 0.6589478126510248,
      "grad_norm": 0.6231852103374036,
      "learning_rate": 0.00015653653424684255,
      "loss": 12.2494,
      "step": 12101
    },
    {
      "epoch": 0.6590022666476077,
      "grad_norm": 0.6667754409013602,
      "learning_rate": 0.0001565292603696025,
      "loss": 12.3361,
      "step": 12102
    },
    {
      "epoch": 0.6590567206441907,
      "grad_norm": 0.5401924422265222,
      "learning_rate": 0.0001565219860527815,
      "loss": 12.2276,
      "step": 12103
    },
    {
      "epoch": 0.6591111746407738,
      "grad_norm": 0.6918791259186888,
      "learning_rate": 0.00015651471129643602,
      "loss": 12.4344,
      "step": 12104
    },
    {
      "epoch": 0.6591656286373568,
      "grad_norm": 0.6604993076391991,
      "learning_rate": 0.0001565074361006227,
      "loss": 12.134,
      "step": 12105
    },
    {
      "epoch": 0.6592200826339398,
      "grad_norm": 0.5515241712824276,
      "learning_rate": 0.00015650016046539806,
      "loss": 12.3196,
      "step": 12106
    },
    {
      "epoch": 0.6592745366305228,
      "grad_norm": 0.6419301595352817,
      "learning_rate": 0.00015649288439081868,
      "loss": 12.0543,
      "step": 12107
    },
    {
      "epoch": 0.6593289906271058,
      "grad_norm": 0.6839742489781926,
      "learning_rate": 0.00015648560787694118,
      "loss": 12.337,
      "step": 12108
    },
    {
      "epoch": 0.6593834446236888,
      "grad_norm": 0.5489400548407735,
      "learning_rate": 0.0001564783309238221,
      "loss": 12.3295,
      "step": 12109
    },
    {
      "epoch": 0.6594378986202719,
      "grad_norm": 0.5588116094373012,
      "learning_rate": 0.0001564710535315181,
      "loss": 12.3351,
      "step": 12110
    },
    {
      "epoch": 0.6594923526168549,
      "grad_norm": 0.6258296469166352,
      "learning_rate": 0.00015646377570008565,
      "loss": 12.3593,
      "step": 12111
    },
    {
      "epoch": 0.6595468066134379,
      "grad_norm": 0.5714303544699514,
      "learning_rate": 0.00015645649742958146,
      "loss": 12.2158,
      "step": 12112
    },
    {
      "epoch": 0.6596012606100209,
      "grad_norm": 0.5690123486052074,
      "learning_rate": 0.00015644921872006205,
      "loss": 12.3059,
      "step": 12113
    },
    {
      "epoch": 0.6596557146066039,
      "grad_norm": 0.5998350963031981,
      "learning_rate": 0.00015644193957158406,
      "loss": 12.2973,
      "step": 12114
    },
    {
      "epoch": 0.6597101686031869,
      "grad_norm": 0.6126527081622569,
      "learning_rate": 0.0001564346599842041,
      "loss": 12.3787,
      "step": 12115
    },
    {
      "epoch": 0.65976462259977,
      "grad_norm": 0.6175210777858374,
      "learning_rate": 0.00015642737995797873,
      "loss": 12.3521,
      "step": 12116
    },
    {
      "epoch": 0.659819076596353,
      "grad_norm": 0.6616911257059388,
      "learning_rate": 0.0001564200994929646,
      "loss": 12.1806,
      "step": 12117
    },
    {
      "epoch": 0.659873530592936,
      "grad_norm": 0.62720735050687,
      "learning_rate": 0.00015641281858921833,
      "loss": 12.34,
      "step": 12118
    },
    {
      "epoch": 0.659927984589519,
      "grad_norm": 0.5898731792627565,
      "learning_rate": 0.00015640553724679648,
      "loss": 12.2155,
      "step": 12119
    },
    {
      "epoch": 0.659982438586102,
      "grad_norm": 0.6070751981197668,
      "learning_rate": 0.00015639825546575576,
      "loss": 12.1565,
      "step": 12120
    },
    {
      "epoch": 0.6600368925826849,
      "grad_norm": 0.6412845745278387,
      "learning_rate": 0.00015639097324615273,
      "loss": 12.5173,
      "step": 12121
    },
    {
      "epoch": 0.660091346579268,
      "grad_norm": 0.6269784738662666,
      "learning_rate": 0.00015638369058804404,
      "loss": 12.2585,
      "step": 12122
    },
    {
      "epoch": 0.660145800575851,
      "grad_norm": 0.5692248201202506,
      "learning_rate": 0.0001563764074914863,
      "loss": 12.1797,
      "step": 12123
    },
    {
      "epoch": 0.660200254572434,
      "grad_norm": 0.5412511925083626,
      "learning_rate": 0.00015636912395653618,
      "loss": 12.3225,
      "step": 12124
    },
    {
      "epoch": 0.660254708569017,
      "grad_norm": 0.5678289684620178,
      "learning_rate": 0.00015636183998325026,
      "loss": 12.332,
      "step": 12125
    },
    {
      "epoch": 0.6603091625656,
      "grad_norm": 0.6016993154089291,
      "learning_rate": 0.00015635455557168527,
      "loss": 12.3073,
      "step": 12126
    },
    {
      "epoch": 0.660363616562183,
      "grad_norm": 0.6483298042378929,
      "learning_rate": 0.00015634727072189782,
      "loss": 12.2413,
      "step": 12127
    },
    {
      "epoch": 0.6604180705587661,
      "grad_norm": 0.5863492923591763,
      "learning_rate": 0.00015633998543394448,
      "loss": 12.2328,
      "step": 12128
    },
    {
      "epoch": 0.6604725245553491,
      "grad_norm": 0.5801873293209294,
      "learning_rate": 0.00015633269970788201,
      "loss": 12.3545,
      "step": 12129
    },
    {
      "epoch": 0.6605269785519321,
      "grad_norm": 0.5478890482606156,
      "learning_rate": 0.00015632541354376698,
      "loss": 12.3379,
      "step": 12130
    },
    {
      "epoch": 0.6605814325485151,
      "grad_norm": 0.5774737062533749,
      "learning_rate": 0.00015631812694165612,
      "loss": 12.3044,
      "step": 12131
    },
    {
      "epoch": 0.6606358865450981,
      "grad_norm": 0.5952639698733143,
      "learning_rate": 0.00015631083990160605,
      "loss": 12.2476,
      "step": 12132
    },
    {
      "epoch": 0.6606903405416811,
      "grad_norm": 0.523753329481436,
      "learning_rate": 0.00015630355242367348,
      "loss": 12.1817,
      "step": 12133
    },
    {
      "epoch": 0.6607447945382642,
      "grad_norm": 0.5474573088759221,
      "learning_rate": 0.000156296264507915,
      "loss": 12.2149,
      "step": 12134
    },
    {
      "epoch": 0.6607992485348472,
      "grad_norm": 0.6304369889579475,
      "learning_rate": 0.0001562889761543873,
      "loss": 12.4091,
      "step": 12135
    },
    {
      "epoch": 0.6608537025314302,
      "grad_norm": 0.6699170088289395,
      "learning_rate": 0.00015628168736314717,
      "loss": 12.4412,
      "step": 12136
    },
    {
      "epoch": 0.6609081565280132,
      "grad_norm": 0.5538506368520014,
      "learning_rate": 0.00015627439813425115,
      "loss": 12.282,
      "step": 12137
    },
    {
      "epoch": 0.6609626105245962,
      "grad_norm": 0.5978091609740449,
      "learning_rate": 0.00015626710846775596,
      "loss": 12.3531,
      "step": 12138
    },
    {
      "epoch": 0.6610170645211793,
      "grad_norm": 0.5623515777462447,
      "learning_rate": 0.00015625981836371833,
      "loss": 12.3091,
      "step": 12139
    },
    {
      "epoch": 0.6610715185177622,
      "grad_norm": 0.5864998538286688,
      "learning_rate": 0.0001562525278221949,
      "loss": 12.1195,
      "step": 12140
    },
    {
      "epoch": 0.6611259725143452,
      "grad_norm": 0.5896243382799916,
      "learning_rate": 0.00015624523684324237,
      "loss": 12.2116,
      "step": 12141
    },
    {
      "epoch": 0.6611804265109282,
      "grad_norm": 0.5491996186435824,
      "learning_rate": 0.0001562379454269175,
      "loss": 12.2354,
      "step": 12142
    },
    {
      "epoch": 0.6612348805075112,
      "grad_norm": 0.6777309207922793,
      "learning_rate": 0.00015623065357327684,
      "loss": 12.2775,
      "step": 12143
    },
    {
      "epoch": 0.6612893345040942,
      "grad_norm": 0.6302362156115535,
      "learning_rate": 0.00015622336128237726,
      "loss": 12.1918,
      "step": 12144
    },
    {
      "epoch": 0.6613437885006773,
      "grad_norm": 0.650374195038451,
      "learning_rate": 0.00015621606855427538,
      "loss": 12.3677,
      "step": 12145
    },
    {
      "epoch": 0.6613982424972603,
      "grad_norm": 0.5932691651104086,
      "learning_rate": 0.0001562087753890279,
      "loss": 12.3541,
      "step": 12146
    },
    {
      "epoch": 0.6614526964938433,
      "grad_norm": 0.5680699252633874,
      "learning_rate": 0.00015620148178669161,
      "loss": 12.3498,
      "step": 12147
    },
    {
      "epoch": 0.6615071504904263,
      "grad_norm": 0.6155619498335252,
      "learning_rate": 0.0001561941877473231,
      "loss": 12.195,
      "step": 12148
    },
    {
      "epoch": 0.6615616044870093,
      "grad_norm": 0.6051882680021209,
      "learning_rate": 0.0001561868932709792,
      "loss": 12.1634,
      "step": 12149
    },
    {
      "epoch": 0.6616160584835923,
      "grad_norm": 0.5986268325000902,
      "learning_rate": 0.00015617959835771662,
      "loss": 12.3676,
      "step": 12150
    },
    {
      "epoch": 0.6616705124801754,
      "grad_norm": 0.5727966647803312,
      "learning_rate": 0.00015617230300759202,
      "loss": 12.2913,
      "step": 12151
    },
    {
      "epoch": 0.6617249664767584,
      "grad_norm": 0.5545194798519972,
      "learning_rate": 0.00015616500722066219,
      "loss": 12.377,
      "step": 12152
    },
    {
      "epoch": 0.6617794204733414,
      "grad_norm": 0.5641449445635649,
      "learning_rate": 0.00015615771099698384,
      "loss": 12.2211,
      "step": 12153
    },
    {
      "epoch": 0.6618338744699244,
      "grad_norm": 0.60254720612745,
      "learning_rate": 0.0001561504143366137,
      "loss": 12.2442,
      "step": 12154
    },
    {
      "epoch": 0.6618883284665074,
      "grad_norm": 0.5907651025492713,
      "learning_rate": 0.00015614311723960854,
      "loss": 12.35,
      "step": 12155
    },
    {
      "epoch": 0.6619427824630904,
      "grad_norm": 0.708172101181777,
      "learning_rate": 0.0001561358197060251,
      "loss": 12.3802,
      "step": 12156
    },
    {
      "epoch": 0.6619972364596735,
      "grad_norm": 0.5675398599531235,
      "learning_rate": 0.0001561285217359201,
      "loss": 12.3175,
      "step": 12157
    },
    {
      "epoch": 0.6620516904562564,
      "grad_norm": 0.6494733439949993,
      "learning_rate": 0.0001561212233293503,
      "loss": 12.2998,
      "step": 12158
    },
    {
      "epoch": 0.6621061444528394,
      "grad_norm": 0.5949033966930486,
      "learning_rate": 0.00015611392448637243,
      "loss": 12.2403,
      "step": 12159
    },
    {
      "epoch": 0.6621605984494224,
      "grad_norm": 0.597294651572877,
      "learning_rate": 0.0001561066252070433,
      "loss": 12.3803,
      "step": 12160
    },
    {
      "epoch": 0.6622150524460054,
      "grad_norm": 0.6428094862347965,
      "learning_rate": 0.00015609932549141966,
      "loss": 12.4209,
      "step": 12161
    },
    {
      "epoch": 0.6622695064425884,
      "grad_norm": 0.6673986827425881,
      "learning_rate": 0.00015609202533955823,
      "loss": 12.3467,
      "step": 12162
    },
    {
      "epoch": 0.6623239604391715,
      "grad_norm": 0.5701563550210614,
      "learning_rate": 0.00015608472475151582,
      "loss": 12.1723,
      "step": 12163
    },
    {
      "epoch": 0.6623784144357545,
      "grad_norm": 0.627956929952825,
      "learning_rate": 0.00015607742372734915,
      "loss": 12.3973,
      "step": 12164
    },
    {
      "epoch": 0.6624328684323375,
      "grad_norm": 0.6038635446801585,
      "learning_rate": 0.00015607012226711507,
      "loss": 12.3268,
      "step": 12165
    },
    {
      "epoch": 0.6624873224289205,
      "grad_norm": 0.5473757396652543,
      "learning_rate": 0.0001560628203708703,
      "loss": 12.2702,
      "step": 12166
    },
    {
      "epoch": 0.6625417764255035,
      "grad_norm": 0.554151613594465,
      "learning_rate": 0.00015605551803867163,
      "loss": 12.2511,
      "step": 12167
    },
    {
      "epoch": 0.6625962304220866,
      "grad_norm": 0.6729198964343531,
      "learning_rate": 0.00015604821527057588,
      "loss": 12.4947,
      "step": 12168
    },
    {
      "epoch": 0.6626506844186696,
      "grad_norm": 0.6180742038416029,
      "learning_rate": 0.00015604091206663977,
      "loss": 12.301,
      "step": 12169
    },
    {
      "epoch": 0.6627051384152526,
      "grad_norm": 0.5713033910454256,
      "learning_rate": 0.00015603360842692015,
      "loss": 12.3176,
      "step": 12170
    },
    {
      "epoch": 0.6627595924118356,
      "grad_norm": 0.599816381001002,
      "learning_rate": 0.0001560263043514738,
      "loss": 12.2241,
      "step": 12171
    },
    {
      "epoch": 0.6628140464084186,
      "grad_norm": 0.5693130389885763,
      "learning_rate": 0.0001560189998403575,
      "loss": 12.353,
      "step": 12172
    },
    {
      "epoch": 0.6628685004050016,
      "grad_norm": 0.5702316741912086,
      "learning_rate": 0.00015601169489362805,
      "loss": 12.1861,
      "step": 12173
    },
    {
      "epoch": 0.6629229544015847,
      "grad_norm": 0.5621961876678578,
      "learning_rate": 0.0001560043895113423,
      "loss": 12.1976,
      "step": 12174
    },
    {
      "epoch": 0.6629774083981677,
      "grad_norm": 0.6012569929692483,
      "learning_rate": 0.000155997083693557,
      "loss": 12.238,
      "step": 12175
    },
    {
      "epoch": 0.6630318623947506,
      "grad_norm": 0.5302971343374439,
      "learning_rate": 0.00015598977744032898,
      "loss": 12.2912,
      "step": 12176
    },
    {
      "epoch": 0.6630863163913336,
      "grad_norm": 0.6054268327809826,
      "learning_rate": 0.00015598247075171507,
      "loss": 12.3391,
      "step": 12177
    },
    {
      "epoch": 0.6631407703879166,
      "grad_norm": 0.6130891768269731,
      "learning_rate": 0.00015597516362777212,
      "loss": 12.3841,
      "step": 12178
    },
    {
      "epoch": 0.6631952243844996,
      "grad_norm": 0.595933876472692,
      "learning_rate": 0.00015596785606855686,
      "loss": 12.3136,
      "step": 12179
    },
    {
      "epoch": 0.6632496783810827,
      "grad_norm": 0.5843406076220462,
      "learning_rate": 0.00015596054807412617,
      "loss": 12.1744,
      "step": 12180
    },
    {
      "epoch": 0.6633041323776657,
      "grad_norm": 0.6031736206179926,
      "learning_rate": 0.00015595323964453687,
      "loss": 12.287,
      "step": 12181
    },
    {
      "epoch": 0.6633585863742487,
      "grad_norm": 0.599656465734208,
      "learning_rate": 0.00015594593077984583,
      "loss": 12.2659,
      "step": 12182
    },
    {
      "epoch": 0.6634130403708317,
      "grad_norm": 0.5738100710286648,
      "learning_rate": 0.00015593862148010983,
      "loss": 12.417,
      "step": 12183
    },
    {
      "epoch": 0.6634674943674147,
      "grad_norm": 0.6222520183921914,
      "learning_rate": 0.0001559313117453857,
      "loss": 12.3374,
      "step": 12184
    },
    {
      "epoch": 0.6635219483639977,
      "grad_norm": 0.543589456981991,
      "learning_rate": 0.00015592400157573034,
      "loss": 12.2794,
      "step": 12185
    },
    {
      "epoch": 0.6635764023605808,
      "grad_norm": 0.6115018237920168,
      "learning_rate": 0.00015591669097120056,
      "loss": 12.2219,
      "step": 12186
    },
    {
      "epoch": 0.6636308563571638,
      "grad_norm": 0.5458482281292727,
      "learning_rate": 0.00015590937993185323,
      "loss": 12.2878,
      "step": 12187
    },
    {
      "epoch": 0.6636853103537468,
      "grad_norm": 0.5843729540859018,
      "learning_rate": 0.0001559020684577452,
      "loss": 12.3356,
      "step": 12188
    },
    {
      "epoch": 0.6637397643503298,
      "grad_norm": 0.6558099047970367,
      "learning_rate": 0.00015589475654893326,
      "loss": 12.1897,
      "step": 12189
    },
    {
      "epoch": 0.6637942183469128,
      "grad_norm": 0.5952426948130463,
      "learning_rate": 0.00015588744420547433,
      "loss": 12.3942,
      "step": 12190
    },
    {
      "epoch": 0.6638486723434958,
      "grad_norm": 0.5839362546215011,
      "learning_rate": 0.0001558801314274253,
      "loss": 12.4069,
      "step": 12191
    },
    {
      "epoch": 0.6639031263400789,
      "grad_norm": 0.8111332595455524,
      "learning_rate": 0.00015587281821484295,
      "loss": 12.4592,
      "step": 12192
    },
    {
      "epoch": 0.6639575803366619,
      "grad_norm": 0.5717171049299086,
      "learning_rate": 0.00015586550456778424,
      "loss": 12.2548,
      "step": 12193
    },
    {
      "epoch": 0.6640120343332448,
      "grad_norm": 0.6587477339077662,
      "learning_rate": 0.00015585819048630597,
      "loss": 12.3692,
      "step": 12194
    },
    {
      "epoch": 0.6640664883298278,
      "grad_norm": 0.7716794105304356,
      "learning_rate": 0.00015585087597046505,
      "loss": 12.1079,
      "step": 12195
    },
    {
      "epoch": 0.6641209423264108,
      "grad_norm": 0.6176557578909904,
      "learning_rate": 0.00015584356102031833,
      "loss": 12.402,
      "step": 12196
    },
    {
      "epoch": 0.6641753963229938,
      "grad_norm": 0.860535049891481,
      "learning_rate": 0.00015583624563592275,
      "loss": 12.201,
      "step": 12197
    },
    {
      "epoch": 0.6642298503195769,
      "grad_norm": 0.6206160971273634,
      "learning_rate": 0.00015582892981733513,
      "loss": 12.32,
      "step": 12198
    },
    {
      "epoch": 0.6642843043161599,
      "grad_norm": 0.5984237246171267,
      "learning_rate": 0.00015582161356461242,
      "loss": 12.2633,
      "step": 12199
    },
    {
      "epoch": 0.6643387583127429,
      "grad_norm": 0.7078818570229009,
      "learning_rate": 0.00015581429687781147,
      "loss": 12.3191,
      "step": 12200
    },
    {
      "epoch": 0.6643932123093259,
      "grad_norm": 0.6105722808230988,
      "learning_rate": 0.00015580697975698917,
      "loss": 12.2233,
      "step": 12201
    },
    {
      "epoch": 0.6644476663059089,
      "grad_norm": 0.6439423699459289,
      "learning_rate": 0.00015579966220220247,
      "loss": 12.3645,
      "step": 12202
    },
    {
      "epoch": 0.664502120302492,
      "grad_norm": 0.6489363836579027,
      "learning_rate": 0.0001557923442135082,
      "loss": 12.3112,
      "step": 12203
    },
    {
      "epoch": 0.664556574299075,
      "grad_norm": 0.6137942271046831,
      "learning_rate": 0.00015578502579096336,
      "loss": 12.32,
      "step": 12204
    },
    {
      "epoch": 0.664611028295658,
      "grad_norm": 0.5813939230459977,
      "learning_rate": 0.00015577770693462475,
      "loss": 12.3516,
      "step": 12205
    },
    {
      "epoch": 0.664665482292241,
      "grad_norm": 0.7177450492577477,
      "learning_rate": 0.00015577038764454936,
      "loss": 12.2759,
      "step": 12206
    },
    {
      "epoch": 0.664719936288824,
      "grad_norm": 0.5908558121589046,
      "learning_rate": 0.00015576306792079408,
      "loss": 12.264,
      "step": 12207
    },
    {
      "epoch": 0.664774390285407,
      "grad_norm": 0.5941370823923019,
      "learning_rate": 0.00015575574776341582,
      "loss": 12.3181,
      "step": 12208
    },
    {
      "epoch": 0.6648288442819901,
      "grad_norm": 0.6088651213946604,
      "learning_rate": 0.00015574842717247154,
      "loss": 12.2413,
      "step": 12209
    },
    {
      "epoch": 0.6648832982785731,
      "grad_norm": 0.6926069129225368,
      "learning_rate": 0.00015574110614801812,
      "loss": 12.2957,
      "step": 12210
    },
    {
      "epoch": 0.6649377522751561,
      "grad_norm": 0.5973284373065706,
      "learning_rate": 0.00015573378469011252,
      "loss": 12.4042,
      "step": 12211
    },
    {
      "epoch": 0.664992206271739,
      "grad_norm": 0.6332900499532489,
      "learning_rate": 0.00015572646279881166,
      "loss": 12.3268,
      "step": 12212
    },
    {
      "epoch": 0.665046660268322,
      "grad_norm": 0.5964173395692406,
      "learning_rate": 0.0001557191404741725,
      "loss": 12.3616,
      "step": 12213
    },
    {
      "epoch": 0.665101114264905,
      "grad_norm": 0.5895920125238943,
      "learning_rate": 0.00015571181771625192,
      "loss": 12.3699,
      "step": 12214
    },
    {
      "epoch": 0.6651555682614881,
      "grad_norm": 0.6704097046230486,
      "learning_rate": 0.00015570449452510693,
      "loss": 12.502,
      "step": 12215
    },
    {
      "epoch": 0.6652100222580711,
      "grad_norm": 0.5794962651301736,
      "learning_rate": 0.00015569717090079444,
      "loss": 12.1452,
      "step": 12216
    },
    {
      "epoch": 0.6652644762546541,
      "grad_norm": 0.5920778321000353,
      "learning_rate": 0.0001556898468433714,
      "loss": 12.1568,
      "step": 12217
    },
    {
      "epoch": 0.6653189302512371,
      "grad_norm": 0.6403214063903778,
      "learning_rate": 0.0001556825223528948,
      "loss": 12.2718,
      "step": 12218
    },
    {
      "epoch": 0.6653733842478201,
      "grad_norm": 0.6007841160667045,
      "learning_rate": 0.00015567519742942153,
      "loss": 12.3545,
      "step": 12219
    },
    {
      "epoch": 0.6654278382444031,
      "grad_norm": 0.6186117578106466,
      "learning_rate": 0.00015566787207300863,
      "loss": 12.3221,
      "step": 12220
    },
    {
      "epoch": 0.6654822922409862,
      "grad_norm": 0.6571254360260127,
      "learning_rate": 0.000155660546283713,
      "loss": 12.4017,
      "step": 12221
    },
    {
      "epoch": 0.6655367462375692,
      "grad_norm": 0.7613889115243458,
      "learning_rate": 0.00015565322006159163,
      "loss": 12.3841,
      "step": 12222
    },
    {
      "epoch": 0.6655912002341522,
      "grad_norm": 0.6264136875700904,
      "learning_rate": 0.00015564589340670147,
      "loss": 12.3541,
      "step": 12223
    },
    {
      "epoch": 0.6656456542307352,
      "grad_norm": 0.5637844084922261,
      "learning_rate": 0.00015563856631909954,
      "loss": 12.1914,
      "step": 12224
    },
    {
      "epoch": 0.6657001082273182,
      "grad_norm": 0.5824889512283538,
      "learning_rate": 0.00015563123879884278,
      "loss": 12.3026,
      "step": 12225
    },
    {
      "epoch": 0.6657545622239012,
      "grad_norm": 0.6220760352001354,
      "learning_rate": 0.00015562391084598818,
      "loss": 12.3276,
      "step": 12226
    },
    {
      "epoch": 0.6658090162204843,
      "grad_norm": 0.5298688423843939,
      "learning_rate": 0.00015561658246059272,
      "loss": 12.2705,
      "step": 12227
    },
    {
      "epoch": 0.6658634702170673,
      "grad_norm": 0.5625501538555766,
      "learning_rate": 0.00015560925364271338,
      "loss": 12.4409,
      "step": 12228
    },
    {
      "epoch": 0.6659179242136503,
      "grad_norm": 0.6663848890519213,
      "learning_rate": 0.00015560192439240716,
      "loss": 12.2582,
      "step": 12229
    },
    {
      "epoch": 0.6659723782102333,
      "grad_norm": 0.5904031545317446,
      "learning_rate": 0.00015559459470973103,
      "loss": 12.1851,
      "step": 12230
    },
    {
      "epoch": 0.6660268322068162,
      "grad_norm": 0.5849650349945908,
      "learning_rate": 0.00015558726459474203,
      "loss": 12.2091,
      "step": 12231
    },
    {
      "epoch": 0.6660812862033992,
      "grad_norm": 0.5324786902705425,
      "learning_rate": 0.00015557993404749715,
      "loss": 12.2008,
      "step": 12232
    },
    {
      "epoch": 0.6661357401999823,
      "grad_norm": 0.5968945370069183,
      "learning_rate": 0.00015557260306805337,
      "loss": 12.4381,
      "step": 12233
    },
    {
      "epoch": 0.6661901941965653,
      "grad_norm": 0.5838181459219236,
      "learning_rate": 0.0001555652716564677,
      "loss": 12.2553,
      "step": 12234
    },
    {
      "epoch": 0.6662446481931483,
      "grad_norm": 0.5742301697411183,
      "learning_rate": 0.00015555793981279718,
      "loss": 12.2566,
      "step": 12235
    },
    {
      "epoch": 0.6662991021897313,
      "grad_norm": 0.5914965836401413,
      "learning_rate": 0.0001555506075370988,
      "loss": 12.3349,
      "step": 12236
    },
    {
      "epoch": 0.6663535561863143,
      "grad_norm": 0.6420592487637078,
      "learning_rate": 0.00015554327482942957,
      "loss": 12.2938,
      "step": 12237
    },
    {
      "epoch": 0.6664080101828974,
      "grad_norm": 0.5607050708073679,
      "learning_rate": 0.00015553594168984654,
      "loss": 12.2412,
      "step": 12238
    },
    {
      "epoch": 0.6664624641794804,
      "grad_norm": 0.5609018375686214,
      "learning_rate": 0.00015552860811840667,
      "loss": 12.2956,
      "step": 12239
    },
    {
      "epoch": 0.6665169181760634,
      "grad_norm": 0.5809352702800051,
      "learning_rate": 0.00015552127411516702,
      "loss": 12.2465,
      "step": 12240
    },
    {
      "epoch": 0.6665713721726464,
      "grad_norm": 0.5842398929846088,
      "learning_rate": 0.0001555139396801847,
      "loss": 12.2001,
      "step": 12241
    },
    {
      "epoch": 0.6666258261692294,
      "grad_norm": 0.5639530150176731,
      "learning_rate": 0.00015550660481351668,
      "loss": 12.3483,
      "step": 12242
    },
    {
      "epoch": 0.6666802801658124,
      "grad_norm": 0.5409246664466759,
      "learning_rate": 0.00015549926951521993,
      "loss": 12.306,
      "step": 12243
    },
    {
      "epoch": 0.6667347341623955,
      "grad_norm": 0.6340991968931381,
      "learning_rate": 0.0001554919337853516,
      "loss": 12.4283,
      "step": 12244
    },
    {
      "epoch": 0.6667891881589785,
      "grad_norm": 0.6306748542962477,
      "learning_rate": 0.00015548459762396863,
      "loss": 12.3561,
      "step": 12245
    },
    {
      "epoch": 0.6668436421555615,
      "grad_norm": 0.6224388651294798,
      "learning_rate": 0.00015547726103112817,
      "loss": 12.407,
      "step": 12246
    },
    {
      "epoch": 0.6668980961521445,
      "grad_norm": 0.524635684961415,
      "learning_rate": 0.00015546992400688724,
      "loss": 12.3067,
      "step": 12247
    },
    {
      "epoch": 0.6669525501487275,
      "grad_norm": 0.589540221009856,
      "learning_rate": 0.00015546258655130285,
      "loss": 12.2554,
      "step": 12248
    },
    {
      "epoch": 0.6670070041453104,
      "grad_norm": 0.5622070308010497,
      "learning_rate": 0.00015545524866443207,
      "loss": 12.1965,
      "step": 12249
    },
    {
      "epoch": 0.6670614581418935,
      "grad_norm": 0.5203199740512348,
      "learning_rate": 0.000155447910346332,
      "loss": 12.2916,
      "step": 12250
    },
    {
      "epoch": 0.6671159121384765,
      "grad_norm": 0.5798442999382576,
      "learning_rate": 0.00015544057159705966,
      "loss": 12.3144,
      "step": 12251
    },
    {
      "epoch": 0.6671703661350595,
      "grad_norm": 0.5390908971582956,
      "learning_rate": 0.00015543323241667216,
      "loss": 12.2795,
      "step": 12252
    },
    {
      "epoch": 0.6672248201316425,
      "grad_norm": 0.5963423017492206,
      "learning_rate": 0.00015542589280522658,
      "loss": 12.2141,
      "step": 12253
    },
    {
      "epoch": 0.6672792741282255,
      "grad_norm": 0.5845700390262353,
      "learning_rate": 0.0001554185527627799,
      "loss": 12.3106,
      "step": 12254
    },
    {
      "epoch": 0.6673337281248085,
      "grad_norm": 0.6140335245300805,
      "learning_rate": 0.0001554112122893893,
      "loss": 12.3444,
      "step": 12255
    },
    {
      "epoch": 0.6673881821213916,
      "grad_norm": 0.6129360063082528,
      "learning_rate": 0.00015540387138511177,
      "loss": 12.3694,
      "step": 12256
    },
    {
      "epoch": 0.6674426361179746,
      "grad_norm": 0.5833285955444938,
      "learning_rate": 0.0001553965300500045,
      "loss": 12.2542,
      "step": 12257
    },
    {
      "epoch": 0.6674970901145576,
      "grad_norm": 0.6353115649335048,
      "learning_rate": 0.0001553891882841245,
      "loss": 12.4273,
      "step": 12258
    },
    {
      "epoch": 0.6675515441111406,
      "grad_norm": 0.5726108332932701,
      "learning_rate": 0.00015538184608752888,
      "loss": 12.1541,
      "step": 12259
    },
    {
      "epoch": 0.6676059981077236,
      "grad_norm": 0.6688286875161022,
      "learning_rate": 0.00015537450346027475,
      "loss": 12.3407,
      "step": 12260
    },
    {
      "epoch": 0.6676604521043066,
      "grad_norm": 0.6635128458029128,
      "learning_rate": 0.0001553671604024192,
      "loss": 12.3662,
      "step": 12261
    },
    {
      "epoch": 0.6677149061008897,
      "grad_norm": 0.616898015401075,
      "learning_rate": 0.0001553598169140193,
      "loss": 12.2972,
      "step": 12262
    },
    {
      "epoch": 0.6677693600974727,
      "grad_norm": 0.6165516831846334,
      "learning_rate": 0.0001553524729951322,
      "loss": 12.1998,
      "step": 12263
    },
    {
      "epoch": 0.6678238140940557,
      "grad_norm": 0.5950312797869388,
      "learning_rate": 0.00015534512864581499,
      "loss": 12.1871,
      "step": 12264
    },
    {
      "epoch": 0.6678782680906387,
      "grad_norm": 0.5886345418949827,
      "learning_rate": 0.00015533778386612478,
      "loss": 12.3738,
      "step": 12265
    },
    {
      "epoch": 0.6679327220872217,
      "grad_norm": 0.5797917759980304,
      "learning_rate": 0.00015533043865611864,
      "loss": 12.259,
      "step": 12266
    },
    {
      "epoch": 0.6679871760838046,
      "grad_norm": 0.6897782017418148,
      "learning_rate": 0.0001553230930158538,
      "loss": 12.3921,
      "step": 12267
    },
    {
      "epoch": 0.6680416300803877,
      "grad_norm": 0.5643376226552791,
      "learning_rate": 0.00015531574694538727,
      "loss": 12.3513,
      "step": 12268
    },
    {
      "epoch": 0.6680960840769707,
      "grad_norm": 0.5793628655866306,
      "learning_rate": 0.00015530840044477619,
      "loss": 12.2912,
      "step": 12269
    },
    {
      "epoch": 0.6681505380735537,
      "grad_norm": 0.524965459789386,
      "learning_rate": 0.00015530105351407777,
      "loss": 12.2108,
      "step": 12270
    },
    {
      "epoch": 0.6682049920701367,
      "grad_norm": 0.588021490527018,
      "learning_rate": 0.00015529370615334904,
      "loss": 12.2733,
      "step": 12271
    },
    {
      "epoch": 0.6682594460667197,
      "grad_norm": 0.5918043083770076,
      "learning_rate": 0.0001552863583626472,
      "loss": 12.3017,
      "step": 12272
    },
    {
      "epoch": 0.6683139000633028,
      "grad_norm": 0.5349154769643079,
      "learning_rate": 0.00015527901014202936,
      "loss": 12.1857,
      "step": 12273
    },
    {
      "epoch": 0.6683683540598858,
      "grad_norm": 0.5826956440854895,
      "learning_rate": 0.00015527166149155268,
      "loss": 12.1369,
      "step": 12274
    },
    {
      "epoch": 0.6684228080564688,
      "grad_norm": 0.6401619839037013,
      "learning_rate": 0.00015526431241127427,
      "loss": 12.3254,
      "step": 12275
    },
    {
      "epoch": 0.6684772620530518,
      "grad_norm": 0.6293240232898998,
      "learning_rate": 0.0001552569629012513,
      "loss": 12.3404,
      "step": 12276
    },
    {
      "epoch": 0.6685317160496348,
      "grad_norm": 0.5878958409192327,
      "learning_rate": 0.00015524961296154092,
      "loss": 12.2896,
      "step": 12277
    },
    {
      "epoch": 0.6685861700462178,
      "grad_norm": 0.5434146624970057,
      "learning_rate": 0.0001552422625922003,
      "loss": 12.2185,
      "step": 12278
    },
    {
      "epoch": 0.6686406240428009,
      "grad_norm": 0.6564827281971444,
      "learning_rate": 0.00015523491179328657,
      "loss": 12.3515,
      "step": 12279
    },
    {
      "epoch": 0.6686950780393839,
      "grad_norm": 0.6563677584404193,
      "learning_rate": 0.0001552275605648569,
      "loss": 12.3971,
      "step": 12280
    },
    {
      "epoch": 0.6687495320359669,
      "grad_norm": 0.639262551518245,
      "learning_rate": 0.00015522020890696846,
      "loss": 12.2402,
      "step": 12281
    },
    {
      "epoch": 0.6688039860325499,
      "grad_norm": 0.7951500546832201,
      "learning_rate": 0.00015521285681967844,
      "loss": 12.2807,
      "step": 12282
    },
    {
      "epoch": 0.6688584400291329,
      "grad_norm": 0.599505447244044,
      "learning_rate": 0.00015520550430304393,
      "loss": 12.3153,
      "step": 12283
    },
    {
      "epoch": 0.6689128940257159,
      "grad_norm": 0.6209193433879819,
      "learning_rate": 0.0001551981513571222,
      "loss": 12.2535,
      "step": 12284
    },
    {
      "epoch": 0.668967348022299,
      "grad_norm": 0.7257855828873232,
      "learning_rate": 0.00015519079798197042,
      "loss": 12.321,
      "step": 12285
    },
    {
      "epoch": 0.669021802018882,
      "grad_norm": 0.5920719331977042,
      "learning_rate": 0.00015518344417764568,
      "loss": 12.2257,
      "step": 12286
    },
    {
      "epoch": 0.6690762560154649,
      "grad_norm": 0.6424559920639874,
      "learning_rate": 0.00015517608994420527,
      "loss": 12.2562,
      "step": 12287
    },
    {
      "epoch": 0.6691307100120479,
      "grad_norm": 0.7367637717941627,
      "learning_rate": 0.0001551687352817063,
      "loss": 12.4215,
      "step": 12288
    },
    {
      "epoch": 0.6691851640086309,
      "grad_norm": 0.6389986361996636,
      "learning_rate": 0.00015516138019020602,
      "loss": 12.3891,
      "step": 12289
    },
    {
      "epoch": 0.6692396180052139,
      "grad_norm": 0.7193207098366835,
      "learning_rate": 0.00015515402466976155,
      "loss": 12.4082,
      "step": 12290
    },
    {
      "epoch": 0.669294072001797,
      "grad_norm": 0.7145208288974416,
      "learning_rate": 0.00015514666872043015,
      "loss": 12.3602,
      "step": 12291
    },
    {
      "epoch": 0.66934852599838,
      "grad_norm": 0.6881150012113569,
      "learning_rate": 0.00015513931234226904,
      "loss": 12.4423,
      "step": 12292
    },
    {
      "epoch": 0.669402979994963,
      "grad_norm": 0.745635936605013,
      "learning_rate": 0.00015513195553533537,
      "loss": 12.3203,
      "step": 12293
    },
    {
      "epoch": 0.669457433991546,
      "grad_norm": 0.6162984481069275,
      "learning_rate": 0.00015512459829968638,
      "loss": 12.2663,
      "step": 12294
    },
    {
      "epoch": 0.669511887988129,
      "grad_norm": 0.6281275905245626,
      "learning_rate": 0.00015511724063537922,
      "loss": 12.3963,
      "step": 12295
    },
    {
      "epoch": 0.669566341984712,
      "grad_norm": 0.7733149049652096,
      "learning_rate": 0.00015510988254247121,
      "loss": 12.3302,
      "step": 12296
    },
    {
      "epoch": 0.6696207959812951,
      "grad_norm": 0.5780855891143521,
      "learning_rate": 0.0001551025240210195,
      "loss": 12.282,
      "step": 12297
    },
    {
      "epoch": 0.6696752499778781,
      "grad_norm": 0.6068608143213838,
      "learning_rate": 0.00015509516507108132,
      "loss": 12.212,
      "step": 12298
    },
    {
      "epoch": 0.6697297039744611,
      "grad_norm": 0.7063211921350746,
      "learning_rate": 0.00015508780569271387,
      "loss": 12.3415,
      "step": 12299
    },
    {
      "epoch": 0.6697841579710441,
      "grad_norm": 0.5802194256618435,
      "learning_rate": 0.00015508044588597442,
      "loss": 12.3357,
      "step": 12300
    },
    {
      "epoch": 0.6698386119676271,
      "grad_norm": 0.6195286163771041,
      "learning_rate": 0.0001550730856509202,
      "loss": 12.3546,
      "step": 12301
    },
    {
      "epoch": 0.6698930659642102,
      "grad_norm": 0.826825832697903,
      "learning_rate": 0.00015506572498760843,
      "loss": 12.3014,
      "step": 12302
    },
    {
      "epoch": 0.6699475199607932,
      "grad_norm": 0.6368084289094414,
      "learning_rate": 0.00015505836389609633,
      "loss": 12.3103,
      "step": 12303
    },
    {
      "epoch": 0.6700019739573762,
      "grad_norm": 0.6280506242277251,
      "learning_rate": 0.00015505100237644118,
      "loss": 12.1298,
      "step": 12304
    },
    {
      "epoch": 0.6700564279539591,
      "grad_norm": 0.6526406731752941,
      "learning_rate": 0.0001550436404287002,
      "loss": 12.4316,
      "step": 12305
    },
    {
      "epoch": 0.6701108819505421,
      "grad_norm": 0.6004196297894798,
      "learning_rate": 0.0001550362780529306,
      "loss": 12.3352,
      "step": 12306
    },
    {
      "epoch": 0.6701653359471251,
      "grad_norm": 0.6262109885265384,
      "learning_rate": 0.00015502891524918973,
      "loss": 12.215,
      "step": 12307
    },
    {
      "epoch": 0.6702197899437082,
      "grad_norm": 0.6828665022831916,
      "learning_rate": 0.00015502155201753477,
      "loss": 12.3009,
      "step": 12308
    },
    {
      "epoch": 0.6702742439402912,
      "grad_norm": 0.5860601334328157,
      "learning_rate": 0.000155014188358023,
      "loss": 12.3318,
      "step": 12309
    },
    {
      "epoch": 0.6703286979368742,
      "grad_norm": 0.6511973526521048,
      "learning_rate": 0.00015500682427071162,
      "loss": 12.4775,
      "step": 12310
    },
    {
      "epoch": 0.6703831519334572,
      "grad_norm": 0.5563454571783759,
      "learning_rate": 0.000154999459755658,
      "loss": 12.2955,
      "step": 12311
    },
    {
      "epoch": 0.6704376059300402,
      "grad_norm": 0.5392087483489257,
      "learning_rate": 0.00015499209481291935,
      "loss": 12.2512,
      "step": 12312
    },
    {
      "epoch": 0.6704920599266232,
      "grad_norm": 0.5933980380777396,
      "learning_rate": 0.00015498472944255296,
      "loss": 12.2228,
      "step": 12313
    },
    {
      "epoch": 0.6705465139232063,
      "grad_norm": 0.578197621670821,
      "learning_rate": 0.0001549773636446161,
      "loss": 12.4391,
      "step": 12314
    },
    {
      "epoch": 0.6706009679197893,
      "grad_norm": 0.5833579940210534,
      "learning_rate": 0.000154969997419166,
      "loss": 12.443,
      "step": 12315
    },
    {
      "epoch": 0.6706554219163723,
      "grad_norm": 0.5603252629323376,
      "learning_rate": 0.00015496263076626,
      "loss": 12.1525,
      "step": 12316
    },
    {
      "epoch": 0.6707098759129553,
      "grad_norm": 0.5888360644780639,
      "learning_rate": 0.0001549552636859554,
      "loss": 12.2737,
      "step": 12317
    },
    {
      "epoch": 0.6707643299095383,
      "grad_norm": 0.6644092677788832,
      "learning_rate": 0.00015494789617830938,
      "loss": 12.1984,
      "step": 12318
    },
    {
      "epoch": 0.6708187839061213,
      "grad_norm": 0.5836970411791585,
      "learning_rate": 0.00015494052824337938,
      "loss": 12.3877,
      "step": 12319
    },
    {
      "epoch": 0.6708732379027044,
      "grad_norm": 0.6424242091880643,
      "learning_rate": 0.00015493315988122258,
      "loss": 12.3825,
      "step": 12320
    },
    {
      "epoch": 0.6709276918992874,
      "grad_norm": 0.6002102631739418,
      "learning_rate": 0.00015492579109189632,
      "loss": 12.1623,
      "step": 12321
    },
    {
      "epoch": 0.6709821458958704,
      "grad_norm": 0.6137470092392461,
      "learning_rate": 0.0001549184218754579,
      "loss": 12.3687,
      "step": 12322
    },
    {
      "epoch": 0.6710365998924533,
      "grad_norm": 0.5601924573585031,
      "learning_rate": 0.00015491105223196464,
      "loss": 12.2654,
      "step": 12323
    },
    {
      "epoch": 0.6710910538890363,
      "grad_norm": 0.6533100680065654,
      "learning_rate": 0.00015490368216147378,
      "loss": 12.3328,
      "step": 12324
    },
    {
      "epoch": 0.6711455078856193,
      "grad_norm": 0.5571867758560064,
      "learning_rate": 0.00015489631166404273,
      "loss": 12.2228,
      "step": 12325
    },
    {
      "epoch": 0.6711999618822024,
      "grad_norm": 0.6720500084489055,
      "learning_rate": 0.0001548889407397287,
      "loss": 12.3046,
      "step": 12326
    },
    {
      "epoch": 0.6712544158787854,
      "grad_norm": 0.5338742799875239,
      "learning_rate": 0.00015488156938858912,
      "loss": 12.2962,
      "step": 12327
    },
    {
      "epoch": 0.6713088698753684,
      "grad_norm": 0.6713263288393614,
      "learning_rate": 0.00015487419761068126,
      "loss": 12.3337,
      "step": 12328
    },
    {
      "epoch": 0.6713633238719514,
      "grad_norm": 0.6455710139817895,
      "learning_rate": 0.00015486682540606238,
      "loss": 12.364,
      "step": 12329
    },
    {
      "epoch": 0.6714177778685344,
      "grad_norm": 0.535391671053815,
      "learning_rate": 0.00015485945277478989,
      "loss": 12.3494,
      "step": 12330
    },
    {
      "epoch": 0.6714722318651174,
      "grad_norm": 0.6048302286197477,
      "learning_rate": 0.0001548520797169211,
      "loss": 12.2211,
      "step": 12331
    },
    {
      "epoch": 0.6715266858617005,
      "grad_norm": 0.5674125889055247,
      "learning_rate": 0.00015484470623251332,
      "loss": 12.1535,
      "step": 12332
    },
    {
      "epoch": 0.6715811398582835,
      "grad_norm": 0.5262918147403375,
      "learning_rate": 0.00015483733232162395,
      "loss": 12.1773,
      "step": 12333
    },
    {
      "epoch": 0.6716355938548665,
      "grad_norm": 0.5512608317557834,
      "learning_rate": 0.00015482995798431025,
      "loss": 12.2709,
      "step": 12334
    },
    {
      "epoch": 0.6716900478514495,
      "grad_norm": 0.5518915582374262,
      "learning_rate": 0.0001548225832206296,
      "loss": 12.2212,
      "step": 12335
    },
    {
      "epoch": 0.6717445018480325,
      "grad_norm": 0.563258202916151,
      "learning_rate": 0.00015481520803063936,
      "loss": 12.3136,
      "step": 12336
    },
    {
      "epoch": 0.6717989558446156,
      "grad_norm": 0.5866718369655832,
      "learning_rate": 0.0001548078324143969,
      "loss": 12.3299,
      "step": 12337
    },
    {
      "epoch": 0.6718534098411986,
      "grad_norm": 0.6200280583867798,
      "learning_rate": 0.0001548004563719595,
      "loss": 12.2475,
      "step": 12338
    },
    {
      "epoch": 0.6719078638377816,
      "grad_norm": 0.6078855752892216,
      "learning_rate": 0.00015479307990338457,
      "loss": 12.2136,
      "step": 12339
    },
    {
      "epoch": 0.6719623178343646,
      "grad_norm": 0.6314598056079747,
      "learning_rate": 0.00015478570300872947,
      "loss": 12.2619,
      "step": 12340
    },
    {
      "epoch": 0.6720167718309475,
      "grad_norm": 0.5881620181870542,
      "learning_rate": 0.00015477832568805156,
      "loss": 12.3938,
      "step": 12341
    },
    {
      "epoch": 0.6720712258275305,
      "grad_norm": 0.6373593538818688,
      "learning_rate": 0.00015477094794140821,
      "loss": 12.2626,
      "step": 12342
    },
    {
      "epoch": 0.6721256798241136,
      "grad_norm": 0.5922390028924963,
      "learning_rate": 0.00015476356976885675,
      "loss": 12.3762,
      "step": 12343
    },
    {
      "epoch": 0.6721801338206966,
      "grad_norm": 0.6446399456565183,
      "learning_rate": 0.0001547561911704546,
      "loss": 12.358,
      "step": 12344
    },
    {
      "epoch": 0.6722345878172796,
      "grad_norm": 0.627724909475266,
      "learning_rate": 0.00015474881214625917,
      "loss": 12.3067,
      "step": 12345
    },
    {
      "epoch": 0.6722890418138626,
      "grad_norm": 0.6242450041274252,
      "learning_rate": 0.00015474143269632773,
      "loss": 12.3418,
      "step": 12346
    },
    {
      "epoch": 0.6723434958104456,
      "grad_norm": 0.5695089624131952,
      "learning_rate": 0.00015473405282071776,
      "loss": 12.315,
      "step": 12347
    },
    {
      "epoch": 0.6723979498070286,
      "grad_norm": 0.6649335811814701,
      "learning_rate": 0.00015472667251948663,
      "loss": 12.4619,
      "step": 12348
    },
    {
      "epoch": 0.6724524038036117,
      "grad_norm": 0.6151808744087764,
      "learning_rate": 0.0001547192917926917,
      "loss": 12.2973,
      "step": 12349
    },
    {
      "epoch": 0.6725068578001947,
      "grad_norm": 0.5903380108730035,
      "learning_rate": 0.00015471191064039038,
      "loss": 12.2254,
      "step": 12350
    },
    {
      "epoch": 0.6725613117967777,
      "grad_norm": 0.5857314014547613,
      "learning_rate": 0.00015470452906264005,
      "loss": 12.3142,
      "step": 12351
    },
    {
      "epoch": 0.6726157657933607,
      "grad_norm": 0.5376113152177633,
      "learning_rate": 0.00015469714705949815,
      "loss": 12.2964,
      "step": 12352
    },
    {
      "epoch": 0.6726702197899437,
      "grad_norm": 0.5985240941302591,
      "learning_rate": 0.00015468976463102208,
      "loss": 12.323,
      "step": 12353
    },
    {
      "epoch": 0.6727246737865267,
      "grad_norm": 0.5841038750145867,
      "learning_rate": 0.0001546823817772692,
      "loss": 12.221,
      "step": 12354
    },
    {
      "epoch": 0.6727791277831098,
      "grad_norm": 0.5958397562184997,
      "learning_rate": 0.00015467499849829697,
      "loss": 12.3587,
      "step": 12355
    },
    {
      "epoch": 0.6728335817796928,
      "grad_norm": 0.6544013573600018,
      "learning_rate": 0.00015466761479416275,
      "loss": 12.2337,
      "step": 12356
    },
    {
      "epoch": 0.6728880357762758,
      "grad_norm": 0.7357460786214085,
      "learning_rate": 0.00015466023066492402,
      "loss": 12.3293,
      "step": 12357
    },
    {
      "epoch": 0.6729424897728588,
      "grad_norm": 0.53074594411693,
      "learning_rate": 0.00015465284611063815,
      "loss": 12.2764,
      "step": 12358
    },
    {
      "epoch": 0.6729969437694417,
      "grad_norm": 0.6640114712340902,
      "learning_rate": 0.0001546454611313626,
      "loss": 12.2647,
      "step": 12359
    },
    {
      "epoch": 0.6730513977660247,
      "grad_norm": 0.5867100703731516,
      "learning_rate": 0.0001546380757271548,
      "loss": 12.3141,
      "step": 12360
    },
    {
      "epoch": 0.6731058517626078,
      "grad_norm": 0.6114819001091513,
      "learning_rate": 0.0001546306898980721,
      "loss": 12.4673,
      "step": 12361
    },
    {
      "epoch": 0.6731603057591908,
      "grad_norm": 0.5919784984722599,
      "learning_rate": 0.00015462330364417203,
      "loss": 12.3433,
      "step": 12362
    },
    {
      "epoch": 0.6732147597557738,
      "grad_norm": 0.6407954340101385,
      "learning_rate": 0.00015461591696551202,
      "loss": 12.1382,
      "step": 12363
    },
    {
      "epoch": 0.6732692137523568,
      "grad_norm": 0.569920810338462,
      "learning_rate": 0.00015460852986214942,
      "loss": 12.3783,
      "step": 12364
    },
    {
      "epoch": 0.6733236677489398,
      "grad_norm": 0.5870295077894265,
      "learning_rate": 0.00015460114233414177,
      "loss": 12.3657,
      "step": 12365
    },
    {
      "epoch": 0.6733781217455228,
      "grad_norm": 0.6636458321676164,
      "learning_rate": 0.00015459375438154645,
      "loss": 12.3468,
      "step": 12366
    },
    {
      "epoch": 0.6734325757421059,
      "grad_norm": 0.6058518127129257,
      "learning_rate": 0.00015458636600442098,
      "loss": 12.3653,
      "step": 12367
    },
    {
      "epoch": 0.6734870297386889,
      "grad_norm": 0.5852369575195829,
      "learning_rate": 0.00015457897720282278,
      "loss": 12.31,
      "step": 12368
    },
    {
      "epoch": 0.6735414837352719,
      "grad_norm": 0.6882324799735526,
      "learning_rate": 0.00015457158797680927,
      "loss": 12.2676,
      "step": 12369
    },
    {
      "epoch": 0.6735959377318549,
      "grad_norm": 0.6390571986339338,
      "learning_rate": 0.00015456419832643795,
      "loss": 12.289,
      "step": 12370
    },
    {
      "epoch": 0.6736503917284379,
      "grad_norm": 0.5658026855278877,
      "learning_rate": 0.00015455680825176624,
      "loss": 12.2175,
      "step": 12371
    },
    {
      "epoch": 0.673704845725021,
      "grad_norm": 0.6516117326196103,
      "learning_rate": 0.00015454941775285168,
      "loss": 12.3273,
      "step": 12372
    },
    {
      "epoch": 0.673759299721604,
      "grad_norm": 0.6616815047039033,
      "learning_rate": 0.0001545420268297517,
      "loss": 12.3157,
      "step": 12373
    },
    {
      "epoch": 0.673813753718187,
      "grad_norm": 0.5321409149239987,
      "learning_rate": 0.00015453463548252376,
      "loss": 12.3777,
      "step": 12374
    },
    {
      "epoch": 0.67386820771477,
      "grad_norm": 0.5706921321363255,
      "learning_rate": 0.00015452724371122536,
      "loss": 12.3715,
      "step": 12375
    },
    {
      "epoch": 0.673922661711353,
      "grad_norm": 0.6419548750079106,
      "learning_rate": 0.00015451985151591395,
      "loss": 12.4302,
      "step": 12376
    },
    {
      "epoch": 0.673977115707936,
      "grad_norm": 0.6001667031937706,
      "learning_rate": 0.00015451245889664701,
      "loss": 12.3792,
      "step": 12377
    },
    {
      "epoch": 0.674031569704519,
      "grad_norm": 0.6269987634290967,
      "learning_rate": 0.00015450506585348213,
      "loss": 12.2882,
      "step": 12378
    },
    {
      "epoch": 0.674086023701102,
      "grad_norm": 0.6149717814168466,
      "learning_rate": 0.00015449767238647665,
      "loss": 12.458,
      "step": 12379
    },
    {
      "epoch": 0.674140477697685,
      "grad_norm": 0.6305814591670665,
      "learning_rate": 0.00015449027849568815,
      "loss": 12.1985,
      "step": 12380
    },
    {
      "epoch": 0.674194931694268,
      "grad_norm": 0.6317839199734646,
      "learning_rate": 0.0001544828841811741,
      "loss": 12.3741,
      "step": 12381
    },
    {
      "epoch": 0.674249385690851,
      "grad_norm": 0.7296820767208071,
      "learning_rate": 0.00015447548944299202,
      "loss": 12.4499,
      "step": 12382
    },
    {
      "epoch": 0.674303839687434,
      "grad_norm": 0.760502440648826,
      "learning_rate": 0.00015446809428119938,
      "loss": 12.4208,
      "step": 12383
    },
    {
      "epoch": 0.6743582936840171,
      "grad_norm": 0.6032206376300737,
      "learning_rate": 0.0001544606986958537,
      "loss": 12.3193,
      "step": 12384
    },
    {
      "epoch": 0.6744127476806001,
      "grad_norm": 0.651116057178426,
      "learning_rate": 0.00015445330268701253,
      "loss": 12.3101,
      "step": 12385
    },
    {
      "epoch": 0.6744672016771831,
      "grad_norm": 0.6649409317668227,
      "learning_rate": 0.00015444590625473333,
      "loss": 12.3424,
      "step": 12386
    },
    {
      "epoch": 0.6745216556737661,
      "grad_norm": 0.6269635487637647,
      "learning_rate": 0.0001544385093990736,
      "loss": 12.3125,
      "step": 12387
    },
    {
      "epoch": 0.6745761096703491,
      "grad_norm": 0.6004238287073058,
      "learning_rate": 0.00015443111212009095,
      "loss": 12.2203,
      "step": 12388
    },
    {
      "epoch": 0.6746305636669321,
      "grad_norm": 0.636585741622858,
      "learning_rate": 0.00015442371441784278,
      "loss": 12.3283,
      "step": 12389
    },
    {
      "epoch": 0.6746850176635152,
      "grad_norm": 0.6314743975811515,
      "learning_rate": 0.0001544163162923867,
      "loss": 12.3442,
      "step": 12390
    },
    {
      "epoch": 0.6747394716600982,
      "grad_norm": 0.5799282392051064,
      "learning_rate": 0.00015440891774378025,
      "loss": 12.4539,
      "step": 12391
    },
    {
      "epoch": 0.6747939256566812,
      "grad_norm": 0.7255747829870278,
      "learning_rate": 0.0001544015187720809,
      "loss": 12.2881,
      "step": 12392
    },
    {
      "epoch": 0.6748483796532642,
      "grad_norm": 0.5439656129914802,
      "learning_rate": 0.00015439411937734625,
      "loss": 12.124,
      "step": 12393
    },
    {
      "epoch": 0.6749028336498472,
      "grad_norm": 0.6537594353864004,
      "learning_rate": 0.0001543867195596338,
      "loss": 12.1749,
      "step": 12394
    },
    {
      "epoch": 0.6749572876464301,
      "grad_norm": 0.6631492345005819,
      "learning_rate": 0.00015437931931900107,
      "loss": 12.2417,
      "step": 12395
    },
    {
      "epoch": 0.6750117416430133,
      "grad_norm": 0.561066732269436,
      "learning_rate": 0.00015437191865550563,
      "loss": 12.2644,
      "step": 12396
    },
    {
      "epoch": 0.6750661956395962,
      "grad_norm": 0.5838921948641176,
      "learning_rate": 0.00015436451756920504,
      "loss": 12.29,
      "step": 12397
    },
    {
      "epoch": 0.6751206496361792,
      "grad_norm": 0.6094728089312846,
      "learning_rate": 0.0001543571160601569,
      "loss": 12.3605,
      "step": 12398
    },
    {
      "epoch": 0.6751751036327622,
      "grad_norm": 0.5575270120681592,
      "learning_rate": 0.00015434971412841867,
      "loss": 12.2974,
      "step": 12399
    },
    {
      "epoch": 0.6752295576293452,
      "grad_norm": 0.6325187505651708,
      "learning_rate": 0.0001543423117740479,
      "loss": 12.2801,
      "step": 12400
    },
    {
      "epoch": 0.6752840116259282,
      "grad_norm": 0.6472180003870803,
      "learning_rate": 0.00015433490899710223,
      "loss": 12.3826,
      "step": 12401
    },
    {
      "epoch": 0.6753384656225113,
      "grad_norm": 0.5868907719763805,
      "learning_rate": 0.0001543275057976392,
      "loss": 12.2332,
      "step": 12402
    },
    {
      "epoch": 0.6753929196190943,
      "grad_norm": 0.6427595911096574,
      "learning_rate": 0.0001543201021757164,
      "loss": 12.4348,
      "step": 12403
    },
    {
      "epoch": 0.6754473736156773,
      "grad_norm": 0.6668338028229127,
      "learning_rate": 0.00015431269813139138,
      "loss": 12.3388,
      "step": 12404
    },
    {
      "epoch": 0.6755018276122603,
      "grad_norm": 0.5775714167901217,
      "learning_rate": 0.00015430529366472167,
      "loss": 12.3564,
      "step": 12405
    },
    {
      "epoch": 0.6755562816088433,
      "grad_norm": 0.5332194175715848,
      "learning_rate": 0.0001542978887757649,
      "loss": 12.267,
      "step": 12406
    },
    {
      "epoch": 0.6756107356054264,
      "grad_norm": 0.5890769707710252,
      "learning_rate": 0.00015429048346457863,
      "loss": 12.3213,
      "step": 12407
    },
    {
      "epoch": 0.6756651896020094,
      "grad_norm": 0.6323650116011595,
      "learning_rate": 0.0001542830777312205,
      "loss": 12.289,
      "step": 12408
    },
    {
      "epoch": 0.6757196435985924,
      "grad_norm": 0.6041259921360344,
      "learning_rate": 0.00015427567157574802,
      "loss": 12.172,
      "step": 12409
    },
    {
      "epoch": 0.6757740975951754,
      "grad_norm": 0.6035886166991528,
      "learning_rate": 0.00015426826499821884,
      "loss": 12.1674,
      "step": 12410
    },
    {
      "epoch": 0.6758285515917584,
      "grad_norm": 0.5608004708899101,
      "learning_rate": 0.00015426085799869052,
      "loss": 12.308,
      "step": 12411
    },
    {
      "epoch": 0.6758830055883414,
      "grad_norm": 0.7775669709518196,
      "learning_rate": 0.00015425345057722064,
      "loss": 12.446,
      "step": 12412
    },
    {
      "epoch": 0.6759374595849245,
      "grad_norm": 0.598766215338868,
      "learning_rate": 0.0001542460427338669,
      "loss": 12.3271,
      "step": 12413
    },
    {
      "epoch": 0.6759919135815075,
      "grad_norm": 0.5941778258183403,
      "learning_rate": 0.00015423863446868677,
      "loss": 12.2538,
      "step": 12414
    },
    {
      "epoch": 0.6760463675780904,
      "grad_norm": 0.5730630987233022,
      "learning_rate": 0.00015423122578173793,
      "loss": 12.2467,
      "step": 12415
    },
    {
      "epoch": 0.6761008215746734,
      "grad_norm": 0.6393883467564998,
      "learning_rate": 0.000154223816673078,
      "loss": 12.3639,
      "step": 12416
    },
    {
      "epoch": 0.6761552755712564,
      "grad_norm": 0.5958537515290336,
      "learning_rate": 0.00015421640714276459,
      "loss": 12.2933,
      "step": 12417
    },
    {
      "epoch": 0.6762097295678394,
      "grad_norm": 0.5473205329362438,
      "learning_rate": 0.00015420899719085528,
      "loss": 12.1924,
      "step": 12418
    },
    {
      "epoch": 0.6762641835644225,
      "grad_norm": 0.6135898548407824,
      "learning_rate": 0.00015420158681740773,
      "loss": 12.3284,
      "step": 12419
    },
    {
      "epoch": 0.6763186375610055,
      "grad_norm": 0.6937169434880651,
      "learning_rate": 0.00015419417602247955,
      "loss": 12.3112,
      "step": 12420
    },
    {
      "epoch": 0.6763730915575885,
      "grad_norm": 0.5813587951963771,
      "learning_rate": 0.00015418676480612837,
      "loss": 12.2987,
      "step": 12421
    },
    {
      "epoch": 0.6764275455541715,
      "grad_norm": 0.5634384409012385,
      "learning_rate": 0.00015417935316841178,
      "loss": 12.2539,
      "step": 12422
    },
    {
      "epoch": 0.6764819995507545,
      "grad_norm": 0.6025612760466049,
      "learning_rate": 0.0001541719411093875,
      "loss": 12.1718,
      "step": 12423
    },
    {
      "epoch": 0.6765364535473375,
      "grad_norm": 0.5833255778249352,
      "learning_rate": 0.00015416452862911312,
      "loss": 12.2218,
      "step": 12424
    },
    {
      "epoch": 0.6765909075439206,
      "grad_norm": 0.5830101697665969,
      "learning_rate": 0.00015415711572764628,
      "loss": 12.2757,
      "step": 12425
    },
    {
      "epoch": 0.6766453615405036,
      "grad_norm": 0.6382187491294317,
      "learning_rate": 0.00015414970240504463,
      "loss": 12.1939,
      "step": 12426
    },
    {
      "epoch": 0.6766998155370866,
      "grad_norm": 0.6245430074794555,
      "learning_rate": 0.0001541422886613658,
      "loss": 12.3249,
      "step": 12427
    },
    {
      "epoch": 0.6767542695336696,
      "grad_norm": 0.5902115680820347,
      "learning_rate": 0.00015413487449666746,
      "loss": 12.3674,
      "step": 12428
    },
    {
      "epoch": 0.6768087235302526,
      "grad_norm": 0.5595368427708826,
      "learning_rate": 0.00015412745991100722,
      "loss": 12.2548,
      "step": 12429
    },
    {
      "epoch": 0.6768631775268356,
      "grad_norm": 0.7746748718973765,
      "learning_rate": 0.00015412004490444282,
      "loss": 12.4776,
      "step": 12430
    },
    {
      "epoch": 0.6769176315234187,
      "grad_norm": 0.5850132877825888,
      "learning_rate": 0.00015411262947703186,
      "loss": 12.2565,
      "step": 12431
    },
    {
      "epoch": 0.6769720855200017,
      "grad_norm": 0.5774462724432337,
      "learning_rate": 0.00015410521362883205,
      "loss": 12.3022,
      "step": 12432
    },
    {
      "epoch": 0.6770265395165846,
      "grad_norm": 0.5587997789708538,
      "learning_rate": 0.00015409779735990096,
      "loss": 12.3016,
      "step": 12433
    },
    {
      "epoch": 0.6770809935131676,
      "grad_norm": 0.7234506787324182,
      "learning_rate": 0.0001540903806702964,
      "loss": 12.2563,
      "step": 12434
    },
    {
      "epoch": 0.6771354475097506,
      "grad_norm": 0.6431977615477494,
      "learning_rate": 0.00015408296356007593,
      "loss": 12.3079,
      "step": 12435
    },
    {
      "epoch": 0.6771899015063337,
      "grad_norm": 0.6397691285259051,
      "learning_rate": 0.00015407554602929726,
      "loss": 12.3851,
      "step": 12436
    },
    {
      "epoch": 0.6772443555029167,
      "grad_norm": 0.5613075305208957,
      "learning_rate": 0.00015406812807801807,
      "loss": 12.274,
      "step": 12437
    },
    {
      "epoch": 0.6772988094994997,
      "grad_norm": 0.5978847918724727,
      "learning_rate": 0.0001540607097062961,
      "loss": 12.2493,
      "step": 12438
    },
    {
      "epoch": 0.6773532634960827,
      "grad_norm": 0.6838034584535284,
      "learning_rate": 0.00015405329091418892,
      "loss": 12.4762,
      "step": 12439
    },
    {
      "epoch": 0.6774077174926657,
      "grad_norm": 0.6147624053315099,
      "learning_rate": 0.00015404587170175432,
      "loss": 12.2885,
      "step": 12440
    },
    {
      "epoch": 0.6774621714892487,
      "grad_norm": 0.566160152720444,
      "learning_rate": 0.00015403845206904995,
      "loss": 12.2939,
      "step": 12441
    },
    {
      "epoch": 0.6775166254858318,
      "grad_norm": 0.6071692350273591,
      "learning_rate": 0.0001540310320161335,
      "loss": 12.2652,
      "step": 12442
    },
    {
      "epoch": 0.6775710794824148,
      "grad_norm": 0.6200172147901956,
      "learning_rate": 0.0001540236115430627,
      "loss": 12.3072,
      "step": 12443
    },
    {
      "epoch": 0.6776255334789978,
      "grad_norm": 0.5818622382070244,
      "learning_rate": 0.00015401619064989524,
      "loss": 12.2654,
      "step": 12444
    },
    {
      "epoch": 0.6776799874755808,
      "grad_norm": 0.669309561790791,
      "learning_rate": 0.00015400876933668883,
      "loss": 12.3663,
      "step": 12445
    },
    {
      "epoch": 0.6777344414721638,
      "grad_norm": 0.5745993725924108,
      "learning_rate": 0.00015400134760350115,
      "loss": 12.2131,
      "step": 12446
    },
    {
      "epoch": 0.6777888954687468,
      "grad_norm": 0.6416078009550086,
      "learning_rate": 0.00015399392545038993,
      "loss": 12.2781,
      "step": 12447
    },
    {
      "epoch": 0.6778433494653299,
      "grad_norm": 0.6656096813777037,
      "learning_rate": 0.00015398650287741294,
      "loss": 12.3451,
      "step": 12448
    },
    {
      "epoch": 0.6778978034619129,
      "grad_norm": 0.5414612288002892,
      "learning_rate": 0.0001539790798846278,
      "loss": 12.2011,
      "step": 12449
    },
    {
      "epoch": 0.6779522574584959,
      "grad_norm": 0.6820187506744111,
      "learning_rate": 0.00015397165647209232,
      "loss": 12.1749,
      "step": 12450
    },
    {
      "epoch": 0.6780067114550788,
      "grad_norm": 0.568131185561711,
      "learning_rate": 0.00015396423263986414,
      "loss": 12.2077,
      "step": 12451
    },
    {
      "epoch": 0.6780611654516618,
      "grad_norm": 0.6201515312515274,
      "learning_rate": 0.00015395680838800106,
      "loss": 12.3235,
      "step": 12452
    },
    {
      "epoch": 0.6781156194482448,
      "grad_norm": 0.576863601249074,
      "learning_rate": 0.00015394938371656082,
      "loss": 12.3372,
      "step": 12453
    },
    {
      "epoch": 0.6781700734448279,
      "grad_norm": 0.6012291927573601,
      "learning_rate": 0.0001539419586256011,
      "loss": 12.4355,
      "step": 12454
    },
    {
      "epoch": 0.6782245274414109,
      "grad_norm": 0.6164387649292629,
      "learning_rate": 0.00015393453311517968,
      "loss": 12.346,
      "step": 12455
    },
    {
      "epoch": 0.6782789814379939,
      "grad_norm": 0.9865572568672095,
      "learning_rate": 0.00015392710718535425,
      "loss": 12.302,
      "step": 12456
    },
    {
      "epoch": 0.6783334354345769,
      "grad_norm": 0.5986802333369561,
      "learning_rate": 0.0001539196808361826,
      "loss": 12.3104,
      "step": 12457
    },
    {
      "epoch": 0.6783878894311599,
      "grad_norm": 0.5640829111191706,
      "learning_rate": 0.0001539122540677225,
      "loss": 12.2395,
      "step": 12458
    },
    {
      "epoch": 0.6784423434277429,
      "grad_norm": 0.556879327150156,
      "learning_rate": 0.00015390482688003166,
      "loss": 12.3582,
      "step": 12459
    },
    {
      "epoch": 0.678496797424326,
      "grad_norm": 0.5968851843906607,
      "learning_rate": 0.0001538973992731678,
      "loss": 12.2607,
      "step": 12460
    },
    {
      "epoch": 0.678551251420909,
      "grad_norm": 0.5580397132881889,
      "learning_rate": 0.00015388997124718878,
      "loss": 12.2164,
      "step": 12461
    },
    {
      "epoch": 0.678605705417492,
      "grad_norm": 0.6033343613214533,
      "learning_rate": 0.00015388254280215228,
      "loss": 12.4007,
      "step": 12462
    },
    {
      "epoch": 0.678660159414075,
      "grad_norm": 0.6286973816578141,
      "learning_rate": 0.0001538751139381161,
      "loss": 12.2404,
      "step": 12463
    },
    {
      "epoch": 0.678714613410658,
      "grad_norm": 0.659599719190087,
      "learning_rate": 0.000153867684655138,
      "loss": 12.5315,
      "step": 12464
    },
    {
      "epoch": 0.678769067407241,
      "grad_norm": 0.5985044281923901,
      "learning_rate": 0.00015386025495327577,
      "loss": 12.1016,
      "step": 12465
    },
    {
      "epoch": 0.6788235214038241,
      "grad_norm": 0.5894161418965819,
      "learning_rate": 0.00015385282483258714,
      "loss": 12.2652,
      "step": 12466
    },
    {
      "epoch": 0.6788779754004071,
      "grad_norm": 0.5674643925713331,
      "learning_rate": 0.00015384539429312989,
      "loss": 12.2766,
      "step": 12467
    },
    {
      "epoch": 0.6789324293969901,
      "grad_norm": 0.6379808068456474,
      "learning_rate": 0.00015383796333496186,
      "loss": 12.2333,
      "step": 12468
    },
    {
      "epoch": 0.678986883393573,
      "grad_norm": 0.5935111162733151,
      "learning_rate": 0.0001538305319581408,
      "loss": 12.3503,
      "step": 12469
    },
    {
      "epoch": 0.679041337390156,
      "grad_norm": 0.6086340132949898,
      "learning_rate": 0.00015382310016272448,
      "loss": 12.2272,
      "step": 12470
    },
    {
      "epoch": 0.6790957913867391,
      "grad_norm": 0.6248968129012789,
      "learning_rate": 0.0001538156679487707,
      "loss": 12.3787,
      "step": 12471
    },
    {
      "epoch": 0.6791502453833221,
      "grad_norm": 0.6803043130004992,
      "learning_rate": 0.00015380823531633729,
      "loss": 12.2579,
      "step": 12472
    },
    {
      "epoch": 0.6792046993799051,
      "grad_norm": 0.656370341075035,
      "learning_rate": 0.000153800802265482,
      "loss": 12.3432,
      "step": 12473
    },
    {
      "epoch": 0.6792591533764881,
      "grad_norm": 0.5485034232369759,
      "learning_rate": 0.00015379336879626262,
      "loss": 12.253,
      "step": 12474
    },
    {
      "epoch": 0.6793136073730711,
      "grad_norm": 0.5865073594312507,
      "learning_rate": 0.00015378593490873702,
      "loss": 12.3455,
      "step": 12475
    },
    {
      "epoch": 0.6793680613696541,
      "grad_norm": 0.6106625314250173,
      "learning_rate": 0.00015377850060296298,
      "loss": 12.25,
      "step": 12476
    },
    {
      "epoch": 0.6794225153662372,
      "grad_norm": 0.6703990987672004,
      "learning_rate": 0.00015377106587899828,
      "loss": 12.3259,
      "step": 12477
    },
    {
      "epoch": 0.6794769693628202,
      "grad_norm": 0.6749449370571579,
      "learning_rate": 0.00015376363073690076,
      "loss": 12.4298,
      "step": 12478
    },
    {
      "epoch": 0.6795314233594032,
      "grad_norm": 0.5267898950623957,
      "learning_rate": 0.00015375619517672822,
      "loss": 12.2494,
      "step": 12479
    },
    {
      "epoch": 0.6795858773559862,
      "grad_norm": 0.5813391501366415,
      "learning_rate": 0.00015374875919853853,
      "loss": 12.2722,
      "step": 12480
    },
    {
      "epoch": 0.6796403313525692,
      "grad_norm": 0.5691942155847314,
      "learning_rate": 0.00015374132280238944,
      "loss": 12.3343,
      "step": 12481
    },
    {
      "epoch": 0.6796947853491522,
      "grad_norm": 0.5597577928871845,
      "learning_rate": 0.00015373388598833882,
      "loss": 12.2779,
      "step": 12482
    },
    {
      "epoch": 0.6797492393457353,
      "grad_norm": 0.5600179115697138,
      "learning_rate": 0.0001537264487564445,
      "loss": 12.3019,
      "step": 12483
    },
    {
      "epoch": 0.6798036933423183,
      "grad_norm": 0.6609106539019437,
      "learning_rate": 0.00015371901110676427,
      "loss": 12.3928,
      "step": 12484
    },
    {
      "epoch": 0.6798581473389013,
      "grad_norm": 0.6196194038765365,
      "learning_rate": 0.00015371157303935604,
      "loss": 12.4834,
      "step": 12485
    },
    {
      "epoch": 0.6799126013354843,
      "grad_norm": 0.7024144124020267,
      "learning_rate": 0.00015370413455427758,
      "loss": 12.3145,
      "step": 12486
    },
    {
      "epoch": 0.6799670553320672,
      "grad_norm": 0.5335364578058331,
      "learning_rate": 0.0001536966956515868,
      "loss": 12.1881,
      "step": 12487
    },
    {
      "epoch": 0.6800215093286502,
      "grad_norm": 0.6308656762616506,
      "learning_rate": 0.00015368925633134145,
      "loss": 12.3981,
      "step": 12488
    },
    {
      "epoch": 0.6800759633252333,
      "grad_norm": 0.5609199684595741,
      "learning_rate": 0.0001536818165935995,
      "loss": 12.1629,
      "step": 12489
    },
    {
      "epoch": 0.6801304173218163,
      "grad_norm": 0.5882023506431455,
      "learning_rate": 0.0001536743764384187,
      "loss": 12.217,
      "step": 12490
    },
    {
      "epoch": 0.6801848713183993,
      "grad_norm": 0.5830886102468975,
      "learning_rate": 0.00015366693586585698,
      "loss": 12.3887,
      "step": 12491
    },
    {
      "epoch": 0.6802393253149823,
      "grad_norm": 0.5933736821972734,
      "learning_rate": 0.00015365949487597215,
      "loss": 12.2738,
      "step": 12492
    },
    {
      "epoch": 0.6802937793115653,
      "grad_norm": 0.6270327180411331,
      "learning_rate": 0.00015365205346882211,
      "loss": 12.4265,
      "step": 12493
    },
    {
      "epoch": 0.6803482333081483,
      "grad_norm": 0.6123050154368799,
      "learning_rate": 0.0001536446116444647,
      "loss": 12.2808,
      "step": 12494
    },
    {
      "epoch": 0.6804026873047314,
      "grad_norm": 0.6149944772394959,
      "learning_rate": 0.00015363716940295776,
      "loss": 12.283,
      "step": 12495
    },
    {
      "epoch": 0.6804571413013144,
      "grad_norm": 0.5767155457877045,
      "learning_rate": 0.00015362972674435923,
      "loss": 12.3108,
      "step": 12496
    },
    {
      "epoch": 0.6805115952978974,
      "grad_norm": 0.6095755670317139,
      "learning_rate": 0.00015362228366872692,
      "loss": 12.3838,
      "step": 12497
    },
    {
      "epoch": 0.6805660492944804,
      "grad_norm": 0.5932684303723733,
      "learning_rate": 0.00015361484017611878,
      "loss": 12.3602,
      "step": 12498
    },
    {
      "epoch": 0.6806205032910634,
      "grad_norm": 0.557865255664125,
      "learning_rate": 0.00015360739626659264,
      "loss": 12.2614,
      "step": 12499
    },
    {
      "epoch": 0.6806749572876464,
      "grad_norm": 0.5466275613168837,
      "learning_rate": 0.00015359995194020635,
      "loss": 12.1963,
      "step": 12500
    },
    {
      "epoch": 0.6807294112842295,
      "grad_norm": 0.5722638998858818,
      "learning_rate": 0.0001535925071970179,
      "loss": 12.3277,
      "step": 12501
    },
    {
      "epoch": 0.6807838652808125,
      "grad_norm": 0.5719742566065302,
      "learning_rate": 0.0001535850620370851,
      "loss": 12.301,
      "step": 12502
    },
    {
      "epoch": 0.6808383192773955,
      "grad_norm": 0.5978564215852396,
      "learning_rate": 0.00015357761646046586,
      "loss": 12.4602,
      "step": 12503
    },
    {
      "epoch": 0.6808927732739785,
      "grad_norm": 0.6845621944319514,
      "learning_rate": 0.00015357017046721815,
      "loss": 12.2861,
      "step": 12504
    },
    {
      "epoch": 0.6809472272705615,
      "grad_norm": 0.5446505254122316,
      "learning_rate": 0.00015356272405739975,
      "loss": 12.2506,
      "step": 12505
    },
    {
      "epoch": 0.6810016812671446,
      "grad_norm": 0.5764664656405027,
      "learning_rate": 0.00015355527723106866,
      "loss": 12.1196,
      "step": 12506
    },
    {
      "epoch": 0.6810561352637275,
      "grad_norm": 0.5824158968217736,
      "learning_rate": 0.00015354782998828272,
      "loss": 12.1073,
      "step": 12507
    },
    {
      "epoch": 0.6811105892603105,
      "grad_norm": 0.5951660676019116,
      "learning_rate": 0.0001535403823290999,
      "loss": 12.1546,
      "step": 12508
    },
    {
      "epoch": 0.6811650432568935,
      "grad_norm": 0.633650384302766,
      "learning_rate": 0.0001535329342535781,
      "loss": 12.3471,
      "step": 12509
    },
    {
      "epoch": 0.6812194972534765,
      "grad_norm": 0.7103507550363984,
      "learning_rate": 0.0001535254857617752,
      "loss": 12.4524,
      "step": 12510
    },
    {
      "epoch": 0.6812739512500595,
      "grad_norm": 0.57847764762371,
      "learning_rate": 0.00015351803685374914,
      "loss": 12.2622,
      "step": 12511
    },
    {
      "epoch": 0.6813284052466426,
      "grad_norm": 0.6403836667625455,
      "learning_rate": 0.00015351058752955784,
      "loss": 12.3535,
      "step": 12512
    },
    {
      "epoch": 0.6813828592432256,
      "grad_norm": 0.7088629683482663,
      "learning_rate": 0.0001535031377892593,
      "loss": 12.3262,
      "step": 12513
    },
    {
      "epoch": 0.6814373132398086,
      "grad_norm": 0.6193844866774646,
      "learning_rate": 0.00015349568763291135,
      "loss": 12.3267,
      "step": 12514
    },
    {
      "epoch": 0.6814917672363916,
      "grad_norm": 0.5998515129373875,
      "learning_rate": 0.00015348823706057196,
      "loss": 12.3904,
      "step": 12515
    },
    {
      "epoch": 0.6815462212329746,
      "grad_norm": 0.6448018105822136,
      "learning_rate": 0.00015348078607229905,
      "loss": 12.4076,
      "step": 12516
    },
    {
      "epoch": 0.6816006752295576,
      "grad_norm": 0.6148133586569113,
      "learning_rate": 0.00015347333466815059,
      "loss": 12.2451,
      "step": 12517
    },
    {
      "epoch": 0.6816551292261407,
      "grad_norm": 0.5678300825860817,
      "learning_rate": 0.00015346588284818454,
      "loss": 12.2692,
      "step": 12518
    },
    {
      "epoch": 0.6817095832227237,
      "grad_norm": 0.5768317706320972,
      "learning_rate": 0.0001534584306124588,
      "loss": 12.3386,
      "step": 12519
    },
    {
      "epoch": 0.6817640372193067,
      "grad_norm": 0.5331376057490201,
      "learning_rate": 0.00015345097796103135,
      "loss": 12.1922,
      "step": 12520
    },
    {
      "epoch": 0.6818184912158897,
      "grad_norm": 0.5544525547019814,
      "learning_rate": 0.0001534435248939601,
      "loss": 12.269,
      "step": 12521
    },
    {
      "epoch": 0.6818729452124727,
      "grad_norm": 0.6193524364573895,
      "learning_rate": 0.00015343607141130304,
      "loss": 12.2663,
      "step": 12522
    },
    {
      "epoch": 0.6819273992090557,
      "grad_norm": 0.5619608059312059,
      "learning_rate": 0.00015342861751311814,
      "loss": 12.3329,
      "step": 12523
    },
    {
      "epoch": 0.6819818532056388,
      "grad_norm": 0.6552963097235771,
      "learning_rate": 0.00015342116319946338,
      "loss": 12.3595,
      "step": 12524
    },
    {
      "epoch": 0.6820363072022217,
      "grad_norm": 0.570274795582616,
      "learning_rate": 0.00015341370847039666,
      "loss": 12.2273,
      "step": 12525
    },
    {
      "epoch": 0.6820907611988047,
      "grad_norm": 0.6239227052665182,
      "learning_rate": 0.000153406253325976,
      "loss": 12.3095,
      "step": 12526
    },
    {
      "epoch": 0.6821452151953877,
      "grad_norm": 0.5745369017605514,
      "learning_rate": 0.00015339879776625937,
      "loss": 12.2778,
      "step": 12527
    },
    {
      "epoch": 0.6821996691919707,
      "grad_norm": 0.6025091065614796,
      "learning_rate": 0.00015339134179130469,
      "loss": 12.248,
      "step": 12528
    },
    {
      "epoch": 0.6822541231885537,
      "grad_norm": 0.6598333268101045,
      "learning_rate": 0.00015338388540117004,
      "loss": 12.1671,
      "step": 12529
    },
    {
      "epoch": 0.6823085771851368,
      "grad_norm": 0.6316947537516462,
      "learning_rate": 0.0001533764285959133,
      "loss": 12.3167,
      "step": 12530
    },
    {
      "epoch": 0.6823630311817198,
      "grad_norm": 0.6344343759849481,
      "learning_rate": 0.0001533689713755925,
      "loss": 12.2439,
      "step": 12531
    },
    {
      "epoch": 0.6824174851783028,
      "grad_norm": 0.6129540352843904,
      "learning_rate": 0.00015336151374026566,
      "loss": 12.1051,
      "step": 12532
    },
    {
      "epoch": 0.6824719391748858,
      "grad_norm": 0.6456101389818156,
      "learning_rate": 0.00015335405568999073,
      "loss": 12.3736,
      "step": 12533
    },
    {
      "epoch": 0.6825263931714688,
      "grad_norm": 0.60867641200438,
      "learning_rate": 0.00015334659722482572,
      "loss": 12.2544,
      "step": 12534
    },
    {
      "epoch": 0.6825808471680518,
      "grad_norm": 0.6173784654404763,
      "learning_rate": 0.0001533391383448286,
      "loss": 12.2428,
      "step": 12535
    },
    {
      "epoch": 0.6826353011646349,
      "grad_norm": 0.6030950147532718,
      "learning_rate": 0.0001533316790500574,
      "loss": 12.282,
      "step": 12536
    },
    {
      "epoch": 0.6826897551612179,
      "grad_norm": 0.6110934537663895,
      "learning_rate": 0.00015332421934057013,
      "loss": 12.304,
      "step": 12537
    },
    {
      "epoch": 0.6827442091578009,
      "grad_norm": 0.6925727134598422,
      "learning_rate": 0.00015331675921642478,
      "loss": 12.3138,
      "step": 12538
    },
    {
      "epoch": 0.6827986631543839,
      "grad_norm": 0.5930059362458303,
      "learning_rate": 0.00015330929867767937,
      "loss": 12.3151,
      "step": 12539
    },
    {
      "epoch": 0.6828531171509669,
      "grad_norm": 0.7085795714441075,
      "learning_rate": 0.00015330183772439195,
      "loss": 12.3222,
      "step": 12540
    },
    {
      "epoch": 0.68290757114755,
      "grad_norm": 0.5369805349876322,
      "learning_rate": 0.00015329437635662046,
      "loss": 12.2943,
      "step": 12541
    },
    {
      "epoch": 0.682962025144133,
      "grad_norm": 0.544593423692748,
      "learning_rate": 0.00015328691457442296,
      "loss": 12.257,
      "step": 12542
    },
    {
      "epoch": 0.683016479140716,
      "grad_norm": 0.5884016852779346,
      "learning_rate": 0.00015327945237785748,
      "loss": 12.261,
      "step": 12543
    },
    {
      "epoch": 0.6830709331372989,
      "grad_norm": 0.5874520382845972,
      "learning_rate": 0.00015327198976698204,
      "loss": 12.3641,
      "step": 12544
    },
    {
      "epoch": 0.6831253871338819,
      "grad_norm": 0.5936986434307064,
      "learning_rate": 0.00015326452674185468,
      "loss": 12.3483,
      "step": 12545
    },
    {
      "epoch": 0.6831798411304649,
      "grad_norm": 0.5729493782893559,
      "learning_rate": 0.00015325706330253348,
      "loss": 12.3482,
      "step": 12546
    },
    {
      "epoch": 0.683234295127048,
      "grad_norm": 0.5379559808268314,
      "learning_rate": 0.00015324959944907636,
      "loss": 12.38,
      "step": 12547
    },
    {
      "epoch": 0.683288749123631,
      "grad_norm": 0.5508562129553998,
      "learning_rate": 0.00015324213518154144,
      "loss": 12.2574,
      "step": 12548
    },
    {
      "epoch": 0.683343203120214,
      "grad_norm": 0.5528210383541169,
      "learning_rate": 0.00015323467049998675,
      "loss": 12.1128,
      "step": 12549
    },
    {
      "epoch": 0.683397657116797,
      "grad_norm": 0.5690346282873607,
      "learning_rate": 0.00015322720540447034,
      "loss": 12.0657,
      "step": 12550
    },
    {
      "epoch": 0.68345211111338,
      "grad_norm": 0.6207858687490257,
      "learning_rate": 0.00015321973989505024,
      "loss": 12.1423,
      "step": 12551
    },
    {
      "epoch": 0.683506565109963,
      "grad_norm": 0.5401235764199944,
      "learning_rate": 0.00015321227397178456,
      "loss": 12.1383,
      "step": 12552
    },
    {
      "epoch": 0.6835610191065461,
      "grad_norm": 0.5639866531793937,
      "learning_rate": 0.00015320480763473127,
      "loss": 12.3375,
      "step": 12553
    },
    {
      "epoch": 0.6836154731031291,
      "grad_norm": 0.5569695080226263,
      "learning_rate": 0.00015319734088394847,
      "loss": 12.2254,
      "step": 12554
    },
    {
      "epoch": 0.6836699270997121,
      "grad_norm": 0.5823886376253445,
      "learning_rate": 0.00015318987371949424,
      "loss": 12.1081,
      "step": 12555
    },
    {
      "epoch": 0.6837243810962951,
      "grad_norm": 0.5831867835423197,
      "learning_rate": 0.0001531824061414266,
      "loss": 12.4263,
      "step": 12556
    },
    {
      "epoch": 0.6837788350928781,
      "grad_norm": 0.6368773183461842,
      "learning_rate": 0.00015317493814980373,
      "loss": 12.3931,
      "step": 12557
    },
    {
      "epoch": 0.6838332890894611,
      "grad_norm": 0.5967235892556965,
      "learning_rate": 0.00015316746974468356,
      "loss": 12.3436,
      "step": 12558
    },
    {
      "epoch": 0.6838877430860442,
      "grad_norm": 0.5611941804067937,
      "learning_rate": 0.00015316000092612425,
      "loss": 12.2935,
      "step": 12559
    },
    {
      "epoch": 0.6839421970826272,
      "grad_norm": 0.5821625925593092,
      "learning_rate": 0.00015315253169418384,
      "loss": 12.3281,
      "step": 12560
    },
    {
      "epoch": 0.6839966510792101,
      "grad_norm": 0.58929587624123,
      "learning_rate": 0.00015314506204892046,
      "loss": 12.2427,
      "step": 12561
    },
    {
      "epoch": 0.6840511050757931,
      "grad_norm": 0.9656728003292488,
      "learning_rate": 0.00015313759199039214,
      "loss": 12.3526,
      "step": 12562
    },
    {
      "epoch": 0.6841055590723761,
      "grad_norm": 0.6415621689917624,
      "learning_rate": 0.000153130121518657,
      "loss": 12.2039,
      "step": 12563
    },
    {
      "epoch": 0.6841600130689591,
      "grad_norm": 0.6300005979679029,
      "learning_rate": 0.00015312265063377315,
      "loss": 12.3009,
      "step": 12564
    },
    {
      "epoch": 0.6842144670655422,
      "grad_norm": 0.5685462441014264,
      "learning_rate": 0.00015311517933579865,
      "loss": 12.2691,
      "step": 12565
    },
    {
      "epoch": 0.6842689210621252,
      "grad_norm": 0.5643867251692329,
      "learning_rate": 0.00015310770762479156,
      "loss": 12.2821,
      "step": 12566
    },
    {
      "epoch": 0.6843233750587082,
      "grad_norm": 0.5868894551224721,
      "learning_rate": 0.00015310023550081008,
      "loss": 12.2581,
      "step": 12567
    },
    {
      "epoch": 0.6843778290552912,
      "grad_norm": 0.5726335847878873,
      "learning_rate": 0.00015309276296391223,
      "loss": 12.2945,
      "step": 12568
    },
    {
      "epoch": 0.6844322830518742,
      "grad_norm": 0.6777200529938219,
      "learning_rate": 0.0001530852900141562,
      "loss": 12.3371,
      "step": 12569
    },
    {
      "epoch": 0.6844867370484573,
      "grad_norm": 0.6001064109162652,
      "learning_rate": 0.00015307781665160005,
      "loss": 12.3353,
      "step": 12570
    },
    {
      "epoch": 0.6845411910450403,
      "grad_norm": 0.5606026837695675,
      "learning_rate": 0.00015307034287630184,
      "loss": 12.3618,
      "step": 12571
    },
    {
      "epoch": 0.6845956450416233,
      "grad_norm": 0.5898496714352334,
      "learning_rate": 0.00015306286868831977,
      "loss": 12.2474,
      "step": 12572
    },
    {
      "epoch": 0.6846500990382063,
      "grad_norm": 0.5889931142568797,
      "learning_rate": 0.00015305539408771193,
      "loss": 12.3571,
      "step": 12573
    },
    {
      "epoch": 0.6847045530347893,
      "grad_norm": 0.6156435547693119,
      "learning_rate": 0.00015304791907453646,
      "loss": 12.2613,
      "step": 12574
    },
    {
      "epoch": 0.6847590070313723,
      "grad_norm": 0.6230231332692246,
      "learning_rate": 0.00015304044364885147,
      "loss": 12.4422,
      "step": 12575
    },
    {
      "epoch": 0.6848134610279554,
      "grad_norm": 0.6871335752250709,
      "learning_rate": 0.0001530329678107151,
      "loss": 12.3385,
      "step": 12576
    },
    {
      "epoch": 0.6848679150245384,
      "grad_norm": 0.623158174408285,
      "learning_rate": 0.00015302549156018545,
      "loss": 12.2042,
      "step": 12577
    },
    {
      "epoch": 0.6849223690211214,
      "grad_norm": 0.5706635418917007,
      "learning_rate": 0.00015301801489732073,
      "loss": 12.1545,
      "step": 12578
    },
    {
      "epoch": 0.6849768230177044,
      "grad_norm": 0.55862280847443,
      "learning_rate": 0.00015301053782217902,
      "loss": 12.2803,
      "step": 12579
    },
    {
      "epoch": 0.6850312770142873,
      "grad_norm": 0.5992289622696629,
      "learning_rate": 0.0001530030603348185,
      "loss": 12.2505,
      "step": 12580
    },
    {
      "epoch": 0.6850857310108703,
      "grad_norm": 0.587073117141727,
      "learning_rate": 0.00015299558243529724,
      "loss": 12.2734,
      "step": 12581
    },
    {
      "epoch": 0.6851401850074534,
      "grad_norm": 0.5766027769184066,
      "learning_rate": 0.00015298810412367348,
      "loss": 12.3116,
      "step": 12582
    },
    {
      "epoch": 0.6851946390040364,
      "grad_norm": 0.6304471982153476,
      "learning_rate": 0.00015298062540000532,
      "loss": 12.3564,
      "step": 12583
    },
    {
      "epoch": 0.6852490930006194,
      "grad_norm": 0.6097819467526905,
      "learning_rate": 0.00015297314626435093,
      "loss": 12.3909,
      "step": 12584
    },
    {
      "epoch": 0.6853035469972024,
      "grad_norm": 0.5925291505144655,
      "learning_rate": 0.0001529656667167685,
      "loss": 12.2947,
      "step": 12585
    },
    {
      "epoch": 0.6853580009937854,
      "grad_norm": 0.5807666138404012,
      "learning_rate": 0.00015295818675731612,
      "loss": 12.2874,
      "step": 12586
    },
    {
      "epoch": 0.6854124549903684,
      "grad_norm": 0.6025571608827972,
      "learning_rate": 0.00015295070638605203,
      "loss": 12.2434,
      "step": 12587
    },
    {
      "epoch": 0.6854669089869515,
      "grad_norm": 0.6237484803353083,
      "learning_rate": 0.00015294322560303436,
      "loss": 12.3145,
      "step": 12588
    },
    {
      "epoch": 0.6855213629835345,
      "grad_norm": 0.5948647440033049,
      "learning_rate": 0.0001529357444083213,
      "loss": 12.2099,
      "step": 12589
    },
    {
      "epoch": 0.6855758169801175,
      "grad_norm": 0.605663960635657,
      "learning_rate": 0.00015292826280197098,
      "loss": 12.2644,
      "step": 12590
    },
    {
      "epoch": 0.6856302709767005,
      "grad_norm": 0.5990931213074819,
      "learning_rate": 0.00015292078078404166,
      "loss": 12.2598,
      "step": 12591
    },
    {
      "epoch": 0.6856847249732835,
      "grad_norm": 0.6073467998974551,
      "learning_rate": 0.00015291329835459146,
      "loss": 12.3719,
      "step": 12592
    },
    {
      "epoch": 0.6857391789698665,
      "grad_norm": 0.614944758172525,
      "learning_rate": 0.00015290581551367856,
      "loss": 12.2056,
      "step": 12593
    },
    {
      "epoch": 0.6857936329664496,
      "grad_norm": 0.6222266110368851,
      "learning_rate": 0.00015289833226136116,
      "loss": 12.2476,
      "step": 12594
    },
    {
      "epoch": 0.6858480869630326,
      "grad_norm": 0.639785126460829,
      "learning_rate": 0.00015289084859769746,
      "loss": 12.4306,
      "step": 12595
    },
    {
      "epoch": 0.6859025409596156,
      "grad_norm": 0.5938562698228961,
      "learning_rate": 0.0001528833645227457,
      "loss": 12.2332,
      "step": 12596
    },
    {
      "epoch": 0.6859569949561986,
      "grad_norm": 0.6253395712765962,
      "learning_rate": 0.00015287588003656398,
      "loss": 12.2702,
      "step": 12597
    },
    {
      "epoch": 0.6860114489527815,
      "grad_norm": 0.7806519859510992,
      "learning_rate": 0.00015286839513921056,
      "loss": 12.2199,
      "step": 12598
    },
    {
      "epoch": 0.6860659029493645,
      "grad_norm": 0.6383790912822903,
      "learning_rate": 0.00015286090983074365,
      "loss": 12.3867,
      "step": 12599
    },
    {
      "epoch": 0.6861203569459476,
      "grad_norm": 0.6354619200163265,
      "learning_rate": 0.0001528534241112214,
      "loss": 12.2405,
      "step": 12600
    },
    {
      "epoch": 0.6861748109425306,
      "grad_norm": 0.7198083446997495,
      "learning_rate": 0.0001528459379807021,
      "loss": 12.1971,
      "step": 12601
    },
    {
      "epoch": 0.6862292649391136,
      "grad_norm": 0.5609639402030633,
      "learning_rate": 0.0001528384514392439,
      "loss": 12.2583,
      "step": 12602
    },
    {
      "epoch": 0.6862837189356966,
      "grad_norm": 0.6590477541121527,
      "learning_rate": 0.00015283096448690504,
      "loss": 12.4317,
      "step": 12603
    },
    {
      "epoch": 0.6863381729322796,
      "grad_norm": 0.6169043747317672,
      "learning_rate": 0.00015282347712374376,
      "loss": 12.2422,
      "step": 12604
    },
    {
      "epoch": 0.6863926269288627,
      "grad_norm": 0.5924528998422541,
      "learning_rate": 0.00015281598934981826,
      "loss": 12.3041,
      "step": 12605
    },
    {
      "epoch": 0.6864470809254457,
      "grad_norm": 0.6317948636202078,
      "learning_rate": 0.00015280850116518673,
      "loss": 12.3189,
      "step": 12606
    },
    {
      "epoch": 0.6865015349220287,
      "grad_norm": 0.6003665146103847,
      "learning_rate": 0.00015280101256990748,
      "loss": 12.2639,
      "step": 12607
    },
    {
      "epoch": 0.6865559889186117,
      "grad_norm": 0.4923224569887765,
      "learning_rate": 0.00015279352356403872,
      "loss": 11.9896,
      "step": 12608
    },
    {
      "epoch": 0.6866104429151947,
      "grad_norm": 0.6174994382831572,
      "learning_rate": 0.00015278603414763863,
      "loss": 12.2954,
      "step": 12609
    },
    {
      "epoch": 0.6866648969117777,
      "grad_norm": 0.5857052679957093,
      "learning_rate": 0.0001527785443207655,
      "loss": 12.3063,
      "step": 12610
    },
    {
      "epoch": 0.6867193509083608,
      "grad_norm": 0.5743478781291048,
      "learning_rate": 0.00015277105408347756,
      "loss": 12.1504,
      "step": 12611
    },
    {
      "epoch": 0.6867738049049438,
      "grad_norm": 0.5559483050548155,
      "learning_rate": 0.00015276356343583305,
      "loss": 12.2322,
      "step": 12612
    },
    {
      "epoch": 0.6868282589015268,
      "grad_norm": 0.6352063878137918,
      "learning_rate": 0.00015275607237789023,
      "loss": 12.338,
      "step": 12613
    },
    {
      "epoch": 0.6868827128981098,
      "grad_norm": 0.561237471566255,
      "learning_rate": 0.00015274858090970735,
      "loss": 12.1346,
      "step": 12614
    },
    {
      "epoch": 0.6869371668946928,
      "grad_norm": 0.6444034241684865,
      "learning_rate": 0.00015274108903134262,
      "loss": 12.4475,
      "step": 12615
    },
    {
      "epoch": 0.6869916208912757,
      "grad_norm": 0.6399563424531881,
      "learning_rate": 0.00015273359674285438,
      "loss": 12.1246,
      "step": 12616
    },
    {
      "epoch": 0.6870460748878588,
      "grad_norm": 0.557675000921174,
      "learning_rate": 0.00015272610404430082,
      "loss": 12.1814,
      "step": 12617
    },
    {
      "epoch": 0.6871005288844418,
      "grad_norm": 0.6547315754685608,
      "learning_rate": 0.0001527186109357402,
      "loss": 12.3489,
      "step": 12618
    },
    {
      "epoch": 0.6871549828810248,
      "grad_norm": 0.6158957717630221,
      "learning_rate": 0.0001527111174172309,
      "loss": 12.2577,
      "step": 12619
    },
    {
      "epoch": 0.6872094368776078,
      "grad_norm": 0.6405001217155041,
      "learning_rate": 0.00015270362348883108,
      "loss": 12.3835,
      "step": 12620
    },
    {
      "epoch": 0.6872638908741908,
      "grad_norm": 0.7138640749459229,
      "learning_rate": 0.00015269612915059903,
      "loss": 12.4221,
      "step": 12621
    },
    {
      "epoch": 0.6873183448707738,
      "grad_norm": 0.5418849717601474,
      "learning_rate": 0.00015268863440259307,
      "loss": 12.3072,
      "step": 12622
    },
    {
      "epoch": 0.6873727988673569,
      "grad_norm": 0.6338427053423075,
      "learning_rate": 0.0001526811392448714,
      "loss": 12.316,
      "step": 12623
    },
    {
      "epoch": 0.6874272528639399,
      "grad_norm": 0.6424556749417951,
      "learning_rate": 0.00015267364367749242,
      "loss": 12.2569,
      "step": 12624
    },
    {
      "epoch": 0.6874817068605229,
      "grad_norm": 0.5318951779941986,
      "learning_rate": 0.00015266614770051434,
      "loss": 12.301,
      "step": 12625
    },
    {
      "epoch": 0.6875361608571059,
      "grad_norm": 0.5824150194263151,
      "learning_rate": 0.00015265865131399543,
      "loss": 12.4017,
      "step": 12626
    },
    {
      "epoch": 0.6875906148536889,
      "grad_norm": 0.6239848450959841,
      "learning_rate": 0.00015265115451799403,
      "loss": 12.3879,
      "step": 12627
    },
    {
      "epoch": 0.6876450688502719,
      "grad_norm": 0.5886551546118839,
      "learning_rate": 0.0001526436573125684,
      "loss": 12.349,
      "step": 12628
    },
    {
      "epoch": 0.687699522846855,
      "grad_norm": 0.6041463662522617,
      "learning_rate": 0.00015263615969777691,
      "loss": 12.2759,
      "step": 12629
    },
    {
      "epoch": 0.687753976843438,
      "grad_norm": 0.6335787004896745,
      "learning_rate": 0.00015262866167367785,
      "loss": 12.2605,
      "step": 12630
    },
    {
      "epoch": 0.687808430840021,
      "grad_norm": 0.5616758429928841,
      "learning_rate": 0.00015262116324032943,
      "loss": 12.2135,
      "step": 12631
    },
    {
      "epoch": 0.687862884836604,
      "grad_norm": 0.5568859215137671,
      "learning_rate": 0.00015261366439779,
      "loss": 12.1521,
      "step": 12632
    },
    {
      "epoch": 0.687917338833187,
      "grad_norm": 0.5301411883490512,
      "learning_rate": 0.00015260616514611793,
      "loss": 12.1824,
      "step": 12633
    },
    {
      "epoch": 0.68797179282977,
      "grad_norm": 0.6112030855999007,
      "learning_rate": 0.00015259866548537144,
      "loss": 12.3687,
      "step": 12634
    },
    {
      "epoch": 0.688026246826353,
      "grad_norm": 0.638658439137997,
      "learning_rate": 0.00015259116541560902,
      "loss": 12.3406,
      "step": 12635
    },
    {
      "epoch": 0.688080700822936,
      "grad_norm": 0.5849601548744854,
      "learning_rate": 0.0001525836649368888,
      "loss": 12.2362,
      "step": 12636
    },
    {
      "epoch": 0.688135154819519,
      "grad_norm": 0.5890796896219497,
      "learning_rate": 0.00015257616404926915,
      "loss": 12.2527,
      "step": 12637
    },
    {
      "epoch": 0.688189608816102,
      "grad_norm": 0.5799865995323636,
      "learning_rate": 0.0001525686627528085,
      "loss": 12.3209,
      "step": 12638
    },
    {
      "epoch": 0.688244062812685,
      "grad_norm": 0.6151652030859852,
      "learning_rate": 0.00015256116104756507,
      "loss": 12.3799,
      "step": 12639
    },
    {
      "epoch": 0.6882985168092681,
      "grad_norm": 0.5264196989976043,
      "learning_rate": 0.0001525536589335973,
      "loss": 12.2559,
      "step": 12640
    },
    {
      "epoch": 0.6883529708058511,
      "grad_norm": 0.5980748544120343,
      "learning_rate": 0.0001525461564109634,
      "loss": 12.2105,
      "step": 12641
    },
    {
      "epoch": 0.6884074248024341,
      "grad_norm": 0.5726667071252022,
      "learning_rate": 0.0001525386534797218,
      "loss": 12.2894,
      "step": 12642
    },
    {
      "epoch": 0.6884618787990171,
      "grad_norm": 0.5679190932449971,
      "learning_rate": 0.0001525311501399308,
      "loss": 12.3174,
      "step": 12643
    },
    {
      "epoch": 0.6885163327956001,
      "grad_norm": 0.5482384017862806,
      "learning_rate": 0.0001525236463916488,
      "loss": 12.2429,
      "step": 12644
    },
    {
      "epoch": 0.6885707867921831,
      "grad_norm": 0.547876635201763,
      "learning_rate": 0.00015251614223493412,
      "loss": 12.3051,
      "step": 12645
    },
    {
      "epoch": 0.6886252407887662,
      "grad_norm": 0.6059517008941783,
      "learning_rate": 0.00015250863766984508,
      "loss": 12.2157,
      "step": 12646
    },
    {
      "epoch": 0.6886796947853492,
      "grad_norm": 0.5977206310025369,
      "learning_rate": 0.00015250113269644005,
      "loss": 12.4078,
      "step": 12647
    },
    {
      "epoch": 0.6887341487819322,
      "grad_norm": 0.5568252330295134,
      "learning_rate": 0.00015249362731477745,
      "loss": 12.3534,
      "step": 12648
    },
    {
      "epoch": 0.6887886027785152,
      "grad_norm": 0.6075483860753537,
      "learning_rate": 0.00015248612152491558,
      "loss": 12.3157,
      "step": 12649
    },
    {
      "epoch": 0.6888430567750982,
      "grad_norm": 0.5951457195496563,
      "learning_rate": 0.00015247861532691285,
      "loss": 12.3323,
      "step": 12650
    },
    {
      "epoch": 0.6888975107716812,
      "grad_norm": 0.6261098231237463,
      "learning_rate": 0.00015247110872082759,
      "loss": 12.37,
      "step": 12651
    },
    {
      "epoch": 0.6889519647682643,
      "grad_norm": 0.651710583732914,
      "learning_rate": 0.0001524636017067182,
      "loss": 12.3662,
      "step": 12652
    },
    {
      "epoch": 0.6890064187648473,
      "grad_norm": 0.6357382069552527,
      "learning_rate": 0.00015245609428464306,
      "loss": 12.4347,
      "step": 12653
    },
    {
      "epoch": 0.6890608727614302,
      "grad_norm": 0.6587152131477861,
      "learning_rate": 0.00015244858645466046,
      "loss": 12.3483,
      "step": 12654
    },
    {
      "epoch": 0.6891153267580132,
      "grad_norm": 0.5647572288398478,
      "learning_rate": 0.00015244107821682894,
      "loss": 12.2754,
      "step": 12655
    },
    {
      "epoch": 0.6891697807545962,
      "grad_norm": 0.5548134170905816,
      "learning_rate": 0.00015243356957120676,
      "loss": 12.2864,
      "step": 12656
    },
    {
      "epoch": 0.6892242347511792,
      "grad_norm": 0.6079633115581555,
      "learning_rate": 0.00015242606051785236,
      "loss": 12.3156,
      "step": 12657
    },
    {
      "epoch": 0.6892786887477623,
      "grad_norm": 0.6751531247949729,
      "learning_rate": 0.00015241855105682412,
      "loss": 12.2497,
      "step": 12658
    },
    {
      "epoch": 0.6893331427443453,
      "grad_norm": 0.5337554194327473,
      "learning_rate": 0.00015241104118818042,
      "loss": 12.311,
      "step": 12659
    },
    {
      "epoch": 0.6893875967409283,
      "grad_norm": 0.6036338919681892,
      "learning_rate": 0.0001524035309119797,
      "loss": 12.2545,
      "step": 12660
    },
    {
      "epoch": 0.6894420507375113,
      "grad_norm": 0.5690388964265504,
      "learning_rate": 0.00015239602022828032,
      "loss": 12.2702,
      "step": 12661
    },
    {
      "epoch": 0.6894965047340943,
      "grad_norm": 0.6315982376221074,
      "learning_rate": 0.0001523885091371407,
      "loss": 12.1872,
      "step": 12662
    },
    {
      "epoch": 0.6895509587306773,
      "grad_norm": 0.560581222380461,
      "learning_rate": 0.00015238099763861926,
      "loss": 12.1801,
      "step": 12663
    },
    {
      "epoch": 0.6896054127272604,
      "grad_norm": 0.5862186620249155,
      "learning_rate": 0.0001523734857327744,
      "loss": 12.1854,
      "step": 12664
    },
    {
      "epoch": 0.6896598667238434,
      "grad_norm": 0.5748225751008038,
      "learning_rate": 0.00015236597341966452,
      "loss": 12.3736,
      "step": 12665
    },
    {
      "epoch": 0.6897143207204264,
      "grad_norm": 0.5517267261087246,
      "learning_rate": 0.00015235846069934804,
      "loss": 12.1587,
      "step": 12666
    },
    {
      "epoch": 0.6897687747170094,
      "grad_norm": 0.7096481690724898,
      "learning_rate": 0.0001523509475718834,
      "loss": 12.2512,
      "step": 12667
    },
    {
      "epoch": 0.6898232287135924,
      "grad_norm": 0.6735552712041467,
      "learning_rate": 0.00015234343403732902,
      "loss": 12.1401,
      "step": 12668
    },
    {
      "epoch": 0.6898776827101754,
      "grad_norm": 0.9097540623374685,
      "learning_rate": 0.0001523359200957433,
      "loss": 12.3809,
      "step": 12669
    },
    {
      "epoch": 0.6899321367067585,
      "grad_norm": 0.6413803080548649,
      "learning_rate": 0.00015232840574718473,
      "loss": 12.2742,
      "step": 12670
    },
    {
      "epoch": 0.6899865907033415,
      "grad_norm": 0.7383597525167909,
      "learning_rate": 0.00015232089099171165,
      "loss": 12.2784,
      "step": 12671
    },
    {
      "epoch": 0.6900410446999244,
      "grad_norm": 0.7636786695259324,
      "learning_rate": 0.00015231337582938257,
      "loss": 12.4001,
      "step": 12672
    },
    {
      "epoch": 0.6900954986965074,
      "grad_norm": 0.6394640382256751,
      "learning_rate": 0.0001523058602602559,
      "loss": 12.2575,
      "step": 12673
    },
    {
      "epoch": 0.6901499526930904,
      "grad_norm": 0.6564638150277223,
      "learning_rate": 0.0001522983442843901,
      "loss": 12.2666,
      "step": 12674
    },
    {
      "epoch": 0.6902044066896735,
      "grad_norm": 0.6819497811738562,
      "learning_rate": 0.0001522908279018436,
      "loss": 12.276,
      "step": 12675
    },
    {
      "epoch": 0.6902588606862565,
      "grad_norm": 0.6309976102996402,
      "learning_rate": 0.00015228331111267487,
      "loss": 12.3654,
      "step": 12676
    },
    {
      "epoch": 0.6903133146828395,
      "grad_norm": 0.632177406445353,
      "learning_rate": 0.0001522757939169423,
      "loss": 12.2867,
      "step": 12677
    },
    {
      "epoch": 0.6903677686794225,
      "grad_norm": 0.6023323477054283,
      "learning_rate": 0.00015226827631470443,
      "loss": 12.3492,
      "step": 12678
    },
    {
      "epoch": 0.6904222226760055,
      "grad_norm": 0.7334268999145962,
      "learning_rate": 0.00015226075830601966,
      "loss": 12.4541,
      "step": 12679
    },
    {
      "epoch": 0.6904766766725885,
      "grad_norm": 0.6600399118868495,
      "learning_rate": 0.0001522532398909465,
      "loss": 12.3805,
      "step": 12680
    },
    {
      "epoch": 0.6905311306691716,
      "grad_norm": 0.6266545126408947,
      "learning_rate": 0.00015224572106954334,
      "loss": 12.1993,
      "step": 12681
    },
    {
      "epoch": 0.6905855846657546,
      "grad_norm": 0.6731096915303995,
      "learning_rate": 0.0001522382018418687,
      "loss": 12.1659,
      "step": 12682
    },
    {
      "epoch": 0.6906400386623376,
      "grad_norm": 0.5633882992088448,
      "learning_rate": 0.00015223068220798105,
      "loss": 12.1641,
      "step": 12683
    },
    {
      "epoch": 0.6906944926589206,
      "grad_norm": 0.6724408702456605,
      "learning_rate": 0.00015222316216793885,
      "loss": 12.3575,
      "step": 12684
    },
    {
      "epoch": 0.6907489466555036,
      "grad_norm": 0.9751431943544033,
      "learning_rate": 0.00015221564172180062,
      "loss": 12.2214,
      "step": 12685
    },
    {
      "epoch": 0.6908034006520866,
      "grad_norm": 0.5675514721348829,
      "learning_rate": 0.00015220812086962475,
      "loss": 12.2358,
      "step": 12686
    },
    {
      "epoch": 0.6908578546486697,
      "grad_norm": 0.5828034922884001,
      "learning_rate": 0.00015220059961146978,
      "loss": 12.2934,
      "step": 12687
    },
    {
      "epoch": 0.6909123086452527,
      "grad_norm": 0.7084947121379525,
      "learning_rate": 0.00015219307794739424,
      "loss": 12.2746,
      "step": 12688
    },
    {
      "epoch": 0.6909667626418357,
      "grad_norm": 0.6315651900269331,
      "learning_rate": 0.00015218555587745653,
      "loss": 12.1951,
      "step": 12689
    },
    {
      "epoch": 0.6910212166384186,
      "grad_norm": 0.6377830658396804,
      "learning_rate": 0.00015217803340171522,
      "loss": 12.2809,
      "step": 12690
    },
    {
      "epoch": 0.6910756706350016,
      "grad_norm": 0.641576023705419,
      "learning_rate": 0.00015217051052022877,
      "loss": 12.2517,
      "step": 12691
    },
    {
      "epoch": 0.6911301246315846,
      "grad_norm": 0.6469535141067794,
      "learning_rate": 0.00015216298723305565,
      "loss": 12.3938,
      "step": 12692
    },
    {
      "epoch": 0.6911845786281677,
      "grad_norm": 0.5669266511751133,
      "learning_rate": 0.0001521554635402544,
      "loss": 12.2574,
      "step": 12693
    },
    {
      "epoch": 0.6912390326247507,
      "grad_norm": 0.6420864976150247,
      "learning_rate": 0.00015214793944188352,
      "loss": 12.2383,
      "step": 12694
    },
    {
      "epoch": 0.6912934866213337,
      "grad_norm": 0.581599782572865,
      "learning_rate": 0.00015214041493800156,
      "loss": 12.1808,
      "step": 12695
    },
    {
      "epoch": 0.6913479406179167,
      "grad_norm": 0.5739199733755143,
      "learning_rate": 0.00015213289002866698,
      "loss": 12.2491,
      "step": 12696
    },
    {
      "epoch": 0.6914023946144997,
      "grad_norm": 0.619162803547792,
      "learning_rate": 0.00015212536471393825,
      "loss": 12.35,
      "step": 12697
    },
    {
      "epoch": 0.6914568486110827,
      "grad_norm": 0.5759863171225407,
      "learning_rate": 0.00015211783899387395,
      "loss": 12.2603,
      "step": 12698
    },
    {
      "epoch": 0.6915113026076658,
      "grad_norm": 0.6136205391072797,
      "learning_rate": 0.00015211031286853263,
      "loss": 12.4047,
      "step": 12699
    },
    {
      "epoch": 0.6915657566042488,
      "grad_norm": 0.5961564416118947,
      "learning_rate": 0.00015210278633797278,
      "loss": 12.22,
      "step": 12700
    },
    {
      "epoch": 0.6916202106008318,
      "grad_norm": 0.5430192216745389,
      "learning_rate": 0.00015209525940225292,
      "loss": 12.2475,
      "step": 12701
    },
    {
      "epoch": 0.6916746645974148,
      "grad_norm": 0.5534574797960371,
      "learning_rate": 0.00015208773206143157,
      "loss": 12.2389,
      "step": 12702
    },
    {
      "epoch": 0.6917291185939978,
      "grad_norm": 0.5240599579137478,
      "learning_rate": 0.00015208020431556725,
      "loss": 12.2449,
      "step": 12703
    },
    {
      "epoch": 0.6917835725905809,
      "grad_norm": 0.5372184729126631,
      "learning_rate": 0.0001520726761647186,
      "loss": 12.1546,
      "step": 12704
    },
    {
      "epoch": 0.6918380265871639,
      "grad_norm": 0.580728450841165,
      "learning_rate": 0.00015206514760894406,
      "loss": 12.3278,
      "step": 12705
    },
    {
      "epoch": 0.6918924805837469,
      "grad_norm": 0.6056883740245131,
      "learning_rate": 0.0001520576186483022,
      "loss": 12.2519,
      "step": 12706
    },
    {
      "epoch": 0.6919469345803299,
      "grad_norm": 0.58052840996791,
      "learning_rate": 0.00015205008928285155,
      "loss": 12.1894,
      "step": 12707
    },
    {
      "epoch": 0.6920013885769128,
      "grad_norm": 0.6161732779139788,
      "learning_rate": 0.0001520425595126507,
      "loss": 12.4466,
      "step": 12708
    },
    {
      "epoch": 0.6920558425734958,
      "grad_norm": 0.590558757040953,
      "learning_rate": 0.00015203502933775815,
      "loss": 12.2084,
      "step": 12709
    },
    {
      "epoch": 0.6921102965700789,
      "grad_norm": 0.6009086221721797,
      "learning_rate": 0.0001520274987582325,
      "loss": 12.2238,
      "step": 12710
    },
    {
      "epoch": 0.6921647505666619,
      "grad_norm": 0.5674633245904497,
      "learning_rate": 0.0001520199677741323,
      "loss": 12.2568,
      "step": 12711
    },
    {
      "epoch": 0.6922192045632449,
      "grad_norm": 0.5529595321026778,
      "learning_rate": 0.0001520124363855161,
      "loss": 12.2639,
      "step": 12712
    },
    {
      "epoch": 0.6922736585598279,
      "grad_norm": 0.6156867194157509,
      "learning_rate": 0.0001520049045924425,
      "loss": 12.4344,
      "step": 12713
    },
    {
      "epoch": 0.6923281125564109,
      "grad_norm": 0.6080044554590114,
      "learning_rate": 0.00015199737239497,
      "loss": 12.335,
      "step": 12714
    },
    {
      "epoch": 0.6923825665529939,
      "grad_norm": 0.5941132220538231,
      "learning_rate": 0.00015198983979315724,
      "loss": 12.3197,
      "step": 12715
    },
    {
      "epoch": 0.692437020549577,
      "grad_norm": 0.5653598233702917,
      "learning_rate": 0.00015198230678706276,
      "loss": 12.1885,
      "step": 12716
    },
    {
      "epoch": 0.69249147454616,
      "grad_norm": 0.5466950124603381,
      "learning_rate": 0.00015197477337674514,
      "loss": 12.3421,
      "step": 12717
    },
    {
      "epoch": 0.692545928542743,
      "grad_norm": 0.6162963914126968,
      "learning_rate": 0.000151967239562263,
      "loss": 12.3732,
      "step": 12718
    },
    {
      "epoch": 0.692600382539326,
      "grad_norm": 0.5681428110322244,
      "learning_rate": 0.00015195970534367484,
      "loss": 12.1208,
      "step": 12719
    },
    {
      "epoch": 0.692654836535909,
      "grad_norm": 0.5765561251357073,
      "learning_rate": 0.0001519521707210393,
      "loss": 12.2899,
      "step": 12720
    },
    {
      "epoch": 0.692709290532492,
      "grad_norm": 0.566818435101275,
      "learning_rate": 0.00015194463569441496,
      "loss": 12.3202,
      "step": 12721
    },
    {
      "epoch": 0.6927637445290751,
      "grad_norm": 0.6627642008500717,
      "learning_rate": 0.00015193710026386047,
      "loss": 12.3357,
      "step": 12722
    },
    {
      "epoch": 0.6928181985256581,
      "grad_norm": 0.6395608584600606,
      "learning_rate": 0.00015192956442943435,
      "loss": 12.2827,
      "step": 12723
    },
    {
      "epoch": 0.6928726525222411,
      "grad_norm": 0.5554920430630873,
      "learning_rate": 0.00015192202819119523,
      "loss": 12.2404,
      "step": 12724
    },
    {
      "epoch": 0.692927106518824,
      "grad_norm": 0.5982854590907775,
      "learning_rate": 0.00015191449154920172,
      "loss": 12.2451,
      "step": 12725
    },
    {
      "epoch": 0.692981560515407,
      "grad_norm": 0.6252918888190135,
      "learning_rate": 0.00015190695450351244,
      "loss": 12.2617,
      "step": 12726
    },
    {
      "epoch": 0.69303601451199,
      "grad_norm": 0.5792879664772468,
      "learning_rate": 0.0001518994170541859,
      "loss": 12.1567,
      "step": 12727
    },
    {
      "epoch": 0.6930904685085731,
      "grad_norm": 0.5816976069993248,
      "learning_rate": 0.00015189187920128084,
      "loss": 12.2965,
      "step": 12728
    },
    {
      "epoch": 0.6931449225051561,
      "grad_norm": 0.5743328304663563,
      "learning_rate": 0.00015188434094485584,
      "loss": 12.2071,
      "step": 12729
    },
    {
      "epoch": 0.6931993765017391,
      "grad_norm": 0.5497556272045456,
      "learning_rate": 0.00015187680228496948,
      "loss": 12.3136,
      "step": 12730
    },
    {
      "epoch": 0.6932538304983221,
      "grad_norm": 0.5326517766493589,
      "learning_rate": 0.00015186926322168043,
      "loss": 12.262,
      "step": 12731
    },
    {
      "epoch": 0.6933082844949051,
      "grad_norm": 0.6330746311860425,
      "learning_rate": 0.00015186172375504726,
      "loss": 12.3068,
      "step": 12732
    },
    {
      "epoch": 0.6933627384914881,
      "grad_norm": 0.6460808031320815,
      "learning_rate": 0.00015185418388512865,
      "loss": 12.3098,
      "step": 12733
    },
    {
      "epoch": 0.6934171924880712,
      "grad_norm": 0.5993161584869566,
      "learning_rate": 0.0001518466436119832,
      "loss": 12.2731,
      "step": 12734
    },
    {
      "epoch": 0.6934716464846542,
      "grad_norm": 0.665898943721501,
      "learning_rate": 0.00015183910293566956,
      "loss": 12.4917,
      "step": 12735
    },
    {
      "epoch": 0.6935261004812372,
      "grad_norm": 0.5514350662347167,
      "learning_rate": 0.00015183156185624635,
      "loss": 12.281,
      "step": 12736
    },
    {
      "epoch": 0.6935805544778202,
      "grad_norm": 0.5824643224684546,
      "learning_rate": 0.00015182402037377222,
      "loss": 12.2971,
      "step": 12737
    },
    {
      "epoch": 0.6936350084744032,
      "grad_norm": 0.5316514992963814,
      "learning_rate": 0.00015181647848830585,
      "loss": 12.2978,
      "step": 12738
    },
    {
      "epoch": 0.6936894624709863,
      "grad_norm": 0.627321558576572,
      "learning_rate": 0.00015180893619990581,
      "loss": 12.3271,
      "step": 12739
    },
    {
      "epoch": 0.6937439164675693,
      "grad_norm": 0.5816828595854907,
      "learning_rate": 0.00015180139350863086,
      "loss": 12.3728,
      "step": 12740
    },
    {
      "epoch": 0.6937983704641523,
      "grad_norm": 0.6047340103704789,
      "learning_rate": 0.00015179385041453952,
      "loss": 12.2363,
      "step": 12741
    },
    {
      "epoch": 0.6938528244607353,
      "grad_norm": 0.5710045945984428,
      "learning_rate": 0.00015178630691769054,
      "loss": 12.1712,
      "step": 12742
    },
    {
      "epoch": 0.6939072784573183,
      "grad_norm": 0.5864954337963909,
      "learning_rate": 0.00015177876301814255,
      "loss": 12.3223,
      "step": 12743
    },
    {
      "epoch": 0.6939617324539012,
      "grad_norm": 0.6000967347826478,
      "learning_rate": 0.0001517712187159542,
      "loss": 12.3434,
      "step": 12744
    },
    {
      "epoch": 0.6940161864504844,
      "grad_norm": 0.5906872874461015,
      "learning_rate": 0.0001517636740111842,
      "loss": 12.2595,
      "step": 12745
    },
    {
      "epoch": 0.6940706404470673,
      "grad_norm": 0.637024524133645,
      "learning_rate": 0.00015175612890389118,
      "loss": 12.4404,
      "step": 12746
    },
    {
      "epoch": 0.6941250944436503,
      "grad_norm": 0.6078531507332336,
      "learning_rate": 0.0001517485833941338,
      "loss": 12.2771,
      "step": 12747
    },
    {
      "epoch": 0.6941795484402333,
      "grad_norm": 0.5362085108950104,
      "learning_rate": 0.00015174103748197078,
      "loss": 12.3344,
      "step": 12748
    },
    {
      "epoch": 0.6942340024368163,
      "grad_norm": 0.591849383692123,
      "learning_rate": 0.0001517334911674608,
      "loss": 12.2298,
      "step": 12749
    },
    {
      "epoch": 0.6942884564333993,
      "grad_norm": 0.5655366449593927,
      "learning_rate": 0.0001517259444506625,
      "loss": 12.3227,
      "step": 12750
    },
    {
      "epoch": 0.6943429104299824,
      "grad_norm": 0.5807513533675587,
      "learning_rate": 0.00015171839733163457,
      "loss": 12.2478,
      "step": 12751
    },
    {
      "epoch": 0.6943973644265654,
      "grad_norm": 0.5495206828814007,
      "learning_rate": 0.0001517108498104357,
      "loss": 12.2208,
      "step": 12752
    },
    {
      "epoch": 0.6944518184231484,
      "grad_norm": 0.5486352146566552,
      "learning_rate": 0.00015170330188712456,
      "loss": 12.2291,
      "step": 12753
    },
    {
      "epoch": 0.6945062724197314,
      "grad_norm": 0.6028736531502734,
      "learning_rate": 0.00015169575356175992,
      "loss": 12.4123,
      "step": 12754
    },
    {
      "epoch": 0.6945607264163144,
      "grad_norm": 0.5180394854006246,
      "learning_rate": 0.00015168820483440042,
      "loss": 12.2678,
      "step": 12755
    },
    {
      "epoch": 0.6946151804128974,
      "grad_norm": 0.6338718124617676,
      "learning_rate": 0.00015168065570510478,
      "loss": 12.2655,
      "step": 12756
    },
    {
      "epoch": 0.6946696344094805,
      "grad_norm": 0.5826209963066219,
      "learning_rate": 0.0001516731061739317,
      "loss": 12.2701,
      "step": 12757
    },
    {
      "epoch": 0.6947240884060635,
      "grad_norm": 0.6083649233191714,
      "learning_rate": 0.00015166555624093986,
      "loss": 12.2946,
      "step": 12758
    },
    {
      "epoch": 0.6947785424026465,
      "grad_norm": 0.5929086951518965,
      "learning_rate": 0.000151658005906188,
      "loss": 12.2975,
      "step": 12759
    },
    {
      "epoch": 0.6948329963992295,
      "grad_norm": 0.6159253227796201,
      "learning_rate": 0.00015165045516973483,
      "loss": 12.4129,
      "step": 12760
    },
    {
      "epoch": 0.6948874503958125,
      "grad_norm": 0.5961923098818454,
      "learning_rate": 0.00015164290403163905,
      "loss": 12.2542,
      "step": 12761
    },
    {
      "epoch": 0.6949419043923954,
      "grad_norm": 0.6082280448345292,
      "learning_rate": 0.0001516353524919594,
      "loss": 12.3274,
      "step": 12762
    },
    {
      "epoch": 0.6949963583889786,
      "grad_norm": 0.5843565354596677,
      "learning_rate": 0.00015162780055075456,
      "loss": 12.3214,
      "step": 12763
    },
    {
      "epoch": 0.6950508123855615,
      "grad_norm": 0.5922646499860922,
      "learning_rate": 0.0001516202482080833,
      "loss": 12.2993,
      "step": 12764
    },
    {
      "epoch": 0.6951052663821445,
      "grad_norm": 0.5319499220696793,
      "learning_rate": 0.00015161269546400434,
      "loss": 12.2324,
      "step": 12765
    },
    {
      "epoch": 0.6951597203787275,
      "grad_norm": 0.7045455639918438,
      "learning_rate": 0.00015160514231857643,
      "loss": 12.2201,
      "step": 12766
    },
    {
      "epoch": 0.6952141743753105,
      "grad_norm": 0.5619467296914059,
      "learning_rate": 0.00015159758877185823,
      "loss": 12.2243,
      "step": 12767
    },
    {
      "epoch": 0.6952686283718935,
      "grad_norm": 0.5465287615022251,
      "learning_rate": 0.00015159003482390857,
      "loss": 12.3618,
      "step": 12768
    },
    {
      "epoch": 0.6953230823684766,
      "grad_norm": 0.5405474447716598,
      "learning_rate": 0.0001515824804747861,
      "loss": 12.1688,
      "step": 12769
    },
    {
      "epoch": 0.6953775363650596,
      "grad_norm": 0.617095133941277,
      "learning_rate": 0.00015157492572454964,
      "loss": 12.2569,
      "step": 12770
    },
    {
      "epoch": 0.6954319903616426,
      "grad_norm": 0.5594754357401277,
      "learning_rate": 0.00015156737057325792,
      "loss": 12.1694,
      "step": 12771
    },
    {
      "epoch": 0.6954864443582256,
      "grad_norm": 0.564470075597679,
      "learning_rate": 0.00015155981502096966,
      "loss": 12.1877,
      "step": 12772
    },
    {
      "epoch": 0.6955408983548086,
      "grad_norm": 0.5394155263359696,
      "learning_rate": 0.00015155225906774363,
      "loss": 12.3133,
      "step": 12773
    },
    {
      "epoch": 0.6955953523513917,
      "grad_norm": 0.5964767067731667,
      "learning_rate": 0.00015154470271363862,
      "loss": 12.2444,
      "step": 12774
    },
    {
      "epoch": 0.6956498063479747,
      "grad_norm": 0.5258330599265519,
      "learning_rate": 0.00015153714595871328,
      "loss": 12.3298,
      "step": 12775
    },
    {
      "epoch": 0.6957042603445577,
      "grad_norm": 0.515769601343884,
      "learning_rate": 0.00015152958880302654,
      "loss": 12.1901,
      "step": 12776
    },
    {
      "epoch": 0.6957587143411407,
      "grad_norm": 0.5625539317088408,
      "learning_rate": 0.00015152203124663705,
      "loss": 12.1423,
      "step": 12777
    },
    {
      "epoch": 0.6958131683377237,
      "grad_norm": 0.5738176088998937,
      "learning_rate": 0.00015151447328960358,
      "loss": 12.3304,
      "step": 12778
    },
    {
      "epoch": 0.6958676223343067,
      "grad_norm": 0.5337821156892026,
      "learning_rate": 0.00015150691493198495,
      "loss": 12.158,
      "step": 12779
    },
    {
      "epoch": 0.6959220763308898,
      "grad_norm": 0.5572367138584677,
      "learning_rate": 0.00015149935617383986,
      "loss": 12.3404,
      "step": 12780
    },
    {
      "epoch": 0.6959765303274728,
      "grad_norm": 0.5785620836035866,
      "learning_rate": 0.0001514917970152272,
      "loss": 12.2895,
      "step": 12781
    },
    {
      "epoch": 0.6960309843240557,
      "grad_norm": 0.5567436442972655,
      "learning_rate": 0.00015148423745620567,
      "loss": 12.2598,
      "step": 12782
    },
    {
      "epoch": 0.6960854383206387,
      "grad_norm": 0.5436685548209514,
      "learning_rate": 0.0001514766774968341,
      "loss": 12.2851,
      "step": 12783
    },
    {
      "epoch": 0.6961398923172217,
      "grad_norm": 0.5620871243842307,
      "learning_rate": 0.00015146911713717122,
      "loss": 12.2629,
      "step": 12784
    },
    {
      "epoch": 0.6961943463138047,
      "grad_norm": 0.5825963197643459,
      "learning_rate": 0.00015146155637727588,
      "loss": 12.2778,
      "step": 12785
    },
    {
      "epoch": 0.6962488003103878,
      "grad_norm": 0.580599403779096,
      "learning_rate": 0.0001514539952172068,
      "loss": 12.293,
      "step": 12786
    },
    {
      "epoch": 0.6963032543069708,
      "grad_norm": 0.5547121852362786,
      "learning_rate": 0.0001514464336570229,
      "loss": 12.1824,
      "step": 12787
    },
    {
      "epoch": 0.6963577083035538,
      "grad_norm": 0.5892606362080496,
      "learning_rate": 0.00015143887169678284,
      "loss": 12.1196,
      "step": 12788
    },
    {
      "epoch": 0.6964121623001368,
      "grad_norm": 0.6486531519357543,
      "learning_rate": 0.0001514313093365455,
      "loss": 12.3475,
      "step": 12789
    },
    {
      "epoch": 0.6964666162967198,
      "grad_norm": 0.5804390280894669,
      "learning_rate": 0.00015142374657636969,
      "loss": 12.3115,
      "step": 12790
    },
    {
      "epoch": 0.6965210702933028,
      "grad_norm": 0.6050143210622061,
      "learning_rate": 0.00015141618341631418,
      "loss": 12.3309,
      "step": 12791
    },
    {
      "epoch": 0.6965755242898859,
      "grad_norm": 0.6440660563413831,
      "learning_rate": 0.0001514086198564378,
      "loss": 12.2225,
      "step": 12792
    },
    {
      "epoch": 0.6966299782864689,
      "grad_norm": 0.6035204658067048,
      "learning_rate": 0.00015140105589679943,
      "loss": 12.3462,
      "step": 12793
    },
    {
      "epoch": 0.6966844322830519,
      "grad_norm": 0.586423227975999,
      "learning_rate": 0.00015139349153745778,
      "loss": 12.1881,
      "step": 12794
    },
    {
      "epoch": 0.6967388862796349,
      "grad_norm": 0.6195614052981723,
      "learning_rate": 0.00015138592677847173,
      "loss": 12.3054,
      "step": 12795
    },
    {
      "epoch": 0.6967933402762179,
      "grad_norm": 0.5695957080876042,
      "learning_rate": 0.0001513783616199001,
      "loss": 12.2431,
      "step": 12796
    },
    {
      "epoch": 0.6968477942728009,
      "grad_norm": 0.5401987111839878,
      "learning_rate": 0.00015137079606180172,
      "loss": 12.2292,
      "step": 12797
    },
    {
      "epoch": 0.696902248269384,
      "grad_norm": 0.644588419023635,
      "learning_rate": 0.0001513632301042354,
      "loss": 12.2898,
      "step": 12798
    },
    {
      "epoch": 0.696956702265967,
      "grad_norm": 0.558414091002706,
      "learning_rate": 0.00015135566374726,
      "loss": 12.2123,
      "step": 12799
    },
    {
      "epoch": 0.69701115626255,
      "grad_norm": 0.5782843434074975,
      "learning_rate": 0.00015134809699093434,
      "loss": 12.3783,
      "step": 12800
    },
    {
      "epoch": 0.6970656102591329,
      "grad_norm": 0.6892559074368685,
      "learning_rate": 0.00015134052983531725,
      "loss": 12.3138,
      "step": 12801
    },
    {
      "epoch": 0.6971200642557159,
      "grad_norm": 0.596134638203521,
      "learning_rate": 0.00015133296228046764,
      "loss": 12.2218,
      "step": 12802
    },
    {
      "epoch": 0.6971745182522989,
      "grad_norm": 0.6657524565185177,
      "learning_rate": 0.00015132539432644425,
      "loss": 12.2724,
      "step": 12803
    },
    {
      "epoch": 0.697228972248882,
      "grad_norm": 0.6904308970357311,
      "learning_rate": 0.00015131782597330602,
      "loss": 12.3158,
      "step": 12804
    },
    {
      "epoch": 0.697283426245465,
      "grad_norm": 0.5646015845227467,
      "learning_rate": 0.00015131025722111177,
      "loss": 12.2422,
      "step": 12805
    },
    {
      "epoch": 0.697337880242048,
      "grad_norm": 0.6003823456063323,
      "learning_rate": 0.00015130268806992037,
      "loss": 12.4381,
      "step": 12806
    },
    {
      "epoch": 0.697392334238631,
      "grad_norm": 0.6052761761039406,
      "learning_rate": 0.0001512951185197906,
      "loss": 12.4855,
      "step": 12807
    },
    {
      "epoch": 0.697446788235214,
      "grad_norm": 0.5215938576523123,
      "learning_rate": 0.00015128754857078141,
      "loss": 12.1122,
      "step": 12808
    },
    {
      "epoch": 0.6975012422317971,
      "grad_norm": 0.5757052319756734,
      "learning_rate": 0.00015127997822295168,
      "loss": 12.3324,
      "step": 12809
    },
    {
      "epoch": 0.6975556962283801,
      "grad_norm": 0.5407436135189331,
      "learning_rate": 0.00015127240747636023,
      "loss": 12.2614,
      "step": 12810
    },
    {
      "epoch": 0.6976101502249631,
      "grad_norm": 0.563649288491365,
      "learning_rate": 0.00015126483633106595,
      "loss": 12.2358,
      "step": 12811
    },
    {
      "epoch": 0.6976646042215461,
      "grad_norm": 0.576318626856709,
      "learning_rate": 0.00015125726478712765,
      "loss": 12.3351,
      "step": 12812
    },
    {
      "epoch": 0.6977190582181291,
      "grad_norm": 0.6966281930802263,
      "learning_rate": 0.00015124969284460428,
      "loss": 12.4885,
      "step": 12813
    },
    {
      "epoch": 0.6977735122147121,
      "grad_norm": 0.7809185908219524,
      "learning_rate": 0.00015124212050355475,
      "loss": 12.3636,
      "step": 12814
    },
    {
      "epoch": 0.6978279662112952,
      "grad_norm": 0.5746577283165907,
      "learning_rate": 0.00015123454776403786,
      "loss": 12.2467,
      "step": 12815
    },
    {
      "epoch": 0.6978824202078782,
      "grad_norm": 0.6024413136284706,
      "learning_rate": 0.00015122697462611258,
      "loss": 12.2904,
      "step": 12816
    },
    {
      "epoch": 0.6979368742044612,
      "grad_norm": 0.5206933345152238,
      "learning_rate": 0.0001512194010898377,
      "loss": 12.248,
      "step": 12817
    },
    {
      "epoch": 0.6979913282010441,
      "grad_norm": 0.5610679148396018,
      "learning_rate": 0.00015121182715527217,
      "loss": 12.3231,
      "step": 12818
    },
    {
      "epoch": 0.6980457821976271,
      "grad_norm": 0.5704520611048067,
      "learning_rate": 0.0001512042528224749,
      "loss": 12.2603,
      "step": 12819
    },
    {
      "epoch": 0.6981002361942101,
      "grad_norm": 0.5341114754262499,
      "learning_rate": 0.00015119667809150475,
      "loss": 12.2551,
      "step": 12820
    },
    {
      "epoch": 0.6981546901907932,
      "grad_norm": 0.5651258746143369,
      "learning_rate": 0.0001511891029624207,
      "loss": 12.3242,
      "step": 12821
    },
    {
      "epoch": 0.6982091441873762,
      "grad_norm": 0.6127180680960584,
      "learning_rate": 0.00015118152743528158,
      "loss": 12.2176,
      "step": 12822
    },
    {
      "epoch": 0.6982635981839592,
      "grad_norm": 0.6178526829964687,
      "learning_rate": 0.0001511739515101463,
      "loss": 12.3267,
      "step": 12823
    },
    {
      "epoch": 0.6983180521805422,
      "grad_norm": 0.5485086221160265,
      "learning_rate": 0.00015116637518707376,
      "loss": 12.2266,
      "step": 12824
    },
    {
      "epoch": 0.6983725061771252,
      "grad_norm": 0.5591931811739885,
      "learning_rate": 0.00015115879846612295,
      "loss": 12.1893,
      "step": 12825
    },
    {
      "epoch": 0.6984269601737082,
      "grad_norm": 0.5504356367908919,
      "learning_rate": 0.00015115122134735278,
      "loss": 12.1897,
      "step": 12826
    },
    {
      "epoch": 0.6984814141702913,
      "grad_norm": 0.5791906704907291,
      "learning_rate": 0.00015114364383082208,
      "loss": 12.3134,
      "step": 12827
    },
    {
      "epoch": 0.6985358681668743,
      "grad_norm": 0.5927965701697596,
      "learning_rate": 0.00015113606591658985,
      "loss": 12.2676,
      "step": 12828
    },
    {
      "epoch": 0.6985903221634573,
      "grad_norm": 0.6514547782037758,
      "learning_rate": 0.00015112848760471498,
      "loss": 12.2984,
      "step": 12829
    },
    {
      "epoch": 0.6986447761600403,
      "grad_norm": 0.5625665172333867,
      "learning_rate": 0.0001511209088952564,
      "loss": 12.2379,
      "step": 12830
    },
    {
      "epoch": 0.6986992301566233,
      "grad_norm": 0.6080219521627901,
      "learning_rate": 0.0001511133297882731,
      "loss": 12.3662,
      "step": 12831
    },
    {
      "epoch": 0.6987536841532063,
      "grad_norm": 0.6208368036638605,
      "learning_rate": 0.00015110575028382396,
      "loss": 12.2077,
      "step": 12832
    },
    {
      "epoch": 0.6988081381497894,
      "grad_norm": 0.5981735476751217,
      "learning_rate": 0.00015109817038196796,
      "loss": 12.3023,
      "step": 12833
    },
    {
      "epoch": 0.6988625921463724,
      "grad_norm": 0.6878859063713462,
      "learning_rate": 0.00015109059008276397,
      "loss": 12.3612,
      "step": 12834
    },
    {
      "epoch": 0.6989170461429554,
      "grad_norm": 0.6536308103657945,
      "learning_rate": 0.000151083009386271,
      "loss": 12.3678,
      "step": 12835
    },
    {
      "epoch": 0.6989715001395383,
      "grad_norm": 0.5334163204272254,
      "learning_rate": 0.00015107542829254802,
      "loss": 12.2648,
      "step": 12836
    },
    {
      "epoch": 0.6990259541361213,
      "grad_norm": 0.5811757267663836,
      "learning_rate": 0.0001510678468016539,
      "loss": 12.2406,
      "step": 12837
    },
    {
      "epoch": 0.6990804081327044,
      "grad_norm": 0.5791601339932474,
      "learning_rate": 0.00015106026491364767,
      "loss": 12.173,
      "step": 12838
    },
    {
      "epoch": 0.6991348621292874,
      "grad_norm": 0.5577669427986419,
      "learning_rate": 0.00015105268262858825,
      "loss": 12.3816,
      "step": 12839
    },
    {
      "epoch": 0.6991893161258704,
      "grad_norm": 0.5523126670538845,
      "learning_rate": 0.00015104509994653457,
      "loss": 12.2392,
      "step": 12840
    },
    {
      "epoch": 0.6992437701224534,
      "grad_norm": 0.5755533090316269,
      "learning_rate": 0.00015103751686754567,
      "loss": 12.2055,
      "step": 12841
    },
    {
      "epoch": 0.6992982241190364,
      "grad_norm": 0.5715341861488942,
      "learning_rate": 0.00015102993339168047,
      "loss": 12.2667,
      "step": 12842
    },
    {
      "epoch": 0.6993526781156194,
      "grad_norm": 0.6337431022952587,
      "learning_rate": 0.00015102234951899795,
      "loss": 12.2897,
      "step": 12843
    },
    {
      "epoch": 0.6994071321122025,
      "grad_norm": 0.5545934302937681,
      "learning_rate": 0.0001510147652495571,
      "loss": 12.208,
      "step": 12844
    },
    {
      "epoch": 0.6994615861087855,
      "grad_norm": 0.5818395953361071,
      "learning_rate": 0.00015100718058341686,
      "loss": 12.3168,
      "step": 12845
    },
    {
      "epoch": 0.6995160401053685,
      "grad_norm": 0.54488203576456,
      "learning_rate": 0.00015099959552063622,
      "loss": 12.2805,
      "step": 12846
    },
    {
      "epoch": 0.6995704941019515,
      "grad_norm": 0.5819073442016384,
      "learning_rate": 0.00015099201006127418,
      "loss": 12.2357,
      "step": 12847
    },
    {
      "epoch": 0.6996249480985345,
      "grad_norm": 0.5980085895394841,
      "learning_rate": 0.00015098442420538973,
      "loss": 12.1713,
      "step": 12848
    },
    {
      "epoch": 0.6996794020951175,
      "grad_norm": 0.5649860888099074,
      "learning_rate": 0.00015097683795304185,
      "loss": 12.2638,
      "step": 12849
    },
    {
      "epoch": 0.6997338560917006,
      "grad_norm": 0.5692873645192698,
      "learning_rate": 0.0001509692513042895,
      "loss": 12.2009,
      "step": 12850
    },
    {
      "epoch": 0.6997883100882836,
      "grad_norm": 0.5254540513369161,
      "learning_rate": 0.00015096166425919175,
      "loss": 12.3878,
      "step": 12851
    },
    {
      "epoch": 0.6998427640848666,
      "grad_norm": 0.559223478872943,
      "learning_rate": 0.00015095407681780753,
      "loss": 12.1803,
      "step": 12852
    },
    {
      "epoch": 0.6998972180814496,
      "grad_norm": 0.5561422820246377,
      "learning_rate": 0.00015094648898019588,
      "loss": 12.2901,
      "step": 12853
    },
    {
      "epoch": 0.6999516720780325,
      "grad_norm": 0.6398204891444215,
      "learning_rate": 0.00015093890074641575,
      "loss": 12.2311,
      "step": 12854
    },
    {
      "epoch": 0.7000061260746155,
      "grad_norm": 0.5730771643651186,
      "learning_rate": 0.0001509313121165262,
      "loss": 12.2579,
      "step": 12855
    },
    {
      "epoch": 0.7000605800711986,
      "grad_norm": 0.5339535864535195,
      "learning_rate": 0.00015092372309058623,
      "loss": 12.1686,
      "step": 12856
    },
    {
      "epoch": 0.7001150340677816,
      "grad_norm": 0.5994247273085027,
      "learning_rate": 0.00015091613366865486,
      "loss": 12.2944,
      "step": 12857
    },
    {
      "epoch": 0.7001694880643646,
      "grad_norm": 0.5083786193557129,
      "learning_rate": 0.0001509085438507911,
      "loss": 12.1098,
      "step": 12858
    },
    {
      "epoch": 0.7002239420609476,
      "grad_norm": 0.5813561783614933,
      "learning_rate": 0.00015090095363705394,
      "loss": 12.3114,
      "step": 12859
    },
    {
      "epoch": 0.7002783960575306,
      "grad_norm": 0.5256842061518319,
      "learning_rate": 0.00015089336302750246,
      "loss": 12.2911,
      "step": 12860
    },
    {
      "epoch": 0.7003328500541136,
      "grad_norm": 0.5854745903233185,
      "learning_rate": 0.00015088577202219562,
      "loss": 12.3942,
      "step": 12861
    },
    {
      "epoch": 0.7003873040506967,
      "grad_norm": 0.6232949445893607,
      "learning_rate": 0.0001508781806211925,
      "loss": 12.3221,
      "step": 12862
    },
    {
      "epoch": 0.7004417580472797,
      "grad_norm": 0.556903770839926,
      "learning_rate": 0.00015087058882455213,
      "loss": 12.3262,
      "step": 12863
    },
    {
      "epoch": 0.7004962120438627,
      "grad_norm": 0.5511222088378669,
      "learning_rate": 0.0001508629966323335,
      "loss": 12.2111,
      "step": 12864
    },
    {
      "epoch": 0.7005506660404457,
      "grad_norm": 0.6933096411771393,
      "learning_rate": 0.0001508554040445957,
      "loss": 12.3443,
      "step": 12865
    },
    {
      "epoch": 0.7006051200370287,
      "grad_norm": 0.6450763766755955,
      "learning_rate": 0.00015084781106139777,
      "loss": 12.2932,
      "step": 12866
    },
    {
      "epoch": 0.7006595740336117,
      "grad_norm": 0.6180854258717581,
      "learning_rate": 0.00015084021768279867,
      "loss": 12.326,
      "step": 12867
    },
    {
      "epoch": 0.7007140280301948,
      "grad_norm": 0.5652729378781475,
      "learning_rate": 0.00015083262390885754,
      "loss": 12.1985,
      "step": 12868
    },
    {
      "epoch": 0.7007684820267778,
      "grad_norm": 0.5832845128774566,
      "learning_rate": 0.0001508250297396334,
      "loss": 12.3935,
      "step": 12869
    },
    {
      "epoch": 0.7008229360233608,
      "grad_norm": 0.6256855422987257,
      "learning_rate": 0.00015081743517518533,
      "loss": 12.3105,
      "step": 12870
    },
    {
      "epoch": 0.7008773900199438,
      "grad_norm": 0.6514861781739107,
      "learning_rate": 0.0001508098402155724,
      "loss": 12.2294,
      "step": 12871
    },
    {
      "epoch": 0.7009318440165268,
      "grad_norm": 0.5976054048957782,
      "learning_rate": 0.00015080224486085355,
      "loss": 12.2549,
      "step": 12872
    },
    {
      "epoch": 0.7009862980131099,
      "grad_norm": 0.6126191865281446,
      "learning_rate": 0.00015079464911108793,
      "loss": 12.2187,
      "step": 12873
    },
    {
      "epoch": 0.7010407520096928,
      "grad_norm": 0.5903689854623854,
      "learning_rate": 0.00015078705296633462,
      "loss": 12.2371,
      "step": 12874
    },
    {
      "epoch": 0.7010952060062758,
      "grad_norm": 0.5812216652704836,
      "learning_rate": 0.00015077945642665269,
      "loss": 12.369,
      "step": 12875
    },
    {
      "epoch": 0.7011496600028588,
      "grad_norm": 0.5534852216522496,
      "learning_rate": 0.00015077185949210117,
      "loss": 12.2853,
      "step": 12876
    },
    {
      "epoch": 0.7012041139994418,
      "grad_norm": 0.6274739036507253,
      "learning_rate": 0.00015076426216273914,
      "loss": 12.1898,
      "step": 12877
    },
    {
      "epoch": 0.7012585679960248,
      "grad_norm": 0.6063997065462186,
      "learning_rate": 0.00015075666443862572,
      "loss": 12.3014,
      "step": 12878
    },
    {
      "epoch": 0.7013130219926079,
      "grad_norm": 0.5217788404855932,
      "learning_rate": 0.00015074906631981999,
      "loss": 12.2941,
      "step": 12879
    },
    {
      "epoch": 0.7013674759891909,
      "grad_norm": 0.5552649160050687,
      "learning_rate": 0.00015074146780638095,
      "loss": 12.1025,
      "step": 12880
    },
    {
      "epoch": 0.7014219299857739,
      "grad_norm": 0.6392569035584885,
      "learning_rate": 0.00015073386889836782,
      "loss": 12.1831,
      "step": 12881
    },
    {
      "epoch": 0.7014763839823569,
      "grad_norm": 0.5521041501633279,
      "learning_rate": 0.00015072626959583956,
      "loss": 12.2143,
      "step": 12882
    },
    {
      "epoch": 0.7015308379789399,
      "grad_norm": 0.6227440240483435,
      "learning_rate": 0.00015071866989885532,
      "loss": 12.3998,
      "step": 12883
    },
    {
      "epoch": 0.7015852919755229,
      "grad_norm": 0.6334247030673338,
      "learning_rate": 0.00015071106980747421,
      "loss": 12.3131,
      "step": 12884
    },
    {
      "epoch": 0.701639745972106,
      "grad_norm": 0.5563039830909307,
      "learning_rate": 0.0001507034693217553,
      "loss": 12.2404,
      "step": 12885
    },
    {
      "epoch": 0.701694199968689,
      "grad_norm": 0.5740373990473525,
      "learning_rate": 0.00015069586844175775,
      "loss": 12.1678,
      "step": 12886
    },
    {
      "epoch": 0.701748653965272,
      "grad_norm": 0.5615508506413522,
      "learning_rate": 0.00015068826716754062,
      "loss": 12.2148,
      "step": 12887
    },
    {
      "epoch": 0.701803107961855,
      "grad_norm": 0.5881964519645851,
      "learning_rate": 0.000150680665499163,
      "loss": 12.3873,
      "step": 12888
    },
    {
      "epoch": 0.701857561958438,
      "grad_norm": 0.6559325462620302,
      "learning_rate": 0.000150673063436684,
      "loss": 12.2265,
      "step": 12889
    },
    {
      "epoch": 0.701912015955021,
      "grad_norm": 0.5351451244591657,
      "learning_rate": 0.0001506654609801628,
      "loss": 12.2559,
      "step": 12890
    },
    {
      "epoch": 0.701966469951604,
      "grad_norm": 0.5569533211512768,
      "learning_rate": 0.00015065785812965848,
      "loss": 12.0566,
      "step": 12891
    },
    {
      "epoch": 0.702020923948187,
      "grad_norm": 0.5937804356031744,
      "learning_rate": 0.00015065025488523017,
      "loss": 12.2113,
      "step": 12892
    },
    {
      "epoch": 0.70207537794477,
      "grad_norm": 0.5659010125123903,
      "learning_rate": 0.00015064265124693698,
      "loss": 12.278,
      "step": 12893
    },
    {
      "epoch": 0.702129831941353,
      "grad_norm": 0.535901319384464,
      "learning_rate": 0.00015063504721483803,
      "loss": 12.3676,
      "step": 12894
    },
    {
      "epoch": 0.702184285937936,
      "grad_norm": 0.5669487525365963,
      "learning_rate": 0.00015062744278899242,
      "loss": 12.3111,
      "step": 12895
    },
    {
      "epoch": 0.702238739934519,
      "grad_norm": 0.6140266146940161,
      "learning_rate": 0.0001506198379694594,
      "loss": 12.276,
      "step": 12896
    },
    {
      "epoch": 0.7022931939311021,
      "grad_norm": 0.5226019933935814,
      "learning_rate": 0.000150612232756298,
      "loss": 12.2468,
      "step": 12897
    },
    {
      "epoch": 0.7023476479276851,
      "grad_norm": 0.5518765132762485,
      "learning_rate": 0.00015060462714956737,
      "loss": 12.3558,
      "step": 12898
    },
    {
      "epoch": 0.7024021019242681,
      "grad_norm": 0.5930158592363907,
      "learning_rate": 0.0001505970211493267,
      "loss": 12.2942,
      "step": 12899
    },
    {
      "epoch": 0.7024565559208511,
      "grad_norm": 0.5990262226635165,
      "learning_rate": 0.00015058941475563507,
      "loss": 12.4069,
      "step": 12900
    },
    {
      "epoch": 0.7025110099174341,
      "grad_norm": 0.6295009447765174,
      "learning_rate": 0.0001505818079685517,
      "loss": 12.2482,
      "step": 12901
    },
    {
      "epoch": 0.7025654639140171,
      "grad_norm": 0.5713781944258155,
      "learning_rate": 0.0001505742007881357,
      "loss": 12.272,
      "step": 12902
    },
    {
      "epoch": 0.7026199179106002,
      "grad_norm": 0.5775952506978804,
      "learning_rate": 0.00015056659321444627,
      "loss": 12.198,
      "step": 12903
    },
    {
      "epoch": 0.7026743719071832,
      "grad_norm": 0.5989314705696425,
      "learning_rate": 0.00015055898524754245,
      "loss": 12.3093,
      "step": 12904
    },
    {
      "epoch": 0.7027288259037662,
      "grad_norm": 0.5691834182800973,
      "learning_rate": 0.00015055137688748356,
      "loss": 12.2635,
      "step": 12905
    },
    {
      "epoch": 0.7027832799003492,
      "grad_norm": 0.5813875692402253,
      "learning_rate": 0.00015054376813432866,
      "loss": 12.2248,
      "step": 12906
    },
    {
      "epoch": 0.7028377338969322,
      "grad_norm": 0.5870360426395933,
      "learning_rate": 0.00015053615898813694,
      "loss": 12.2217,
      "step": 12907
    },
    {
      "epoch": 0.7028921878935153,
      "grad_norm": 0.5343732930374057,
      "learning_rate": 0.00015052854944896756,
      "loss": 12.3343,
      "step": 12908
    },
    {
      "epoch": 0.7029466418900983,
      "grad_norm": 0.5699443719681166,
      "learning_rate": 0.0001505209395168797,
      "loss": 12.223,
      "step": 12909
    },
    {
      "epoch": 0.7030010958866812,
      "grad_norm": 0.5973307979328459,
      "learning_rate": 0.00015051332919193257,
      "loss": 12.2756,
      "step": 12910
    },
    {
      "epoch": 0.7030555498832642,
      "grad_norm": 0.6120749129557116,
      "learning_rate": 0.0001505057184741853,
      "loss": 12.3103,
      "step": 12911
    },
    {
      "epoch": 0.7031100038798472,
      "grad_norm": 0.6953633712044719,
      "learning_rate": 0.0001504981073636971,
      "loss": 12.3662,
      "step": 12912
    },
    {
      "epoch": 0.7031644578764302,
      "grad_norm": 0.5138596022468547,
      "learning_rate": 0.00015049049586052718,
      "loss": 12.253,
      "step": 12913
    },
    {
      "epoch": 0.7032189118730133,
      "grad_norm": 0.6047293103686023,
      "learning_rate": 0.00015048288396473468,
      "loss": 12.4362,
      "step": 12914
    },
    {
      "epoch": 0.7032733658695963,
      "grad_norm": 0.5717537391124353,
      "learning_rate": 0.00015047527167637878,
      "loss": 12.1667,
      "step": 12915
    },
    {
      "epoch": 0.7033278198661793,
      "grad_norm": 0.5407187604839495,
      "learning_rate": 0.0001504676589955187,
      "loss": 12.1886,
      "step": 12916
    },
    {
      "epoch": 0.7033822738627623,
      "grad_norm": 0.6358353512639061,
      "learning_rate": 0.00015046004592221367,
      "loss": 12.2309,
      "step": 12917
    },
    {
      "epoch": 0.7034367278593453,
      "grad_norm": 0.5882551571751676,
      "learning_rate": 0.00015045243245652285,
      "loss": 12.3412,
      "step": 12918
    },
    {
      "epoch": 0.7034911818559283,
      "grad_norm": 0.5664833841054566,
      "learning_rate": 0.00015044481859850545,
      "loss": 12.3031,
      "step": 12919
    },
    {
      "epoch": 0.7035456358525114,
      "grad_norm": 0.6640648902820085,
      "learning_rate": 0.00015043720434822067,
      "loss": 12.1166,
      "step": 12920
    },
    {
      "epoch": 0.7036000898490944,
      "grad_norm": 0.5332762140500776,
      "learning_rate": 0.0001504295897057278,
      "loss": 12.213,
      "step": 12921
    },
    {
      "epoch": 0.7036545438456774,
      "grad_norm": 0.6033862431121884,
      "learning_rate": 0.0001504219746710859,
      "loss": 12.1348,
      "step": 12922
    },
    {
      "epoch": 0.7037089978422604,
      "grad_norm": 0.6008975227578273,
      "learning_rate": 0.0001504143592443543,
      "loss": 12.2651,
      "step": 12923
    },
    {
      "epoch": 0.7037634518388434,
      "grad_norm": 0.5713175980307648,
      "learning_rate": 0.00015040674342559217,
      "loss": 12.1195,
      "step": 12924
    },
    {
      "epoch": 0.7038179058354264,
      "grad_norm": 0.6870336688388138,
      "learning_rate": 0.00015039912721485877,
      "loss": 12.2627,
      "step": 12925
    },
    {
      "epoch": 0.7038723598320095,
      "grad_norm": 0.550509285896728,
      "learning_rate": 0.0001503915106122133,
      "loss": 12.2699,
      "step": 12926
    },
    {
      "epoch": 0.7039268138285925,
      "grad_norm": 0.5183542937646016,
      "learning_rate": 0.000150383893617715,
      "loss": 12.1956,
      "step": 12927
    },
    {
      "epoch": 0.7039812678251754,
      "grad_norm": 0.505524050927356,
      "learning_rate": 0.0001503762762314231,
      "loss": 12.2436,
      "step": 12928
    },
    {
      "epoch": 0.7040357218217584,
      "grad_norm": 0.5528750624664851,
      "learning_rate": 0.0001503686584533968,
      "loss": 12.2574,
      "step": 12929
    },
    {
      "epoch": 0.7040901758183414,
      "grad_norm": 0.6847928417111844,
      "learning_rate": 0.00015036104028369538,
      "loss": 12.2566,
      "step": 12930
    },
    {
      "epoch": 0.7041446298149244,
      "grad_norm": 0.5811133448734164,
      "learning_rate": 0.00015035342172237806,
      "loss": 12.193,
      "step": 12931
    },
    {
      "epoch": 0.7041990838115075,
      "grad_norm": 0.5146003924536093,
      "learning_rate": 0.0001503458027695041,
      "loss": 12.2794,
      "step": 12932
    },
    {
      "epoch": 0.7042535378080905,
      "grad_norm": 0.6257737359358769,
      "learning_rate": 0.00015033818342513272,
      "loss": 12.2411,
      "step": 12933
    },
    {
      "epoch": 0.7043079918046735,
      "grad_norm": 0.5742448206754013,
      "learning_rate": 0.00015033056368932322,
      "loss": 12.3275,
      "step": 12934
    },
    {
      "epoch": 0.7043624458012565,
      "grad_norm": 0.5797591318656884,
      "learning_rate": 0.00015032294356213477,
      "loss": 12.1942,
      "step": 12935
    },
    {
      "epoch": 0.7044168997978395,
      "grad_norm": 0.7451421687573521,
      "learning_rate": 0.0001503153230436267,
      "loss": 12.226,
      "step": 12936
    },
    {
      "epoch": 0.7044713537944225,
      "grad_norm": 0.539841668324749,
      "learning_rate": 0.00015030770213385824,
      "loss": 12.272,
      "step": 12937
    },
    {
      "epoch": 0.7045258077910056,
      "grad_norm": 0.5935585281767789,
      "learning_rate": 0.00015030008083288864,
      "loss": 12.3097,
      "step": 12938
    },
    {
      "epoch": 0.7045802617875886,
      "grad_norm": 0.5511736584624088,
      "learning_rate": 0.0001502924591407772,
      "loss": 12.2108,
      "step": 12939
    },
    {
      "epoch": 0.7046347157841716,
      "grad_norm": 0.5534412404344856,
      "learning_rate": 0.00015028483705758312,
      "loss": 12.2039,
      "step": 12940
    },
    {
      "epoch": 0.7046891697807546,
      "grad_norm": 0.6563150821077004,
      "learning_rate": 0.00015027721458336574,
      "loss": 12.2303,
      "step": 12941
    },
    {
      "epoch": 0.7047436237773376,
      "grad_norm": 0.6366833265609215,
      "learning_rate": 0.00015026959171818432,
      "loss": 12.4365,
      "step": 12942
    },
    {
      "epoch": 0.7047980777739207,
      "grad_norm": 0.6358540136371424,
      "learning_rate": 0.0001502619684620981,
      "loss": 12.3262,
      "step": 12943
    },
    {
      "epoch": 0.7048525317705037,
      "grad_norm": 0.6537474371778512,
      "learning_rate": 0.00015025434481516638,
      "loss": 12.3111,
      "step": 12944
    },
    {
      "epoch": 0.7049069857670867,
      "grad_norm": 0.5743493906745762,
      "learning_rate": 0.00015024672077744845,
      "loss": 12.2707,
      "step": 12945
    },
    {
      "epoch": 0.7049614397636697,
      "grad_norm": 0.5473693388988641,
      "learning_rate": 0.00015023909634900363,
      "loss": 12.2449,
      "step": 12946
    },
    {
      "epoch": 0.7050158937602526,
      "grad_norm": 0.6103975608154453,
      "learning_rate": 0.00015023147152989113,
      "loss": 12.3103,
      "step": 12947
    },
    {
      "epoch": 0.7050703477568356,
      "grad_norm": 0.5814312456830363,
      "learning_rate": 0.00015022384632017033,
      "loss": 12.2969,
      "step": 12948
    },
    {
      "epoch": 0.7051248017534187,
      "grad_norm": 0.5968487505951352,
      "learning_rate": 0.00015021622071990045,
      "loss": 12.2258,
      "step": 12949
    },
    {
      "epoch": 0.7051792557500017,
      "grad_norm": 0.5681675401207187,
      "learning_rate": 0.00015020859472914078,
      "loss": 12.083,
      "step": 12950
    },
    {
      "epoch": 0.7052337097465847,
      "grad_norm": 0.5611788321327241,
      "learning_rate": 0.0001502009683479507,
      "loss": 12.2544,
      "step": 12951
    },
    {
      "epoch": 0.7052881637431677,
      "grad_norm": 0.6028495282539408,
      "learning_rate": 0.00015019334157638948,
      "loss": 12.0151,
      "step": 12952
    },
    {
      "epoch": 0.7053426177397507,
      "grad_norm": 0.6158821652859549,
      "learning_rate": 0.00015018571441451642,
      "loss": 12.3645,
      "step": 12953
    },
    {
      "epoch": 0.7053970717363337,
      "grad_norm": 0.5943618216782155,
      "learning_rate": 0.00015017808686239079,
      "loss": 12.2499,
      "step": 12954
    },
    {
      "epoch": 0.7054515257329168,
      "grad_norm": 0.5629309510246335,
      "learning_rate": 0.00015017045892007195,
      "loss": 12.3606,
      "step": 12955
    },
    {
      "epoch": 0.7055059797294998,
      "grad_norm": 0.6314402368385341,
      "learning_rate": 0.0001501628305876192,
      "loss": 12.0944,
      "step": 12956
    },
    {
      "epoch": 0.7055604337260828,
      "grad_norm": 0.6365115644400484,
      "learning_rate": 0.00015015520186509193,
      "loss": 12.3064,
      "step": 12957
    },
    {
      "epoch": 0.7056148877226658,
      "grad_norm": 0.5965986296655351,
      "learning_rate": 0.00015014757275254932,
      "loss": 12.1364,
      "step": 12958
    },
    {
      "epoch": 0.7056693417192488,
      "grad_norm": 0.5948973063017181,
      "learning_rate": 0.00015013994325005085,
      "loss": 12.3672,
      "step": 12959
    },
    {
      "epoch": 0.7057237957158318,
      "grad_norm": 0.5884444949376633,
      "learning_rate": 0.00015013231335765572,
      "loss": 12.2143,
      "step": 12960
    },
    {
      "epoch": 0.7057782497124149,
      "grad_norm": 0.6133597645062747,
      "learning_rate": 0.0001501246830754233,
      "loss": 12.3034,
      "step": 12961
    },
    {
      "epoch": 0.7058327037089979,
      "grad_norm": 0.5898220414273352,
      "learning_rate": 0.000150117052403413,
      "loss": 12.2611,
      "step": 12962
    },
    {
      "epoch": 0.7058871577055809,
      "grad_norm": 0.5961760638203288,
      "learning_rate": 0.00015010942134168403,
      "loss": 12.1966,
      "step": 12963
    },
    {
      "epoch": 0.7059416117021639,
      "grad_norm": 0.5601188165322548,
      "learning_rate": 0.00015010178989029584,
      "loss": 12.3401,
      "step": 12964
    },
    {
      "epoch": 0.7059960656987468,
      "grad_norm": 0.6761852903853308,
      "learning_rate": 0.00015009415804930772,
      "loss": 12.328,
      "step": 12965
    },
    {
      "epoch": 0.7060505196953298,
      "grad_norm": 0.8251909509115992,
      "learning_rate": 0.000150086525818779,
      "loss": 12.3578,
      "step": 12966
    },
    {
      "epoch": 0.7061049736919129,
      "grad_norm": 0.5408091436796602,
      "learning_rate": 0.00015007889319876912,
      "loss": 12.2561,
      "step": 12967
    },
    {
      "epoch": 0.7061594276884959,
      "grad_norm": 0.5507950813247795,
      "learning_rate": 0.00015007126018933733,
      "loss": 12.1793,
      "step": 12968
    },
    {
      "epoch": 0.7062138816850789,
      "grad_norm": 0.5232231954991038,
      "learning_rate": 0.000150063626790543,
      "loss": 12.2478,
      "step": 12969
    },
    {
      "epoch": 0.7062683356816619,
      "grad_norm": 0.5860145048316517,
      "learning_rate": 0.00015005599300244556,
      "loss": 12.3502,
      "step": 12970
    },
    {
      "epoch": 0.7063227896782449,
      "grad_norm": 0.5740070028883291,
      "learning_rate": 0.0001500483588251043,
      "loss": 12.201,
      "step": 12971
    },
    {
      "epoch": 0.706377243674828,
      "grad_norm": 0.5277369607087119,
      "learning_rate": 0.0001500407242585786,
      "loss": 12.2771,
      "step": 12972
    },
    {
      "epoch": 0.706431697671411,
      "grad_norm": 0.6978047578361452,
      "learning_rate": 0.00015003308930292784,
      "loss": 12.2717,
      "step": 12973
    },
    {
      "epoch": 0.706486151667994,
      "grad_norm": 0.5385232740976329,
      "learning_rate": 0.00015002545395821138,
      "loss": 12.1565,
      "step": 12974
    },
    {
      "epoch": 0.706540605664577,
      "grad_norm": 0.5768496412226605,
      "learning_rate": 0.0001500178182244886,
      "loss": 12.2259,
      "step": 12975
    },
    {
      "epoch": 0.70659505966116,
      "grad_norm": 0.6189081619794492,
      "learning_rate": 0.0001500101821018189,
      "loss": 12.1882,
      "step": 12976
    },
    {
      "epoch": 0.706649513657743,
      "grad_norm": 0.630691109976242,
      "learning_rate": 0.00015000254559026163,
      "loss": 12.1659,
      "step": 12977
    },
    {
      "epoch": 0.7067039676543261,
      "grad_norm": 0.5674145907848038,
      "learning_rate": 0.00014999490868987617,
      "loss": 12.2894,
      "step": 12978
    },
    {
      "epoch": 0.7067584216509091,
      "grad_norm": 0.5967081606199218,
      "learning_rate": 0.00014998727140072194,
      "loss": 12.3154,
      "step": 12979
    },
    {
      "epoch": 0.7068128756474921,
      "grad_norm": 0.5814432648578499,
      "learning_rate": 0.00014997963372285827,
      "loss": 12.2798,
      "step": 12980
    },
    {
      "epoch": 0.7068673296440751,
      "grad_norm": 0.7079190267375342,
      "learning_rate": 0.0001499719956563446,
      "loss": 12.2989,
      "step": 12981
    },
    {
      "epoch": 0.706921783640658,
      "grad_norm": 0.5976573686600499,
      "learning_rate": 0.0001499643572012403,
      "loss": 12.3619,
      "step": 12982
    },
    {
      "epoch": 0.706976237637241,
      "grad_norm": 0.5945798656008229,
      "learning_rate": 0.00014995671835760482,
      "loss": 12.2903,
      "step": 12983
    },
    {
      "epoch": 0.7070306916338241,
      "grad_norm": 0.7134044262037385,
      "learning_rate": 0.0001499490791254975,
      "loss": 12.315,
      "step": 12984
    },
    {
      "epoch": 0.7070851456304071,
      "grad_norm": 0.7298247499418934,
      "learning_rate": 0.00014994143950497775,
      "loss": 12.2258,
      "step": 12985
    },
    {
      "epoch": 0.7071395996269901,
      "grad_norm": 0.5753519644815357,
      "learning_rate": 0.000149933799496105,
      "loss": 12.2707,
      "step": 12986
    },
    {
      "epoch": 0.7071940536235731,
      "grad_norm": 0.5618107193737906,
      "learning_rate": 0.00014992615909893865,
      "loss": 12.2128,
      "step": 12987
    },
    {
      "epoch": 0.7072485076201561,
      "grad_norm": 0.6589114753084558,
      "learning_rate": 0.0001499185183135381,
      "loss": 12.3593,
      "step": 12988
    },
    {
      "epoch": 0.7073029616167391,
      "grad_norm": 0.6024408905521136,
      "learning_rate": 0.0001499108771399628,
      "loss": 12.1083,
      "step": 12989
    },
    {
      "epoch": 0.7073574156133222,
      "grad_norm": 0.5998225459431465,
      "learning_rate": 0.00014990323557827214,
      "loss": 12.2451,
      "step": 12990
    },
    {
      "epoch": 0.7074118696099052,
      "grad_norm": 0.6596737008732433,
      "learning_rate": 0.00014989559362852555,
      "loss": 12.352,
      "step": 12991
    },
    {
      "epoch": 0.7074663236064882,
      "grad_norm": 0.5640430440547669,
      "learning_rate": 0.0001498879512907825,
      "loss": 12.2266,
      "step": 12992
    },
    {
      "epoch": 0.7075207776030712,
      "grad_norm": 0.58905805846539,
      "learning_rate": 0.0001498803085651023,
      "loss": 12.2615,
      "step": 12993
    },
    {
      "epoch": 0.7075752315996542,
      "grad_norm": 0.5459263943702819,
      "learning_rate": 0.0001498726654515445,
      "loss": 12.2659,
      "step": 12994
    },
    {
      "epoch": 0.7076296855962372,
      "grad_norm": 0.64321390044219,
      "learning_rate": 0.0001498650219501685,
      "loss": 12.3126,
      "step": 12995
    },
    {
      "epoch": 0.7076841395928203,
      "grad_norm": 0.6119570222522704,
      "learning_rate": 0.00014985737806103372,
      "loss": 12.3096,
      "step": 12996
    },
    {
      "epoch": 0.7077385935894033,
      "grad_norm": 0.594510800282877,
      "learning_rate": 0.0001498497337841996,
      "loss": 12.3027,
      "step": 12997
    },
    {
      "epoch": 0.7077930475859863,
      "grad_norm": 0.6117166303249918,
      "learning_rate": 0.0001498420891197256,
      "loss": 12.0491,
      "step": 12998
    },
    {
      "epoch": 0.7078475015825693,
      "grad_norm": 0.5886463053915209,
      "learning_rate": 0.00014983444406767112,
      "loss": 12.3004,
      "step": 12999
    },
    {
      "epoch": 0.7079019555791523,
      "grad_norm": 0.5637629847214728,
      "learning_rate": 0.00014982679862809566,
      "loss": 12.2073,
      "step": 13000
    },
    {
      "epoch": 0.7079564095757352,
      "grad_norm": 0.8458027872746456,
      "learning_rate": 0.00014981915280105868,
      "loss": 12.3988,
      "step": 13001
    },
    {
      "epoch": 0.7080108635723183,
      "grad_norm": 0.6073435712822226,
      "learning_rate": 0.0001498115065866196,
      "loss": 12.2968,
      "step": 13002
    },
    {
      "epoch": 0.7080653175689013,
      "grad_norm": 0.574713441835352,
      "learning_rate": 0.0001498038599848379,
      "loss": 12.3245,
      "step": 13003
    },
    {
      "epoch": 0.7081197715654843,
      "grad_norm": 0.5738326787390379,
      "learning_rate": 0.000149796212995773,
      "loss": 12.0705,
      "step": 13004
    },
    {
      "epoch": 0.7081742255620673,
      "grad_norm": 0.6681791939973973,
      "learning_rate": 0.00014978856561948442,
      "loss": 12.3879,
      "step": 13005
    },
    {
      "epoch": 0.7082286795586503,
      "grad_norm": 0.5903997054499869,
      "learning_rate": 0.00014978091785603162,
      "loss": 12.3157,
      "step": 13006
    },
    {
      "epoch": 0.7082831335552334,
      "grad_norm": 0.6755493009094661,
      "learning_rate": 0.00014977326970547406,
      "loss": 12.3751,
      "step": 13007
    },
    {
      "epoch": 0.7083375875518164,
      "grad_norm": 0.6520915349052323,
      "learning_rate": 0.00014976562116787115,
      "loss": 12.2351,
      "step": 13008
    },
    {
      "epoch": 0.7083920415483994,
      "grad_norm": 0.6333760318312311,
      "learning_rate": 0.00014975797224328247,
      "loss": 12.1828,
      "step": 13009
    },
    {
      "epoch": 0.7084464955449824,
      "grad_norm": 0.6331195754530802,
      "learning_rate": 0.00014975032293176744,
      "loss": 12.3569,
      "step": 13010
    },
    {
      "epoch": 0.7085009495415654,
      "grad_norm": 0.6345875524219766,
      "learning_rate": 0.00014974267323338556,
      "loss": 12.3998,
      "step": 13011
    },
    {
      "epoch": 0.7085554035381484,
      "grad_norm": 0.580569784650918,
      "learning_rate": 0.00014973502314819635,
      "loss": 12.2017,
      "step": 13012
    },
    {
      "epoch": 0.7086098575347315,
      "grad_norm": 0.5794457087183509,
      "learning_rate": 0.0001497273726762592,
      "loss": 12.2481,
      "step": 13013
    },
    {
      "epoch": 0.7086643115313145,
      "grad_norm": 0.566031642380644,
      "learning_rate": 0.0001497197218176337,
      "loss": 12.17,
      "step": 13014
    },
    {
      "epoch": 0.7087187655278975,
      "grad_norm": 0.5701972777924958,
      "learning_rate": 0.00014971207057237927,
      "loss": 12.2852,
      "step": 13015
    },
    {
      "epoch": 0.7087732195244805,
      "grad_norm": 0.6256922142544816,
      "learning_rate": 0.00014970441894055544,
      "loss": 12.3115,
      "step": 13016
    },
    {
      "epoch": 0.7088276735210635,
      "grad_norm": 0.6354118579123414,
      "learning_rate": 0.00014969676692222174,
      "loss": 12.3488,
      "step": 13017
    },
    {
      "epoch": 0.7088821275176465,
      "grad_norm": 0.6177833774317564,
      "learning_rate": 0.00014968911451743766,
      "loss": 12.3751,
      "step": 13018
    },
    {
      "epoch": 0.7089365815142296,
      "grad_norm": 0.6051671506499325,
      "learning_rate": 0.00014968146172626264,
      "loss": 12.1438,
      "step": 13019
    },
    {
      "epoch": 0.7089910355108126,
      "grad_norm": 0.5834906437141846,
      "learning_rate": 0.0001496738085487563,
      "loss": 12.2348,
      "step": 13020
    },
    {
      "epoch": 0.7090454895073955,
      "grad_norm": 0.579844554082125,
      "learning_rate": 0.00014966615498497804,
      "loss": 12.297,
      "step": 13021
    },
    {
      "epoch": 0.7090999435039785,
      "grad_norm": 0.5442590529551725,
      "learning_rate": 0.00014965850103498747,
      "loss": 12.2521,
      "step": 13022
    },
    {
      "epoch": 0.7091543975005615,
      "grad_norm": 0.5210616076593362,
      "learning_rate": 0.00014965084669884407,
      "loss": 11.9946,
      "step": 13023
    },
    {
      "epoch": 0.7092088514971445,
      "grad_norm": 0.6417333566925977,
      "learning_rate": 0.00014964319197660736,
      "loss": 12.4075,
      "step": 13024
    },
    {
      "epoch": 0.7092633054937276,
      "grad_norm": 0.5371747489792881,
      "learning_rate": 0.00014963553686833683,
      "loss": 12.1904,
      "step": 13025
    },
    {
      "epoch": 0.7093177594903106,
      "grad_norm": 0.5548915066555968,
      "learning_rate": 0.00014962788137409206,
      "loss": 12.1782,
      "step": 13026
    },
    {
      "epoch": 0.7093722134868936,
      "grad_norm": 0.5334034909808053,
      "learning_rate": 0.00014962022549393255,
      "loss": 12.1223,
      "step": 13027
    },
    {
      "epoch": 0.7094266674834766,
      "grad_norm": 0.5738456004784469,
      "learning_rate": 0.00014961256922791787,
      "loss": 12.1724,
      "step": 13028
    },
    {
      "epoch": 0.7094811214800596,
      "grad_norm": 0.6147623973023848,
      "learning_rate": 0.00014960491257610752,
      "loss": 12.3882,
      "step": 13029
    },
    {
      "epoch": 0.7095355754766426,
      "grad_norm": 0.5406251727724115,
      "learning_rate": 0.00014959725553856105,
      "loss": 12.1801,
      "step": 13030
    },
    {
      "epoch": 0.7095900294732257,
      "grad_norm": 0.5821260739638942,
      "learning_rate": 0.000149589598115338,
      "loss": 12.2504,
      "step": 13031
    },
    {
      "epoch": 0.7096444834698087,
      "grad_norm": 0.555126981255046,
      "learning_rate": 0.00014958194030649792,
      "loss": 12.1343,
      "step": 13032
    },
    {
      "epoch": 0.7096989374663917,
      "grad_norm": 0.5697013295420689,
      "learning_rate": 0.00014957428211210036,
      "loss": 12.3033,
      "step": 13033
    },
    {
      "epoch": 0.7097533914629747,
      "grad_norm": 0.5492315961312766,
      "learning_rate": 0.00014956662353220488,
      "loss": 12.3455,
      "step": 13034
    },
    {
      "epoch": 0.7098078454595577,
      "grad_norm": 0.5918307712403531,
      "learning_rate": 0.00014955896456687103,
      "loss": 12.2083,
      "step": 13035
    },
    {
      "epoch": 0.7098622994561407,
      "grad_norm": 0.6530977518570804,
      "learning_rate": 0.00014955130521615833,
      "loss": 12.2589,
      "step": 13036
    },
    {
      "epoch": 0.7099167534527238,
      "grad_norm": 0.6120186751995188,
      "learning_rate": 0.00014954364548012638,
      "loss": 12.2314,
      "step": 13037
    },
    {
      "epoch": 0.7099712074493068,
      "grad_norm": 0.5867757951906781,
      "learning_rate": 0.00014953598535883474,
      "loss": 12.3824,
      "step": 13038
    },
    {
      "epoch": 0.7100256614458897,
      "grad_norm": 0.6451329228108325,
      "learning_rate": 0.00014952832485234295,
      "loss": 12.3503,
      "step": 13039
    },
    {
      "epoch": 0.7100801154424727,
      "grad_norm": 0.6861941361354333,
      "learning_rate": 0.00014952066396071062,
      "loss": 12.3653,
      "step": 13040
    },
    {
      "epoch": 0.7101345694390557,
      "grad_norm": 0.6513061150663569,
      "learning_rate": 0.0001495130026839973,
      "loss": 12.2643,
      "step": 13041
    },
    {
      "epoch": 0.7101890234356388,
      "grad_norm": 0.6295887315128024,
      "learning_rate": 0.0001495053410222626,
      "loss": 12.1674,
      "step": 13042
    },
    {
      "epoch": 0.7102434774322218,
      "grad_norm": 0.6510454621330697,
      "learning_rate": 0.00014949767897556602,
      "loss": 12.2018,
      "step": 13043
    },
    {
      "epoch": 0.7102979314288048,
      "grad_norm": 0.6875565394105327,
      "learning_rate": 0.00014949001654396719,
      "loss": 12.3256,
      "step": 13044
    },
    {
      "epoch": 0.7103523854253878,
      "grad_norm": 0.5398634943151209,
      "learning_rate": 0.0001494823537275257,
      "loss": 12.2262,
      "step": 13045
    },
    {
      "epoch": 0.7104068394219708,
      "grad_norm": 0.543330019868829,
      "learning_rate": 0.00014947469052630115,
      "loss": 12.2737,
      "step": 13046
    },
    {
      "epoch": 0.7104612934185538,
      "grad_norm": 0.6116969575632312,
      "learning_rate": 0.0001494670269403531,
      "loss": 12.2623,
      "step": 13047
    },
    {
      "epoch": 0.7105157474151369,
      "grad_norm": 0.577409531377027,
      "learning_rate": 0.00014945936296974113,
      "loss": 12.3066,
      "step": 13048
    },
    {
      "epoch": 0.7105702014117199,
      "grad_norm": 0.6337211061727935,
      "learning_rate": 0.0001494516986145249,
      "loss": 12.2312,
      "step": 13049
    },
    {
      "epoch": 0.7106246554083029,
      "grad_norm": 0.6188852759508668,
      "learning_rate": 0.00014944403387476393,
      "loss": 12.1674,
      "step": 13050
    },
    {
      "epoch": 0.7106791094048859,
      "grad_norm": 0.6080463199956461,
      "learning_rate": 0.00014943636875051788,
      "loss": 12.2713,
      "step": 13051
    },
    {
      "epoch": 0.7107335634014689,
      "grad_norm": 0.5719060777051059,
      "learning_rate": 0.00014942870324184633,
      "loss": 12.248,
      "step": 13052
    },
    {
      "epoch": 0.7107880173980519,
      "grad_norm": 0.5589157114552834,
      "learning_rate": 0.0001494210373488089,
      "loss": 12.2863,
      "step": 13053
    },
    {
      "epoch": 0.710842471394635,
      "grad_norm": 0.6891559546184134,
      "learning_rate": 0.00014941337107146518,
      "loss": 12.2425,
      "step": 13054
    },
    {
      "epoch": 0.710896925391218,
      "grad_norm": 0.5982489360119315,
      "learning_rate": 0.0001494057044098748,
      "loss": 12.2747,
      "step": 13055
    },
    {
      "epoch": 0.710951379387801,
      "grad_norm": 0.6009505560195769,
      "learning_rate": 0.00014939803736409738,
      "loss": 12.2842,
      "step": 13056
    },
    {
      "epoch": 0.711005833384384,
      "grad_norm": 0.6199360893918832,
      "learning_rate": 0.00014939036993419255,
      "loss": 12.2094,
      "step": 13057
    },
    {
      "epoch": 0.7110602873809669,
      "grad_norm": 0.6178935479060033,
      "learning_rate": 0.00014938270212021994,
      "loss": 12.323,
      "step": 13058
    },
    {
      "epoch": 0.7111147413775499,
      "grad_norm": 0.6180419467020031,
      "learning_rate": 0.0001493750339222391,
      "loss": 12.4192,
      "step": 13059
    },
    {
      "epoch": 0.711169195374133,
      "grad_norm": 0.61773533913744,
      "learning_rate": 0.00014936736534030978,
      "loss": 12.3702,
      "step": 13060
    },
    {
      "epoch": 0.711223649370716,
      "grad_norm": 0.6224002643765801,
      "learning_rate": 0.0001493596963744915,
      "loss": 12.3371,
      "step": 13061
    },
    {
      "epoch": 0.711278103367299,
      "grad_norm": 0.6259537112445535,
      "learning_rate": 0.00014935202702484395,
      "loss": 12.3679,
      "step": 13062
    },
    {
      "epoch": 0.711332557363882,
      "grad_norm": 0.5328991389879344,
      "learning_rate": 0.00014934435729142682,
      "loss": 12.269,
      "step": 13063
    },
    {
      "epoch": 0.711387011360465,
      "grad_norm": 0.5898400658269457,
      "learning_rate": 0.0001493366871742996,
      "loss": 12.4181,
      "step": 13064
    },
    {
      "epoch": 0.711441465357048,
      "grad_norm": 0.5628544206358188,
      "learning_rate": 0.0001493290166735221,
      "loss": 12.3083,
      "step": 13065
    },
    {
      "epoch": 0.7114959193536311,
      "grad_norm": 0.6712292466408655,
      "learning_rate": 0.00014932134578915385,
      "loss": 12.2852,
      "step": 13066
    },
    {
      "epoch": 0.7115503733502141,
      "grad_norm": 0.5481225982918251,
      "learning_rate": 0.00014931367452125456,
      "loss": 12.0959,
      "step": 13067
    },
    {
      "epoch": 0.7116048273467971,
      "grad_norm": 0.6030042765006167,
      "learning_rate": 0.00014930600286988387,
      "loss": 12.2808,
      "step": 13068
    },
    {
      "epoch": 0.7116592813433801,
      "grad_norm": 0.6391940261618729,
      "learning_rate": 0.00014929833083510144,
      "loss": 12.4102,
      "step": 13069
    },
    {
      "epoch": 0.7117137353399631,
      "grad_norm": 0.5354168607988496,
      "learning_rate": 0.00014929065841696686,
      "loss": 12.2758,
      "step": 13070
    },
    {
      "epoch": 0.7117681893365461,
      "grad_norm": 0.596410067524563,
      "learning_rate": 0.00014928298561553996,
      "loss": 12.3488,
      "step": 13071
    },
    {
      "epoch": 0.7118226433331292,
      "grad_norm": 0.5435545899850704,
      "learning_rate": 0.00014927531243088022,
      "loss": 12.0993,
      "step": 13072
    },
    {
      "epoch": 0.7118770973297122,
      "grad_norm": 0.5464907779246453,
      "learning_rate": 0.00014926763886304744,
      "loss": 12.2565,
      "step": 13073
    },
    {
      "epoch": 0.7119315513262952,
      "grad_norm": 0.5507232020598494,
      "learning_rate": 0.0001492599649121012,
      "loss": 12.2084,
      "step": 13074
    },
    {
      "epoch": 0.7119860053228781,
      "grad_norm": 0.576496083537236,
      "learning_rate": 0.0001492522905781012,
      "loss": 12.4546,
      "step": 13075
    },
    {
      "epoch": 0.7120404593194611,
      "grad_norm": 0.5174273315209944,
      "learning_rate": 0.00014924461586110716,
      "loss": 12.2471,
      "step": 13076
    },
    {
      "epoch": 0.7120949133160442,
      "grad_norm": 0.5563814598625784,
      "learning_rate": 0.00014923694076117872,
      "loss": 12.1744,
      "step": 13077
    },
    {
      "epoch": 0.7121493673126272,
      "grad_norm": 0.5938209262601601,
      "learning_rate": 0.00014922926527837556,
      "loss": 12.4136,
      "step": 13078
    },
    {
      "epoch": 0.7122038213092102,
      "grad_norm": 0.6440486117840959,
      "learning_rate": 0.00014922158941275742,
      "loss": 12.2879,
      "step": 13079
    },
    {
      "epoch": 0.7122582753057932,
      "grad_norm": 0.5424085110763885,
      "learning_rate": 0.0001492139131643839,
      "loss": 12.2847,
      "step": 13080
    },
    {
      "epoch": 0.7123127293023762,
      "grad_norm": 0.6288430281756898,
      "learning_rate": 0.00014920623653331473,
      "loss": 12.2262,
      "step": 13081
    },
    {
      "epoch": 0.7123671832989592,
      "grad_norm": 0.55112138539526,
      "learning_rate": 0.00014919855951960964,
      "loss": 12.193,
      "step": 13082
    },
    {
      "epoch": 0.7124216372955423,
      "grad_norm": 0.6103232989314326,
      "learning_rate": 0.00014919088212332832,
      "loss": 12.3006,
      "step": 13083
    },
    {
      "epoch": 0.7124760912921253,
      "grad_norm": 0.585852252955573,
      "learning_rate": 0.00014918320434453044,
      "loss": 12.2011,
      "step": 13084
    },
    {
      "epoch": 0.7125305452887083,
      "grad_norm": 0.6023320204235458,
      "learning_rate": 0.00014917552618327567,
      "loss": 12.183,
      "step": 13085
    },
    {
      "epoch": 0.7125849992852913,
      "grad_norm": 0.637684131987905,
      "learning_rate": 0.00014916784763962382,
      "loss": 12.1952,
      "step": 13086
    },
    {
      "epoch": 0.7126394532818743,
      "grad_norm": 0.5635832480007723,
      "learning_rate": 0.0001491601687136345,
      "loss": 12.2105,
      "step": 13087
    },
    {
      "epoch": 0.7126939072784573,
      "grad_norm": 0.5277875554433332,
      "learning_rate": 0.00014915248940536747,
      "loss": 12.2859,
      "step": 13088
    },
    {
      "epoch": 0.7127483612750404,
      "grad_norm": 0.7686941264622156,
      "learning_rate": 0.00014914480971488247,
      "loss": 12.3042,
      "step": 13089
    },
    {
      "epoch": 0.7128028152716234,
      "grad_norm": 0.5777273873310947,
      "learning_rate": 0.00014913712964223917,
      "loss": 12.3386,
      "step": 13090
    },
    {
      "epoch": 0.7128572692682064,
      "grad_norm": 0.5662398880449121,
      "learning_rate": 0.0001491294491874973,
      "loss": 12.0751,
      "step": 13091
    },
    {
      "epoch": 0.7129117232647894,
      "grad_norm": 0.5694475424596657,
      "learning_rate": 0.00014912176835071657,
      "loss": 12.2462,
      "step": 13092
    },
    {
      "epoch": 0.7129661772613723,
      "grad_norm": 0.6194296392446067,
      "learning_rate": 0.00014911408713195678,
      "loss": 12.3051,
      "step": 13093
    },
    {
      "epoch": 0.7130206312579553,
      "grad_norm": 0.5929804883936958,
      "learning_rate": 0.0001491064055312776,
      "loss": 11.9636,
      "step": 13094
    },
    {
      "epoch": 0.7130750852545384,
      "grad_norm": 0.577121793325591,
      "learning_rate": 0.00014909872354873876,
      "loss": 12.1692,
      "step": 13095
    },
    {
      "epoch": 0.7131295392511214,
      "grad_norm": 0.5399081423465084,
      "learning_rate": 0.00014909104118440003,
      "loss": 12.1337,
      "step": 13096
    },
    {
      "epoch": 0.7131839932477044,
      "grad_norm": 0.6129261066771551,
      "learning_rate": 0.00014908335843832112,
      "loss": 12.2635,
      "step": 13097
    },
    {
      "epoch": 0.7132384472442874,
      "grad_norm": 0.5557186330224296,
      "learning_rate": 0.0001490756753105618,
      "loss": 12.152,
      "step": 13098
    },
    {
      "epoch": 0.7132929012408704,
      "grad_norm": 0.6522389101871415,
      "learning_rate": 0.00014906799180118178,
      "loss": 12.222,
      "step": 13099
    },
    {
      "epoch": 0.7133473552374534,
      "grad_norm": 0.6201291721778343,
      "learning_rate": 0.00014906030791024083,
      "loss": 12.4092,
      "step": 13100
    },
    {
      "epoch": 0.7134018092340365,
      "grad_norm": 0.5769656708175186,
      "learning_rate": 0.0001490526236377987,
      "loss": 12.1597,
      "step": 13101
    },
    {
      "epoch": 0.7134562632306195,
      "grad_norm": 0.6034623352492604,
      "learning_rate": 0.00014904493898391515,
      "loss": 12.2657,
      "step": 13102
    },
    {
      "epoch": 0.7135107172272025,
      "grad_norm": 0.7048369446631927,
      "learning_rate": 0.00014903725394864993,
      "loss": 12.3768,
      "step": 13103
    },
    {
      "epoch": 0.7135651712237855,
      "grad_norm": 0.5917872807045255,
      "learning_rate": 0.00014902956853206275,
      "loss": 12.1509,
      "step": 13104
    },
    {
      "epoch": 0.7136196252203685,
      "grad_norm": 0.6228402490002504,
      "learning_rate": 0.0001490218827342135,
      "loss": 12.1816,
      "step": 13105
    },
    {
      "epoch": 0.7136740792169516,
      "grad_norm": 0.6237825639120929,
      "learning_rate": 0.00014901419655516182,
      "loss": 12.2323,
      "step": 13106
    },
    {
      "epoch": 0.7137285332135346,
      "grad_norm": 0.6437731143746762,
      "learning_rate": 0.00014900650999496754,
      "loss": 12.1785,
      "step": 13107
    },
    {
      "epoch": 0.7137829872101176,
      "grad_norm": 0.5712009786525403,
      "learning_rate": 0.00014899882305369043,
      "loss": 12.2221,
      "step": 13108
    },
    {
      "epoch": 0.7138374412067006,
      "grad_norm": 0.5935683808044501,
      "learning_rate": 0.00014899113573139022,
      "loss": 12.2265,
      "step": 13109
    },
    {
      "epoch": 0.7138918952032836,
      "grad_norm": 0.6016139400333811,
      "learning_rate": 0.00014898344802812677,
      "loss": 12.2309,
      "step": 13110
    },
    {
      "epoch": 0.7139463491998665,
      "grad_norm": 0.5515894574044553,
      "learning_rate": 0.0001489757599439598,
      "loss": 12.2961,
      "step": 13111
    },
    {
      "epoch": 0.7140008031964497,
      "grad_norm": 0.6157520966673266,
      "learning_rate": 0.0001489680714789491,
      "loss": 12.3123,
      "step": 13112
    },
    {
      "epoch": 0.7140552571930326,
      "grad_norm": 0.6139323540701279,
      "learning_rate": 0.00014896038263315445,
      "loss": 12.1918,
      "step": 13113
    },
    {
      "epoch": 0.7141097111896156,
      "grad_norm": 0.6524316381419333,
      "learning_rate": 0.00014895269340663568,
      "loss": 12.3856,
      "step": 13114
    },
    {
      "epoch": 0.7141641651861986,
      "grad_norm": 0.564740400972851,
      "learning_rate": 0.00014894500379945252,
      "loss": 12.401,
      "step": 13115
    },
    {
      "epoch": 0.7142186191827816,
      "grad_norm": 0.621305851648461,
      "learning_rate": 0.00014893731381166486,
      "loss": 12.2224,
      "step": 13116
    },
    {
      "epoch": 0.7142730731793646,
      "grad_norm": 0.6056844793845901,
      "learning_rate": 0.0001489296234433324,
      "loss": 12.2252,
      "step": 13117
    },
    {
      "epoch": 0.7143275271759477,
      "grad_norm": 0.565175529177596,
      "learning_rate": 0.000148921932694515,
      "loss": 12.2674,
      "step": 13118
    },
    {
      "epoch": 0.7143819811725307,
      "grad_norm": 0.6050541903974955,
      "learning_rate": 0.00014891424156527241,
      "loss": 12.3528,
      "step": 13119
    },
    {
      "epoch": 0.7144364351691137,
      "grad_norm": 0.5742183932203648,
      "learning_rate": 0.0001489065500556645,
      "loss": 12.2321,
      "step": 13120
    },
    {
      "epoch": 0.7144908891656967,
      "grad_norm": 0.6274447021592812,
      "learning_rate": 0.00014889885816575106,
      "loss": 12.2845,
      "step": 13121
    },
    {
      "epoch": 0.7145453431622797,
      "grad_norm": 0.6539372288535807,
      "learning_rate": 0.00014889116589559192,
      "loss": 12.2235,
      "step": 13122
    },
    {
      "epoch": 0.7145997971588627,
      "grad_norm": 0.5398132389690338,
      "learning_rate": 0.00014888347324524685,
      "loss": 12.1618,
      "step": 13123
    },
    {
      "epoch": 0.7146542511554458,
      "grad_norm": 0.6012845841839606,
      "learning_rate": 0.0001488757802147757,
      "loss": 12.4002,
      "step": 13124
    },
    {
      "epoch": 0.7147087051520288,
      "grad_norm": 0.6481016820248655,
      "learning_rate": 0.00014886808680423825,
      "loss": 12.3224,
      "step": 13125
    },
    {
      "epoch": 0.7147631591486118,
      "grad_norm": 0.5473743337050713,
      "learning_rate": 0.0001488603930136944,
      "loss": 12.2754,
      "step": 13126
    },
    {
      "epoch": 0.7148176131451948,
      "grad_norm": 0.5585932175414414,
      "learning_rate": 0.00014885269884320394,
      "loss": 12.3176,
      "step": 13127
    },
    {
      "epoch": 0.7148720671417778,
      "grad_norm": 0.6098644156874881,
      "learning_rate": 0.00014884500429282672,
      "loss": 12.3161,
      "step": 13128
    },
    {
      "epoch": 0.7149265211383607,
      "grad_norm": 0.5900891649967697,
      "learning_rate": 0.0001488373093626225,
      "loss": 12.3236,
      "step": 13129
    },
    {
      "epoch": 0.7149809751349439,
      "grad_norm": 0.52526226559206,
      "learning_rate": 0.0001488296140526512,
      "loss": 12.2481,
      "step": 13130
    },
    {
      "epoch": 0.7150354291315268,
      "grad_norm": 0.6211944112719395,
      "learning_rate": 0.00014882191836297263,
      "loss": 12.355,
      "step": 13131
    },
    {
      "epoch": 0.7150898831281098,
      "grad_norm": 0.6088171060766341,
      "learning_rate": 0.00014881422229364667,
      "loss": 12.2572,
      "step": 13132
    },
    {
      "epoch": 0.7151443371246928,
      "grad_norm": 0.5867369564056631,
      "learning_rate": 0.0001488065258447331,
      "loss": 12.366,
      "step": 13133
    },
    {
      "epoch": 0.7151987911212758,
      "grad_norm": 0.6565966460145465,
      "learning_rate": 0.00014879882901629182,
      "loss": 12.0463,
      "step": 13134
    },
    {
      "epoch": 0.7152532451178588,
      "grad_norm": 0.8062366477423585,
      "learning_rate": 0.00014879113180838263,
      "loss": 12.2164,
      "step": 13135
    },
    {
      "epoch": 0.7153076991144419,
      "grad_norm": 0.8015097363835164,
      "learning_rate": 0.0001487834342210654,
      "loss": 12.4014,
      "step": 13136
    },
    {
      "epoch": 0.7153621531110249,
      "grad_norm": 0.5663030303378246,
      "learning_rate": 0.00014877573625440006,
      "loss": 12.3439,
      "step": 13137
    },
    {
      "epoch": 0.7154166071076079,
      "grad_norm": 0.6414869394241983,
      "learning_rate": 0.0001487680379084464,
      "loss": 12.3671,
      "step": 13138
    },
    {
      "epoch": 0.7154710611041909,
      "grad_norm": 0.599459862838817,
      "learning_rate": 0.0001487603391832643,
      "loss": 12.2809,
      "step": 13139
    },
    {
      "epoch": 0.7155255151007739,
      "grad_norm": 0.6084045336182192,
      "learning_rate": 0.0001487526400789136,
      "loss": 12.366,
      "step": 13140
    },
    {
      "epoch": 0.715579969097357,
      "grad_norm": 0.5717786606274167,
      "learning_rate": 0.0001487449405954542,
      "loss": 12.1181,
      "step": 13141
    },
    {
      "epoch": 0.71563442309394,
      "grad_norm": 0.5944221913914374,
      "learning_rate": 0.00014873724073294597,
      "loss": 12.3322,
      "step": 13142
    },
    {
      "epoch": 0.715688877090523,
      "grad_norm": 0.6191905531252168,
      "learning_rate": 0.0001487295404914488,
      "loss": 12.1768,
      "step": 13143
    },
    {
      "epoch": 0.715743331087106,
      "grad_norm": 0.5948946545639103,
      "learning_rate": 0.00014872183987102254,
      "loss": 12.138,
      "step": 13144
    },
    {
      "epoch": 0.715797785083689,
      "grad_norm": 0.5729785875130822,
      "learning_rate": 0.0001487141388717271,
      "loss": 12.2828,
      "step": 13145
    },
    {
      "epoch": 0.715852239080272,
      "grad_norm": 0.5673965410686297,
      "learning_rate": 0.00014870643749362233,
      "loss": 12.33,
      "step": 13146
    },
    {
      "epoch": 0.7159066930768551,
      "grad_norm": 0.5663436986398205,
      "learning_rate": 0.00014869873573676812,
      "loss": 12.3317,
      "step": 13147
    },
    {
      "epoch": 0.715961147073438,
      "grad_norm": 0.5832471979785514,
      "learning_rate": 0.00014869103360122437,
      "loss": 12.2599,
      "step": 13148
    },
    {
      "epoch": 0.716015601070021,
      "grad_norm": 0.6042795355125038,
      "learning_rate": 0.00014868333108705102,
      "loss": 12.3361,
      "step": 13149
    },
    {
      "epoch": 0.716070055066604,
      "grad_norm": 0.5654725103208902,
      "learning_rate": 0.0001486756281943079,
      "loss": 12.1197,
      "step": 13150
    },
    {
      "epoch": 0.716124509063187,
      "grad_norm": 0.5867311364311525,
      "learning_rate": 0.00014866792492305493,
      "loss": 12.2859,
      "step": 13151
    },
    {
      "epoch": 0.71617896305977,
      "grad_norm": 0.6016682905288424,
      "learning_rate": 0.00014866022127335202,
      "loss": 12.2406,
      "step": 13152
    },
    {
      "epoch": 0.7162334170563531,
      "grad_norm": 0.6115550769668805,
      "learning_rate": 0.00014865251724525906,
      "loss": 12.2205,
      "step": 13153
    },
    {
      "epoch": 0.7162878710529361,
      "grad_norm": 0.5630660440481695,
      "learning_rate": 0.00014864481283883598,
      "loss": 12.1391,
      "step": 13154
    },
    {
      "epoch": 0.7163423250495191,
      "grad_norm": 0.5638294256497566,
      "learning_rate": 0.00014863710805414267,
      "loss": 12.3224,
      "step": 13155
    },
    {
      "epoch": 0.7163967790461021,
      "grad_norm": 0.560701108490784,
      "learning_rate": 0.00014862940289123904,
      "loss": 12.1735,
      "step": 13156
    },
    {
      "epoch": 0.7164512330426851,
      "grad_norm": 0.541192260738824,
      "learning_rate": 0.00014862169735018504,
      "loss": 12.1591,
      "step": 13157
    },
    {
      "epoch": 0.7165056870392681,
      "grad_norm": 0.6230640920391687,
      "learning_rate": 0.00014861399143104053,
      "loss": 12.2217,
      "step": 13158
    },
    {
      "epoch": 0.7165601410358512,
      "grad_norm": 0.5660894693307399,
      "learning_rate": 0.0001486062851338655,
      "loss": 12.2232,
      "step": 13159
    },
    {
      "epoch": 0.7166145950324342,
      "grad_norm": 0.5447717408784877,
      "learning_rate": 0.00014859857845871984,
      "loss": 12.3504,
      "step": 13160
    },
    {
      "epoch": 0.7166690490290172,
      "grad_norm": 0.5989972551874758,
      "learning_rate": 0.00014859087140566351,
      "loss": 12.3824,
      "step": 13161
    },
    {
      "epoch": 0.7167235030256002,
      "grad_norm": 0.6462447002724582,
      "learning_rate": 0.0001485831639747564,
      "loss": 12.3014,
      "step": 13162
    },
    {
      "epoch": 0.7167779570221832,
      "grad_norm": 0.5806368639453152,
      "learning_rate": 0.00014857545616605842,
      "loss": 12.2635,
      "step": 13163
    },
    {
      "epoch": 0.7168324110187662,
      "grad_norm": 0.5597040487622332,
      "learning_rate": 0.00014856774797962957,
      "loss": 12.1284,
      "step": 13164
    },
    {
      "epoch": 0.7168868650153493,
      "grad_norm": 0.6358123005292217,
      "learning_rate": 0.0001485600394155298,
      "loss": 12.3169,
      "step": 13165
    },
    {
      "epoch": 0.7169413190119323,
      "grad_norm": 0.6256827556331831,
      "learning_rate": 0.00014855233047381897,
      "loss": 12.2201,
      "step": 13166
    },
    {
      "epoch": 0.7169957730085152,
      "grad_norm": 0.6116325432530314,
      "learning_rate": 0.00014854462115455712,
      "loss": 12.3764,
      "step": 13167
    },
    {
      "epoch": 0.7170502270050982,
      "grad_norm": 0.5973785607260411,
      "learning_rate": 0.00014853691145780412,
      "loss": 12.3064,
      "step": 13168
    },
    {
      "epoch": 0.7171046810016812,
      "grad_norm": 0.6229178221685887,
      "learning_rate": 0.00014852920138361996,
      "loss": 12.2634,
      "step": 13169
    },
    {
      "epoch": 0.7171591349982642,
      "grad_norm": 0.5018641979222833,
      "learning_rate": 0.00014852149093206462,
      "loss": 12.227,
      "step": 13170
    },
    {
      "epoch": 0.7172135889948473,
      "grad_norm": 0.6245439842759595,
      "learning_rate": 0.00014851378010319801,
      "loss": 12.1027,
      "step": 13171
    },
    {
      "epoch": 0.7172680429914303,
      "grad_norm": 0.5944257970056329,
      "learning_rate": 0.00014850606889708012,
      "loss": 12.3463,
      "step": 13172
    },
    {
      "epoch": 0.7173224969880133,
      "grad_norm": 0.5401040438062852,
      "learning_rate": 0.00014849835731377088,
      "loss": 12.2422,
      "step": 13173
    },
    {
      "epoch": 0.7173769509845963,
      "grad_norm": 0.5836619995670721,
      "learning_rate": 0.0001484906453533303,
      "loss": 12.2505,
      "step": 13174
    },
    {
      "epoch": 0.7174314049811793,
      "grad_norm": 0.6015006881115633,
      "learning_rate": 0.00014848293301581835,
      "loss": 12.195,
      "step": 13175
    },
    {
      "epoch": 0.7174858589777624,
      "grad_norm": 0.6136890602982735,
      "learning_rate": 0.00014847522030129494,
      "loss": 12.3676,
      "step": 13176
    },
    {
      "epoch": 0.7175403129743454,
      "grad_norm": 0.622763211404351,
      "learning_rate": 0.0001484675072098201,
      "loss": 12.3623,
      "step": 13177
    },
    {
      "epoch": 0.7175947669709284,
      "grad_norm": 0.5902021660932876,
      "learning_rate": 0.00014845979374145384,
      "loss": 12.3369,
      "step": 13178
    },
    {
      "epoch": 0.7176492209675114,
      "grad_norm": 0.6462075351516074,
      "learning_rate": 0.00014845207989625604,
      "loss": 12.3864,
      "step": 13179
    },
    {
      "epoch": 0.7177036749640944,
      "grad_norm": 0.5608681041428895,
      "learning_rate": 0.00014844436567428674,
      "loss": 12.0696,
      "step": 13180
    },
    {
      "epoch": 0.7177581289606774,
      "grad_norm": 0.6024320792955538,
      "learning_rate": 0.00014843665107560597,
      "loss": 12.419,
      "step": 13181
    },
    {
      "epoch": 0.7178125829572605,
      "grad_norm": 0.5527658114202388,
      "learning_rate": 0.00014842893610027367,
      "loss": 12.3359,
      "step": 13182
    },
    {
      "epoch": 0.7178670369538435,
      "grad_norm": 0.585011551195655,
      "learning_rate": 0.00014842122074834984,
      "loss": 12.2481,
      "step": 13183
    },
    {
      "epoch": 0.7179214909504265,
      "grad_norm": 0.6151587175440556,
      "learning_rate": 0.00014841350501989446,
      "loss": 12.2827,
      "step": 13184
    },
    {
      "epoch": 0.7179759449470094,
      "grad_norm": 0.5769188019599946,
      "learning_rate": 0.00014840578891496753,
      "loss": 12.2817,
      "step": 13185
    },
    {
      "epoch": 0.7180303989435924,
      "grad_norm": 0.621828495887503,
      "learning_rate": 0.00014839807243362908,
      "loss": 12.2329,
      "step": 13186
    },
    {
      "epoch": 0.7180848529401754,
      "grad_norm": 0.5963198844843429,
      "learning_rate": 0.00014839035557593912,
      "loss": 12.3039,
      "step": 13187
    },
    {
      "epoch": 0.7181393069367585,
      "grad_norm": 0.5708228204132709,
      "learning_rate": 0.00014838263834195765,
      "loss": 12.3559,
      "step": 13188
    },
    {
      "epoch": 0.7181937609333415,
      "grad_norm": 0.5727996839170717,
      "learning_rate": 0.00014837492073174467,
      "loss": 12.2922,
      "step": 13189
    },
    {
      "epoch": 0.7182482149299245,
      "grad_norm": 0.6060897473951321,
      "learning_rate": 0.00014836720274536016,
      "loss": 12.4252,
      "step": 13190
    },
    {
      "epoch": 0.7183026689265075,
      "grad_norm": 0.5326123536533812,
      "learning_rate": 0.00014835948438286418,
      "loss": 12.1688,
      "step": 13191
    },
    {
      "epoch": 0.7183571229230905,
      "grad_norm": 0.5509354815979086,
      "learning_rate": 0.00014835176564431674,
      "loss": 12.256,
      "step": 13192
    },
    {
      "epoch": 0.7184115769196735,
      "grad_norm": 0.6110671982352849,
      "learning_rate": 0.0001483440465297779,
      "loss": 12.2245,
      "step": 13193
    },
    {
      "epoch": 0.7184660309162566,
      "grad_norm": 0.5516487381645298,
      "learning_rate": 0.0001483363270393076,
      "loss": 12.2766,
      "step": 13194
    },
    {
      "epoch": 0.7185204849128396,
      "grad_norm": 0.5437664696658203,
      "learning_rate": 0.00014832860717296594,
      "loss": 12.2439,
      "step": 13195
    },
    {
      "epoch": 0.7185749389094226,
      "grad_norm": 0.5607613825596112,
      "learning_rate": 0.0001483208869308129,
      "loss": 12.4613,
      "step": 13196
    },
    {
      "epoch": 0.7186293929060056,
      "grad_norm": 0.5516449388920995,
      "learning_rate": 0.00014831316631290856,
      "loss": 12.2504,
      "step": 13197
    },
    {
      "epoch": 0.7186838469025886,
      "grad_norm": 0.589519706303882,
      "learning_rate": 0.00014830544531931292,
      "loss": 12.3524,
      "step": 13198
    },
    {
      "epoch": 0.7187383008991716,
      "grad_norm": 0.5644158485740307,
      "learning_rate": 0.00014829772395008606,
      "loss": 12.2768,
      "step": 13199
    },
    {
      "epoch": 0.7187927548957547,
      "grad_norm": 0.565257166737133,
      "learning_rate": 0.00014829000220528799,
      "loss": 12.3203,
      "step": 13200
    },
    {
      "epoch": 0.7188472088923377,
      "grad_norm": 0.5562113769820356,
      "learning_rate": 0.00014828228008497877,
      "loss": 12.2482,
      "step": 13201
    },
    {
      "epoch": 0.7189016628889207,
      "grad_norm": 0.5630262812387138,
      "learning_rate": 0.00014827455758921842,
      "loss": 12.2811,
      "step": 13202
    },
    {
      "epoch": 0.7189561168855036,
      "grad_norm": 0.5694366223079674,
      "learning_rate": 0.00014826683471806703,
      "loss": 12.2332,
      "step": 13203
    },
    {
      "epoch": 0.7190105708820866,
      "grad_norm": 0.6049096069216633,
      "learning_rate": 0.00014825911147158465,
      "loss": 12.2179,
      "step": 13204
    },
    {
      "epoch": 0.7190650248786696,
      "grad_norm": 0.5414694234678251,
      "learning_rate": 0.0001482513878498313,
      "loss": 12.2502,
      "step": 13205
    },
    {
      "epoch": 0.7191194788752527,
      "grad_norm": 0.5530293618529364,
      "learning_rate": 0.0001482436638528671,
      "loss": 12.1771,
      "step": 13206
    },
    {
      "epoch": 0.7191739328718357,
      "grad_norm": 0.605115901589326,
      "learning_rate": 0.00014823593948075202,
      "loss": 12.3434,
      "step": 13207
    },
    {
      "epoch": 0.7192283868684187,
      "grad_norm": 0.5936913368093815,
      "learning_rate": 0.00014822821473354624,
      "loss": 12.2302,
      "step": 13208
    },
    {
      "epoch": 0.7192828408650017,
      "grad_norm": 0.5730714795630938,
      "learning_rate": 0.0001482204896113098,
      "loss": 12.2835,
      "step": 13209
    },
    {
      "epoch": 0.7193372948615847,
      "grad_norm": 0.5703306686028046,
      "learning_rate": 0.0001482127641141027,
      "loss": 12.2479,
      "step": 13210
    },
    {
      "epoch": 0.7193917488581678,
      "grad_norm": 0.574174984669217,
      "learning_rate": 0.00014820503824198507,
      "loss": 12.2809,
      "step": 13211
    },
    {
      "epoch": 0.7194462028547508,
      "grad_norm": 0.5336652865060002,
      "learning_rate": 0.00014819731199501696,
      "loss": 12.2294,
      "step": 13212
    },
    {
      "epoch": 0.7195006568513338,
      "grad_norm": 0.556654346812789,
      "learning_rate": 0.00014818958537325848,
      "loss": 12.1503,
      "step": 13213
    },
    {
      "epoch": 0.7195551108479168,
      "grad_norm": 0.5763679539817065,
      "learning_rate": 0.00014818185837676972,
      "loss": 12.2594,
      "step": 13214
    },
    {
      "epoch": 0.7196095648444998,
      "grad_norm": 0.5844597639610618,
      "learning_rate": 0.00014817413100561076,
      "loss": 12.2629,
      "step": 13215
    },
    {
      "epoch": 0.7196640188410828,
      "grad_norm": 0.5921512871316501,
      "learning_rate": 0.00014816640325984165,
      "loss": 12.2631,
      "step": 13216
    },
    {
      "epoch": 0.7197184728376659,
      "grad_norm": 0.5898121302300365,
      "learning_rate": 0.0001481586751395225,
      "loss": 12.35,
      "step": 13217
    },
    {
      "epoch": 0.7197729268342489,
      "grad_norm": 0.5358922462860523,
      "learning_rate": 0.00014815094664471343,
      "loss": 12.2799,
      "step": 13218
    },
    {
      "epoch": 0.7198273808308319,
      "grad_norm": 0.6353500122755664,
      "learning_rate": 0.00014814321777547454,
      "loss": 12.3604,
      "step": 13219
    },
    {
      "epoch": 0.7198818348274149,
      "grad_norm": 0.6809386098176096,
      "learning_rate": 0.0001481354885318659,
      "loss": 12.1429,
      "step": 13220
    },
    {
      "epoch": 0.7199362888239978,
      "grad_norm": 0.5314956806716192,
      "learning_rate": 0.00014812775891394762,
      "loss": 12.1481,
      "step": 13221
    },
    {
      "epoch": 0.7199907428205808,
      "grad_norm": 0.5688952356004566,
      "learning_rate": 0.00014812002892177983,
      "loss": 12.2585,
      "step": 13222
    },
    {
      "epoch": 0.720045196817164,
      "grad_norm": 0.5839960551532886,
      "learning_rate": 0.00014811229855542265,
      "loss": 12.2537,
      "step": 13223
    },
    {
      "epoch": 0.7200996508137469,
      "grad_norm": 0.626351737694393,
      "learning_rate": 0.0001481045678149361,
      "loss": 12.2205,
      "step": 13224
    },
    {
      "epoch": 0.7201541048103299,
      "grad_norm": 0.6407856048354745,
      "learning_rate": 0.0001480968367003804,
      "loss": 12.356,
      "step": 13225
    },
    {
      "epoch": 0.7202085588069129,
      "grad_norm": 0.6536082447989517,
      "learning_rate": 0.00014808910521181564,
      "loss": 12.3344,
      "step": 13226
    },
    {
      "epoch": 0.7202630128034959,
      "grad_norm": 0.6131272889251808,
      "learning_rate": 0.00014808137334930193,
      "loss": 12.2439,
      "step": 13227
    },
    {
      "epoch": 0.7203174668000789,
      "grad_norm": 0.6580668437691843,
      "learning_rate": 0.0001480736411128994,
      "loss": 12.2149,
      "step": 13228
    },
    {
      "epoch": 0.720371920796662,
      "grad_norm": 0.5741574367765172,
      "learning_rate": 0.00014806590850266818,
      "loss": 12.2251,
      "step": 13229
    },
    {
      "epoch": 0.720426374793245,
      "grad_norm": 0.5474613052768861,
      "learning_rate": 0.00014805817551866838,
      "loss": 12.3027,
      "step": 13230
    },
    {
      "epoch": 0.720480828789828,
      "grad_norm": 0.6422235053459793,
      "learning_rate": 0.00014805044216096016,
      "loss": 12.3378,
      "step": 13231
    },
    {
      "epoch": 0.720535282786411,
      "grad_norm": 0.597895121175297,
      "learning_rate": 0.00014804270842960364,
      "loss": 12.3361,
      "step": 13232
    },
    {
      "epoch": 0.720589736782994,
      "grad_norm": 0.6785533887594158,
      "learning_rate": 0.00014803497432465897,
      "loss": 12.379,
      "step": 13233
    },
    {
      "epoch": 0.720644190779577,
      "grad_norm": 0.6109394549575866,
      "learning_rate": 0.00014802723984618624,
      "loss": 12.3102,
      "step": 13234
    },
    {
      "epoch": 0.7206986447761601,
      "grad_norm": 0.6062002142767499,
      "learning_rate": 0.00014801950499424571,
      "loss": 12.3901,
      "step": 13235
    },
    {
      "epoch": 0.7207530987727431,
      "grad_norm": 0.6146312263798974,
      "learning_rate": 0.00014801176976889742,
      "loss": 12.27,
      "step": 13236
    },
    {
      "epoch": 0.7208075527693261,
      "grad_norm": 0.5439968721697996,
      "learning_rate": 0.00014800403417020155,
      "loss": 12.2037,
      "step": 13237
    },
    {
      "epoch": 0.7208620067659091,
      "grad_norm": 0.5652952590775139,
      "learning_rate": 0.00014799629819821824,
      "loss": 12.2625,
      "step": 13238
    },
    {
      "epoch": 0.720916460762492,
      "grad_norm": 0.5736293661004036,
      "learning_rate": 0.00014798856185300772,
      "loss": 12.3362,
      "step": 13239
    },
    {
      "epoch": 0.7209709147590752,
      "grad_norm": 0.6988017105721162,
      "learning_rate": 0.00014798082513463006,
      "loss": 12.2804,
      "step": 13240
    },
    {
      "epoch": 0.7210253687556581,
      "grad_norm": 0.5507195086396749,
      "learning_rate": 0.00014797308804314545,
      "loss": 12.2721,
      "step": 13241
    },
    {
      "epoch": 0.7210798227522411,
      "grad_norm": 0.5989828390022323,
      "learning_rate": 0.00014796535057861408,
      "loss": 12.2255,
      "step": 13242
    },
    {
      "epoch": 0.7211342767488241,
      "grad_norm": 0.5783072500642104,
      "learning_rate": 0.0001479576127410961,
      "loss": 12.3047,
      "step": 13243
    },
    {
      "epoch": 0.7211887307454071,
      "grad_norm": 0.5611434128093166,
      "learning_rate": 0.0001479498745306517,
      "loss": 12.1689,
      "step": 13244
    },
    {
      "epoch": 0.7212431847419901,
      "grad_norm": 0.5748784280245217,
      "learning_rate": 0.00014794213594734098,
      "loss": 12.2031,
      "step": 13245
    },
    {
      "epoch": 0.7212976387385732,
      "grad_norm": 0.5399655039572351,
      "learning_rate": 0.00014793439699122422,
      "loss": 12.2459,
      "step": 13246
    },
    {
      "epoch": 0.7213520927351562,
      "grad_norm": 0.6441317220669062,
      "learning_rate": 0.00014792665766236155,
      "loss": 12.3354,
      "step": 13247
    },
    {
      "epoch": 0.7214065467317392,
      "grad_norm": 0.5339434294189568,
      "learning_rate": 0.00014791891796081313,
      "loss": 12.2029,
      "step": 13248
    },
    {
      "epoch": 0.7214610007283222,
      "grad_norm": 0.5497203591337948,
      "learning_rate": 0.00014791117788663918,
      "loss": 12.28,
      "step": 13249
    },
    {
      "epoch": 0.7215154547249052,
      "grad_norm": 0.5884085579228641,
      "learning_rate": 0.00014790343743989988,
      "loss": 12.4046,
      "step": 13250
    },
    {
      "epoch": 0.7215699087214882,
      "grad_norm": 0.5542525975959941,
      "learning_rate": 0.00014789569662065539,
      "loss": 12.239,
      "step": 13251
    },
    {
      "epoch": 0.7216243627180713,
      "grad_norm": 0.5225190570253677,
      "learning_rate": 0.00014788795542896595,
      "loss": 12.2479,
      "step": 13252
    },
    {
      "epoch": 0.7216788167146543,
      "grad_norm": 0.5464471759603579,
      "learning_rate": 0.00014788021386489173,
      "loss": 12.2732,
      "step": 13253
    },
    {
      "epoch": 0.7217332707112373,
      "grad_norm": 0.6026126866474447,
      "learning_rate": 0.00014787247192849296,
      "loss": 12.338,
      "step": 13254
    },
    {
      "epoch": 0.7217877247078203,
      "grad_norm": 0.5680906729215941,
      "learning_rate": 0.0001478647296198298,
      "loss": 12.2654,
      "step": 13255
    },
    {
      "epoch": 0.7218421787044033,
      "grad_norm": 0.5279643367138486,
      "learning_rate": 0.00014785698693896247,
      "loss": 12.281,
      "step": 13256
    },
    {
      "epoch": 0.7218966327009863,
      "grad_norm": 0.5780639737499039,
      "learning_rate": 0.00014784924388595118,
      "loss": 12.1744,
      "step": 13257
    },
    {
      "epoch": 0.7219510866975694,
      "grad_norm": 0.6157191538415512,
      "learning_rate": 0.00014784150046085617,
      "loss": 12.2092,
      "step": 13258
    },
    {
      "epoch": 0.7220055406941523,
      "grad_norm": 0.594849991840615,
      "learning_rate": 0.00014783375666373765,
      "loss": 12.3156,
      "step": 13259
    },
    {
      "epoch": 0.7220599946907353,
      "grad_norm": 0.5610084800496234,
      "learning_rate": 0.00014782601249465577,
      "loss": 12.3049,
      "step": 13260
    },
    {
      "epoch": 0.7221144486873183,
      "grad_norm": 0.5138308536565745,
      "learning_rate": 0.0001478182679536708,
      "loss": 12.1668,
      "step": 13261
    },
    {
      "epoch": 0.7221689026839013,
      "grad_norm": 0.5857148696720267,
      "learning_rate": 0.00014781052304084296,
      "loss": 12.2645,
      "step": 13262
    },
    {
      "epoch": 0.7222233566804843,
      "grad_norm": 0.5869236689159668,
      "learning_rate": 0.00014780277775623248,
      "loss": 12.2645,
      "step": 13263
    },
    {
      "epoch": 0.7222778106770674,
      "grad_norm": 0.5861017227555576,
      "learning_rate": 0.00014779503209989963,
      "loss": 12.1274,
      "step": 13264
    },
    {
      "epoch": 0.7223322646736504,
      "grad_norm": 0.5319448586885308,
      "learning_rate": 0.00014778728607190453,
      "loss": 12.2314,
      "step": 13265
    },
    {
      "epoch": 0.7223867186702334,
      "grad_norm": 0.5311790313561285,
      "learning_rate": 0.0001477795396723075,
      "loss": 12.1801,
      "step": 13266
    },
    {
      "epoch": 0.7224411726668164,
      "grad_norm": 0.6732219693817844,
      "learning_rate": 0.00014777179290116873,
      "loss": 12.3139,
      "step": 13267
    },
    {
      "epoch": 0.7224956266633994,
      "grad_norm": 0.6493652353578988,
      "learning_rate": 0.00014776404575854855,
      "loss": 12.3059,
      "step": 13268
    },
    {
      "epoch": 0.7225500806599824,
      "grad_norm": 0.5937044833129376,
      "learning_rate": 0.0001477562982445071,
      "loss": 12.364,
      "step": 13269
    },
    {
      "epoch": 0.7226045346565655,
      "grad_norm": 0.6198658920216181,
      "learning_rate": 0.00014774855035910468,
      "loss": 12.2827,
      "step": 13270
    },
    {
      "epoch": 0.7226589886531485,
      "grad_norm": 0.5732476439334573,
      "learning_rate": 0.00014774080210240153,
      "loss": 12.2738,
      "step": 13271
    },
    {
      "epoch": 0.7227134426497315,
      "grad_norm": 0.5360260553007256,
      "learning_rate": 0.00014773305347445787,
      "loss": 12.111,
      "step": 13272
    },
    {
      "epoch": 0.7227678966463145,
      "grad_norm": 0.5275770907062345,
      "learning_rate": 0.000147725304475334,
      "loss": 12.2252,
      "step": 13273
    },
    {
      "epoch": 0.7228223506428975,
      "grad_norm": 0.5172824207596717,
      "learning_rate": 0.00014771755510509016,
      "loss": 12.2106,
      "step": 13274
    },
    {
      "epoch": 0.7228768046394806,
      "grad_norm": 0.5721249521602304,
      "learning_rate": 0.0001477098053637866,
      "loss": 12.199,
      "step": 13275
    },
    {
      "epoch": 0.7229312586360636,
      "grad_norm": 0.7191742987674118,
      "learning_rate": 0.0001477020552514836,
      "loss": 12.2984,
      "step": 13276
    },
    {
      "epoch": 0.7229857126326465,
      "grad_norm": 0.6365763140364471,
      "learning_rate": 0.00014769430476824143,
      "loss": 12.2329,
      "step": 13277
    },
    {
      "epoch": 0.7230401666292295,
      "grad_norm": 0.551456274285001,
      "learning_rate": 0.0001476865539141203,
      "loss": 12.2782,
      "step": 13278
    },
    {
      "epoch": 0.7230946206258125,
      "grad_norm": 0.5991760905410665,
      "learning_rate": 0.0001476788026891806,
      "loss": 12.2426,
      "step": 13279
    },
    {
      "epoch": 0.7231490746223955,
      "grad_norm": 0.5544333248898721,
      "learning_rate": 0.0001476710510934825,
      "loss": 12.299,
      "step": 13280
    },
    {
      "epoch": 0.7232035286189786,
      "grad_norm": 0.5394336831800627,
      "learning_rate": 0.0001476632991270863,
      "loss": 12.2804,
      "step": 13281
    },
    {
      "epoch": 0.7232579826155616,
      "grad_norm": 0.6062299792911675,
      "learning_rate": 0.00014765554679005232,
      "loss": 12.2861,
      "step": 13282
    },
    {
      "epoch": 0.7233124366121446,
      "grad_norm": 0.5730035830507569,
      "learning_rate": 0.00014764779408244077,
      "loss": 12.2278,
      "step": 13283
    },
    {
      "epoch": 0.7233668906087276,
      "grad_norm": 0.6156500355967656,
      "learning_rate": 0.00014764004100431202,
      "loss": 12.2724,
      "step": 13284
    },
    {
      "epoch": 0.7234213446053106,
      "grad_norm": 0.6188118186143371,
      "learning_rate": 0.0001476322875557263,
      "loss": 12.3845,
      "step": 13285
    },
    {
      "epoch": 0.7234757986018936,
      "grad_norm": 0.5600796231001508,
      "learning_rate": 0.00014762453373674394,
      "loss": 12.3946,
      "step": 13286
    },
    {
      "epoch": 0.7235302525984767,
      "grad_norm": 0.5966923620651808,
      "learning_rate": 0.00014761677954742525,
      "loss": 12.2449,
      "step": 13287
    },
    {
      "epoch": 0.7235847065950597,
      "grad_norm": 0.6581313433360012,
      "learning_rate": 0.00014760902498783045,
      "loss": 12.2262,
      "step": 13288
    },
    {
      "epoch": 0.7236391605916427,
      "grad_norm": 0.6462196705353955,
      "learning_rate": 0.0001476012700580199,
      "loss": 12.1701,
      "step": 13289
    },
    {
      "epoch": 0.7236936145882257,
      "grad_norm": 0.5686442508026062,
      "learning_rate": 0.0001475935147580539,
      "loss": 12.3473,
      "step": 13290
    },
    {
      "epoch": 0.7237480685848087,
      "grad_norm": 0.6936995965357283,
      "learning_rate": 0.00014758575908799273,
      "loss": 12.161,
      "step": 13291
    },
    {
      "epoch": 0.7238025225813917,
      "grad_norm": 0.5734359477612324,
      "learning_rate": 0.00014757800304789672,
      "loss": 12.2128,
      "step": 13292
    },
    {
      "epoch": 0.7238569765779748,
      "grad_norm": 0.5543159894192236,
      "learning_rate": 0.00014757024663782618,
      "loss": 12.2305,
      "step": 13293
    },
    {
      "epoch": 0.7239114305745578,
      "grad_norm": 0.7523444100322242,
      "learning_rate": 0.00014756248985784145,
      "loss": 12.2388,
      "step": 13294
    },
    {
      "epoch": 0.7239658845711407,
      "grad_norm": 0.6761115842844122,
      "learning_rate": 0.0001475547327080028,
      "loss": 12.5183,
      "step": 13295
    },
    {
      "epoch": 0.7240203385677237,
      "grad_norm": 0.6478389352027981,
      "learning_rate": 0.00014754697518837057,
      "loss": 12.2586,
      "step": 13296
    },
    {
      "epoch": 0.7240747925643067,
      "grad_norm": 0.6431879080137026,
      "learning_rate": 0.0001475392172990051,
      "loss": 12.3238,
      "step": 13297
    },
    {
      "epoch": 0.7241292465608897,
      "grad_norm": 0.5920708792499044,
      "learning_rate": 0.0001475314590399667,
      "loss": 12.2876,
      "step": 13298
    },
    {
      "epoch": 0.7241837005574728,
      "grad_norm": 0.5144758746792878,
      "learning_rate": 0.00014752370041131571,
      "loss": 12.118,
      "step": 13299
    },
    {
      "epoch": 0.7242381545540558,
      "grad_norm": 0.5594016258185207,
      "learning_rate": 0.00014751594141311245,
      "loss": 12.171,
      "step": 13300
    },
    {
      "epoch": 0.7242926085506388,
      "grad_norm": 0.5727034735694334,
      "learning_rate": 0.00014750818204541726,
      "loss": 12.2966,
      "step": 13301
    },
    {
      "epoch": 0.7243470625472218,
      "grad_norm": 0.5423280565884221,
      "learning_rate": 0.00014750042230829048,
      "loss": 12.2054,
      "step": 13302
    },
    {
      "epoch": 0.7244015165438048,
      "grad_norm": 0.6132969743187309,
      "learning_rate": 0.00014749266220179244,
      "loss": 12.2604,
      "step": 13303
    },
    {
      "epoch": 0.7244559705403878,
      "grad_norm": 0.5824401764590079,
      "learning_rate": 0.00014748490172598351,
      "loss": 12.3298,
      "step": 13304
    },
    {
      "epoch": 0.7245104245369709,
      "grad_norm": 0.6718562011950225,
      "learning_rate": 0.000147477140880924,
      "loss": 12.2995,
      "step": 13305
    },
    {
      "epoch": 0.7245648785335539,
      "grad_norm": 0.6296253155808531,
      "learning_rate": 0.0001474693796666743,
      "loss": 12.2741,
      "step": 13306
    },
    {
      "epoch": 0.7246193325301369,
      "grad_norm": 0.5643711473386575,
      "learning_rate": 0.00014746161808329474,
      "loss": 12.3123,
      "step": 13307
    },
    {
      "epoch": 0.7246737865267199,
      "grad_norm": 0.5789895661930375,
      "learning_rate": 0.00014745385613084569,
      "loss": 12.3231,
      "step": 13308
    },
    {
      "epoch": 0.7247282405233029,
      "grad_norm": 0.6649865105288295,
      "learning_rate": 0.0001474460938093875,
      "loss": 12.3596,
      "step": 13309
    },
    {
      "epoch": 0.724782694519886,
      "grad_norm": 0.6178714964656502,
      "learning_rate": 0.00014743833111898052,
      "loss": 12.252,
      "step": 13310
    },
    {
      "epoch": 0.724837148516469,
      "grad_norm": 0.5727155683070335,
      "learning_rate": 0.00014743056805968508,
      "loss": 12.2908,
      "step": 13311
    },
    {
      "epoch": 0.724891602513052,
      "grad_norm": 0.6811695965625274,
      "learning_rate": 0.00014742280463156163,
      "loss": 12.4003,
      "step": 13312
    },
    {
      "epoch": 0.724946056509635,
      "grad_norm": 0.606043872375847,
      "learning_rate": 0.0001474150408346705,
      "loss": 12.3047,
      "step": 13313
    },
    {
      "epoch": 0.7250005105062179,
      "grad_norm": 0.5663335087891896,
      "learning_rate": 0.00014740727666907207,
      "loss": 12.2953,
      "step": 13314
    },
    {
      "epoch": 0.7250549645028009,
      "grad_norm": 0.6290111840184379,
      "learning_rate": 0.00014739951213482667,
      "loss": 12.2735,
      "step": 13315
    },
    {
      "epoch": 0.725109418499384,
      "grad_norm": 0.6244440368051256,
      "learning_rate": 0.0001473917472319947,
      "loss": 12.2807,
      "step": 13316
    },
    {
      "epoch": 0.725163872495967,
      "grad_norm": 0.5204893675027747,
      "learning_rate": 0.00014738398196063662,
      "loss": 12.1904,
      "step": 13317
    },
    {
      "epoch": 0.72521832649255,
      "grad_norm": 0.6421462501017908,
      "learning_rate": 0.00014737621632081272,
      "loss": 12.2922,
      "step": 13318
    },
    {
      "epoch": 0.725272780489133,
      "grad_norm": 0.6330231138389201,
      "learning_rate": 0.0001473684503125834,
      "loss": 12.3048,
      "step": 13319
    },
    {
      "epoch": 0.725327234485716,
      "grad_norm": 0.5362822429633152,
      "learning_rate": 0.0001473606839360091,
      "loss": 12.2355,
      "step": 13320
    },
    {
      "epoch": 0.725381688482299,
      "grad_norm": 0.5748616604842707,
      "learning_rate": 0.00014735291719115016,
      "loss": 12.2656,
      "step": 13321
    },
    {
      "epoch": 0.7254361424788821,
      "grad_norm": 0.6475297889755877,
      "learning_rate": 0.00014734515007806698,
      "loss": 12.3399,
      "step": 13322
    },
    {
      "epoch": 0.7254905964754651,
      "grad_norm": 0.6503247222470766,
      "learning_rate": 0.00014733738259682,
      "loss": 12.3816,
      "step": 13323
    },
    {
      "epoch": 0.7255450504720481,
      "grad_norm": 0.5656483255703779,
      "learning_rate": 0.0001473296147474696,
      "loss": 12.2718,
      "step": 13324
    },
    {
      "epoch": 0.7255995044686311,
      "grad_norm": 0.6695739183253075,
      "learning_rate": 0.00014732184653007616,
      "loss": 12.1371,
      "step": 13325
    },
    {
      "epoch": 0.7256539584652141,
      "grad_norm": 0.6144707598948669,
      "learning_rate": 0.00014731407794470013,
      "loss": 12.2903,
      "step": 13326
    },
    {
      "epoch": 0.7257084124617971,
      "grad_norm": 0.6436012618615664,
      "learning_rate": 0.00014730630899140184,
      "loss": 12.2941,
      "step": 13327
    },
    {
      "epoch": 0.7257628664583802,
      "grad_norm": 0.5990201294905514,
      "learning_rate": 0.00014729853967024176,
      "loss": 12.3202,
      "step": 13328
    },
    {
      "epoch": 0.7258173204549632,
      "grad_norm": 0.558011434881847,
      "learning_rate": 0.00014729076998128038,
      "loss": 12.3215,
      "step": 13329
    },
    {
      "epoch": 0.7258717744515462,
      "grad_norm": 0.5789380298918541,
      "learning_rate": 0.000147282999924578,
      "loss": 12.2409,
      "step": 13330
    },
    {
      "epoch": 0.7259262284481292,
      "grad_norm": 0.6535672422004971,
      "learning_rate": 0.00014727522950019507,
      "loss": 12.2585,
      "step": 13331
    },
    {
      "epoch": 0.7259806824447121,
      "grad_norm": 0.5473404565670977,
      "learning_rate": 0.00014726745870819203,
      "loss": 12.063,
      "step": 13332
    },
    {
      "epoch": 0.7260351364412951,
      "grad_norm": 0.5768166524166972,
      "learning_rate": 0.0001472596875486293,
      "loss": 12.3775,
      "step": 13333
    },
    {
      "epoch": 0.7260895904378782,
      "grad_norm": 0.605122601726206,
      "learning_rate": 0.00014725191602156735,
      "loss": 12.3055,
      "step": 13334
    },
    {
      "epoch": 0.7261440444344612,
      "grad_norm": 0.5977671889604294,
      "learning_rate": 0.00014724414412706655,
      "loss": 12.2939,
      "step": 13335
    },
    {
      "epoch": 0.7261984984310442,
      "grad_norm": 0.5656575048123962,
      "learning_rate": 0.00014723637186518738,
      "loss": 12.2937,
      "step": 13336
    },
    {
      "epoch": 0.7262529524276272,
      "grad_norm": 0.7920585669939834,
      "learning_rate": 0.0001472285992359902,
      "loss": 12.4134,
      "step": 13337
    },
    {
      "epoch": 0.7263074064242102,
      "grad_norm": 0.6288762888571692,
      "learning_rate": 0.0001472208262395356,
      "loss": 12.4197,
      "step": 13338
    },
    {
      "epoch": 0.7263618604207932,
      "grad_norm": 0.5799542013846901,
      "learning_rate": 0.00014721305287588386,
      "loss": 12.2353,
      "step": 13339
    },
    {
      "epoch": 0.7264163144173763,
      "grad_norm": 0.5553553616139497,
      "learning_rate": 0.00014720527914509557,
      "loss": 12.2557,
      "step": 13340
    },
    {
      "epoch": 0.7264707684139593,
      "grad_norm": 0.5912935785724808,
      "learning_rate": 0.00014719750504723107,
      "loss": 12.2559,
      "step": 13341
    },
    {
      "epoch": 0.7265252224105423,
      "grad_norm": 0.5412870648624158,
      "learning_rate": 0.00014718973058235087,
      "loss": 12.0724,
      "step": 13342
    },
    {
      "epoch": 0.7265796764071253,
      "grad_norm": 0.5471026390853213,
      "learning_rate": 0.0001471819557505154,
      "loss": 12.1997,
      "step": 13343
    },
    {
      "epoch": 0.7266341304037083,
      "grad_norm": 0.6702127640155227,
      "learning_rate": 0.0001471741805517851,
      "loss": 12.4182,
      "step": 13344
    },
    {
      "epoch": 0.7266885844002914,
      "grad_norm": 0.583732138964941,
      "learning_rate": 0.00014716640498622054,
      "loss": 12.2193,
      "step": 13345
    },
    {
      "epoch": 0.7267430383968744,
      "grad_norm": 0.571671078235295,
      "learning_rate": 0.00014715862905388205,
      "loss": 12.2459,
      "step": 13346
    },
    {
      "epoch": 0.7267974923934574,
      "grad_norm": 0.7012614857505217,
      "learning_rate": 0.00014715085275483015,
      "loss": 12.2849,
      "step": 13347
    },
    {
      "epoch": 0.7268519463900404,
      "grad_norm": 0.6271694136899435,
      "learning_rate": 0.00014714307608912534,
      "loss": 12.2322,
      "step": 13348
    },
    {
      "epoch": 0.7269064003866234,
      "grad_norm": 0.6199557536061102,
      "learning_rate": 0.00014713529905682805,
      "loss": 12.1896,
      "step": 13349
    },
    {
      "epoch": 0.7269608543832063,
      "grad_norm": 0.7010623743553475,
      "learning_rate": 0.00014712752165799877,
      "loss": 12.4026,
      "step": 13350
    },
    {
      "epoch": 0.7270153083797894,
      "grad_norm": 0.6694734843494442,
      "learning_rate": 0.00014711974389269798,
      "loss": 12.1728,
      "step": 13351
    },
    {
      "epoch": 0.7270697623763724,
      "grad_norm": 0.6172901433277747,
      "learning_rate": 0.00014711196576098616,
      "loss": 12.2305,
      "step": 13352
    },
    {
      "epoch": 0.7271242163729554,
      "grad_norm": 0.6393667883302725,
      "learning_rate": 0.00014710418726292382,
      "loss": 12.2514,
      "step": 13353
    },
    {
      "epoch": 0.7271786703695384,
      "grad_norm": 0.8042984169721533,
      "learning_rate": 0.00014709640839857138,
      "loss": 12.2913,
      "step": 13354
    },
    {
      "epoch": 0.7272331243661214,
      "grad_norm": 0.5605753639003496,
      "learning_rate": 0.00014708862916798938,
      "loss": 12.3283,
      "step": 13355
    },
    {
      "epoch": 0.7272875783627044,
      "grad_norm": 0.8140781294933005,
      "learning_rate": 0.00014708084957123832,
      "loss": 12.388,
      "step": 13356
    },
    {
      "epoch": 0.7273420323592875,
      "grad_norm": 0.6234743849624682,
      "learning_rate": 0.00014707306960837866,
      "loss": 12.2007,
      "step": 13357
    },
    {
      "epoch": 0.7273964863558705,
      "grad_norm": 0.5595606397637554,
      "learning_rate": 0.00014706528927947092,
      "loss": 12.2607,
      "step": 13358
    },
    {
      "epoch": 0.7274509403524535,
      "grad_norm": 0.5624929559642442,
      "learning_rate": 0.00014705750858457557,
      "loss": 12.155,
      "step": 13359
    },
    {
      "epoch": 0.7275053943490365,
      "grad_norm": 0.7250644876285313,
      "learning_rate": 0.0001470497275237532,
      "loss": 12.232,
      "step": 13360
    },
    {
      "epoch": 0.7275598483456195,
      "grad_norm": 0.5341427046039439,
      "learning_rate": 0.0001470419460970642,
      "loss": 12.1898,
      "step": 13361
    },
    {
      "epoch": 0.7276143023422025,
      "grad_norm": 0.6067037064595763,
      "learning_rate": 0.00014703416430456918,
      "loss": 12.2517,
      "step": 13362
    },
    {
      "epoch": 0.7276687563387856,
      "grad_norm": 0.6619249652306931,
      "learning_rate": 0.0001470263821463286,
      "loss": 12.1289,
      "step": 13363
    },
    {
      "epoch": 0.7277232103353686,
      "grad_norm": 0.539140862548324,
      "learning_rate": 0.00014701859962240298,
      "loss": 12.2192,
      "step": 13364
    },
    {
      "epoch": 0.7277776643319516,
      "grad_norm": 0.6013055091812155,
      "learning_rate": 0.00014701081673285286,
      "loss": 12.2707,
      "step": 13365
    },
    {
      "epoch": 0.7278321183285346,
      "grad_norm": 0.5978285453649279,
      "learning_rate": 0.00014700303347773872,
      "loss": 12.2326,
      "step": 13366
    },
    {
      "epoch": 0.7278865723251176,
      "grad_norm": 0.5711855928250466,
      "learning_rate": 0.00014699524985712113,
      "loss": 12.3082,
      "step": 13367
    },
    {
      "epoch": 0.7279410263217005,
      "grad_norm": 0.5499721625163349,
      "learning_rate": 0.0001469874658710606,
      "loss": 12.134,
      "step": 13368
    },
    {
      "epoch": 0.7279954803182836,
      "grad_norm": 0.6450866127583862,
      "learning_rate": 0.00014697968151961764,
      "loss": 12.3181,
      "step": 13369
    },
    {
      "epoch": 0.7280499343148666,
      "grad_norm": 0.6587570915991241,
      "learning_rate": 0.00014697189680285282,
      "loss": 12.1078,
      "step": 13370
    },
    {
      "epoch": 0.7281043883114496,
      "grad_norm": 0.6449739855760339,
      "learning_rate": 0.0001469641117208266,
      "loss": 12.3408,
      "step": 13371
    },
    {
      "epoch": 0.7281588423080326,
      "grad_norm": 0.5962989797445586,
      "learning_rate": 0.00014695632627359962,
      "loss": 12.0901,
      "step": 13372
    },
    {
      "epoch": 0.7282132963046156,
      "grad_norm": 0.6419537874928303,
      "learning_rate": 0.00014694854046123237,
      "loss": 12.3548,
      "step": 13373
    },
    {
      "epoch": 0.7282677503011986,
      "grad_norm": 0.6217723276634614,
      "learning_rate": 0.0001469407542837854,
      "loss": 12.3115,
      "step": 13374
    },
    {
      "epoch": 0.7283222042977817,
      "grad_norm": 0.5143051049374879,
      "learning_rate": 0.00014693296774131924,
      "loss": 12.1744,
      "step": 13375
    },
    {
      "epoch": 0.7283766582943647,
      "grad_norm": 0.7747848484144629,
      "learning_rate": 0.00014692518083389442,
      "loss": 12.3623,
      "step": 13376
    },
    {
      "epoch": 0.7284311122909477,
      "grad_norm": 0.6839726523055348,
      "learning_rate": 0.00014691739356157156,
      "loss": 12.1832,
      "step": 13377
    },
    {
      "epoch": 0.7284855662875307,
      "grad_norm": 0.5831370886270943,
      "learning_rate": 0.00014690960592441118,
      "loss": 12.2775,
      "step": 13378
    },
    {
      "epoch": 0.7285400202841137,
      "grad_norm": 0.6356590097338246,
      "learning_rate": 0.00014690181792247382,
      "loss": 12.2817,
      "step": 13379
    },
    {
      "epoch": 0.7285944742806968,
      "grad_norm": 0.6111939040509912,
      "learning_rate": 0.0001468940295558201,
      "loss": 12.32,
      "step": 13380
    },
    {
      "epoch": 0.7286489282772798,
      "grad_norm": 0.6661048890883065,
      "learning_rate": 0.0001468862408245105,
      "loss": 12.4681,
      "step": 13381
    },
    {
      "epoch": 0.7287033822738628,
      "grad_norm": 0.701286526648347,
      "learning_rate": 0.0001468784517286056,
      "loss": 12.3801,
      "step": 13382
    },
    {
      "epoch": 0.7287578362704458,
      "grad_norm": 0.6065563264179271,
      "learning_rate": 0.00014687066226816607,
      "loss": 12.1662,
      "step": 13383
    },
    {
      "epoch": 0.7288122902670288,
      "grad_norm": 0.6226504736642833,
      "learning_rate": 0.00014686287244325237,
      "loss": 12.2861,
      "step": 13384
    },
    {
      "epoch": 0.7288667442636118,
      "grad_norm": 0.6341344589623714,
      "learning_rate": 0.00014685508225392515,
      "loss": 12.3233,
      "step": 13385
    },
    {
      "epoch": 0.7289211982601949,
      "grad_norm": 0.6283968099363282,
      "learning_rate": 0.00014684729170024493,
      "loss": 12.2036,
      "step": 13386
    },
    {
      "epoch": 0.7289756522567779,
      "grad_norm": 0.5482355338897202,
      "learning_rate": 0.00014683950078227232,
      "loss": 12.1719,
      "step": 13387
    },
    {
      "epoch": 0.7290301062533608,
      "grad_norm": 0.5844528949461678,
      "learning_rate": 0.00014683170950006785,
      "loss": 12.1159,
      "step": 13388
    },
    {
      "epoch": 0.7290845602499438,
      "grad_norm": 0.6758906069619615,
      "learning_rate": 0.0001468239178536922,
      "loss": 12.4005,
      "step": 13389
    },
    {
      "epoch": 0.7291390142465268,
      "grad_norm": 0.5703040967089928,
      "learning_rate": 0.00014681612584320592,
      "loss": 12.2006,
      "step": 13390
    },
    {
      "epoch": 0.7291934682431098,
      "grad_norm": 0.5527985593987613,
      "learning_rate": 0.00014680833346866957,
      "loss": 12.1996,
      "step": 13391
    },
    {
      "epoch": 0.7292479222396929,
      "grad_norm": 0.5777786022154163,
      "learning_rate": 0.00014680054073014378,
      "loss": 12.3138,
      "step": 13392
    },
    {
      "epoch": 0.7293023762362759,
      "grad_norm": 0.546652622965535,
      "learning_rate": 0.00014679274762768911,
      "loss": 12.2746,
      "step": 13393
    },
    {
      "epoch": 0.7293568302328589,
      "grad_norm": 0.5675033988841082,
      "learning_rate": 0.00014678495416136622,
      "loss": 12.2041,
      "step": 13394
    },
    {
      "epoch": 0.7294112842294419,
      "grad_norm": 0.5321126755557474,
      "learning_rate": 0.00014677716033123568,
      "loss": 12.1684,
      "step": 13395
    },
    {
      "epoch": 0.7294657382260249,
      "grad_norm": 0.5728224465252162,
      "learning_rate": 0.0001467693661373581,
      "loss": 12.18,
      "step": 13396
    },
    {
      "epoch": 0.7295201922226079,
      "grad_norm": 0.5844124602179768,
      "learning_rate": 0.00014676157157979407,
      "loss": 12.3234,
      "step": 13397
    },
    {
      "epoch": 0.729574646219191,
      "grad_norm": 0.5655242088204449,
      "learning_rate": 0.00014675377665860422,
      "loss": 12.2359,
      "step": 13398
    },
    {
      "epoch": 0.729629100215774,
      "grad_norm": 0.5145490553683293,
      "learning_rate": 0.00014674598137384915,
      "loss": 12.1504,
      "step": 13399
    },
    {
      "epoch": 0.729683554212357,
      "grad_norm": 0.5206155089772565,
      "learning_rate": 0.0001467381857255895,
      "loss": 12.3104,
      "step": 13400
    },
    {
      "epoch": 0.72973800820894,
      "grad_norm": 0.5732089652040417,
      "learning_rate": 0.0001467303897138859,
      "loss": 12.3735,
      "step": 13401
    },
    {
      "epoch": 0.729792462205523,
      "grad_norm": 0.5893013514173729,
      "learning_rate": 0.00014672259333879892,
      "loss": 12.3674,
      "step": 13402
    },
    {
      "epoch": 0.729846916202106,
      "grad_norm": 0.5453283379493583,
      "learning_rate": 0.00014671479660038921,
      "loss": 12.3758,
      "step": 13403
    },
    {
      "epoch": 0.7299013701986891,
      "grad_norm": 0.5553399861077735,
      "learning_rate": 0.00014670699949871746,
      "loss": 12.2429,
      "step": 13404
    },
    {
      "epoch": 0.729955824195272,
      "grad_norm": 0.5781061733118386,
      "learning_rate": 0.00014669920203384422,
      "loss": 12.325,
      "step": 13405
    },
    {
      "epoch": 0.730010278191855,
      "grad_norm": 0.6196937505905702,
      "learning_rate": 0.00014669140420583017,
      "loss": 12.3664,
      "step": 13406
    },
    {
      "epoch": 0.730064732188438,
      "grad_norm": 0.5884521889356856,
      "learning_rate": 0.0001466836060147359,
      "loss": 12.2777,
      "step": 13407
    },
    {
      "epoch": 0.730119186185021,
      "grad_norm": 0.672981152642552,
      "learning_rate": 0.00014667580746062208,
      "loss": 12.359,
      "step": 13408
    },
    {
      "epoch": 0.7301736401816041,
      "grad_norm": 0.5582101153546333,
      "learning_rate": 0.00014666800854354938,
      "loss": 12.3577,
      "step": 13409
    },
    {
      "epoch": 0.7302280941781871,
      "grad_norm": 0.573828160756465,
      "learning_rate": 0.00014666020926357843,
      "loss": 12.2839,
      "step": 13410
    },
    {
      "epoch": 0.7302825481747701,
      "grad_norm": 0.5411774195844762,
      "learning_rate": 0.00014665240962076983,
      "loss": 12.3829,
      "step": 13411
    },
    {
      "epoch": 0.7303370021713531,
      "grad_norm": 0.5269134935140668,
      "learning_rate": 0.0001466446096151843,
      "loss": 12.1819,
      "step": 13412
    },
    {
      "epoch": 0.7303914561679361,
      "grad_norm": 0.5500051890399976,
      "learning_rate": 0.00014663680924688243,
      "loss": 12.2836,
      "step": 13413
    },
    {
      "epoch": 0.7304459101645191,
      "grad_norm": 0.5691310142421843,
      "learning_rate": 0.00014662900851592493,
      "loss": 12.0934,
      "step": 13414
    },
    {
      "epoch": 0.7305003641611022,
      "grad_norm": 0.5971972214596137,
      "learning_rate": 0.00014662120742237245,
      "loss": 12.2853,
      "step": 13415
    },
    {
      "epoch": 0.7305548181576852,
      "grad_norm": 0.5193671565095754,
      "learning_rate": 0.00014661340596628563,
      "loss": 12.2221,
      "step": 13416
    },
    {
      "epoch": 0.7306092721542682,
      "grad_norm": 0.6140800043494785,
      "learning_rate": 0.00014660560414772516,
      "loss": 12.2681,
      "step": 13417
    },
    {
      "epoch": 0.7306637261508512,
      "grad_norm": 0.5116223750237714,
      "learning_rate": 0.00014659780196675168,
      "loss": 12.0489,
      "step": 13418
    },
    {
      "epoch": 0.7307181801474342,
      "grad_norm": 0.5778048595206212,
      "learning_rate": 0.00014658999942342588,
      "loss": 12.2313,
      "step": 13419
    },
    {
      "epoch": 0.7307726341440172,
      "grad_norm": 0.5833576137200138,
      "learning_rate": 0.00014658219651780843,
      "loss": 12.3763,
      "step": 13420
    },
    {
      "epoch": 0.7308270881406003,
      "grad_norm": 0.5915683602865042,
      "learning_rate": 0.00014657439324996,
      "loss": 12.233,
      "step": 13421
    },
    {
      "epoch": 0.7308815421371833,
      "grad_norm": 0.5454616527970554,
      "learning_rate": 0.0001465665896199413,
      "loss": 12.0943,
      "step": 13422
    },
    {
      "epoch": 0.7309359961337663,
      "grad_norm": 0.5649149517234928,
      "learning_rate": 0.000146558785627813,
      "loss": 12.0765,
      "step": 13423
    },
    {
      "epoch": 0.7309904501303492,
      "grad_norm": 0.5516421296053254,
      "learning_rate": 0.00014655098127363574,
      "loss": 12.167,
      "step": 13424
    },
    {
      "epoch": 0.7310449041269322,
      "grad_norm": 0.5613095485317154,
      "learning_rate": 0.00014654317655747026,
      "loss": 12.3,
      "step": 13425
    },
    {
      "epoch": 0.7310993581235152,
      "grad_norm": 0.6169308152750115,
      "learning_rate": 0.00014653537147937723,
      "loss": 12.2698,
      "step": 13426
    },
    {
      "epoch": 0.7311538121200983,
      "grad_norm": 0.5861048880352672,
      "learning_rate": 0.00014652756603941735,
      "loss": 12.1751,
      "step": 13427
    },
    {
      "epoch": 0.7312082661166813,
      "grad_norm": 0.6425101239536442,
      "learning_rate": 0.00014651976023765133,
      "loss": 12.2381,
      "step": 13428
    },
    {
      "epoch": 0.7312627201132643,
      "grad_norm": 0.5962613060442913,
      "learning_rate": 0.00014651195407413984,
      "loss": 12.1845,
      "step": 13429
    },
    {
      "epoch": 0.7313171741098473,
      "grad_norm": 0.5802762355006588,
      "learning_rate": 0.0001465041475489436,
      "loss": 12.3437,
      "step": 13430
    },
    {
      "epoch": 0.7313716281064303,
      "grad_norm": 0.6089072591955769,
      "learning_rate": 0.00014649634066212327,
      "loss": 12.2565,
      "step": 13431
    },
    {
      "epoch": 0.7314260821030133,
      "grad_norm": 0.5502931701197896,
      "learning_rate": 0.00014648853341373965,
      "loss": 12.2332,
      "step": 13432
    },
    {
      "epoch": 0.7314805360995964,
      "grad_norm": 0.624853396439485,
      "learning_rate": 0.00014648072580385337,
      "loss": 12.1673,
      "step": 13433
    },
    {
      "epoch": 0.7315349900961794,
      "grad_norm": 0.7874743943545011,
      "learning_rate": 0.00014647291783252518,
      "loss": 12.4053,
      "step": 13434
    },
    {
      "epoch": 0.7315894440927624,
      "grad_norm": 0.6302379277319068,
      "learning_rate": 0.0001464651094998158,
      "loss": 12.3989,
      "step": 13435
    },
    {
      "epoch": 0.7316438980893454,
      "grad_norm": 0.5415596819633678,
      "learning_rate": 0.00014645730080578592,
      "loss": 12.1859,
      "step": 13436
    },
    {
      "epoch": 0.7316983520859284,
      "grad_norm": 0.6349985580522457,
      "learning_rate": 0.00014644949175049627,
      "loss": 12.2451,
      "step": 13437
    },
    {
      "epoch": 0.7317528060825114,
      "grad_norm": 0.5329773431001347,
      "learning_rate": 0.0001464416823340076,
      "loss": 12.2649,
      "step": 13438
    },
    {
      "epoch": 0.7318072600790945,
      "grad_norm": 0.6348334862342842,
      "learning_rate": 0.00014643387255638062,
      "loss": 12.2546,
      "step": 13439
    },
    {
      "epoch": 0.7318617140756775,
      "grad_norm": 0.5907952876566426,
      "learning_rate": 0.00014642606241767605,
      "loss": 12.263,
      "step": 13440
    },
    {
      "epoch": 0.7319161680722605,
      "grad_norm": 0.674265035975136,
      "learning_rate": 0.00014641825191795464,
      "loss": 12.3306,
      "step": 13441
    },
    {
      "epoch": 0.7319706220688434,
      "grad_norm": 0.5746427347799835,
      "learning_rate": 0.0001464104410572771,
      "loss": 12.2673,
      "step": 13442
    },
    {
      "epoch": 0.7320250760654264,
      "grad_norm": 0.9009121743624731,
      "learning_rate": 0.0001464026298357042,
      "loss": 12.3275,
      "step": 13443
    },
    {
      "epoch": 0.7320795300620095,
      "grad_norm": 0.8717593714396172,
      "learning_rate": 0.00014639481825329668,
      "loss": 12.4716,
      "step": 13444
    },
    {
      "epoch": 0.7321339840585925,
      "grad_norm": 0.5369226230477494,
      "learning_rate": 0.0001463870063101153,
      "loss": 12.1858,
      "step": 13445
    },
    {
      "epoch": 0.7321884380551755,
      "grad_norm": 0.7186729605792112,
      "learning_rate": 0.0001463791940062207,
      "loss": 12.3621,
      "step": 13446
    },
    {
      "epoch": 0.7322428920517585,
      "grad_norm": 0.6834082893354678,
      "learning_rate": 0.00014637138134167377,
      "loss": 12.2566,
      "step": 13447
    },
    {
      "epoch": 0.7322973460483415,
      "grad_norm": 0.5394081582792316,
      "learning_rate": 0.00014636356831653518,
      "loss": 12.2781,
      "step": 13448
    },
    {
      "epoch": 0.7323518000449245,
      "grad_norm": 0.7195940712977363,
      "learning_rate": 0.0001463557549308657,
      "loss": 12.3361,
      "step": 13449
    },
    {
      "epoch": 0.7324062540415076,
      "grad_norm": 0.6233426938470448,
      "learning_rate": 0.00014634794118472612,
      "loss": 12.3518,
      "step": 13450
    },
    {
      "epoch": 0.7324607080380906,
      "grad_norm": 0.5872558256245961,
      "learning_rate": 0.00014634012707817718,
      "loss": 12.3114,
      "step": 13451
    },
    {
      "epoch": 0.7325151620346736,
      "grad_norm": 0.5707993151414409,
      "learning_rate": 0.00014633231261127963,
      "loss": 12.0597,
      "step": 13452
    },
    {
      "epoch": 0.7325696160312566,
      "grad_norm": 0.750199557350371,
      "learning_rate": 0.0001463244977840942,
      "loss": 12.3008,
      "step": 13453
    },
    {
      "epoch": 0.7326240700278396,
      "grad_norm": 0.5962135364609051,
      "learning_rate": 0.00014631668259668177,
      "loss": 12.2241,
      "step": 13454
    },
    {
      "epoch": 0.7326785240244226,
      "grad_norm": 0.5914875532602344,
      "learning_rate": 0.00014630886704910306,
      "loss": 12.2123,
      "step": 13455
    },
    {
      "epoch": 0.7327329780210057,
      "grad_norm": 0.6519128543769156,
      "learning_rate": 0.00014630105114141878,
      "loss": 12.1622,
      "step": 13456
    },
    {
      "epoch": 0.7327874320175887,
      "grad_norm": 0.5820134384859893,
      "learning_rate": 0.0001462932348736898,
      "loss": 12.169,
      "step": 13457
    },
    {
      "epoch": 0.7328418860141717,
      "grad_norm": 0.6285044893524522,
      "learning_rate": 0.00014628541824597685,
      "loss": 12.3364,
      "step": 13458
    },
    {
      "epoch": 0.7328963400107547,
      "grad_norm": 0.5651507324870145,
      "learning_rate": 0.0001462776012583407,
      "loss": 12.2208,
      "step": 13459
    },
    {
      "epoch": 0.7329507940073376,
      "grad_norm": 0.6813457185451129,
      "learning_rate": 0.0001462697839108422,
      "loss": 12.3105,
      "step": 13460
    },
    {
      "epoch": 0.7330052480039206,
      "grad_norm": 0.6321546183033216,
      "learning_rate": 0.0001462619662035421,
      "loss": 12.298,
      "step": 13461
    },
    {
      "epoch": 0.7330597020005037,
      "grad_norm": 0.5713403422122469,
      "learning_rate": 0.00014625414813650115,
      "loss": 12.1503,
      "step": 13462
    },
    {
      "epoch": 0.7331141559970867,
      "grad_norm": 0.5789797059059206,
      "learning_rate": 0.0001462463297097802,
      "loss": 12.3081,
      "step": 13463
    },
    {
      "epoch": 0.7331686099936697,
      "grad_norm": 0.6189844941170249,
      "learning_rate": 0.00014623851092344006,
      "loss": 12.2846,
      "step": 13464
    },
    {
      "epoch": 0.7332230639902527,
      "grad_norm": 0.6421197750744856,
      "learning_rate": 0.0001462306917775415,
      "loss": 12.302,
      "step": 13465
    },
    {
      "epoch": 0.7332775179868357,
      "grad_norm": 0.6500344081161316,
      "learning_rate": 0.00014622287227214533,
      "loss": 12.0938,
      "step": 13466
    },
    {
      "epoch": 0.7333319719834187,
      "grad_norm": 0.5543130907603288,
      "learning_rate": 0.00014621505240731234,
      "loss": 12.1314,
      "step": 13467
    },
    {
      "epoch": 0.7333864259800018,
      "grad_norm": 0.6276524792591983,
      "learning_rate": 0.00014620723218310334,
      "loss": 12.2113,
      "step": 13468
    },
    {
      "epoch": 0.7334408799765848,
      "grad_norm": 0.5433848208031679,
      "learning_rate": 0.0001461994115995792,
      "loss": 12.2855,
      "step": 13469
    },
    {
      "epoch": 0.7334953339731678,
      "grad_norm": 0.5441465096653444,
      "learning_rate": 0.00014619159065680065,
      "loss": 12.209,
      "step": 13470
    },
    {
      "epoch": 0.7335497879697508,
      "grad_norm": 0.5901166753899868,
      "learning_rate": 0.00014618376935482855,
      "loss": 12.341,
      "step": 13471
    },
    {
      "epoch": 0.7336042419663338,
      "grad_norm": 0.5837915344088275,
      "learning_rate": 0.00014617594769372372,
      "loss": 12.2193,
      "step": 13472
    },
    {
      "epoch": 0.7336586959629168,
      "grad_norm": 0.5186080736607912,
      "learning_rate": 0.000146168125673547,
      "loss": 12.2442,
      "step": 13473
    },
    {
      "epoch": 0.7337131499594999,
      "grad_norm": 0.5407521525804058,
      "learning_rate": 0.00014616030329435913,
      "loss": 12.2598,
      "step": 13474
    },
    {
      "epoch": 0.7337676039560829,
      "grad_norm": 0.4887043957052942,
      "learning_rate": 0.00014615248055622107,
      "loss": 12.1829,
      "step": 13475
    },
    {
      "epoch": 0.7338220579526659,
      "grad_norm": 0.5614679674040877,
      "learning_rate": 0.00014614465745919355,
      "loss": 12.3229,
      "step": 13476
    },
    {
      "epoch": 0.7338765119492489,
      "grad_norm": 0.631059913970582,
      "learning_rate": 0.00014613683400333742,
      "loss": 12.4136,
      "step": 13477
    },
    {
      "epoch": 0.7339309659458318,
      "grad_norm": 0.5837309056326555,
      "learning_rate": 0.00014612901018871356,
      "loss": 12.0118,
      "step": 13478
    },
    {
      "epoch": 0.733985419942415,
      "grad_norm": 0.6021308541180563,
      "learning_rate": 0.00014612118601538276,
      "loss": 12.3842,
      "step": 13479
    },
    {
      "epoch": 0.7340398739389979,
      "grad_norm": 0.6012447827579785,
      "learning_rate": 0.0001461133614834059,
      "loss": 12.3327,
      "step": 13480
    },
    {
      "epoch": 0.7340943279355809,
      "grad_norm": 0.5456836130548256,
      "learning_rate": 0.00014610553659284378,
      "loss": 12.2408,
      "step": 13481
    },
    {
      "epoch": 0.7341487819321639,
      "grad_norm": 0.5359770041758793,
      "learning_rate": 0.0001460977113437573,
      "loss": 12.0626,
      "step": 13482
    },
    {
      "epoch": 0.7342032359287469,
      "grad_norm": 0.5577707378911068,
      "learning_rate": 0.00014608988573620727,
      "loss": 12.3124,
      "step": 13483
    },
    {
      "epoch": 0.7342576899253299,
      "grad_norm": 0.5949322494568023,
      "learning_rate": 0.00014608205977025457,
      "loss": 12.1902,
      "step": 13484
    },
    {
      "epoch": 0.734312143921913,
      "grad_norm": 0.5502749814129778,
      "learning_rate": 0.00014607423344596004,
      "loss": 12.2777,
      "step": 13485
    },
    {
      "epoch": 0.734366597918496,
      "grad_norm": 0.632209872094979,
      "learning_rate": 0.00014606640676338457,
      "loss": 12.4037,
      "step": 13486
    },
    {
      "epoch": 0.734421051915079,
      "grad_norm": 0.557195963153172,
      "learning_rate": 0.00014605857972258896,
      "loss": 12.391,
      "step": 13487
    },
    {
      "epoch": 0.734475505911662,
      "grad_norm": 0.5943557669758027,
      "learning_rate": 0.00014605075232363413,
      "loss": 12.217,
      "step": 13488
    },
    {
      "epoch": 0.734529959908245,
      "grad_norm": 0.5943023457895839,
      "learning_rate": 0.0001460429245665809,
      "loss": 12.3385,
      "step": 13489
    },
    {
      "epoch": 0.734584413904828,
      "grad_norm": 0.7428476430130446,
      "learning_rate": 0.0001460350964514902,
      "loss": 12.2624,
      "step": 13490
    },
    {
      "epoch": 0.7346388679014111,
      "grad_norm": 0.5815614808212614,
      "learning_rate": 0.0001460272679784229,
      "loss": 12.2735,
      "step": 13491
    },
    {
      "epoch": 0.7346933218979941,
      "grad_norm": 0.5817982523327593,
      "learning_rate": 0.00014601943914743977,
      "loss": 12.332,
      "step": 13492
    },
    {
      "epoch": 0.7347477758945771,
      "grad_norm": 0.6430540982028815,
      "learning_rate": 0.00014601160995860178,
      "loss": 12.2342,
      "step": 13493
    },
    {
      "epoch": 0.7348022298911601,
      "grad_norm": 0.5849969352651594,
      "learning_rate": 0.00014600378041196982,
      "loss": 12.2903,
      "step": 13494
    },
    {
      "epoch": 0.7348566838877431,
      "grad_norm": 0.5780107159072119,
      "learning_rate": 0.00014599595050760475,
      "loss": 12.2955,
      "step": 13495
    },
    {
      "epoch": 0.734911137884326,
      "grad_norm": 0.5634969389091683,
      "learning_rate": 0.00014598812024556746,
      "loss": 12.2938,
      "step": 13496
    },
    {
      "epoch": 0.7349655918809092,
      "grad_norm": 0.5658156766141947,
      "learning_rate": 0.0001459802896259188,
      "loss": 12.2406,
      "step": 13497
    },
    {
      "epoch": 0.7350200458774921,
      "grad_norm": 0.5327649821036103,
      "learning_rate": 0.00014597245864871974,
      "loss": 12.3137,
      "step": 13498
    },
    {
      "epoch": 0.7350744998740751,
      "grad_norm": 0.5806255240789654,
      "learning_rate": 0.0001459646273140311,
      "loss": 12.2654,
      "step": 13499
    },
    {
      "epoch": 0.7351289538706581,
      "grad_norm": 0.6076178268505733,
      "learning_rate": 0.00014595679562191382,
      "loss": 12.1846,
      "step": 13500
    },
    {
      "epoch": 0.7351834078672411,
      "grad_norm": 0.5687460574489392,
      "learning_rate": 0.0001459489635724288,
      "loss": 12.1725,
      "step": 13501
    },
    {
      "epoch": 0.7352378618638241,
      "grad_norm": 0.8070955169800478,
      "learning_rate": 0.0001459411311656369,
      "loss": 12.2549,
      "step": 13502
    },
    {
      "epoch": 0.7352923158604072,
      "grad_norm": 0.6600330966428078,
      "learning_rate": 0.00014593329840159908,
      "loss": 12.2134,
      "step": 13503
    },
    {
      "epoch": 0.7353467698569902,
      "grad_norm": 0.5385361695723251,
      "learning_rate": 0.00014592546528037625,
      "loss": 11.9943,
      "step": 13504
    },
    {
      "epoch": 0.7354012238535732,
      "grad_norm": 0.5557788743641016,
      "learning_rate": 0.00014591763180202928,
      "loss": 12.2433,
      "step": 13505
    },
    {
      "epoch": 0.7354556778501562,
      "grad_norm": 0.5609249760965059,
      "learning_rate": 0.00014590979796661913,
      "loss": 12.0992,
      "step": 13506
    },
    {
      "epoch": 0.7355101318467392,
      "grad_norm": 0.5638842934128018,
      "learning_rate": 0.00014590196377420667,
      "loss": 12.2798,
      "step": 13507
    },
    {
      "epoch": 0.7355645858433222,
      "grad_norm": 0.5813055251948805,
      "learning_rate": 0.0001458941292248528,
      "loss": 12.4461,
      "step": 13508
    },
    {
      "epoch": 0.7356190398399053,
      "grad_norm": 0.6686401426161527,
      "learning_rate": 0.00014588629431861857,
      "loss": 12.4359,
      "step": 13509
    },
    {
      "epoch": 0.7356734938364883,
      "grad_norm": 0.5796752659915911,
      "learning_rate": 0.00014587845905556478,
      "loss": 12.2123,
      "step": 13510
    },
    {
      "epoch": 0.7357279478330713,
      "grad_norm": 0.5726235025614373,
      "learning_rate": 0.0001458706234357524,
      "loss": 12.1894,
      "step": 13511
    },
    {
      "epoch": 0.7357824018296543,
      "grad_norm": 0.5743503966196527,
      "learning_rate": 0.00014586278745924234,
      "loss": 12.209,
      "step": 13512
    },
    {
      "epoch": 0.7358368558262373,
      "grad_norm": 0.6588986599637868,
      "learning_rate": 0.00014585495112609558,
      "loss": 12.3586,
      "step": 13513
    },
    {
      "epoch": 0.7358913098228204,
      "grad_norm": 0.5738888272484045,
      "learning_rate": 0.00014584711443637298,
      "loss": 12.2194,
      "step": 13514
    },
    {
      "epoch": 0.7359457638194034,
      "grad_norm": 0.5493036599841871,
      "learning_rate": 0.00014583927739013558,
      "loss": 12.1909,
      "step": 13515
    },
    {
      "epoch": 0.7360002178159863,
      "grad_norm": 0.5487234894005516,
      "learning_rate": 0.00014583143998744426,
      "loss": 12.2235,
      "step": 13516
    },
    {
      "epoch": 0.7360546718125693,
      "grad_norm": 0.5138716167966478,
      "learning_rate": 0.00014582360222835998,
      "loss": 12.2678,
      "step": 13517
    },
    {
      "epoch": 0.7361091258091523,
      "grad_norm": 0.5461539491689609,
      "learning_rate": 0.0001458157641129437,
      "loss": 12.2597,
      "step": 13518
    },
    {
      "epoch": 0.7361635798057353,
      "grad_norm": 0.5466894629268688,
      "learning_rate": 0.0001458079256412563,
      "loss": 12.2962,
      "step": 13519
    },
    {
      "epoch": 0.7362180338023184,
      "grad_norm": 0.5822005527745485,
      "learning_rate": 0.00014580008681335885,
      "loss": 12.2964,
      "step": 13520
    },
    {
      "epoch": 0.7362724877989014,
      "grad_norm": 0.5800259077540941,
      "learning_rate": 0.00014579224762931224,
      "loss": 12.2507,
      "step": 13521
    },
    {
      "epoch": 0.7363269417954844,
      "grad_norm": 0.6048416669196667,
      "learning_rate": 0.0001457844080891774,
      "loss": 12.1804,
      "step": 13522
    },
    {
      "epoch": 0.7363813957920674,
      "grad_norm": 0.5683179294002261,
      "learning_rate": 0.00014577656819301534,
      "loss": 12.276,
      "step": 13523
    },
    {
      "epoch": 0.7364358497886504,
      "grad_norm": 0.6042767266846831,
      "learning_rate": 0.000145768727940887,
      "loss": 12.3961,
      "step": 13524
    },
    {
      "epoch": 0.7364903037852334,
      "grad_norm": 0.5971000332059202,
      "learning_rate": 0.00014576088733285334,
      "loss": 12.1241,
      "step": 13525
    },
    {
      "epoch": 0.7365447577818165,
      "grad_norm": 0.5960415410498545,
      "learning_rate": 0.00014575304636897538,
      "loss": 12.2755,
      "step": 13526
    },
    {
      "epoch": 0.7365992117783995,
      "grad_norm": 0.5501720767126754,
      "learning_rate": 0.00014574520504931403,
      "loss": 12.2194,
      "step": 13527
    },
    {
      "epoch": 0.7366536657749825,
      "grad_norm": 0.5525318178591357,
      "learning_rate": 0.00014573736337393031,
      "loss": 12.2742,
      "step": 13528
    },
    {
      "epoch": 0.7367081197715655,
      "grad_norm": 0.5931735852925304,
      "learning_rate": 0.0001457295213428852,
      "loss": 12.2892,
      "step": 13529
    },
    {
      "epoch": 0.7367625737681485,
      "grad_norm": 0.574905675319439,
      "learning_rate": 0.00014572167895623962,
      "loss": 12.2478,
      "step": 13530
    },
    {
      "epoch": 0.7368170277647315,
      "grad_norm": 0.5614724178224977,
      "learning_rate": 0.00014571383621405463,
      "loss": 12.2513,
      "step": 13531
    },
    {
      "epoch": 0.7368714817613146,
      "grad_norm": 0.5164268210770245,
      "learning_rate": 0.0001457059931163912,
      "loss": 12.3159,
      "step": 13532
    },
    {
      "epoch": 0.7369259357578976,
      "grad_norm": 0.5551055122116084,
      "learning_rate": 0.00014569814966331027,
      "loss": 12.2811,
      "step": 13533
    },
    {
      "epoch": 0.7369803897544805,
      "grad_norm": 0.7238589607973043,
      "learning_rate": 0.00014569030585487286,
      "loss": 12.2553,
      "step": 13534
    },
    {
      "epoch": 0.7370348437510635,
      "grad_norm": 0.6121042465176388,
      "learning_rate": 0.00014568246169114,
      "loss": 12.2991,
      "step": 13535
    },
    {
      "epoch": 0.7370892977476465,
      "grad_norm": 0.5443821640990063,
      "learning_rate": 0.00014567461717217262,
      "loss": 12.2379,
      "step": 13536
    },
    {
      "epoch": 0.7371437517442295,
      "grad_norm": 0.5899600381512372,
      "learning_rate": 0.00014566677229803178,
      "loss": 12.2119,
      "step": 13537
    },
    {
      "epoch": 0.7371982057408126,
      "grad_norm": 0.5833225779741186,
      "learning_rate": 0.00014565892706877847,
      "loss": 12.3063,
      "step": 13538
    },
    {
      "epoch": 0.7372526597373956,
      "grad_norm": 0.5632297981423725,
      "learning_rate": 0.00014565108148447366,
      "loss": 12.2568,
      "step": 13539
    },
    {
      "epoch": 0.7373071137339786,
      "grad_norm": 0.5456236956874208,
      "learning_rate": 0.0001456432355451784,
      "loss": 12.2761,
      "step": 13540
    },
    {
      "epoch": 0.7373615677305616,
      "grad_norm": 0.5495704263049151,
      "learning_rate": 0.00014563538925095368,
      "loss": 12.2297,
      "step": 13541
    },
    {
      "epoch": 0.7374160217271446,
      "grad_norm": 0.5689853068119046,
      "learning_rate": 0.0001456275426018605,
      "loss": 12.2283,
      "step": 13542
    },
    {
      "epoch": 0.7374704757237277,
      "grad_norm": 0.5180836306877948,
      "learning_rate": 0.00014561969559795995,
      "loss": 12.1401,
      "step": 13543
    },
    {
      "epoch": 0.7375249297203107,
      "grad_norm": 0.5718271747766805,
      "learning_rate": 0.00014561184823931296,
      "loss": 12.3157,
      "step": 13544
    },
    {
      "epoch": 0.7375793837168937,
      "grad_norm": 0.5746504330019291,
      "learning_rate": 0.0001456040005259806,
      "loss": 12.1731,
      "step": 13545
    },
    {
      "epoch": 0.7376338377134767,
      "grad_norm": 0.6065461245363518,
      "learning_rate": 0.00014559615245802386,
      "loss": 12.352,
      "step": 13546
    },
    {
      "epoch": 0.7376882917100597,
      "grad_norm": 0.5784360166514776,
      "learning_rate": 0.00014558830403550382,
      "loss": 12.3759,
      "step": 13547
    },
    {
      "epoch": 0.7377427457066427,
      "grad_norm": 0.5602189288940183,
      "learning_rate": 0.00014558045525848147,
      "loss": 12.1386,
      "step": 13548
    },
    {
      "epoch": 0.7377971997032258,
      "grad_norm": 0.5670399784838877,
      "learning_rate": 0.00014557260612701786,
      "loss": 12.2799,
      "step": 13549
    },
    {
      "epoch": 0.7378516536998088,
      "grad_norm": 0.5896749388285503,
      "learning_rate": 0.00014556475664117405,
      "loss": 12.3304,
      "step": 13550
    },
    {
      "epoch": 0.7379061076963918,
      "grad_norm": 0.5927587892809875,
      "learning_rate": 0.00014555690680101102,
      "loss": 12.3231,
      "step": 13551
    },
    {
      "epoch": 0.7379605616929747,
      "grad_norm": 0.5778156056704327,
      "learning_rate": 0.00014554905660658983,
      "loss": 12.2669,
      "step": 13552
    },
    {
      "epoch": 0.7380150156895577,
      "grad_norm": 0.6212925554269098,
      "learning_rate": 0.00014554120605797156,
      "loss": 12.268,
      "step": 13553
    },
    {
      "epoch": 0.7380694696861407,
      "grad_norm": 0.5703826293906922,
      "learning_rate": 0.0001455333551552172,
      "loss": 12.1634,
      "step": 13554
    },
    {
      "epoch": 0.7381239236827238,
      "grad_norm": 0.5892926596679476,
      "learning_rate": 0.00014552550389838791,
      "loss": 12.362,
      "step": 13555
    },
    {
      "epoch": 0.7381783776793068,
      "grad_norm": 0.6057136352760333,
      "learning_rate": 0.00014551765228754463,
      "loss": 12.2202,
      "step": 13556
    },
    {
      "epoch": 0.7382328316758898,
      "grad_norm": 0.5703198221862579,
      "learning_rate": 0.00014550980032274841,
      "loss": 12.2498,
      "step": 13557
    },
    {
      "epoch": 0.7382872856724728,
      "grad_norm": 0.627692916894544,
      "learning_rate": 0.00014550194800406037,
      "loss": 12.3628,
      "step": 13558
    },
    {
      "epoch": 0.7383417396690558,
      "grad_norm": 0.6016477926686616,
      "learning_rate": 0.0001454940953315416,
      "loss": 12.1268,
      "step": 13559
    },
    {
      "epoch": 0.7383961936656388,
      "grad_norm": 0.6573337693225969,
      "learning_rate": 0.00014548624230525307,
      "loss": 12.3016,
      "step": 13560
    },
    {
      "epoch": 0.7384506476622219,
      "grad_norm": 0.575658458983648,
      "learning_rate": 0.00014547838892525592,
      "loss": 12.2689,
      "step": 13561
    },
    {
      "epoch": 0.7385051016588049,
      "grad_norm": 0.583711733930323,
      "learning_rate": 0.00014547053519161116,
      "loss": 12.2254,
      "step": 13562
    },
    {
      "epoch": 0.7385595556553879,
      "grad_norm": 0.6919837322214447,
      "learning_rate": 0.0001454626811043799,
      "loss": 12.4156,
      "step": 13563
    },
    {
      "epoch": 0.7386140096519709,
      "grad_norm": 0.5667989851626005,
      "learning_rate": 0.0001454548266636232,
      "loss": 12.2173,
      "step": 13564
    },
    {
      "epoch": 0.7386684636485539,
      "grad_norm": 0.5967847312016634,
      "learning_rate": 0.00014544697186940218,
      "loss": 12.3047,
      "step": 13565
    },
    {
      "epoch": 0.7387229176451369,
      "grad_norm": 0.5783482130215254,
      "learning_rate": 0.00014543911672177786,
      "loss": 12.393,
      "step": 13566
    },
    {
      "epoch": 0.73877737164172,
      "grad_norm": 0.6326278799926038,
      "learning_rate": 0.00014543126122081138,
      "loss": 12.3051,
      "step": 13567
    },
    {
      "epoch": 0.738831825638303,
      "grad_norm": 0.5975802047422007,
      "learning_rate": 0.00014542340536656374,
      "loss": 12.1247,
      "step": 13568
    },
    {
      "epoch": 0.738886279634886,
      "grad_norm": 0.6201015768827072,
      "learning_rate": 0.0001454155491590961,
      "loss": 12.2745,
      "step": 13569
    },
    {
      "epoch": 0.738940733631469,
      "grad_norm": 0.5977105967525396,
      "learning_rate": 0.00014540769259846953,
      "loss": 12.3551,
      "step": 13570
    },
    {
      "epoch": 0.7389951876280519,
      "grad_norm": 0.5992047393148686,
      "learning_rate": 0.00014539983568474517,
      "loss": 12.2941,
      "step": 13571
    },
    {
      "epoch": 0.7390496416246349,
      "grad_norm": 0.5986098983608372,
      "learning_rate": 0.00014539197841798403,
      "loss": 12.2243,
      "step": 13572
    },
    {
      "epoch": 0.739104095621218,
      "grad_norm": 0.562071760476364,
      "learning_rate": 0.00014538412079824728,
      "loss": 12.3107,
      "step": 13573
    },
    {
      "epoch": 0.739158549617801,
      "grad_norm": 0.5992289735675804,
      "learning_rate": 0.00014537626282559596,
      "loss": 12.1891,
      "step": 13574
    },
    {
      "epoch": 0.739213003614384,
      "grad_norm": 0.5848912184261011,
      "learning_rate": 0.00014536840450009124,
      "loss": 12.1678,
      "step": 13575
    },
    {
      "epoch": 0.739267457610967,
      "grad_norm": 0.5714703977133936,
      "learning_rate": 0.0001453605458217942,
      "loss": 12.2443,
      "step": 13576
    },
    {
      "epoch": 0.73932191160755,
      "grad_norm": 0.5760218558487907,
      "learning_rate": 0.00014535268679076595,
      "loss": 12.3124,
      "step": 13577
    },
    {
      "epoch": 0.7393763656041331,
      "grad_norm": 0.6493899149913374,
      "learning_rate": 0.00014534482740706758,
      "loss": 12.4153,
      "step": 13578
    },
    {
      "epoch": 0.7394308196007161,
      "grad_norm": 0.5886954646525789,
      "learning_rate": 0.00014533696767076023,
      "loss": 12.331,
      "step": 13579
    },
    {
      "epoch": 0.7394852735972991,
      "grad_norm": 0.5521788953301661,
      "learning_rate": 0.00014532910758190503,
      "loss": 12.2524,
      "step": 13580
    },
    {
      "epoch": 0.7395397275938821,
      "grad_norm": 0.5263278183981711,
      "learning_rate": 0.0001453212471405631,
      "loss": 12.3494,
      "step": 13581
    },
    {
      "epoch": 0.7395941815904651,
      "grad_norm": 0.5995637429608626,
      "learning_rate": 0.00014531338634679553,
      "loss": 12.2439,
      "step": 13582
    },
    {
      "epoch": 0.7396486355870481,
      "grad_norm": 0.6126319407071071,
      "learning_rate": 0.00014530552520066348,
      "loss": 12.3076,
      "step": 13583
    },
    {
      "epoch": 0.7397030895836312,
      "grad_norm": 0.5829210288313339,
      "learning_rate": 0.00014529766370222807,
      "loss": 12.1716,
      "step": 13584
    },
    {
      "epoch": 0.7397575435802142,
      "grad_norm": 0.5936945728405048,
      "learning_rate": 0.0001452898018515504,
      "loss": 12.2531,
      "step": 13585
    },
    {
      "epoch": 0.7398119975767972,
      "grad_norm": 0.5985701987761367,
      "learning_rate": 0.00014528193964869168,
      "loss": 12.1612,
      "step": 13586
    },
    {
      "epoch": 0.7398664515733802,
      "grad_norm": 0.5641246242590785,
      "learning_rate": 0.00014527407709371298,
      "loss": 12.1089,
      "step": 13587
    },
    {
      "epoch": 0.7399209055699631,
      "grad_norm": 0.6151635949583258,
      "learning_rate": 0.00014526621418667546,
      "loss": 12.2131,
      "step": 13588
    },
    {
      "epoch": 0.7399753595665461,
      "grad_norm": 0.6159449104099786,
      "learning_rate": 0.0001452583509276403,
      "loss": 12.302,
      "step": 13589
    },
    {
      "epoch": 0.7400298135631292,
      "grad_norm": 0.6151269426786296,
      "learning_rate": 0.00014525048731666858,
      "loss": 12.2723,
      "step": 13590
    },
    {
      "epoch": 0.7400842675597122,
      "grad_norm": 0.657999967620929,
      "learning_rate": 0.00014524262335382149,
      "loss": 12.1819,
      "step": 13591
    },
    {
      "epoch": 0.7401387215562952,
      "grad_norm": 0.5417844327833512,
      "learning_rate": 0.00014523475903916016,
      "loss": 12.2666,
      "step": 13592
    },
    {
      "epoch": 0.7401931755528782,
      "grad_norm": 0.659413396793381,
      "learning_rate": 0.00014522689437274577,
      "loss": 12.3194,
      "step": 13593
    },
    {
      "epoch": 0.7402476295494612,
      "grad_norm": 0.6522191671164157,
      "learning_rate": 0.00014521902935463947,
      "loss": 12.2682,
      "step": 13594
    },
    {
      "epoch": 0.7403020835460442,
      "grad_norm": 0.5712575409989621,
      "learning_rate": 0.00014521116398490241,
      "loss": 12.2528,
      "step": 13595
    },
    {
      "epoch": 0.7403565375426273,
      "grad_norm": 0.5925659529071102,
      "learning_rate": 0.00014520329826359576,
      "loss": 12.1879,
      "step": 13596
    },
    {
      "epoch": 0.7404109915392103,
      "grad_norm": 0.6085032636756643,
      "learning_rate": 0.00014519543219078068,
      "loss": 12.2377,
      "step": 13597
    },
    {
      "epoch": 0.7404654455357933,
      "grad_norm": 0.548249484685204,
      "learning_rate": 0.00014518756576651834,
      "loss": 12.3519,
      "step": 13598
    },
    {
      "epoch": 0.7405198995323763,
      "grad_norm": 0.5862348515271124,
      "learning_rate": 0.0001451796989908699,
      "loss": 12.3024,
      "step": 13599
    },
    {
      "epoch": 0.7405743535289593,
      "grad_norm": 0.5617435120340949,
      "learning_rate": 0.00014517183186389657,
      "loss": 12.3191,
      "step": 13600
    },
    {
      "epoch": 0.7406288075255423,
      "grad_norm": 0.5561225287567393,
      "learning_rate": 0.00014516396438565948,
      "loss": 12.2676,
      "step": 13601
    },
    {
      "epoch": 0.7406832615221254,
      "grad_norm": 0.6384067084986527,
      "learning_rate": 0.00014515609655621985,
      "loss": 12.3782,
      "step": 13602
    },
    {
      "epoch": 0.7407377155187084,
      "grad_norm": 0.530157034517588,
      "learning_rate": 0.00014514822837563882,
      "loss": 12.2542,
      "step": 13603
    },
    {
      "epoch": 0.7407921695152914,
      "grad_norm": 0.5767821558742593,
      "learning_rate": 0.00014514035984397757,
      "loss": 12.2971,
      "step": 13604
    },
    {
      "epoch": 0.7408466235118744,
      "grad_norm": 0.6036355364934566,
      "learning_rate": 0.00014513249096129735,
      "loss": 12.2806,
      "step": 13605
    },
    {
      "epoch": 0.7409010775084574,
      "grad_norm": 0.6545513956221317,
      "learning_rate": 0.0001451246217276593,
      "loss": 12.0547,
      "step": 13606
    },
    {
      "epoch": 0.7409555315050403,
      "grad_norm": 0.670134175781721,
      "learning_rate": 0.00014511675214312462,
      "loss": 12.1504,
      "step": 13607
    },
    {
      "epoch": 0.7410099855016234,
      "grad_norm": 0.6100622110757251,
      "learning_rate": 0.00014510888220775454,
      "loss": 12.2609,
      "step": 13608
    },
    {
      "epoch": 0.7410644394982064,
      "grad_norm": 0.8321259426406292,
      "learning_rate": 0.00014510101192161018,
      "loss": 12.1772,
      "step": 13609
    },
    {
      "epoch": 0.7411188934947894,
      "grad_norm": 0.5704049760476888,
      "learning_rate": 0.00014509314128475283,
      "loss": 12.2398,
      "step": 13610
    },
    {
      "epoch": 0.7411733474913724,
      "grad_norm": 0.5847939629309116,
      "learning_rate": 0.00014508527029724366,
      "loss": 12.2684,
      "step": 13611
    },
    {
      "epoch": 0.7412278014879554,
      "grad_norm": 0.5778186625862041,
      "learning_rate": 0.00014507739895914382,
      "loss": 12.2927,
      "step": 13612
    },
    {
      "epoch": 0.7412822554845385,
      "grad_norm": 0.6271163582322844,
      "learning_rate": 0.00014506952727051458,
      "loss": 12.2311,
      "step": 13613
    },
    {
      "epoch": 0.7413367094811215,
      "grad_norm": 0.569653045038958,
      "learning_rate": 0.00014506165523141712,
      "loss": 12.28,
      "step": 13614
    },
    {
      "epoch": 0.7413911634777045,
      "grad_norm": 0.6158185408598926,
      "learning_rate": 0.0001450537828419127,
      "loss": 12.2618,
      "step": 13615
    },
    {
      "epoch": 0.7414456174742875,
      "grad_norm": 0.6184929667207226,
      "learning_rate": 0.0001450459101020625,
      "loss": 12.2997,
      "step": 13616
    },
    {
      "epoch": 0.7415000714708705,
      "grad_norm": 0.5651499843666569,
      "learning_rate": 0.00014503803701192776,
      "loss": 12.228,
      "step": 13617
    },
    {
      "epoch": 0.7415545254674535,
      "grad_norm": 0.5706628227808465,
      "learning_rate": 0.00014503016357156969,
      "loss": 12.264,
      "step": 13618
    },
    {
      "epoch": 0.7416089794640366,
      "grad_norm": 0.5585073128263942,
      "learning_rate": 0.0001450222897810495,
      "loss": 12.2749,
      "step": 13619
    },
    {
      "epoch": 0.7416634334606196,
      "grad_norm": 0.6289826168740338,
      "learning_rate": 0.00014501441564042847,
      "loss": 12.3302,
      "step": 13620
    },
    {
      "epoch": 0.7417178874572026,
      "grad_norm": 0.5512294887972594,
      "learning_rate": 0.00014500654114976778,
      "loss": 12.2143,
      "step": 13621
    },
    {
      "epoch": 0.7417723414537856,
      "grad_norm": 0.6371616506621154,
      "learning_rate": 0.00014499866630912866,
      "loss": 12.4681,
      "step": 13622
    },
    {
      "epoch": 0.7418267954503686,
      "grad_norm": 0.5869525221861992,
      "learning_rate": 0.00014499079111857235,
      "loss": 12.212,
      "step": 13623
    },
    {
      "epoch": 0.7418812494469516,
      "grad_norm": 0.5600136938066297,
      "learning_rate": 0.00014498291557816012,
      "loss": 12.2693,
      "step": 13624
    },
    {
      "epoch": 0.7419357034435347,
      "grad_norm": 0.6598005867431953,
      "learning_rate": 0.00014497503968795324,
      "loss": 12.3359,
      "step": 13625
    },
    {
      "epoch": 0.7419901574401176,
      "grad_norm": 0.5939656040106014,
      "learning_rate": 0.00014496716344801288,
      "loss": 12.2251,
      "step": 13626
    },
    {
      "epoch": 0.7420446114367006,
      "grad_norm": 0.6090690177906303,
      "learning_rate": 0.0001449592868584003,
      "loss": 12.2218,
      "step": 13627
    },
    {
      "epoch": 0.7420990654332836,
      "grad_norm": 0.5774584014151134,
      "learning_rate": 0.00014495140991917674,
      "loss": 12.3191,
      "step": 13628
    },
    {
      "epoch": 0.7421535194298666,
      "grad_norm": 0.6263921067003643,
      "learning_rate": 0.0001449435326304035,
      "loss": 12.3032,
      "step": 13629
    },
    {
      "epoch": 0.7422079734264496,
      "grad_norm": 0.7222604042773875,
      "learning_rate": 0.00014493565499214183,
      "loss": 12.237,
      "step": 13630
    },
    {
      "epoch": 0.7422624274230327,
      "grad_norm": 0.6076500489786401,
      "learning_rate": 0.00014492777700445296,
      "loss": 12.3552,
      "step": 13631
    },
    {
      "epoch": 0.7423168814196157,
      "grad_norm": 0.6058780080649863,
      "learning_rate": 0.0001449198986673982,
      "loss": 12.2474,
      "step": 13632
    },
    {
      "epoch": 0.7423713354161987,
      "grad_norm": 0.5490641811425729,
      "learning_rate": 0.00014491201998103874,
      "loss": 12.2425,
      "step": 13633
    },
    {
      "epoch": 0.7424257894127817,
      "grad_norm": 0.5456156945385467,
      "learning_rate": 0.00014490414094543589,
      "loss": 12.1875,
      "step": 13634
    },
    {
      "epoch": 0.7424802434093647,
      "grad_norm": 0.6296439828779367,
      "learning_rate": 0.00014489626156065087,
      "loss": 12.244,
      "step": 13635
    },
    {
      "epoch": 0.7425346974059477,
      "grad_norm": 0.6227359546068482,
      "learning_rate": 0.00014488838182674503,
      "loss": 12.2818,
      "step": 13636
    },
    {
      "epoch": 0.7425891514025308,
      "grad_norm": 0.5959937363619823,
      "learning_rate": 0.00014488050174377962,
      "loss": 12.3168,
      "step": 13637
    },
    {
      "epoch": 0.7426436053991138,
      "grad_norm": 0.6016861050673398,
      "learning_rate": 0.00014487262131181587,
      "loss": 12.333,
      "step": 13638
    },
    {
      "epoch": 0.7426980593956968,
      "grad_norm": 0.5950905794185968,
      "learning_rate": 0.0001448647405309151,
      "loss": 12.3405,
      "step": 13639
    },
    {
      "epoch": 0.7427525133922798,
      "grad_norm": 0.5464111091884458,
      "learning_rate": 0.0001448568594011386,
      "loss": 12.1223,
      "step": 13640
    },
    {
      "epoch": 0.7428069673888628,
      "grad_norm": 0.7861765047156204,
      "learning_rate": 0.0001448489779225476,
      "loss": 12.2474,
      "step": 13641
    },
    {
      "epoch": 0.7428614213854458,
      "grad_norm": 0.6003190220576142,
      "learning_rate": 0.00014484109609520345,
      "loss": 12.2989,
      "step": 13642
    },
    {
      "epoch": 0.7429158753820289,
      "grad_norm": 0.5756507794999683,
      "learning_rate": 0.00014483321391916746,
      "loss": 12.2264,
      "step": 13643
    },
    {
      "epoch": 0.7429703293786118,
      "grad_norm": 0.6686777632634148,
      "learning_rate": 0.0001448253313945008,
      "loss": 12.2312,
      "step": 13644
    },
    {
      "epoch": 0.7430247833751948,
      "grad_norm": 0.6014856625238543,
      "learning_rate": 0.00014481744852126485,
      "loss": 12.1709,
      "step": 13645
    },
    {
      "epoch": 0.7430792373717778,
      "grad_norm": 0.6274728754383695,
      "learning_rate": 0.00014480956529952095,
      "loss": 12.1732,
      "step": 13646
    },
    {
      "epoch": 0.7431336913683608,
      "grad_norm": 0.6578097771737071,
      "learning_rate": 0.00014480168172933036,
      "loss": 12.1725,
      "step": 13647
    },
    {
      "epoch": 0.7431881453649439,
      "grad_norm": 0.6243897320773598,
      "learning_rate": 0.00014479379781075438,
      "loss": 12.2669,
      "step": 13648
    },
    {
      "epoch": 0.7432425993615269,
      "grad_norm": 0.5644404409732856,
      "learning_rate": 0.00014478591354385428,
      "loss": 12.2525,
      "step": 13649
    },
    {
      "epoch": 0.7432970533581099,
      "grad_norm": 0.5899929628289414,
      "learning_rate": 0.00014477802892869142,
      "loss": 12.3192,
      "step": 13650
    },
    {
      "epoch": 0.7433515073546929,
      "grad_norm": 0.6204189531105295,
      "learning_rate": 0.0001447701439653271,
      "loss": 12.2539,
      "step": 13651
    },
    {
      "epoch": 0.7434059613512759,
      "grad_norm": 0.5978774508167788,
      "learning_rate": 0.00014476225865382264,
      "loss": 12.2512,
      "step": 13652
    },
    {
      "epoch": 0.7434604153478589,
      "grad_norm": 0.5607476504150831,
      "learning_rate": 0.00014475437299423937,
      "loss": 12.4198,
      "step": 13653
    },
    {
      "epoch": 0.743514869344442,
      "grad_norm": 0.7064522212229157,
      "learning_rate": 0.00014474648698663856,
      "loss": 12.3364,
      "step": 13654
    },
    {
      "epoch": 0.743569323341025,
      "grad_norm": 0.5292842905349883,
      "learning_rate": 0.00014473860063108157,
      "loss": 12.2842,
      "step": 13655
    },
    {
      "epoch": 0.743623777337608,
      "grad_norm": 0.6604362435987455,
      "learning_rate": 0.0001447307139276297,
      "loss": 12.2738,
      "step": 13656
    },
    {
      "epoch": 0.743678231334191,
      "grad_norm": 0.6411131805921383,
      "learning_rate": 0.00014472282687634432,
      "loss": 12.3607,
      "step": 13657
    },
    {
      "epoch": 0.743732685330774,
      "grad_norm": 0.5740431821060552,
      "learning_rate": 0.00014471493947728673,
      "loss": 12.2158,
      "step": 13658
    },
    {
      "epoch": 0.743787139327357,
      "grad_norm": 0.650154708336648,
      "learning_rate": 0.00014470705173051827,
      "loss": 12.3459,
      "step": 13659
    },
    {
      "epoch": 0.7438415933239401,
      "grad_norm": 0.6166673090641452,
      "learning_rate": 0.0001446991636361003,
      "loss": 12.2296,
      "step": 13660
    },
    {
      "epoch": 0.7438960473205231,
      "grad_norm": 0.6630240435802148,
      "learning_rate": 0.00014469127519409414,
      "loss": 12.2581,
      "step": 13661
    },
    {
      "epoch": 0.743950501317106,
      "grad_norm": 0.6285502485037263,
      "learning_rate": 0.0001446833864045611,
      "loss": 12.3641,
      "step": 13662
    },
    {
      "epoch": 0.744004955313689,
      "grad_norm": 0.6243840836632044,
      "learning_rate": 0.00014467549726756256,
      "loss": 12.2185,
      "step": 13663
    },
    {
      "epoch": 0.744059409310272,
      "grad_norm": 0.594732247309961,
      "learning_rate": 0.00014466760778315986,
      "loss": 12.1288,
      "step": 13664
    },
    {
      "epoch": 0.744113863306855,
      "grad_norm": 0.5717664112534702,
      "learning_rate": 0.00014465971795141436,
      "loss": 12.2183,
      "step": 13665
    },
    {
      "epoch": 0.7441683173034381,
      "grad_norm": 0.6705905603648593,
      "learning_rate": 0.0001446518277723874,
      "loss": 12.3219,
      "step": 13666
    },
    {
      "epoch": 0.7442227713000211,
      "grad_norm": 0.6370616353146389,
      "learning_rate": 0.00014464393724614033,
      "loss": 12.3651,
      "step": 13667
    },
    {
      "epoch": 0.7442772252966041,
      "grad_norm": 0.558520232425937,
      "learning_rate": 0.00014463604637273453,
      "loss": 12.2007,
      "step": 13668
    },
    {
      "epoch": 0.7443316792931871,
      "grad_norm": 0.6498084660186766,
      "learning_rate": 0.00014462815515223137,
      "loss": 12.3098,
      "step": 13669
    },
    {
      "epoch": 0.7443861332897701,
      "grad_norm": 0.7689024411652365,
      "learning_rate": 0.00014462026358469214,
      "loss": 12.232,
      "step": 13670
    },
    {
      "epoch": 0.7444405872863531,
      "grad_norm": 0.5720853515258976,
      "learning_rate": 0.0001446123716701783,
      "loss": 12.2595,
      "step": 13671
    },
    {
      "epoch": 0.7444950412829362,
      "grad_norm": 0.5988148788680487,
      "learning_rate": 0.00014460447940875114,
      "loss": 12.3409,
      "step": 13672
    },
    {
      "epoch": 0.7445494952795192,
      "grad_norm": 0.7856904621089689,
      "learning_rate": 0.0001445965868004721,
      "loss": 12.3772,
      "step": 13673
    },
    {
      "epoch": 0.7446039492761022,
      "grad_norm": 0.7043842824252822,
      "learning_rate": 0.0001445886938454025,
      "loss": 12.2397,
      "step": 13674
    },
    {
      "epoch": 0.7446584032726852,
      "grad_norm": 0.5913036266727861,
      "learning_rate": 0.00014458080054360374,
      "loss": 12.2747,
      "step": 13675
    },
    {
      "epoch": 0.7447128572692682,
      "grad_norm": 0.6960974142798146,
      "learning_rate": 0.00014457290689513723,
      "loss": 12.3003,
      "step": 13676
    },
    {
      "epoch": 0.7447673112658513,
      "grad_norm": 0.6087124615164062,
      "learning_rate": 0.00014456501290006427,
      "loss": 12.2339,
      "step": 13677
    },
    {
      "epoch": 0.7448217652624343,
      "grad_norm": 0.5571231214937553,
      "learning_rate": 0.00014455711855844636,
      "loss": 12.1596,
      "step": 13678
    },
    {
      "epoch": 0.7448762192590173,
      "grad_norm": 0.5879599096930934,
      "learning_rate": 0.00014454922387034476,
      "loss": 12.2023,
      "step": 13679
    },
    {
      "epoch": 0.7449306732556003,
      "grad_norm": 0.6587926585973795,
      "learning_rate": 0.00014454132883582097,
      "loss": 12.3255,
      "step": 13680
    },
    {
      "epoch": 0.7449851272521832,
      "grad_norm": 0.5565501984200791,
      "learning_rate": 0.00014453343345493633,
      "loss": 12.2422,
      "step": 13681
    },
    {
      "epoch": 0.7450395812487662,
      "grad_norm": 0.6054602481476327,
      "learning_rate": 0.00014452553772775225,
      "loss": 12.1699,
      "step": 13682
    },
    {
      "epoch": 0.7450940352453493,
      "grad_norm": 0.6502136729803876,
      "learning_rate": 0.00014451764165433008,
      "loss": 12.0922,
      "step": 13683
    },
    {
      "epoch": 0.7451484892419323,
      "grad_norm": 0.5510847853537779,
      "learning_rate": 0.0001445097452347313,
      "loss": 12.1959,
      "step": 13684
    },
    {
      "epoch": 0.7452029432385153,
      "grad_norm": 0.5850713060797119,
      "learning_rate": 0.00014450184846901724,
      "loss": 12.1878,
      "step": 13685
    },
    {
      "epoch": 0.7452573972350983,
      "grad_norm": 0.6319530785149593,
      "learning_rate": 0.00014449395135724937,
      "loss": 12.264,
      "step": 13686
    },
    {
      "epoch": 0.7453118512316813,
      "grad_norm": 0.554872046596875,
      "learning_rate": 0.0001444860538994891,
      "loss": 12.1727,
      "step": 13687
    },
    {
      "epoch": 0.7453663052282643,
      "grad_norm": 0.5718499206730574,
      "learning_rate": 0.00014447815609579777,
      "loss": 12.1314,
      "step": 13688
    },
    {
      "epoch": 0.7454207592248474,
      "grad_norm": 0.5848613944881548,
      "learning_rate": 0.00014447025794623686,
      "loss": 12.189,
      "step": 13689
    },
    {
      "epoch": 0.7454752132214304,
      "grad_norm": 0.5843558986303549,
      "learning_rate": 0.00014446235945086774,
      "loss": 12.3481,
      "step": 13690
    },
    {
      "epoch": 0.7455296672180134,
      "grad_norm": 0.5790485101159226,
      "learning_rate": 0.0001444544606097519,
      "loss": 12.2546,
      "step": 13691
    },
    {
      "epoch": 0.7455841212145964,
      "grad_norm": 0.5908234248192212,
      "learning_rate": 0.0001444465614229507,
      "loss": 12.1865,
      "step": 13692
    },
    {
      "epoch": 0.7456385752111794,
      "grad_norm": 0.5979218535232742,
      "learning_rate": 0.00014443866189052559,
      "loss": 12.2428,
      "step": 13693
    },
    {
      "epoch": 0.7456930292077624,
      "grad_norm": 0.5620026112583111,
      "learning_rate": 0.00014443076201253795,
      "loss": 12.3382,
      "step": 13694
    },
    {
      "epoch": 0.7457474832043455,
      "grad_norm": 0.6823118955874878,
      "learning_rate": 0.00014442286178904928,
      "loss": 12.325,
      "step": 13695
    },
    {
      "epoch": 0.7458019372009285,
      "grad_norm": 0.6067039441439942,
      "learning_rate": 0.000144414961220121,
      "loss": 12.3408,
      "step": 13696
    },
    {
      "epoch": 0.7458563911975115,
      "grad_norm": 0.7007921904958164,
      "learning_rate": 0.00014440706030581456,
      "loss": 12.3337,
      "step": 13697
    },
    {
      "epoch": 0.7459108451940945,
      "grad_norm": 0.6113298266092536,
      "learning_rate": 0.00014439915904619134,
      "loss": 12.2801,
      "step": 13698
    },
    {
      "epoch": 0.7459652991906774,
      "grad_norm": 0.5916639524324443,
      "learning_rate": 0.00014439125744131282,
      "loss": 12.1633,
      "step": 13699
    },
    {
      "epoch": 0.7460197531872604,
      "grad_norm": 0.604444000681318,
      "learning_rate": 0.0001443833554912404,
      "loss": 12.315,
      "step": 13700
    },
    {
      "epoch": 0.7460742071838435,
      "grad_norm": 0.5307951087338542,
      "learning_rate": 0.0001443754531960356,
      "loss": 12.1628,
      "step": 13701
    },
    {
      "epoch": 0.7461286611804265,
      "grad_norm": 0.6424748872034338,
      "learning_rate": 0.00014436755055575984,
      "loss": 12.2518,
      "step": 13702
    },
    {
      "epoch": 0.7461831151770095,
      "grad_norm": 0.5858372985468218,
      "learning_rate": 0.00014435964757047458,
      "loss": 12.1831,
      "step": 13703
    },
    {
      "epoch": 0.7462375691735925,
      "grad_norm": 0.7016042162035304,
      "learning_rate": 0.00014435174424024124,
      "loss": 12.3785,
      "step": 13704
    },
    {
      "epoch": 0.7462920231701755,
      "grad_norm": 0.5351112745115311,
      "learning_rate": 0.00014434384056512126,
      "loss": 12.0123,
      "step": 13705
    },
    {
      "epoch": 0.7463464771667585,
      "grad_norm": 0.6004125768755129,
      "learning_rate": 0.00014433593654517618,
      "loss": 12.2861,
      "step": 13706
    },
    {
      "epoch": 0.7464009311633416,
      "grad_norm": 0.5993177925145355,
      "learning_rate": 0.00014432803218046746,
      "loss": 12.265,
      "step": 13707
    },
    {
      "epoch": 0.7464553851599246,
      "grad_norm": 0.50355922016749,
      "learning_rate": 0.00014432012747105647,
      "loss": 12.0824,
      "step": 13708
    },
    {
      "epoch": 0.7465098391565076,
      "grad_norm": 0.5574649579642977,
      "learning_rate": 0.00014431222241700475,
      "loss": 12.1988,
      "step": 13709
    },
    {
      "epoch": 0.7465642931530906,
      "grad_norm": 0.6904411913829945,
      "learning_rate": 0.00014430431701837376,
      "loss": 12.2252,
      "step": 13710
    },
    {
      "epoch": 0.7466187471496736,
      "grad_norm": 0.5940225992786953,
      "learning_rate": 0.00014429641127522495,
      "loss": 11.9661,
      "step": 13711
    },
    {
      "epoch": 0.7466732011462567,
      "grad_norm": 0.5574148268284755,
      "learning_rate": 0.00014428850518761986,
      "loss": 12.318,
      "step": 13712
    },
    {
      "epoch": 0.7467276551428397,
      "grad_norm": 0.5236783246576344,
      "learning_rate": 0.0001442805987556199,
      "loss": 12.0991,
      "step": 13713
    },
    {
      "epoch": 0.7467821091394227,
      "grad_norm": 0.5820653271899563,
      "learning_rate": 0.0001442726919792866,
      "loss": 12.3288,
      "step": 13714
    },
    {
      "epoch": 0.7468365631360057,
      "grad_norm": 0.5932316400669257,
      "learning_rate": 0.0001442647848586814,
      "loss": 12.3073,
      "step": 13715
    },
    {
      "epoch": 0.7468910171325887,
      "grad_norm": 0.6467755435657985,
      "learning_rate": 0.0001442568773938658,
      "loss": 12.1978,
      "step": 13716
    },
    {
      "epoch": 0.7469454711291716,
      "grad_norm": 0.5516042158353747,
      "learning_rate": 0.00014424896958490133,
      "loss": 12.0761,
      "step": 13717
    },
    {
      "epoch": 0.7469999251257547,
      "grad_norm": 0.5613663038502804,
      "learning_rate": 0.00014424106143184944,
      "loss": 12.187,
      "step": 13718
    },
    {
      "epoch": 0.7470543791223377,
      "grad_norm": 0.6013135202519371,
      "learning_rate": 0.00014423315293477163,
      "loss": 12.3413,
      "step": 13719
    },
    {
      "epoch": 0.7471088331189207,
      "grad_norm": 0.5628113838646053,
      "learning_rate": 0.0001442252440937294,
      "loss": 12.2496,
      "step": 13720
    },
    {
      "epoch": 0.7471632871155037,
      "grad_norm": 0.5904699220530979,
      "learning_rate": 0.00014421733490878425,
      "loss": 12.2199,
      "step": 13721
    },
    {
      "epoch": 0.7472177411120867,
      "grad_norm": 0.6257160288719434,
      "learning_rate": 0.00014420942537999773,
      "loss": 12.2555,
      "step": 13722
    },
    {
      "epoch": 0.7472721951086697,
      "grad_norm": 0.5439262444923114,
      "learning_rate": 0.00014420151550743125,
      "loss": 12.2134,
      "step": 13723
    },
    {
      "epoch": 0.7473266491052528,
      "grad_norm": 0.6459417849868265,
      "learning_rate": 0.00014419360529114642,
      "loss": 12.2833,
      "step": 13724
    },
    {
      "epoch": 0.7473811031018358,
      "grad_norm": 0.62630629000918,
      "learning_rate": 0.00014418569473120468,
      "loss": 12.2979,
      "step": 13725
    },
    {
      "epoch": 0.7474355570984188,
      "grad_norm": 0.5777191040932425,
      "learning_rate": 0.00014417778382766757,
      "loss": 12.211,
      "step": 13726
    },
    {
      "epoch": 0.7474900110950018,
      "grad_norm": 0.6563049547344048,
      "learning_rate": 0.00014416987258059663,
      "loss": 12.3766,
      "step": 13727
    },
    {
      "epoch": 0.7475444650915848,
      "grad_norm": 0.530946466034582,
      "learning_rate": 0.0001441619609900533,
      "loss": 12.187,
      "step": 13728
    },
    {
      "epoch": 0.7475989190881678,
      "grad_norm": 0.5851608738144949,
      "learning_rate": 0.0001441540490560992,
      "loss": 12.1812,
      "step": 13729
    },
    {
      "epoch": 0.7476533730847509,
      "grad_norm": 0.623454123044338,
      "learning_rate": 0.0001441461367787958,
      "loss": 12.2312,
      "step": 13730
    },
    {
      "epoch": 0.7477078270813339,
      "grad_norm": 0.6151173822258862,
      "learning_rate": 0.00014413822415820465,
      "loss": 12.272,
      "step": 13731
    },
    {
      "epoch": 0.7477622810779169,
      "grad_norm": 0.6270601805006041,
      "learning_rate": 0.00014413031119438723,
      "loss": 12.2757,
      "step": 13732
    },
    {
      "epoch": 0.7478167350744999,
      "grad_norm": 0.5982856037425696,
      "learning_rate": 0.00014412239788740513,
      "loss": 12.2053,
      "step": 13733
    },
    {
      "epoch": 0.7478711890710829,
      "grad_norm": 0.6499148875444776,
      "learning_rate": 0.00014411448423731985,
      "loss": 12.3002,
      "step": 13734
    },
    {
      "epoch": 0.7479256430676658,
      "grad_norm": 0.566224351321186,
      "learning_rate": 0.00014410657024419295,
      "loss": 12.1763,
      "step": 13735
    },
    {
      "epoch": 0.747980097064249,
      "grad_norm": 0.6627051566300732,
      "learning_rate": 0.00014409865590808598,
      "loss": 12.2677,
      "step": 13736
    },
    {
      "epoch": 0.7480345510608319,
      "grad_norm": 0.646344114628114,
      "learning_rate": 0.00014409074122906048,
      "loss": 12.2699,
      "step": 13737
    },
    {
      "epoch": 0.7480890050574149,
      "grad_norm": 0.5937235560428703,
      "learning_rate": 0.00014408282620717794,
      "loss": 12.2404,
      "step": 13738
    },
    {
      "epoch": 0.7481434590539979,
      "grad_norm": 0.5806014220082066,
      "learning_rate": 0.00014407491084249995,
      "loss": 12.2855,
      "step": 13739
    },
    {
      "epoch": 0.7481979130505809,
      "grad_norm": 0.5659780798715816,
      "learning_rate": 0.0001440669951350881,
      "loss": 12.3017,
      "step": 13740
    },
    {
      "epoch": 0.7482523670471639,
      "grad_norm": 0.6227277218313976,
      "learning_rate": 0.00014405907908500388,
      "loss": 12.2454,
      "step": 13741
    },
    {
      "epoch": 0.748306821043747,
      "grad_norm": 0.6664710496279762,
      "learning_rate": 0.0001440511626923089,
      "loss": 12.3328,
      "step": 13742
    },
    {
      "epoch": 0.74836127504033,
      "grad_norm": 0.5970770618016265,
      "learning_rate": 0.00014404324595706464,
      "loss": 12.2713,
      "step": 13743
    },
    {
      "epoch": 0.748415729036913,
      "grad_norm": 0.578291694869162,
      "learning_rate": 0.00014403532887933274,
      "loss": 12.3062,
      "step": 13744
    },
    {
      "epoch": 0.748470183033496,
      "grad_norm": 0.6416824753124646,
      "learning_rate": 0.00014402741145917475,
      "loss": 12.0758,
      "step": 13745
    },
    {
      "epoch": 0.748524637030079,
      "grad_norm": 0.5655552786892055,
      "learning_rate": 0.00014401949369665222,
      "loss": 12.1638,
      "step": 13746
    },
    {
      "epoch": 0.7485790910266621,
      "grad_norm": 0.6710049036354784,
      "learning_rate": 0.00014401157559182674,
      "loss": 12.285,
      "step": 13747
    },
    {
      "epoch": 0.7486335450232451,
      "grad_norm": 0.6801326451669022,
      "learning_rate": 0.00014400365714475986,
      "loss": 12.3959,
      "step": 13748
    },
    {
      "epoch": 0.7486879990198281,
      "grad_norm": 0.5368301698649847,
      "learning_rate": 0.00014399573835551313,
      "loss": 12.2605,
      "step": 13749
    },
    {
      "epoch": 0.7487424530164111,
      "grad_norm": 0.6540587730560741,
      "learning_rate": 0.00014398781922414817,
      "loss": 12.3747,
      "step": 13750
    },
    {
      "epoch": 0.7487969070129941,
      "grad_norm": 0.6016638900317841,
      "learning_rate": 0.00014397989975072656,
      "loss": 12.3126,
      "step": 13751
    },
    {
      "epoch": 0.748851361009577,
      "grad_norm": 0.555668289888147,
      "learning_rate": 0.0001439719799353099,
      "loss": 12.3146,
      "step": 13752
    },
    {
      "epoch": 0.7489058150061602,
      "grad_norm": 0.6973329845200757,
      "learning_rate": 0.00014396405977795972,
      "loss": 12.2642,
      "step": 13753
    },
    {
      "epoch": 0.7489602690027432,
      "grad_norm": 0.5280970056108852,
      "learning_rate": 0.00014395613927873765,
      "loss": 12.2146,
      "step": 13754
    },
    {
      "epoch": 0.7490147229993261,
      "grad_norm": 0.5899667507355091,
      "learning_rate": 0.00014394821843770526,
      "loss": 12.2097,
      "step": 13755
    },
    {
      "epoch": 0.7490691769959091,
      "grad_norm": 0.5202820450953016,
      "learning_rate": 0.00014394029725492416,
      "loss": 12.2154,
      "step": 13756
    },
    {
      "epoch": 0.7491236309924921,
      "grad_norm": 0.5798168456983885,
      "learning_rate": 0.00014393237573045596,
      "loss": 12.2218,
      "step": 13757
    },
    {
      "epoch": 0.7491780849890751,
      "grad_norm": 0.5757177192193262,
      "learning_rate": 0.0001439244538643622,
      "loss": 12.1298,
      "step": 13758
    },
    {
      "epoch": 0.7492325389856582,
      "grad_norm": 0.5439117109561773,
      "learning_rate": 0.00014391653165670454,
      "loss": 12.2106,
      "step": 13759
    },
    {
      "epoch": 0.7492869929822412,
      "grad_norm": 0.5665992283629767,
      "learning_rate": 0.00014390860910754453,
      "loss": 12.2342,
      "step": 13760
    },
    {
      "epoch": 0.7493414469788242,
      "grad_norm": 0.610271080039589,
      "learning_rate": 0.00014390068621694387,
      "loss": 12.2926,
      "step": 13761
    },
    {
      "epoch": 0.7493959009754072,
      "grad_norm": 0.5701767578829343,
      "learning_rate": 0.0001438927629849641,
      "loss": 12.23,
      "step": 13762
    },
    {
      "epoch": 0.7494503549719902,
      "grad_norm": 0.7850785013536542,
      "learning_rate": 0.00014388483941166682,
      "loss": 12.2746,
      "step": 13763
    },
    {
      "epoch": 0.7495048089685732,
      "grad_norm": 0.6326742294350892,
      "learning_rate": 0.0001438769154971137,
      "loss": 12.2667,
      "step": 13764
    },
    {
      "epoch": 0.7495592629651563,
      "grad_norm": 0.5210535123007397,
      "learning_rate": 0.0001438689912413663,
      "loss": 12.2196,
      "step": 13765
    },
    {
      "epoch": 0.7496137169617393,
      "grad_norm": 0.5766175886374203,
      "learning_rate": 0.00014386106664448625,
      "loss": 12.2693,
      "step": 13766
    },
    {
      "epoch": 0.7496681709583223,
      "grad_norm": 0.6178574912384032,
      "learning_rate": 0.00014385314170653523,
      "loss": 12.252,
      "step": 13767
    },
    {
      "epoch": 0.7497226249549053,
      "grad_norm": 0.5915591393770658,
      "learning_rate": 0.0001438452164275748,
      "loss": 12.2551,
      "step": 13768
    },
    {
      "epoch": 0.7497770789514883,
      "grad_norm": 0.6269592287242665,
      "learning_rate": 0.00014383729080766664,
      "loss": 12.2719,
      "step": 13769
    },
    {
      "epoch": 0.7498315329480713,
      "grad_norm": 0.5609303240083141,
      "learning_rate": 0.00014382936484687233,
      "loss": 12.1725,
      "step": 13770
    },
    {
      "epoch": 0.7498859869446544,
      "grad_norm": 0.6438885235253843,
      "learning_rate": 0.00014382143854525353,
      "loss": 12.333,
      "step": 13771
    },
    {
      "epoch": 0.7499404409412374,
      "grad_norm": 0.5705368582587276,
      "learning_rate": 0.00014381351190287188,
      "loss": 12.2322,
      "step": 13772
    },
    {
      "epoch": 0.7499948949378203,
      "grad_norm": 0.6498876610872063,
      "learning_rate": 0.000143805584919789,
      "loss": 12.3526,
      "step": 13773
    },
    {
      "epoch": 0.7500493489344033,
      "grad_norm": 0.5829378708822657,
      "learning_rate": 0.00014379765759606658,
      "loss": 12.146,
      "step": 13774
    },
    {
      "epoch": 0.7501038029309863,
      "grad_norm": 0.5781624917520148,
      "learning_rate": 0.00014378972993176622,
      "loss": 12.2652,
      "step": 13775
    },
    {
      "epoch": 0.7501582569275693,
      "grad_norm": 0.5605017438627762,
      "learning_rate": 0.00014378180192694957,
      "loss": 12.1909,
      "step": 13776
    },
    {
      "epoch": 0.7502127109241524,
      "grad_norm": 0.5589222728784693,
      "learning_rate": 0.00014377387358167828,
      "loss": 12.159,
      "step": 13777
    },
    {
      "epoch": 0.7502671649207354,
      "grad_norm": 0.5655748541001888,
      "learning_rate": 0.00014376594489601402,
      "loss": 12.1719,
      "step": 13778
    },
    {
      "epoch": 0.7503216189173184,
      "grad_norm": 0.9229516569274044,
      "learning_rate": 0.00014375801587001842,
      "loss": 12.1961,
      "step": 13779
    },
    {
      "epoch": 0.7503760729139014,
      "grad_norm": 0.5190587246726092,
      "learning_rate": 0.00014375008650375313,
      "loss": 12.1197,
      "step": 13780
    },
    {
      "epoch": 0.7504305269104844,
      "grad_norm": 0.5842286466771999,
      "learning_rate": 0.0001437421567972799,
      "loss": 12.0913,
      "step": 13781
    },
    {
      "epoch": 0.7504849809070675,
      "grad_norm": 0.5366466263553302,
      "learning_rate": 0.00014373422675066023,
      "loss": 12.1292,
      "step": 13782
    },
    {
      "epoch": 0.7505394349036505,
      "grad_norm": 0.6292072846094948,
      "learning_rate": 0.00014372629636395597,
      "loss": 12.3445,
      "step": 13783
    },
    {
      "epoch": 0.7505938889002335,
      "grad_norm": 0.6518660654831823,
      "learning_rate": 0.00014371836563722865,
      "loss": 12.3205,
      "step": 13784
    },
    {
      "epoch": 0.7506483428968165,
      "grad_norm": 0.5621702149563956,
      "learning_rate": 0.00014371043457054,
      "loss": 12.3394,
      "step": 13785
    },
    {
      "epoch": 0.7507027968933995,
      "grad_norm": 0.5716000303103266,
      "learning_rate": 0.00014370250316395167,
      "loss": 12.2124,
      "step": 13786
    },
    {
      "epoch": 0.7507572508899825,
      "grad_norm": 0.527542076552203,
      "learning_rate": 0.00014369457141752534,
      "loss": 12.2448,
      "step": 13787
    },
    {
      "epoch": 0.7508117048865656,
      "grad_norm": 0.6303658246608846,
      "learning_rate": 0.0001436866393313227,
      "loss": 12.3021,
      "step": 13788
    },
    {
      "epoch": 0.7508661588831486,
      "grad_norm": 0.5276474677797767,
      "learning_rate": 0.00014367870690540544,
      "loss": 12.2299,
      "step": 13789
    },
    {
      "epoch": 0.7509206128797316,
      "grad_norm": 0.6712221790438491,
      "learning_rate": 0.00014367077413983523,
      "loss": 12.2085,
      "step": 13790
    },
    {
      "epoch": 0.7509750668763145,
      "grad_norm": 0.534081814151848,
      "learning_rate": 0.00014366284103467373,
      "loss": 12.1906,
      "step": 13791
    },
    {
      "epoch": 0.7510295208728975,
      "grad_norm": 0.5951491738825837,
      "learning_rate": 0.0001436549075899827,
      "loss": 12.2514,
      "step": 13792
    },
    {
      "epoch": 0.7510839748694805,
      "grad_norm": 0.6803997052913654,
      "learning_rate": 0.00014364697380582375,
      "loss": 12.2647,
      "step": 13793
    },
    {
      "epoch": 0.7511384288660636,
      "grad_norm": 0.5874316064236298,
      "learning_rate": 0.00014363903968225863,
      "loss": 12.3495,
      "step": 13794
    },
    {
      "epoch": 0.7511928828626466,
      "grad_norm": 0.5356708641919317,
      "learning_rate": 0.000143631105219349,
      "loss": 12.2118,
      "step": 13795
    },
    {
      "epoch": 0.7512473368592296,
      "grad_norm": 0.6295249693626224,
      "learning_rate": 0.00014362317041715657,
      "loss": 12.1805,
      "step": 13796
    },
    {
      "epoch": 0.7513017908558126,
      "grad_norm": 1.03868303664006,
      "learning_rate": 0.00014361523527574307,
      "loss": 12.259,
      "step": 13797
    },
    {
      "epoch": 0.7513562448523956,
      "grad_norm": 0.6257302393342892,
      "learning_rate": 0.0001436072997951702,
      "loss": 12.257,
      "step": 13798
    },
    {
      "epoch": 0.7514106988489786,
      "grad_norm": 0.623335943617568,
      "learning_rate": 0.00014359936397549965,
      "loss": 12.3552,
      "step": 13799
    },
    {
      "epoch": 0.7514651528455617,
      "grad_norm": 0.5684967086416157,
      "learning_rate": 0.00014359142781679313,
      "loss": 12.169,
      "step": 13800
    },
    {
      "epoch": 0.7515196068421447,
      "grad_norm": 0.5688960037551076,
      "learning_rate": 0.00014358349131911234,
      "loss": 12.3163,
      "step": 13801
    },
    {
      "epoch": 0.7515740608387277,
      "grad_norm": 0.5842685732263988,
      "learning_rate": 0.00014357555448251902,
      "loss": 12.2271,
      "step": 13802
    },
    {
      "epoch": 0.7516285148353107,
      "grad_norm": 0.6143806368491119,
      "learning_rate": 0.00014356761730707489,
      "loss": 12.3893,
      "step": 13803
    },
    {
      "epoch": 0.7516829688318937,
      "grad_norm": 0.6554362600326484,
      "learning_rate": 0.00014355967979284167,
      "loss": 12.2427,
      "step": 13804
    },
    {
      "epoch": 0.7517374228284767,
      "grad_norm": 0.6311453226982938,
      "learning_rate": 0.00014355174193988107,
      "loss": 12.3582,
      "step": 13805
    },
    {
      "epoch": 0.7517918768250598,
      "grad_norm": 0.6346421035651231,
      "learning_rate": 0.0001435438037482548,
      "loss": 12.2786,
      "step": 13806
    },
    {
      "epoch": 0.7518463308216428,
      "grad_norm": 0.6117921300048075,
      "learning_rate": 0.00014353586521802461,
      "loss": 12.2096,
      "step": 13807
    },
    {
      "epoch": 0.7519007848182258,
      "grad_norm": 0.6019476080731878,
      "learning_rate": 0.0001435279263492523,
      "loss": 12.3236,
      "step": 13808
    },
    {
      "epoch": 0.7519552388148087,
      "grad_norm": 0.7514865124107512,
      "learning_rate": 0.00014351998714199943,
      "loss": 12.3547,
      "step": 13809
    },
    {
      "epoch": 0.7520096928113917,
      "grad_norm": 0.5646877462273913,
      "learning_rate": 0.0001435120475963279,
      "loss": 12.2639,
      "step": 13810
    },
    {
      "epoch": 0.7520641468079748,
      "grad_norm": 0.5891752717887218,
      "learning_rate": 0.0001435041077122994,
      "loss": 12.3077,
      "step": 13811
    },
    {
      "epoch": 0.7521186008045578,
      "grad_norm": 0.5862291042415606,
      "learning_rate": 0.00014349616748997562,
      "loss": 12.2445,
      "step": 13812
    },
    {
      "epoch": 0.7521730548011408,
      "grad_norm": 0.614589247223749,
      "learning_rate": 0.00014348822692941842,
      "loss": 12.3795,
      "step": 13813
    },
    {
      "epoch": 0.7522275087977238,
      "grad_norm": 0.5813380916070603,
      "learning_rate": 0.00014348028603068942,
      "loss": 12.2293,
      "step": 13814
    },
    {
      "epoch": 0.7522819627943068,
      "grad_norm": 0.63080789984502,
      "learning_rate": 0.0001434723447938504,
      "loss": 12.2895,
      "step": 13815
    },
    {
      "epoch": 0.7523364167908898,
      "grad_norm": 0.5631956520829646,
      "learning_rate": 0.00014346440321896318,
      "loss": 12.2301,
      "step": 13816
    },
    {
      "epoch": 0.7523908707874729,
      "grad_norm": 0.5276970829388432,
      "learning_rate": 0.00014345646130608944,
      "loss": 12.0721,
      "step": 13817
    },
    {
      "epoch": 0.7524453247840559,
      "grad_norm": 0.5801900867009577,
      "learning_rate": 0.00014344851905529103,
      "loss": 12.2662,
      "step": 13818
    },
    {
      "epoch": 0.7524997787806389,
      "grad_norm": 0.620601218168845,
      "learning_rate": 0.0001434405764666296,
      "loss": 12.3432,
      "step": 13819
    },
    {
      "epoch": 0.7525542327772219,
      "grad_norm": 0.5993138059551066,
      "learning_rate": 0.00014343263354016695,
      "loss": 12.246,
      "step": 13820
    },
    {
      "epoch": 0.7526086867738049,
      "grad_norm": 0.5793428284321268,
      "learning_rate": 0.00014342469027596487,
      "loss": 12.2635,
      "step": 13821
    },
    {
      "epoch": 0.7526631407703879,
      "grad_norm": 0.626149109871221,
      "learning_rate": 0.00014341674667408513,
      "loss": 12.0145,
      "step": 13822
    },
    {
      "epoch": 0.752717594766971,
      "grad_norm": 0.6405112662502047,
      "learning_rate": 0.0001434088027345895,
      "loss": 12.463,
      "step": 13823
    },
    {
      "epoch": 0.752772048763554,
      "grad_norm": 0.6023957152504208,
      "learning_rate": 0.00014340085845753972,
      "loss": 12.1497,
      "step": 13824
    },
    {
      "epoch": 0.752826502760137,
      "grad_norm": 0.6355489896387884,
      "learning_rate": 0.0001433929138429976,
      "loss": 12.459,
      "step": 13825
    },
    {
      "epoch": 0.75288095675672,
      "grad_norm": 0.6425241440639089,
      "learning_rate": 0.00014338496889102487,
      "loss": 12.311,
      "step": 13826
    },
    {
      "epoch": 0.752935410753303,
      "grad_norm": 0.6167990707777722,
      "learning_rate": 0.00014337702360168336,
      "loss": 12.1696,
      "step": 13827
    },
    {
      "epoch": 0.7529898647498859,
      "grad_norm": 0.6539399435350298,
      "learning_rate": 0.00014336907797503487,
      "loss": 12.2048,
      "step": 13828
    },
    {
      "epoch": 0.753044318746469,
      "grad_norm": 0.6056473383463385,
      "learning_rate": 0.00014336113201114114,
      "loss": 12.1473,
      "step": 13829
    },
    {
      "epoch": 0.753098772743052,
      "grad_norm": 0.5535186851656467,
      "learning_rate": 0.00014335318571006398,
      "loss": 12.2927,
      "step": 13830
    },
    {
      "epoch": 0.753153226739635,
      "grad_norm": 0.6022086596989811,
      "learning_rate": 0.00014334523907186513,
      "loss": 12.2001,
      "step": 13831
    },
    {
      "epoch": 0.753207680736218,
      "grad_norm": 0.591273556250499,
      "learning_rate": 0.00014333729209660648,
      "loss": 12.199,
      "step": 13832
    },
    {
      "epoch": 0.753262134732801,
      "grad_norm": 0.5760100490032448,
      "learning_rate": 0.00014332934478434982,
      "loss": 12.1534,
      "step": 13833
    },
    {
      "epoch": 0.753316588729384,
      "grad_norm": 0.6027884470696825,
      "learning_rate": 0.00014332139713515684,
      "loss": 12.3531,
      "step": 13834
    },
    {
      "epoch": 0.7533710427259671,
      "grad_norm": 0.6205928858416357,
      "learning_rate": 0.00014331344914908942,
      "loss": 12.2737,
      "step": 13835
    },
    {
      "epoch": 0.7534254967225501,
      "grad_norm": 0.5886492463854501,
      "learning_rate": 0.00014330550082620937,
      "loss": 12.263,
      "step": 13836
    },
    {
      "epoch": 0.7534799507191331,
      "grad_norm": 0.5914234570101967,
      "learning_rate": 0.00014329755216657849,
      "loss": 12.0383,
      "step": 13837
    },
    {
      "epoch": 0.7535344047157161,
      "grad_norm": 0.6410193708975239,
      "learning_rate": 0.00014328960317025856,
      "loss": 12.281,
      "step": 13838
    },
    {
      "epoch": 0.7535888587122991,
      "grad_norm": 0.5314330429599432,
      "learning_rate": 0.00014328165383731145,
      "loss": 12.2602,
      "step": 13839
    },
    {
      "epoch": 0.7536433127088821,
      "grad_norm": 0.540957434758258,
      "learning_rate": 0.0001432737041677989,
      "loss": 12.2461,
      "step": 13840
    },
    {
      "epoch": 0.7536977667054652,
      "grad_norm": 0.5432546815104532,
      "learning_rate": 0.00014326575416178278,
      "loss": 12.1897,
      "step": 13841
    },
    {
      "epoch": 0.7537522207020482,
      "grad_norm": 0.5541501713891164,
      "learning_rate": 0.00014325780381932492,
      "loss": 12.164,
      "step": 13842
    },
    {
      "epoch": 0.7538066746986312,
      "grad_norm": 0.7697095586952005,
      "learning_rate": 0.0001432498531404871,
      "loss": 12.1623,
      "step": 13843
    },
    {
      "epoch": 0.7538611286952142,
      "grad_norm": 0.49945314514455,
      "learning_rate": 0.0001432419021253312,
      "loss": 12.2043,
      "step": 13844
    },
    {
      "epoch": 0.7539155826917971,
      "grad_norm": 0.5374440022342415,
      "learning_rate": 0.000143233950773919,
      "loss": 12.1716,
      "step": 13845
    },
    {
      "epoch": 0.7539700366883803,
      "grad_norm": 0.7339500382749483,
      "learning_rate": 0.00014322599908631232,
      "loss": 12.3037,
      "step": 13846
    },
    {
      "epoch": 0.7540244906849632,
      "grad_norm": 0.5815419891307385,
      "learning_rate": 0.00014321804706257307,
      "loss": 12.2847,
      "step": 13847
    },
    {
      "epoch": 0.7540789446815462,
      "grad_norm": 0.5297258689843342,
      "learning_rate": 0.00014321009470276303,
      "loss": 12.1873,
      "step": 13848
    },
    {
      "epoch": 0.7541333986781292,
      "grad_norm": 0.6692806498141518,
      "learning_rate": 0.000143202142006944,
      "loss": 12.3004,
      "step": 13849
    },
    {
      "epoch": 0.7541878526747122,
      "grad_norm": 0.517943893944968,
      "learning_rate": 0.00014319418897517792,
      "loss": 12.1011,
      "step": 13850
    },
    {
      "epoch": 0.7542423066712952,
      "grad_norm": 0.5772763106403177,
      "learning_rate": 0.00014318623560752658,
      "loss": 12.3086,
      "step": 13851
    },
    {
      "epoch": 0.7542967606678783,
      "grad_norm": 0.6149695725607878,
      "learning_rate": 0.0001431782819040518,
      "loss": 12.2392,
      "step": 13852
    },
    {
      "epoch": 0.7543512146644613,
      "grad_norm": 0.6400262055660713,
      "learning_rate": 0.00014317032786481548,
      "loss": 12.4411,
      "step": 13853
    },
    {
      "epoch": 0.7544056686610443,
      "grad_norm": 0.5576986266208344,
      "learning_rate": 0.00014316237348987944,
      "loss": 12.2759,
      "step": 13854
    },
    {
      "epoch": 0.7544601226576273,
      "grad_norm": 0.6692229239456416,
      "learning_rate": 0.00014315441877930555,
      "loss": 12.2574,
      "step": 13855
    },
    {
      "epoch": 0.7545145766542103,
      "grad_norm": 0.6281993479852913,
      "learning_rate": 0.0001431464637331557,
      "loss": 12.3431,
      "step": 13856
    },
    {
      "epoch": 0.7545690306507933,
      "grad_norm": 0.5816207313651164,
      "learning_rate": 0.00014313850835149166,
      "loss": 12.1529,
      "step": 13857
    },
    {
      "epoch": 0.7546234846473764,
      "grad_norm": 0.6051148573253917,
      "learning_rate": 0.00014313055263437535,
      "loss": 12.3998,
      "step": 13858
    },
    {
      "epoch": 0.7546779386439594,
      "grad_norm": 0.557618965382382,
      "learning_rate": 0.00014312259658186865,
      "loss": 12.2016,
      "step": 13859
    },
    {
      "epoch": 0.7547323926405424,
      "grad_norm": 0.5795972098301391,
      "learning_rate": 0.0001431146401940334,
      "loss": 12.2748,
      "step": 13860
    },
    {
      "epoch": 0.7547868466371254,
      "grad_norm": 0.5661625741258683,
      "learning_rate": 0.00014310668347093146,
      "loss": 12.1477,
      "step": 13861
    },
    {
      "epoch": 0.7548413006337084,
      "grad_norm": 0.5847540858969881,
      "learning_rate": 0.00014309872641262475,
      "loss": 12.2328,
      "step": 13862
    },
    {
      "epoch": 0.7548957546302913,
      "grad_norm": 0.5954478677911897,
      "learning_rate": 0.00014309076901917512,
      "loss": 12.2756,
      "step": 13863
    },
    {
      "epoch": 0.7549502086268745,
      "grad_norm": 0.5042730004768833,
      "learning_rate": 0.00014308281129064442,
      "loss": 12.159,
      "step": 13864
    },
    {
      "epoch": 0.7550046626234574,
      "grad_norm": 0.6458120398598242,
      "learning_rate": 0.00014307485322709456,
      "loss": 12.1989,
      "step": 13865
    },
    {
      "epoch": 0.7550591166200404,
      "grad_norm": 0.5819826806870918,
      "learning_rate": 0.00014306689482858738,
      "loss": 12.1708,
      "step": 13866
    },
    {
      "epoch": 0.7551135706166234,
      "grad_norm": 0.552542592845485,
      "learning_rate": 0.00014305893609518487,
      "loss": 12.1174,
      "step": 13867
    },
    {
      "epoch": 0.7551680246132064,
      "grad_norm": 0.6422678536801202,
      "learning_rate": 0.00014305097702694883,
      "loss": 12.2365,
      "step": 13868
    },
    {
      "epoch": 0.7552224786097894,
      "grad_norm": 0.5456608259492738,
      "learning_rate": 0.00014304301762394116,
      "loss": 12.2812,
      "step": 13869
    },
    {
      "epoch": 0.7552769326063725,
      "grad_norm": 0.5815679736576249,
      "learning_rate": 0.00014303505788622374,
      "loss": 12.2574,
      "step": 13870
    },
    {
      "epoch": 0.7553313866029555,
      "grad_norm": 0.6311052988627255,
      "learning_rate": 0.00014302709781385855,
      "loss": 12.2308,
      "step": 13871
    },
    {
      "epoch": 0.7553858405995385,
      "grad_norm": 0.6006336352942205,
      "learning_rate": 0.0001430191374069074,
      "loss": 12.3778,
      "step": 13872
    },
    {
      "epoch": 0.7554402945961215,
      "grad_norm": 0.6255333097598189,
      "learning_rate": 0.00014301117666543225,
      "loss": 11.9867,
      "step": 13873
    },
    {
      "epoch": 0.7554947485927045,
      "grad_norm": 0.5808851218717448,
      "learning_rate": 0.00014300321558949496,
      "loss": 12.172,
      "step": 13874
    },
    {
      "epoch": 0.7555492025892875,
      "grad_norm": 0.661725093009846,
      "learning_rate": 0.00014299525417915744,
      "loss": 12.3623,
      "step": 13875
    },
    {
      "epoch": 0.7556036565858706,
      "grad_norm": 0.6244849203626905,
      "learning_rate": 0.00014298729243448162,
      "loss": 12.2022,
      "step": 13876
    },
    {
      "epoch": 0.7556581105824536,
      "grad_norm": 0.5769886983567915,
      "learning_rate": 0.00014297933035552942,
      "loss": 12.2396,
      "step": 13877
    },
    {
      "epoch": 0.7557125645790366,
      "grad_norm": 0.5965097791921065,
      "learning_rate": 0.00014297136794236273,
      "loss": 12.2386,
      "step": 13878
    },
    {
      "epoch": 0.7557670185756196,
      "grad_norm": 0.5921318795590744,
      "learning_rate": 0.00014296340519504347,
      "loss": 12.2327,
      "step": 13879
    },
    {
      "epoch": 0.7558214725722026,
      "grad_norm": 0.6021718374457375,
      "learning_rate": 0.00014295544211363357,
      "loss": 12.2171,
      "step": 13880
    },
    {
      "epoch": 0.7558759265687857,
      "grad_norm": 0.5942547115161655,
      "learning_rate": 0.00014294747869819495,
      "loss": 12.2583,
      "step": 13881
    },
    {
      "epoch": 0.7559303805653687,
      "grad_norm": 0.5705233666422782,
      "learning_rate": 0.00014293951494878955,
      "loss": 12.1982,
      "step": 13882
    },
    {
      "epoch": 0.7559848345619516,
      "grad_norm": 0.6233055567094691,
      "learning_rate": 0.00014293155086547927,
      "loss": 12.3178,
      "step": 13883
    },
    {
      "epoch": 0.7560392885585346,
      "grad_norm": 0.5574722512875832,
      "learning_rate": 0.00014292358644832603,
      "loss": 12.4212,
      "step": 13884
    },
    {
      "epoch": 0.7560937425551176,
      "grad_norm": 0.6029716387807774,
      "learning_rate": 0.0001429156216973918,
      "loss": 12.4054,
      "step": 13885
    },
    {
      "epoch": 0.7561481965517006,
      "grad_norm": 0.588774065224866,
      "learning_rate": 0.00014290765661273847,
      "loss": 12.2661,
      "step": 13886
    },
    {
      "epoch": 0.7562026505482837,
      "grad_norm": 0.5495595440606598,
      "learning_rate": 0.00014289969119442804,
      "loss": 12.364,
      "step": 13887
    },
    {
      "epoch": 0.7562571045448667,
      "grad_norm": 0.6343815607337305,
      "learning_rate": 0.00014289172544252246,
      "loss": 12.2642,
      "step": 13888
    },
    {
      "epoch": 0.7563115585414497,
      "grad_norm": 0.5732501696621901,
      "learning_rate": 0.00014288375935708357,
      "loss": 12.2799,
      "step": 13889
    },
    {
      "epoch": 0.7563660125380327,
      "grad_norm": 0.5839071557316391,
      "learning_rate": 0.0001428757929381734,
      "loss": 12.1891,
      "step": 13890
    },
    {
      "epoch": 0.7564204665346157,
      "grad_norm": 0.6679324286207482,
      "learning_rate": 0.00014286782618585383,
      "loss": 12.253,
      "step": 13891
    },
    {
      "epoch": 0.7564749205311987,
      "grad_norm": 0.5527326954974784,
      "learning_rate": 0.00014285985910018689,
      "loss": 12.2295,
      "step": 13892
    },
    {
      "epoch": 0.7565293745277818,
      "grad_norm": 0.5980791290120517,
      "learning_rate": 0.0001428518916812345,
      "loss": 12.3769,
      "step": 13893
    },
    {
      "epoch": 0.7565838285243648,
      "grad_norm": 0.6069665356579306,
      "learning_rate": 0.00014284392392905861,
      "loss": 12.2317,
      "step": 13894
    },
    {
      "epoch": 0.7566382825209478,
      "grad_norm": 0.5593302708857323,
      "learning_rate": 0.0001428359558437212,
      "loss": 12.1659,
      "step": 13895
    },
    {
      "epoch": 0.7566927365175308,
      "grad_norm": 0.6431685721579361,
      "learning_rate": 0.0001428279874252842,
      "loss": 12.3359,
      "step": 13896
    },
    {
      "epoch": 0.7567471905141138,
      "grad_norm": 0.6286285307146453,
      "learning_rate": 0.00014282001867380953,
      "loss": 12.261,
      "step": 13897
    },
    {
      "epoch": 0.7568016445106968,
      "grad_norm": 0.5815019618216938,
      "learning_rate": 0.00014281204958935929,
      "loss": 12.2054,
      "step": 13898
    },
    {
      "epoch": 0.7568560985072799,
      "grad_norm": 0.6448256396057778,
      "learning_rate": 0.00014280408017199535,
      "loss": 12.2712,
      "step": 13899
    },
    {
      "epoch": 0.7569105525038629,
      "grad_norm": 0.5632502677636745,
      "learning_rate": 0.0001427961104217797,
      "loss": 12.3028,
      "step": 13900
    },
    {
      "epoch": 0.7569650065004458,
      "grad_norm": 0.5939995631430055,
      "learning_rate": 0.00014278814033877432,
      "loss": 12.1262,
      "step": 13901
    },
    {
      "epoch": 0.7570194604970288,
      "grad_norm": 0.610425379765044,
      "learning_rate": 0.0001427801699230412,
      "loss": 12.2151,
      "step": 13902
    },
    {
      "epoch": 0.7570739144936118,
      "grad_norm": 0.5515460328017994,
      "learning_rate": 0.0001427721991746423,
      "loss": 12.2251,
      "step": 13903
    },
    {
      "epoch": 0.7571283684901948,
      "grad_norm": 0.6204862537814627,
      "learning_rate": 0.00014276422809363957,
      "loss": 12.2634,
      "step": 13904
    },
    {
      "epoch": 0.7571828224867779,
      "grad_norm": 0.582736281918877,
      "learning_rate": 0.00014275625668009508,
      "loss": 12.2089,
      "step": 13905
    },
    {
      "epoch": 0.7572372764833609,
      "grad_norm": 0.5936936567696945,
      "learning_rate": 0.00014274828493407072,
      "loss": 12.2002,
      "step": 13906
    },
    {
      "epoch": 0.7572917304799439,
      "grad_norm": 0.5530303776248853,
      "learning_rate": 0.00014274031285562856,
      "loss": 12.2321,
      "step": 13907
    },
    {
      "epoch": 0.7573461844765269,
      "grad_norm": 0.5826100861165144,
      "learning_rate": 0.00014273234044483054,
      "loss": 12.3413,
      "step": 13908
    },
    {
      "epoch": 0.7574006384731099,
      "grad_norm": 0.5578995915437032,
      "learning_rate": 0.00014272436770173868,
      "loss": 12.2443,
      "step": 13909
    },
    {
      "epoch": 0.7574550924696929,
      "grad_norm": 0.5545115358555107,
      "learning_rate": 0.000142716394626415,
      "loss": 12.1602,
      "step": 13910
    },
    {
      "epoch": 0.757509546466276,
      "grad_norm": 0.6002801986882448,
      "learning_rate": 0.00014270842121892144,
      "loss": 12.2704,
      "step": 13911
    },
    {
      "epoch": 0.757564000462859,
      "grad_norm": 0.5875392696566134,
      "learning_rate": 0.00014270044747932006,
      "loss": 12.3167,
      "step": 13912
    },
    {
      "epoch": 0.757618454459442,
      "grad_norm": 0.6594591775214861,
      "learning_rate": 0.00014269247340767283,
      "loss": 12.3784,
      "step": 13913
    },
    {
      "epoch": 0.757672908456025,
      "grad_norm": 0.5947682770062622,
      "learning_rate": 0.00014268449900404175,
      "loss": 12.3146,
      "step": 13914
    },
    {
      "epoch": 0.757727362452608,
      "grad_norm": 0.6568834496108618,
      "learning_rate": 0.00014267652426848887,
      "loss": 12.1571,
      "step": 13915
    },
    {
      "epoch": 0.7577818164491911,
      "grad_norm": 0.6560720446581242,
      "learning_rate": 0.00014266854920107617,
      "loss": 12.1478,
      "step": 13916
    },
    {
      "epoch": 0.7578362704457741,
      "grad_norm": 0.5738304909572517,
      "learning_rate": 0.0001426605738018657,
      "loss": 12.2459,
      "step": 13917
    },
    {
      "epoch": 0.757890724442357,
      "grad_norm": 0.560306964319945,
      "learning_rate": 0.0001426525980709194,
      "loss": 12.3494,
      "step": 13918
    },
    {
      "epoch": 0.75794517843894,
      "grad_norm": 0.6763115699225631,
      "learning_rate": 0.00014264462200829937,
      "loss": 12.3879,
      "step": 13919
    },
    {
      "epoch": 0.757999632435523,
      "grad_norm": 0.5703290243771721,
      "learning_rate": 0.00014263664561406763,
      "loss": 12.3699,
      "step": 13920
    },
    {
      "epoch": 0.758054086432106,
      "grad_norm": 0.6625014384246614,
      "learning_rate": 0.00014262866888828615,
      "loss": 12.2679,
      "step": 13921
    },
    {
      "epoch": 0.7581085404286891,
      "grad_norm": 0.5769931413845019,
      "learning_rate": 0.000142620691831017,
      "loss": 12.2372,
      "step": 13922
    },
    {
      "epoch": 0.7581629944252721,
      "grad_norm": 0.5949061529616914,
      "learning_rate": 0.00014261271444232222,
      "loss": 12.2714,
      "step": 13923
    },
    {
      "epoch": 0.7582174484218551,
      "grad_norm": 0.5496679640085099,
      "learning_rate": 0.00014260473672226382,
      "loss": 12.2477,
      "step": 13924
    },
    {
      "epoch": 0.7582719024184381,
      "grad_norm": 0.6493771055430315,
      "learning_rate": 0.00014259675867090384,
      "loss": 12.3036,
      "step": 13925
    },
    {
      "epoch": 0.7583263564150211,
      "grad_norm": 0.5665578504631344,
      "learning_rate": 0.00014258878028830432,
      "loss": 12.3569,
      "step": 13926
    },
    {
      "epoch": 0.7583808104116041,
      "grad_norm": 0.553654878253341,
      "learning_rate": 0.0001425808015745273,
      "loss": 12.2306,
      "step": 13927
    },
    {
      "epoch": 0.7584352644081872,
      "grad_norm": 0.6336767633083233,
      "learning_rate": 0.00014257282252963483,
      "loss": 12.2879,
      "step": 13928
    },
    {
      "epoch": 0.7584897184047702,
      "grad_norm": 0.6957955270092684,
      "learning_rate": 0.00014256484315368896,
      "loss": 12.3103,
      "step": 13929
    },
    {
      "epoch": 0.7585441724013532,
      "grad_norm": 0.6296785833482869,
      "learning_rate": 0.00014255686344675171,
      "loss": 12.2726,
      "step": 13930
    },
    {
      "epoch": 0.7585986263979362,
      "grad_norm": 0.6146755812721879,
      "learning_rate": 0.00014254888340888518,
      "loss": 12.3361,
      "step": 13931
    },
    {
      "epoch": 0.7586530803945192,
      "grad_norm": 0.5695074475030054,
      "learning_rate": 0.00014254090304015136,
      "loss": 12.3646,
      "step": 13932
    },
    {
      "epoch": 0.7587075343911022,
      "grad_norm": 0.5958776096596177,
      "learning_rate": 0.00014253292234061237,
      "loss": 12.2512,
      "step": 13933
    },
    {
      "epoch": 0.7587619883876853,
      "grad_norm": 0.579919117409958,
      "learning_rate": 0.00014252494131033027,
      "loss": 12.2253,
      "step": 13934
    },
    {
      "epoch": 0.7588164423842683,
      "grad_norm": 0.5915421701951903,
      "learning_rate": 0.00014251695994936704,
      "loss": 12.2334,
      "step": 13935
    },
    {
      "epoch": 0.7588708963808513,
      "grad_norm": 0.6499479162687332,
      "learning_rate": 0.00014250897825778482,
      "loss": 12.4318,
      "step": 13936
    },
    {
      "epoch": 0.7589253503774342,
      "grad_norm": 0.5295834541077248,
      "learning_rate": 0.00014250099623564565,
      "loss": 12.1041,
      "step": 13937
    },
    {
      "epoch": 0.7589798043740172,
      "grad_norm": 0.5251911988552174,
      "learning_rate": 0.00014249301388301165,
      "loss": 12.2811,
      "step": 13938
    },
    {
      "epoch": 0.7590342583706002,
      "grad_norm": 0.6043959090041436,
      "learning_rate": 0.00014248503119994484,
      "loss": 12.246,
      "step": 13939
    },
    {
      "epoch": 0.7590887123671833,
      "grad_norm": 0.577799259201176,
      "learning_rate": 0.00014247704818650723,
      "loss": 12.1886,
      "step": 13940
    },
    {
      "epoch": 0.7591431663637663,
      "grad_norm": 0.6230601576377133,
      "learning_rate": 0.00014246906484276104,
      "loss": 12.2068,
      "step": 13941
    },
    {
      "epoch": 0.7591976203603493,
      "grad_norm": 0.6102940610921074,
      "learning_rate": 0.00014246108116876824,
      "loss": 12.1497,
      "step": 13942
    },
    {
      "epoch": 0.7592520743569323,
      "grad_norm": 0.638796409373731,
      "learning_rate": 0.00014245309716459098,
      "loss": 12.1821,
      "step": 13943
    },
    {
      "epoch": 0.7593065283535153,
      "grad_norm": 0.60346867893492,
      "learning_rate": 0.00014244511283029133,
      "loss": 12.2799,
      "step": 13944
    },
    {
      "epoch": 0.7593609823500984,
      "grad_norm": 0.5569590445930767,
      "learning_rate": 0.00014243712816593132,
      "loss": 12.1409,
      "step": 13945
    },
    {
      "epoch": 0.7594154363466814,
      "grad_norm": 0.6095387293968589,
      "learning_rate": 0.0001424291431715731,
      "loss": 12.3211,
      "step": 13946
    },
    {
      "epoch": 0.7594698903432644,
      "grad_norm": 0.6054626111032342,
      "learning_rate": 0.00014242115784727873,
      "loss": 12.281,
      "step": 13947
    },
    {
      "epoch": 0.7595243443398474,
      "grad_norm": 0.5724240846205761,
      "learning_rate": 0.00014241317219311036,
      "loss": 12.375,
      "step": 13948
    },
    {
      "epoch": 0.7595787983364304,
      "grad_norm": 0.561696636977615,
      "learning_rate": 0.00014240518620913005,
      "loss": 12.2647,
      "step": 13949
    },
    {
      "epoch": 0.7596332523330134,
      "grad_norm": 0.5821499988373755,
      "learning_rate": 0.0001423971998953999,
      "loss": 12.1901,
      "step": 13950
    },
    {
      "epoch": 0.7596877063295965,
      "grad_norm": 0.5776864442082547,
      "learning_rate": 0.000142389213251982,
      "loss": 12.2944,
      "step": 13951
    },
    {
      "epoch": 0.7597421603261795,
      "grad_norm": 0.5354845840539186,
      "learning_rate": 0.00014238122627893845,
      "loss": 12.1452,
      "step": 13952
    },
    {
      "epoch": 0.7597966143227625,
      "grad_norm": 0.5960205210876889,
      "learning_rate": 0.0001423732389763314,
      "loss": 12.0842,
      "step": 13953
    },
    {
      "epoch": 0.7598510683193455,
      "grad_norm": 0.6156684709857669,
      "learning_rate": 0.00014236525134422295,
      "loss": 12.2972,
      "step": 13954
    },
    {
      "epoch": 0.7599055223159285,
      "grad_norm": 0.5610810124186224,
      "learning_rate": 0.00014235726338267517,
      "loss": 12.2933,
      "step": 13955
    },
    {
      "epoch": 0.7599599763125114,
      "grad_norm": 0.6195395184747021,
      "learning_rate": 0.00014234927509175022,
      "loss": 12.2857,
      "step": 13956
    },
    {
      "epoch": 0.7600144303090945,
      "grad_norm": 0.5602682088640545,
      "learning_rate": 0.00014234128647151018,
      "loss": 12.093,
      "step": 13957
    },
    {
      "epoch": 0.7600688843056775,
      "grad_norm": 0.722577336954991,
      "learning_rate": 0.00014233329752201722,
      "loss": 12.2452,
      "step": 13958
    },
    {
      "epoch": 0.7601233383022605,
      "grad_norm": 0.5471100875030792,
      "learning_rate": 0.00014232530824333348,
      "loss": 12.2115,
      "step": 13959
    },
    {
      "epoch": 0.7601777922988435,
      "grad_norm": 0.6606957761725386,
      "learning_rate": 0.000142317318635521,
      "loss": 12.2199,
      "step": 13960
    },
    {
      "epoch": 0.7602322462954265,
      "grad_norm": 0.5268687412814926,
      "learning_rate": 0.00014230932869864195,
      "loss": 12.092,
      "step": 13961
    },
    {
      "epoch": 0.7602867002920095,
      "grad_norm": 0.607743751189324,
      "learning_rate": 0.00014230133843275847,
      "loss": 12.2734,
      "step": 13962
    },
    {
      "epoch": 0.7603411542885926,
      "grad_norm": 0.5600065514983386,
      "learning_rate": 0.00014229334783793268,
      "loss": 12.0704,
      "step": 13963
    },
    {
      "epoch": 0.7603956082851756,
      "grad_norm": 0.5695028254072064,
      "learning_rate": 0.00014228535691422674,
      "loss": 12.2109,
      "step": 13964
    },
    {
      "epoch": 0.7604500622817586,
      "grad_norm": 0.5815235519116067,
      "learning_rate": 0.0001422773656617028,
      "loss": 12.1929,
      "step": 13965
    },
    {
      "epoch": 0.7605045162783416,
      "grad_norm": 0.5251375307774158,
      "learning_rate": 0.00014226937408042293,
      "loss": 12.2333,
      "step": 13966
    },
    {
      "epoch": 0.7605589702749246,
      "grad_norm": 0.5857315453039587,
      "learning_rate": 0.00014226138217044937,
      "loss": 12.2082,
      "step": 13967
    },
    {
      "epoch": 0.7606134242715076,
      "grad_norm": 0.577963728586446,
      "learning_rate": 0.00014225338993184417,
      "loss": 12.2695,
      "step": 13968
    },
    {
      "epoch": 0.7606678782680907,
      "grad_norm": 0.5513473816586468,
      "learning_rate": 0.0001422453973646695,
      "loss": 12.3888,
      "step": 13969
    },
    {
      "epoch": 0.7607223322646737,
      "grad_norm": 0.6016650383635523,
      "learning_rate": 0.0001422374044689876,
      "loss": 12.1849,
      "step": 13970
    },
    {
      "epoch": 0.7607767862612567,
      "grad_norm": 0.6370471484189492,
      "learning_rate": 0.00014222941124486052,
      "loss": 12.3189,
      "step": 13971
    },
    {
      "epoch": 0.7608312402578397,
      "grad_norm": 0.5499723366235015,
      "learning_rate": 0.00014222141769235048,
      "loss": 12.2014,
      "step": 13972
    },
    {
      "epoch": 0.7608856942544227,
      "grad_norm": 0.6698610911475733,
      "learning_rate": 0.0001422134238115196,
      "loss": 12.2091,
      "step": 13973
    },
    {
      "epoch": 0.7609401482510056,
      "grad_norm": 0.599557631459671,
      "learning_rate": 0.00014220542960243003,
      "loss": 12.2612,
      "step": 13974
    },
    {
      "epoch": 0.7609946022475887,
      "grad_norm": 0.6551527802649024,
      "learning_rate": 0.000142197435065144,
      "loss": 12.2865,
      "step": 13975
    },
    {
      "epoch": 0.7610490562441717,
      "grad_norm": 0.6157311467055842,
      "learning_rate": 0.00014218944019972363,
      "loss": 12.3061,
      "step": 13976
    },
    {
      "epoch": 0.7611035102407547,
      "grad_norm": 0.5781439219031981,
      "learning_rate": 0.0001421814450062311,
      "loss": 12.1868,
      "step": 13977
    },
    {
      "epoch": 0.7611579642373377,
      "grad_norm": 0.5501211790005208,
      "learning_rate": 0.00014217344948472857,
      "loss": 12.2197,
      "step": 13978
    },
    {
      "epoch": 0.7612124182339207,
      "grad_norm": 0.5240842381295177,
      "learning_rate": 0.00014216545363527822,
      "loss": 12.1797,
      "step": 13979
    },
    {
      "epoch": 0.7612668722305038,
      "grad_norm": 0.6189210529605181,
      "learning_rate": 0.00014215745745794226,
      "loss": 12.2871,
      "step": 13980
    },
    {
      "epoch": 0.7613213262270868,
      "grad_norm": 0.5951223275828108,
      "learning_rate": 0.0001421494609527828,
      "loss": 12.3688,
      "step": 13981
    },
    {
      "epoch": 0.7613757802236698,
      "grad_norm": 0.5921341999404179,
      "learning_rate": 0.0001421414641198621,
      "loss": 12.3034,
      "step": 13982
    },
    {
      "epoch": 0.7614302342202528,
      "grad_norm": 0.655618968539555,
      "learning_rate": 0.00014213346695924229,
      "loss": 12.3406,
      "step": 13983
    },
    {
      "epoch": 0.7614846882168358,
      "grad_norm": 0.5432306320692075,
      "learning_rate": 0.00014212546947098558,
      "loss": 12.24,
      "step": 13984
    },
    {
      "epoch": 0.7615391422134188,
      "grad_norm": 0.6036883548019074,
      "learning_rate": 0.00014211747165515415,
      "loss": 12.0426,
      "step": 13985
    },
    {
      "epoch": 0.7615935962100019,
      "grad_norm": 0.5124128687561235,
      "learning_rate": 0.00014210947351181018,
      "loss": 12.1231,
      "step": 13986
    },
    {
      "epoch": 0.7616480502065849,
      "grad_norm": 0.5640171129791517,
      "learning_rate": 0.0001421014750410159,
      "loss": 12.2459,
      "step": 13987
    },
    {
      "epoch": 0.7617025042031679,
      "grad_norm": 0.5533932551479256,
      "learning_rate": 0.0001420934762428335,
      "loss": 12.2373,
      "step": 13988
    },
    {
      "epoch": 0.7617569581997509,
      "grad_norm": 0.7231313373660987,
      "learning_rate": 0.00014208547711732516,
      "loss": 12.2069,
      "step": 13989
    },
    {
      "epoch": 0.7618114121963339,
      "grad_norm": 0.5883848264970278,
      "learning_rate": 0.0001420774776645531,
      "loss": 12.2925,
      "step": 13990
    },
    {
      "epoch": 0.7618658661929169,
      "grad_norm": 0.6112306520637002,
      "learning_rate": 0.00014206947788457952,
      "loss": 12.2716,
      "step": 13991
    },
    {
      "epoch": 0.7619203201895,
      "grad_norm": 0.5651211551615413,
      "learning_rate": 0.00014206147777746662,
      "loss": 12.315,
      "step": 13992
    },
    {
      "epoch": 0.761974774186083,
      "grad_norm": 0.5509869436075503,
      "learning_rate": 0.0001420534773432766,
      "loss": 12.346,
      "step": 13993
    },
    {
      "epoch": 0.7620292281826659,
      "grad_norm": 0.5918797097019901,
      "learning_rate": 0.00014204547658207173,
      "loss": 12.276,
      "step": 13994
    },
    {
      "epoch": 0.7620836821792489,
      "grad_norm": 0.6708119074943824,
      "learning_rate": 0.00014203747549391412,
      "loss": 12.3249,
      "step": 13995
    },
    {
      "epoch": 0.7621381361758319,
      "grad_norm": 0.6152818078894708,
      "learning_rate": 0.0001420294740788661,
      "loss": 12.153,
      "step": 13996
    },
    {
      "epoch": 0.7621925901724149,
      "grad_norm": 0.6318583507656009,
      "learning_rate": 0.00014202147233698982,
      "loss": 12.3252,
      "step": 13997
    },
    {
      "epoch": 0.762247044168998,
      "grad_norm": 0.5700684553606541,
      "learning_rate": 0.0001420134702683475,
      "loss": 12.2328,
      "step": 13998
    },
    {
      "epoch": 0.762301498165581,
      "grad_norm": 0.5645040981476561,
      "learning_rate": 0.00014200546787300144,
      "loss": 12.1694,
      "step": 13999
    },
    {
      "epoch": 0.762355952162164,
      "grad_norm": 0.688139703247829,
      "learning_rate": 0.00014199746515101383,
      "loss": 12.3021,
      "step": 14000
    },
    {
      "epoch": 0.762410406158747,
      "grad_norm": 0.9161755471060709,
      "learning_rate": 0.0001419894621024468,
      "loss": 12.309,
      "step": 14001
    },
    {
      "epoch": 0.76246486015533,
      "grad_norm": 0.6852488173033148,
      "learning_rate": 0.00014198145872736272,
      "loss": 12.3005,
      "step": 14002
    },
    {
      "epoch": 0.762519314151913,
      "grad_norm": 0.6742687447311347,
      "learning_rate": 0.0001419734550258238,
      "loss": 12.3142,
      "step": 14003
    },
    {
      "epoch": 0.7625737681484961,
      "grad_norm": 0.6925915386799684,
      "learning_rate": 0.00014196545099789224,
      "loss": 12.2864,
      "step": 14004
    },
    {
      "epoch": 0.7626282221450791,
      "grad_norm": 0.595301777501327,
      "learning_rate": 0.00014195744664363032,
      "loss": 12.0688,
      "step": 14005
    },
    {
      "epoch": 0.7626826761416621,
      "grad_norm": 0.5281359073325815,
      "learning_rate": 0.00014194944196310017,
      "loss": 12.2558,
      "step": 14006
    },
    {
      "epoch": 0.7627371301382451,
      "grad_norm": 0.5523714966665472,
      "learning_rate": 0.00014194143695636421,
      "loss": 12.2697,
      "step": 14007
    },
    {
      "epoch": 0.7627915841348281,
      "grad_norm": 0.6080070137033171,
      "learning_rate": 0.00014193343162348455,
      "loss": 12.2527,
      "step": 14008
    },
    {
      "epoch": 0.762846038131411,
      "grad_norm": 0.6627933267237143,
      "learning_rate": 0.00014192542596452355,
      "loss": 12.0337,
      "step": 14009
    },
    {
      "epoch": 0.7629004921279942,
      "grad_norm": 0.6805001156429517,
      "learning_rate": 0.0001419174199795434,
      "loss": 12.2574,
      "step": 14010
    },
    {
      "epoch": 0.7629549461245771,
      "grad_norm": 0.6161146392108356,
      "learning_rate": 0.00014190941366860633,
      "loss": 12.1328,
      "step": 14011
    },
    {
      "epoch": 0.7630094001211601,
      "grad_norm": 0.5934805056499477,
      "learning_rate": 0.0001419014070317746,
      "loss": 12.2775,
      "step": 14012
    },
    {
      "epoch": 0.7630638541177431,
      "grad_norm": 0.6779302289590717,
      "learning_rate": 0.00014189340006911055,
      "loss": 12.3298,
      "step": 14013
    },
    {
      "epoch": 0.7631183081143261,
      "grad_norm": 0.6070414779581786,
      "learning_rate": 0.00014188539278067643,
      "loss": 12.219,
      "step": 14014
    },
    {
      "epoch": 0.7631727621109092,
      "grad_norm": 0.6993064018098617,
      "learning_rate": 0.0001418773851665344,
      "loss": 12.3345,
      "step": 14015
    },
    {
      "epoch": 0.7632272161074922,
      "grad_norm": 0.6305373858898516,
      "learning_rate": 0.00014186937722674683,
      "loss": 12.1297,
      "step": 14016
    },
    {
      "epoch": 0.7632816701040752,
      "grad_norm": 0.6013170403095079,
      "learning_rate": 0.00014186136896137598,
      "loss": 12.265,
      "step": 14017
    },
    {
      "epoch": 0.7633361241006582,
      "grad_norm": 0.5879012594309635,
      "learning_rate": 0.00014185336037048408,
      "loss": 12.2657,
      "step": 14018
    },
    {
      "epoch": 0.7633905780972412,
      "grad_norm": 0.5452413532638681,
      "learning_rate": 0.00014184535145413344,
      "loss": 12.1823,
      "step": 14019
    },
    {
      "epoch": 0.7634450320938242,
      "grad_norm": 0.6873152094810032,
      "learning_rate": 0.00014183734221238635,
      "loss": 12.4098,
      "step": 14020
    },
    {
      "epoch": 0.7634994860904073,
      "grad_norm": 0.6416778737618622,
      "learning_rate": 0.00014182933264530502,
      "loss": 12.283,
      "step": 14021
    },
    {
      "epoch": 0.7635539400869903,
      "grad_norm": 0.6437179415426925,
      "learning_rate": 0.00014182132275295182,
      "loss": 12.3234,
      "step": 14022
    },
    {
      "epoch": 0.7636083940835733,
      "grad_norm": 0.556334606335069,
      "learning_rate": 0.00014181331253538897,
      "loss": 12.2012,
      "step": 14023
    },
    {
      "epoch": 0.7636628480801563,
      "grad_norm": 0.5543179396361387,
      "learning_rate": 0.00014180530199267883,
      "loss": 12.2312,
      "step": 14024
    },
    {
      "epoch": 0.7637173020767393,
      "grad_norm": 0.5500295422833468,
      "learning_rate": 0.00014179729112488365,
      "loss": 12.1844,
      "step": 14025
    },
    {
      "epoch": 0.7637717560733223,
      "grad_norm": 0.5425630994642155,
      "learning_rate": 0.0001417892799320657,
      "loss": 12.2491,
      "step": 14026
    },
    {
      "epoch": 0.7638262100699054,
      "grad_norm": 0.6669730471823939,
      "learning_rate": 0.00014178126841428733,
      "loss": 12.3245,
      "step": 14027
    },
    {
      "epoch": 0.7638806640664884,
      "grad_norm": 0.6372441479102375,
      "learning_rate": 0.00014177325657161079,
      "loss": 12.121,
      "step": 14028
    },
    {
      "epoch": 0.7639351180630714,
      "grad_norm": 0.5936749439861729,
      "learning_rate": 0.00014176524440409838,
      "loss": 12.1716,
      "step": 14029
    },
    {
      "epoch": 0.7639895720596543,
      "grad_norm": 0.5548252061186839,
      "learning_rate": 0.00014175723191181246,
      "loss": 12.2135,
      "step": 14030
    },
    {
      "epoch": 0.7640440260562373,
      "grad_norm": 0.6081908661433257,
      "learning_rate": 0.00014174921909481528,
      "loss": 12.1114,
      "step": 14031
    },
    {
      "epoch": 0.7640984800528203,
      "grad_norm": 0.5962901142755918,
      "learning_rate": 0.00014174120595316918,
      "loss": 12.3284,
      "step": 14032
    },
    {
      "epoch": 0.7641529340494034,
      "grad_norm": 0.6069543878257613,
      "learning_rate": 0.00014173319248693647,
      "loss": 12.2171,
      "step": 14033
    },
    {
      "epoch": 0.7642073880459864,
      "grad_norm": 0.5340676390046819,
      "learning_rate": 0.00014172517869617942,
      "loss": 12.1292,
      "step": 14034
    },
    {
      "epoch": 0.7642618420425694,
      "grad_norm": 0.5802220742321209,
      "learning_rate": 0.00014171716458096043,
      "loss": 12.0764,
      "step": 14035
    },
    {
      "epoch": 0.7643162960391524,
      "grad_norm": 0.5425549092592509,
      "learning_rate": 0.00014170915014134175,
      "loss": 12.1105,
      "step": 14036
    },
    {
      "epoch": 0.7643707500357354,
      "grad_norm": 0.6521458879338885,
      "learning_rate": 0.00014170113537738572,
      "loss": 12.3549,
      "step": 14037
    },
    {
      "epoch": 0.7644252040323184,
      "grad_norm": 0.5721445904275433,
      "learning_rate": 0.00014169312028915467,
      "loss": 12.2843,
      "step": 14038
    },
    {
      "epoch": 0.7644796580289015,
      "grad_norm": 0.6204254611658935,
      "learning_rate": 0.00014168510487671095,
      "loss": 12.1998,
      "step": 14039
    },
    {
      "epoch": 0.7645341120254845,
      "grad_norm": 0.5713946970086126,
      "learning_rate": 0.00014167708914011683,
      "loss": 12.3153,
      "step": 14040
    },
    {
      "epoch": 0.7645885660220675,
      "grad_norm": 0.5532519282768511,
      "learning_rate": 0.00014166907307943468,
      "loss": 12.1961,
      "step": 14041
    },
    {
      "epoch": 0.7646430200186505,
      "grad_norm": 0.5569680627789245,
      "learning_rate": 0.00014166105669472686,
      "loss": 12.2786,
      "step": 14042
    },
    {
      "epoch": 0.7646974740152335,
      "grad_norm": 0.5536494835696807,
      "learning_rate": 0.00014165303998605567,
      "loss": 12.2799,
      "step": 14043
    },
    {
      "epoch": 0.7647519280118165,
      "grad_norm": 0.681863473872942,
      "learning_rate": 0.00014164502295348344,
      "loss": 12.2401,
      "step": 14044
    },
    {
      "epoch": 0.7648063820083996,
      "grad_norm": 0.5696680388291402,
      "learning_rate": 0.00014163700559707251,
      "loss": 12.0994,
      "step": 14045
    },
    {
      "epoch": 0.7648608360049826,
      "grad_norm": 0.5586529453896714,
      "learning_rate": 0.00014162898791688527,
      "loss": 12.2236,
      "step": 14046
    },
    {
      "epoch": 0.7649152900015656,
      "grad_norm": 0.6305781565597973,
      "learning_rate": 0.000141620969912984,
      "loss": 12.3263,
      "step": 14047
    },
    {
      "epoch": 0.7649697439981485,
      "grad_norm": 0.5753330948747856,
      "learning_rate": 0.00014161295158543117,
      "loss": 12.3274,
      "step": 14048
    },
    {
      "epoch": 0.7650241979947315,
      "grad_norm": 0.5359807136362528,
      "learning_rate": 0.000141604932934289,
      "loss": 12.2058,
      "step": 14049
    },
    {
      "epoch": 0.7650786519913146,
      "grad_norm": 0.5956265092053136,
      "learning_rate": 0.00014159691395961986,
      "loss": 12.2686,
      "step": 14050
    },
    {
      "epoch": 0.7651331059878976,
      "grad_norm": 0.5829785198998221,
      "learning_rate": 0.0001415888946614862,
      "loss": 12.2445,
      "step": 14051
    },
    {
      "epoch": 0.7651875599844806,
      "grad_norm": 0.5570639705465623,
      "learning_rate": 0.00014158087503995029,
      "loss": 12.2025,
      "step": 14052
    },
    {
      "epoch": 0.7652420139810636,
      "grad_norm": 0.6362187285845966,
      "learning_rate": 0.00014157285509507452,
      "loss": 12.3384,
      "step": 14053
    },
    {
      "epoch": 0.7652964679776466,
      "grad_norm": 0.5309991837853475,
      "learning_rate": 0.00014156483482692127,
      "loss": 12.2158,
      "step": 14054
    },
    {
      "epoch": 0.7653509219742296,
      "grad_norm": 0.5545637757522944,
      "learning_rate": 0.00014155681423555288,
      "loss": 12.1136,
      "step": 14055
    },
    {
      "epoch": 0.7654053759708127,
      "grad_norm": 0.6145129134489931,
      "learning_rate": 0.00014154879332103174,
      "loss": 12.1744,
      "step": 14056
    },
    {
      "epoch": 0.7654598299673957,
      "grad_norm": 0.5123633665586574,
      "learning_rate": 0.00014154077208342023,
      "loss": 12.1632,
      "step": 14057
    },
    {
      "epoch": 0.7655142839639787,
      "grad_norm": 0.5908039187960368,
      "learning_rate": 0.00014153275052278068,
      "loss": 12.2842,
      "step": 14058
    },
    {
      "epoch": 0.7655687379605617,
      "grad_norm": 0.5412909082059538,
      "learning_rate": 0.00014152472863917555,
      "loss": 12.0648,
      "step": 14059
    },
    {
      "epoch": 0.7656231919571447,
      "grad_norm": 0.6130141862215384,
      "learning_rate": 0.00014151670643266712,
      "loss": 12.1749,
      "step": 14060
    },
    {
      "epoch": 0.7656776459537277,
      "grad_norm": 0.6770562027867211,
      "learning_rate": 0.0001415086839033178,
      "loss": 12.4342,
      "step": 14061
    },
    {
      "epoch": 0.7657320999503108,
      "grad_norm": 0.6104892574412778,
      "learning_rate": 0.00014150066105119002,
      "loss": 12.3244,
      "step": 14062
    },
    {
      "epoch": 0.7657865539468938,
      "grad_norm": 0.556694748689259,
      "learning_rate": 0.00014149263787634615,
      "loss": 12.1292,
      "step": 14063
    },
    {
      "epoch": 0.7658410079434768,
      "grad_norm": 0.532048703486904,
      "learning_rate": 0.00014148461437884857,
      "loss": 12.1891,
      "step": 14064
    },
    {
      "epoch": 0.7658954619400598,
      "grad_norm": 0.5730215255960645,
      "learning_rate": 0.0001414765905587597,
      "loss": 12.272,
      "step": 14065
    },
    {
      "epoch": 0.7659499159366427,
      "grad_norm": 0.7000139721440017,
      "learning_rate": 0.00014146856641614184,
      "loss": 12.1002,
      "step": 14066
    },
    {
      "epoch": 0.7660043699332257,
      "grad_norm": 0.6843534527260816,
      "learning_rate": 0.00014146054195105748,
      "loss": 12.3384,
      "step": 14067
    },
    {
      "epoch": 0.7660588239298088,
      "grad_norm": 0.5634617117221637,
      "learning_rate": 0.00014145251716356897,
      "loss": 12.3222,
      "step": 14068
    },
    {
      "epoch": 0.7661132779263918,
      "grad_norm": 0.5687279399361462,
      "learning_rate": 0.00014144449205373877,
      "loss": 12.2644,
      "step": 14069
    },
    {
      "epoch": 0.7661677319229748,
      "grad_norm": 0.6051485761650688,
      "learning_rate": 0.00014143646662162927,
      "loss": 12.2573,
      "step": 14070
    },
    {
      "epoch": 0.7662221859195578,
      "grad_norm": 0.5839936942791063,
      "learning_rate": 0.00014142844086730282,
      "loss": 12.2485,
      "step": 14071
    },
    {
      "epoch": 0.7662766399161408,
      "grad_norm": 0.5517736427541252,
      "learning_rate": 0.00014142041479082185,
      "loss": 12.116,
      "step": 14072
    },
    {
      "epoch": 0.7663310939127238,
      "grad_norm": 0.5854388894353925,
      "learning_rate": 0.00014141238839224883,
      "loss": 12.3393,
      "step": 14073
    },
    {
      "epoch": 0.7663855479093069,
      "grad_norm": 0.6314116918440781,
      "learning_rate": 0.0001414043616716461,
      "loss": 12.3267,
      "step": 14074
    },
    {
      "epoch": 0.7664400019058899,
      "grad_norm": 0.5714963377609871,
      "learning_rate": 0.00014139633462907614,
      "loss": 12.2609,
      "step": 14075
    },
    {
      "epoch": 0.7664944559024729,
      "grad_norm": 0.6137410946530638,
      "learning_rate": 0.00014138830726460132,
      "loss": 12.2717,
      "step": 14076
    },
    {
      "epoch": 0.7665489098990559,
      "grad_norm": 0.5546721402047556,
      "learning_rate": 0.0001413802795782841,
      "loss": 12.3057,
      "step": 14077
    },
    {
      "epoch": 0.7666033638956389,
      "grad_norm": 0.6403782377884639,
      "learning_rate": 0.00014137225157018684,
      "loss": 12.2847,
      "step": 14078
    },
    {
      "epoch": 0.766657817892222,
      "grad_norm": 0.611941356520709,
      "learning_rate": 0.00014136422324037207,
      "loss": 12.3098,
      "step": 14079
    },
    {
      "epoch": 0.766712271888805,
      "grad_norm": 0.6050520607106942,
      "learning_rate": 0.00014135619458890215,
      "loss": 12.3527,
      "step": 14080
    },
    {
      "epoch": 0.766766725885388,
      "grad_norm": 0.5664621409668334,
      "learning_rate": 0.0001413481656158395,
      "loss": 12.2701,
      "step": 14081
    },
    {
      "epoch": 0.766821179881971,
      "grad_norm": 0.5631607069562086,
      "learning_rate": 0.00014134013632124658,
      "loss": 12.1803,
      "step": 14082
    },
    {
      "epoch": 0.766875633878554,
      "grad_norm": 0.5437281052141982,
      "learning_rate": 0.00014133210670518585,
      "loss": 12.1745,
      "step": 14083
    },
    {
      "epoch": 0.766930087875137,
      "grad_norm": 0.5991515782094939,
      "learning_rate": 0.0001413240767677197,
      "loss": 12.2471,
      "step": 14084
    },
    {
      "epoch": 0.76698454187172,
      "grad_norm": 0.5363419925328808,
      "learning_rate": 0.00014131604650891063,
      "loss": 12.1673,
      "step": 14085
    },
    {
      "epoch": 0.767038995868303,
      "grad_norm": 0.6946396140367467,
      "learning_rate": 0.00014130801592882107,
      "loss": 12.2711,
      "step": 14086
    },
    {
      "epoch": 0.767093449864886,
      "grad_norm": 0.6315343447460228,
      "learning_rate": 0.00014129998502751342,
      "loss": 12.2857,
      "step": 14087
    },
    {
      "epoch": 0.767147903861469,
      "grad_norm": 0.5829576252690439,
      "learning_rate": 0.00014129195380505017,
      "loss": 12.225,
      "step": 14088
    },
    {
      "epoch": 0.767202357858052,
      "grad_norm": 0.6156299807501912,
      "learning_rate": 0.0001412839222614937,
      "loss": 12.2652,
      "step": 14089
    },
    {
      "epoch": 0.767256811854635,
      "grad_norm": 0.6254524138851091,
      "learning_rate": 0.00014127589039690663,
      "loss": 12.2521,
      "step": 14090
    },
    {
      "epoch": 0.7673112658512181,
      "grad_norm": 0.6034966930210365,
      "learning_rate": 0.00014126785821135126,
      "loss": 12.2941,
      "step": 14091
    },
    {
      "epoch": 0.7673657198478011,
      "grad_norm": 0.6437585849859505,
      "learning_rate": 0.00014125982570489012,
      "loss": 12.202,
      "step": 14092
    },
    {
      "epoch": 0.7674201738443841,
      "grad_norm": 0.5373181078765519,
      "learning_rate": 0.00014125179287758564,
      "loss": 12.1046,
      "step": 14093
    },
    {
      "epoch": 0.7674746278409671,
      "grad_norm": 0.6572518223600609,
      "learning_rate": 0.0001412437597295003,
      "loss": 12.4171,
      "step": 14094
    },
    {
      "epoch": 0.7675290818375501,
      "grad_norm": 0.6232379859928363,
      "learning_rate": 0.00014123572626069657,
      "loss": 12.2784,
      "step": 14095
    },
    {
      "epoch": 0.7675835358341331,
      "grad_norm": 0.6577703484439251,
      "learning_rate": 0.00014122769247123694,
      "loss": 12.199,
      "step": 14096
    },
    {
      "epoch": 0.7676379898307162,
      "grad_norm": 0.5299622761718101,
      "learning_rate": 0.00014121965836118384,
      "loss": 12.2502,
      "step": 14097
    },
    {
      "epoch": 0.7676924438272992,
      "grad_norm": 0.6044188745898461,
      "learning_rate": 0.00014121162393059976,
      "loss": 12.2248,
      "step": 14098
    },
    {
      "epoch": 0.7677468978238822,
      "grad_norm": 0.6542289909898396,
      "learning_rate": 0.0001412035891795472,
      "loss": 12.1898,
      "step": 14099
    },
    {
      "epoch": 0.7678013518204652,
      "grad_norm": 0.5732423193009067,
      "learning_rate": 0.0001411955541080886,
      "loss": 12.3215,
      "step": 14100
    },
    {
      "epoch": 0.7678558058170482,
      "grad_norm": 0.6485999873433785,
      "learning_rate": 0.00014118751871628645,
      "loss": 12.341,
      "step": 14101
    },
    {
      "epoch": 0.7679102598136311,
      "grad_norm": 0.6092602520664506,
      "learning_rate": 0.0001411794830042033,
      "loss": 12.2059,
      "step": 14102
    },
    {
      "epoch": 0.7679647138102143,
      "grad_norm": 0.5593846752229331,
      "learning_rate": 0.00014117144697190154,
      "loss": 12.2195,
      "step": 14103
    },
    {
      "epoch": 0.7680191678067972,
      "grad_norm": 0.5751117900315875,
      "learning_rate": 0.00014116341061944372,
      "loss": 12.1362,
      "step": 14104
    },
    {
      "epoch": 0.7680736218033802,
      "grad_norm": 0.6063510707625296,
      "learning_rate": 0.00014115537394689232,
      "loss": 12.2299,
      "step": 14105
    },
    {
      "epoch": 0.7681280757999632,
      "grad_norm": 0.597977108682404,
      "learning_rate": 0.0001411473369543098,
      "loss": 12.0905,
      "step": 14106
    },
    {
      "epoch": 0.7681825297965462,
      "grad_norm": 0.5748477244036646,
      "learning_rate": 0.0001411392996417587,
      "loss": 12.2832,
      "step": 14107
    },
    {
      "epoch": 0.7682369837931292,
      "grad_norm": 0.6037677483892898,
      "learning_rate": 0.00014113126200930153,
      "loss": 12.2684,
      "step": 14108
    },
    {
      "epoch": 0.7682914377897123,
      "grad_norm": 0.6156244411231219,
      "learning_rate": 0.00014112322405700076,
      "loss": 12.2037,
      "step": 14109
    },
    {
      "epoch": 0.7683458917862953,
      "grad_norm": 0.6140588551870635,
      "learning_rate": 0.0001411151857849189,
      "loss": 12.3049,
      "step": 14110
    },
    {
      "epoch": 0.7684003457828783,
      "grad_norm": 0.6866071509031823,
      "learning_rate": 0.00014110714719311847,
      "loss": 12.1586,
      "step": 14111
    },
    {
      "epoch": 0.7684547997794613,
      "grad_norm": 0.6114492089687162,
      "learning_rate": 0.00014109910828166196,
      "loss": 12.2286,
      "step": 14112
    },
    {
      "epoch": 0.7685092537760443,
      "grad_norm": 0.5802988526973467,
      "learning_rate": 0.0001410910690506119,
      "loss": 12.2486,
      "step": 14113
    },
    {
      "epoch": 0.7685637077726274,
      "grad_norm": 0.6874522798193882,
      "learning_rate": 0.00014108302950003077,
      "loss": 12.1485,
      "step": 14114
    },
    {
      "epoch": 0.7686181617692104,
      "grad_norm": 0.6517853430434413,
      "learning_rate": 0.00014107498962998114,
      "loss": 12.317,
      "step": 14115
    },
    {
      "epoch": 0.7686726157657934,
      "grad_norm": 0.6469399529099216,
      "learning_rate": 0.0001410669494405255,
      "loss": 12.1699,
      "step": 14116
    },
    {
      "epoch": 0.7687270697623764,
      "grad_norm": 0.6214515150946013,
      "learning_rate": 0.00014105890893172636,
      "loss": 12.3012,
      "step": 14117
    },
    {
      "epoch": 0.7687815237589594,
      "grad_norm": 0.6367902841549076,
      "learning_rate": 0.00014105086810364625,
      "loss": 12.3051,
      "step": 14118
    },
    {
      "epoch": 0.7688359777555424,
      "grad_norm": 0.5850425284936396,
      "learning_rate": 0.00014104282695634771,
      "loss": 12.2222,
      "step": 14119
    },
    {
      "epoch": 0.7688904317521255,
      "grad_norm": 0.6108114424575637,
      "learning_rate": 0.0001410347854898933,
      "loss": 12.2159,
      "step": 14120
    },
    {
      "epoch": 0.7689448857487085,
      "grad_norm": 0.6282073225540852,
      "learning_rate": 0.00014102674370434546,
      "loss": 12.2697,
      "step": 14121
    },
    {
      "epoch": 0.7689993397452914,
      "grad_norm": 0.6241786619890566,
      "learning_rate": 0.00014101870159976683,
      "loss": 12.1934,
      "step": 14122
    },
    {
      "epoch": 0.7690537937418744,
      "grad_norm": 0.5502129213330372,
      "learning_rate": 0.00014101065917621988,
      "loss": 12.1528,
      "step": 14123
    },
    {
      "epoch": 0.7691082477384574,
      "grad_norm": 0.6558359504268575,
      "learning_rate": 0.00014100261643376717,
      "loss": 12.1864,
      "step": 14124
    },
    {
      "epoch": 0.7691627017350404,
      "grad_norm": 0.610417619158311,
      "learning_rate": 0.00014099457337247127,
      "loss": 12.201,
      "step": 14125
    },
    {
      "epoch": 0.7692171557316235,
      "grad_norm": 0.5636968506474196,
      "learning_rate": 0.00014098652999239462,
      "loss": 12.1071,
      "step": 14126
    },
    {
      "epoch": 0.7692716097282065,
      "grad_norm": 0.6876986715271385,
      "learning_rate": 0.0001409784862935999,
      "loss": 12.2103,
      "step": 14127
    },
    {
      "epoch": 0.7693260637247895,
      "grad_norm": 0.5493141280887939,
      "learning_rate": 0.00014097044227614954,
      "loss": 12.2582,
      "step": 14128
    },
    {
      "epoch": 0.7693805177213725,
      "grad_norm": 0.5951341346995529,
      "learning_rate": 0.0001409623979401062,
      "loss": 12.2833,
      "step": 14129
    },
    {
      "epoch": 0.7694349717179555,
      "grad_norm": 0.6742088084590737,
      "learning_rate": 0.00014095435328553239,
      "loss": 12.3141,
      "step": 14130
    },
    {
      "epoch": 0.7694894257145385,
      "grad_norm": 0.5384175217702536,
      "learning_rate": 0.00014094630831249064,
      "loss": 12.2142,
      "step": 14131
    },
    {
      "epoch": 0.7695438797111216,
      "grad_norm": 0.6060570398527729,
      "learning_rate": 0.00014093826302104351,
      "loss": 12.2129,
      "step": 14132
    },
    {
      "epoch": 0.7695983337077046,
      "grad_norm": 0.7023543612414251,
      "learning_rate": 0.0001409302174112536,
      "loss": 12.4134,
      "step": 14133
    },
    {
      "epoch": 0.7696527877042876,
      "grad_norm": 0.5842896036172063,
      "learning_rate": 0.00014092217148318348,
      "loss": 12.3908,
      "step": 14134
    },
    {
      "epoch": 0.7697072417008706,
      "grad_norm": 0.5804160737393887,
      "learning_rate": 0.00014091412523689566,
      "loss": 12.2307,
      "step": 14135
    },
    {
      "epoch": 0.7697616956974536,
      "grad_norm": 0.5472274719740815,
      "learning_rate": 0.00014090607867245274,
      "loss": 12.1742,
      "step": 14136
    },
    {
      "epoch": 0.7698161496940366,
      "grad_norm": 0.5146133222305155,
      "learning_rate": 0.00014089803178991733,
      "loss": 12.2048,
      "step": 14137
    },
    {
      "epoch": 0.7698706036906197,
      "grad_norm": 0.5899090148481555,
      "learning_rate": 0.00014088998458935192,
      "loss": 12.2604,
      "step": 14138
    },
    {
      "epoch": 0.7699250576872027,
      "grad_norm": 0.5819490009014724,
      "learning_rate": 0.00014088193707081914,
      "loss": 12.2092,
      "step": 14139
    },
    {
      "epoch": 0.7699795116837856,
      "grad_norm": 0.585003880961486,
      "learning_rate": 0.0001408738892343816,
      "loss": 12.1618,
      "step": 14140
    },
    {
      "epoch": 0.7700339656803686,
      "grad_norm": 0.6022137293508225,
      "learning_rate": 0.0001408658410801018,
      "loss": 12.2765,
      "step": 14141
    },
    {
      "epoch": 0.7700884196769516,
      "grad_norm": 0.5975999931093471,
      "learning_rate": 0.0001408577926080424,
      "loss": 12.208,
      "step": 14142
    },
    {
      "epoch": 0.7701428736735346,
      "grad_norm": 0.5542571069872959,
      "learning_rate": 0.00014084974381826592,
      "loss": 12.2463,
      "step": 14143
    },
    {
      "epoch": 0.7701973276701177,
      "grad_norm": 0.5281431689300452,
      "learning_rate": 0.000140841694710835,
      "loss": 12.2376,
      "step": 14144
    },
    {
      "epoch": 0.7702517816667007,
      "grad_norm": 0.5943061625491666,
      "learning_rate": 0.0001408336452858122,
      "loss": 12.2756,
      "step": 14145
    },
    {
      "epoch": 0.7703062356632837,
      "grad_norm": 0.5421811990530448,
      "learning_rate": 0.00014082559554326015,
      "loss": 12.0254,
      "step": 14146
    },
    {
      "epoch": 0.7703606896598667,
      "grad_norm": 0.6361321674266639,
      "learning_rate": 0.0001408175454832414,
      "loss": 12.2234,
      "step": 14147
    },
    {
      "epoch": 0.7704151436564497,
      "grad_norm": 0.5642016530538166,
      "learning_rate": 0.00014080949510581858,
      "loss": 12.2879,
      "step": 14148
    },
    {
      "epoch": 0.7704695976530328,
      "grad_norm": 0.5787333057483416,
      "learning_rate": 0.00014080144441105429,
      "loss": 12.2408,
      "step": 14149
    },
    {
      "epoch": 0.7705240516496158,
      "grad_norm": 0.5170654352262323,
      "learning_rate": 0.00014079339339901113,
      "loss": 12.2432,
      "step": 14150
    },
    {
      "epoch": 0.7705785056461988,
      "grad_norm": 0.6319362718158542,
      "learning_rate": 0.00014078534206975166,
      "loss": 12.328,
      "step": 14151
    },
    {
      "epoch": 0.7706329596427818,
      "grad_norm": 0.628696957200546,
      "learning_rate": 0.0001407772904233386,
      "loss": 12.313,
      "step": 14152
    },
    {
      "epoch": 0.7706874136393648,
      "grad_norm": 0.5423506725771897,
      "learning_rate": 0.00014076923845983443,
      "loss": 12.2563,
      "step": 14153
    },
    {
      "epoch": 0.7707418676359478,
      "grad_norm": 0.5830858369473833,
      "learning_rate": 0.00014076118617930186,
      "loss": 12.2093,
      "step": 14154
    },
    {
      "epoch": 0.7707963216325309,
      "grad_norm": 0.6498214355393194,
      "learning_rate": 0.00014075313358180344,
      "loss": 12.2451,
      "step": 14155
    },
    {
      "epoch": 0.7708507756291139,
      "grad_norm": 0.6162319238332199,
      "learning_rate": 0.00014074508066740183,
      "loss": 12.1171,
      "step": 14156
    },
    {
      "epoch": 0.7709052296256969,
      "grad_norm": 0.6190695713664903,
      "learning_rate": 0.00014073702743615964,
      "loss": 12.2088,
      "step": 14157
    },
    {
      "epoch": 0.7709596836222798,
      "grad_norm": 0.5412261050077144,
      "learning_rate": 0.0001407289738881395,
      "loss": 12.0418,
      "step": 14158
    },
    {
      "epoch": 0.7710141376188628,
      "grad_norm": 0.7319502020643702,
      "learning_rate": 0.00014072092002340405,
      "loss": 12.3812,
      "step": 14159
    },
    {
      "epoch": 0.7710685916154458,
      "grad_norm": 0.6467337001986067,
      "learning_rate": 0.00014071286584201587,
      "loss": 12.3046,
      "step": 14160
    },
    {
      "epoch": 0.7711230456120289,
      "grad_norm": 0.63723613660386,
      "learning_rate": 0.00014070481134403762,
      "loss": 12.3337,
      "step": 14161
    },
    {
      "epoch": 0.7711774996086119,
      "grad_norm": 0.541520470381976,
      "learning_rate": 0.00014069675652953194,
      "loss": 12.1234,
      "step": 14162
    },
    {
      "epoch": 0.7712319536051949,
      "grad_norm": 0.6029623074529031,
      "learning_rate": 0.00014068870139856142,
      "loss": 12.3006,
      "step": 14163
    },
    {
      "epoch": 0.7712864076017779,
      "grad_norm": 0.5985857701434745,
      "learning_rate": 0.00014068064595118877,
      "loss": 12.2511,
      "step": 14164
    },
    {
      "epoch": 0.7713408615983609,
      "grad_norm": 0.550235811414424,
      "learning_rate": 0.0001406725901874766,
      "loss": 12.1389,
      "step": 14165
    },
    {
      "epoch": 0.7713953155949439,
      "grad_norm": 0.5984638770455566,
      "learning_rate": 0.00014066453410748752,
      "loss": 12.3201,
      "step": 14166
    },
    {
      "epoch": 0.771449769591527,
      "grad_norm": 0.5862567048207609,
      "learning_rate": 0.00014065647771128423,
      "loss": 12.3524,
      "step": 14167
    },
    {
      "epoch": 0.77150422358811,
      "grad_norm": 0.562059377621077,
      "learning_rate": 0.00014064842099892935,
      "loss": 12.2515,
      "step": 14168
    },
    {
      "epoch": 0.771558677584693,
      "grad_norm": 0.5315770319244236,
      "learning_rate": 0.00014064036397048551,
      "loss": 12.0774,
      "step": 14169
    },
    {
      "epoch": 0.771613131581276,
      "grad_norm": 0.6436814701280071,
      "learning_rate": 0.0001406323066260154,
      "loss": 12.2492,
      "step": 14170
    },
    {
      "epoch": 0.771667585577859,
      "grad_norm": 0.5141619178170451,
      "learning_rate": 0.00014062424896558166,
      "loss": 12.216,
      "step": 14171
    },
    {
      "epoch": 0.771722039574442,
      "grad_norm": 0.601453934681363,
      "learning_rate": 0.00014061619098924694,
      "loss": 12.0353,
      "step": 14172
    },
    {
      "epoch": 0.7717764935710251,
      "grad_norm": 0.7781834634976749,
      "learning_rate": 0.00014060813269707392,
      "loss": 12.272,
      "step": 14173
    },
    {
      "epoch": 0.7718309475676081,
      "grad_norm": 0.6037550044731685,
      "learning_rate": 0.00014060007408912525,
      "loss": 12.2854,
      "step": 14174
    },
    {
      "epoch": 0.771885401564191,
      "grad_norm": 0.6248294027085685,
      "learning_rate": 0.0001405920151654636,
      "loss": 12.2883,
      "step": 14175
    },
    {
      "epoch": 0.771939855560774,
      "grad_norm": 0.5940242997523166,
      "learning_rate": 0.00014058395592615158,
      "loss": 12.2134,
      "step": 14176
    },
    {
      "epoch": 0.771994309557357,
      "grad_norm": 0.6932183762839871,
      "learning_rate": 0.00014057589637125198,
      "loss": 12.1776,
      "step": 14177
    },
    {
      "epoch": 0.77204876355394,
      "grad_norm": 0.5124592144785853,
      "learning_rate": 0.00014056783650082737,
      "loss": 12.0275,
      "step": 14178
    },
    {
      "epoch": 0.7721032175505231,
      "grad_norm": 0.5714334986091909,
      "learning_rate": 0.00014055977631494045,
      "loss": 12.2715,
      "step": 14179
    },
    {
      "epoch": 0.7721576715471061,
      "grad_norm": 0.5712169349242016,
      "learning_rate": 0.00014055171581365397,
      "loss": 12.2596,
      "step": 14180
    },
    {
      "epoch": 0.7722121255436891,
      "grad_norm": 0.6319480061558428,
      "learning_rate": 0.00014054365499703046,
      "loss": 12.3137,
      "step": 14181
    },
    {
      "epoch": 0.7722665795402721,
      "grad_norm": 0.5785141653358192,
      "learning_rate": 0.00014053559386513275,
      "loss": 12.2219,
      "step": 14182
    },
    {
      "epoch": 0.7723210335368551,
      "grad_norm": 0.5591661426535585,
      "learning_rate": 0.00014052753241802342,
      "loss": 12.2081,
      "step": 14183
    },
    {
      "epoch": 0.7723754875334382,
      "grad_norm": 0.5548912954895845,
      "learning_rate": 0.00014051947065576526,
      "loss": 12.2108,
      "step": 14184
    },
    {
      "epoch": 0.7724299415300212,
      "grad_norm": 0.569668227503653,
      "learning_rate": 0.00014051140857842086,
      "loss": 12.2756,
      "step": 14185
    },
    {
      "epoch": 0.7724843955266042,
      "grad_norm": 0.525580456092135,
      "learning_rate": 0.000140503346186053,
      "loss": 12.3377,
      "step": 14186
    },
    {
      "epoch": 0.7725388495231872,
      "grad_norm": 0.5478331742854908,
      "learning_rate": 0.00014049528347872425,
      "loss": 12.2428,
      "step": 14187
    },
    {
      "epoch": 0.7725933035197702,
      "grad_norm": 0.572341978220355,
      "learning_rate": 0.00014048722045649742,
      "loss": 12.2341,
      "step": 14188
    },
    {
      "epoch": 0.7726477575163532,
      "grad_norm": 0.6014348989134071,
      "learning_rate": 0.0001404791571194352,
      "loss": 12.2087,
      "step": 14189
    },
    {
      "epoch": 0.7727022115129363,
      "grad_norm": 0.5457895419471461,
      "learning_rate": 0.00014047109346760023,
      "loss": 12.2166,
      "step": 14190
    },
    {
      "epoch": 0.7727566655095193,
      "grad_norm": 0.6426511252714703,
      "learning_rate": 0.00014046302950105529,
      "loss": 12.2607,
      "step": 14191
    },
    {
      "epoch": 0.7728111195061023,
      "grad_norm": 0.5161979350729351,
      "learning_rate": 0.000140454965219863,
      "loss": 12.1732,
      "step": 14192
    },
    {
      "epoch": 0.7728655735026853,
      "grad_norm": 0.554545149023617,
      "learning_rate": 0.00014044690062408612,
      "loss": 12.2118,
      "step": 14193
    },
    {
      "epoch": 0.7729200274992682,
      "grad_norm": 0.5719533718854034,
      "learning_rate": 0.00014043883571378737,
      "loss": 12.22,
      "step": 14194
    },
    {
      "epoch": 0.7729744814958512,
      "grad_norm": 0.6718674667684825,
      "learning_rate": 0.0001404307704890295,
      "loss": 12.0822,
      "step": 14195
    },
    {
      "epoch": 0.7730289354924343,
      "grad_norm": 0.5855615166821195,
      "learning_rate": 0.00014042270494987513,
      "loss": 12.3239,
      "step": 14196
    },
    {
      "epoch": 0.7730833894890173,
      "grad_norm": 0.6017993497722051,
      "learning_rate": 0.000140414639096387,
      "loss": 12.3342,
      "step": 14197
    },
    {
      "epoch": 0.7731378434856003,
      "grad_norm": 0.5411438881222047,
      "learning_rate": 0.00014040657292862792,
      "loss": 12.3142,
      "step": 14198
    },
    {
      "epoch": 0.7731922974821833,
      "grad_norm": 0.5559677870703806,
      "learning_rate": 0.0001403985064466605,
      "loss": 12.2001,
      "step": 14199
    },
    {
      "epoch": 0.7732467514787663,
      "grad_norm": 0.5760873129276473,
      "learning_rate": 0.00014039043965054758,
      "loss": 12.2604,
      "step": 14200
    },
    {
      "epoch": 0.7733012054753493,
      "grad_norm": 0.5805544309998965,
      "learning_rate": 0.00014038237254035177,
      "loss": 12.3626,
      "step": 14201
    },
    {
      "epoch": 0.7733556594719324,
      "grad_norm": 0.5615832662673361,
      "learning_rate": 0.00014037430511613588,
      "loss": 12.0235,
      "step": 14202
    },
    {
      "epoch": 0.7734101134685154,
      "grad_norm": 0.6065320858204621,
      "learning_rate": 0.00014036623737796261,
      "loss": 12.3079,
      "step": 14203
    },
    {
      "epoch": 0.7734645674650984,
      "grad_norm": 0.5479637189418413,
      "learning_rate": 0.0001403581693258947,
      "loss": 12.2065,
      "step": 14204
    },
    {
      "epoch": 0.7735190214616814,
      "grad_norm": 0.5847645596608256,
      "learning_rate": 0.00014035010095999497,
      "loss": 12.1909,
      "step": 14205
    },
    {
      "epoch": 0.7735734754582644,
      "grad_norm": 0.5972784753038021,
      "learning_rate": 0.00014034203228032604,
      "loss": 12.2817,
      "step": 14206
    },
    {
      "epoch": 0.7736279294548474,
      "grad_norm": 0.6331906216266722,
      "learning_rate": 0.0001403339632869507,
      "loss": 12.2867,
      "step": 14207
    },
    {
      "epoch": 0.7736823834514305,
      "grad_norm": 0.5132859404573481,
      "learning_rate": 0.00014032589397993168,
      "loss": 12.1685,
      "step": 14208
    },
    {
      "epoch": 0.7737368374480135,
      "grad_norm": 0.5589622425109604,
      "learning_rate": 0.00014031782435933174,
      "loss": 12.1366,
      "step": 14209
    },
    {
      "epoch": 0.7737912914445965,
      "grad_norm": 0.5858976298498404,
      "learning_rate": 0.00014030975442521365,
      "loss": 12.1649,
      "step": 14210
    },
    {
      "epoch": 0.7738457454411795,
      "grad_norm": 0.6674804842623974,
      "learning_rate": 0.0001403016841776402,
      "loss": 12.4075,
      "step": 14211
    },
    {
      "epoch": 0.7739001994377624,
      "grad_norm": 0.5939039505378468,
      "learning_rate": 0.00014029361361667407,
      "loss": 12.1996,
      "step": 14212
    },
    {
      "epoch": 0.7739546534343456,
      "grad_norm": 0.5616736497767092,
      "learning_rate": 0.000140285542742378,
      "loss": 12.3147,
      "step": 14213
    },
    {
      "epoch": 0.7740091074309285,
      "grad_norm": 0.6052812911670153,
      "learning_rate": 0.00014027747155481482,
      "loss": 12.3252,
      "step": 14214
    },
    {
      "epoch": 0.7740635614275115,
      "grad_norm": 0.612917274492108,
      "learning_rate": 0.00014026940005404726,
      "loss": 12.2219,
      "step": 14215
    },
    {
      "epoch": 0.7741180154240945,
      "grad_norm": 0.546998223405516,
      "learning_rate": 0.00014026132824013812,
      "loss": 12.2989,
      "step": 14216
    },
    {
      "epoch": 0.7741724694206775,
      "grad_norm": 0.5894059670789069,
      "learning_rate": 0.00014025325611315012,
      "loss": 12.2427,
      "step": 14217
    },
    {
      "epoch": 0.7742269234172605,
      "grad_norm": 0.5957066757457238,
      "learning_rate": 0.00014024518367314602,
      "loss": 12.3165,
      "step": 14218
    },
    {
      "epoch": 0.7742813774138436,
      "grad_norm": 0.5836536982041985,
      "learning_rate": 0.00014023711092018868,
      "loss": 12.2283,
      "step": 14219
    },
    {
      "epoch": 0.7743358314104266,
      "grad_norm": 0.5368407291439166,
      "learning_rate": 0.00014022903785434077,
      "loss": 12.351,
      "step": 14220
    },
    {
      "epoch": 0.7743902854070096,
      "grad_norm": 0.5384549553801493,
      "learning_rate": 0.00014022096447566514,
      "loss": 12.2068,
      "step": 14221
    },
    {
      "epoch": 0.7744447394035926,
      "grad_norm": 0.5645116287805855,
      "learning_rate": 0.00014021289078422456,
      "loss": 12.1967,
      "step": 14222
    },
    {
      "epoch": 0.7744991934001756,
      "grad_norm": 0.5907865445463186,
      "learning_rate": 0.00014020481678008175,
      "loss": 12.252,
      "step": 14223
    },
    {
      "epoch": 0.7745536473967586,
      "grad_norm": 0.6491946219544488,
      "learning_rate": 0.0001401967424632996,
      "loss": 12.2809,
      "step": 14224
    },
    {
      "epoch": 0.7746081013933417,
      "grad_norm": 0.562889073485214,
      "learning_rate": 0.0001401886678339408,
      "loss": 12.1292,
      "step": 14225
    },
    {
      "epoch": 0.7746625553899247,
      "grad_norm": 0.6158908521325102,
      "learning_rate": 0.00014018059289206818,
      "loss": 12.2408,
      "step": 14226
    },
    {
      "epoch": 0.7747170093865077,
      "grad_norm": 0.6743988176690096,
      "learning_rate": 0.00014017251763774456,
      "loss": 12.2302,
      "step": 14227
    },
    {
      "epoch": 0.7747714633830907,
      "grad_norm": 0.5750539418203042,
      "learning_rate": 0.0001401644420710327,
      "loss": 12.3108,
      "step": 14228
    },
    {
      "epoch": 0.7748259173796737,
      "grad_norm": 0.6184375139837525,
      "learning_rate": 0.00014015636619199535,
      "loss": 12.3028,
      "step": 14229
    },
    {
      "epoch": 0.7748803713762566,
      "grad_norm": 0.610642355628531,
      "learning_rate": 0.0001401482900006954,
      "loss": 12.1486,
      "step": 14230
    },
    {
      "epoch": 0.7749348253728398,
      "grad_norm": 0.5821790796481626,
      "learning_rate": 0.0001401402134971956,
      "loss": 12.2153,
      "step": 14231
    },
    {
      "epoch": 0.7749892793694227,
      "grad_norm": 0.5664819681648965,
      "learning_rate": 0.0001401321366815588,
      "loss": 12.2911,
      "step": 14232
    },
    {
      "epoch": 0.7750437333660057,
      "grad_norm": 0.5699499311143544,
      "learning_rate": 0.00014012405955384776,
      "loss": 12.1542,
      "step": 14233
    },
    {
      "epoch": 0.7750981873625887,
      "grad_norm": 0.6004532193029597,
      "learning_rate": 0.0001401159821141253,
      "loss": 12.2844,
      "step": 14234
    },
    {
      "epoch": 0.7751526413591717,
      "grad_norm": 0.5606771204873505,
      "learning_rate": 0.00014010790436245425,
      "loss": 12.2204,
      "step": 14235
    },
    {
      "epoch": 0.7752070953557547,
      "grad_norm": 0.5976923703128083,
      "learning_rate": 0.00014009982629889736,
      "loss": 12.2957,
      "step": 14236
    },
    {
      "epoch": 0.7752615493523378,
      "grad_norm": 0.6291384584779743,
      "learning_rate": 0.00014009174792351756,
      "loss": 12.1954,
      "step": 14237
    },
    {
      "epoch": 0.7753160033489208,
      "grad_norm": 0.5841736611008006,
      "learning_rate": 0.00014008366923637757,
      "loss": 12.2286,
      "step": 14238
    },
    {
      "epoch": 0.7753704573455038,
      "grad_norm": 0.6188229355529823,
      "learning_rate": 0.00014007559023754027,
      "loss": 12.2629,
      "step": 14239
    },
    {
      "epoch": 0.7754249113420868,
      "grad_norm": 0.6582155521706322,
      "learning_rate": 0.00014006751092706843,
      "loss": 12.2669,
      "step": 14240
    },
    {
      "epoch": 0.7754793653386698,
      "grad_norm": 0.643902293261855,
      "learning_rate": 0.00014005943130502492,
      "loss": 12.3276,
      "step": 14241
    },
    {
      "epoch": 0.7755338193352528,
      "grad_norm": 0.6682038792668181,
      "learning_rate": 0.00014005135137147256,
      "loss": 12.2784,
      "step": 14242
    },
    {
      "epoch": 0.7755882733318359,
      "grad_norm": 0.5877777782070308,
      "learning_rate": 0.00014004327112647418,
      "loss": 12.3538,
      "step": 14243
    },
    {
      "epoch": 0.7756427273284189,
      "grad_norm": 0.5785124014962444,
      "learning_rate": 0.0001400351905700926,
      "loss": 12.1579,
      "step": 14244
    },
    {
      "epoch": 0.7756971813250019,
      "grad_norm": 0.6505713308803961,
      "learning_rate": 0.00014002710970239062,
      "loss": 12.1179,
      "step": 14245
    },
    {
      "epoch": 0.7757516353215849,
      "grad_norm": 0.6449881450005565,
      "learning_rate": 0.00014001902852343122,
      "loss": 12.3544,
      "step": 14246
    },
    {
      "epoch": 0.7758060893181679,
      "grad_norm": 0.5479162926644555,
      "learning_rate": 0.00014001094703327706,
      "loss": 12.1618,
      "step": 14247
    },
    {
      "epoch": 0.775860543314751,
      "grad_norm": 0.6190528133158492,
      "learning_rate": 0.00014000286523199108,
      "loss": 12.2099,
      "step": 14248
    },
    {
      "epoch": 0.775914997311334,
      "grad_norm": 0.7418593316007284,
      "learning_rate": 0.00013999478311963614,
      "loss": 12.3345,
      "step": 14249
    },
    {
      "epoch": 0.775969451307917,
      "grad_norm": 0.5577836545952066,
      "learning_rate": 0.00013998670069627505,
      "loss": 12.3324,
      "step": 14250
    },
    {
      "epoch": 0.7760239053044999,
      "grad_norm": 0.5608169550998805,
      "learning_rate": 0.00013997861796197068,
      "loss": 12.124,
      "step": 14251
    },
    {
      "epoch": 0.7760783593010829,
      "grad_norm": 0.632786855936109,
      "learning_rate": 0.00013997053491678584,
      "loss": 12.4297,
      "step": 14252
    },
    {
      "epoch": 0.7761328132976659,
      "grad_norm": 0.5433203133912633,
      "learning_rate": 0.00013996245156078343,
      "loss": 12.072,
      "step": 14253
    },
    {
      "epoch": 0.776187267294249,
      "grad_norm": 0.5843187705344397,
      "learning_rate": 0.0001399543678940263,
      "loss": 12.2276,
      "step": 14254
    },
    {
      "epoch": 0.776241721290832,
      "grad_norm": 0.5815658943635917,
      "learning_rate": 0.0001399462839165773,
      "loss": 12.2323,
      "step": 14255
    },
    {
      "epoch": 0.776296175287415,
      "grad_norm": 0.5538371414048853,
      "learning_rate": 0.00013993819962849932,
      "loss": 12.1063,
      "step": 14256
    },
    {
      "epoch": 0.776350629283998,
      "grad_norm": 0.5472741001888909,
      "learning_rate": 0.00013993011502985519,
      "loss": 12.1826,
      "step": 14257
    },
    {
      "epoch": 0.776405083280581,
      "grad_norm": 0.6017028137201669,
      "learning_rate": 0.00013992203012070775,
      "loss": 12.1747,
      "step": 14258
    },
    {
      "epoch": 0.776459537277164,
      "grad_norm": 0.5861421427275709,
      "learning_rate": 0.00013991394490111994,
      "loss": 12.1843,
      "step": 14259
    },
    {
      "epoch": 0.7765139912737471,
      "grad_norm": 0.6023377412435076,
      "learning_rate": 0.0001399058593711546,
      "loss": 12.1658,
      "step": 14260
    },
    {
      "epoch": 0.7765684452703301,
      "grad_norm": 0.5754821585939178,
      "learning_rate": 0.00013989777353087463,
      "loss": 12.1961,
      "step": 14261
    },
    {
      "epoch": 0.7766228992669131,
      "grad_norm": 0.6475523071910245,
      "learning_rate": 0.00013988968738034286,
      "loss": 12.2166,
      "step": 14262
    },
    {
      "epoch": 0.7766773532634961,
      "grad_norm": 0.5220931415276219,
      "learning_rate": 0.00013988160091962218,
      "loss": 12.0853,
      "step": 14263
    },
    {
      "epoch": 0.7767318072600791,
      "grad_norm": 0.570755149486794,
      "learning_rate": 0.00013987351414877547,
      "loss": 12.1619,
      "step": 14264
    },
    {
      "epoch": 0.7767862612566621,
      "grad_norm": 0.6100661674420887,
      "learning_rate": 0.00013986542706786564,
      "loss": 12.2246,
      "step": 14265
    },
    {
      "epoch": 0.7768407152532452,
      "grad_norm": 0.5887754471812554,
      "learning_rate": 0.00013985733967695563,
      "loss": 12.2059,
      "step": 14266
    },
    {
      "epoch": 0.7768951692498282,
      "grad_norm": 0.6519893463475808,
      "learning_rate": 0.0001398492519761082,
      "loss": 12.1453,
      "step": 14267
    },
    {
      "epoch": 0.7769496232464111,
      "grad_norm": 0.5964378271102649,
      "learning_rate": 0.0001398411639653863,
      "loss": 12.2564,
      "step": 14268
    },
    {
      "epoch": 0.7770040772429941,
      "grad_norm": 0.5732917751538955,
      "learning_rate": 0.0001398330756448528,
      "loss": 12.2191,
      "step": 14269
    },
    {
      "epoch": 0.7770585312395771,
      "grad_norm": 0.5754070704973367,
      "learning_rate": 0.0001398249870145707,
      "loss": 12.1242,
      "step": 14270
    },
    {
      "epoch": 0.7771129852361601,
      "grad_norm": 0.6793722475374825,
      "learning_rate": 0.0001398168980746028,
      "loss": 12.3338,
      "step": 14271
    },
    {
      "epoch": 0.7771674392327432,
      "grad_norm": 0.5648411282352749,
      "learning_rate": 0.00013980880882501199,
      "loss": 12.218,
      "step": 14272
    },
    {
      "epoch": 0.7772218932293262,
      "grad_norm": 0.6331849677299348,
      "learning_rate": 0.00013980071926586124,
      "loss": 12.4167,
      "step": 14273
    },
    {
      "epoch": 0.7772763472259092,
      "grad_norm": 0.578231429227074,
      "learning_rate": 0.0001397926293972134,
      "loss": 12.2841,
      "step": 14274
    },
    {
      "epoch": 0.7773308012224922,
      "grad_norm": 0.6491037203032933,
      "learning_rate": 0.0001397845392191314,
      "loss": 12.3289,
      "step": 14275
    },
    {
      "epoch": 0.7773852552190752,
      "grad_norm": 0.5303799869679332,
      "learning_rate": 0.00013977644873167816,
      "loss": 12.1669,
      "step": 14276
    },
    {
      "epoch": 0.7774397092156582,
      "grad_norm": 0.5901335407418612,
      "learning_rate": 0.0001397683579349166,
      "loss": 12.1261,
      "step": 14277
    },
    {
      "epoch": 0.7774941632122413,
      "grad_norm": 0.5676359507328063,
      "learning_rate": 0.00013976026682890958,
      "loss": 12.1885,
      "step": 14278
    },
    {
      "epoch": 0.7775486172088243,
      "grad_norm": 0.534057470171481,
      "learning_rate": 0.0001397521754137201,
      "loss": 12.1992,
      "step": 14279
    },
    {
      "epoch": 0.7776030712054073,
      "grad_norm": 0.5814240032332899,
      "learning_rate": 0.000139744083689411,
      "loss": 12.1737,
      "step": 14280
    },
    {
      "epoch": 0.7776575252019903,
      "grad_norm": 0.5591124285685799,
      "learning_rate": 0.00013973599165604527,
      "loss": 12.2514,
      "step": 14281
    },
    {
      "epoch": 0.7777119791985733,
      "grad_norm": 0.5714511589156374,
      "learning_rate": 0.0001397278993136858,
      "loss": 12.2969,
      "step": 14282
    },
    {
      "epoch": 0.7777664331951564,
      "grad_norm": 0.5185389079557104,
      "learning_rate": 0.0001397198066623955,
      "loss": 12.2458,
      "step": 14283
    },
    {
      "epoch": 0.7778208871917394,
      "grad_norm": 0.5416837860102734,
      "learning_rate": 0.00013971171370223736,
      "loss": 12.292,
      "step": 14284
    },
    {
      "epoch": 0.7778753411883224,
      "grad_norm": 0.6085444604486575,
      "learning_rate": 0.00013970362043327423,
      "loss": 12.2255,
      "step": 14285
    },
    {
      "epoch": 0.7779297951849053,
      "grad_norm": 0.6118832655342363,
      "learning_rate": 0.00013969552685556914,
      "loss": 12.249,
      "step": 14286
    },
    {
      "epoch": 0.7779842491814883,
      "grad_norm": 0.583875910795135,
      "learning_rate": 0.0001396874329691849,
      "loss": 12.2431,
      "step": 14287
    },
    {
      "epoch": 0.7780387031780713,
      "grad_norm": 0.6199842096758743,
      "learning_rate": 0.0001396793387741846,
      "loss": 12.237,
      "step": 14288
    },
    {
      "epoch": 0.7780931571746544,
      "grad_norm": 0.5951711257796768,
      "learning_rate": 0.00013967124427063108,
      "loss": 12.0649,
      "step": 14289
    },
    {
      "epoch": 0.7781476111712374,
      "grad_norm": 0.6533099070438597,
      "learning_rate": 0.0001396631494585873,
      "loss": 12.3251,
      "step": 14290
    },
    {
      "epoch": 0.7782020651678204,
      "grad_norm": 0.6306942750243015,
      "learning_rate": 0.00013965505433811623,
      "loss": 12.2949,
      "step": 14291
    },
    {
      "epoch": 0.7782565191644034,
      "grad_norm": 0.5373222830184418,
      "learning_rate": 0.0001396469589092808,
      "loss": 12.1354,
      "step": 14292
    },
    {
      "epoch": 0.7783109731609864,
      "grad_norm": 0.5551774573109617,
      "learning_rate": 0.00013963886317214398,
      "loss": 12.2581,
      "step": 14293
    },
    {
      "epoch": 0.7783654271575694,
      "grad_norm": 0.5963154606421266,
      "learning_rate": 0.00013963076712676873,
      "loss": 12.3286,
      "step": 14294
    },
    {
      "epoch": 0.7784198811541525,
      "grad_norm": 0.5861963744730597,
      "learning_rate": 0.00013962267077321795,
      "loss": 12.2763,
      "step": 14295
    },
    {
      "epoch": 0.7784743351507355,
      "grad_norm": 0.6176149298969914,
      "learning_rate": 0.00013961457411155466,
      "loss": 12.3619,
      "step": 14296
    },
    {
      "epoch": 0.7785287891473185,
      "grad_norm": 0.5448354456452081,
      "learning_rate": 0.00013960647714184182,
      "loss": 12.2151,
      "step": 14297
    },
    {
      "epoch": 0.7785832431439015,
      "grad_norm": 0.520422704062557,
      "learning_rate": 0.0001395983798641423,
      "loss": 12.2499,
      "step": 14298
    },
    {
      "epoch": 0.7786376971404845,
      "grad_norm": 0.557202372867009,
      "learning_rate": 0.0001395902822785192,
      "loss": 12.165,
      "step": 14299
    },
    {
      "epoch": 0.7786921511370675,
      "grad_norm": 0.5961029491895187,
      "learning_rate": 0.00013958218438503542,
      "loss": 12.3071,
      "step": 14300
    },
    {
      "epoch": 0.7787466051336506,
      "grad_norm": 0.5826837314749691,
      "learning_rate": 0.00013957408618375393,
      "loss": 12.2213,
      "step": 14301
    },
    {
      "epoch": 0.7788010591302336,
      "grad_norm": 0.5420504530064048,
      "learning_rate": 0.0001395659876747377,
      "loss": 12.2102,
      "step": 14302
    },
    {
      "epoch": 0.7788555131268166,
      "grad_norm": 0.5346947413027613,
      "learning_rate": 0.00013955788885804972,
      "loss": 12.1678,
      "step": 14303
    },
    {
      "epoch": 0.7789099671233995,
      "grad_norm": 0.5882148826393877,
      "learning_rate": 0.00013954978973375294,
      "loss": 12.1439,
      "step": 14304
    },
    {
      "epoch": 0.7789644211199825,
      "grad_norm": 0.5857740211823618,
      "learning_rate": 0.0001395416903019104,
      "loss": 12.2248,
      "step": 14305
    },
    {
      "epoch": 0.7790188751165655,
      "grad_norm": 0.5330991902072838,
      "learning_rate": 0.00013953359056258503,
      "loss": 12.1562,
      "step": 14306
    },
    {
      "epoch": 0.7790733291131486,
      "grad_norm": 0.6525952363657743,
      "learning_rate": 0.00013952549051583982,
      "loss": 12.281,
      "step": 14307
    },
    {
      "epoch": 0.7791277831097316,
      "grad_norm": 0.6381722528438454,
      "learning_rate": 0.0001395173901617378,
      "loss": 12.2519,
      "step": 14308
    },
    {
      "epoch": 0.7791822371063146,
      "grad_norm": 0.5788636699692825,
      "learning_rate": 0.00013950928950034187,
      "loss": 12.1706,
      "step": 14309
    },
    {
      "epoch": 0.7792366911028976,
      "grad_norm": 0.5714525079864948,
      "learning_rate": 0.00013950118853171513,
      "loss": 12.1567,
      "step": 14310
    },
    {
      "epoch": 0.7792911450994806,
      "grad_norm": 0.5847814821575158,
      "learning_rate": 0.0001394930872559205,
      "loss": 12.274,
      "step": 14311
    },
    {
      "epoch": 0.7793455990960636,
      "grad_norm": 0.6262641473386918,
      "learning_rate": 0.000139484985673021,
      "loss": 12.3492,
      "step": 14312
    },
    {
      "epoch": 0.7794000530926467,
      "grad_norm": 0.6105688553660109,
      "learning_rate": 0.00013947688378307963,
      "loss": 12.1608,
      "step": 14313
    },
    {
      "epoch": 0.7794545070892297,
      "grad_norm": 0.5249630319727343,
      "learning_rate": 0.0001394687815861594,
      "loss": 12.0978,
      "step": 14314
    },
    {
      "epoch": 0.7795089610858127,
      "grad_norm": 0.6093801425256825,
      "learning_rate": 0.00013946067908232333,
      "loss": 12.1557,
      "step": 14315
    },
    {
      "epoch": 0.7795634150823957,
      "grad_norm": 0.6559736581795284,
      "learning_rate": 0.00013945257627163437,
      "loss": 12.2336,
      "step": 14316
    },
    {
      "epoch": 0.7796178690789787,
      "grad_norm": 0.5574043885221901,
      "learning_rate": 0.00013944447315415557,
      "loss": 12.1308,
      "step": 14317
    },
    {
      "epoch": 0.7796723230755618,
      "grad_norm": 0.5868701210809807,
      "learning_rate": 0.00013943636972994991,
      "loss": 12.2638,
      "step": 14318
    },
    {
      "epoch": 0.7797267770721448,
      "grad_norm": 0.6012516255490127,
      "learning_rate": 0.00013942826599908044,
      "loss": 12.2809,
      "step": 14319
    },
    {
      "epoch": 0.7797812310687278,
      "grad_norm": 0.5372097078881665,
      "learning_rate": 0.00013942016196161016,
      "loss": 12.2253,
      "step": 14320
    },
    {
      "epoch": 0.7798356850653108,
      "grad_norm": 0.545134903719249,
      "learning_rate": 0.00013941205761760212,
      "loss": 12.2653,
      "step": 14321
    },
    {
      "epoch": 0.7798901390618938,
      "grad_norm": 0.5730787720955979,
      "learning_rate": 0.0001394039529671193,
      "loss": 12.2101,
      "step": 14322
    },
    {
      "epoch": 0.7799445930584767,
      "grad_norm": 0.5340799709141614,
      "learning_rate": 0.0001393958480102247,
      "loss": 12.1503,
      "step": 14323
    },
    {
      "epoch": 0.7799990470550598,
      "grad_norm": 0.562407601107663,
      "learning_rate": 0.0001393877427469814,
      "loss": 12.2644,
      "step": 14324
    },
    {
      "epoch": 0.7800535010516428,
      "grad_norm": 0.6181723575845991,
      "learning_rate": 0.0001393796371774524,
      "loss": 12.124,
      "step": 14325
    },
    {
      "epoch": 0.7801079550482258,
      "grad_norm": 0.509798324698095,
      "learning_rate": 0.00013937153130170075,
      "loss": 12.2401,
      "step": 14326
    },
    {
      "epoch": 0.7801624090448088,
      "grad_norm": 0.587458779265688,
      "learning_rate": 0.00013936342511978946,
      "loss": 12.126,
      "step": 14327
    },
    {
      "epoch": 0.7802168630413918,
      "grad_norm": 0.6251305508340274,
      "learning_rate": 0.00013935531863178157,
      "loss": 12.1942,
      "step": 14328
    },
    {
      "epoch": 0.7802713170379748,
      "grad_norm": 0.5939469560259177,
      "learning_rate": 0.00013934721183774015,
      "loss": 12.2734,
      "step": 14329
    },
    {
      "epoch": 0.7803257710345579,
      "grad_norm": 0.584601867857084,
      "learning_rate": 0.00013933910473772816,
      "loss": 12.2859,
      "step": 14330
    },
    {
      "epoch": 0.7803802250311409,
      "grad_norm": 0.5854926109302606,
      "learning_rate": 0.00013933099733180876,
      "loss": 12.2367,
      "step": 14331
    },
    {
      "epoch": 0.7804346790277239,
      "grad_norm": 0.6588983389598322,
      "learning_rate": 0.00013932288962004486,
      "loss": 12.209,
      "step": 14332
    },
    {
      "epoch": 0.7804891330243069,
      "grad_norm": 0.6337923522917029,
      "learning_rate": 0.00013931478160249966,
      "loss": 12.237,
      "step": 14333
    },
    {
      "epoch": 0.7805435870208899,
      "grad_norm": 0.587236476231902,
      "learning_rate": 0.00013930667327923606,
      "loss": 12.2099,
      "step": 14334
    },
    {
      "epoch": 0.7805980410174729,
      "grad_norm": 0.5244300732635503,
      "learning_rate": 0.00013929856465031716,
      "loss": 12.1203,
      "step": 14335
    },
    {
      "epoch": 0.780652495014056,
      "grad_norm": 0.5159189318018038,
      "learning_rate": 0.00013929045571580608,
      "loss": 12.2196,
      "step": 14336
    },
    {
      "epoch": 0.780706949010639,
      "grad_norm": 0.6009509071692117,
      "learning_rate": 0.00013928234647576581,
      "loss": 12.2065,
      "step": 14337
    },
    {
      "epoch": 0.780761403007222,
      "grad_norm": 0.62828740155288,
      "learning_rate": 0.00013927423693025942,
      "loss": 12.2639,
      "step": 14338
    },
    {
      "epoch": 0.780815857003805,
      "grad_norm": 0.5782629456373587,
      "learning_rate": 0.00013926612707935,
      "loss": 12.3063,
      "step": 14339
    },
    {
      "epoch": 0.780870311000388,
      "grad_norm": 0.6183466777482532,
      "learning_rate": 0.00013925801692310058,
      "loss": 12.1923,
      "step": 14340
    },
    {
      "epoch": 0.7809247649969709,
      "grad_norm": 0.5903083877025544,
      "learning_rate": 0.00013924990646157424,
      "loss": 12.2672,
      "step": 14341
    },
    {
      "epoch": 0.780979218993554,
      "grad_norm": 0.6053161053985386,
      "learning_rate": 0.00013924179569483401,
      "loss": 12.2701,
      "step": 14342
    },
    {
      "epoch": 0.781033672990137,
      "grad_norm": 0.5986010917087732,
      "learning_rate": 0.00013923368462294303,
      "loss": 12.2788,
      "step": 14343
    },
    {
      "epoch": 0.78108812698672,
      "grad_norm": 0.6302334667450733,
      "learning_rate": 0.00013922557324596435,
      "loss": 12.3288,
      "step": 14344
    },
    {
      "epoch": 0.781142580983303,
      "grad_norm": 0.564817574612964,
      "learning_rate": 0.000139217461563961,
      "loss": 12.2768,
      "step": 14345
    },
    {
      "epoch": 0.781197034979886,
      "grad_norm": 0.6042381395814973,
      "learning_rate": 0.00013920934957699612,
      "loss": 12.0756,
      "step": 14346
    },
    {
      "epoch": 0.7812514889764691,
      "grad_norm": 0.7369856283803117,
      "learning_rate": 0.00013920123728513274,
      "loss": 12.2836,
      "step": 14347
    },
    {
      "epoch": 0.7813059429730521,
      "grad_norm": 0.6592566362375767,
      "learning_rate": 0.00013919312468843397,
      "loss": 12.2566,
      "step": 14348
    },
    {
      "epoch": 0.7813603969696351,
      "grad_norm": 0.6098850365522455,
      "learning_rate": 0.0001391850117869629,
      "loss": 12.2025,
      "step": 14349
    },
    {
      "epoch": 0.7814148509662181,
      "grad_norm": 0.6014968474919596,
      "learning_rate": 0.0001391768985807826,
      "loss": 12.2522,
      "step": 14350
    },
    {
      "epoch": 0.7814693049628011,
      "grad_norm": 0.5381136132046134,
      "learning_rate": 0.00013916878506995618,
      "loss": 12.2215,
      "step": 14351
    },
    {
      "epoch": 0.7815237589593841,
      "grad_norm": 0.5348296555678123,
      "learning_rate": 0.0001391606712545467,
      "loss": 12.2183,
      "step": 14352
    },
    {
      "epoch": 0.7815782129559672,
      "grad_norm": 0.6507633060534262,
      "learning_rate": 0.00013915255713461727,
      "loss": 12.2612,
      "step": 14353
    },
    {
      "epoch": 0.7816326669525502,
      "grad_norm": 0.5396809649087732,
      "learning_rate": 0.000139144442710231,
      "loss": 12.1339,
      "step": 14354
    },
    {
      "epoch": 0.7816871209491332,
      "grad_norm": 0.5438368503662775,
      "learning_rate": 0.000139136327981451,
      "loss": 12.3099,
      "step": 14355
    },
    {
      "epoch": 0.7817415749457162,
      "grad_norm": 0.659822970082171,
      "learning_rate": 0.00013912821294834033,
      "loss": 12.2638,
      "step": 14356
    },
    {
      "epoch": 0.7817960289422992,
      "grad_norm": 0.4852121890389048,
      "learning_rate": 0.00013912009761096213,
      "loss": 12.1385,
      "step": 14357
    },
    {
      "epoch": 0.7818504829388822,
      "grad_norm": 0.515123968159365,
      "learning_rate": 0.00013911198196937946,
      "loss": 12.2034,
      "step": 14358
    },
    {
      "epoch": 0.7819049369354653,
      "grad_norm": 0.5601852691409116,
      "learning_rate": 0.00013910386602365547,
      "loss": 12.2976,
      "step": 14359
    },
    {
      "epoch": 0.7819593909320482,
      "grad_norm": 0.5218995324590391,
      "learning_rate": 0.00013909574977385327,
      "loss": 12.2155,
      "step": 14360
    },
    {
      "epoch": 0.7820138449286312,
      "grad_norm": 0.7009425777978717,
      "learning_rate": 0.00013908763322003595,
      "loss": 12.1889,
      "step": 14361
    },
    {
      "epoch": 0.7820682989252142,
      "grad_norm": 0.5250741002436927,
      "learning_rate": 0.00013907951636226665,
      "loss": 12.2074,
      "step": 14362
    },
    {
      "epoch": 0.7821227529217972,
      "grad_norm": 0.5484504796190243,
      "learning_rate": 0.00013907139920060847,
      "loss": 12.2898,
      "step": 14363
    },
    {
      "epoch": 0.7821772069183802,
      "grad_norm": 0.639001255662936,
      "learning_rate": 0.00013906328173512455,
      "loss": 12.215,
      "step": 14364
    },
    {
      "epoch": 0.7822316609149633,
      "grad_norm": 0.5844299666446592,
      "learning_rate": 0.000139055163965878,
      "loss": 12.1908,
      "step": 14365
    },
    {
      "epoch": 0.7822861149115463,
      "grad_norm": 0.6061975542541682,
      "learning_rate": 0.00013904704589293192,
      "loss": 12.2301,
      "step": 14366
    },
    {
      "epoch": 0.7823405689081293,
      "grad_norm": 0.6185399821651211,
      "learning_rate": 0.00013903892751634947,
      "loss": 12.2471,
      "step": 14367
    },
    {
      "epoch": 0.7823950229047123,
      "grad_norm": 0.6760511265559507,
      "learning_rate": 0.0001390308088361938,
      "loss": 12.233,
      "step": 14368
    },
    {
      "epoch": 0.7824494769012953,
      "grad_norm": 0.5388520380861961,
      "learning_rate": 0.000139022689852528,
      "loss": 12.2188,
      "step": 14369
    },
    {
      "epoch": 0.7825039308978783,
      "grad_norm": 0.5564895987143336,
      "learning_rate": 0.0001390145705654152,
      "loss": 12.106,
      "step": 14370
    },
    {
      "epoch": 0.7825583848944614,
      "grad_norm": 0.5215538856075096,
      "learning_rate": 0.00013900645097491857,
      "loss": 12.1252,
      "step": 14371
    },
    {
      "epoch": 0.7826128388910444,
      "grad_norm": 0.6067388147613065,
      "learning_rate": 0.00013899833108110127,
      "loss": 12.1462,
      "step": 14372
    },
    {
      "epoch": 0.7826672928876274,
      "grad_norm": 0.5499048916495173,
      "learning_rate": 0.00013899021088402638,
      "loss": 12.1765,
      "step": 14373
    },
    {
      "epoch": 0.7827217468842104,
      "grad_norm": 0.6123862168122999,
      "learning_rate": 0.00013898209038375704,
      "loss": 12.1465,
      "step": 14374
    },
    {
      "epoch": 0.7827762008807934,
      "grad_norm": 0.5353567283212902,
      "learning_rate": 0.00013897396958035646,
      "loss": 12.3048,
      "step": 14375
    },
    {
      "epoch": 0.7828306548773764,
      "grad_norm": 0.7451386834832064,
      "learning_rate": 0.00013896584847388776,
      "loss": 12.3199,
      "step": 14376
    },
    {
      "epoch": 0.7828851088739595,
      "grad_norm": 0.5468178378836177,
      "learning_rate": 0.0001389577270644141,
      "loss": 12.2477,
      "step": 14377
    },
    {
      "epoch": 0.7829395628705424,
      "grad_norm": 0.561119411847658,
      "learning_rate": 0.00013894960535199862,
      "loss": 12.0196,
      "step": 14378
    },
    {
      "epoch": 0.7829940168671254,
      "grad_norm": 0.5808741207692449,
      "learning_rate": 0.00013894148333670444,
      "loss": 12.1076,
      "step": 14379
    },
    {
      "epoch": 0.7830484708637084,
      "grad_norm": 0.6466017916541937,
      "learning_rate": 0.00013893336101859479,
      "loss": 12.3068,
      "step": 14380
    },
    {
      "epoch": 0.7831029248602914,
      "grad_norm": 0.6033825408993769,
      "learning_rate": 0.00013892523839773274,
      "loss": 12.1595,
      "step": 14381
    },
    {
      "epoch": 0.7831573788568745,
      "grad_norm": 0.6134304033155082,
      "learning_rate": 0.00013891711547418158,
      "loss": 12.3442,
      "step": 14382
    },
    {
      "epoch": 0.7832118328534575,
      "grad_norm": 0.516047645087405,
      "learning_rate": 0.0001389089922480044,
      "loss": 12.241,
      "step": 14383
    },
    {
      "epoch": 0.7832662868500405,
      "grad_norm": 0.5306019795143434,
      "learning_rate": 0.0001389008687192643,
      "loss": 12.2247,
      "step": 14384
    },
    {
      "epoch": 0.7833207408466235,
      "grad_norm": 0.6731617967193122,
      "learning_rate": 0.00013889274488802458,
      "loss": 12.157,
      "step": 14385
    },
    {
      "epoch": 0.7833751948432065,
      "grad_norm": 0.5768927983386264,
      "learning_rate": 0.00013888462075434832,
      "loss": 12.1987,
      "step": 14386
    },
    {
      "epoch": 0.7834296488397895,
      "grad_norm": 0.5348814636384673,
      "learning_rate": 0.00013887649631829879,
      "loss": 12.2369,
      "step": 14387
    },
    {
      "epoch": 0.7834841028363726,
      "grad_norm": 0.5322337299843027,
      "learning_rate": 0.00013886837157993904,
      "loss": 12.1423,
      "step": 14388
    },
    {
      "epoch": 0.7835385568329556,
      "grad_norm": 0.5413549190692319,
      "learning_rate": 0.00013886024653933233,
      "loss": 12.2067,
      "step": 14389
    },
    {
      "epoch": 0.7835930108295386,
      "grad_norm": 0.5780044416754666,
      "learning_rate": 0.0001388521211965418,
      "loss": 12.1955,
      "step": 14390
    },
    {
      "epoch": 0.7836474648261216,
      "grad_norm": 0.5817049120825359,
      "learning_rate": 0.0001388439955516307,
      "loss": 12.2339,
      "step": 14391
    },
    {
      "epoch": 0.7837019188227046,
      "grad_norm": 0.5970867677148517,
      "learning_rate": 0.0001388358696046622,
      "loss": 12.3324,
      "step": 14392
    },
    {
      "epoch": 0.7837563728192876,
      "grad_norm": 0.6115114126909592,
      "learning_rate": 0.00013882774335569943,
      "loss": 12.1704,
      "step": 14393
    },
    {
      "epoch": 0.7838108268158707,
      "grad_norm": 0.6358443448694101,
      "learning_rate": 0.00013881961680480562,
      "loss": 12.2058,
      "step": 14394
    },
    {
      "epoch": 0.7838652808124537,
      "grad_norm": 0.5083773725240279,
      "learning_rate": 0.00013881148995204394,
      "loss": 12.0843,
      "step": 14395
    },
    {
      "epoch": 0.7839197348090367,
      "grad_norm": 0.5771923630494237,
      "learning_rate": 0.0001388033627974776,
      "loss": 12.2925,
      "step": 14396
    },
    {
      "epoch": 0.7839741888056196,
      "grad_norm": 0.7033857158513953,
      "learning_rate": 0.00013879523534116988,
      "loss": 12.3006,
      "step": 14397
    },
    {
      "epoch": 0.7840286428022026,
      "grad_norm": 0.5933514222197236,
      "learning_rate": 0.00013878710758318384,
      "loss": 12.2273,
      "step": 14398
    },
    {
      "epoch": 0.7840830967987856,
      "grad_norm": 0.5768914323508817,
      "learning_rate": 0.00013877897952358276,
      "loss": 12.1288,
      "step": 14399
    },
    {
      "epoch": 0.7841375507953687,
      "grad_norm": 0.5646886542575501,
      "learning_rate": 0.00013877085116242982,
      "loss": 12.2829,
      "step": 14400
    },
    {
      "epoch": 0.7841920047919517,
      "grad_norm": 0.6477289200019736,
      "learning_rate": 0.00013876272249978823,
      "loss": 12.3978,
      "step": 14401
    },
    {
      "epoch": 0.7842464587885347,
      "grad_norm": 0.5648206891195865,
      "learning_rate": 0.00013875459353572124,
      "loss": 12.1633,
      "step": 14402
    },
    {
      "epoch": 0.7843009127851177,
      "grad_norm": 0.6490668828600686,
      "learning_rate": 0.00013874646427029203,
      "loss": 12.2609,
      "step": 14403
    },
    {
      "epoch": 0.7843553667817007,
      "grad_norm": 0.5890545313527348,
      "learning_rate": 0.00013873833470356381,
      "loss": 12.3098,
      "step": 14404
    },
    {
      "epoch": 0.7844098207782837,
      "grad_norm": 0.6127329938396254,
      "learning_rate": 0.00013873020483559978,
      "loss": 12.3745,
      "step": 14405
    },
    {
      "epoch": 0.7844642747748668,
      "grad_norm": 0.5983140066662879,
      "learning_rate": 0.00013872207466646323,
      "loss": 12.2806,
      "step": 14406
    },
    {
      "epoch": 0.7845187287714498,
      "grad_norm": 0.5606783047599062,
      "learning_rate": 0.0001387139441962173,
      "loss": 12.0884,
      "step": 14407
    },
    {
      "epoch": 0.7845731827680328,
      "grad_norm": 0.6118712385980577,
      "learning_rate": 0.00013870581342492527,
      "loss": 12.2015,
      "step": 14408
    },
    {
      "epoch": 0.7846276367646158,
      "grad_norm": 0.5752553693397917,
      "learning_rate": 0.00013869768235265034,
      "loss": 12.262,
      "step": 14409
    },
    {
      "epoch": 0.7846820907611988,
      "grad_norm": 0.6494019440602582,
      "learning_rate": 0.00013868955097945572,
      "loss": 12.2359,
      "step": 14410
    },
    {
      "epoch": 0.7847365447577818,
      "grad_norm": 0.5540211353726932,
      "learning_rate": 0.00013868141930540467,
      "loss": 12.2692,
      "step": 14411
    },
    {
      "epoch": 0.7847909987543649,
      "grad_norm": 0.5986349560833598,
      "learning_rate": 0.00013867328733056043,
      "loss": 12.1355,
      "step": 14412
    },
    {
      "epoch": 0.7848454527509479,
      "grad_norm": 0.5163497163917473,
      "learning_rate": 0.00013866515505498619,
      "loss": 12.2954,
      "step": 14413
    },
    {
      "epoch": 0.7848999067475309,
      "grad_norm": 0.6433436001488466,
      "learning_rate": 0.00013865702247874525,
      "loss": 12.2901,
      "step": 14414
    },
    {
      "epoch": 0.7849543607441138,
      "grad_norm": 0.581890366088506,
      "learning_rate": 0.0001386488896019008,
      "loss": 12.0715,
      "step": 14415
    },
    {
      "epoch": 0.7850088147406968,
      "grad_norm": 0.6102902133868333,
      "learning_rate": 0.0001386407564245161,
      "loss": 12.3666,
      "step": 14416
    },
    {
      "epoch": 0.7850632687372799,
      "grad_norm": 0.5804923766002399,
      "learning_rate": 0.00013863262294665444,
      "loss": 12.3214,
      "step": 14417
    },
    {
      "epoch": 0.7851177227338629,
      "grad_norm": 0.6068979406349941,
      "learning_rate": 0.00013862448916837901,
      "loss": 12.41,
      "step": 14418
    },
    {
      "epoch": 0.7851721767304459,
      "grad_norm": 0.6058305883785644,
      "learning_rate": 0.00013861635508975305,
      "loss": 12.224,
      "step": 14419
    },
    {
      "epoch": 0.7852266307270289,
      "grad_norm": 0.5933602069707344,
      "learning_rate": 0.00013860822071083985,
      "loss": 12.2893,
      "step": 14420
    },
    {
      "epoch": 0.7852810847236119,
      "grad_norm": 0.5838506041908982,
      "learning_rate": 0.00013860008603170266,
      "loss": 12.219,
      "step": 14421
    },
    {
      "epoch": 0.7853355387201949,
      "grad_norm": 0.565446422535431,
      "learning_rate": 0.0001385919510524047,
      "loss": 12.1994,
      "step": 14422
    },
    {
      "epoch": 0.785389992716778,
      "grad_norm": 0.591349648141618,
      "learning_rate": 0.00013858381577300927,
      "loss": 12.3008,
      "step": 14423
    },
    {
      "epoch": 0.785444446713361,
      "grad_norm": 0.6212443898915186,
      "learning_rate": 0.00013857568019357962,
      "loss": 12.3911,
      "step": 14424
    },
    {
      "epoch": 0.785498900709944,
      "grad_norm": 0.5605166019234723,
      "learning_rate": 0.00013856754431417905,
      "loss": 12.297,
      "step": 14425
    },
    {
      "epoch": 0.785553354706527,
      "grad_norm": 0.5557117201016185,
      "learning_rate": 0.00013855940813487075,
      "loss": 12.2484,
      "step": 14426
    },
    {
      "epoch": 0.78560780870311,
      "grad_norm": 0.5896367418056946,
      "learning_rate": 0.00013855127165571804,
      "loss": 12.0708,
      "step": 14427
    },
    {
      "epoch": 0.785662262699693,
      "grad_norm": 0.5566213369949221,
      "learning_rate": 0.00013854313487678414,
      "loss": 12.2472,
      "step": 14428
    },
    {
      "epoch": 0.7857167166962761,
      "grad_norm": 0.5948199089283874,
      "learning_rate": 0.0001385349977981324,
      "loss": 12.3153,
      "step": 14429
    },
    {
      "epoch": 0.7857711706928591,
      "grad_norm": 0.6235122010792709,
      "learning_rate": 0.00013852686041982603,
      "loss": 12.3027,
      "step": 14430
    },
    {
      "epoch": 0.7858256246894421,
      "grad_norm": 0.5084880832530156,
      "learning_rate": 0.00013851872274192833,
      "loss": 12.1476,
      "step": 14431
    },
    {
      "epoch": 0.785880078686025,
      "grad_norm": 0.5662903553664386,
      "learning_rate": 0.00013851058476450264,
      "loss": 12.1704,
      "step": 14432
    },
    {
      "epoch": 0.785934532682608,
      "grad_norm": 0.5153724619182718,
      "learning_rate": 0.0001385024464876121,
      "loss": 12.1991,
      "step": 14433
    },
    {
      "epoch": 0.785988986679191,
      "grad_norm": 0.608915473834492,
      "learning_rate": 0.00013849430791132013,
      "loss": 12.4918,
      "step": 14434
    },
    {
      "epoch": 0.7860434406757741,
      "grad_norm": 0.5316088716633605,
      "learning_rate": 0.00013848616903568996,
      "loss": 12.1875,
      "step": 14435
    },
    {
      "epoch": 0.7860978946723571,
      "grad_norm": 0.5752852247644937,
      "learning_rate": 0.00013847802986078486,
      "loss": 12.2453,
      "step": 14436
    },
    {
      "epoch": 0.7861523486689401,
      "grad_norm": 0.5561773848748036,
      "learning_rate": 0.0001384698903866682,
      "loss": 12.1688,
      "step": 14437
    },
    {
      "epoch": 0.7862068026655231,
      "grad_norm": 0.5808394601058333,
      "learning_rate": 0.00013846175061340318,
      "loss": 12.2476,
      "step": 14438
    },
    {
      "epoch": 0.7862612566621061,
      "grad_norm": 0.6028946704665702,
      "learning_rate": 0.00013845361054105312,
      "loss": 12.2198,
      "step": 14439
    },
    {
      "epoch": 0.7863157106586891,
      "grad_norm": 0.5933014981190774,
      "learning_rate": 0.00013844547016968138,
      "loss": 12.2097,
      "step": 14440
    },
    {
      "epoch": 0.7863701646552722,
      "grad_norm": 0.5770644222918464,
      "learning_rate": 0.0001384373294993512,
      "loss": 12.1455,
      "step": 14441
    },
    {
      "epoch": 0.7864246186518552,
      "grad_norm": 0.6156425720653473,
      "learning_rate": 0.00013842918853012592,
      "loss": 12.2782,
      "step": 14442
    },
    {
      "epoch": 0.7864790726484382,
      "grad_norm": 0.6109416698177631,
      "learning_rate": 0.0001384210472620688,
      "loss": 12.2674,
      "step": 14443
    },
    {
      "epoch": 0.7865335266450212,
      "grad_norm": 0.6145744345463588,
      "learning_rate": 0.00013841290569524314,
      "loss": 12.2003,
      "step": 14444
    },
    {
      "epoch": 0.7865879806416042,
      "grad_norm": 0.5651575047053573,
      "learning_rate": 0.00013840476382971233,
      "loss": 12.2573,
      "step": 14445
    },
    {
      "epoch": 0.7866424346381872,
      "grad_norm": 0.569776001009847,
      "learning_rate": 0.00013839662166553965,
      "loss": 12.2178,
      "step": 14446
    },
    {
      "epoch": 0.7866968886347703,
      "grad_norm": 0.5834825925351157,
      "learning_rate": 0.00013838847920278838,
      "loss": 12.203,
      "step": 14447
    },
    {
      "epoch": 0.7867513426313533,
      "grad_norm": 0.5820569556884193,
      "learning_rate": 0.00013838033644152185,
      "loss": 12.1508,
      "step": 14448
    },
    {
      "epoch": 0.7868057966279363,
      "grad_norm": 0.549526625619789,
      "learning_rate": 0.0001383721933818034,
      "loss": 12.1994,
      "step": 14449
    },
    {
      "epoch": 0.7868602506245193,
      "grad_norm": 0.6419107847056853,
      "learning_rate": 0.0001383640500236963,
      "loss": 12.1019,
      "step": 14450
    },
    {
      "epoch": 0.7869147046211022,
      "grad_norm": 0.5877121778475324,
      "learning_rate": 0.00013835590636726396,
      "loss": 12.2655,
      "step": 14451
    },
    {
      "epoch": 0.7869691586176853,
      "grad_norm": 0.5575112325526567,
      "learning_rate": 0.00013834776241256966,
      "loss": 12.223,
      "step": 14452
    },
    {
      "epoch": 0.7870236126142683,
      "grad_norm": 0.5705576045447335,
      "learning_rate": 0.0001383396181596767,
      "loss": 12.2809,
      "step": 14453
    },
    {
      "epoch": 0.7870780666108513,
      "grad_norm": 0.5420703228607463,
      "learning_rate": 0.00013833147360864848,
      "loss": 12.2297,
      "step": 14454
    },
    {
      "epoch": 0.7871325206074343,
      "grad_norm": 0.5484969588430767,
      "learning_rate": 0.00013832332875954824,
      "loss": 12.2913,
      "step": 14455
    },
    {
      "epoch": 0.7871869746040173,
      "grad_norm": 0.549863675343978,
      "learning_rate": 0.00013831518361243937,
      "loss": 12.2342,
      "step": 14456
    },
    {
      "epoch": 0.7872414286006003,
      "grad_norm": 0.5468044821046292,
      "learning_rate": 0.00013830703816738527,
      "loss": 12.2085,
      "step": 14457
    },
    {
      "epoch": 0.7872958825971834,
      "grad_norm": 0.6595316857062661,
      "learning_rate": 0.00013829889242444917,
      "loss": 12.1952,
      "step": 14458
    },
    {
      "epoch": 0.7873503365937664,
      "grad_norm": 0.6309748753681473,
      "learning_rate": 0.0001382907463836945,
      "loss": 12.1863,
      "step": 14459
    },
    {
      "epoch": 0.7874047905903494,
      "grad_norm": 0.528524171845402,
      "learning_rate": 0.0001382826000451845,
      "loss": 12.1068,
      "step": 14460
    },
    {
      "epoch": 0.7874592445869324,
      "grad_norm": 0.6553348815463083,
      "learning_rate": 0.00013827445340898258,
      "loss": 12.2282,
      "step": 14461
    },
    {
      "epoch": 0.7875136985835154,
      "grad_norm": 0.5770544173966313,
      "learning_rate": 0.00013826630647515218,
      "loss": 12.2327,
      "step": 14462
    },
    {
      "epoch": 0.7875681525800984,
      "grad_norm": 0.6103420761265705,
      "learning_rate": 0.0001382581592437565,
      "loss": 12.3382,
      "step": 14463
    },
    {
      "epoch": 0.7876226065766815,
      "grad_norm": 0.6456961095391563,
      "learning_rate": 0.00013825001171485895,
      "loss": 12.2109,
      "step": 14464
    },
    {
      "epoch": 0.7876770605732645,
      "grad_norm": 0.6069538803998202,
      "learning_rate": 0.0001382418638885229,
      "loss": 12.2003,
      "step": 14465
    },
    {
      "epoch": 0.7877315145698475,
      "grad_norm": 0.6036311575066308,
      "learning_rate": 0.00013823371576481174,
      "loss": 12.2489,
      "step": 14466
    },
    {
      "epoch": 0.7877859685664305,
      "grad_norm": 0.6012697959525011,
      "learning_rate": 0.00013822556734378877,
      "loss": 12.2291,
      "step": 14467
    },
    {
      "epoch": 0.7878404225630135,
      "grad_norm": 0.5750414494310325,
      "learning_rate": 0.00013821741862551738,
      "loss": 12.1316,
      "step": 14468
    },
    {
      "epoch": 0.7878948765595964,
      "grad_norm": 0.6562236907687204,
      "learning_rate": 0.00013820926961006092,
      "loss": 12.1454,
      "step": 14469
    },
    {
      "epoch": 0.7879493305561796,
      "grad_norm": 0.61036382134781,
      "learning_rate": 0.0001382011202974828,
      "loss": 12.3017,
      "step": 14470
    },
    {
      "epoch": 0.7880037845527625,
      "grad_norm": 0.5865091461942048,
      "learning_rate": 0.00013819297068784633,
      "loss": 12.2436,
      "step": 14471
    },
    {
      "epoch": 0.7880582385493455,
      "grad_norm": 0.6258204578276099,
      "learning_rate": 0.00013818482078121496,
      "loss": 12.297,
      "step": 14472
    },
    {
      "epoch": 0.7881126925459285,
      "grad_norm": 0.5526784225643813,
      "learning_rate": 0.00013817667057765197,
      "loss": 12.263,
      "step": 14473
    },
    {
      "epoch": 0.7881671465425115,
      "grad_norm": 0.6303789008627477,
      "learning_rate": 0.0001381685200772208,
      "loss": 12.0504,
      "step": 14474
    },
    {
      "epoch": 0.7882216005390945,
      "grad_norm": 0.5270705695551884,
      "learning_rate": 0.00013816036927998484,
      "loss": 12.0476,
      "step": 14475
    },
    {
      "epoch": 0.7882760545356776,
      "grad_norm": 0.5661338193897029,
      "learning_rate": 0.00013815221818600743,
      "loss": 12.3261,
      "step": 14476
    },
    {
      "epoch": 0.7883305085322606,
      "grad_norm": 0.4920199077371078,
      "learning_rate": 0.000138144066795352,
      "loss": 12.1692,
      "step": 14477
    },
    {
      "epoch": 0.7883849625288436,
      "grad_norm": 0.5405871902740343,
      "learning_rate": 0.00013813591510808187,
      "loss": 12.1211,
      "step": 14478
    },
    {
      "epoch": 0.7884394165254266,
      "grad_norm": 0.5315923941485983,
      "learning_rate": 0.00013812776312426047,
      "loss": 12.0371,
      "step": 14479
    },
    {
      "epoch": 0.7884938705220096,
      "grad_norm": 0.5127041828760139,
      "learning_rate": 0.0001381196108439512,
      "loss": 12.135,
      "step": 14480
    },
    {
      "epoch": 0.7885483245185927,
      "grad_norm": 0.5459664774774081,
      "learning_rate": 0.00013811145826721747,
      "loss": 12.1413,
      "step": 14481
    },
    {
      "epoch": 0.7886027785151757,
      "grad_norm": 0.6777493278458847,
      "learning_rate": 0.00013810330539412263,
      "loss": 12.2048,
      "step": 14482
    },
    {
      "epoch": 0.7886572325117587,
      "grad_norm": 0.567214388297729,
      "learning_rate": 0.0001380951522247301,
      "loss": 12.194,
      "step": 14483
    },
    {
      "epoch": 0.7887116865083417,
      "grad_norm": 0.5948900536613386,
      "learning_rate": 0.00013808699875910327,
      "loss": 12.1971,
      "step": 14484
    },
    {
      "epoch": 0.7887661405049247,
      "grad_norm": 0.552835977401995,
      "learning_rate": 0.00013807884499730555,
      "loss": 12.2779,
      "step": 14485
    },
    {
      "epoch": 0.7888205945015077,
      "grad_norm": 0.6584396299441723,
      "learning_rate": 0.00013807069093940035,
      "loss": 12.3761,
      "step": 14486
    },
    {
      "epoch": 0.7888750484980908,
      "grad_norm": 0.5307595145590913,
      "learning_rate": 0.0001380625365854511,
      "loss": 12.2804,
      "step": 14487
    },
    {
      "epoch": 0.7889295024946738,
      "grad_norm": 0.5491257047139692,
      "learning_rate": 0.00013805438193552114,
      "loss": 12.2452,
      "step": 14488
    },
    {
      "epoch": 0.7889839564912567,
      "grad_norm": 0.5955458673578021,
      "learning_rate": 0.00013804622698967392,
      "loss": 12.1676,
      "step": 14489
    },
    {
      "epoch": 0.7890384104878397,
      "grad_norm": 0.5192351249345226,
      "learning_rate": 0.0001380380717479729,
      "loss": 12.3022,
      "step": 14490
    },
    {
      "epoch": 0.7890928644844227,
      "grad_norm": 0.4887155507006782,
      "learning_rate": 0.00013802991621048141,
      "loss": 12.192,
      "step": 14491
    },
    {
      "epoch": 0.7891473184810057,
      "grad_norm": 0.5547709732773544,
      "learning_rate": 0.00013802176037726297,
      "loss": 12.2312,
      "step": 14492
    },
    {
      "epoch": 0.7892017724775888,
      "grad_norm": 0.556762358309864,
      "learning_rate": 0.00013801360424838089,
      "loss": 12.2581,
      "step": 14493
    },
    {
      "epoch": 0.7892562264741718,
      "grad_norm": 0.5308349164149758,
      "learning_rate": 0.00013800544782389867,
      "loss": 12.2044,
      "step": 14494
    },
    {
      "epoch": 0.7893106804707548,
      "grad_norm": 0.5452192484856463,
      "learning_rate": 0.00013799729110387972,
      "loss": 12.2182,
      "step": 14495
    },
    {
      "epoch": 0.7893651344673378,
      "grad_norm": 0.5342171146676084,
      "learning_rate": 0.00013798913408838746,
      "loss": 12.1737,
      "step": 14496
    },
    {
      "epoch": 0.7894195884639208,
      "grad_norm": 0.5880953024934228,
      "learning_rate": 0.00013798097677748534,
      "loss": 12.1635,
      "step": 14497
    },
    {
      "epoch": 0.7894740424605038,
      "grad_norm": 0.5501073488750367,
      "learning_rate": 0.00013797281917123674,
      "loss": 12.1803,
      "step": 14498
    },
    {
      "epoch": 0.7895284964570869,
      "grad_norm": 0.5466462771813189,
      "learning_rate": 0.00013796466126970514,
      "loss": 12.282,
      "step": 14499
    },
    {
      "epoch": 0.7895829504536699,
      "grad_norm": 0.555337501442166,
      "learning_rate": 0.00013795650307295396,
      "loss": 12.2572,
      "step": 14500
    },
    {
      "epoch": 0.7896374044502529,
      "grad_norm": 0.5644067977646973,
      "learning_rate": 0.00013794834458104665,
      "loss": 12.2426,
      "step": 14501
    },
    {
      "epoch": 0.7896918584468359,
      "grad_norm": 0.6290705505458757,
      "learning_rate": 0.00013794018579404668,
      "loss": 12.3258,
      "step": 14502
    },
    {
      "epoch": 0.7897463124434189,
      "grad_norm": 0.5530317475999645,
      "learning_rate": 0.00013793202671201745,
      "loss": 12.2766,
      "step": 14503
    },
    {
      "epoch": 0.7898007664400019,
      "grad_norm": 0.5526295277722347,
      "learning_rate": 0.0001379238673350224,
      "loss": 12.3009,
      "step": 14504
    },
    {
      "epoch": 0.789855220436585,
      "grad_norm": 0.5924239444057052,
      "learning_rate": 0.000137915707663125,
      "loss": 12.1161,
      "step": 14505
    },
    {
      "epoch": 0.789909674433168,
      "grad_norm": 0.59420567157587,
      "learning_rate": 0.0001379075476963887,
      "loss": 12.1919,
      "step": 14506
    },
    {
      "epoch": 0.789964128429751,
      "grad_norm": 0.6298810966617565,
      "learning_rate": 0.00013789938743487696,
      "loss": 12.2766,
      "step": 14507
    },
    {
      "epoch": 0.7900185824263339,
      "grad_norm": 0.5476263842566301,
      "learning_rate": 0.00013789122687865325,
      "loss": 12.2857,
      "step": 14508
    },
    {
      "epoch": 0.7900730364229169,
      "grad_norm": 0.555176190503391,
      "learning_rate": 0.00013788306602778097,
      "loss": 12.132,
      "step": 14509
    },
    {
      "epoch": 0.7901274904194999,
      "grad_norm": 0.6204025876595295,
      "learning_rate": 0.00013787490488232363,
      "loss": 12.2649,
      "step": 14510
    },
    {
      "epoch": 0.790181944416083,
      "grad_norm": 0.5461093843848734,
      "learning_rate": 0.00013786674344234466,
      "loss": 12.1079,
      "step": 14511
    },
    {
      "epoch": 0.790236398412666,
      "grad_norm": 0.5836535420064173,
      "learning_rate": 0.00013785858170790754,
      "loss": 11.9077,
      "step": 14512
    },
    {
      "epoch": 0.790290852409249,
      "grad_norm": 0.5692528585134783,
      "learning_rate": 0.0001378504196790758,
      "loss": 12.3145,
      "step": 14513
    },
    {
      "epoch": 0.790345306405832,
      "grad_norm": 0.526143377711485,
      "learning_rate": 0.00013784225735591278,
      "loss": 12.1412,
      "step": 14514
    },
    {
      "epoch": 0.790399760402415,
      "grad_norm": 0.6780130437929691,
      "learning_rate": 0.00013783409473848207,
      "loss": 12.1476,
      "step": 14515
    },
    {
      "epoch": 0.7904542143989981,
      "grad_norm": 0.5763331530144483,
      "learning_rate": 0.00013782593182684705,
      "loss": 12.2956,
      "step": 14516
    },
    {
      "epoch": 0.7905086683955811,
      "grad_norm": 0.5775096015189632,
      "learning_rate": 0.00013781776862107126,
      "loss": 12.259,
      "step": 14517
    },
    {
      "epoch": 0.7905631223921641,
      "grad_norm": 0.5617552123384292,
      "learning_rate": 0.00013780960512121818,
      "loss": 12.0761,
      "step": 14518
    },
    {
      "epoch": 0.7906175763887471,
      "grad_norm": 0.715470699824914,
      "learning_rate": 0.00013780144132735124,
      "loss": 12.267,
      "step": 14519
    },
    {
      "epoch": 0.7906720303853301,
      "grad_norm": 0.6163374199941223,
      "learning_rate": 0.00013779327723953397,
      "loss": 12.1038,
      "step": 14520
    },
    {
      "epoch": 0.7907264843819131,
      "grad_norm": 0.5935665540501485,
      "learning_rate": 0.0001377851128578298,
      "loss": 12.2294,
      "step": 14521
    },
    {
      "epoch": 0.7907809383784962,
      "grad_norm": 0.5838932995567909,
      "learning_rate": 0.0001377769481823023,
      "loss": 12.312,
      "step": 14522
    },
    {
      "epoch": 0.7908353923750792,
      "grad_norm": 0.6095538760565489,
      "learning_rate": 0.00013776878321301492,
      "loss": 12.1626,
      "step": 14523
    },
    {
      "epoch": 0.7908898463716622,
      "grad_norm": 0.5922649772685893,
      "learning_rate": 0.00013776061795003113,
      "loss": 12.3035,
      "step": 14524
    },
    {
      "epoch": 0.7909443003682451,
      "grad_norm": 0.5515954714545865,
      "learning_rate": 0.00013775245239341444,
      "loss": 12.315,
      "step": 14525
    },
    {
      "epoch": 0.7909987543648281,
      "grad_norm": 0.6539394474403301,
      "learning_rate": 0.00013774428654322836,
      "loss": 12.242,
      "step": 14526
    },
    {
      "epoch": 0.7910532083614111,
      "grad_norm": 0.5634951214555385,
      "learning_rate": 0.00013773612039953635,
      "loss": 12.26,
      "step": 14527
    },
    {
      "epoch": 0.7911076623579942,
      "grad_norm": 0.6319106731830428,
      "learning_rate": 0.000137727953962402,
      "loss": 12.3335,
      "step": 14528
    },
    {
      "epoch": 0.7911621163545772,
      "grad_norm": 0.5989947043730212,
      "learning_rate": 0.00013771978723188869,
      "loss": 12.0249,
      "step": 14529
    },
    {
      "epoch": 0.7912165703511602,
      "grad_norm": 0.658527852097458,
      "learning_rate": 0.00013771162020806004,
      "loss": 12.1369,
      "step": 14530
    },
    {
      "epoch": 0.7912710243477432,
      "grad_norm": 0.6143727627790688,
      "learning_rate": 0.00013770345289097945,
      "loss": 12.2429,
      "step": 14531
    },
    {
      "epoch": 0.7913254783443262,
      "grad_norm": 0.7332441044745701,
      "learning_rate": 0.00013769528528071053,
      "loss": 12.2045,
      "step": 14532
    },
    {
      "epoch": 0.7913799323409092,
      "grad_norm": 0.7027518226233722,
      "learning_rate": 0.00013768711737731674,
      "loss": 12.3577,
      "step": 14533
    },
    {
      "epoch": 0.7914343863374923,
      "grad_norm": 0.6661778673562905,
      "learning_rate": 0.0001376789491808616,
      "loss": 12.2812,
      "step": 14534
    },
    {
      "epoch": 0.7914888403340753,
      "grad_norm": 0.7305109231819773,
      "learning_rate": 0.00013767078069140863,
      "loss": 12.3905,
      "step": 14535
    },
    {
      "epoch": 0.7915432943306583,
      "grad_norm": 0.678911449401082,
      "learning_rate": 0.00013766261190902138,
      "loss": 12.2408,
      "step": 14536
    },
    {
      "epoch": 0.7915977483272413,
      "grad_norm": 0.6530486210619075,
      "learning_rate": 0.0001376544428337633,
      "loss": 12.1734,
      "step": 14537
    },
    {
      "epoch": 0.7916522023238243,
      "grad_norm": 0.5827242495444377,
      "learning_rate": 0.00013764627346569798,
      "loss": 12.2727,
      "step": 14538
    },
    {
      "epoch": 0.7917066563204073,
      "grad_norm": 0.542965175998758,
      "learning_rate": 0.00013763810380488893,
      "loss": 12.2165,
      "step": 14539
    },
    {
      "epoch": 0.7917611103169904,
      "grad_norm": 0.6735407450418386,
      "learning_rate": 0.00013762993385139967,
      "loss": 12.2434,
      "step": 14540
    },
    {
      "epoch": 0.7918155643135734,
      "grad_norm": 0.5408198038923513,
      "learning_rate": 0.0001376217636052937,
      "loss": 12.2286,
      "step": 14541
    },
    {
      "epoch": 0.7918700183101564,
      "grad_norm": 0.6335727041039692,
      "learning_rate": 0.00013761359306663465,
      "loss": 12.3902,
      "step": 14542
    },
    {
      "epoch": 0.7919244723067393,
      "grad_norm": 0.7766853164870627,
      "learning_rate": 0.00013760542223548595,
      "loss": 12.3514,
      "step": 14543
    },
    {
      "epoch": 0.7919789263033223,
      "grad_norm": 0.5963601706181519,
      "learning_rate": 0.00013759725111191118,
      "loss": 12.1699,
      "step": 14544
    },
    {
      "epoch": 0.7920333802999053,
      "grad_norm": 0.641213799629075,
      "learning_rate": 0.0001375890796959739,
      "loss": 12.32,
      "step": 14545
    },
    {
      "epoch": 0.7920878342964884,
      "grad_norm": 0.6972634743512368,
      "learning_rate": 0.00013758090798773762,
      "loss": 12.3032,
      "step": 14546
    },
    {
      "epoch": 0.7921422882930714,
      "grad_norm": 0.6017069234143751,
      "learning_rate": 0.00013757273598726587,
      "loss": 12.3563,
      "step": 14547
    },
    {
      "epoch": 0.7921967422896544,
      "grad_norm": 0.6360018090555419,
      "learning_rate": 0.00013756456369462227,
      "loss": 12.2611,
      "step": 14548
    },
    {
      "epoch": 0.7922511962862374,
      "grad_norm": 0.6503478151736607,
      "learning_rate": 0.00013755639110987032,
      "loss": 12.3143,
      "step": 14549
    },
    {
      "epoch": 0.7923056502828204,
      "grad_norm": 0.5543365015342491,
      "learning_rate": 0.00013754821823307354,
      "loss": 12.3949,
      "step": 14550
    },
    {
      "epoch": 0.7923601042794035,
      "grad_norm": 0.5973243221748423,
      "learning_rate": 0.00013754004506429554,
      "loss": 12.3585,
      "step": 14551
    },
    {
      "epoch": 0.7924145582759865,
      "grad_norm": 0.7117129070923307,
      "learning_rate": 0.00013753187160359985,
      "loss": 12.2683,
      "step": 14552
    },
    {
      "epoch": 0.7924690122725695,
      "grad_norm": 0.5653042724525108,
      "learning_rate": 0.00013752369785105005,
      "loss": 12.2011,
      "step": 14553
    },
    {
      "epoch": 0.7925234662691525,
      "grad_norm": 0.6353979891028753,
      "learning_rate": 0.00013751552380670968,
      "loss": 12.2549,
      "step": 14554
    },
    {
      "epoch": 0.7925779202657355,
      "grad_norm": 0.5861169980674861,
      "learning_rate": 0.00013750734947064227,
      "loss": 12.2593,
      "step": 14555
    },
    {
      "epoch": 0.7926323742623185,
      "grad_norm": 0.6516937965590943,
      "learning_rate": 0.00013749917484291144,
      "loss": 12.2449,
      "step": 14556
    },
    {
      "epoch": 0.7926868282589016,
      "grad_norm": 0.6707502137605735,
      "learning_rate": 0.00013749099992358077,
      "loss": 12.2887,
      "step": 14557
    },
    {
      "epoch": 0.7927412822554846,
      "grad_norm": 0.5770933413979192,
      "learning_rate": 0.00013748282471271375,
      "loss": 12.2399,
      "step": 14558
    },
    {
      "epoch": 0.7927957362520676,
      "grad_norm": 0.6133565514888829,
      "learning_rate": 0.000137474649210374,
      "loss": 12.1271,
      "step": 14559
    },
    {
      "epoch": 0.7928501902486506,
      "grad_norm": 0.565282292880067,
      "learning_rate": 0.00013746647341662512,
      "loss": 12.0827,
      "step": 14560
    },
    {
      "epoch": 0.7929046442452335,
      "grad_norm": 0.630819282440226,
      "learning_rate": 0.00013745829733153065,
      "loss": 12.2273,
      "step": 14561
    },
    {
      "epoch": 0.7929590982418165,
      "grad_norm": 0.5287941385974477,
      "learning_rate": 0.00013745012095515418,
      "loss": 12.3124,
      "step": 14562
    },
    {
      "epoch": 0.7930135522383996,
      "grad_norm": 0.6013911595860558,
      "learning_rate": 0.00013744194428755933,
      "loss": 12.1449,
      "step": 14563
    },
    {
      "epoch": 0.7930680062349826,
      "grad_norm": 0.5908970621387647,
      "learning_rate": 0.0001374337673288096,
      "loss": 11.9763,
      "step": 14564
    },
    {
      "epoch": 0.7931224602315656,
      "grad_norm": 0.5678961642577054,
      "learning_rate": 0.0001374255900789686,
      "loss": 12.1764,
      "step": 14565
    },
    {
      "epoch": 0.7931769142281486,
      "grad_norm": 0.5223695709824954,
      "learning_rate": 0.00013741741253809996,
      "loss": 12.0156,
      "step": 14566
    },
    {
      "epoch": 0.7932313682247316,
      "grad_norm": 0.564503233628171,
      "learning_rate": 0.0001374092347062672,
      "loss": 12.188,
      "step": 14567
    },
    {
      "epoch": 0.7932858222213146,
      "grad_norm": 0.5416228729888778,
      "learning_rate": 0.00013740105658353405,
      "loss": 12.1661,
      "step": 14568
    },
    {
      "epoch": 0.7933402762178977,
      "grad_norm": 0.6134468990697433,
      "learning_rate": 0.00013739287816996395,
      "loss": 12.3041,
      "step": 14569
    },
    {
      "epoch": 0.7933947302144807,
      "grad_norm": 0.5865723569145507,
      "learning_rate": 0.00013738469946562054,
      "loss": 12.1678,
      "step": 14570
    },
    {
      "epoch": 0.7934491842110637,
      "grad_norm": 0.5917678061680537,
      "learning_rate": 0.00013737652047056745,
      "loss": 12.1629,
      "step": 14571
    },
    {
      "epoch": 0.7935036382076467,
      "grad_norm": 0.6067411146767691,
      "learning_rate": 0.0001373683411848683,
      "loss": 12.2797,
      "step": 14572
    },
    {
      "epoch": 0.7935580922042297,
      "grad_norm": 0.5688816523572685,
      "learning_rate": 0.00013736016160858667,
      "loss": 12.1704,
      "step": 14573
    },
    {
      "epoch": 0.7936125462008127,
      "grad_norm": 0.5348887968152709,
      "learning_rate": 0.00013735198174178614,
      "loss": 12.1637,
      "step": 14574
    },
    {
      "epoch": 0.7936670001973958,
      "grad_norm": 0.5976971969257272,
      "learning_rate": 0.0001373438015845303,
      "loss": 12.2787,
      "step": 14575
    },
    {
      "epoch": 0.7937214541939788,
      "grad_norm": 0.6372535098151176,
      "learning_rate": 0.00013733562113688283,
      "loss": 12.2731,
      "step": 14576
    },
    {
      "epoch": 0.7937759081905618,
      "grad_norm": 0.5856623029227053,
      "learning_rate": 0.0001373274403989073,
      "loss": 12.1469,
      "step": 14577
    },
    {
      "epoch": 0.7938303621871448,
      "grad_norm": 0.5340829343077999,
      "learning_rate": 0.00013731925937066736,
      "loss": 12.2518,
      "step": 14578
    },
    {
      "epoch": 0.7938848161837277,
      "grad_norm": 0.6135248305708401,
      "learning_rate": 0.0001373110780522266,
      "loss": 12.2766,
      "step": 14579
    },
    {
      "epoch": 0.7939392701803107,
      "grad_norm": 0.5980173322439528,
      "learning_rate": 0.0001373028964436486,
      "loss": 12.1919,
      "step": 14580
    },
    {
      "epoch": 0.7939937241768938,
      "grad_norm": 0.6206658138611297,
      "learning_rate": 0.00013729471454499703,
      "loss": 12.1507,
      "step": 14581
    },
    {
      "epoch": 0.7940481781734768,
      "grad_norm": 0.5367388619002081,
      "learning_rate": 0.0001372865323563355,
      "loss": 12.2038,
      "step": 14582
    },
    {
      "epoch": 0.7941026321700598,
      "grad_norm": 0.5240897993979492,
      "learning_rate": 0.00013727834987772768,
      "loss": 12.2194,
      "step": 14583
    },
    {
      "epoch": 0.7941570861666428,
      "grad_norm": 0.647420197970378,
      "learning_rate": 0.0001372701671092371,
      "loss": 12.2785,
      "step": 14584
    },
    {
      "epoch": 0.7942115401632258,
      "grad_norm": 0.5684613022836379,
      "learning_rate": 0.0001372619840509275,
      "loss": 12.2735,
      "step": 14585
    },
    {
      "epoch": 0.7942659941598089,
      "grad_norm": 0.5945683442190239,
      "learning_rate": 0.00013725380070286246,
      "loss": 12.2594,
      "step": 14586
    },
    {
      "epoch": 0.7943204481563919,
      "grad_norm": 0.6259752328008525,
      "learning_rate": 0.0001372456170651056,
      "loss": 12.1878,
      "step": 14587
    },
    {
      "epoch": 0.7943749021529749,
      "grad_norm": 0.546480691542995,
      "learning_rate": 0.00013723743313772058,
      "loss": 12.2561,
      "step": 14588
    },
    {
      "epoch": 0.7944293561495579,
      "grad_norm": 0.5150143310049871,
      "learning_rate": 0.00013722924892077106,
      "loss": 12.1557,
      "step": 14589
    },
    {
      "epoch": 0.7944838101461409,
      "grad_norm": 0.561310388062198,
      "learning_rate": 0.00013722106441432064,
      "loss": 12.1778,
      "step": 14590
    },
    {
      "epoch": 0.7945382641427239,
      "grad_norm": 0.606486418543721,
      "learning_rate": 0.00013721287961843297,
      "loss": 12.2936,
      "step": 14591
    },
    {
      "epoch": 0.794592718139307,
      "grad_norm": 0.5837914612089911,
      "learning_rate": 0.00013720469453317173,
      "loss": 12.3163,
      "step": 14592
    },
    {
      "epoch": 0.79464717213589,
      "grad_norm": 0.6707863662057902,
      "learning_rate": 0.00013719650915860053,
      "loss": 12.1247,
      "step": 14593
    },
    {
      "epoch": 0.794701626132473,
      "grad_norm": 0.5614619023691085,
      "learning_rate": 0.00013718832349478305,
      "loss": 12.3049,
      "step": 14594
    },
    {
      "epoch": 0.794756080129056,
      "grad_norm": 0.5097509032435517,
      "learning_rate": 0.0001371801375417829,
      "loss": 12.164,
      "step": 14595
    },
    {
      "epoch": 0.794810534125639,
      "grad_norm": 0.5976077493113269,
      "learning_rate": 0.00013717195129966378,
      "loss": 12.2807,
      "step": 14596
    },
    {
      "epoch": 0.794864988122222,
      "grad_norm": 0.5392292790465633,
      "learning_rate": 0.00013716376476848933,
      "loss": 12.2907,
      "step": 14597
    },
    {
      "epoch": 0.794919442118805,
      "grad_norm": 0.5623035202767939,
      "learning_rate": 0.00013715557794832323,
      "loss": 12.2697,
      "step": 14598
    },
    {
      "epoch": 0.794973896115388,
      "grad_norm": 0.5536073700263109,
      "learning_rate": 0.00013714739083922912,
      "loss": 12.1871,
      "step": 14599
    },
    {
      "epoch": 0.795028350111971,
      "grad_norm": 0.5485839640207039,
      "learning_rate": 0.00013713920344127068,
      "loss": 12.2897,
      "step": 14600
    },
    {
      "epoch": 0.795082804108554,
      "grad_norm": 0.5211499804336917,
      "learning_rate": 0.00013713101575451152,
      "loss": 12.1034,
      "step": 14601
    },
    {
      "epoch": 0.795137258105137,
      "grad_norm": 0.5511440782227762,
      "learning_rate": 0.00013712282777901538,
      "loss": 12.2477,
      "step": 14602
    },
    {
      "epoch": 0.79519171210172,
      "grad_norm": 0.5648455952058448,
      "learning_rate": 0.00013711463951484592,
      "loss": 12.1965,
      "step": 14603
    },
    {
      "epoch": 0.7952461660983031,
      "grad_norm": 0.6047438965263254,
      "learning_rate": 0.0001371064509620668,
      "loss": 12.2426,
      "step": 14604
    },
    {
      "epoch": 0.7953006200948861,
      "grad_norm": 0.5210930126052459,
      "learning_rate": 0.0001370982621207417,
      "loss": 12.2967,
      "step": 14605
    },
    {
      "epoch": 0.7953550740914691,
      "grad_norm": 0.6753107385324894,
      "learning_rate": 0.00013709007299093426,
      "loss": 12.233,
      "step": 14606
    },
    {
      "epoch": 0.7954095280880521,
      "grad_norm": 0.5903172203022329,
      "learning_rate": 0.0001370818835727082,
      "loss": 12.1532,
      "step": 14607
    },
    {
      "epoch": 0.7954639820846351,
      "grad_norm": 0.5400132903393716,
      "learning_rate": 0.00013707369386612721,
      "loss": 12.1621,
      "step": 14608
    },
    {
      "epoch": 0.7955184360812181,
      "grad_norm": 0.5691708873764322,
      "learning_rate": 0.00013706550387125493,
      "loss": 12.2356,
      "step": 14609
    },
    {
      "epoch": 0.7955728900778012,
      "grad_norm": 0.545545193497752,
      "learning_rate": 0.0001370573135881551,
      "loss": 12.2555,
      "step": 14610
    },
    {
      "epoch": 0.7956273440743842,
      "grad_norm": 0.5469260876252976,
      "learning_rate": 0.0001370491230168914,
      "loss": 12.1018,
      "step": 14611
    },
    {
      "epoch": 0.7956817980709672,
      "grad_norm": 0.571494057494282,
      "learning_rate": 0.0001370409321575275,
      "loss": 12.1387,
      "step": 14612
    },
    {
      "epoch": 0.7957362520675502,
      "grad_norm": 0.5837025395276881,
      "learning_rate": 0.00013703274101012708,
      "loss": 12.2117,
      "step": 14613
    },
    {
      "epoch": 0.7957907060641332,
      "grad_norm": 0.6528327502031022,
      "learning_rate": 0.00013702454957475386,
      "loss": 12.2818,
      "step": 14614
    },
    {
      "epoch": 0.7958451600607163,
      "grad_norm": 0.6390175326589221,
      "learning_rate": 0.00013701635785147152,
      "loss": 12.097,
      "step": 14615
    },
    {
      "epoch": 0.7958996140572993,
      "grad_norm": 0.6133716093343833,
      "learning_rate": 0.0001370081658403438,
      "loss": 12.2751,
      "step": 14616
    },
    {
      "epoch": 0.7959540680538822,
      "grad_norm": 0.5489687858980747,
      "learning_rate": 0.00013699997354143438,
      "loss": 12.2439,
      "step": 14617
    },
    {
      "epoch": 0.7960085220504652,
      "grad_norm": 0.6197973559483165,
      "learning_rate": 0.00013699178095480697,
      "loss": 12.2906,
      "step": 14618
    },
    {
      "epoch": 0.7960629760470482,
      "grad_norm": 0.7095353639430027,
      "learning_rate": 0.0001369835880805252,
      "loss": 12.1462,
      "step": 14619
    },
    {
      "epoch": 0.7961174300436312,
      "grad_norm": 0.5870978258985311,
      "learning_rate": 0.0001369753949186529,
      "loss": 12.1074,
      "step": 14620
    },
    {
      "epoch": 0.7961718840402143,
      "grad_norm": 0.6594864495359288,
      "learning_rate": 0.00013696720146925373,
      "loss": 12.368,
      "step": 14621
    },
    {
      "epoch": 0.7962263380367973,
      "grad_norm": 0.6184783874708606,
      "learning_rate": 0.0001369590077323914,
      "loss": 12.3256,
      "step": 14622
    },
    {
      "epoch": 0.7962807920333803,
      "grad_norm": 0.5951694789565457,
      "learning_rate": 0.00013695081370812963,
      "loss": 12.1502,
      "step": 14623
    },
    {
      "epoch": 0.7963352460299633,
      "grad_norm": 0.607457031537557,
      "learning_rate": 0.00013694261939653214,
      "loss": 12.1754,
      "step": 14624
    },
    {
      "epoch": 0.7963897000265463,
      "grad_norm": 0.5191105385985078,
      "learning_rate": 0.0001369344247976626,
      "loss": 12.1142,
      "step": 14625
    },
    {
      "epoch": 0.7964441540231293,
      "grad_norm": 0.5516297677297756,
      "learning_rate": 0.0001369262299115848,
      "loss": 12.2765,
      "step": 14626
    },
    {
      "epoch": 0.7964986080197124,
      "grad_norm": 0.5515463986744881,
      "learning_rate": 0.00013691803473836247,
      "loss": 12.2683,
      "step": 14627
    },
    {
      "epoch": 0.7965530620162954,
      "grad_norm": 0.5851600829755839,
      "learning_rate": 0.00013690983927805932,
      "loss": 12.2109,
      "step": 14628
    },
    {
      "epoch": 0.7966075160128784,
      "grad_norm": 0.6260586049833458,
      "learning_rate": 0.00013690164353073904,
      "loss": 12.29,
      "step": 14629
    },
    {
      "epoch": 0.7966619700094614,
      "grad_norm": 0.5304233546419216,
      "learning_rate": 0.0001368934474964654,
      "loss": 12.2011,
      "step": 14630
    },
    {
      "epoch": 0.7967164240060444,
      "grad_norm": 0.6180773390482837,
      "learning_rate": 0.0001368852511753021,
      "loss": 12.2119,
      "step": 14631
    },
    {
      "epoch": 0.7967708780026274,
      "grad_norm": 0.5630688489731834,
      "learning_rate": 0.00013687705456731295,
      "loss": 12.2902,
      "step": 14632
    },
    {
      "epoch": 0.7968253319992105,
      "grad_norm": 0.6090926735929871,
      "learning_rate": 0.00013686885767256164,
      "loss": 12.1717,
      "step": 14633
    },
    {
      "epoch": 0.7968797859957935,
      "grad_norm": 0.5872552028357287,
      "learning_rate": 0.00013686066049111185,
      "loss": 12.269,
      "step": 14634
    },
    {
      "epoch": 0.7969342399923764,
      "grad_norm": 0.5742787047807977,
      "learning_rate": 0.00013685246302302744,
      "loss": 12.2413,
      "step": 14635
    },
    {
      "epoch": 0.7969886939889594,
      "grad_norm": 0.6247212389621019,
      "learning_rate": 0.00013684426526837205,
      "loss": 12.2067,
      "step": 14636
    },
    {
      "epoch": 0.7970431479855424,
      "grad_norm": 0.5868775930778958,
      "learning_rate": 0.0001368360672272095,
      "loss": 12.2478,
      "step": 14637
    },
    {
      "epoch": 0.7970976019821254,
      "grad_norm": 0.5846150678878385,
      "learning_rate": 0.00013682786889960354,
      "loss": 12.2411,
      "step": 14638
    },
    {
      "epoch": 0.7971520559787085,
      "grad_norm": 0.5843841367905177,
      "learning_rate": 0.00013681967028561785,
      "loss": 12.2898,
      "step": 14639
    },
    {
      "epoch": 0.7972065099752915,
      "grad_norm": 0.5868475342754406,
      "learning_rate": 0.00013681147138531625,
      "loss": 12.3007,
      "step": 14640
    },
    {
      "epoch": 0.7972609639718745,
      "grad_norm": 0.583273108266093,
      "learning_rate": 0.00013680327219876248,
      "loss": 12.315,
      "step": 14641
    },
    {
      "epoch": 0.7973154179684575,
      "grad_norm": 0.600712103579576,
      "learning_rate": 0.00013679507272602027,
      "loss": 12.2416,
      "step": 14642
    },
    {
      "epoch": 0.7973698719650405,
      "grad_norm": 0.5903813153092471,
      "learning_rate": 0.0001367868729671534,
      "loss": 12.1484,
      "step": 14643
    },
    {
      "epoch": 0.7974243259616235,
      "grad_norm": 0.6057061535598088,
      "learning_rate": 0.00013677867292222567,
      "loss": 12.2062,
      "step": 14644
    },
    {
      "epoch": 0.7974787799582066,
      "grad_norm": 0.5382834472760614,
      "learning_rate": 0.00013677047259130082,
      "loss": 12.1268,
      "step": 14645
    },
    {
      "epoch": 0.7975332339547896,
      "grad_norm": 0.6063616100416661,
      "learning_rate": 0.00013676227197444258,
      "loss": 12.2374,
      "step": 14646
    },
    {
      "epoch": 0.7975876879513726,
      "grad_norm": 0.5621407926219465,
      "learning_rate": 0.00013675407107171473,
      "loss": 12.2684,
      "step": 14647
    },
    {
      "epoch": 0.7976421419479556,
      "grad_norm": 0.5651809829118901,
      "learning_rate": 0.00013674586988318108,
      "loss": 12.3008,
      "step": 14648
    },
    {
      "epoch": 0.7976965959445386,
      "grad_norm": 0.7892357277511804,
      "learning_rate": 0.0001367376684089054,
      "loss": 12.313,
      "step": 14649
    },
    {
      "epoch": 0.7977510499411217,
      "grad_norm": 0.6506794636249186,
      "learning_rate": 0.00013672946664895145,
      "loss": 12.4271,
      "step": 14650
    },
    {
      "epoch": 0.7978055039377047,
      "grad_norm": 0.5041891292585133,
      "learning_rate": 0.00013672126460338298,
      "loss": 12.2156,
      "step": 14651
    },
    {
      "epoch": 0.7978599579342877,
      "grad_norm": 0.7401483260152929,
      "learning_rate": 0.00013671306227226385,
      "loss": 12.2375,
      "step": 14652
    },
    {
      "epoch": 0.7979144119308706,
      "grad_norm": 0.6222369533899695,
      "learning_rate": 0.00013670485965565772,
      "loss": 12.276,
      "step": 14653
    },
    {
      "epoch": 0.7979688659274536,
      "grad_norm": 0.6692280187649381,
      "learning_rate": 0.00013669665675362848,
      "loss": 12.3599,
      "step": 14654
    },
    {
      "epoch": 0.7980233199240366,
      "grad_norm": 0.5884085203012612,
      "learning_rate": 0.0001366884535662399,
      "loss": 12.4401,
      "step": 14655
    },
    {
      "epoch": 0.7980777739206197,
      "grad_norm": 0.5589894629310427,
      "learning_rate": 0.00013668025009355573,
      "loss": 12.1091,
      "step": 14656
    },
    {
      "epoch": 0.7981322279172027,
      "grad_norm": 0.5326014670977347,
      "learning_rate": 0.0001366720463356398,
      "loss": 12.1856,
      "step": 14657
    },
    {
      "epoch": 0.7981866819137857,
      "grad_norm": 0.6594920539581905,
      "learning_rate": 0.00013666384229255585,
      "loss": 12.336,
      "step": 14658
    },
    {
      "epoch": 0.7982411359103687,
      "grad_norm": 0.5766616140343678,
      "learning_rate": 0.00013665563796436776,
      "loss": 12.111,
      "step": 14659
    },
    {
      "epoch": 0.7982955899069517,
      "grad_norm": 0.6231565003777191,
      "learning_rate": 0.00013664743335113926,
      "loss": 12.3686,
      "step": 14660
    },
    {
      "epoch": 0.7983500439035347,
      "grad_norm": 0.6206882350344767,
      "learning_rate": 0.00013663922845293417,
      "loss": 12.1138,
      "step": 14661
    },
    {
      "epoch": 0.7984044979001178,
      "grad_norm": 0.5864199324503923,
      "learning_rate": 0.00013663102326981632,
      "loss": 12.2635,
      "step": 14662
    },
    {
      "epoch": 0.7984589518967008,
      "grad_norm": 0.5555180848908318,
      "learning_rate": 0.00013662281780184947,
      "loss": 12.1993,
      "step": 14663
    },
    {
      "epoch": 0.7985134058932838,
      "grad_norm": 0.5393908024567289,
      "learning_rate": 0.00013661461204909746,
      "loss": 12.2018,
      "step": 14664
    },
    {
      "epoch": 0.7985678598898668,
      "grad_norm": 0.5869442261183992,
      "learning_rate": 0.00013660640601162406,
      "loss": 12.3016,
      "step": 14665
    },
    {
      "epoch": 0.7986223138864498,
      "grad_norm": 0.5705150655125303,
      "learning_rate": 0.0001365981996894931,
      "loss": 12.1733,
      "step": 14666
    },
    {
      "epoch": 0.7986767678830328,
      "grad_norm": 0.5564124217432502,
      "learning_rate": 0.00013658999308276845,
      "loss": 12.2509,
      "step": 14667
    },
    {
      "epoch": 0.7987312218796159,
      "grad_norm": 0.543176948218349,
      "learning_rate": 0.00013658178619151384,
      "loss": 12.1485,
      "step": 14668
    },
    {
      "epoch": 0.7987856758761989,
      "grad_norm": 0.5580168309808128,
      "learning_rate": 0.00013657357901579315,
      "loss": 12.1323,
      "step": 14669
    },
    {
      "epoch": 0.7988401298727819,
      "grad_norm": 0.5844524466135493,
      "learning_rate": 0.00013656537155567016,
      "loss": 12.1079,
      "step": 14670
    },
    {
      "epoch": 0.7988945838693648,
      "grad_norm": 0.5497642901168213,
      "learning_rate": 0.0001365571638112087,
      "loss": 12.2781,
      "step": 14671
    },
    {
      "epoch": 0.7989490378659478,
      "grad_norm": 0.5574194895445754,
      "learning_rate": 0.00013654895578247262,
      "loss": 12.2041,
      "step": 14672
    },
    {
      "epoch": 0.7990034918625308,
      "grad_norm": 0.5767024085857305,
      "learning_rate": 0.00013654074746952572,
      "loss": 12.2425,
      "step": 14673
    },
    {
      "epoch": 0.7990579458591139,
      "grad_norm": 0.6576026837736271,
      "learning_rate": 0.00013653253887243184,
      "loss": 12.3398,
      "step": 14674
    },
    {
      "epoch": 0.7991123998556969,
      "grad_norm": 0.6085438305694729,
      "learning_rate": 0.00013652432999125484,
      "loss": 12.229,
      "step": 14675
    },
    {
      "epoch": 0.7991668538522799,
      "grad_norm": 0.5624586386412984,
      "learning_rate": 0.0001365161208260585,
      "loss": 12.0903,
      "step": 14676
    },
    {
      "epoch": 0.7992213078488629,
      "grad_norm": 0.5549898217069293,
      "learning_rate": 0.00013650791137690668,
      "loss": 12.1677,
      "step": 14677
    },
    {
      "epoch": 0.7992757618454459,
      "grad_norm": 0.5676483063775526,
      "learning_rate": 0.00013649970164386323,
      "loss": 12.2304,
      "step": 14678
    },
    {
      "epoch": 0.7993302158420289,
      "grad_norm": 0.5308701782521362,
      "learning_rate": 0.00013649149162699197,
      "loss": 12.261,
      "step": 14679
    },
    {
      "epoch": 0.799384669838612,
      "grad_norm": 0.6357462521824179,
      "learning_rate": 0.00013648328132635676,
      "loss": 12.3132,
      "step": 14680
    },
    {
      "epoch": 0.799439123835195,
      "grad_norm": 0.5580459839955783,
      "learning_rate": 0.00013647507074202142,
      "loss": 12.2095,
      "step": 14681
    },
    {
      "epoch": 0.799493577831778,
      "grad_norm": 0.5323776944845773,
      "learning_rate": 0.0001364668598740498,
      "loss": 12.259,
      "step": 14682
    },
    {
      "epoch": 0.799548031828361,
      "grad_norm": 0.5452627919182734,
      "learning_rate": 0.0001364586487225058,
      "loss": 12.1575,
      "step": 14683
    },
    {
      "epoch": 0.799602485824944,
      "grad_norm": 0.6235269652424776,
      "learning_rate": 0.00013645043728745325,
      "loss": 12.2087,
      "step": 14684
    },
    {
      "epoch": 0.7996569398215271,
      "grad_norm": 0.6513695487707094,
      "learning_rate": 0.00013644222556895592,
      "loss": 12.3251,
      "step": 14685
    },
    {
      "epoch": 0.7997113938181101,
      "grad_norm": 0.541478799476608,
      "learning_rate": 0.00013643401356707777,
      "loss": 12.2063,
      "step": 14686
    },
    {
      "epoch": 0.7997658478146931,
      "grad_norm": 0.6221379479706041,
      "learning_rate": 0.00013642580128188264,
      "loss": 12.2197,
      "step": 14687
    },
    {
      "epoch": 0.7998203018112761,
      "grad_norm": 0.5977131685243543,
      "learning_rate": 0.00013641758871343432,
      "loss": 12.2846,
      "step": 14688
    },
    {
      "epoch": 0.799874755807859,
      "grad_norm": 0.5986729314260992,
      "learning_rate": 0.00013640937586179678,
      "loss": 12.2008,
      "step": 14689
    },
    {
      "epoch": 0.799929209804442,
      "grad_norm": 0.6037694902876016,
      "learning_rate": 0.0001364011627270338,
      "loss": 12.3244,
      "step": 14690
    },
    {
      "epoch": 0.7999836638010251,
      "grad_norm": 0.5716463867704794,
      "learning_rate": 0.00013639294930920925,
      "loss": 12.2065,
      "step": 14691
    },
    {
      "epoch": 0.8000381177976081,
      "grad_norm": 0.6101374572745445,
      "learning_rate": 0.00013638473560838706,
      "loss": 12.2074,
      "step": 14692
    },
    {
      "epoch": 0.8000925717941911,
      "grad_norm": 0.5804774150756093,
      "learning_rate": 0.00013637652162463103,
      "loss": 12.1971,
      "step": 14693
    },
    {
      "epoch": 0.8001470257907741,
      "grad_norm": 0.5712723197657905,
      "learning_rate": 0.0001363683073580051,
      "loss": 12.1784,
      "step": 14694
    },
    {
      "epoch": 0.8002014797873571,
      "grad_norm": 0.5736267925533555,
      "learning_rate": 0.0001363600928085731,
      "loss": 12.1645,
      "step": 14695
    },
    {
      "epoch": 0.8002559337839401,
      "grad_norm": 0.5704183696042504,
      "learning_rate": 0.0001363518779763989,
      "loss": 12.2962,
      "step": 14696
    },
    {
      "epoch": 0.8003103877805232,
      "grad_norm": 0.5422945074682746,
      "learning_rate": 0.00013634366286154642,
      "loss": 12.2148,
      "step": 14697
    },
    {
      "epoch": 0.8003648417771062,
      "grad_norm": 0.538070683148117,
      "learning_rate": 0.00013633544746407953,
      "loss": 12.2103,
      "step": 14698
    },
    {
      "epoch": 0.8004192957736892,
      "grad_norm": 0.5463898898573379,
      "learning_rate": 0.00013632723178406213,
      "loss": 12.1015,
      "step": 14699
    },
    {
      "epoch": 0.8004737497702722,
      "grad_norm": 0.6780580468037041,
      "learning_rate": 0.00013631901582155807,
      "loss": 12.2762,
      "step": 14700
    },
    {
      "epoch": 0.8005282037668552,
      "grad_norm": 0.5277520574048612,
      "learning_rate": 0.00013631079957663122,
      "loss": 12.2817,
      "step": 14701
    },
    {
      "epoch": 0.8005826577634382,
      "grad_norm": 0.5393645235090414,
      "learning_rate": 0.00013630258304934552,
      "loss": 12.146,
      "step": 14702
    },
    {
      "epoch": 0.8006371117600213,
      "grad_norm": 0.5997930595758416,
      "learning_rate": 0.00013629436623976483,
      "loss": 12.2559,
      "step": 14703
    },
    {
      "epoch": 0.8006915657566043,
      "grad_norm": 0.5373163929300773,
      "learning_rate": 0.0001362861491479531,
      "loss": 12.1077,
      "step": 14704
    },
    {
      "epoch": 0.8007460197531873,
      "grad_norm": 0.5806005994554888,
      "learning_rate": 0.00013627793177397416,
      "loss": 12.3336,
      "step": 14705
    },
    {
      "epoch": 0.8008004737497703,
      "grad_norm": 0.6518520429407361,
      "learning_rate": 0.00013626971411789197,
      "loss": 12.2488,
      "step": 14706
    },
    {
      "epoch": 0.8008549277463533,
      "grad_norm": 0.5552547751323605,
      "learning_rate": 0.00013626149617977035,
      "loss": 12.2116,
      "step": 14707
    },
    {
      "epoch": 0.8009093817429362,
      "grad_norm": 0.551120271148661,
      "learning_rate": 0.00013625327795967326,
      "loss": 12.2319,
      "step": 14708
    },
    {
      "epoch": 0.8009638357395193,
      "grad_norm": 0.5855379834522427,
      "learning_rate": 0.00013624505945766466,
      "loss": 12.1803,
      "step": 14709
    },
    {
      "epoch": 0.8010182897361023,
      "grad_norm": 0.5753752303480557,
      "learning_rate": 0.00013623684067380835,
      "loss": 12.2338,
      "step": 14710
    },
    {
      "epoch": 0.8010727437326853,
      "grad_norm": 0.5330703433849421,
      "learning_rate": 0.0001362286216081683,
      "loss": 12.3258,
      "step": 14711
    },
    {
      "epoch": 0.8011271977292683,
      "grad_norm": 0.5619976654850436,
      "learning_rate": 0.00013622040226080842,
      "loss": 12.2137,
      "step": 14712
    },
    {
      "epoch": 0.8011816517258513,
      "grad_norm": 0.600438451497241,
      "learning_rate": 0.00013621218263179259,
      "loss": 12.1938,
      "step": 14713
    },
    {
      "epoch": 0.8012361057224343,
      "grad_norm": 0.6901686333060042,
      "learning_rate": 0.0001362039627211848,
      "loss": 12.058,
      "step": 14714
    },
    {
      "epoch": 0.8012905597190174,
      "grad_norm": 0.5017285782650442,
      "learning_rate": 0.0001361957425290489,
      "loss": 12.2392,
      "step": 14715
    },
    {
      "epoch": 0.8013450137156004,
      "grad_norm": 0.5516965967674011,
      "learning_rate": 0.00013618752205544885,
      "loss": 12.138,
      "step": 14716
    },
    {
      "epoch": 0.8013994677121834,
      "grad_norm": 0.6359579853789539,
      "learning_rate": 0.00013617930130044854,
      "loss": 12.2751,
      "step": 14717
    },
    {
      "epoch": 0.8014539217087664,
      "grad_norm": 0.6236073792954583,
      "learning_rate": 0.0001361710802641119,
      "loss": 12.131,
      "step": 14718
    },
    {
      "epoch": 0.8015083757053494,
      "grad_norm": 0.5946363293936324,
      "learning_rate": 0.0001361628589465029,
      "loss": 12.2467,
      "step": 14719
    },
    {
      "epoch": 0.8015628297019325,
      "grad_norm": 0.5702754809065214,
      "learning_rate": 0.00013615463734768546,
      "loss": 12.2673,
      "step": 14720
    },
    {
      "epoch": 0.8016172836985155,
      "grad_norm": 0.5655799635243692,
      "learning_rate": 0.00013614641546772348,
      "loss": 12.1912,
      "step": 14721
    },
    {
      "epoch": 0.8016717376950985,
      "grad_norm": 0.5317017544142133,
      "learning_rate": 0.0001361381933066809,
      "loss": 12.2049,
      "step": 14722
    },
    {
      "epoch": 0.8017261916916815,
      "grad_norm": 0.6350850966842021,
      "learning_rate": 0.00013612997086462169,
      "loss": 12.302,
      "step": 14723
    },
    {
      "epoch": 0.8017806456882645,
      "grad_norm": 0.6144874801838794,
      "learning_rate": 0.00013612174814160976,
      "loss": 12.2868,
      "step": 14724
    },
    {
      "epoch": 0.8018350996848475,
      "grad_norm": 0.5321833507643821,
      "learning_rate": 0.00013611352513770905,
      "loss": 12.0525,
      "step": 14725
    },
    {
      "epoch": 0.8018895536814306,
      "grad_norm": 0.5718843327099883,
      "learning_rate": 0.00013610530185298353,
      "loss": 12.1852,
      "step": 14726
    },
    {
      "epoch": 0.8019440076780135,
      "grad_norm": 0.6070868198744772,
      "learning_rate": 0.0001360970782874971,
      "loss": 12.1785,
      "step": 14727
    },
    {
      "epoch": 0.8019984616745965,
      "grad_norm": 0.5306665602802828,
      "learning_rate": 0.00013608885444131374,
      "loss": 12.3216,
      "step": 14728
    },
    {
      "epoch": 0.8020529156711795,
      "grad_norm": 0.6288825474118335,
      "learning_rate": 0.0001360806303144974,
      "loss": 12.3004,
      "step": 14729
    },
    {
      "epoch": 0.8021073696677625,
      "grad_norm": 0.5613030264008502,
      "learning_rate": 0.00013607240590711206,
      "loss": 12.2871,
      "step": 14730
    },
    {
      "epoch": 0.8021618236643455,
      "grad_norm": 0.5322723276921058,
      "learning_rate": 0.0001360641812192216,
      "loss": 12.1806,
      "step": 14731
    },
    {
      "epoch": 0.8022162776609286,
      "grad_norm": 0.584965786831096,
      "learning_rate": 0.00013605595625089005,
      "loss": 12.1354,
      "step": 14732
    },
    {
      "epoch": 0.8022707316575116,
      "grad_norm": 0.561919891996372,
      "learning_rate": 0.00013604773100218132,
      "loss": 12.2726,
      "step": 14733
    },
    {
      "epoch": 0.8023251856540946,
      "grad_norm": 0.697282815185284,
      "learning_rate": 0.0001360395054731594,
      "loss": 12.3293,
      "step": 14734
    },
    {
      "epoch": 0.8023796396506776,
      "grad_norm": 0.5649339931331062,
      "learning_rate": 0.0001360312796638882,
      "loss": 12.0921,
      "step": 14735
    },
    {
      "epoch": 0.8024340936472606,
      "grad_norm": 0.5525623322911426,
      "learning_rate": 0.0001360230535744318,
      "loss": 12.2729,
      "step": 14736
    },
    {
      "epoch": 0.8024885476438436,
      "grad_norm": 0.5735865225673554,
      "learning_rate": 0.00013601482720485404,
      "loss": 12.06,
      "step": 14737
    },
    {
      "epoch": 0.8025430016404267,
      "grad_norm": 0.6089480738355258,
      "learning_rate": 0.00013600660055521896,
      "loss": 12.2962,
      "step": 14738
    },
    {
      "epoch": 0.8025974556370097,
      "grad_norm": 0.6060095300813723,
      "learning_rate": 0.00013599837362559053,
      "loss": 12.1462,
      "step": 14739
    },
    {
      "epoch": 0.8026519096335927,
      "grad_norm": 0.5992632558396295,
      "learning_rate": 0.0001359901464160327,
      "loss": 12.3345,
      "step": 14740
    },
    {
      "epoch": 0.8027063636301757,
      "grad_norm": 0.646098819998647,
      "learning_rate": 0.00013598191892660942,
      "loss": 12.2307,
      "step": 14741
    },
    {
      "epoch": 0.8027608176267587,
      "grad_norm": 0.590717453674496,
      "learning_rate": 0.00013597369115738475,
      "loss": 12.2248,
      "step": 14742
    },
    {
      "epoch": 0.8028152716233417,
      "grad_norm": 0.5790171924040252,
      "learning_rate": 0.00013596546310842259,
      "loss": 12.3083,
      "step": 14743
    },
    {
      "epoch": 0.8028697256199248,
      "grad_norm": 0.5690108975248733,
      "learning_rate": 0.000135957234779787,
      "loss": 12.1001,
      "step": 14744
    },
    {
      "epoch": 0.8029241796165077,
      "grad_norm": 0.6598419449912617,
      "learning_rate": 0.00013594900617154188,
      "loss": 12.2677,
      "step": 14745
    },
    {
      "epoch": 0.8029786336130907,
      "grad_norm": 0.6457502835703006,
      "learning_rate": 0.00013594077728375128,
      "loss": 12.2013,
      "step": 14746
    },
    {
      "epoch": 0.8030330876096737,
      "grad_norm": 0.5676104527087016,
      "learning_rate": 0.00013593254811647916,
      "loss": 12.2458,
      "step": 14747
    },
    {
      "epoch": 0.8030875416062567,
      "grad_norm": 0.5815299697852133,
      "learning_rate": 0.00013592431866978955,
      "loss": 12.0223,
      "step": 14748
    },
    {
      "epoch": 0.8031419956028398,
      "grad_norm": 0.635907774790155,
      "learning_rate": 0.00013591608894374642,
      "loss": 12.1705,
      "step": 14749
    },
    {
      "epoch": 0.8031964495994228,
      "grad_norm": 0.632519027187312,
      "learning_rate": 0.00013590785893841372,
      "loss": 12.2368,
      "step": 14750
    },
    {
      "epoch": 0.8032509035960058,
      "grad_norm": 0.5971202957757051,
      "learning_rate": 0.00013589962865385546,
      "loss": 12.2808,
      "step": 14751
    },
    {
      "epoch": 0.8033053575925888,
      "grad_norm": 0.6267980859454244,
      "learning_rate": 0.00013589139809013572,
      "loss": 12.3181,
      "step": 14752
    },
    {
      "epoch": 0.8033598115891718,
      "grad_norm": 0.5696313738145158,
      "learning_rate": 0.00013588316724731842,
      "loss": 12.1311,
      "step": 14753
    },
    {
      "epoch": 0.8034142655857548,
      "grad_norm": 0.5387721506325479,
      "learning_rate": 0.00013587493612546764,
      "loss": 12.2173,
      "step": 14754
    },
    {
      "epoch": 0.8034687195823379,
      "grad_norm": 0.5487563950306569,
      "learning_rate": 0.00013586670472464732,
      "loss": 12.2189,
      "step": 14755
    },
    {
      "epoch": 0.8035231735789209,
      "grad_norm": 0.6023344704593654,
      "learning_rate": 0.00013585847304492144,
      "loss": 12.2929,
      "step": 14756
    },
    {
      "epoch": 0.8035776275755039,
      "grad_norm": 0.68285705582827,
      "learning_rate": 0.00013585024108635408,
      "loss": 12.2312,
      "step": 14757
    },
    {
      "epoch": 0.8036320815720869,
      "grad_norm": 0.5828683432730125,
      "learning_rate": 0.00013584200884900926,
      "loss": 12.166,
      "step": 14758
    },
    {
      "epoch": 0.8036865355686699,
      "grad_norm": 0.598886249716344,
      "learning_rate": 0.00013583377633295097,
      "loss": 12.272,
      "step": 14759
    },
    {
      "epoch": 0.8037409895652529,
      "grad_norm": 0.574350861635556,
      "learning_rate": 0.00013582554353824323,
      "loss": 12.2154,
      "step": 14760
    },
    {
      "epoch": 0.803795443561836,
      "grad_norm": 0.531639449220211,
      "learning_rate": 0.00013581731046495004,
      "loss": 12.0654,
      "step": 14761
    },
    {
      "epoch": 0.803849897558419,
      "grad_norm": 0.5760371469392463,
      "learning_rate": 0.00013580907711313543,
      "loss": 12.1291,
      "step": 14762
    },
    {
      "epoch": 0.803904351555002,
      "grad_norm": 0.6355944627676456,
      "learning_rate": 0.00013580084348286344,
      "loss": 12.2196,
      "step": 14763
    },
    {
      "epoch": 0.8039588055515849,
      "grad_norm": 0.6236060806198317,
      "learning_rate": 0.00013579260957419812,
      "loss": 12.1445,
      "step": 14764
    },
    {
      "epoch": 0.8040132595481679,
      "grad_norm": 0.5654738248795809,
      "learning_rate": 0.0001357843753872034,
      "loss": 12.3111,
      "step": 14765
    },
    {
      "epoch": 0.8040677135447509,
      "grad_norm": 0.6123689053793384,
      "learning_rate": 0.0001357761409219434,
      "loss": 12.2449,
      "step": 14766
    },
    {
      "epoch": 0.804122167541334,
      "grad_norm": 0.6004670492334654,
      "learning_rate": 0.00013576790617848215,
      "loss": 12.2578,
      "step": 14767
    },
    {
      "epoch": 0.804176621537917,
      "grad_norm": 0.5645618267114271,
      "learning_rate": 0.00013575967115688365,
      "loss": 12.2648,
      "step": 14768
    },
    {
      "epoch": 0.8042310755345,
      "grad_norm": 0.5500008223839482,
      "learning_rate": 0.00013575143585721196,
      "loss": 12.2283,
      "step": 14769
    },
    {
      "epoch": 0.804285529531083,
      "grad_norm": 0.6051696895269583,
      "learning_rate": 0.0001357432002795311,
      "loss": 12.3082,
      "step": 14770
    },
    {
      "epoch": 0.804339983527666,
      "grad_norm": 0.5952390426204499,
      "learning_rate": 0.00013573496442390511,
      "loss": 12.1966,
      "step": 14771
    },
    {
      "epoch": 0.804394437524249,
      "grad_norm": 0.6088605323849096,
      "learning_rate": 0.00013572672829039806,
      "loss": 12.3442,
      "step": 14772
    },
    {
      "epoch": 0.8044488915208321,
      "grad_norm": 0.5993152028924302,
      "learning_rate": 0.00013571849187907396,
      "loss": 12.1041,
      "step": 14773
    },
    {
      "epoch": 0.8045033455174151,
      "grad_norm": 0.5218338456393441,
      "learning_rate": 0.0001357102551899969,
      "loss": 12.1031,
      "step": 14774
    },
    {
      "epoch": 0.8045577995139981,
      "grad_norm": 0.609926418028468,
      "learning_rate": 0.0001357020182232309,
      "loss": 12.3269,
      "step": 14775
    },
    {
      "epoch": 0.8046122535105811,
      "grad_norm": 0.6470069008285002,
      "learning_rate": 0.00013569378097884,
      "loss": 12.2054,
      "step": 14776
    },
    {
      "epoch": 0.8046667075071641,
      "grad_norm": 0.5562287041283757,
      "learning_rate": 0.0001356855434568883,
      "loss": 12.171,
      "step": 14777
    },
    {
      "epoch": 0.8047211615037471,
      "grad_norm": 0.5383874909013538,
      "learning_rate": 0.00013567730565743982,
      "loss": 12.2667,
      "step": 14778
    },
    {
      "epoch": 0.8047756155003302,
      "grad_norm": 0.5781141534823001,
      "learning_rate": 0.00013566906758055863,
      "loss": 12.0728,
      "step": 14779
    },
    {
      "epoch": 0.8048300694969132,
      "grad_norm": 0.5767436399068844,
      "learning_rate": 0.00013566082922630878,
      "loss": 12.181,
      "step": 14780
    },
    {
      "epoch": 0.8048845234934962,
      "grad_norm": 0.6601181379244292,
      "learning_rate": 0.00013565259059475436,
      "loss": 12.1892,
      "step": 14781
    },
    {
      "epoch": 0.8049389774900791,
      "grad_norm": 0.5180753674094193,
      "learning_rate": 0.00013564435168595938,
      "loss": 12.2281,
      "step": 14782
    },
    {
      "epoch": 0.8049934314866621,
      "grad_norm": 0.5461455256382294,
      "learning_rate": 0.000135636112499988,
      "loss": 12.1856,
      "step": 14783
    },
    {
      "epoch": 0.8050478854832452,
      "grad_norm": 0.5651300856132475,
      "learning_rate": 0.0001356278730369042,
      "loss": 12.1808,
      "step": 14784
    },
    {
      "epoch": 0.8051023394798282,
      "grad_norm": 0.5762346524051141,
      "learning_rate": 0.00013561963329677208,
      "loss": 12.0586,
      "step": 14785
    },
    {
      "epoch": 0.8051567934764112,
      "grad_norm": 0.5566516236897829,
      "learning_rate": 0.0001356113932796557,
      "loss": 12.2405,
      "step": 14786
    },
    {
      "epoch": 0.8052112474729942,
      "grad_norm": 0.5272763168734077,
      "learning_rate": 0.0001356031529856192,
      "loss": 12.2018,
      "step": 14787
    },
    {
      "epoch": 0.8052657014695772,
      "grad_norm": 0.5277452420280747,
      "learning_rate": 0.00013559491241472657,
      "loss": 12.1506,
      "step": 14788
    },
    {
      "epoch": 0.8053201554661602,
      "grad_norm": 0.5549414438819789,
      "learning_rate": 0.00013558667156704195,
      "loss": 12.2929,
      "step": 14789
    },
    {
      "epoch": 0.8053746094627433,
      "grad_norm": 0.573943349349434,
      "learning_rate": 0.00013557843044262942,
      "loss": 12.2114,
      "step": 14790
    },
    {
      "epoch": 0.8054290634593263,
      "grad_norm": 0.631727434172093,
      "learning_rate": 0.000135570189041553,
      "loss": 12.0948,
      "step": 14791
    },
    {
      "epoch": 0.8054835174559093,
      "grad_norm": 0.5886082743780373,
      "learning_rate": 0.00013556194736387688,
      "loss": 12.2512,
      "step": 14792
    },
    {
      "epoch": 0.8055379714524923,
      "grad_norm": 0.5156967125767348,
      "learning_rate": 0.00013555370540966507,
      "loss": 12.1234,
      "step": 14793
    },
    {
      "epoch": 0.8055924254490753,
      "grad_norm": 0.5906966524320194,
      "learning_rate": 0.00013554546317898168,
      "loss": 12.1704,
      "step": 14794
    },
    {
      "epoch": 0.8056468794456583,
      "grad_norm": 0.6407555076229271,
      "learning_rate": 0.00013553722067189084,
      "loss": 12.1559,
      "step": 14795
    },
    {
      "epoch": 0.8057013334422414,
      "grad_norm": 0.7499130923304811,
      "learning_rate": 0.00013552897788845656,
      "loss": 12.1958,
      "step": 14796
    },
    {
      "epoch": 0.8057557874388244,
      "grad_norm": 0.5890173368065736,
      "learning_rate": 0.00013552073482874302,
      "loss": 12.1924,
      "step": 14797
    },
    {
      "epoch": 0.8058102414354074,
      "grad_norm": 0.5424754999109495,
      "learning_rate": 0.0001355124914928143,
      "loss": 12.1568,
      "step": 14798
    },
    {
      "epoch": 0.8058646954319904,
      "grad_norm": 0.5570372008104348,
      "learning_rate": 0.00013550424788073446,
      "loss": 12.1763,
      "step": 14799
    },
    {
      "epoch": 0.8059191494285733,
      "grad_norm": 0.6144758166704617,
      "learning_rate": 0.00013549600399256762,
      "loss": 12.2377,
      "step": 14800
    },
    {
      "epoch": 0.8059736034251563,
      "grad_norm": 0.5475933792469335,
      "learning_rate": 0.00013548775982837795,
      "loss": 12.1686,
      "step": 14801
    },
    {
      "epoch": 0.8060280574217394,
      "grad_norm": 0.532083181554464,
      "learning_rate": 0.0001354795153882295,
      "loss": 12.1992,
      "step": 14802
    },
    {
      "epoch": 0.8060825114183224,
      "grad_norm": 0.569394156356611,
      "learning_rate": 0.00013547127067218637,
      "loss": 12.2908,
      "step": 14803
    },
    {
      "epoch": 0.8061369654149054,
      "grad_norm": 0.5876466479875222,
      "learning_rate": 0.0001354630256803127,
      "loss": 12.376,
      "step": 14804
    },
    {
      "epoch": 0.8061914194114884,
      "grad_norm": 0.6446041649581469,
      "learning_rate": 0.00013545478041267258,
      "loss": 12.3708,
      "step": 14805
    },
    {
      "epoch": 0.8062458734080714,
      "grad_norm": 0.5745533624203241,
      "learning_rate": 0.00013544653486933017,
      "loss": 12.2191,
      "step": 14806
    },
    {
      "epoch": 0.8063003274046544,
      "grad_norm": 0.5876921213626856,
      "learning_rate": 0.00013543828905034953,
      "loss": 12.0911,
      "step": 14807
    },
    {
      "epoch": 0.8063547814012375,
      "grad_norm": 0.61122104079932,
      "learning_rate": 0.00013543004295579481,
      "loss": 12.3174,
      "step": 14808
    },
    {
      "epoch": 0.8064092353978205,
      "grad_norm": 0.5693585541111951,
      "learning_rate": 0.00013542179658573018,
      "loss": 12.3254,
      "step": 14809
    },
    {
      "epoch": 0.8064636893944035,
      "grad_norm": 0.6509018058521152,
      "learning_rate": 0.00013541354994021972,
      "loss": 12.1927,
      "step": 14810
    },
    {
      "epoch": 0.8065181433909865,
      "grad_norm": 0.524792413116897,
      "learning_rate": 0.0001354053030193275,
      "loss": 12.1193,
      "step": 14811
    },
    {
      "epoch": 0.8065725973875695,
      "grad_norm": 0.6622307559286874,
      "learning_rate": 0.0001353970558231177,
      "loss": 12.1634,
      "step": 14812
    },
    {
      "epoch": 0.8066270513841525,
      "grad_norm": 0.6240020101914101,
      "learning_rate": 0.00013538880835165453,
      "loss": 12.1128,
      "step": 14813
    },
    {
      "epoch": 0.8066815053807356,
      "grad_norm": 0.6213612845211237,
      "learning_rate": 0.000135380560605002,
      "loss": 12.2167,
      "step": 14814
    },
    {
      "epoch": 0.8067359593773186,
      "grad_norm": 0.5891375759818152,
      "learning_rate": 0.00013537231258322434,
      "loss": 12.2565,
      "step": 14815
    },
    {
      "epoch": 0.8067904133739016,
      "grad_norm": 0.6213675246525658,
      "learning_rate": 0.00013536406428638558,
      "loss": 12.1733,
      "step": 14816
    },
    {
      "epoch": 0.8068448673704846,
      "grad_norm": 0.5792737305087725,
      "learning_rate": 0.00013535581571454995,
      "loss": 12.2326,
      "step": 14817
    },
    {
      "epoch": 0.8068993213670675,
      "grad_norm": 0.5503879796337302,
      "learning_rate": 0.00013534756686778157,
      "loss": 12.1473,
      "step": 14818
    },
    {
      "epoch": 0.8069537753636506,
      "grad_norm": 0.5875674071397632,
      "learning_rate": 0.0001353393177461446,
      "loss": 12.3264,
      "step": 14819
    },
    {
      "epoch": 0.8070082293602336,
      "grad_norm": 0.6658764117400607,
      "learning_rate": 0.00013533106834970319,
      "loss": 12.3356,
      "step": 14820
    },
    {
      "epoch": 0.8070626833568166,
      "grad_norm": 0.5598078870331447,
      "learning_rate": 0.00013532281867852144,
      "loss": 12.056,
      "step": 14821
    },
    {
      "epoch": 0.8071171373533996,
      "grad_norm": 0.5427067549451772,
      "learning_rate": 0.00013531456873266352,
      "loss": 12.2513,
      "step": 14822
    },
    {
      "epoch": 0.8071715913499826,
      "grad_norm": 0.6039181316269996,
      "learning_rate": 0.00013530631851219358,
      "loss": 12.104,
      "step": 14823
    },
    {
      "epoch": 0.8072260453465656,
      "grad_norm": 0.5514512877987486,
      "learning_rate": 0.00013529806801717583,
      "loss": 12.0721,
      "step": 14824
    },
    {
      "epoch": 0.8072804993431487,
      "grad_norm": 0.5026969335243999,
      "learning_rate": 0.00013528981724767434,
      "loss": 12.1453,
      "step": 14825
    },
    {
      "epoch": 0.8073349533397317,
      "grad_norm": 0.6399354209506974,
      "learning_rate": 0.00013528156620375335,
      "loss": 12.3137,
      "step": 14826
    },
    {
      "epoch": 0.8073894073363147,
      "grad_norm": 0.5228089435675359,
      "learning_rate": 0.00013527331488547698,
      "loss": 12.1478,
      "step": 14827
    },
    {
      "epoch": 0.8074438613328977,
      "grad_norm": 0.5234410976027418,
      "learning_rate": 0.00013526506329290933,
      "loss": 12.2074,
      "step": 14828
    },
    {
      "epoch": 0.8074983153294807,
      "grad_norm": 0.529908021814124,
      "learning_rate": 0.00013525681142611472,
      "loss": 12.1225,
      "step": 14829
    },
    {
      "epoch": 0.8075527693260637,
      "grad_norm": 0.544371633644651,
      "learning_rate": 0.00013524855928515717,
      "loss": 12.0765,
      "step": 14830
    },
    {
      "epoch": 0.8076072233226468,
      "grad_norm": 0.5615786923535416,
      "learning_rate": 0.00013524030687010096,
      "loss": 12.2037,
      "step": 14831
    },
    {
      "epoch": 0.8076616773192298,
      "grad_norm": 0.5504412742251322,
      "learning_rate": 0.0001352320541810102,
      "loss": 12.1708,
      "step": 14832
    },
    {
      "epoch": 0.8077161313158128,
      "grad_norm": 0.7066085094534744,
      "learning_rate": 0.00013522380121794907,
      "loss": 12.1918,
      "step": 14833
    },
    {
      "epoch": 0.8077705853123958,
      "grad_norm": 0.6199773081342997,
      "learning_rate": 0.00013521554798098172,
      "loss": 12.276,
      "step": 14834
    },
    {
      "epoch": 0.8078250393089788,
      "grad_norm": 0.5580291655636943,
      "learning_rate": 0.00013520729447017243,
      "loss": 12.3218,
      "step": 14835
    },
    {
      "epoch": 0.8078794933055617,
      "grad_norm": 0.5219495895617315,
      "learning_rate": 0.0001351990406855853,
      "loss": 12.1418,
      "step": 14836
    },
    {
      "epoch": 0.8079339473021449,
      "grad_norm": 0.6113277869483955,
      "learning_rate": 0.00013519078662728448,
      "loss": 12.1674,
      "step": 14837
    },
    {
      "epoch": 0.8079884012987278,
      "grad_norm": 0.5109110687457398,
      "learning_rate": 0.00013518253229533424,
      "loss": 12.3098,
      "step": 14838
    },
    {
      "epoch": 0.8080428552953108,
      "grad_norm": 0.557004942818786,
      "learning_rate": 0.0001351742776897987,
      "loss": 12.1824,
      "step": 14839
    },
    {
      "epoch": 0.8080973092918938,
      "grad_norm": 0.6010657488392056,
      "learning_rate": 0.00013516602281074213,
      "loss": 12.2227,
      "step": 14840
    },
    {
      "epoch": 0.8081517632884768,
      "grad_norm": 0.6358189902556056,
      "learning_rate": 0.00013515776765822863,
      "loss": 12.277,
      "step": 14841
    },
    {
      "epoch": 0.8082062172850598,
      "grad_norm": 0.5751197052084914,
      "learning_rate": 0.00013514951223232244,
      "loss": 12.2324,
      "step": 14842
    },
    {
      "epoch": 0.8082606712816429,
      "grad_norm": 0.5700495119696227,
      "learning_rate": 0.00013514125653308777,
      "loss": 12.2499,
      "step": 14843
    },
    {
      "epoch": 0.8083151252782259,
      "grad_norm": 0.5918496963055511,
      "learning_rate": 0.00013513300056058877,
      "loss": 12.2725,
      "step": 14844
    },
    {
      "epoch": 0.8083695792748089,
      "grad_norm": 0.5626414254597144,
      "learning_rate": 0.00013512474431488967,
      "loss": 12.2727,
      "step": 14845
    },
    {
      "epoch": 0.8084240332713919,
      "grad_norm": 0.5642979068634242,
      "learning_rate": 0.00013511648779605465,
      "loss": 12.156,
      "step": 14846
    },
    {
      "epoch": 0.8084784872679749,
      "grad_norm": 0.5493409572701059,
      "learning_rate": 0.00013510823100414796,
      "loss": 12.2125,
      "step": 14847
    },
    {
      "epoch": 0.8085329412645579,
      "grad_norm": 0.5689975122058666,
      "learning_rate": 0.00013509997393923377,
      "loss": 12.2015,
      "step": 14848
    },
    {
      "epoch": 0.808587395261141,
      "grad_norm": 0.49182099649915784,
      "learning_rate": 0.0001350917166013763,
      "loss": 12.1185,
      "step": 14849
    },
    {
      "epoch": 0.808641849257724,
      "grad_norm": 0.5247699829593893,
      "learning_rate": 0.00013508345899063975,
      "loss": 12.1663,
      "step": 14850
    },
    {
      "epoch": 0.808696303254307,
      "grad_norm": 0.5642604245222994,
      "learning_rate": 0.00013507520110708833,
      "loss": 12.1613,
      "step": 14851
    },
    {
      "epoch": 0.80875075725089,
      "grad_norm": 0.5587763435778694,
      "learning_rate": 0.00013506694295078628,
      "loss": 12.2153,
      "step": 14852
    },
    {
      "epoch": 0.808805211247473,
      "grad_norm": 0.5574095281181681,
      "learning_rate": 0.0001350586845217978,
      "loss": 12.267,
      "step": 14853
    },
    {
      "epoch": 0.8088596652440561,
      "grad_norm": 0.5716751867881421,
      "learning_rate": 0.0001350504258201871,
      "loss": 12.1996,
      "step": 14854
    },
    {
      "epoch": 0.808914119240639,
      "grad_norm": 0.5550756421929837,
      "learning_rate": 0.00013504216684601843,
      "loss": 12.1243,
      "step": 14855
    },
    {
      "epoch": 0.808968573237222,
      "grad_norm": 0.5443095253722837,
      "learning_rate": 0.00013503390759935597,
      "loss": 11.9836,
      "step": 14856
    },
    {
      "epoch": 0.809023027233805,
      "grad_norm": 0.6122039032601948,
      "learning_rate": 0.00013502564808026398,
      "loss": 12.2784,
      "step": 14857
    },
    {
      "epoch": 0.809077481230388,
      "grad_norm": 0.5787822118596989,
      "learning_rate": 0.00013501738828880668,
      "loss": 12.086,
      "step": 14858
    },
    {
      "epoch": 0.809131935226971,
      "grad_norm": 0.6197521341959868,
      "learning_rate": 0.0001350091282250483,
      "loss": 12.3282,
      "step": 14859
    },
    {
      "epoch": 0.8091863892235541,
      "grad_norm": 0.5613037311489635,
      "learning_rate": 0.00013500086788905305,
      "loss": 12.1889,
      "step": 14860
    },
    {
      "epoch": 0.8092408432201371,
      "grad_norm": 0.5834880990877368,
      "learning_rate": 0.00013499260728088518,
      "loss": 12.2518,
      "step": 14861
    },
    {
      "epoch": 0.8092952972167201,
      "grad_norm": 0.6046503559268155,
      "learning_rate": 0.00013498434640060896,
      "loss": 12.1995,
      "step": 14862
    },
    {
      "epoch": 0.8093497512133031,
      "grad_norm": 0.5478187529728412,
      "learning_rate": 0.00013497608524828857,
      "loss": 12.2138,
      "step": 14863
    },
    {
      "epoch": 0.8094042052098861,
      "grad_norm": 0.5791711463562509,
      "learning_rate": 0.0001349678238239883,
      "loss": 12.0796,
      "step": 14864
    },
    {
      "epoch": 0.8094586592064691,
      "grad_norm": 0.5683328909188236,
      "learning_rate": 0.00013495956212777237,
      "loss": 12.1985,
      "step": 14865
    },
    {
      "epoch": 0.8095131132030522,
      "grad_norm": 0.5371246737714352,
      "learning_rate": 0.00013495130015970497,
      "loss": 12.0929,
      "step": 14866
    },
    {
      "epoch": 0.8095675671996352,
      "grad_norm": 0.5835081530600631,
      "learning_rate": 0.00013494303791985045,
      "loss": 12.1738,
      "step": 14867
    },
    {
      "epoch": 0.8096220211962182,
      "grad_norm": 0.5857274196314827,
      "learning_rate": 0.00013493477540827298,
      "loss": 12.2489,
      "step": 14868
    },
    {
      "epoch": 0.8096764751928012,
      "grad_norm": 0.5554600483101491,
      "learning_rate": 0.00013492651262503685,
      "loss": 12.1106,
      "step": 14869
    },
    {
      "epoch": 0.8097309291893842,
      "grad_norm": 0.6650260925920747,
      "learning_rate": 0.00013491824957020628,
      "loss": 12.4283,
      "step": 14870
    },
    {
      "epoch": 0.8097853831859672,
      "grad_norm": 0.5837912278619439,
      "learning_rate": 0.00013490998624384558,
      "loss": 12.1607,
      "step": 14871
    },
    {
      "epoch": 0.8098398371825503,
      "grad_norm": 0.5421872387535078,
      "learning_rate": 0.0001349017226460189,
      "loss": 12.2336,
      "step": 14872
    },
    {
      "epoch": 0.8098942911791333,
      "grad_norm": 0.7267994171336976,
      "learning_rate": 0.00013489345877679067,
      "loss": 12.2144,
      "step": 14873
    },
    {
      "epoch": 0.8099487451757162,
      "grad_norm": 0.544225159839763,
      "learning_rate": 0.000134885194636225,
      "loss": 12.3555,
      "step": 14874
    },
    {
      "epoch": 0.8100031991722992,
      "grad_norm": 0.6004138775591223,
      "learning_rate": 0.00013487693022438624,
      "loss": 12.2701,
      "step": 14875
    },
    {
      "epoch": 0.8100576531688822,
      "grad_norm": 0.6246154775919707,
      "learning_rate": 0.0001348686655413386,
      "loss": 12.2617,
      "step": 14876
    },
    {
      "epoch": 0.8101121071654652,
      "grad_norm": 0.5404264875377471,
      "learning_rate": 0.00013486040058714632,
      "loss": 12.2787,
      "step": 14877
    },
    {
      "epoch": 0.8101665611620483,
      "grad_norm": 0.550133321004403,
      "learning_rate": 0.00013485213536187378,
      "loss": 12.2487,
      "step": 14878
    },
    {
      "epoch": 0.8102210151586313,
      "grad_norm": 0.5770492698911255,
      "learning_rate": 0.00013484386986558516,
      "loss": 12.2898,
      "step": 14879
    },
    {
      "epoch": 0.8102754691552143,
      "grad_norm": 0.7417205778001171,
      "learning_rate": 0.0001348356040983448,
      "loss": 12.324,
      "step": 14880
    },
    {
      "epoch": 0.8103299231517973,
      "grad_norm": 0.5670109763720117,
      "learning_rate": 0.00013482733806021693,
      "loss": 12.1479,
      "step": 14881
    },
    {
      "epoch": 0.8103843771483803,
      "grad_norm": 0.6161954924055218,
      "learning_rate": 0.00013481907175126582,
      "loss": 12.3676,
      "step": 14882
    },
    {
      "epoch": 0.8104388311449634,
      "grad_norm": 0.5659342862521876,
      "learning_rate": 0.00013481080517155578,
      "loss": 12.1375,
      "step": 14883
    },
    {
      "epoch": 0.8104932851415464,
      "grad_norm": 0.5521924434604571,
      "learning_rate": 0.00013480253832115108,
      "loss": 12.1946,
      "step": 14884
    },
    {
      "epoch": 0.8105477391381294,
      "grad_norm": 0.5282392078953213,
      "learning_rate": 0.000134794271200116,
      "loss": 12.1081,
      "step": 14885
    },
    {
      "epoch": 0.8106021931347124,
      "grad_norm": 0.5869486267483298,
      "learning_rate": 0.00013478600380851486,
      "loss": 12.1544,
      "step": 14886
    },
    {
      "epoch": 0.8106566471312954,
      "grad_norm": 0.6393240761158134,
      "learning_rate": 0.00013477773614641188,
      "loss": 12.2286,
      "step": 14887
    },
    {
      "epoch": 0.8107111011278784,
      "grad_norm": 0.5865865463438562,
      "learning_rate": 0.0001347694682138714,
      "loss": 12.1453,
      "step": 14888
    },
    {
      "epoch": 0.8107655551244615,
      "grad_norm": 0.6034268114450104,
      "learning_rate": 0.0001347612000109577,
      "loss": 12.1117,
      "step": 14889
    },
    {
      "epoch": 0.8108200091210445,
      "grad_norm": 0.5755438394885927,
      "learning_rate": 0.0001347529315377351,
      "loss": 12.2393,
      "step": 14890
    },
    {
      "epoch": 0.8108744631176275,
      "grad_norm": 0.5920600215118633,
      "learning_rate": 0.00013474466279426788,
      "loss": 12.2165,
      "step": 14891
    },
    {
      "epoch": 0.8109289171142104,
      "grad_norm": 0.5869621005908414,
      "learning_rate": 0.00013473639378062035,
      "loss": 12.1462,
      "step": 14892
    },
    {
      "epoch": 0.8109833711107934,
      "grad_norm": 0.6218140884335485,
      "learning_rate": 0.00013472812449685675,
      "loss": 12.2385,
      "step": 14893
    },
    {
      "epoch": 0.8110378251073764,
      "grad_norm": 0.6585378622196876,
      "learning_rate": 0.00013471985494304143,
      "loss": 12.2975,
      "step": 14894
    },
    {
      "epoch": 0.8110922791039595,
      "grad_norm": 0.603287054489097,
      "learning_rate": 0.00013471158511923876,
      "loss": 12.1148,
      "step": 14895
    },
    {
      "epoch": 0.8111467331005425,
      "grad_norm": 0.5996342050759436,
      "learning_rate": 0.00013470331502551293,
      "loss": 12.2267,
      "step": 14896
    },
    {
      "epoch": 0.8112011870971255,
      "grad_norm": 0.5561600021162297,
      "learning_rate": 0.00013469504466192831,
      "loss": 12.227,
      "step": 14897
    },
    {
      "epoch": 0.8112556410937085,
      "grad_norm": 0.5989965552789527,
      "learning_rate": 0.0001346867740285492,
      "loss": 12.1776,
      "step": 14898
    },
    {
      "epoch": 0.8113100950902915,
      "grad_norm": 0.5785238524484136,
      "learning_rate": 0.00013467850312543994,
      "loss": 12.2839,
      "step": 14899
    },
    {
      "epoch": 0.8113645490868745,
      "grad_norm": 0.6157536243138646,
      "learning_rate": 0.0001346702319526648,
      "loss": 12.216,
      "step": 14900
    },
    {
      "epoch": 0.8114190030834576,
      "grad_norm": 0.6139216252470333,
      "learning_rate": 0.00013466196051028814,
      "loss": 12.3419,
      "step": 14901
    },
    {
      "epoch": 0.8114734570800406,
      "grad_norm": 0.565349217936788,
      "learning_rate": 0.00013465368879837425,
      "loss": 12.2088,
      "step": 14902
    },
    {
      "epoch": 0.8115279110766236,
      "grad_norm": 0.610329727565721,
      "learning_rate": 0.00013464541681698747,
      "loss": 12.246,
      "step": 14903
    },
    {
      "epoch": 0.8115823650732066,
      "grad_norm": 0.6121022967448893,
      "learning_rate": 0.0001346371445661921,
      "loss": 12.1796,
      "step": 14904
    },
    {
      "epoch": 0.8116368190697896,
      "grad_norm": 0.5470315206379218,
      "learning_rate": 0.00013462887204605253,
      "loss": 12.0974,
      "step": 14905
    },
    {
      "epoch": 0.8116912730663726,
      "grad_norm": 0.5627392323847954,
      "learning_rate": 0.00013462059925663299,
      "loss": 12.1349,
      "step": 14906
    },
    {
      "epoch": 0.8117457270629557,
      "grad_norm": 0.6150381921746256,
      "learning_rate": 0.0001346123261979979,
      "loss": 12.1735,
      "step": 14907
    },
    {
      "epoch": 0.8118001810595387,
      "grad_norm": 0.539941217460702,
      "learning_rate": 0.00013460405287021155,
      "loss": 12.1675,
      "step": 14908
    },
    {
      "epoch": 0.8118546350561217,
      "grad_norm": 0.6506378340672261,
      "learning_rate": 0.0001345957792733383,
      "loss": 12.1344,
      "step": 14909
    },
    {
      "epoch": 0.8119090890527046,
      "grad_norm": 0.5772058084352601,
      "learning_rate": 0.00013458750540744244,
      "loss": 12.2413,
      "step": 14910
    },
    {
      "epoch": 0.8119635430492876,
      "grad_norm": 0.5675206244256016,
      "learning_rate": 0.00013457923127258833,
      "loss": 12.0752,
      "step": 14911
    },
    {
      "epoch": 0.8120179970458706,
      "grad_norm": 0.5806601939090454,
      "learning_rate": 0.00013457095686884033,
      "loss": 12.2102,
      "step": 14912
    },
    {
      "epoch": 0.8120724510424537,
      "grad_norm": 0.6070178312564786,
      "learning_rate": 0.00013456268219626277,
      "loss": 12.2825,
      "step": 14913
    },
    {
      "epoch": 0.8121269050390367,
      "grad_norm": 0.6414391844447898,
      "learning_rate": 0.00013455440725492,
      "loss": 12.1771,
      "step": 14914
    },
    {
      "epoch": 0.8121813590356197,
      "grad_norm": 0.593889169013522,
      "learning_rate": 0.00013454613204487637,
      "loss": 12.1321,
      "step": 14915
    },
    {
      "epoch": 0.8122358130322027,
      "grad_norm": 0.6188895270570187,
      "learning_rate": 0.00013453785656619623,
      "loss": 12.3201,
      "step": 14916
    },
    {
      "epoch": 0.8122902670287857,
      "grad_norm": 0.6319531156353801,
      "learning_rate": 0.00013452958081894392,
      "loss": 12.2275,
      "step": 14917
    },
    {
      "epoch": 0.8123447210253688,
      "grad_norm": 0.6033406417544306,
      "learning_rate": 0.0001345213048031838,
      "loss": 12.2757,
      "step": 14918
    },
    {
      "epoch": 0.8123991750219518,
      "grad_norm": 0.5136940686796113,
      "learning_rate": 0.00013451302851898023,
      "loss": 12.141,
      "step": 14919
    },
    {
      "epoch": 0.8124536290185348,
      "grad_norm": 0.5767843735213666,
      "learning_rate": 0.00013450475196639754,
      "loss": 12.2928,
      "step": 14920
    },
    {
      "epoch": 0.8125080830151178,
      "grad_norm": 0.5360562112845915,
      "learning_rate": 0.00013449647514550013,
      "loss": 12.2221,
      "step": 14921
    },
    {
      "epoch": 0.8125625370117008,
      "grad_norm": 0.5779506182159675,
      "learning_rate": 0.00013448819805635234,
      "loss": 12.184,
      "step": 14922
    },
    {
      "epoch": 0.8126169910082838,
      "grad_norm": 0.5536109511509439,
      "learning_rate": 0.0001344799206990185,
      "loss": 12.2786,
      "step": 14923
    },
    {
      "epoch": 0.8126714450048669,
      "grad_norm": 0.530682294119379,
      "learning_rate": 0.0001344716430735631,
      "loss": 12.1823,
      "step": 14924
    },
    {
      "epoch": 0.8127258990014499,
      "grad_norm": 0.5865372297258536,
      "learning_rate": 0.00013446336518005037,
      "loss": 12.1718,
      "step": 14925
    },
    {
      "epoch": 0.8127803529980329,
      "grad_norm": 0.5676681573906274,
      "learning_rate": 0.00013445508701854473,
      "loss": 12.1888,
      "step": 14926
    },
    {
      "epoch": 0.8128348069946159,
      "grad_norm": 0.5783128967423121,
      "learning_rate": 0.00013444680858911055,
      "loss": 12.2015,
      "step": 14927
    },
    {
      "epoch": 0.8128892609911988,
      "grad_norm": 0.6538533256948675,
      "learning_rate": 0.00013443852989181222,
      "loss": 12.3719,
      "step": 14928
    },
    {
      "epoch": 0.8129437149877818,
      "grad_norm": 0.5702116895106648,
      "learning_rate": 0.0001344302509267141,
      "loss": 12.2315,
      "step": 14929
    },
    {
      "epoch": 0.8129981689843649,
      "grad_norm": 0.5575423623731566,
      "learning_rate": 0.0001344219716938806,
      "loss": 12.1814,
      "step": 14930
    },
    {
      "epoch": 0.8130526229809479,
      "grad_norm": 0.6088179847352079,
      "learning_rate": 0.00013441369219337605,
      "loss": 12.1685,
      "step": 14931
    },
    {
      "epoch": 0.8131070769775309,
      "grad_norm": 0.607212284364403,
      "learning_rate": 0.00013440541242526485,
      "loss": 12.3098,
      "step": 14932
    },
    {
      "epoch": 0.8131615309741139,
      "grad_norm": 0.567310137687238,
      "learning_rate": 0.00013439713238961142,
      "loss": 12.0795,
      "step": 14933
    },
    {
      "epoch": 0.8132159849706969,
      "grad_norm": 0.5280748988570373,
      "learning_rate": 0.0001343888520864801,
      "loss": 12.2618,
      "step": 14934
    },
    {
      "epoch": 0.8132704389672799,
      "grad_norm": 0.6022799459452628,
      "learning_rate": 0.00013438057151593532,
      "loss": 12.2013,
      "step": 14935
    },
    {
      "epoch": 0.813324892963863,
      "grad_norm": 0.5395486494308462,
      "learning_rate": 0.00013437229067804146,
      "loss": 12.1533,
      "step": 14936
    },
    {
      "epoch": 0.813379346960446,
      "grad_norm": 0.5345849608595724,
      "learning_rate": 0.00013436400957286286,
      "loss": 12.2119,
      "step": 14937
    },
    {
      "epoch": 0.813433800957029,
      "grad_norm": 0.5223620841321481,
      "learning_rate": 0.00013435572820046397,
      "loss": 12.1625,
      "step": 14938
    },
    {
      "epoch": 0.813488254953612,
      "grad_norm": 0.6776491326508257,
      "learning_rate": 0.0001343474465609092,
      "loss": 12.2583,
      "step": 14939
    },
    {
      "epoch": 0.813542708950195,
      "grad_norm": 0.5517732802269188,
      "learning_rate": 0.00013433916465426294,
      "loss": 12.069,
      "step": 14940
    },
    {
      "epoch": 0.813597162946778,
      "grad_norm": 0.5566442191844937,
      "learning_rate": 0.00013433088248058955,
      "loss": 12.2793,
      "step": 14941
    },
    {
      "epoch": 0.8136516169433611,
      "grad_norm": 0.5713057966290501,
      "learning_rate": 0.00013432260003995347,
      "loss": 12.1679,
      "step": 14942
    },
    {
      "epoch": 0.8137060709399441,
      "grad_norm": 0.5897516315170662,
      "learning_rate": 0.00013431431733241907,
      "loss": 12.1071,
      "step": 14943
    },
    {
      "epoch": 0.8137605249365271,
      "grad_norm": 0.5827581462897125,
      "learning_rate": 0.00013430603435805077,
      "loss": 12.3552,
      "step": 14944
    },
    {
      "epoch": 0.8138149789331101,
      "grad_norm": 0.5386720004141335,
      "learning_rate": 0.00013429775111691304,
      "loss": 12.1711,
      "step": 14945
    },
    {
      "epoch": 0.813869432929693,
      "grad_norm": 0.6672047986879622,
      "learning_rate": 0.00013428946760907025,
      "loss": 12.0904,
      "step": 14946
    },
    {
      "epoch": 0.813923886926276,
      "grad_norm": 0.570405414930972,
      "learning_rate": 0.00013428118383458678,
      "loss": 12.1777,
      "step": 14947
    },
    {
      "epoch": 0.8139783409228591,
      "grad_norm": 0.5348817899525419,
      "learning_rate": 0.00013427289979352707,
      "loss": 12.181,
      "step": 14948
    },
    {
      "epoch": 0.8140327949194421,
      "grad_norm": 0.5564605962690318,
      "learning_rate": 0.00013426461548595556,
      "loss": 12.2239,
      "step": 14949
    },
    {
      "epoch": 0.8140872489160251,
      "grad_norm": 0.5826757439277491,
      "learning_rate": 0.00013425633091193666,
      "loss": 12.0763,
      "step": 14950
    },
    {
      "epoch": 0.8141417029126081,
      "grad_norm": 0.6392072154759486,
      "learning_rate": 0.00013424804607153478,
      "loss": 12.2083,
      "step": 14951
    },
    {
      "epoch": 0.8141961569091911,
      "grad_norm": 0.5278847783302267,
      "learning_rate": 0.00013423976096481435,
      "loss": 12.0883,
      "step": 14952
    },
    {
      "epoch": 0.8142506109057742,
      "grad_norm": 0.5667153367275108,
      "learning_rate": 0.00013423147559183982,
      "loss": 12.1424,
      "step": 14953
    },
    {
      "epoch": 0.8143050649023572,
      "grad_norm": 0.6059042657487198,
      "learning_rate": 0.00013422318995267554,
      "loss": 12.1883,
      "step": 14954
    },
    {
      "epoch": 0.8143595188989402,
      "grad_norm": 0.584830107343991,
      "learning_rate": 0.00013421490404738604,
      "loss": 12.2183,
      "step": 14955
    },
    {
      "epoch": 0.8144139728955232,
      "grad_norm": 0.6678900363954223,
      "learning_rate": 0.0001342066178760357,
      "loss": 12.2575,
      "step": 14956
    },
    {
      "epoch": 0.8144684268921062,
      "grad_norm": 0.563722864109056,
      "learning_rate": 0.00013419833143868897,
      "loss": 12.1862,
      "step": 14957
    },
    {
      "epoch": 0.8145228808886892,
      "grad_norm": 0.6231865053320498,
      "learning_rate": 0.00013419004473541027,
      "loss": 12.1494,
      "step": 14958
    },
    {
      "epoch": 0.8145773348852723,
      "grad_norm": 0.6414733001046949,
      "learning_rate": 0.000134181757766264,
      "loss": 12.2958,
      "step": 14959
    },
    {
      "epoch": 0.8146317888818553,
      "grad_norm": 0.5652240401860344,
      "learning_rate": 0.0001341734705313147,
      "loss": 12.1767,
      "step": 14960
    },
    {
      "epoch": 0.8146862428784383,
      "grad_norm": 0.5747060543173036,
      "learning_rate": 0.0001341651830306268,
      "loss": 12.1846,
      "step": 14961
    },
    {
      "epoch": 0.8147406968750213,
      "grad_norm": 0.6027332207345627,
      "learning_rate": 0.00013415689526426465,
      "loss": 12.1885,
      "step": 14962
    },
    {
      "epoch": 0.8147951508716043,
      "grad_norm": 0.5982557446880276,
      "learning_rate": 0.00013414860723229277,
      "loss": 12.173,
      "step": 14963
    },
    {
      "epoch": 0.8148496048681872,
      "grad_norm": 0.557802973894175,
      "learning_rate": 0.00013414031893477558,
      "loss": 12.2135,
      "step": 14964
    },
    {
      "epoch": 0.8149040588647704,
      "grad_norm": 0.5448109877067271,
      "learning_rate": 0.00013413203037177754,
      "loss": 12.146,
      "step": 14965
    },
    {
      "epoch": 0.8149585128613533,
      "grad_norm": 0.5656869069438613,
      "learning_rate": 0.00013412374154336316,
      "loss": 12.1739,
      "step": 14966
    },
    {
      "epoch": 0.8150129668579363,
      "grad_norm": 0.5997702753121104,
      "learning_rate": 0.0001341154524495968,
      "loss": 12.2298,
      "step": 14967
    },
    {
      "epoch": 0.8150674208545193,
      "grad_norm": 0.5748158705419266,
      "learning_rate": 0.00013410716309054295,
      "loss": 12.3095,
      "step": 14968
    },
    {
      "epoch": 0.8151218748511023,
      "grad_norm": 0.6540648899325642,
      "learning_rate": 0.00013409887346626612,
      "loss": 12.1783,
      "step": 14969
    },
    {
      "epoch": 0.8151763288476853,
      "grad_norm": 0.5447230236480691,
      "learning_rate": 0.0001340905835768307,
      "loss": 12.2082,
      "step": 14970
    },
    {
      "epoch": 0.8152307828442684,
      "grad_norm": 0.5399966726816193,
      "learning_rate": 0.0001340822934223012,
      "loss": 12.118,
      "step": 14971
    },
    {
      "epoch": 0.8152852368408514,
      "grad_norm": 0.5574187249050045,
      "learning_rate": 0.00013407400300274207,
      "loss": 12.1641,
      "step": 14972
    },
    {
      "epoch": 0.8153396908374344,
      "grad_norm": 0.5774856992493318,
      "learning_rate": 0.00013406571231821775,
      "loss": 12.2274,
      "step": 14973
    },
    {
      "epoch": 0.8153941448340174,
      "grad_norm": 0.5995070765895913,
      "learning_rate": 0.00013405742136879278,
      "loss": 12.2242,
      "step": 14974
    },
    {
      "epoch": 0.8154485988306004,
      "grad_norm": 0.5695205010593473,
      "learning_rate": 0.00013404913015453157,
      "loss": 12.1707,
      "step": 14975
    },
    {
      "epoch": 0.8155030528271834,
      "grad_norm": 0.5626070337798693,
      "learning_rate": 0.00013404083867549863,
      "loss": 12.2188,
      "step": 14976
    },
    {
      "epoch": 0.8155575068237665,
      "grad_norm": 0.6446567968174487,
      "learning_rate": 0.0001340325469317584,
      "loss": 12.254,
      "step": 14977
    },
    {
      "epoch": 0.8156119608203495,
      "grad_norm": 0.578028180559197,
      "learning_rate": 0.00013402425492337538,
      "loss": 12.0871,
      "step": 14978
    },
    {
      "epoch": 0.8156664148169325,
      "grad_norm": 0.5707444550148375,
      "learning_rate": 0.00013401596265041405,
      "loss": 12.2326,
      "step": 14979
    },
    {
      "epoch": 0.8157208688135155,
      "grad_norm": 0.5790218064683561,
      "learning_rate": 0.0001340076701129389,
      "loss": 12.1651,
      "step": 14980
    },
    {
      "epoch": 0.8157753228100985,
      "grad_norm": 0.5530111516710103,
      "learning_rate": 0.0001339993773110144,
      "loss": 12.2381,
      "step": 14981
    },
    {
      "epoch": 0.8158297768066815,
      "grad_norm": 0.5525339506735362,
      "learning_rate": 0.00013399108424470504,
      "loss": 12.1943,
      "step": 14982
    },
    {
      "epoch": 0.8158842308032646,
      "grad_norm": 0.6791189230415343,
      "learning_rate": 0.0001339827909140753,
      "loss": 12.2272,
      "step": 14983
    },
    {
      "epoch": 0.8159386847998475,
      "grad_norm": 0.8677941890684707,
      "learning_rate": 0.00013397449731918968,
      "loss": 12.2352,
      "step": 14984
    },
    {
      "epoch": 0.8159931387964305,
      "grad_norm": 0.5423445874305333,
      "learning_rate": 0.00013396620346011267,
      "loss": 12.2357,
      "step": 14985
    },
    {
      "epoch": 0.8160475927930135,
      "grad_norm": 0.5993168559027767,
      "learning_rate": 0.00013395790933690878,
      "loss": 12.2122,
      "step": 14986
    },
    {
      "epoch": 0.8161020467895965,
      "grad_norm": 0.5884806571904675,
      "learning_rate": 0.0001339496149496425,
      "loss": 12.2743,
      "step": 14987
    },
    {
      "epoch": 0.8161565007861796,
      "grad_norm": 0.5554630343945989,
      "learning_rate": 0.00013394132029837828,
      "loss": 11.9055,
      "step": 14988
    },
    {
      "epoch": 0.8162109547827626,
      "grad_norm": 0.5215621664224854,
      "learning_rate": 0.0001339330253831807,
      "loss": 12.1985,
      "step": 14989
    },
    {
      "epoch": 0.8162654087793456,
      "grad_norm": 0.6540481921438157,
      "learning_rate": 0.0001339247302041142,
      "loss": 12.2831,
      "step": 14990
    },
    {
      "epoch": 0.8163198627759286,
      "grad_norm": 0.6157703375357999,
      "learning_rate": 0.00013391643476124335,
      "loss": 12.262,
      "step": 14991
    },
    {
      "epoch": 0.8163743167725116,
      "grad_norm": 0.6485824430647104,
      "learning_rate": 0.00013390813905463255,
      "loss": 12.2356,
      "step": 14992
    },
    {
      "epoch": 0.8164287707690946,
      "grad_norm": 0.5571332481181887,
      "learning_rate": 0.0001338998430843464,
      "loss": 12.2123,
      "step": 14993
    },
    {
      "epoch": 0.8164832247656777,
      "grad_norm": 0.6334643771864544,
      "learning_rate": 0.0001338915468504494,
      "loss": 12.2523,
      "step": 14994
    },
    {
      "epoch": 0.8165376787622607,
      "grad_norm": 0.5118559097275878,
      "learning_rate": 0.00013388325035300605,
      "loss": 12.1212,
      "step": 14995
    },
    {
      "epoch": 0.8165921327588437,
      "grad_norm": 0.5741352405113275,
      "learning_rate": 0.00013387495359208087,
      "loss": 12.1782,
      "step": 14996
    },
    {
      "epoch": 0.8166465867554267,
      "grad_norm": 0.7133859869683401,
      "learning_rate": 0.00013386665656773834,
      "loss": 12.1758,
      "step": 14997
    },
    {
      "epoch": 0.8167010407520097,
      "grad_norm": 0.6902711214672571,
      "learning_rate": 0.00013385835928004302,
      "loss": 12.2727,
      "step": 14998
    },
    {
      "epoch": 0.8167554947485927,
      "grad_norm": 0.5711108832715581,
      "learning_rate": 0.00013385006172905942,
      "loss": 12.0462,
      "step": 14999
    },
    {
      "epoch": 0.8168099487451758,
      "grad_norm": 0.7150747556761515,
      "learning_rate": 0.00013384176391485205,
      "loss": 12.3261,
      "step": 15000
    },
    {
      "epoch": 0.8168644027417588,
      "grad_norm": 0.5812774549161445,
      "learning_rate": 0.0001338334658374855,
      "loss": 12.1661,
      "step": 15001
    },
    {
      "epoch": 0.8169188567383417,
      "grad_norm": 0.5824205643528076,
      "learning_rate": 0.0001338251674970242,
      "loss": 12.1429,
      "step": 15002
    },
    {
      "epoch": 0.8169733107349247,
      "grad_norm": 0.5720606638834883,
      "learning_rate": 0.00013381686889353273,
      "loss": 12.2057,
      "step": 15003
    },
    {
      "epoch": 0.8170277647315077,
      "grad_norm": 0.5880657890604791,
      "learning_rate": 0.00013380857002707563,
      "loss": 12.3358,
      "step": 15004
    },
    {
      "epoch": 0.8170822187280907,
      "grad_norm": 0.6258769532904722,
      "learning_rate": 0.0001338002708977174,
      "loss": 12.2655,
      "step": 15005
    },
    {
      "epoch": 0.8171366727246738,
      "grad_norm": 0.6101374276518876,
      "learning_rate": 0.00013379197150552262,
      "loss": 12.2291,
      "step": 15006
    },
    {
      "epoch": 0.8171911267212568,
      "grad_norm": 0.512928837530879,
      "learning_rate": 0.0001337836718505558,
      "loss": 12.1752,
      "step": 15007
    },
    {
      "epoch": 0.8172455807178398,
      "grad_norm": 0.5762590538623538,
      "learning_rate": 0.00013377537193288145,
      "loss": 12.1972,
      "step": 15008
    },
    {
      "epoch": 0.8173000347144228,
      "grad_norm": 0.5424482566110911,
      "learning_rate": 0.00013376707175256417,
      "loss": 12.0931,
      "step": 15009
    },
    {
      "epoch": 0.8173544887110058,
      "grad_norm": 0.5571591334733318,
      "learning_rate": 0.00013375877130966847,
      "loss": 12.2593,
      "step": 15010
    },
    {
      "epoch": 0.8174089427075888,
      "grad_norm": 0.5990429172530969,
      "learning_rate": 0.00013375047060425893,
      "loss": 12.3188,
      "step": 15011
    },
    {
      "epoch": 0.8174633967041719,
      "grad_norm": 0.5771953018397764,
      "learning_rate": 0.00013374216963640004,
      "loss": 12.1905,
      "step": 15012
    },
    {
      "epoch": 0.8175178507007549,
      "grad_norm": 0.5856012728256729,
      "learning_rate": 0.0001337338684061564,
      "loss": 12.2053,
      "step": 15013
    },
    {
      "epoch": 0.8175723046973379,
      "grad_norm": 0.5262269838088067,
      "learning_rate": 0.0001337255669135925,
      "loss": 12.2862,
      "step": 15014
    },
    {
      "epoch": 0.8176267586939209,
      "grad_norm": 0.553391820209899,
      "learning_rate": 0.000133717265158773,
      "loss": 12.2014,
      "step": 15015
    },
    {
      "epoch": 0.8176812126905039,
      "grad_norm": 0.568890866266896,
      "learning_rate": 0.00013370896314176235,
      "loss": 12.2971,
      "step": 15016
    },
    {
      "epoch": 0.817735666687087,
      "grad_norm": 0.616745452688003,
      "learning_rate": 0.00013370066086262517,
      "loss": 12.232,
      "step": 15017
    },
    {
      "epoch": 0.81779012068367,
      "grad_norm": 0.5888482042030274,
      "learning_rate": 0.00013369235832142598,
      "loss": 12.2087,
      "step": 15018
    },
    {
      "epoch": 0.817844574680253,
      "grad_norm": 0.606699835744316,
      "learning_rate": 0.00013368405551822935,
      "loss": 12.2074,
      "step": 15019
    },
    {
      "epoch": 0.817899028676836,
      "grad_norm": 0.6539207177383145,
      "learning_rate": 0.00013367575245309987,
      "loss": 12.1792,
      "step": 15020
    },
    {
      "epoch": 0.8179534826734189,
      "grad_norm": 0.5783875286795106,
      "learning_rate": 0.0001336674491261021,
      "loss": 12.2332,
      "step": 15021
    },
    {
      "epoch": 0.8180079366700019,
      "grad_norm": 0.6137768782047479,
      "learning_rate": 0.0001336591455373006,
      "loss": 12.1272,
      "step": 15022
    },
    {
      "epoch": 0.818062390666585,
      "grad_norm": 0.6339487854126397,
      "learning_rate": 0.00013365084168675994,
      "loss": 12.1809,
      "step": 15023
    },
    {
      "epoch": 0.818116844663168,
      "grad_norm": 0.57824407345384,
      "learning_rate": 0.00013364253757454467,
      "loss": 12.3144,
      "step": 15024
    },
    {
      "epoch": 0.818171298659751,
      "grad_norm": 0.5894692927948438,
      "learning_rate": 0.00013363423320071938,
      "loss": 12.2668,
      "step": 15025
    },
    {
      "epoch": 0.818225752656334,
      "grad_norm": 0.5856085183522631,
      "learning_rate": 0.00013362592856534873,
      "loss": 12.1982,
      "step": 15026
    },
    {
      "epoch": 0.818280206652917,
      "grad_norm": 0.5610742843623008,
      "learning_rate": 0.00013361762366849715,
      "loss": 12.2298,
      "step": 15027
    },
    {
      "epoch": 0.8183346606495,
      "grad_norm": 0.540835830705468,
      "learning_rate": 0.00013360931851022931,
      "loss": 12.19,
      "step": 15028
    },
    {
      "epoch": 0.8183891146460831,
      "grad_norm": 0.5510492670300929,
      "learning_rate": 0.00013360101309060974,
      "loss": 12.3049,
      "step": 15029
    },
    {
      "epoch": 0.8184435686426661,
      "grad_norm": 0.5300153454770301,
      "learning_rate": 0.0001335927074097031,
      "loss": 12.014,
      "step": 15030
    },
    {
      "epoch": 0.8184980226392491,
      "grad_norm": 0.5576465141288676,
      "learning_rate": 0.0001335844014675739,
      "loss": 12.1904,
      "step": 15031
    },
    {
      "epoch": 0.8185524766358321,
      "grad_norm": 0.5183661440156276,
      "learning_rate": 0.0001335760952642868,
      "loss": 12.2148,
      "step": 15032
    },
    {
      "epoch": 0.8186069306324151,
      "grad_norm": 0.5266003458757883,
      "learning_rate": 0.00013356778879990632,
      "loss": 12.1942,
      "step": 15033
    },
    {
      "epoch": 0.8186613846289981,
      "grad_norm": 0.5466301509703855,
      "learning_rate": 0.0001335594820744971,
      "loss": 12.1378,
      "step": 15034
    },
    {
      "epoch": 0.8187158386255812,
      "grad_norm": 0.5023351188898104,
      "learning_rate": 0.00013355117508812372,
      "loss": 12.106,
      "step": 15035
    },
    {
      "epoch": 0.8187702926221642,
      "grad_norm": 0.660955828481278,
      "learning_rate": 0.00013354286784085078,
      "loss": 12.1578,
      "step": 15036
    },
    {
      "epoch": 0.8188247466187472,
      "grad_norm": 0.5585541576290538,
      "learning_rate": 0.00013353456033274286,
      "loss": 12.1009,
      "step": 15037
    },
    {
      "epoch": 0.8188792006153301,
      "grad_norm": 0.5276413850534565,
      "learning_rate": 0.0001335262525638646,
      "loss": 12.2114,
      "step": 15038
    },
    {
      "epoch": 0.8189336546119131,
      "grad_norm": 0.6796528046414316,
      "learning_rate": 0.00013351794453428056,
      "loss": 12.5493,
      "step": 15039
    },
    {
      "epoch": 0.8189881086084961,
      "grad_norm": 0.5857017757416788,
      "learning_rate": 0.00013350963624405538,
      "loss": 12.196,
      "step": 15040
    },
    {
      "epoch": 0.8190425626050792,
      "grad_norm": 0.5514624615441971,
      "learning_rate": 0.00013350132769325362,
      "loss": 12.152,
      "step": 15041
    },
    {
      "epoch": 0.8190970166016622,
      "grad_norm": 0.5636971843317338,
      "learning_rate": 0.00013349301888193992,
      "loss": 12.2295,
      "step": 15042
    },
    {
      "epoch": 0.8191514705982452,
      "grad_norm": 0.5006540132671965,
      "learning_rate": 0.0001334847098101789,
      "loss": 12.1136,
      "step": 15043
    },
    {
      "epoch": 0.8192059245948282,
      "grad_norm": 0.5326584324047584,
      "learning_rate": 0.00013347640047803517,
      "loss": 12.065,
      "step": 15044
    },
    {
      "epoch": 0.8192603785914112,
      "grad_norm": 0.5372718477066166,
      "learning_rate": 0.00013346809088557332,
      "loss": 12.2216,
      "step": 15045
    },
    {
      "epoch": 0.8193148325879942,
      "grad_norm": 0.5359071036987115,
      "learning_rate": 0.000133459781032858,
      "loss": 12.1195,
      "step": 15046
    },
    {
      "epoch": 0.8193692865845773,
      "grad_norm": 0.5939841514566832,
      "learning_rate": 0.00013345147091995378,
      "loss": 12.1763,
      "step": 15047
    },
    {
      "epoch": 0.8194237405811603,
      "grad_norm": 0.5482117414915757,
      "learning_rate": 0.00013344316054692533,
      "loss": 12.176,
      "step": 15048
    },
    {
      "epoch": 0.8194781945777433,
      "grad_norm": 0.5910514392352234,
      "learning_rate": 0.0001334348499138373,
      "loss": 12.2198,
      "step": 15049
    },
    {
      "epoch": 0.8195326485743263,
      "grad_norm": 0.5148094608300227,
      "learning_rate": 0.00013342653902075418,
      "loss": 12.2085,
      "step": 15050
    },
    {
      "epoch": 0.8195871025709093,
      "grad_norm": 0.6351328799642851,
      "learning_rate": 0.00013341822786774076,
      "loss": 12.2853,
      "step": 15051
    },
    {
      "epoch": 0.8196415565674924,
      "grad_norm": 0.5943046099493197,
      "learning_rate": 0.00013340991645486157,
      "loss": 12.1503,
      "step": 15052
    },
    {
      "epoch": 0.8196960105640754,
      "grad_norm": 0.7340126164156044,
      "learning_rate": 0.00013340160478218126,
      "loss": 12.3678,
      "step": 15053
    },
    {
      "epoch": 0.8197504645606584,
      "grad_norm": 0.5561954548461967,
      "learning_rate": 0.00013339329284976447,
      "loss": 12.2121,
      "step": 15054
    },
    {
      "epoch": 0.8198049185572414,
      "grad_norm": 0.5936471684288838,
      "learning_rate": 0.00013338498065767587,
      "loss": 12.2002,
      "step": 15055
    },
    {
      "epoch": 0.8198593725538244,
      "grad_norm": 0.5767221429828392,
      "learning_rate": 0.00013337666820598001,
      "loss": 12.2062,
      "step": 15056
    },
    {
      "epoch": 0.8199138265504073,
      "grad_norm": 0.5520368662964245,
      "learning_rate": 0.0001333683554947416,
      "loss": 12.2063,
      "step": 15057
    },
    {
      "epoch": 0.8199682805469904,
      "grad_norm": 0.5753584754006462,
      "learning_rate": 0.00013336004252402527,
      "loss": 12.1897,
      "step": 15058
    },
    {
      "epoch": 0.8200227345435734,
      "grad_norm": 0.533969251635747,
      "learning_rate": 0.00013335172929389565,
      "loss": 12.1252,
      "step": 15059
    },
    {
      "epoch": 0.8200771885401564,
      "grad_norm": 0.5427429013322442,
      "learning_rate": 0.0001333434158044174,
      "loss": 12.1885,
      "step": 15060
    },
    {
      "epoch": 0.8201316425367394,
      "grad_norm": 0.4952467730062113,
      "learning_rate": 0.00013333510205565516,
      "loss": 12.0779,
      "step": 15061
    },
    {
      "epoch": 0.8201860965333224,
      "grad_norm": 0.5345066578384771,
      "learning_rate": 0.00013332678804767358,
      "loss": 12.2012,
      "step": 15062
    },
    {
      "epoch": 0.8202405505299054,
      "grad_norm": 0.552949495917493,
      "learning_rate": 0.00013331847378053726,
      "loss": 12.1347,
      "step": 15063
    },
    {
      "epoch": 0.8202950045264885,
      "grad_norm": 0.5550958159476888,
      "learning_rate": 0.00013331015925431095,
      "loss": 12.2529,
      "step": 15064
    },
    {
      "epoch": 0.8203494585230715,
      "grad_norm": 0.5544262032236434,
      "learning_rate": 0.00013330184446905922,
      "loss": 12.0939,
      "step": 15065
    },
    {
      "epoch": 0.8204039125196545,
      "grad_norm": 0.5958917087716364,
      "learning_rate": 0.00013329352942484678,
      "loss": 12.1504,
      "step": 15066
    },
    {
      "epoch": 0.8204583665162375,
      "grad_norm": 0.550229500277586,
      "learning_rate": 0.0001332852141217383,
      "loss": 12.101,
      "step": 15067
    },
    {
      "epoch": 0.8205128205128205,
      "grad_norm": 0.5225662194495485,
      "learning_rate": 0.00013327689855979836,
      "loss": 12.1023,
      "step": 15068
    },
    {
      "epoch": 0.8205672745094035,
      "grad_norm": 0.530911218491028,
      "learning_rate": 0.00013326858273909168,
      "loss": 12.2245,
      "step": 15069
    },
    {
      "epoch": 0.8206217285059866,
      "grad_norm": 0.5554527417021022,
      "learning_rate": 0.00013326026665968295,
      "loss": 12.1633,
      "step": 15070
    },
    {
      "epoch": 0.8206761825025696,
      "grad_norm": 0.6875066978735751,
      "learning_rate": 0.00013325195032163682,
      "loss": 12.2125,
      "step": 15071
    },
    {
      "epoch": 0.8207306364991526,
      "grad_norm": 0.6042554096651395,
      "learning_rate": 0.00013324363372501795,
      "loss": 12.2184,
      "step": 15072
    },
    {
      "epoch": 0.8207850904957356,
      "grad_norm": 0.5588330743067605,
      "learning_rate": 0.00013323531686989097,
      "loss": 12.1346,
      "step": 15073
    },
    {
      "epoch": 0.8208395444923186,
      "grad_norm": 0.6216993891491206,
      "learning_rate": 0.0001332269997563206,
      "loss": 12.2326,
      "step": 15074
    },
    {
      "epoch": 0.8208939984889015,
      "grad_norm": 0.5679478272925854,
      "learning_rate": 0.00013321868238437153,
      "loss": 12.2848,
      "step": 15075
    },
    {
      "epoch": 0.8209484524854846,
      "grad_norm": 0.6401907090556928,
      "learning_rate": 0.0001332103647541084,
      "loss": 12.2149,
      "step": 15076
    },
    {
      "epoch": 0.8210029064820676,
      "grad_norm": 0.5509524927768689,
      "learning_rate": 0.00013320204686559592,
      "loss": 12.1914,
      "step": 15077
    },
    {
      "epoch": 0.8210573604786506,
      "grad_norm": 0.5333250539046567,
      "learning_rate": 0.00013319372871889874,
      "loss": 12.1979,
      "step": 15078
    },
    {
      "epoch": 0.8211118144752336,
      "grad_norm": 0.6510537470053677,
      "learning_rate": 0.00013318541031408156,
      "loss": 12.1866,
      "step": 15079
    },
    {
      "epoch": 0.8211662684718166,
      "grad_norm": 0.5456246146264664,
      "learning_rate": 0.00013317709165120903,
      "loss": 12.1768,
      "step": 15080
    },
    {
      "epoch": 0.8212207224683996,
      "grad_norm": 0.60234366708656,
      "learning_rate": 0.0001331687727303459,
      "loss": 12.173,
      "step": 15081
    },
    {
      "epoch": 0.8212751764649827,
      "grad_norm": 0.5435572532492777,
      "learning_rate": 0.00013316045355155689,
      "loss": 12.2481,
      "step": 15082
    },
    {
      "epoch": 0.8213296304615657,
      "grad_norm": 0.5782641007791894,
      "learning_rate": 0.00013315213411490656,
      "loss": 12.126,
      "step": 15083
    },
    {
      "epoch": 0.8213840844581487,
      "grad_norm": 0.5774753450730651,
      "learning_rate": 0.0001331438144204597,
      "loss": 12.0706,
      "step": 15084
    },
    {
      "epoch": 0.8214385384547317,
      "grad_norm": 0.6488453649299252,
      "learning_rate": 0.00013313549446828096,
      "loss": 12.1743,
      "step": 15085
    },
    {
      "epoch": 0.8214929924513147,
      "grad_norm": 0.5730644068548515,
      "learning_rate": 0.00013312717425843508,
      "loss": 12.218,
      "step": 15086
    },
    {
      "epoch": 0.8215474464478978,
      "grad_norm": 0.5404712651524503,
      "learning_rate": 0.00013311885379098674,
      "loss": 12.1142,
      "step": 15087
    },
    {
      "epoch": 0.8216019004444808,
      "grad_norm": 0.5496206771070863,
      "learning_rate": 0.00013311053306600066,
      "loss": 12.2041,
      "step": 15088
    },
    {
      "epoch": 0.8216563544410638,
      "grad_norm": 0.5862083753367804,
      "learning_rate": 0.00013310221208354147,
      "loss": 12.2,
      "step": 15089
    },
    {
      "epoch": 0.8217108084376468,
      "grad_norm": 0.6033512458007323,
      "learning_rate": 0.00013309389084367396,
      "loss": 12.2182,
      "step": 15090
    },
    {
      "epoch": 0.8217652624342298,
      "grad_norm": 0.576839526949148,
      "learning_rate": 0.00013308556934646276,
      "loss": 12.3081,
      "step": 15091
    },
    {
      "epoch": 0.8218197164308128,
      "grad_norm": 0.5115106817823334,
      "learning_rate": 0.0001330772475919727,
      "loss": 12.1187,
      "step": 15092
    },
    {
      "epoch": 0.8218741704273959,
      "grad_norm": 0.6423138685815555,
      "learning_rate": 0.00013306892558026837,
      "loss": 12.3313,
      "step": 15093
    },
    {
      "epoch": 0.8219286244239788,
      "grad_norm": 0.5606849061577902,
      "learning_rate": 0.00013306060331141456,
      "loss": 12.1125,
      "step": 15094
    },
    {
      "epoch": 0.8219830784205618,
      "grad_norm": 0.5475736926996045,
      "learning_rate": 0.0001330522807854759,
      "loss": 12.2019,
      "step": 15095
    },
    {
      "epoch": 0.8220375324171448,
      "grad_norm": 0.5095605580907949,
      "learning_rate": 0.00013304395800251722,
      "loss": 12.1341,
      "step": 15096
    },
    {
      "epoch": 0.8220919864137278,
      "grad_norm": 0.5374397129569775,
      "learning_rate": 0.00013303563496260313,
      "loss": 12.304,
      "step": 15097
    },
    {
      "epoch": 0.8221464404103108,
      "grad_norm": 0.5965306221315761,
      "learning_rate": 0.00013302731166579842,
      "loss": 12.108,
      "step": 15098
    },
    {
      "epoch": 0.8222008944068939,
      "grad_norm": 0.6075627779084403,
      "learning_rate": 0.00013301898811216782,
      "loss": 12.2044,
      "step": 15099
    },
    {
      "epoch": 0.8222553484034769,
      "grad_norm": 0.5798476561085326,
      "learning_rate": 0.000133010664301776,
      "loss": 12.1951,
      "step": 15100
    },
    {
      "epoch": 0.8223098024000599,
      "grad_norm": 0.5656120162879976,
      "learning_rate": 0.00013300234023468774,
      "loss": 12.2176,
      "step": 15101
    },
    {
      "epoch": 0.8223642563966429,
      "grad_norm": 0.5392765374070314,
      "learning_rate": 0.00013299401591096774,
      "loss": 12.3709,
      "step": 15102
    },
    {
      "epoch": 0.8224187103932259,
      "grad_norm": 0.5474554976638221,
      "learning_rate": 0.00013298569133068073,
      "loss": 12.1893,
      "step": 15103
    },
    {
      "epoch": 0.8224731643898089,
      "grad_norm": 0.5922941430634817,
      "learning_rate": 0.00013297736649389147,
      "loss": 12.2515,
      "step": 15104
    },
    {
      "epoch": 0.822527618386392,
      "grad_norm": 0.6390417808924062,
      "learning_rate": 0.00013296904140066467,
      "loss": 12.2673,
      "step": 15105
    },
    {
      "epoch": 0.822582072382975,
      "grad_norm": 0.5320100230544484,
      "learning_rate": 0.00013296071605106507,
      "loss": 12.1982,
      "step": 15106
    },
    {
      "epoch": 0.822636526379558,
      "grad_norm": 0.574697692543034,
      "learning_rate": 0.00013295239044515742,
      "loss": 12.2736,
      "step": 15107
    },
    {
      "epoch": 0.822690980376141,
      "grad_norm": 0.5725886367874117,
      "learning_rate": 0.00013294406458300644,
      "loss": 12.1297,
      "step": 15108
    },
    {
      "epoch": 0.822745434372724,
      "grad_norm": 0.6607264694600522,
      "learning_rate": 0.0001329357384646769,
      "loss": 12.2569,
      "step": 15109
    },
    {
      "epoch": 0.822799888369307,
      "grad_norm": 0.5687599519704036,
      "learning_rate": 0.00013292741209023354,
      "loss": 12.2327,
      "step": 15110
    },
    {
      "epoch": 0.8228543423658901,
      "grad_norm": 0.5317134091370369,
      "learning_rate": 0.0001329190854597411,
      "loss": 12.1822,
      "step": 15111
    },
    {
      "epoch": 0.822908796362473,
      "grad_norm": 0.5366673126182483,
      "learning_rate": 0.00013291075857326434,
      "loss": 12.1526,
      "step": 15112
    },
    {
      "epoch": 0.822963250359056,
      "grad_norm": 0.6491831756234533,
      "learning_rate": 0.00013290243143086798,
      "loss": 12.2972,
      "step": 15113
    },
    {
      "epoch": 0.823017704355639,
      "grad_norm": 0.5172806571766189,
      "learning_rate": 0.00013289410403261682,
      "loss": 12.0933,
      "step": 15114
    },
    {
      "epoch": 0.823072158352222,
      "grad_norm": 0.5683596383075902,
      "learning_rate": 0.00013288577637857556,
      "loss": 12.2948,
      "step": 15115
    },
    {
      "epoch": 0.823126612348805,
      "grad_norm": 0.5875142686688732,
      "learning_rate": 0.00013287744846880904,
      "loss": 12.2356,
      "step": 15116
    },
    {
      "epoch": 0.8231810663453881,
      "grad_norm": 0.570076875886587,
      "learning_rate": 0.00013286912030338193,
      "loss": 12.3245,
      "step": 15117
    },
    {
      "epoch": 0.8232355203419711,
      "grad_norm": 0.6519852031719279,
      "learning_rate": 0.000132860791882359,
      "loss": 12.1499,
      "step": 15118
    },
    {
      "epoch": 0.8232899743385541,
      "grad_norm": 0.5983918322104361,
      "learning_rate": 0.0001328524632058051,
      "loss": 12.1896,
      "step": 15119
    },
    {
      "epoch": 0.8233444283351371,
      "grad_norm": 0.5932760518870501,
      "learning_rate": 0.0001328441342737849,
      "loss": 12.198,
      "step": 15120
    },
    {
      "epoch": 0.8233988823317201,
      "grad_norm": 0.59185708375135,
      "learning_rate": 0.00013283580508636323,
      "loss": 12.1776,
      "step": 15121
    },
    {
      "epoch": 0.8234533363283032,
      "grad_norm": 0.5858903392430636,
      "learning_rate": 0.0001328274756436048,
      "loss": 12.1823,
      "step": 15122
    },
    {
      "epoch": 0.8235077903248862,
      "grad_norm": 0.5887880925684659,
      "learning_rate": 0.00013281914594557442,
      "loss": 12.0918,
      "step": 15123
    },
    {
      "epoch": 0.8235622443214692,
      "grad_norm": 0.540027316882084,
      "learning_rate": 0.00013281081599233688,
      "loss": 12.1714,
      "step": 15124
    },
    {
      "epoch": 0.8236166983180522,
      "grad_norm": 0.5667513931707916,
      "learning_rate": 0.0001328024857839569,
      "loss": 12.1175,
      "step": 15125
    },
    {
      "epoch": 0.8236711523146352,
      "grad_norm": 0.6075293531496816,
      "learning_rate": 0.00013279415532049932,
      "loss": 12.1341,
      "step": 15126
    },
    {
      "epoch": 0.8237256063112182,
      "grad_norm": 0.6328301413775694,
      "learning_rate": 0.0001327858246020289,
      "loss": 12.2516,
      "step": 15127
    },
    {
      "epoch": 0.8237800603078013,
      "grad_norm": 0.5389076487138665,
      "learning_rate": 0.0001327774936286104,
      "loss": 12.3203,
      "step": 15128
    },
    {
      "epoch": 0.8238345143043843,
      "grad_norm": 0.5732364690525088,
      "learning_rate": 0.00013276916240030856,
      "loss": 12.2238,
      "step": 15129
    },
    {
      "epoch": 0.8238889683009673,
      "grad_norm": 0.6314953539941783,
      "learning_rate": 0.00013276083091718827,
      "loss": 12.236,
      "step": 15130
    },
    {
      "epoch": 0.8239434222975502,
      "grad_norm": 0.5614318534426168,
      "learning_rate": 0.00013275249917931424,
      "loss": 12.0909,
      "step": 15131
    },
    {
      "epoch": 0.8239978762941332,
      "grad_norm": 0.5909689994430131,
      "learning_rate": 0.00013274416718675133,
      "loss": 12.1413,
      "step": 15132
    },
    {
      "epoch": 0.8240523302907162,
      "grad_norm": 0.634909287782436,
      "learning_rate": 0.00013273583493956425,
      "loss": 12.3205,
      "step": 15133
    },
    {
      "epoch": 0.8241067842872993,
      "grad_norm": 0.5915683255470073,
      "learning_rate": 0.00013272750243781782,
      "loss": 12.0764,
      "step": 15134
    },
    {
      "epoch": 0.8241612382838823,
      "grad_norm": 0.6220220305808242,
      "learning_rate": 0.00013271916968157684,
      "loss": 12.2925,
      "step": 15135
    },
    {
      "epoch": 0.8242156922804653,
      "grad_norm": 0.5997585424873934,
      "learning_rate": 0.00013271083667090614,
      "loss": 12.1521,
      "step": 15136
    },
    {
      "epoch": 0.8242701462770483,
      "grad_norm": 0.5278769519062395,
      "learning_rate": 0.00013270250340587046,
      "loss": 12.1131,
      "step": 15137
    },
    {
      "epoch": 0.8243246002736313,
      "grad_norm": 0.524995503437437,
      "learning_rate": 0.00013269416988653468,
      "loss": 12.2127,
      "step": 15138
    },
    {
      "epoch": 0.8243790542702143,
      "grad_norm": 0.6492659430294754,
      "learning_rate": 0.0001326858361129635,
      "loss": 12.2993,
      "step": 15139
    },
    {
      "epoch": 0.8244335082667974,
      "grad_norm": 0.5796193885415427,
      "learning_rate": 0.00013267750208522175,
      "loss": 12.2585,
      "step": 15140
    },
    {
      "epoch": 0.8244879622633804,
      "grad_norm": 0.5702422691628836,
      "learning_rate": 0.00013266916780337433,
      "loss": 12.1578,
      "step": 15141
    },
    {
      "epoch": 0.8245424162599634,
      "grad_norm": 0.6843938991700117,
      "learning_rate": 0.00013266083326748596,
      "loss": 12.4301,
      "step": 15142
    },
    {
      "epoch": 0.8245968702565464,
      "grad_norm": 0.6428570758362845,
      "learning_rate": 0.00013265249847762146,
      "loss": 12.1424,
      "step": 15143
    },
    {
      "epoch": 0.8246513242531294,
      "grad_norm": 0.5287730976454701,
      "learning_rate": 0.00013264416343384568,
      "loss": 12.2458,
      "step": 15144
    },
    {
      "epoch": 0.8247057782497124,
      "grad_norm": 0.5955981955761246,
      "learning_rate": 0.0001326358281362234,
      "loss": 12.1539,
      "step": 15145
    },
    {
      "epoch": 0.8247602322462955,
      "grad_norm": 0.5641366752885426,
      "learning_rate": 0.0001326274925848194,
      "loss": 12.1635,
      "step": 15146
    },
    {
      "epoch": 0.8248146862428785,
      "grad_norm": 0.5511840238603661,
      "learning_rate": 0.00013261915677969862,
      "loss": 12.2499,
      "step": 15147
    },
    {
      "epoch": 0.8248691402394615,
      "grad_norm": 0.5706898292626512,
      "learning_rate": 0.00013261082072092578,
      "loss": 12.1368,
      "step": 15148
    },
    {
      "epoch": 0.8249235942360444,
      "grad_norm": 0.6157811933476967,
      "learning_rate": 0.00013260248440856572,
      "loss": 12.2726,
      "step": 15149
    },
    {
      "epoch": 0.8249780482326274,
      "grad_norm": 0.5371187183576617,
      "learning_rate": 0.00013259414784268328,
      "loss": 12.159,
      "step": 15150
    },
    {
      "epoch": 0.8250325022292105,
      "grad_norm": 0.5935211585790439,
      "learning_rate": 0.00013258581102334327,
      "loss": 12.0911,
      "step": 15151
    },
    {
      "epoch": 0.8250869562257935,
      "grad_norm": 0.5473734635512906,
      "learning_rate": 0.00013257747395061054,
      "loss": 12.2132,
      "step": 15152
    },
    {
      "epoch": 0.8251414102223765,
      "grad_norm": 0.5040749155652381,
      "learning_rate": 0.00013256913662454994,
      "loss": 12.1254,
      "step": 15153
    },
    {
      "epoch": 0.8251958642189595,
      "grad_norm": 0.5500514667341897,
      "learning_rate": 0.00013256079904522622,
      "loss": 12.2533,
      "step": 15154
    },
    {
      "epoch": 0.8252503182155425,
      "grad_norm": 0.5382832797192473,
      "learning_rate": 0.00013255246121270429,
      "loss": 12.1713,
      "step": 15155
    },
    {
      "epoch": 0.8253047722121255,
      "grad_norm": 0.5520042361371003,
      "learning_rate": 0.00013254412312704896,
      "loss": 12.1337,
      "step": 15156
    },
    {
      "epoch": 0.8253592262087086,
      "grad_norm": 0.569443223317474,
      "learning_rate": 0.00013253578478832507,
      "loss": 12.2426,
      "step": 15157
    },
    {
      "epoch": 0.8254136802052916,
      "grad_norm": 0.6446110993756711,
      "learning_rate": 0.00013252744619659745,
      "loss": 12.1518,
      "step": 15158
    },
    {
      "epoch": 0.8254681342018746,
      "grad_norm": 0.6038693948816372,
      "learning_rate": 0.00013251910735193097,
      "loss": 12.2149,
      "step": 15159
    },
    {
      "epoch": 0.8255225881984576,
      "grad_norm": 0.6525901872902011,
      "learning_rate": 0.00013251076825439043,
      "loss": 12.3306,
      "step": 15160
    },
    {
      "epoch": 0.8255770421950406,
      "grad_norm": 0.6481041452764463,
      "learning_rate": 0.00013250242890404076,
      "loss": 12.283,
      "step": 15161
    },
    {
      "epoch": 0.8256314961916236,
      "grad_norm": 0.5856391893384809,
      "learning_rate": 0.0001324940893009467,
      "loss": 12.0143,
      "step": 15162
    },
    {
      "epoch": 0.8256859501882067,
      "grad_norm": 0.5459209242515835,
      "learning_rate": 0.00013248574944517316,
      "loss": 12.1153,
      "step": 15163
    },
    {
      "epoch": 0.8257404041847897,
      "grad_norm": 0.6465502579114393,
      "learning_rate": 0.00013247740933678502,
      "loss": 12.1841,
      "step": 15164
    },
    {
      "epoch": 0.8257948581813727,
      "grad_norm": 0.7028110202838673,
      "learning_rate": 0.00013246906897584706,
      "loss": 12.1794,
      "step": 15165
    },
    {
      "epoch": 0.8258493121779557,
      "grad_norm": 0.5582118795351031,
      "learning_rate": 0.00013246072836242418,
      "loss": 12.1924,
      "step": 15166
    },
    {
      "epoch": 0.8259037661745386,
      "grad_norm": 0.6080292099406148,
      "learning_rate": 0.00013245238749658123,
      "loss": 12.2244,
      "step": 15167
    },
    {
      "epoch": 0.8259582201711216,
      "grad_norm": 0.6377228813808977,
      "learning_rate": 0.0001324440463783831,
      "loss": 12.0584,
      "step": 15168
    },
    {
      "epoch": 0.8260126741677047,
      "grad_norm": 0.5637368797575435,
      "learning_rate": 0.0001324357050078946,
      "loss": 12.2439,
      "step": 15169
    },
    {
      "epoch": 0.8260671281642877,
      "grad_norm": 0.5521350397117724,
      "learning_rate": 0.00013242736338518063,
      "loss": 12.1303,
      "step": 15170
    },
    {
      "epoch": 0.8261215821608707,
      "grad_norm": 0.5868245001879185,
      "learning_rate": 0.00013241902151030604,
      "loss": 12.1829,
      "step": 15171
    },
    {
      "epoch": 0.8261760361574537,
      "grad_norm": 0.5677736365759718,
      "learning_rate": 0.00013241067938333568,
      "loss": 12.1276,
      "step": 15172
    },
    {
      "epoch": 0.8262304901540367,
      "grad_norm": 0.5734079441240211,
      "learning_rate": 0.00013240233700433448,
      "loss": 12.2542,
      "step": 15173
    },
    {
      "epoch": 0.8262849441506197,
      "grad_norm": 0.5641702940489937,
      "learning_rate": 0.00013239399437336725,
      "loss": 12.3364,
      "step": 15174
    },
    {
      "epoch": 0.8263393981472028,
      "grad_norm": 0.5934181207445682,
      "learning_rate": 0.00013238565149049892,
      "loss": 12.2717,
      "step": 15175
    },
    {
      "epoch": 0.8263938521437858,
      "grad_norm": 0.5179909032091367,
      "learning_rate": 0.0001323773083557943,
      "loss": 12.2665,
      "step": 15176
    },
    {
      "epoch": 0.8264483061403688,
      "grad_norm": 0.5937253578214855,
      "learning_rate": 0.00013236896496931832,
      "loss": 12.3528,
      "step": 15177
    },
    {
      "epoch": 0.8265027601369518,
      "grad_norm": 0.502476889232368,
      "learning_rate": 0.0001323606213311358,
      "loss": 12.1983,
      "step": 15178
    },
    {
      "epoch": 0.8265572141335348,
      "grad_norm": 0.531901728632106,
      "learning_rate": 0.0001323522774413117,
      "loss": 11.9762,
      "step": 15179
    },
    {
      "epoch": 0.8266116681301178,
      "grad_norm": 0.5294521390225176,
      "learning_rate": 0.00013234393329991084,
      "loss": 12.1733,
      "step": 15180
    },
    {
      "epoch": 0.8266661221267009,
      "grad_norm": 0.5465460122389068,
      "learning_rate": 0.00013233558890699814,
      "loss": 12.0676,
      "step": 15181
    },
    {
      "epoch": 0.8267205761232839,
      "grad_norm": 0.549624763802986,
      "learning_rate": 0.00013232724426263853,
      "loss": 12.2706,
      "step": 15182
    },
    {
      "epoch": 0.8267750301198669,
      "grad_norm": 0.5855428195237342,
      "learning_rate": 0.00013231889936689677,
      "loss": 12.2152,
      "step": 15183
    },
    {
      "epoch": 0.8268294841164499,
      "grad_norm": 0.48878624781299707,
      "learning_rate": 0.00013231055421983787,
      "loss": 12.1706,
      "step": 15184
    },
    {
      "epoch": 0.8268839381130328,
      "grad_norm": 0.6439542590151076,
      "learning_rate": 0.00013230220882152666,
      "loss": 12.2569,
      "step": 15185
    },
    {
      "epoch": 0.826938392109616,
      "grad_norm": 0.5333101585897986,
      "learning_rate": 0.00013229386317202806,
      "loss": 12.0302,
      "step": 15186
    },
    {
      "epoch": 0.8269928461061989,
      "grad_norm": 0.6082116074025128,
      "learning_rate": 0.000132285517271407,
      "loss": 12.2354,
      "step": 15187
    },
    {
      "epoch": 0.8270473001027819,
      "grad_norm": 0.5978135139139774,
      "learning_rate": 0.00013227717111972834,
      "loss": 12.2528,
      "step": 15188
    },
    {
      "epoch": 0.8271017540993649,
      "grad_norm": 0.5878083936155488,
      "learning_rate": 0.0001322688247170569,
      "loss": 12.3464,
      "step": 15189
    },
    {
      "epoch": 0.8271562080959479,
      "grad_norm": 0.5858863888880815,
      "learning_rate": 0.00013226047806345773,
      "loss": 12.3016,
      "step": 15190
    },
    {
      "epoch": 0.8272106620925309,
      "grad_norm": 0.589348190036878,
      "learning_rate": 0.00013225213115899569,
      "loss": 12.1782,
      "step": 15191
    },
    {
      "epoch": 0.827265116089114,
      "grad_norm": 0.5216357234796167,
      "learning_rate": 0.00013224378400373564,
      "loss": 12.1173,
      "step": 15192
    },
    {
      "epoch": 0.827319570085697,
      "grad_norm": 0.5453169370096311,
      "learning_rate": 0.00013223543659774255,
      "loss": 12.1557,
      "step": 15193
    },
    {
      "epoch": 0.82737402408228,
      "grad_norm": 1.8532061690992983,
      "learning_rate": 0.00013222708894108126,
      "loss": 12.2979,
      "step": 15194
    },
    {
      "epoch": 0.827428478078863,
      "grad_norm": 0.5981247067276574,
      "learning_rate": 0.00013221874103381672,
      "loss": 12.1334,
      "step": 15195
    },
    {
      "epoch": 0.827482932075446,
      "grad_norm": 0.5542933873060595,
      "learning_rate": 0.00013221039287601388,
      "loss": 12.1865,
      "step": 15196
    },
    {
      "epoch": 0.827537386072029,
      "grad_norm": 0.5801329723495282,
      "learning_rate": 0.00013220204446773762,
      "loss": 12.2615,
      "step": 15197
    },
    {
      "epoch": 0.8275918400686121,
      "grad_norm": 0.7613352726613339,
      "learning_rate": 0.00013219369580905289,
      "loss": 12.1892,
      "step": 15198
    },
    {
      "epoch": 0.8276462940651951,
      "grad_norm": 0.6650532458101823,
      "learning_rate": 0.00013218534690002452,
      "loss": 12.2418,
      "step": 15199
    },
    {
      "epoch": 0.8277007480617781,
      "grad_norm": 0.6317919173574159,
      "learning_rate": 0.0001321769977407175,
      "loss": 12.1991,
      "step": 15200
    },
    {
      "epoch": 0.8277552020583611,
      "grad_norm": 0.5618309537187812,
      "learning_rate": 0.00013216864833119675,
      "loss": 12.1656,
      "step": 15201
    },
    {
      "epoch": 0.827809656054944,
      "grad_norm": 0.6232714602395312,
      "learning_rate": 0.00013216029867152724,
      "loss": 12.1143,
      "step": 15202
    },
    {
      "epoch": 0.827864110051527,
      "grad_norm": 0.5737033519336524,
      "learning_rate": 0.00013215194876177385,
      "loss": 12.1755,
      "step": 15203
    },
    {
      "epoch": 0.8279185640481102,
      "grad_norm": 0.70282100669915,
      "learning_rate": 0.00013214359860200148,
      "loss": 12.1138,
      "step": 15204
    },
    {
      "epoch": 0.8279730180446931,
      "grad_norm": 0.757833350297887,
      "learning_rate": 0.00013213524819227513,
      "loss": 12.187,
      "step": 15205
    },
    {
      "epoch": 0.8280274720412761,
      "grad_norm": 0.5789596533234671,
      "learning_rate": 0.00013212689753265965,
      "loss": 12.2284,
      "step": 15206
    },
    {
      "epoch": 0.8280819260378591,
      "grad_norm": 0.641352209545092,
      "learning_rate": 0.00013211854662322007,
      "loss": 12.3184,
      "step": 15207
    },
    {
      "epoch": 0.8281363800344421,
      "grad_norm": 0.5660627888936081,
      "learning_rate": 0.00013211019546402128,
      "loss": 12.1388,
      "step": 15208
    },
    {
      "epoch": 0.8281908340310251,
      "grad_norm": 0.5537899501912319,
      "learning_rate": 0.0001321018440551282,
      "loss": 12.2371,
      "step": 15209
    },
    {
      "epoch": 0.8282452880276082,
      "grad_norm": 0.7017780050980083,
      "learning_rate": 0.0001320934923966058,
      "loss": 12.2134,
      "step": 15210
    },
    {
      "epoch": 0.8282997420241912,
      "grad_norm": 0.629825583850304,
      "learning_rate": 0.00013208514048851903,
      "loss": 12.1223,
      "step": 15211
    },
    {
      "epoch": 0.8283541960207742,
      "grad_norm": 0.5852813246540232,
      "learning_rate": 0.00013207678833093285,
      "loss": 12.193,
      "step": 15212
    },
    {
      "epoch": 0.8284086500173572,
      "grad_norm": 0.654363836888728,
      "learning_rate": 0.00013206843592391217,
      "loss": 12.2933,
      "step": 15213
    },
    {
      "epoch": 0.8284631040139402,
      "grad_norm": 0.6444913765173723,
      "learning_rate": 0.00013206008326752198,
      "loss": 12.1784,
      "step": 15214
    },
    {
      "epoch": 0.8285175580105232,
      "grad_norm": 0.7632623438059727,
      "learning_rate": 0.00013205173036182717,
      "loss": 12.2421,
      "step": 15215
    },
    {
      "epoch": 0.8285720120071063,
      "grad_norm": 0.6093109181145168,
      "learning_rate": 0.00013204337720689274,
      "loss": 12.2275,
      "step": 15216
    },
    {
      "epoch": 0.8286264660036893,
      "grad_norm": 0.5887353236091951,
      "learning_rate": 0.0001320350238027836,
      "loss": 12.2388,
      "step": 15217
    },
    {
      "epoch": 0.8286809200002723,
      "grad_norm": 0.5890757678442023,
      "learning_rate": 0.00013202667014956481,
      "loss": 12.2197,
      "step": 15218
    },
    {
      "epoch": 0.8287353739968553,
      "grad_norm": 0.8375274103573752,
      "learning_rate": 0.0001320183162473012,
      "loss": 12.2183,
      "step": 15219
    },
    {
      "epoch": 0.8287898279934383,
      "grad_norm": 0.6257245685057095,
      "learning_rate": 0.0001320099620960578,
      "loss": 12.1336,
      "step": 15220
    },
    {
      "epoch": 0.8288442819900214,
      "grad_norm": 0.5952583992488294,
      "learning_rate": 0.00013200160769589962,
      "loss": 12.1276,
      "step": 15221
    },
    {
      "epoch": 0.8288987359866044,
      "grad_norm": 0.6480649554934045,
      "learning_rate": 0.00013199325304689153,
      "loss": 12.1835,
      "step": 15222
    },
    {
      "epoch": 0.8289531899831873,
      "grad_norm": 0.6080343846993349,
      "learning_rate": 0.00013198489814909855,
      "loss": 12.1816,
      "step": 15223
    },
    {
      "epoch": 0.8290076439797703,
      "grad_norm": 0.5783908524821597,
      "learning_rate": 0.00013197654300258564,
      "loss": 12.3538,
      "step": 15224
    },
    {
      "epoch": 0.8290620979763533,
      "grad_norm": 0.6053366205027224,
      "learning_rate": 0.0001319681876074178,
      "loss": 12.1792,
      "step": 15225
    },
    {
      "epoch": 0.8291165519729363,
      "grad_norm": 0.6561232963435246,
      "learning_rate": 0.0001319598319636599,
      "loss": 12.3573,
      "step": 15226
    },
    {
      "epoch": 0.8291710059695194,
      "grad_norm": 0.5575796826423648,
      "learning_rate": 0.00013195147607137703,
      "loss": 12.3334,
      "step": 15227
    },
    {
      "epoch": 0.8292254599661024,
      "grad_norm": 0.613532357704036,
      "learning_rate": 0.00013194311993063412,
      "loss": 12.1937,
      "step": 15228
    },
    {
      "epoch": 0.8292799139626854,
      "grad_norm": 0.6084414020114763,
      "learning_rate": 0.00013193476354149617,
      "loss": 12.213,
      "step": 15229
    },
    {
      "epoch": 0.8293343679592684,
      "grad_norm": 0.5463433763684223,
      "learning_rate": 0.00013192640690402811,
      "loss": 12.2368,
      "step": 15230
    },
    {
      "epoch": 0.8293888219558514,
      "grad_norm": 0.5157030945118751,
      "learning_rate": 0.00013191805001829495,
      "loss": 12.1926,
      "step": 15231
    },
    {
      "epoch": 0.8294432759524344,
      "grad_norm": 0.6152722825800643,
      "learning_rate": 0.00013190969288436172,
      "loss": 12.4055,
      "step": 15232
    },
    {
      "epoch": 0.8294977299490175,
      "grad_norm": 0.6345939630072156,
      "learning_rate": 0.00013190133550229334,
      "loss": 12.1898,
      "step": 15233
    },
    {
      "epoch": 0.8295521839456005,
      "grad_norm": 0.5770984625351699,
      "learning_rate": 0.00013189297787215485,
      "loss": 12.2821,
      "step": 15234
    },
    {
      "epoch": 0.8296066379421835,
      "grad_norm": 0.5530313175997955,
      "learning_rate": 0.00013188461999401118,
      "loss": 12.2363,
      "step": 15235
    },
    {
      "epoch": 0.8296610919387665,
      "grad_norm": 0.5676731684196313,
      "learning_rate": 0.0001318762618679274,
      "loss": 12.1333,
      "step": 15236
    },
    {
      "epoch": 0.8297155459353495,
      "grad_norm": 0.5897430392692317,
      "learning_rate": 0.00013186790349396842,
      "loss": 12.2853,
      "step": 15237
    },
    {
      "epoch": 0.8297699999319325,
      "grad_norm": 0.6143112745421574,
      "learning_rate": 0.0001318595448721993,
      "loss": 12.1977,
      "step": 15238
    },
    {
      "epoch": 0.8298244539285156,
      "grad_norm": 0.5961912593192616,
      "learning_rate": 0.000131851186002685,
      "loss": 12.2267,
      "step": 15239
    },
    {
      "epoch": 0.8298789079250986,
      "grad_norm": 0.5864432196548776,
      "learning_rate": 0.00013184282688549057,
      "loss": 12.0463,
      "step": 15240
    },
    {
      "epoch": 0.8299333619216815,
      "grad_norm": 0.550662140713574,
      "learning_rate": 0.00013183446752068094,
      "loss": 12.1639,
      "step": 15241
    },
    {
      "epoch": 0.8299878159182645,
      "grad_norm": 0.5855212190237381,
      "learning_rate": 0.00013182610790832118,
      "loss": 12.1433,
      "step": 15242
    },
    {
      "epoch": 0.8300422699148475,
      "grad_norm": 0.5995853240988076,
      "learning_rate": 0.00013181774804847627,
      "loss": 12.1656,
      "step": 15243
    },
    {
      "epoch": 0.8300967239114305,
      "grad_norm": 0.5616551083670744,
      "learning_rate": 0.00013180938794121118,
      "loss": 12.2573,
      "step": 15244
    },
    {
      "epoch": 0.8301511779080136,
      "grad_norm": 0.7404378825252865,
      "learning_rate": 0.000131801027586591,
      "loss": 12.3811,
      "step": 15245
    },
    {
      "epoch": 0.8302056319045966,
      "grad_norm": 0.6166144410712183,
      "learning_rate": 0.00013179266698468064,
      "loss": 12.3776,
      "step": 15246
    },
    {
      "epoch": 0.8302600859011796,
      "grad_norm": 0.5533530192204831,
      "learning_rate": 0.00013178430613554522,
      "loss": 12.1586,
      "step": 15247
    },
    {
      "epoch": 0.8303145398977626,
      "grad_norm": 0.5844910546885216,
      "learning_rate": 0.0001317759450392497,
      "loss": 12.3598,
      "step": 15248
    },
    {
      "epoch": 0.8303689938943456,
      "grad_norm": 0.6002406804984849,
      "learning_rate": 0.0001317675836958591,
      "loss": 12.2845,
      "step": 15249
    },
    {
      "epoch": 0.8304234478909286,
      "grad_norm": 0.4980976925076202,
      "learning_rate": 0.0001317592221054384,
      "loss": 12.1897,
      "step": 15250
    },
    {
      "epoch": 0.8304779018875117,
      "grad_norm": 0.6031308926270025,
      "learning_rate": 0.0001317508602680527,
      "loss": 12.2017,
      "step": 15251
    },
    {
      "epoch": 0.8305323558840947,
      "grad_norm": 0.6147733627442982,
      "learning_rate": 0.00013174249818376699,
      "loss": 12.2611,
      "step": 15252
    },
    {
      "epoch": 0.8305868098806777,
      "grad_norm": 0.5800706246651661,
      "learning_rate": 0.00013173413585264632,
      "loss": 12.3297,
      "step": 15253
    },
    {
      "epoch": 0.8306412638772607,
      "grad_norm": 0.5807351910438452,
      "learning_rate": 0.00013172577327475563,
      "loss": 12.3804,
      "step": 15254
    },
    {
      "epoch": 0.8306957178738437,
      "grad_norm": 0.617417880066591,
      "learning_rate": 0.00013171741045016002,
      "loss": 12.1884,
      "step": 15255
    },
    {
      "epoch": 0.8307501718704268,
      "grad_norm": 0.6256639537422343,
      "learning_rate": 0.00013170904737892452,
      "loss": 12.2294,
      "step": 15256
    },
    {
      "epoch": 0.8308046258670098,
      "grad_norm": 0.5417802792206997,
      "learning_rate": 0.00013170068406111413,
      "loss": 12.2778,
      "step": 15257
    },
    {
      "epoch": 0.8308590798635928,
      "grad_norm": 0.6743405301489448,
      "learning_rate": 0.00013169232049679394,
      "loss": 12.1945,
      "step": 15258
    },
    {
      "epoch": 0.8309135338601757,
      "grad_norm": 0.637377462288876,
      "learning_rate": 0.00013168395668602893,
      "loss": 12.2543,
      "step": 15259
    },
    {
      "epoch": 0.8309679878567587,
      "grad_norm": 0.5322625723022238,
      "learning_rate": 0.00013167559262888413,
      "loss": 12.1928,
      "step": 15260
    },
    {
      "epoch": 0.8310224418533417,
      "grad_norm": 0.6171535140221914,
      "learning_rate": 0.00013166722832542465,
      "loss": 12.2262,
      "step": 15261
    },
    {
      "epoch": 0.8310768958499248,
      "grad_norm": 0.8597285154023284,
      "learning_rate": 0.00013165886377571547,
      "loss": 12.2184,
      "step": 15262
    },
    {
      "epoch": 0.8311313498465078,
      "grad_norm": 0.5779101303293248,
      "learning_rate": 0.00013165049897982168,
      "loss": 12.2524,
      "step": 15263
    },
    {
      "epoch": 0.8311858038430908,
      "grad_norm": 0.5739459923844983,
      "learning_rate": 0.0001316421339378083,
      "loss": 12.2599,
      "step": 15264
    },
    {
      "epoch": 0.8312402578396738,
      "grad_norm": 0.6661241609743747,
      "learning_rate": 0.00013163376864974038,
      "loss": 12.2557,
      "step": 15265
    },
    {
      "epoch": 0.8312947118362568,
      "grad_norm": 0.5379554045635468,
      "learning_rate": 0.00013162540311568294,
      "loss": 12.1778,
      "step": 15266
    },
    {
      "epoch": 0.8313491658328398,
      "grad_norm": 0.5524540929241767,
      "learning_rate": 0.00013161703733570107,
      "loss": 12.1688,
      "step": 15267
    },
    {
      "epoch": 0.8314036198294229,
      "grad_norm": 0.6297163416154375,
      "learning_rate": 0.00013160867130985985,
      "loss": 12.2242,
      "step": 15268
    },
    {
      "epoch": 0.8314580738260059,
      "grad_norm": 0.9573883676541012,
      "learning_rate": 0.00013160030503822428,
      "loss": 12.1995,
      "step": 15269
    },
    {
      "epoch": 0.8315125278225889,
      "grad_norm": 0.6181746204171851,
      "learning_rate": 0.00013159193852085944,
      "loss": 12.1728,
      "step": 15270
    },
    {
      "epoch": 0.8315669818191719,
      "grad_norm": 0.5946896226480637,
      "learning_rate": 0.00013158357175783038,
      "loss": 12.051,
      "step": 15271
    },
    {
      "epoch": 0.8316214358157549,
      "grad_norm": 0.5841329282239158,
      "learning_rate": 0.00013157520474920214,
      "loss": 12.1945,
      "step": 15272
    },
    {
      "epoch": 0.8316758898123379,
      "grad_norm": 0.6162149903340335,
      "learning_rate": 0.00013156683749503988,
      "loss": 12.186,
      "step": 15273
    },
    {
      "epoch": 0.831730343808921,
      "grad_norm": 0.5633195206483552,
      "learning_rate": 0.00013155846999540858,
      "loss": 12.1581,
      "step": 15274
    },
    {
      "epoch": 0.831784797805504,
      "grad_norm": 0.5604639867880522,
      "learning_rate": 0.0001315501022503733,
      "loss": 12.0926,
      "step": 15275
    },
    {
      "epoch": 0.831839251802087,
      "grad_norm": 0.5308652778688698,
      "learning_rate": 0.00013154173425999915,
      "loss": 12.0024,
      "step": 15276
    },
    {
      "epoch": 0.83189370579867,
      "grad_norm": 0.5447564444093732,
      "learning_rate": 0.00013153336602435113,
      "loss": 12.1712,
      "step": 15277
    },
    {
      "epoch": 0.8319481597952529,
      "grad_norm": 0.5386724699747754,
      "learning_rate": 0.00013152499754349445,
      "loss": 11.9927,
      "step": 15278
    },
    {
      "epoch": 0.8320026137918359,
      "grad_norm": 0.5797490134641445,
      "learning_rate": 0.00013151662881749407,
      "loss": 12.2172,
      "step": 15279
    },
    {
      "epoch": 0.832057067788419,
      "grad_norm": 0.626816413248062,
      "learning_rate": 0.0001315082598464151,
      "loss": 12.1729,
      "step": 15280
    },
    {
      "epoch": 0.832111521785002,
      "grad_norm": 0.5420185299027364,
      "learning_rate": 0.0001314998906303226,
      "loss": 12.1501,
      "step": 15281
    },
    {
      "epoch": 0.832165975781585,
      "grad_norm": 0.6286576341012895,
      "learning_rate": 0.00013149152116928169,
      "loss": 12.3009,
      "step": 15282
    },
    {
      "epoch": 0.832220429778168,
      "grad_norm": 0.5347714267872193,
      "learning_rate": 0.00013148315146335743,
      "loss": 12.094,
      "step": 15283
    },
    {
      "epoch": 0.832274883774751,
      "grad_norm": 0.5982064880757932,
      "learning_rate": 0.0001314747815126149,
      "loss": 12.0955,
      "step": 15284
    },
    {
      "epoch": 0.8323293377713341,
      "grad_norm": 0.5360239055325735,
      "learning_rate": 0.00013146641131711918,
      "loss": 12.207,
      "step": 15285
    },
    {
      "epoch": 0.8323837917679171,
      "grad_norm": 0.6032727350194691,
      "learning_rate": 0.0001314580408769354,
      "loss": 12.2156,
      "step": 15286
    },
    {
      "epoch": 0.8324382457645001,
      "grad_norm": 0.5264867464022237,
      "learning_rate": 0.00013144967019212858,
      "loss": 12.2824,
      "step": 15287
    },
    {
      "epoch": 0.8324926997610831,
      "grad_norm": 0.614179478401167,
      "learning_rate": 0.00013144129926276387,
      "loss": 12.2034,
      "step": 15288
    },
    {
      "epoch": 0.8325471537576661,
      "grad_norm": 0.585713319770856,
      "learning_rate": 0.00013143292808890633,
      "loss": 12.1851,
      "step": 15289
    },
    {
      "epoch": 0.8326016077542491,
      "grad_norm": 0.5740739803646323,
      "learning_rate": 0.00013142455667062108,
      "loss": 12.01,
      "step": 15290
    },
    {
      "epoch": 0.8326560617508322,
      "grad_norm": 0.5796548730360052,
      "learning_rate": 0.00013141618500797322,
      "loss": 12.244,
      "step": 15291
    },
    {
      "epoch": 0.8327105157474152,
      "grad_norm": 0.5597923489671587,
      "learning_rate": 0.0001314078131010278,
      "loss": 12.2179,
      "step": 15292
    },
    {
      "epoch": 0.8327649697439982,
      "grad_norm": 0.6321308750754813,
      "learning_rate": 0.00013139944094985,
      "loss": 12.2756,
      "step": 15293
    },
    {
      "epoch": 0.8328194237405812,
      "grad_norm": 0.5446385936148014,
      "learning_rate": 0.00013139106855450486,
      "loss": 12.2949,
      "step": 15294
    },
    {
      "epoch": 0.8328738777371641,
      "grad_norm": 0.6349345381171592,
      "learning_rate": 0.00013138269591505752,
      "loss": 12.0296,
      "step": 15295
    },
    {
      "epoch": 0.8329283317337471,
      "grad_norm": 0.6889425832186824,
      "learning_rate": 0.00013137432303157305,
      "loss": 12.3,
      "step": 15296
    },
    {
      "epoch": 0.8329827857303302,
      "grad_norm": 0.5147927373090182,
      "learning_rate": 0.0001313659499041166,
      "loss": 12.211,
      "step": 15297
    },
    {
      "epoch": 0.8330372397269132,
      "grad_norm": 0.6220664367543297,
      "learning_rate": 0.00013135757653275326,
      "loss": 12.3032,
      "step": 15298
    },
    {
      "epoch": 0.8330916937234962,
      "grad_norm": 0.5922621718306847,
      "learning_rate": 0.00013134920291754814,
      "loss": 12.1767,
      "step": 15299
    },
    {
      "epoch": 0.8331461477200792,
      "grad_norm": 0.598265551750958,
      "learning_rate": 0.00013134082905856637,
      "loss": 12.1843,
      "step": 15300
    },
    {
      "epoch": 0.8332006017166622,
      "grad_norm": 0.6566281451981195,
      "learning_rate": 0.00013133245495587306,
      "loss": 12.1329,
      "step": 15301
    },
    {
      "epoch": 0.8332550557132452,
      "grad_norm": 0.6198392400520069,
      "learning_rate": 0.0001313240806095333,
      "loss": 12.2163,
      "step": 15302
    },
    {
      "epoch": 0.8333095097098283,
      "grad_norm": 0.5089681050707612,
      "learning_rate": 0.0001313157060196123,
      "loss": 12.1203,
      "step": 15303
    },
    {
      "epoch": 0.8333639637064113,
      "grad_norm": 0.5946276409729907,
      "learning_rate": 0.00013130733118617505,
      "loss": 12.3139,
      "step": 15304
    },
    {
      "epoch": 0.8334184177029943,
      "grad_norm": 0.6768073905113495,
      "learning_rate": 0.00013129895610928678,
      "loss": 12.0899,
      "step": 15305
    },
    {
      "epoch": 0.8334728716995773,
      "grad_norm": 0.5706898431899106,
      "learning_rate": 0.00013129058078901256,
      "loss": 12.2588,
      "step": 15306
    },
    {
      "epoch": 0.8335273256961603,
      "grad_norm": 0.5295503593770585,
      "learning_rate": 0.00013128220522541753,
      "loss": 12.2527,
      "step": 15307
    },
    {
      "epoch": 0.8335817796927433,
      "grad_norm": 0.5914096601533734,
      "learning_rate": 0.00013127382941856687,
      "loss": 12.304,
      "step": 15308
    },
    {
      "epoch": 0.8336362336893264,
      "grad_norm": 0.555249289629514,
      "learning_rate": 0.0001312654533685256,
      "loss": 12.0958,
      "step": 15309
    },
    {
      "epoch": 0.8336906876859094,
      "grad_norm": 0.553953495231632,
      "learning_rate": 0.00013125707707535897,
      "loss": 12.0996,
      "step": 15310
    },
    {
      "epoch": 0.8337451416824924,
      "grad_norm": 0.5426767378037972,
      "learning_rate": 0.00013124870053913206,
      "loss": 12.1959,
      "step": 15311
    },
    {
      "epoch": 0.8337995956790754,
      "grad_norm": 0.5638715609519398,
      "learning_rate": 0.00013124032375991,
      "loss": 12.1801,
      "step": 15312
    },
    {
      "epoch": 0.8338540496756583,
      "grad_norm": 0.5968174649338724,
      "learning_rate": 0.00013123194673775798,
      "loss": 12.2152,
      "step": 15313
    },
    {
      "epoch": 0.8339085036722413,
      "grad_norm": 0.5734240715647675,
      "learning_rate": 0.00013122356947274107,
      "loss": 12.1478,
      "step": 15314
    },
    {
      "epoch": 0.8339629576688244,
      "grad_norm": 0.611642059701702,
      "learning_rate": 0.00013121519196492444,
      "loss": 12.1398,
      "step": 15315
    },
    {
      "epoch": 0.8340174116654074,
      "grad_norm": 0.6082665454481369,
      "learning_rate": 0.00013120681421437325,
      "loss": 12.0678,
      "step": 15316
    },
    {
      "epoch": 0.8340718656619904,
      "grad_norm": 0.6145617171266933,
      "learning_rate": 0.00013119843622115264,
      "loss": 12.2043,
      "step": 15317
    },
    {
      "epoch": 0.8341263196585734,
      "grad_norm": 0.5730138743482941,
      "learning_rate": 0.0001311900579853278,
      "loss": 12.3287,
      "step": 15318
    },
    {
      "epoch": 0.8341807736551564,
      "grad_norm": 0.5975019952444647,
      "learning_rate": 0.00013118167950696382,
      "loss": 12.1496,
      "step": 15319
    },
    {
      "epoch": 0.8342352276517395,
      "grad_norm": 0.5874154839513164,
      "learning_rate": 0.00013117330078612582,
      "loss": 12.1702,
      "step": 15320
    },
    {
      "epoch": 0.8342896816483225,
      "grad_norm": 0.5368851943471138,
      "learning_rate": 0.00013116492182287904,
      "loss": 12.1665,
      "step": 15321
    },
    {
      "epoch": 0.8343441356449055,
      "grad_norm": 0.6005496640293485,
      "learning_rate": 0.0001311565426172886,
      "loss": 12.1575,
      "step": 15322
    },
    {
      "epoch": 0.8343985896414885,
      "grad_norm": 0.5216150312343807,
      "learning_rate": 0.00013114816316941967,
      "loss": 12.1666,
      "step": 15323
    },
    {
      "epoch": 0.8344530436380715,
      "grad_norm": 0.5951216910198904,
      "learning_rate": 0.0001311397834793374,
      "loss": 12.1675,
      "step": 15324
    },
    {
      "epoch": 0.8345074976346545,
      "grad_norm": 0.6309894694156547,
      "learning_rate": 0.00013113140354710693,
      "loss": 12.2063,
      "step": 15325
    },
    {
      "epoch": 0.8345619516312376,
      "grad_norm": 0.5681904907277748,
      "learning_rate": 0.00013112302337279342,
      "loss": 12.1512,
      "step": 15326
    },
    {
      "epoch": 0.8346164056278206,
      "grad_norm": 0.5735928927624965,
      "learning_rate": 0.00013111464295646212,
      "loss": 12.3276,
      "step": 15327
    },
    {
      "epoch": 0.8346708596244036,
      "grad_norm": 0.6591312024956174,
      "learning_rate": 0.00013110626229817813,
      "loss": 12.224,
      "step": 15328
    },
    {
      "epoch": 0.8347253136209866,
      "grad_norm": 0.6444292844076731,
      "learning_rate": 0.0001310978813980066,
      "loss": 12.2489,
      "step": 15329
    },
    {
      "epoch": 0.8347797676175696,
      "grad_norm": 0.6733190185260566,
      "learning_rate": 0.00013108950025601275,
      "loss": 12.1378,
      "step": 15330
    },
    {
      "epoch": 0.8348342216141525,
      "grad_norm": 0.5369684470576751,
      "learning_rate": 0.00013108111887226174,
      "loss": 12.0915,
      "step": 15331
    },
    {
      "epoch": 0.8348886756107357,
      "grad_norm": 0.6256900431355148,
      "learning_rate": 0.0001310727372468187,
      "loss": 12.2504,
      "step": 15332
    },
    {
      "epoch": 0.8349431296073186,
      "grad_norm": 0.6406448205533162,
      "learning_rate": 0.0001310643553797489,
      "loss": 12.1589,
      "step": 15333
    },
    {
      "epoch": 0.8349975836039016,
      "grad_norm": 0.5376786196408638,
      "learning_rate": 0.0001310559732711174,
      "loss": 12.2238,
      "step": 15334
    },
    {
      "epoch": 0.8350520376004846,
      "grad_norm": 0.5783970591286314,
      "learning_rate": 0.0001310475909209895,
      "loss": 12.3498,
      "step": 15335
    },
    {
      "epoch": 0.8351064915970676,
      "grad_norm": 0.5942024240935793,
      "learning_rate": 0.0001310392083294303,
      "loss": 12.188,
      "step": 15336
    },
    {
      "epoch": 0.8351609455936506,
      "grad_norm": 0.60545358347949,
      "learning_rate": 0.00013103082549650497,
      "loss": 12.2307,
      "step": 15337
    },
    {
      "epoch": 0.8352153995902337,
      "grad_norm": 0.5636018467561122,
      "learning_rate": 0.00013102244242227878,
      "loss": 12.2584,
      "step": 15338
    },
    {
      "epoch": 0.8352698535868167,
      "grad_norm": 0.580316800840767,
      "learning_rate": 0.0001310140591068169,
      "loss": 12.1868,
      "step": 15339
    },
    {
      "epoch": 0.8353243075833997,
      "grad_norm": 0.5418777539446611,
      "learning_rate": 0.00013100567555018446,
      "loss": 12.1379,
      "step": 15340
    },
    {
      "epoch": 0.8353787615799827,
      "grad_norm": 0.6088070564598652,
      "learning_rate": 0.0001309972917524467,
      "loss": 12.1415,
      "step": 15341
    },
    {
      "epoch": 0.8354332155765657,
      "grad_norm": 0.6252243895763365,
      "learning_rate": 0.00013098890771366878,
      "loss": 12.2393,
      "step": 15342
    },
    {
      "epoch": 0.8354876695731487,
      "grad_norm": 0.5364906707347072,
      "learning_rate": 0.00013098052343391597,
      "loss": 12.147,
      "step": 15343
    },
    {
      "epoch": 0.8355421235697318,
      "grad_norm": 0.5666736471551951,
      "learning_rate": 0.00013097213891325336,
      "loss": 12.1947,
      "step": 15344
    },
    {
      "epoch": 0.8355965775663148,
      "grad_norm": 0.5833266248887412,
      "learning_rate": 0.00013096375415174623,
      "loss": 12.1789,
      "step": 15345
    },
    {
      "epoch": 0.8356510315628978,
      "grad_norm": 0.6036746488446625,
      "learning_rate": 0.00013095536914945973,
      "loss": 12.2361,
      "step": 15346
    },
    {
      "epoch": 0.8357054855594808,
      "grad_norm": 0.5709164918817299,
      "learning_rate": 0.00013094698390645913,
      "loss": 12.1975,
      "step": 15347
    },
    {
      "epoch": 0.8357599395560638,
      "grad_norm": 0.5415393121643536,
      "learning_rate": 0.00013093859842280955,
      "loss": 12.1258,
      "step": 15348
    },
    {
      "epoch": 0.8358143935526468,
      "grad_norm": 0.6444672849336692,
      "learning_rate": 0.00013093021269857625,
      "loss": 12.1467,
      "step": 15349
    },
    {
      "epoch": 0.8358688475492299,
      "grad_norm": 0.6121385670183787,
      "learning_rate": 0.00013092182673382445,
      "loss": 12.2194,
      "step": 15350
    },
    {
      "epoch": 0.8359233015458128,
      "grad_norm": 0.5742302760383521,
      "learning_rate": 0.0001309134405286193,
      "loss": 12.1077,
      "step": 15351
    },
    {
      "epoch": 0.8359777555423958,
      "grad_norm": 0.5597706966472342,
      "learning_rate": 0.00013090505408302612,
      "loss": 12.2605,
      "step": 15352
    },
    {
      "epoch": 0.8360322095389788,
      "grad_norm": 0.6120210383203647,
      "learning_rate": 0.00013089666739711,
      "loss": 12.3513,
      "step": 15353
    },
    {
      "epoch": 0.8360866635355618,
      "grad_norm": 0.6357714430157161,
      "learning_rate": 0.00013088828047093623,
      "loss": 12.22,
      "step": 15354
    },
    {
      "epoch": 0.8361411175321449,
      "grad_norm": 0.560241191145675,
      "learning_rate": 0.00013087989330457,
      "loss": 12.1131,
      "step": 15355
    },
    {
      "epoch": 0.8361955715287279,
      "grad_norm": 0.558665134696943,
      "learning_rate": 0.00013087150589807656,
      "loss": 12.2699,
      "step": 15356
    },
    {
      "epoch": 0.8362500255253109,
      "grad_norm": 0.6030718684554315,
      "learning_rate": 0.00013086311825152112,
      "loss": 12.164,
      "step": 15357
    },
    {
      "epoch": 0.8363044795218939,
      "grad_norm": 0.5626851183441052,
      "learning_rate": 0.00013085473036496888,
      "loss": 12.2383,
      "step": 15358
    },
    {
      "epoch": 0.8363589335184769,
      "grad_norm": 0.60581591719412,
      "learning_rate": 0.00013084634223848506,
      "loss": 12.1086,
      "step": 15359
    },
    {
      "epoch": 0.8364133875150599,
      "grad_norm": 0.6441289035553069,
      "learning_rate": 0.00013083795387213495,
      "loss": 12.1012,
      "step": 15360
    },
    {
      "epoch": 0.836467841511643,
      "grad_norm": 0.5432182200872989,
      "learning_rate": 0.0001308295652659837,
      "loss": 12.2727,
      "step": 15361
    },
    {
      "epoch": 0.836522295508226,
      "grad_norm": 0.5112965914274825,
      "learning_rate": 0.00013082117642009662,
      "loss": 12.1125,
      "step": 15362
    },
    {
      "epoch": 0.836576749504809,
      "grad_norm": 0.5197450256056448,
      "learning_rate": 0.0001308127873345389,
      "loss": 12.117,
      "step": 15363
    },
    {
      "epoch": 0.836631203501392,
      "grad_norm": 0.5860435056081319,
      "learning_rate": 0.00013080439800937575,
      "loss": 12.2292,
      "step": 15364
    },
    {
      "epoch": 0.836685657497975,
      "grad_norm": 0.5629178413356406,
      "learning_rate": 0.00013079600844467242,
      "loss": 12.196,
      "step": 15365
    },
    {
      "epoch": 0.836740111494558,
      "grad_norm": 0.5758380845386492,
      "learning_rate": 0.0001307876186404942,
      "loss": 12.2736,
      "step": 15366
    },
    {
      "epoch": 0.8367945654911411,
      "grad_norm": 0.5345267079541275,
      "learning_rate": 0.0001307792285969063,
      "loss": 12.2351,
      "step": 15367
    },
    {
      "epoch": 0.836849019487724,
      "grad_norm": 0.6010343621198327,
      "learning_rate": 0.00013077083831397395,
      "loss": 12.1085,
      "step": 15368
    },
    {
      "epoch": 0.836903473484307,
      "grad_norm": 0.5567987453339249,
      "learning_rate": 0.00013076244779176244,
      "loss": 12.2886,
      "step": 15369
    },
    {
      "epoch": 0.83695792748089,
      "grad_norm": 0.575988620818282,
      "learning_rate": 0.0001307540570303369,
      "loss": 12.0892,
      "step": 15370
    },
    {
      "epoch": 0.837012381477473,
      "grad_norm": 0.5461307537724167,
      "learning_rate": 0.00013074566602976268,
      "loss": 12.1839,
      "step": 15371
    },
    {
      "epoch": 0.837066835474056,
      "grad_norm": 0.5741964053298968,
      "learning_rate": 0.000130737274790105,
      "loss": 12.288,
      "step": 15372
    },
    {
      "epoch": 0.8371212894706391,
      "grad_norm": 0.5969391944247079,
      "learning_rate": 0.00013072888331142914,
      "loss": 12.0641,
      "step": 15373
    },
    {
      "epoch": 0.8371757434672221,
      "grad_norm": 0.5987646358058737,
      "learning_rate": 0.00013072049159380033,
      "loss": 12.2112,
      "step": 15374
    },
    {
      "epoch": 0.8372301974638051,
      "grad_norm": 0.5455711004770989,
      "learning_rate": 0.0001307120996372838,
      "loss": 12.2458,
      "step": 15375
    },
    {
      "epoch": 0.8372846514603881,
      "grad_norm": 0.5414312185891778,
      "learning_rate": 0.00013070370744194487,
      "loss": 12.127,
      "step": 15376
    },
    {
      "epoch": 0.8373391054569711,
      "grad_norm": 0.6305935747972193,
      "learning_rate": 0.00013069531500784873,
      "loss": 12.2119,
      "step": 15377
    },
    {
      "epoch": 0.8373935594535541,
      "grad_norm": 0.5685400442083921,
      "learning_rate": 0.00013068692233506068,
      "loss": 12.2422,
      "step": 15378
    },
    {
      "epoch": 0.8374480134501372,
      "grad_norm": 0.5949894077370246,
      "learning_rate": 0.000130678529423646,
      "loss": 12.1692,
      "step": 15379
    },
    {
      "epoch": 0.8375024674467202,
      "grad_norm": 0.6653462622638929,
      "learning_rate": 0.00013067013627366991,
      "loss": 12.2144,
      "step": 15380
    },
    {
      "epoch": 0.8375569214433032,
      "grad_norm": 0.5731677512947161,
      "learning_rate": 0.00013066174288519768,
      "loss": 12.1916,
      "step": 15381
    },
    {
      "epoch": 0.8376113754398862,
      "grad_norm": 0.6319538726760888,
      "learning_rate": 0.0001306533492582946,
      "loss": 12.2809,
      "step": 15382
    },
    {
      "epoch": 0.8376658294364692,
      "grad_norm": 0.6846801198465237,
      "learning_rate": 0.00013064495539302594,
      "loss": 12.0877,
      "step": 15383
    },
    {
      "epoch": 0.8377202834330522,
      "grad_norm": 0.534280489313378,
      "learning_rate": 0.000130636561289457,
      "loss": 12.076,
      "step": 15384
    },
    {
      "epoch": 0.8377747374296353,
      "grad_norm": 0.5713743213754297,
      "learning_rate": 0.000130628166947653,
      "loss": 12.1155,
      "step": 15385
    },
    {
      "epoch": 0.8378291914262183,
      "grad_norm": 0.6238325355933657,
      "learning_rate": 0.0001306197723676792,
      "loss": 12.1667,
      "step": 15386
    },
    {
      "epoch": 0.8378836454228012,
      "grad_norm": 0.6029480253780464,
      "learning_rate": 0.00013061137754960094,
      "loss": 12.1733,
      "step": 15387
    },
    {
      "epoch": 0.8379380994193842,
      "grad_norm": 0.5375203365801448,
      "learning_rate": 0.0001306029824934835,
      "loss": 12.2736,
      "step": 15388
    },
    {
      "epoch": 0.8379925534159672,
      "grad_norm": 0.6180323788492011,
      "learning_rate": 0.00013059458719939215,
      "loss": 12.2002,
      "step": 15389
    },
    {
      "epoch": 0.8380470074125503,
      "grad_norm": 0.5576984568599684,
      "learning_rate": 0.0001305861916673921,
      "loss": 12.0192,
      "step": 15390
    },
    {
      "epoch": 0.8381014614091333,
      "grad_norm": 0.5571576789776306,
      "learning_rate": 0.00013057779589754876,
      "loss": 12.1686,
      "step": 15391
    },
    {
      "epoch": 0.8381559154057163,
      "grad_norm": 0.5723703608319274,
      "learning_rate": 0.0001305693998899273,
      "loss": 12.1266,
      "step": 15392
    },
    {
      "epoch": 0.8382103694022993,
      "grad_norm": 0.5639093049927778,
      "learning_rate": 0.00013056100364459305,
      "loss": 12.1069,
      "step": 15393
    },
    {
      "epoch": 0.8382648233988823,
      "grad_norm": 0.5571453029303685,
      "learning_rate": 0.00013055260716161136,
      "loss": 12.1868,
      "step": 15394
    },
    {
      "epoch": 0.8383192773954653,
      "grad_norm": 0.5848548378507161,
      "learning_rate": 0.00013054421044104744,
      "loss": 12.2119,
      "step": 15395
    },
    {
      "epoch": 0.8383737313920484,
      "grad_norm": 0.5833350880830424,
      "learning_rate": 0.00013053581348296663,
      "loss": 12.2092,
      "step": 15396
    },
    {
      "epoch": 0.8384281853886314,
      "grad_norm": 0.564726235079284,
      "learning_rate": 0.0001305274162874342,
      "loss": 12.2672,
      "step": 15397
    },
    {
      "epoch": 0.8384826393852144,
      "grad_norm": 0.5640123189749882,
      "learning_rate": 0.00013051901885451544,
      "loss": 12.1969,
      "step": 15398
    },
    {
      "epoch": 0.8385370933817974,
      "grad_norm": 0.5567158354042706,
      "learning_rate": 0.00013051062118427575,
      "loss": 12.1581,
      "step": 15399
    },
    {
      "epoch": 0.8385915473783804,
      "grad_norm": 0.6333351307124527,
      "learning_rate": 0.0001305022232767803,
      "loss": 12.2515,
      "step": 15400
    },
    {
      "epoch": 0.8386460013749634,
      "grad_norm": 0.6196519026665442,
      "learning_rate": 0.00013049382513209446,
      "loss": 12.157,
      "step": 15401
    },
    {
      "epoch": 0.8387004553715465,
      "grad_norm": 0.6613005026286023,
      "learning_rate": 0.0001304854267502835,
      "loss": 12.2158,
      "step": 15402
    },
    {
      "epoch": 0.8387549093681295,
      "grad_norm": 0.5624156990839256,
      "learning_rate": 0.00013047702813141274,
      "loss": 12.1756,
      "step": 15403
    },
    {
      "epoch": 0.8388093633647125,
      "grad_norm": 0.6391498061034914,
      "learning_rate": 0.00013046862927554756,
      "loss": 12.1817,
      "step": 15404
    },
    {
      "epoch": 0.8388638173612954,
      "grad_norm": 0.7324785085813693,
      "learning_rate": 0.00013046023018275314,
      "loss": 12.4085,
      "step": 15405
    },
    {
      "epoch": 0.8389182713578784,
      "grad_norm": 0.5810583470765037,
      "learning_rate": 0.00013045183085309492,
      "loss": 12.3102,
      "step": 15406
    },
    {
      "epoch": 0.8389727253544614,
      "grad_norm": 0.5512366258667796,
      "learning_rate": 0.00013044343128663813,
      "loss": 12.1526,
      "step": 15407
    },
    {
      "epoch": 0.8390271793510445,
      "grad_norm": 0.5588903997378186,
      "learning_rate": 0.0001304350314834481,
      "loss": 12.1694,
      "step": 15408
    },
    {
      "epoch": 0.8390816333476275,
      "grad_norm": 0.5795679654907878,
      "learning_rate": 0.00013042663144359015,
      "loss": 12.3495,
      "step": 15409
    },
    {
      "epoch": 0.8391360873442105,
      "grad_norm": 0.5369781674763817,
      "learning_rate": 0.00013041823116712964,
      "loss": 12.2644,
      "step": 15410
    },
    {
      "epoch": 0.8391905413407935,
      "grad_norm": 0.5698494013959302,
      "learning_rate": 0.00013040983065413185,
      "loss": 12.1736,
      "step": 15411
    },
    {
      "epoch": 0.8392449953373765,
      "grad_norm": 0.5609634761067283,
      "learning_rate": 0.00013040142990466212,
      "loss": 12.3333,
      "step": 15412
    },
    {
      "epoch": 0.8392994493339595,
      "grad_norm": 0.5922172919484425,
      "learning_rate": 0.0001303930289187858,
      "loss": 12.1139,
      "step": 15413
    },
    {
      "epoch": 0.8393539033305426,
      "grad_norm": 0.5710970376219491,
      "learning_rate": 0.00013038462769656816,
      "loss": 12.2493,
      "step": 15414
    },
    {
      "epoch": 0.8394083573271256,
      "grad_norm": 0.6130269140220768,
      "learning_rate": 0.00013037622623807458,
      "loss": 12.3838,
      "step": 15415
    },
    {
      "epoch": 0.8394628113237086,
      "grad_norm": 0.5685305558318331,
      "learning_rate": 0.00013036782454337034,
      "loss": 12.264,
      "step": 15416
    },
    {
      "epoch": 0.8395172653202916,
      "grad_norm": 0.6027775225125119,
      "learning_rate": 0.00013035942261252083,
      "loss": 12.0201,
      "step": 15417
    },
    {
      "epoch": 0.8395717193168746,
      "grad_norm": 0.5522397647824678,
      "learning_rate": 0.00013035102044559133,
      "loss": 12.0813,
      "step": 15418
    },
    {
      "epoch": 0.8396261733134577,
      "grad_norm": 0.6259116033321801,
      "learning_rate": 0.00013034261804264726,
      "loss": 12.2236,
      "step": 15419
    },
    {
      "epoch": 0.8396806273100407,
      "grad_norm": 0.5905157049713758,
      "learning_rate": 0.00013033421540375385,
      "loss": 12.3173,
      "step": 15420
    },
    {
      "epoch": 0.8397350813066237,
      "grad_norm": 0.5151391159511745,
      "learning_rate": 0.0001303258125289765,
      "loss": 12.1515,
      "step": 15421
    },
    {
      "epoch": 0.8397895353032067,
      "grad_norm": 0.531701959746869,
      "learning_rate": 0.00013031740941838057,
      "loss": 12.1666,
      "step": 15422
    },
    {
      "epoch": 0.8398439892997897,
      "grad_norm": 0.522400648155579,
      "learning_rate": 0.00013030900607203136,
      "loss": 12.1277,
      "step": 15423
    },
    {
      "epoch": 0.8398984432963726,
      "grad_norm": 0.6362661145573922,
      "learning_rate": 0.00013030060248999425,
      "loss": 12.2302,
      "step": 15424
    },
    {
      "epoch": 0.8399528972929557,
      "grad_norm": 0.5138634705111699,
      "learning_rate": 0.00013029219867233458,
      "loss": 12.1477,
      "step": 15425
    },
    {
      "epoch": 0.8400073512895387,
      "grad_norm": 0.5504408190011557,
      "learning_rate": 0.00013028379461911766,
      "loss": 12.2609,
      "step": 15426
    },
    {
      "epoch": 0.8400618052861217,
      "grad_norm": 0.5817229386487759,
      "learning_rate": 0.0001302753903304089,
      "loss": 12.1662,
      "step": 15427
    },
    {
      "epoch": 0.8401162592827047,
      "grad_norm": 0.5779589417910154,
      "learning_rate": 0.00013026698580627364,
      "loss": 12.1039,
      "step": 15428
    },
    {
      "epoch": 0.8401707132792877,
      "grad_norm": 0.6501390033485781,
      "learning_rate": 0.00013025858104677722,
      "loss": 12.3038,
      "step": 15429
    },
    {
      "epoch": 0.8402251672758707,
      "grad_norm": 0.547251099236945,
      "learning_rate": 0.00013025017605198494,
      "loss": 12.1442,
      "step": 15430
    },
    {
      "epoch": 0.8402796212724538,
      "grad_norm": 0.5299319663603849,
      "learning_rate": 0.00013024177082196226,
      "loss": 12.1717,
      "step": 15431
    },
    {
      "epoch": 0.8403340752690368,
      "grad_norm": 0.5703350863581309,
      "learning_rate": 0.00013023336535677454,
      "loss": 12.2853,
      "step": 15432
    },
    {
      "epoch": 0.8403885292656198,
      "grad_norm": 0.6139029056887727,
      "learning_rate": 0.00013022495965648705,
      "loss": 12.2486,
      "step": 15433
    },
    {
      "epoch": 0.8404429832622028,
      "grad_norm": 0.5488466788920311,
      "learning_rate": 0.00013021655372116525,
      "loss": 12.1232,
      "step": 15434
    },
    {
      "epoch": 0.8404974372587858,
      "grad_norm": 0.5072533971790376,
      "learning_rate": 0.0001302081475508744,
      "loss": 12.1123,
      "step": 15435
    },
    {
      "epoch": 0.8405518912553688,
      "grad_norm": 0.6553603035867362,
      "learning_rate": 0.00013019974114567993,
      "loss": 12.4077,
      "step": 15436
    },
    {
      "epoch": 0.8406063452519519,
      "grad_norm": 0.5963926708645846,
      "learning_rate": 0.00013019133450564724,
      "loss": 12.0707,
      "step": 15437
    },
    {
      "epoch": 0.8406607992485349,
      "grad_norm": 0.5766582613875916,
      "learning_rate": 0.00013018292763084167,
      "loss": 12.1169,
      "step": 15438
    },
    {
      "epoch": 0.8407152532451179,
      "grad_norm": 0.598765168143679,
      "learning_rate": 0.0001301745205213286,
      "loss": 12.2108,
      "step": 15439
    },
    {
      "epoch": 0.8407697072417009,
      "grad_norm": 0.6183119765379161,
      "learning_rate": 0.0001301661131771734,
      "loss": 12.3027,
      "step": 15440
    },
    {
      "epoch": 0.8408241612382839,
      "grad_norm": 0.5438659222201025,
      "learning_rate": 0.0001301577055984414,
      "loss": 12.2832,
      "step": 15441
    },
    {
      "epoch": 0.8408786152348668,
      "grad_norm": 0.5569020818739407,
      "learning_rate": 0.00013014929778519806,
      "loss": 12.1564,
      "step": 15442
    },
    {
      "epoch": 0.84093306923145,
      "grad_norm": 0.5396355165755193,
      "learning_rate": 0.00013014088973750874,
      "loss": 12.2076,
      "step": 15443
    },
    {
      "epoch": 0.8409875232280329,
      "grad_norm": 0.5594209164770022,
      "learning_rate": 0.00013013248145543878,
      "loss": 12.1998,
      "step": 15444
    },
    {
      "epoch": 0.8410419772246159,
      "grad_norm": 0.5582406828411369,
      "learning_rate": 0.00013012407293905363,
      "loss": 12.0757,
      "step": 15445
    },
    {
      "epoch": 0.8410964312211989,
      "grad_norm": 0.557935986177484,
      "learning_rate": 0.00013011566418841858,
      "loss": 12.1023,
      "step": 15446
    },
    {
      "epoch": 0.8411508852177819,
      "grad_norm": 0.5590660230542867,
      "learning_rate": 0.00013010725520359908,
      "loss": 12.2449,
      "step": 15447
    },
    {
      "epoch": 0.8412053392143649,
      "grad_norm": 0.5686932609582025,
      "learning_rate": 0.00013009884598466054,
      "loss": 12.1602,
      "step": 15448
    },
    {
      "epoch": 0.841259793210948,
      "grad_norm": 0.5438917008388785,
      "learning_rate": 0.00013009043653166834,
      "loss": 12.1989,
      "step": 15449
    },
    {
      "epoch": 0.841314247207531,
      "grad_norm": 0.5770990080511804,
      "learning_rate": 0.00013008202684468786,
      "loss": 12.0269,
      "step": 15450
    },
    {
      "epoch": 0.841368701204114,
      "grad_norm": 0.6950577279601987,
      "learning_rate": 0.00013007361692378446,
      "loss": 12.2033,
      "step": 15451
    },
    {
      "epoch": 0.841423155200697,
      "grad_norm": 0.5215834135230242,
      "learning_rate": 0.00013006520676902357,
      "loss": 12.153,
      "step": 15452
    },
    {
      "epoch": 0.84147760919728,
      "grad_norm": 0.5919430674869965,
      "learning_rate": 0.00013005679638047058,
      "loss": 12.2235,
      "step": 15453
    },
    {
      "epoch": 0.8415320631938631,
      "grad_norm": 0.5878741611778164,
      "learning_rate": 0.00013004838575819097,
      "loss": 12.1629,
      "step": 15454
    },
    {
      "epoch": 0.8415865171904461,
      "grad_norm": 0.5506223431101179,
      "learning_rate": 0.00013003997490225003,
      "loss": 12.1377,
      "step": 15455
    },
    {
      "epoch": 0.8416409711870291,
      "grad_norm": 0.5477777108898231,
      "learning_rate": 0.0001300315638127132,
      "loss": 12.1777,
      "step": 15456
    },
    {
      "epoch": 0.8416954251836121,
      "grad_norm": 0.6021798307343533,
      "learning_rate": 0.00013002315248964588,
      "loss": 12.058,
      "step": 15457
    },
    {
      "epoch": 0.8417498791801951,
      "grad_norm": 0.6091936693650849,
      "learning_rate": 0.00013001474093311352,
      "loss": 12.1401,
      "step": 15458
    },
    {
      "epoch": 0.841804333176778,
      "grad_norm": 0.590264553535219,
      "learning_rate": 0.0001300063291431815,
      "loss": 12.1855,
      "step": 15459
    },
    {
      "epoch": 0.8418587871733612,
      "grad_norm": 0.5543263624857145,
      "learning_rate": 0.00012999791711991522,
      "loss": 12.1857,
      "step": 15460
    },
    {
      "epoch": 0.8419132411699441,
      "grad_norm": 0.6258587587841732,
      "learning_rate": 0.00012998950486338014,
      "loss": 12.1809,
      "step": 15461
    },
    {
      "epoch": 0.8419676951665271,
      "grad_norm": 0.576115813718609,
      "learning_rate": 0.0001299810923736416,
      "loss": 12.1628,
      "step": 15462
    },
    {
      "epoch": 0.8420221491631101,
      "grad_norm": 0.5779529876001324,
      "learning_rate": 0.00012997267965076504,
      "loss": 12.1651,
      "step": 15463
    },
    {
      "epoch": 0.8420766031596931,
      "grad_norm": 0.6170858443746692,
      "learning_rate": 0.00012996426669481593,
      "loss": 12.3123,
      "step": 15464
    },
    {
      "epoch": 0.8421310571562761,
      "grad_norm": 0.5993499195332175,
      "learning_rate": 0.00012995585350585967,
      "loss": 12.2695,
      "step": 15465
    },
    {
      "epoch": 0.8421855111528592,
      "grad_norm": 0.5826163014390762,
      "learning_rate": 0.00012994744008396167,
      "loss": 12.2115,
      "step": 15466
    },
    {
      "epoch": 0.8422399651494422,
      "grad_norm": 0.5385510345677272,
      "learning_rate": 0.00012993902642918732,
      "loss": 12.2976,
      "step": 15467
    },
    {
      "epoch": 0.8422944191460252,
      "grad_norm": 0.5659184008479512,
      "learning_rate": 0.0001299306125416021,
      "loss": 12.1892,
      "step": 15468
    },
    {
      "epoch": 0.8423488731426082,
      "grad_norm": 0.5205833855431005,
      "learning_rate": 0.00012992219842127142,
      "loss": 12.1612,
      "step": 15469
    },
    {
      "epoch": 0.8424033271391912,
      "grad_norm": 0.5737317949815772,
      "learning_rate": 0.0001299137840682607,
      "loss": 12.2158,
      "step": 15470
    },
    {
      "epoch": 0.8424577811357742,
      "grad_norm": 0.5923701455972223,
      "learning_rate": 0.00012990536948263536,
      "loss": 12.1834,
      "step": 15471
    },
    {
      "epoch": 0.8425122351323573,
      "grad_norm": 0.5610356626591934,
      "learning_rate": 0.00012989695466446088,
      "loss": 12.1868,
      "step": 15472
    },
    {
      "epoch": 0.8425666891289403,
      "grad_norm": 0.5593317641200841,
      "learning_rate": 0.00012988853961380268,
      "loss": 12.1865,
      "step": 15473
    },
    {
      "epoch": 0.8426211431255233,
      "grad_norm": 0.5609214500978,
      "learning_rate": 0.00012988012433072616,
      "loss": 12.2068,
      "step": 15474
    },
    {
      "epoch": 0.8426755971221063,
      "grad_norm": 0.6189225809724805,
      "learning_rate": 0.00012987170881529678,
      "loss": 12.3564,
      "step": 15475
    },
    {
      "epoch": 0.8427300511186893,
      "grad_norm": 0.5810532995223517,
      "learning_rate": 0.00012986329306757997,
      "loss": 12.225,
      "step": 15476
    },
    {
      "epoch": 0.8427845051152723,
      "grad_norm": 0.6477006315994673,
      "learning_rate": 0.00012985487708764122,
      "loss": 12.3078,
      "step": 15477
    },
    {
      "epoch": 0.8428389591118554,
      "grad_norm": 0.5325454748651055,
      "learning_rate": 0.0001298464608755459,
      "loss": 12.2116,
      "step": 15478
    },
    {
      "epoch": 0.8428934131084383,
      "grad_norm": 0.5630427183844071,
      "learning_rate": 0.0001298380444313595,
      "loss": 12.2595,
      "step": 15479
    },
    {
      "epoch": 0.8429478671050213,
      "grad_norm": 0.5153174542840714,
      "learning_rate": 0.0001298296277551475,
      "loss": 12.144,
      "step": 15480
    },
    {
      "epoch": 0.8430023211016043,
      "grad_norm": 0.5812089814295625,
      "learning_rate": 0.00012982121084697529,
      "loss": 12.289,
      "step": 15481
    },
    {
      "epoch": 0.8430567750981873,
      "grad_norm": 0.5651481908263806,
      "learning_rate": 0.00012981279370690834,
      "loss": 12.1724,
      "step": 15482
    },
    {
      "epoch": 0.8431112290947703,
      "grad_norm": 0.5993511716762084,
      "learning_rate": 0.00012980437633501214,
      "loss": 12.1811,
      "step": 15483
    },
    {
      "epoch": 0.8431656830913534,
      "grad_norm": 0.5576741439576591,
      "learning_rate": 0.00012979595873135205,
      "loss": 12.1552,
      "step": 15484
    },
    {
      "epoch": 0.8432201370879364,
      "grad_norm": 0.5825885230304538,
      "learning_rate": 0.00012978754089599363,
      "loss": 12.2838,
      "step": 15485
    },
    {
      "epoch": 0.8432745910845194,
      "grad_norm": 0.5981931555447255,
      "learning_rate": 0.00012977912282900232,
      "loss": 12.1833,
      "step": 15486
    },
    {
      "epoch": 0.8433290450811024,
      "grad_norm": 0.5922220565819358,
      "learning_rate": 0.00012977070453044348,
      "loss": 12.1535,
      "step": 15487
    },
    {
      "epoch": 0.8433834990776854,
      "grad_norm": 0.5485480344051351,
      "learning_rate": 0.00012976228600038273,
      "loss": 12.2509,
      "step": 15488
    },
    {
      "epoch": 0.8434379530742685,
      "grad_norm": 0.5948772804151026,
      "learning_rate": 0.00012975386723888542,
      "loss": 12.1709,
      "step": 15489
    },
    {
      "epoch": 0.8434924070708515,
      "grad_norm": 0.6364488969136304,
      "learning_rate": 0.00012974544824601703,
      "loss": 12.0577,
      "step": 15490
    },
    {
      "epoch": 0.8435468610674345,
      "grad_norm": 0.5738671439296887,
      "learning_rate": 0.00012973702902184306,
      "loss": 12.2015,
      "step": 15491
    },
    {
      "epoch": 0.8436013150640175,
      "grad_norm": 0.5646197147628447,
      "learning_rate": 0.00012972860956642895,
      "loss": 12.1925,
      "step": 15492
    },
    {
      "epoch": 0.8436557690606005,
      "grad_norm": 0.527313632438477,
      "learning_rate": 0.00012972018987984023,
      "loss": 12.1908,
      "step": 15493
    },
    {
      "epoch": 0.8437102230571835,
      "grad_norm": 0.5621058442006267,
      "learning_rate": 0.00012971176996214232,
      "loss": 12.189,
      "step": 15494
    },
    {
      "epoch": 0.8437646770537666,
      "grad_norm": 0.5441499521332711,
      "learning_rate": 0.00012970334981340063,
      "loss": 12.1707,
      "step": 15495
    },
    {
      "epoch": 0.8438191310503496,
      "grad_norm": 0.6113276956688619,
      "learning_rate": 0.0001296949294336808,
      "loss": 12.1549,
      "step": 15496
    },
    {
      "epoch": 0.8438735850469326,
      "grad_norm": 0.6162383652250739,
      "learning_rate": 0.00012968650882304818,
      "loss": 12.2017,
      "step": 15497
    },
    {
      "epoch": 0.8439280390435155,
      "grad_norm": 0.5558501593387551,
      "learning_rate": 0.00012967808798156828,
      "loss": 12.1558,
      "step": 15498
    },
    {
      "epoch": 0.8439824930400985,
      "grad_norm": 0.527880376040958,
      "learning_rate": 0.00012966966690930665,
      "loss": 12.2283,
      "step": 15499
    },
    {
      "epoch": 0.8440369470366815,
      "grad_norm": 0.5775924687961809,
      "learning_rate": 0.00012966124560632867,
      "loss": 12.3681,
      "step": 15500
    },
    {
      "epoch": 0.8440914010332646,
      "grad_norm": 0.5918371650924008,
      "learning_rate": 0.00012965282407269982,
      "loss": 12.0314,
      "step": 15501
    },
    {
      "epoch": 0.8441458550298476,
      "grad_norm": 0.5889430522237789,
      "learning_rate": 0.0001296444023084857,
      "loss": 12.1733,
      "step": 15502
    },
    {
      "epoch": 0.8442003090264306,
      "grad_norm": 0.630949062021415,
      "learning_rate": 0.0001296359803137517,
      "loss": 12.1696,
      "step": 15503
    },
    {
      "epoch": 0.8442547630230136,
      "grad_norm": 0.6027404325392326,
      "learning_rate": 0.00012962755808856342,
      "loss": 12.2744,
      "step": 15504
    },
    {
      "epoch": 0.8443092170195966,
      "grad_norm": 0.6058898368980901,
      "learning_rate": 0.00012961913563298624,
      "loss": 12.3292,
      "step": 15505
    },
    {
      "epoch": 0.8443636710161796,
      "grad_norm": 0.5531306126172333,
      "learning_rate": 0.0001296107129470857,
      "loss": 12.2541,
      "step": 15506
    },
    {
      "epoch": 0.8444181250127627,
      "grad_norm": 0.6083152632010047,
      "learning_rate": 0.00012960229003092724,
      "loss": 12.163,
      "step": 15507
    },
    {
      "epoch": 0.8444725790093457,
      "grad_norm": 0.5771152076523544,
      "learning_rate": 0.00012959386688457642,
      "loss": 12.3182,
      "step": 15508
    },
    {
      "epoch": 0.8445270330059287,
      "grad_norm": 0.5904188640337819,
      "learning_rate": 0.00012958544350809878,
      "loss": 12.0135,
      "step": 15509
    },
    {
      "epoch": 0.8445814870025117,
      "grad_norm": 0.511318850188779,
      "learning_rate": 0.00012957701990155975,
      "loss": 12.1324,
      "step": 15510
    },
    {
      "epoch": 0.8446359409990947,
      "grad_norm": 0.549514494740283,
      "learning_rate": 0.00012956859606502486,
      "loss": 12.1693,
      "step": 15511
    },
    {
      "epoch": 0.8446903949956777,
      "grad_norm": 0.6198624919438058,
      "learning_rate": 0.00012956017199855957,
      "loss": 12.3404,
      "step": 15512
    },
    {
      "epoch": 0.8447448489922608,
      "grad_norm": 0.5654961144625223,
      "learning_rate": 0.00012955174770222944,
      "loss": 12.3016,
      "step": 15513
    },
    {
      "epoch": 0.8447993029888438,
      "grad_norm": 0.5328895178588599,
      "learning_rate": 0.00012954332317609995,
      "loss": 12.1564,
      "step": 15514
    },
    {
      "epoch": 0.8448537569854268,
      "grad_norm": 0.593522704545548,
      "learning_rate": 0.0001295348984202367,
      "loss": 12.2934,
      "step": 15515
    },
    {
      "epoch": 0.8449082109820097,
      "grad_norm": 0.5430808014371415,
      "learning_rate": 0.00012952647343470505,
      "loss": 12.191,
      "step": 15516
    },
    {
      "epoch": 0.8449626649785927,
      "grad_norm": 0.5265172841458723,
      "learning_rate": 0.00012951804821957063,
      "loss": 12.2325,
      "step": 15517
    },
    {
      "epoch": 0.8450171189751757,
      "grad_norm": 0.6839327055872534,
      "learning_rate": 0.00012950962277489885,
      "loss": 12.3049,
      "step": 15518
    },
    {
      "epoch": 0.8450715729717588,
      "grad_norm": 0.5473030469567395,
      "learning_rate": 0.00012950119710075536,
      "loss": 12.2636,
      "step": 15519
    },
    {
      "epoch": 0.8451260269683418,
      "grad_norm": 0.5500510581940654,
      "learning_rate": 0.00012949277119720564,
      "loss": 12.2041,
      "step": 15520
    },
    {
      "epoch": 0.8451804809649248,
      "grad_norm": 0.586482402015144,
      "learning_rate": 0.00012948434506431514,
      "loss": 12.2955,
      "step": 15521
    },
    {
      "epoch": 0.8452349349615078,
      "grad_norm": 0.5724440277061343,
      "learning_rate": 0.00012947591870214945,
      "loss": 12.2484,
      "step": 15522
    },
    {
      "epoch": 0.8452893889580908,
      "grad_norm": 0.5696594400728073,
      "learning_rate": 0.00012946749211077406,
      "loss": 12.193,
      "step": 15523
    },
    {
      "epoch": 0.8453438429546739,
      "grad_norm": 0.5209405674486619,
      "learning_rate": 0.00012945906529025447,
      "loss": 12.0911,
      "step": 15524
    },
    {
      "epoch": 0.8453982969512569,
      "grad_norm": 0.570044338500723,
      "learning_rate": 0.00012945063824065632,
      "loss": 12.2399,
      "step": 15525
    },
    {
      "epoch": 0.8454527509478399,
      "grad_norm": 0.5518226214837841,
      "learning_rate": 0.00012944221096204502,
      "loss": 12.2789,
      "step": 15526
    },
    {
      "epoch": 0.8455072049444229,
      "grad_norm": 0.5511632782042666,
      "learning_rate": 0.00012943378345448616,
      "loss": 12.0836,
      "step": 15527
    },
    {
      "epoch": 0.8455616589410059,
      "grad_norm": 0.6136098803437449,
      "learning_rate": 0.00012942535571804526,
      "loss": 12.1695,
      "step": 15528
    },
    {
      "epoch": 0.8456161129375889,
      "grad_norm": 0.6071253304180788,
      "learning_rate": 0.00012941692775278785,
      "loss": 12.2983,
      "step": 15529
    },
    {
      "epoch": 0.845670566934172,
      "grad_norm": 0.5204147345078632,
      "learning_rate": 0.0001294084995587795,
      "loss": 12.1814,
      "step": 15530
    },
    {
      "epoch": 0.845725020930755,
      "grad_norm": 0.5472504695583279,
      "learning_rate": 0.0001294000711360857,
      "loss": 12.2124,
      "step": 15531
    },
    {
      "epoch": 0.845779474927338,
      "grad_norm": 0.6406576106211804,
      "learning_rate": 0.00012939164248477206,
      "loss": 12.3914,
      "step": 15532
    },
    {
      "epoch": 0.845833928923921,
      "grad_norm": 0.5236012866958867,
      "learning_rate": 0.00012938321360490406,
      "loss": 12.0712,
      "step": 15533
    },
    {
      "epoch": 0.845888382920504,
      "grad_norm": 0.5797560648484227,
      "learning_rate": 0.00012937478449654726,
      "loss": 12.1934,
      "step": 15534
    },
    {
      "epoch": 0.8459428369170869,
      "grad_norm": 0.5685664703573508,
      "learning_rate": 0.00012936635515976722,
      "loss": 12.1307,
      "step": 15535
    },
    {
      "epoch": 0.84599729091367,
      "grad_norm": 0.6221459849482847,
      "learning_rate": 0.00012935792559462945,
      "loss": 12.1631,
      "step": 15536
    },
    {
      "epoch": 0.846051744910253,
      "grad_norm": 0.5876772033125665,
      "learning_rate": 0.00012934949580119953,
      "loss": 12.1943,
      "step": 15537
    },
    {
      "epoch": 0.846106198906836,
      "grad_norm": 0.759882435895185,
      "learning_rate": 0.00012934106577954305,
      "loss": 12.3563,
      "step": 15538
    },
    {
      "epoch": 0.846160652903419,
      "grad_norm": 0.5415582133886058,
      "learning_rate": 0.00012933263552972549,
      "loss": 12.1682,
      "step": 15539
    },
    {
      "epoch": 0.846215106900002,
      "grad_norm": 0.5903793304696313,
      "learning_rate": 0.00012932420505181241,
      "loss": 12.2252,
      "step": 15540
    },
    {
      "epoch": 0.846269560896585,
      "grad_norm": 0.5849930079909551,
      "learning_rate": 0.00012931577434586943,
      "loss": 12.3335,
      "step": 15541
    },
    {
      "epoch": 0.8463240148931681,
      "grad_norm": 0.6003166564253619,
      "learning_rate": 0.00012930734341196206,
      "loss": 12.1644,
      "step": 15542
    },
    {
      "epoch": 0.8463784688897511,
      "grad_norm": 0.6097396533940193,
      "learning_rate": 0.00012929891225015586,
      "loss": 12.1985,
      "step": 15543
    },
    {
      "epoch": 0.8464329228863341,
      "grad_norm": 0.5938451907882728,
      "learning_rate": 0.00012929048086051645,
      "loss": 12.2239,
      "step": 15544
    },
    {
      "epoch": 0.8464873768829171,
      "grad_norm": 0.7848091599738762,
      "learning_rate": 0.0001292820492431093,
      "loss": 12.3416,
      "step": 15545
    },
    {
      "epoch": 0.8465418308795001,
      "grad_norm": 0.5677497381016352,
      "learning_rate": 0.00012927361739800005,
      "loss": 12.1728,
      "step": 15546
    },
    {
      "epoch": 0.8465962848760831,
      "grad_norm": 0.6104022337169018,
      "learning_rate": 0.00012926518532525424,
      "loss": 12.1725,
      "step": 15547
    },
    {
      "epoch": 0.8466507388726662,
      "grad_norm": 0.5022216219713065,
      "learning_rate": 0.00012925675302493745,
      "loss": 12.1429,
      "step": 15548
    },
    {
      "epoch": 0.8467051928692492,
      "grad_norm": 0.5848567111984192,
      "learning_rate": 0.00012924832049711525,
      "loss": 12.2029,
      "step": 15549
    },
    {
      "epoch": 0.8467596468658322,
      "grad_norm": 0.5081974836950981,
      "learning_rate": 0.00012923988774185316,
      "loss": 11.9883,
      "step": 15550
    },
    {
      "epoch": 0.8468141008624152,
      "grad_norm": 0.6221118858893209,
      "learning_rate": 0.00012923145475921683,
      "loss": 12.2106,
      "step": 15551
    },
    {
      "epoch": 0.8468685548589981,
      "grad_norm": 0.6175250867818555,
      "learning_rate": 0.00012922302154927179,
      "loss": 12.3017,
      "step": 15552
    },
    {
      "epoch": 0.8469230088555812,
      "grad_norm": 0.5352118342810327,
      "learning_rate": 0.00012921458811208366,
      "loss": 12.1494,
      "step": 15553
    },
    {
      "epoch": 0.8469774628521642,
      "grad_norm": 0.5872967934663972,
      "learning_rate": 0.00012920615444771797,
      "loss": 12.0825,
      "step": 15554
    },
    {
      "epoch": 0.8470319168487472,
      "grad_norm": 0.5933345426004941,
      "learning_rate": 0.00012919772055624038,
      "loss": 12.2394,
      "step": 15555
    },
    {
      "epoch": 0.8470863708453302,
      "grad_norm": 0.6272768455911071,
      "learning_rate": 0.00012918928643771633,
      "loss": 12.2328,
      "step": 15556
    },
    {
      "epoch": 0.8471408248419132,
      "grad_norm": 0.5198233088362648,
      "learning_rate": 0.00012918085209221155,
      "loss": 12.0769,
      "step": 15557
    },
    {
      "epoch": 0.8471952788384962,
      "grad_norm": 0.6244403370285655,
      "learning_rate": 0.00012917241751979154,
      "loss": 12.1873,
      "step": 15558
    },
    {
      "epoch": 0.8472497328350793,
      "grad_norm": 0.522512235489971,
      "learning_rate": 0.00012916398272052194,
      "loss": 12.1636,
      "step": 15559
    },
    {
      "epoch": 0.8473041868316623,
      "grad_norm": 0.5587611000537016,
      "learning_rate": 0.00012915554769446833,
      "loss": 12.3209,
      "step": 15560
    },
    {
      "epoch": 0.8473586408282453,
      "grad_norm": 0.5404422738375619,
      "learning_rate": 0.00012914711244169626,
      "loss": 12.2373,
      "step": 15561
    },
    {
      "epoch": 0.8474130948248283,
      "grad_norm": 0.5233974892533286,
      "learning_rate": 0.00012913867696227136,
      "loss": 12.1196,
      "step": 15562
    },
    {
      "epoch": 0.8474675488214113,
      "grad_norm": 0.5540664811886048,
      "learning_rate": 0.00012913024125625925,
      "loss": 12.201,
      "step": 15563
    },
    {
      "epoch": 0.8475220028179943,
      "grad_norm": 0.585623802648054,
      "learning_rate": 0.00012912180532372548,
      "loss": 12.0897,
      "step": 15564
    },
    {
      "epoch": 0.8475764568145774,
      "grad_norm": 0.5162575407180071,
      "learning_rate": 0.0001291133691647357,
      "loss": 12.097,
      "step": 15565
    },
    {
      "epoch": 0.8476309108111604,
      "grad_norm": 0.6065161656591413,
      "learning_rate": 0.00012910493277935544,
      "loss": 12.1424,
      "step": 15566
    },
    {
      "epoch": 0.8476853648077434,
      "grad_norm": 0.6120489582519593,
      "learning_rate": 0.00012909649616765033,
      "loss": 12.1668,
      "step": 15567
    },
    {
      "epoch": 0.8477398188043264,
      "grad_norm": 0.610427532528256,
      "learning_rate": 0.00012908805932968602,
      "loss": 12.2514,
      "step": 15568
    },
    {
      "epoch": 0.8477942728009094,
      "grad_norm": 0.5880797631439723,
      "learning_rate": 0.00012907962226552807,
      "loss": 12.1468,
      "step": 15569
    },
    {
      "epoch": 0.8478487267974923,
      "grad_norm": 0.7011077942022087,
      "learning_rate": 0.00012907118497524213,
      "loss": 12.3227,
      "step": 15570
    },
    {
      "epoch": 0.8479031807940755,
      "grad_norm": 0.5620293371921344,
      "learning_rate": 0.00012906274745889374,
      "loss": 12.1715,
      "step": 15571
    },
    {
      "epoch": 0.8479576347906584,
      "grad_norm": 0.6079063552721871,
      "learning_rate": 0.00012905430971654858,
      "loss": 12.1763,
      "step": 15572
    },
    {
      "epoch": 0.8480120887872414,
      "grad_norm": 0.6148687010089448,
      "learning_rate": 0.0001290458717482722,
      "loss": 12.2441,
      "step": 15573
    },
    {
      "epoch": 0.8480665427838244,
      "grad_norm": 0.5995826169807735,
      "learning_rate": 0.00012903743355413024,
      "loss": 12.19,
      "step": 15574
    },
    {
      "epoch": 0.8481209967804074,
      "grad_norm": 0.5907546417641577,
      "learning_rate": 0.0001290289951341884,
      "loss": 12.1121,
      "step": 15575
    },
    {
      "epoch": 0.8481754507769904,
      "grad_norm": 0.5300240353852979,
      "learning_rate": 0.00012902055648851218,
      "loss": 12.134,
      "step": 15576
    },
    {
      "epoch": 0.8482299047735735,
      "grad_norm": 0.6731591510323991,
      "learning_rate": 0.00012901211761716724,
      "loss": 12.3143,
      "step": 15577
    },
    {
      "epoch": 0.8482843587701565,
      "grad_norm": 0.6375915340003899,
      "learning_rate": 0.0001290036785202192,
      "loss": 12.2515,
      "step": 15578
    },
    {
      "epoch": 0.8483388127667395,
      "grad_norm": 0.5813097313194986,
      "learning_rate": 0.00012899523919773372,
      "loss": 12.2116,
      "step": 15579
    },
    {
      "epoch": 0.8483932667633225,
      "grad_norm": 0.5742185102082507,
      "learning_rate": 0.0001289867996497764,
      "loss": 12.1474,
      "step": 15580
    },
    {
      "epoch": 0.8484477207599055,
      "grad_norm": 0.590523813660125,
      "learning_rate": 0.00012897835987641285,
      "loss": 12.106,
      "step": 15581
    },
    {
      "epoch": 0.8485021747564885,
      "grad_norm": 0.6000171184383458,
      "learning_rate": 0.0001289699198777087,
      "loss": 12.154,
      "step": 15582
    },
    {
      "epoch": 0.8485566287530716,
      "grad_norm": 0.5719366141175086,
      "learning_rate": 0.00012896147965372963,
      "loss": 12.2698,
      "step": 15583
    },
    {
      "epoch": 0.8486110827496546,
      "grad_norm": 0.5386123198262065,
      "learning_rate": 0.00012895303920454118,
      "loss": 12.2407,
      "step": 15584
    },
    {
      "epoch": 0.8486655367462376,
      "grad_norm": 0.5296364039346468,
      "learning_rate": 0.0001289445985302091,
      "loss": 12.155,
      "step": 15585
    },
    {
      "epoch": 0.8487199907428206,
      "grad_norm": 0.6482716971909245,
      "learning_rate": 0.00012893615763079894,
      "loss": 12.2155,
      "step": 15586
    },
    {
      "epoch": 0.8487744447394036,
      "grad_norm": 0.7109488935622398,
      "learning_rate": 0.00012892771650637637,
      "loss": 12.1994,
      "step": 15587
    },
    {
      "epoch": 0.8488288987359867,
      "grad_norm": 0.593128389693315,
      "learning_rate": 0.00012891927515700703,
      "loss": 12.1713,
      "step": 15588
    },
    {
      "epoch": 0.8488833527325697,
      "grad_norm": 0.6519114749932312,
      "learning_rate": 0.00012891083358275654,
      "loss": 12.1332,
      "step": 15589
    },
    {
      "epoch": 0.8489378067291526,
      "grad_norm": 0.6107853226476072,
      "learning_rate": 0.00012890239178369058,
      "loss": 12.2532,
      "step": 15590
    },
    {
      "epoch": 0.8489922607257356,
      "grad_norm": 0.6797629066198152,
      "learning_rate": 0.00012889394975987476,
      "loss": 12.1826,
      "step": 15591
    },
    {
      "epoch": 0.8490467147223186,
      "grad_norm": 0.6361362221930984,
      "learning_rate": 0.00012888550751137475,
      "loss": 12.1707,
      "step": 15592
    },
    {
      "epoch": 0.8491011687189016,
      "grad_norm": 0.5971055886039516,
      "learning_rate": 0.0001288770650382562,
      "loss": 12.2782,
      "step": 15593
    },
    {
      "epoch": 0.8491556227154847,
      "grad_norm": 0.5947171593968466,
      "learning_rate": 0.00012886862234058475,
      "loss": 12.0976,
      "step": 15594
    },
    {
      "epoch": 0.8492100767120677,
      "grad_norm": 0.6335157038701741,
      "learning_rate": 0.00012886017941842604,
      "loss": 12.1543,
      "step": 15595
    },
    {
      "epoch": 0.8492645307086507,
      "grad_norm": 0.5592617264152265,
      "learning_rate": 0.00012885173627184571,
      "loss": 12.1406,
      "step": 15596
    },
    {
      "epoch": 0.8493189847052337,
      "grad_norm": 0.6160812598362347,
      "learning_rate": 0.00012884329290090949,
      "loss": 12.1859,
      "step": 15597
    },
    {
      "epoch": 0.8493734387018167,
      "grad_norm": 0.5659969854376821,
      "learning_rate": 0.00012883484930568294,
      "loss": 12.1066,
      "step": 15598
    },
    {
      "epoch": 0.8494278926983997,
      "grad_norm": 0.6664842479810961,
      "learning_rate": 0.0001288264054862318,
      "loss": 12.096,
      "step": 15599
    },
    {
      "epoch": 0.8494823466949828,
      "grad_norm": 0.5276228134648259,
      "learning_rate": 0.00012881796144262168,
      "loss": 12.1223,
      "step": 15600
    },
    {
      "epoch": 0.8495368006915658,
      "grad_norm": 0.5805855702523423,
      "learning_rate": 0.00012880951717491828,
      "loss": 12.2562,
      "step": 15601
    },
    {
      "epoch": 0.8495912546881488,
      "grad_norm": 0.5383428403603512,
      "learning_rate": 0.00012880107268318722,
      "loss": 12.0346,
      "step": 15602
    },
    {
      "epoch": 0.8496457086847318,
      "grad_norm": 0.5522025055589089,
      "learning_rate": 0.00012879262796749422,
      "loss": 12.1612,
      "step": 15603
    },
    {
      "epoch": 0.8497001626813148,
      "grad_norm": 0.583298513098549,
      "learning_rate": 0.00012878418302790488,
      "loss": 12.1072,
      "step": 15604
    },
    {
      "epoch": 0.8497546166778978,
      "grad_norm": 0.6016494841639478,
      "learning_rate": 0.0001287757378644849,
      "loss": 12.3494,
      "step": 15605
    },
    {
      "epoch": 0.8498090706744809,
      "grad_norm": 0.5713091203784781,
      "learning_rate": 0.00012876729247729998,
      "loss": 12.0923,
      "step": 15606
    },
    {
      "epoch": 0.8498635246710639,
      "grad_norm": 0.5960425318662682,
      "learning_rate": 0.00012875884686641578,
      "loss": 12.0263,
      "step": 15607
    },
    {
      "epoch": 0.8499179786676468,
      "grad_norm": 0.5389275337863298,
      "learning_rate": 0.00012875040103189791,
      "loss": 12.1331,
      "step": 15608
    },
    {
      "epoch": 0.8499724326642298,
      "grad_norm": 0.5523982837647737,
      "learning_rate": 0.00012874195497381217,
      "loss": 12.1424,
      "step": 15609
    },
    {
      "epoch": 0.8500268866608128,
      "grad_norm": 0.5161903641796735,
      "learning_rate": 0.0001287335086922241,
      "loss": 12.1487,
      "step": 15610
    },
    {
      "epoch": 0.8500813406573958,
      "grad_norm": 0.544290717987044,
      "learning_rate": 0.0001287250621871995,
      "loss": 12.1779,
      "step": 15611
    },
    {
      "epoch": 0.8501357946539789,
      "grad_norm": 0.5652310341744862,
      "learning_rate": 0.00012871661545880398,
      "loss": 12.2929,
      "step": 15612
    },
    {
      "epoch": 0.8501902486505619,
      "grad_norm": 0.5724418405069429,
      "learning_rate": 0.00012870816850710326,
      "loss": 12.1614,
      "step": 15613
    },
    {
      "epoch": 0.8502447026471449,
      "grad_norm": 0.5898243654550441,
      "learning_rate": 0.000128699721332163,
      "loss": 12.229,
      "step": 15614
    },
    {
      "epoch": 0.8502991566437279,
      "grad_norm": 0.5731288486052545,
      "learning_rate": 0.0001286912739340489,
      "loss": 12.183,
      "step": 15615
    },
    {
      "epoch": 0.8503536106403109,
      "grad_norm": 0.5246223185961127,
      "learning_rate": 0.00012868282631282662,
      "loss": 12.1769,
      "step": 15616
    },
    {
      "epoch": 0.8504080646368939,
      "grad_norm": 0.5049356988743475,
      "learning_rate": 0.00012867437846856186,
      "loss": 12.1367,
      "step": 15617
    },
    {
      "epoch": 0.850462518633477,
      "grad_norm": 0.5481648482068753,
      "learning_rate": 0.00012866593040132036,
      "loss": 12.1416,
      "step": 15618
    },
    {
      "epoch": 0.85051697263006,
      "grad_norm": 0.6076905164308919,
      "learning_rate": 0.00012865748211116776,
      "loss": 12.2317,
      "step": 15619
    },
    {
      "epoch": 0.850571426626643,
      "grad_norm": 0.507112466697084,
      "learning_rate": 0.00012864903359816979,
      "loss": 12.0967,
      "step": 15620
    },
    {
      "epoch": 0.850625880623226,
      "grad_norm": 0.5362129911701178,
      "learning_rate": 0.0001286405848623921,
      "loss": 12.0546,
      "step": 15621
    },
    {
      "epoch": 0.850680334619809,
      "grad_norm": 0.5950019792773952,
      "learning_rate": 0.00012863213590390044,
      "loss": 12.0381,
      "step": 15622
    },
    {
      "epoch": 0.8507347886163921,
      "grad_norm": 0.5447907140074338,
      "learning_rate": 0.0001286236867227605,
      "loss": 12.1993,
      "step": 15623
    },
    {
      "epoch": 0.8507892426129751,
      "grad_norm": 0.5149878848651488,
      "learning_rate": 0.00012861523731903796,
      "loss": 12.0942,
      "step": 15624
    },
    {
      "epoch": 0.850843696609558,
      "grad_norm": 0.5531567677264669,
      "learning_rate": 0.00012860678769279854,
      "loss": 12.2105,
      "step": 15625
    },
    {
      "epoch": 0.850898150606141,
      "grad_norm": 0.5594788185985377,
      "learning_rate": 0.00012859833784410792,
      "loss": 12.1554,
      "step": 15626
    },
    {
      "epoch": 0.850952604602724,
      "grad_norm": 0.5571718604967746,
      "learning_rate": 0.00012858988777303184,
      "loss": 12.2235,
      "step": 15627
    },
    {
      "epoch": 0.851007058599307,
      "grad_norm": 0.7899774324453996,
      "learning_rate": 0.000128581437479636,
      "loss": 12.5309,
      "step": 15628
    },
    {
      "epoch": 0.8510615125958901,
      "grad_norm": 0.5659755971424956,
      "learning_rate": 0.00012857298696398613,
      "loss": 12.2907,
      "step": 15629
    },
    {
      "epoch": 0.8511159665924731,
      "grad_norm": 0.61763375246358,
      "learning_rate": 0.00012856453622614791,
      "loss": 12.2098,
      "step": 15630
    },
    {
      "epoch": 0.8511704205890561,
      "grad_norm": 0.5680070609100679,
      "learning_rate": 0.00012855608526618706,
      "loss": 12.1818,
      "step": 15631
    },
    {
      "epoch": 0.8512248745856391,
      "grad_norm": 0.6291213052887918,
      "learning_rate": 0.0001285476340841693,
      "loss": 12.1265,
      "step": 15632
    },
    {
      "epoch": 0.8512793285822221,
      "grad_norm": 0.5967570939808631,
      "learning_rate": 0.00012853918268016033,
      "loss": 12.1694,
      "step": 15633
    },
    {
      "epoch": 0.8513337825788051,
      "grad_norm": 0.5645989960623116,
      "learning_rate": 0.0001285307310542259,
      "loss": 12.123,
      "step": 15634
    },
    {
      "epoch": 0.8513882365753882,
      "grad_norm": 0.5923944416190499,
      "learning_rate": 0.00012852227920643177,
      "loss": 12.1184,
      "step": 15635
    },
    {
      "epoch": 0.8514426905719712,
      "grad_norm": 0.5343867541462105,
      "learning_rate": 0.00012851382713684358,
      "loss": 12.2321,
      "step": 15636
    },
    {
      "epoch": 0.8514971445685542,
      "grad_norm": 0.5317908527102665,
      "learning_rate": 0.00012850537484552707,
      "loss": 12.0625,
      "step": 15637
    },
    {
      "epoch": 0.8515515985651372,
      "grad_norm": 0.6008950192481002,
      "learning_rate": 0.00012849692233254798,
      "loss": 12.2511,
      "step": 15638
    },
    {
      "epoch": 0.8516060525617202,
      "grad_norm": 0.5891196291828917,
      "learning_rate": 0.00012848846959797206,
      "loss": 12.2549,
      "step": 15639
    },
    {
      "epoch": 0.8516605065583032,
      "grad_norm": 0.6080059299665412,
      "learning_rate": 0.00012848001664186504,
      "loss": 12.2132,
      "step": 15640
    },
    {
      "epoch": 0.8517149605548863,
      "grad_norm": 0.6157269450140649,
      "learning_rate": 0.00012847156346429262,
      "loss": 12.3017,
      "step": 15641
    },
    {
      "epoch": 0.8517694145514693,
      "grad_norm": 0.5276589367835588,
      "learning_rate": 0.00012846311006532054,
      "loss": 12.2921,
      "step": 15642
    },
    {
      "epoch": 0.8518238685480523,
      "grad_norm": 0.5142565017966974,
      "learning_rate": 0.00012845465644501454,
      "loss": 12.1849,
      "step": 15643
    },
    {
      "epoch": 0.8518783225446352,
      "grad_norm": 0.5449135733343222,
      "learning_rate": 0.00012844620260344037,
      "loss": 12.1698,
      "step": 15644
    },
    {
      "epoch": 0.8519327765412182,
      "grad_norm": 0.6327389332555616,
      "learning_rate": 0.00012843774854066377,
      "loss": 12.1762,
      "step": 15645
    },
    {
      "epoch": 0.8519872305378012,
      "grad_norm": 0.5907704006134956,
      "learning_rate": 0.00012842929425675048,
      "loss": 12.2365,
      "step": 15646
    },
    {
      "epoch": 0.8520416845343843,
      "grad_norm": 0.6351242368170997,
      "learning_rate": 0.0001284208397517662,
      "loss": 12.1271,
      "step": 15647
    },
    {
      "epoch": 0.8520961385309673,
      "grad_norm": 0.5151588337199436,
      "learning_rate": 0.0001284123850257767,
      "loss": 12.1418,
      "step": 15648
    },
    {
      "epoch": 0.8521505925275503,
      "grad_norm": 0.5683816891671283,
      "learning_rate": 0.00012840393007884776,
      "loss": 12.1423,
      "step": 15649
    },
    {
      "epoch": 0.8522050465241333,
      "grad_norm": 0.6980382453882409,
      "learning_rate": 0.00012839547491104507,
      "loss": 12.3939,
      "step": 15650
    },
    {
      "epoch": 0.8522595005207163,
      "grad_norm": 0.5468550971970724,
      "learning_rate": 0.00012838701952243439,
      "loss": 12.1403,
      "step": 15651
    },
    {
      "epoch": 0.8523139545172993,
      "grad_norm": 0.5720309246905906,
      "learning_rate": 0.00012837856391308152,
      "loss": 12.1732,
      "step": 15652
    },
    {
      "epoch": 0.8523684085138824,
      "grad_norm": 0.5901994933747409,
      "learning_rate": 0.00012837010808305216,
      "loss": 12.1811,
      "step": 15653
    },
    {
      "epoch": 0.8524228625104654,
      "grad_norm": 0.6417524330224629,
      "learning_rate": 0.00012836165203241207,
      "loss": 12.1895,
      "step": 15654
    },
    {
      "epoch": 0.8524773165070484,
      "grad_norm": 0.5508830074943798,
      "learning_rate": 0.00012835319576122705,
      "loss": 12.2433,
      "step": 15655
    },
    {
      "epoch": 0.8525317705036314,
      "grad_norm": 0.4847934473438708,
      "learning_rate": 0.00012834473926956282,
      "loss": 12.1649,
      "step": 15656
    },
    {
      "epoch": 0.8525862245002144,
      "grad_norm": 0.5106405252112488,
      "learning_rate": 0.0001283362825574851,
      "loss": 12.1226,
      "step": 15657
    },
    {
      "epoch": 0.8526406784967975,
      "grad_norm": 0.6224454278859773,
      "learning_rate": 0.00012832782562505974,
      "loss": 12.4138,
      "step": 15658
    },
    {
      "epoch": 0.8526951324933805,
      "grad_norm": 0.5774410000543585,
      "learning_rate": 0.00012831936847235243,
      "loss": 12.185,
      "step": 15659
    },
    {
      "epoch": 0.8527495864899635,
      "grad_norm": 0.5379968708739294,
      "learning_rate": 0.000128310911099429,
      "loss": 12.121,
      "step": 15660
    },
    {
      "epoch": 0.8528040404865465,
      "grad_norm": 0.5884777063588783,
      "learning_rate": 0.00012830245350635514,
      "loss": 12.2266,
      "step": 15661
    },
    {
      "epoch": 0.8528584944831294,
      "grad_norm": 0.5666623086845319,
      "learning_rate": 0.0001282939956931967,
      "loss": 12.2207,
      "step": 15662
    },
    {
      "epoch": 0.8529129484797124,
      "grad_norm": 0.5922210812425717,
      "learning_rate": 0.00012828553766001935,
      "loss": 12.2144,
      "step": 15663
    },
    {
      "epoch": 0.8529674024762955,
      "grad_norm": 0.5464399974466867,
      "learning_rate": 0.00012827707940688896,
      "loss": 12.3001,
      "step": 15664
    },
    {
      "epoch": 0.8530218564728785,
      "grad_norm": 0.5641139397080874,
      "learning_rate": 0.00012826862093387123,
      "loss": 12.1005,
      "step": 15665
    },
    {
      "epoch": 0.8530763104694615,
      "grad_norm": 0.5914279130202791,
      "learning_rate": 0.00012826016224103196,
      "loss": 12.2504,
      "step": 15666
    },
    {
      "epoch": 0.8531307644660445,
      "grad_norm": 0.5375514541952663,
      "learning_rate": 0.00012825170332843696,
      "loss": 12.18,
      "step": 15667
    },
    {
      "epoch": 0.8531852184626275,
      "grad_norm": 0.5536914518116253,
      "learning_rate": 0.00012824324419615196,
      "loss": 12.0696,
      "step": 15668
    },
    {
      "epoch": 0.8532396724592105,
      "grad_norm": 0.608791386538355,
      "learning_rate": 0.00012823478484424273,
      "loss": 12.2929,
      "step": 15669
    },
    {
      "epoch": 0.8532941264557936,
      "grad_norm": 0.5442849581551492,
      "learning_rate": 0.0001282263252727751,
      "loss": 12.2019,
      "step": 15670
    },
    {
      "epoch": 0.8533485804523766,
      "grad_norm": 0.5579861998109426,
      "learning_rate": 0.00012821786548181485,
      "loss": 12.193,
      "step": 15671
    },
    {
      "epoch": 0.8534030344489596,
      "grad_norm": 0.5299834636731653,
      "learning_rate": 0.00012820940547142773,
      "loss": 12.2381,
      "step": 15672
    },
    {
      "epoch": 0.8534574884455426,
      "grad_norm": 0.7834986858693507,
      "learning_rate": 0.00012820094524167955,
      "loss": 12.1198,
      "step": 15673
    },
    {
      "epoch": 0.8535119424421256,
      "grad_norm": 0.602285605010258,
      "learning_rate": 0.00012819248479263606,
      "loss": 12.2375,
      "step": 15674
    },
    {
      "epoch": 0.8535663964387086,
      "grad_norm": 0.5712320288841317,
      "learning_rate": 0.00012818402412436312,
      "loss": 12.1941,
      "step": 15675
    },
    {
      "epoch": 0.8536208504352917,
      "grad_norm": 0.5429152513655475,
      "learning_rate": 0.00012817556323692646,
      "loss": 12.0621,
      "step": 15676
    },
    {
      "epoch": 0.8536753044318747,
      "grad_norm": 0.6057068946766633,
      "learning_rate": 0.0001281671021303919,
      "loss": 12.2741,
      "step": 15677
    },
    {
      "epoch": 0.8537297584284577,
      "grad_norm": 0.6539577442174344,
      "learning_rate": 0.00012815864080482523,
      "loss": 12.2394,
      "step": 15678
    },
    {
      "epoch": 0.8537842124250407,
      "grad_norm": 0.6474510632946274,
      "learning_rate": 0.00012815017926029224,
      "loss": 12.2298,
      "step": 15679
    },
    {
      "epoch": 0.8538386664216236,
      "grad_norm": 0.6012156573147364,
      "learning_rate": 0.00012814171749685874,
      "loss": 12.2179,
      "step": 15680
    },
    {
      "epoch": 0.8538931204182066,
      "grad_norm": 0.5748682195613363,
      "learning_rate": 0.00012813325551459055,
      "loss": 12.0987,
      "step": 15681
    },
    {
      "epoch": 0.8539475744147897,
      "grad_norm": 0.5932538036371817,
      "learning_rate": 0.0001281247933135534,
      "loss": 12.2372,
      "step": 15682
    },
    {
      "epoch": 0.8540020284113727,
      "grad_norm": 0.5569302402712455,
      "learning_rate": 0.00012811633089381317,
      "loss": 12.1693,
      "step": 15683
    },
    {
      "epoch": 0.8540564824079557,
      "grad_norm": 0.9870608844038362,
      "learning_rate": 0.00012810786825543562,
      "loss": 12.1499,
      "step": 15684
    },
    {
      "epoch": 0.8541109364045387,
      "grad_norm": 0.6173207121444124,
      "learning_rate": 0.0001280994053984866,
      "loss": 12.2683,
      "step": 15685
    },
    {
      "epoch": 0.8541653904011217,
      "grad_norm": 0.6280964771873433,
      "learning_rate": 0.0001280909423230319,
      "loss": 12.2718,
      "step": 15686
    },
    {
      "epoch": 0.8542198443977048,
      "grad_norm": 0.619197910380965,
      "learning_rate": 0.00012808247902913725,
      "loss": 12.2047,
      "step": 15687
    },
    {
      "epoch": 0.8542742983942878,
      "grad_norm": 0.6570766982144078,
      "learning_rate": 0.0001280740155168686,
      "loss": 12.314,
      "step": 15688
    },
    {
      "epoch": 0.8543287523908708,
      "grad_norm": 0.7308763519060135,
      "learning_rate": 0.00012806555178629167,
      "loss": 12.1709,
      "step": 15689
    },
    {
      "epoch": 0.8543832063874538,
      "grad_norm": 0.6809149470383755,
      "learning_rate": 0.00012805708783747232,
      "loss": 12.1298,
      "step": 15690
    },
    {
      "epoch": 0.8544376603840368,
      "grad_norm": 0.5754298989534565,
      "learning_rate": 0.00012804862367047637,
      "loss": 12.1261,
      "step": 15691
    },
    {
      "epoch": 0.8544921143806198,
      "grad_norm": 0.645798903559039,
      "learning_rate": 0.00012804015928536956,
      "loss": 12.1758,
      "step": 15692
    },
    {
      "epoch": 0.8545465683772029,
      "grad_norm": 0.5825265322505782,
      "learning_rate": 0.00012803169468221778,
      "loss": 12.1584,
      "step": 15693
    },
    {
      "epoch": 0.8546010223737859,
      "grad_norm": 0.5700802449068311,
      "learning_rate": 0.00012802322986108687,
      "loss": 12.1773,
      "step": 15694
    },
    {
      "epoch": 0.8546554763703689,
      "grad_norm": 0.5854575405293855,
      "learning_rate": 0.0001280147648220426,
      "loss": 12.1983,
      "step": 15695
    },
    {
      "epoch": 0.8547099303669519,
      "grad_norm": 0.559123078988021,
      "learning_rate": 0.00012800629956515088,
      "loss": 12.1285,
      "step": 15696
    },
    {
      "epoch": 0.8547643843635349,
      "grad_norm": 0.5305252493834495,
      "learning_rate": 0.0001279978340904774,
      "loss": 12.1489,
      "step": 15697
    },
    {
      "epoch": 0.8548188383601178,
      "grad_norm": 0.6947880528017051,
      "learning_rate": 0.00012798936839808811,
      "loss": 12.2499,
      "step": 15698
    },
    {
      "epoch": 0.854873292356701,
      "grad_norm": 0.5427173645269607,
      "learning_rate": 0.00012798090248804876,
      "loss": 12.0906,
      "step": 15699
    },
    {
      "epoch": 0.854927746353284,
      "grad_norm": 0.5586751572827647,
      "learning_rate": 0.00012797243636042527,
      "loss": 12.1936,
      "step": 15700
    },
    {
      "epoch": 0.8549822003498669,
      "grad_norm": 0.5557568132196792,
      "learning_rate": 0.0001279639700152834,
      "loss": 12.2273,
      "step": 15701
    },
    {
      "epoch": 0.8550366543464499,
      "grad_norm": 0.5682109416724617,
      "learning_rate": 0.00012795550345268903,
      "loss": 12.2523,
      "step": 15702
    },
    {
      "epoch": 0.8550911083430329,
      "grad_norm": 0.5987692325981113,
      "learning_rate": 0.00012794703667270795,
      "loss": 12.2351,
      "step": 15703
    },
    {
      "epoch": 0.8551455623396159,
      "grad_norm": 0.5738180654741157,
      "learning_rate": 0.00012793856967540602,
      "loss": 12.2044,
      "step": 15704
    },
    {
      "epoch": 0.855200016336199,
      "grad_norm": 0.556910543590465,
      "learning_rate": 0.00012793010246084908,
      "loss": 12.1197,
      "step": 15705
    },
    {
      "epoch": 0.855254470332782,
      "grad_norm": 0.6121439251457784,
      "learning_rate": 0.00012792163502910303,
      "loss": 12.0928,
      "step": 15706
    },
    {
      "epoch": 0.855308924329365,
      "grad_norm": 0.564178283270504,
      "learning_rate": 0.00012791316738023365,
      "loss": 12.211,
      "step": 15707
    },
    {
      "epoch": 0.855363378325948,
      "grad_norm": 0.5654129343631402,
      "learning_rate": 0.00012790469951430678,
      "loss": 12.2054,
      "step": 15708
    },
    {
      "epoch": 0.855417832322531,
      "grad_norm": 0.5129254702848438,
      "learning_rate": 0.00012789623143138828,
      "loss": 12.2229,
      "step": 15709
    },
    {
      "epoch": 0.855472286319114,
      "grad_norm": 0.5557932268999569,
      "learning_rate": 0.000127887763131544,
      "loss": 12.2176,
      "step": 15710
    },
    {
      "epoch": 0.8555267403156971,
      "grad_norm": 0.5486285626359496,
      "learning_rate": 0.00012787929461483983,
      "loss": 12.2681,
      "step": 15711
    },
    {
      "epoch": 0.8555811943122801,
      "grad_norm": 0.5652838634167313,
      "learning_rate": 0.0001278708258813416,
      "loss": 12.1061,
      "step": 15712
    },
    {
      "epoch": 0.8556356483088631,
      "grad_norm": 0.5872954549608131,
      "learning_rate": 0.0001278623569311151,
      "loss": 12.0819,
      "step": 15713
    },
    {
      "epoch": 0.8556901023054461,
      "grad_norm": 0.5543051339598092,
      "learning_rate": 0.00012785388776422626,
      "loss": 12.1606,
      "step": 15714
    },
    {
      "epoch": 0.8557445563020291,
      "grad_norm": 0.6378223338713711,
      "learning_rate": 0.0001278454183807409,
      "loss": 12.202,
      "step": 15715
    },
    {
      "epoch": 0.855799010298612,
      "grad_norm": 0.5744170509509097,
      "learning_rate": 0.00012783694878072495,
      "loss": 12.1183,
      "step": 15716
    },
    {
      "epoch": 0.8558534642951952,
      "grad_norm": 0.5258546467076047,
      "learning_rate": 0.00012782847896424418,
      "loss": 12.1676,
      "step": 15717
    },
    {
      "epoch": 0.8559079182917781,
      "grad_norm": 0.6485700887433437,
      "learning_rate": 0.00012782000893136452,
      "loss": 12.3775,
      "step": 15718
    },
    {
      "epoch": 0.8559623722883611,
      "grad_norm": 0.5813874713791451,
      "learning_rate": 0.00012781153868215178,
      "loss": 12.3445,
      "step": 15719
    },
    {
      "epoch": 0.8560168262849441,
      "grad_norm": 0.5446866086229591,
      "learning_rate": 0.00012780306821667185,
      "loss": 12.1281,
      "step": 15720
    },
    {
      "epoch": 0.8560712802815271,
      "grad_norm": 0.617440955219744,
      "learning_rate": 0.00012779459753499062,
      "loss": 12.3763,
      "step": 15721
    },
    {
      "epoch": 0.8561257342781102,
      "grad_norm": 0.5485880847095598,
      "learning_rate": 0.0001277861266371739,
      "loss": 12.1894,
      "step": 15722
    },
    {
      "epoch": 0.8561801882746932,
      "grad_norm": 0.6267972681339404,
      "learning_rate": 0.00012777765552328765,
      "loss": 12.1803,
      "step": 15723
    },
    {
      "epoch": 0.8562346422712762,
      "grad_norm": 0.6028696142522535,
      "learning_rate": 0.00012776918419339764,
      "loss": 12.0883,
      "step": 15724
    },
    {
      "epoch": 0.8562890962678592,
      "grad_norm": 0.5685786219339078,
      "learning_rate": 0.00012776071264756985,
      "loss": 12.1525,
      "step": 15725
    },
    {
      "epoch": 0.8563435502644422,
      "grad_norm": 0.7374933028055527,
      "learning_rate": 0.00012775224088587005,
      "loss": 12.2591,
      "step": 15726
    },
    {
      "epoch": 0.8563980042610252,
      "grad_norm": 0.5980493259753991,
      "learning_rate": 0.0001277437689083642,
      "loss": 12.1731,
      "step": 15727
    },
    {
      "epoch": 0.8564524582576083,
      "grad_norm": 0.5843647706304774,
      "learning_rate": 0.00012773529671511816,
      "loss": 12.1551,
      "step": 15728
    },
    {
      "epoch": 0.8565069122541913,
      "grad_norm": 0.5503302278270538,
      "learning_rate": 0.00012772682430619778,
      "loss": 12.2297,
      "step": 15729
    },
    {
      "epoch": 0.8565613662507743,
      "grad_norm": 0.6475393445604003,
      "learning_rate": 0.000127718351681669,
      "loss": 12.2689,
      "step": 15730
    },
    {
      "epoch": 0.8566158202473573,
      "grad_norm": 0.5359188086214569,
      "learning_rate": 0.0001277098788415976,
      "loss": 12.1432,
      "step": 15731
    },
    {
      "epoch": 0.8566702742439403,
      "grad_norm": 0.563354163807259,
      "learning_rate": 0.0001277014057860496,
      "loss": 12.2512,
      "step": 15732
    },
    {
      "epoch": 0.8567247282405233,
      "grad_norm": 0.5281041117124711,
      "learning_rate": 0.0001276929325150908,
      "loss": 12.1194,
      "step": 15733
    },
    {
      "epoch": 0.8567791822371064,
      "grad_norm": 0.556022921641765,
      "learning_rate": 0.00012768445902878713,
      "loss": 12.1332,
      "step": 15734
    },
    {
      "epoch": 0.8568336362336894,
      "grad_norm": 0.5349451535607578,
      "learning_rate": 0.00012767598532720443,
      "loss": 12.0863,
      "step": 15735
    },
    {
      "epoch": 0.8568880902302723,
      "grad_norm": 0.608521578203048,
      "learning_rate": 0.00012766751141040866,
      "loss": 12.2161,
      "step": 15736
    },
    {
      "epoch": 0.8569425442268553,
      "grad_norm": 0.543320017849001,
      "learning_rate": 0.00012765903727846565,
      "loss": 12.1176,
      "step": 15737
    },
    {
      "epoch": 0.8569969982234383,
      "grad_norm": 0.5457695761142336,
      "learning_rate": 0.00012765056293144133,
      "loss": 12.2448,
      "step": 15738
    },
    {
      "epoch": 0.8570514522200213,
      "grad_norm": 0.554684247229752,
      "learning_rate": 0.0001276420883694016,
      "loss": 12.0115,
      "step": 15739
    },
    {
      "epoch": 0.8571059062166044,
      "grad_norm": 0.5336282304607382,
      "learning_rate": 0.00012763361359241238,
      "loss": 12.0986,
      "step": 15740
    },
    {
      "epoch": 0.8571603602131874,
      "grad_norm": 0.5906900125833782,
      "learning_rate": 0.00012762513860053955,
      "loss": 12.1907,
      "step": 15741
    },
    {
      "epoch": 0.8572148142097704,
      "grad_norm": 0.5163197838124347,
      "learning_rate": 0.00012761666339384896,
      "loss": 12.1659,
      "step": 15742
    },
    {
      "epoch": 0.8572692682063534,
      "grad_norm": 0.5845795882841365,
      "learning_rate": 0.0001276081879724066,
      "loss": 12.3958,
      "step": 15743
    },
    {
      "epoch": 0.8573237222029364,
      "grad_norm": 0.5321494538582507,
      "learning_rate": 0.00012759971233627834,
      "loss": 12.1918,
      "step": 15744
    },
    {
      "epoch": 0.8573781761995194,
      "grad_norm": 0.5166924507597535,
      "learning_rate": 0.00012759123648553006,
      "loss": 12.1325,
      "step": 15745
    },
    {
      "epoch": 0.8574326301961025,
      "grad_norm": 0.587566718750353,
      "learning_rate": 0.00012758276042022776,
      "loss": 12.1939,
      "step": 15746
    },
    {
      "epoch": 0.8574870841926855,
      "grad_norm": 0.6003595372439998,
      "learning_rate": 0.0001275742841404372,
      "loss": 12.2588,
      "step": 15747
    },
    {
      "epoch": 0.8575415381892685,
      "grad_norm": 0.7960268013196092,
      "learning_rate": 0.00012756580764622445,
      "loss": 12.4471,
      "step": 15748
    },
    {
      "epoch": 0.8575959921858515,
      "grad_norm": 0.5824598169241241,
      "learning_rate": 0.00012755733093765533,
      "loss": 12.2527,
      "step": 15749
    },
    {
      "epoch": 0.8576504461824345,
      "grad_norm": 0.551996715410952,
      "learning_rate": 0.00012754885401479582,
      "loss": 12.107,
      "step": 15750
    },
    {
      "epoch": 0.8577049001790175,
      "grad_norm": 0.5434241010597629,
      "learning_rate": 0.00012754037687771178,
      "loss": 12.1144,
      "step": 15751
    },
    {
      "epoch": 0.8577593541756006,
      "grad_norm": 0.6147048798357491,
      "learning_rate": 0.00012753189952646916,
      "loss": 12.1863,
      "step": 15752
    },
    {
      "epoch": 0.8578138081721836,
      "grad_norm": 0.6034774893716371,
      "learning_rate": 0.00012752342196113383,
      "loss": 12.2386,
      "step": 15753
    },
    {
      "epoch": 0.8578682621687665,
      "grad_norm": 0.5959562157298051,
      "learning_rate": 0.0001275149441817718,
      "loss": 12.2741,
      "step": 15754
    },
    {
      "epoch": 0.8579227161653495,
      "grad_norm": 0.5231375209394874,
      "learning_rate": 0.0001275064661884489,
      "loss": 12.1576,
      "step": 15755
    },
    {
      "epoch": 0.8579771701619325,
      "grad_norm": 0.5972949129909878,
      "learning_rate": 0.00012749798798123116,
      "loss": 12.1434,
      "step": 15756
    },
    {
      "epoch": 0.8580316241585156,
      "grad_norm": 0.6122546872739701,
      "learning_rate": 0.00012748950956018444,
      "loss": 12.1824,
      "step": 15757
    },
    {
      "epoch": 0.8580860781550986,
      "grad_norm": 0.548827539391232,
      "learning_rate": 0.00012748103092537466,
      "loss": 12.1858,
      "step": 15758
    },
    {
      "epoch": 0.8581405321516816,
      "grad_norm": 0.5268961504095557,
      "learning_rate": 0.00012747255207686778,
      "loss": 12.1409,
      "step": 15759
    },
    {
      "epoch": 0.8581949861482646,
      "grad_norm": 0.6139714662781754,
      "learning_rate": 0.00012746407301472974,
      "loss": 12.2219,
      "step": 15760
    },
    {
      "epoch": 0.8582494401448476,
      "grad_norm": 0.5731399770059618,
      "learning_rate": 0.00012745559373902648,
      "loss": 12.2019,
      "step": 15761
    },
    {
      "epoch": 0.8583038941414306,
      "grad_norm": 0.7382380608364358,
      "learning_rate": 0.0001274471142498239,
      "loss": 12.0339,
      "step": 15762
    },
    {
      "epoch": 0.8583583481380137,
      "grad_norm": 0.6025951957591087,
      "learning_rate": 0.00012743863454718797,
      "loss": 12.1943,
      "step": 15763
    },
    {
      "epoch": 0.8584128021345967,
      "grad_norm": 0.5695672727653155,
      "learning_rate": 0.00012743015463118458,
      "loss": 12.3162,
      "step": 15764
    },
    {
      "epoch": 0.8584672561311797,
      "grad_norm": 0.5282926940371087,
      "learning_rate": 0.0001274216745018797,
      "loss": 12.1697,
      "step": 15765
    },
    {
      "epoch": 0.8585217101277627,
      "grad_norm": 0.6295977037542692,
      "learning_rate": 0.00012741319415933934,
      "loss": 12.4558,
      "step": 15766
    },
    {
      "epoch": 0.8585761641243457,
      "grad_norm": 0.5456157031705988,
      "learning_rate": 0.00012740471360362938,
      "loss": 12.213,
      "step": 15767
    },
    {
      "epoch": 0.8586306181209287,
      "grad_norm": 0.5722528479712246,
      "learning_rate": 0.00012739623283481572,
      "loss": 12.0039,
      "step": 15768
    },
    {
      "epoch": 0.8586850721175118,
      "grad_norm": 0.5259898355911858,
      "learning_rate": 0.0001273877518529644,
      "loss": 12.1635,
      "step": 15769
    },
    {
      "epoch": 0.8587395261140948,
      "grad_norm": 0.5325491294898629,
      "learning_rate": 0.00012737927065814127,
      "loss": 12.216,
      "step": 15770
    },
    {
      "epoch": 0.8587939801106778,
      "grad_norm": 0.5831718666607144,
      "learning_rate": 0.00012737078925041244,
      "loss": 12.2157,
      "step": 15771
    },
    {
      "epoch": 0.8588484341072607,
      "grad_norm": 0.5362693585664796,
      "learning_rate": 0.0001273623076298437,
      "loss": 12.0752,
      "step": 15772
    },
    {
      "epoch": 0.8589028881038437,
      "grad_norm": 0.5317601893405478,
      "learning_rate": 0.00012735382579650106,
      "loss": 12.1272,
      "step": 15773
    },
    {
      "epoch": 0.8589573421004267,
      "grad_norm": 0.5385207556616182,
      "learning_rate": 0.0001273453437504505,
      "loss": 12.2153,
      "step": 15774
    },
    {
      "epoch": 0.8590117960970098,
      "grad_norm": 0.5966096170591533,
      "learning_rate": 0.00012733686149175795,
      "loss": 12.3078,
      "step": 15775
    },
    {
      "epoch": 0.8590662500935928,
      "grad_norm": 0.5946618340368106,
      "learning_rate": 0.00012732837902048943,
      "loss": 12.1621,
      "step": 15776
    },
    {
      "epoch": 0.8591207040901758,
      "grad_norm": 0.5682263517375123,
      "learning_rate": 0.00012731989633671078,
      "loss": 12.2022,
      "step": 15777
    },
    {
      "epoch": 0.8591751580867588,
      "grad_norm": 0.6371370179585477,
      "learning_rate": 0.00012731141344048808,
      "loss": 12.2738,
      "step": 15778
    },
    {
      "epoch": 0.8592296120833418,
      "grad_norm": 0.562737486394942,
      "learning_rate": 0.00012730293033188722,
      "loss": 12.2651,
      "step": 15779
    },
    {
      "epoch": 0.8592840660799248,
      "grad_norm": 0.5862226570010126,
      "learning_rate": 0.0001272944470109742,
      "loss": 12.2035,
      "step": 15780
    },
    {
      "epoch": 0.8593385200765079,
      "grad_norm": 0.5991217322704121,
      "learning_rate": 0.000127285963477815,
      "loss": 12.213,
      "step": 15781
    },
    {
      "epoch": 0.8593929740730909,
      "grad_norm": 0.5343548554741508,
      "learning_rate": 0.00012727747973247558,
      "loss": 12.1085,
      "step": 15782
    },
    {
      "epoch": 0.8594474280696739,
      "grad_norm": 0.6527294044364248,
      "learning_rate": 0.0001272689957750219,
      "loss": 12.2033,
      "step": 15783
    },
    {
      "epoch": 0.8595018820662569,
      "grad_norm": 0.5903937465005267,
      "learning_rate": 0.0001272605116055199,
      "loss": 12.221,
      "step": 15784
    },
    {
      "epoch": 0.8595563360628399,
      "grad_norm": 0.5348246661864136,
      "learning_rate": 0.00012725202722403561,
      "loss": 12.1709,
      "step": 15785
    },
    {
      "epoch": 0.8596107900594229,
      "grad_norm": 0.5726522511258787,
      "learning_rate": 0.000127243542630635,
      "loss": 12.1589,
      "step": 15786
    },
    {
      "epoch": 0.859665244056006,
      "grad_norm": 0.5883931333080112,
      "learning_rate": 0.00012723505782538403,
      "loss": 12.1653,
      "step": 15787
    },
    {
      "epoch": 0.859719698052589,
      "grad_norm": 0.5167501422138584,
      "learning_rate": 0.00012722657280834866,
      "loss": 12.1629,
      "step": 15788
    },
    {
      "epoch": 0.859774152049172,
      "grad_norm": 0.5965163354165692,
      "learning_rate": 0.00012721808757959493,
      "loss": 12.1697,
      "step": 15789
    },
    {
      "epoch": 0.859828606045755,
      "grad_norm": 0.6206399193062977,
      "learning_rate": 0.00012720960213918875,
      "loss": 12.1275,
      "step": 15790
    },
    {
      "epoch": 0.8598830600423379,
      "grad_norm": 0.5041028697890138,
      "learning_rate": 0.00012720111648719618,
      "loss": 11.9535,
      "step": 15791
    },
    {
      "epoch": 0.859937514038921,
      "grad_norm": 0.5346675298940814,
      "learning_rate": 0.0001271926306236831,
      "loss": 12.1904,
      "step": 15792
    },
    {
      "epoch": 0.859991968035504,
      "grad_norm": 0.6794951925620087,
      "learning_rate": 0.00012718414454871563,
      "loss": 12.2198,
      "step": 15793
    },
    {
      "epoch": 0.860046422032087,
      "grad_norm": 0.5376773252891305,
      "learning_rate": 0.00012717565826235967,
      "loss": 12.196,
      "step": 15794
    },
    {
      "epoch": 0.86010087602867,
      "grad_norm": 0.572556725984343,
      "learning_rate": 0.00012716717176468125,
      "loss": 12.1767,
      "step": 15795
    },
    {
      "epoch": 0.860155330025253,
      "grad_norm": 0.5859922266043762,
      "learning_rate": 0.00012715868505574633,
      "loss": 12.2814,
      "step": 15796
    },
    {
      "epoch": 0.860209784021836,
      "grad_norm": 0.5422575776659524,
      "learning_rate": 0.00012715019813562092,
      "loss": 12.1301,
      "step": 15797
    },
    {
      "epoch": 0.8602642380184191,
      "grad_norm": 0.563991747041213,
      "learning_rate": 0.00012714171100437101,
      "loss": 12.2544,
      "step": 15798
    },
    {
      "epoch": 0.8603186920150021,
      "grad_norm": 0.5510860391557669,
      "learning_rate": 0.00012713322366206262,
      "loss": 12.1942,
      "step": 15799
    },
    {
      "epoch": 0.8603731460115851,
      "grad_norm": 0.5276618685659437,
      "learning_rate": 0.00012712473610876173,
      "loss": 12.1224,
      "step": 15800
    },
    {
      "epoch": 0.8604276000081681,
      "grad_norm": 0.5539203820098891,
      "learning_rate": 0.00012711624834453434,
      "loss": 12.2318,
      "step": 15801
    },
    {
      "epoch": 0.8604820540047511,
      "grad_norm": 0.5027654839906905,
      "learning_rate": 0.00012710776036944644,
      "loss": 12.1669,
      "step": 15802
    },
    {
      "epoch": 0.8605365080013341,
      "grad_norm": 0.5413982573232451,
      "learning_rate": 0.00012709927218356408,
      "loss": 12.1924,
      "step": 15803
    },
    {
      "epoch": 0.8605909619979172,
      "grad_norm": 0.5670345239907993,
      "learning_rate": 0.0001270907837869532,
      "loss": 12.2394,
      "step": 15804
    },
    {
      "epoch": 0.8606454159945002,
      "grad_norm": 0.6040295601424962,
      "learning_rate": 0.00012708229517967987,
      "loss": 12.2011,
      "step": 15805
    },
    {
      "epoch": 0.8606998699910832,
      "grad_norm": 0.5927785994065459,
      "learning_rate": 0.00012707380636181004,
      "loss": 12.2454,
      "step": 15806
    },
    {
      "epoch": 0.8607543239876662,
      "grad_norm": 0.6117346037164473,
      "learning_rate": 0.0001270653173334098,
      "loss": 12.3725,
      "step": 15807
    },
    {
      "epoch": 0.8608087779842492,
      "grad_norm": 0.5259833447062889,
      "learning_rate": 0.0001270568280945451,
      "loss": 12.1544,
      "step": 15808
    },
    {
      "epoch": 0.8608632319808321,
      "grad_norm": 0.5304112166595931,
      "learning_rate": 0.00012704833864528195,
      "loss": 12.1403,
      "step": 15809
    },
    {
      "epoch": 0.8609176859774152,
      "grad_norm": 0.5797000367923341,
      "learning_rate": 0.00012703984898568637,
      "loss": 12.182,
      "step": 15810
    },
    {
      "epoch": 0.8609721399739982,
      "grad_norm": 0.6167607347967838,
      "learning_rate": 0.0001270313591158244,
      "loss": 12.2032,
      "step": 15811
    },
    {
      "epoch": 0.8610265939705812,
      "grad_norm": 0.6009461728701405,
      "learning_rate": 0.00012702286903576207,
      "loss": 12.2631,
      "step": 15812
    },
    {
      "epoch": 0.8610810479671642,
      "grad_norm": 0.5570489244263896,
      "learning_rate": 0.00012701437874556537,
      "loss": 12.2562,
      "step": 15813
    },
    {
      "epoch": 0.8611355019637472,
      "grad_norm": 0.5456738299269331,
      "learning_rate": 0.0001270058882453003,
      "loss": 12.1073,
      "step": 15814
    },
    {
      "epoch": 0.8611899559603302,
      "grad_norm": 0.5308395603172099,
      "learning_rate": 0.00012699739753503295,
      "loss": 12.3165,
      "step": 15815
    },
    {
      "epoch": 0.8612444099569133,
      "grad_norm": 0.5629932726871376,
      "learning_rate": 0.0001269889066148293,
      "loss": 12.0754,
      "step": 15816
    },
    {
      "epoch": 0.8612988639534963,
      "grad_norm": 0.5583612163321409,
      "learning_rate": 0.00012698041548475539,
      "loss": 12.055,
      "step": 15817
    },
    {
      "epoch": 0.8613533179500793,
      "grad_norm": 0.5271745262645791,
      "learning_rate": 0.00012697192414487724,
      "loss": 12.2464,
      "step": 15818
    },
    {
      "epoch": 0.8614077719466623,
      "grad_norm": 0.7819952104852348,
      "learning_rate": 0.00012696343259526087,
      "loss": 12.2299,
      "step": 15819
    },
    {
      "epoch": 0.8614622259432453,
      "grad_norm": 0.5765961651972086,
      "learning_rate": 0.00012695494083597234,
      "loss": 12.0144,
      "step": 15820
    },
    {
      "epoch": 0.8615166799398284,
      "grad_norm": 0.5198545087241869,
      "learning_rate": 0.00012694644886707766,
      "loss": 12.0843,
      "step": 15821
    },
    {
      "epoch": 0.8615711339364114,
      "grad_norm": 0.5480677269766174,
      "learning_rate": 0.0001269379566886429,
      "loss": 12.1978,
      "step": 15822
    },
    {
      "epoch": 0.8616255879329944,
      "grad_norm": 0.5826026538724887,
      "learning_rate": 0.00012692946430073408,
      "loss": 12.1008,
      "step": 15823
    },
    {
      "epoch": 0.8616800419295774,
      "grad_norm": 0.6127132481854947,
      "learning_rate": 0.00012692097170341718,
      "loss": 12.0893,
      "step": 15824
    },
    {
      "epoch": 0.8617344959261604,
      "grad_norm": 0.5628802050208016,
      "learning_rate": 0.0001269124788967583,
      "loss": 12.2108,
      "step": 15825
    },
    {
      "epoch": 0.8617889499227434,
      "grad_norm": 0.5169071358368816,
      "learning_rate": 0.00012690398588082347,
      "loss": 12.163,
      "step": 15826
    },
    {
      "epoch": 0.8618434039193265,
      "grad_norm": 0.6288968216432084,
      "learning_rate": 0.00012689549265567878,
      "loss": 12.2239,
      "step": 15827
    },
    {
      "epoch": 0.8618978579159094,
      "grad_norm": 0.5522019003477887,
      "learning_rate": 0.0001268869992213902,
      "loss": 12.232,
      "step": 15828
    },
    {
      "epoch": 0.8619523119124924,
      "grad_norm": 0.560181845302788,
      "learning_rate": 0.0001268785055780238,
      "loss": 12.1887,
      "step": 15829
    },
    {
      "epoch": 0.8620067659090754,
      "grad_norm": 0.6260370544594249,
      "learning_rate": 0.0001268700117256456,
      "loss": 12.1765,
      "step": 15830
    },
    {
      "epoch": 0.8620612199056584,
      "grad_norm": 0.5681163602908822,
      "learning_rate": 0.00012686151766432173,
      "loss": 12.164,
      "step": 15831
    },
    {
      "epoch": 0.8621156739022414,
      "grad_norm": 0.5935928659798232,
      "learning_rate": 0.0001268530233941182,
      "loss": 12.1859,
      "step": 15832
    },
    {
      "epoch": 0.8621701278988245,
      "grad_norm": 0.5675365643912833,
      "learning_rate": 0.00012684452891510104,
      "loss": 12.1672,
      "step": 15833
    },
    {
      "epoch": 0.8622245818954075,
      "grad_norm": 0.6448365470006007,
      "learning_rate": 0.00012683603422733631,
      "loss": 12.2005,
      "step": 15834
    },
    {
      "epoch": 0.8622790358919905,
      "grad_norm": 0.5880991860675321,
      "learning_rate": 0.0001268275393308901,
      "loss": 12.088,
      "step": 15835
    },
    {
      "epoch": 0.8623334898885735,
      "grad_norm": 0.5695140928744413,
      "learning_rate": 0.0001268190442258284,
      "loss": 12.0658,
      "step": 15836
    },
    {
      "epoch": 0.8623879438851565,
      "grad_norm": 0.5976367558594671,
      "learning_rate": 0.00012681054891221737,
      "loss": 12.1741,
      "step": 15837
    },
    {
      "epoch": 0.8624423978817395,
      "grad_norm": 0.588921568249969,
      "learning_rate": 0.000126802053390123,
      "loss": 12.1703,
      "step": 15838
    },
    {
      "epoch": 0.8624968518783226,
      "grad_norm": 0.5867597648512988,
      "learning_rate": 0.00012679355765961135,
      "loss": 12.2174,
      "step": 15839
    },
    {
      "epoch": 0.8625513058749056,
      "grad_norm": 0.6466506954919735,
      "learning_rate": 0.00012678506172074852,
      "loss": 12.3293,
      "step": 15840
    },
    {
      "epoch": 0.8626057598714886,
      "grad_norm": 0.5361124196749647,
      "learning_rate": 0.00012677656557360053,
      "loss": 12.1079,
      "step": 15841
    },
    {
      "epoch": 0.8626602138680716,
      "grad_norm": 0.6141638826558271,
      "learning_rate": 0.00012676806921823352,
      "loss": 12.2044,
      "step": 15842
    },
    {
      "epoch": 0.8627146678646546,
      "grad_norm": 0.5573530110558617,
      "learning_rate": 0.0001267595726547135,
      "loss": 12.0857,
      "step": 15843
    },
    {
      "epoch": 0.8627691218612376,
      "grad_norm": 0.5294474626887843,
      "learning_rate": 0.00012675107588310653,
      "loss": 12.1398,
      "step": 15844
    },
    {
      "epoch": 0.8628235758578207,
      "grad_norm": 0.547586828976658,
      "learning_rate": 0.00012674257890347873,
      "loss": 12.2056,
      "step": 15845
    },
    {
      "epoch": 0.8628780298544036,
      "grad_norm": 0.5949898096535011,
      "learning_rate": 0.00012673408171589616,
      "loss": 12.357,
      "step": 15846
    },
    {
      "epoch": 0.8629324838509866,
      "grad_norm": 0.6332146130789029,
      "learning_rate": 0.00012672558432042487,
      "loss": 12.138,
      "step": 15847
    },
    {
      "epoch": 0.8629869378475696,
      "grad_norm": 0.5961911285767854,
      "learning_rate": 0.00012671708671713097,
      "loss": 12.1459,
      "step": 15848
    },
    {
      "epoch": 0.8630413918441526,
      "grad_norm": 0.5387073458048571,
      "learning_rate": 0.00012670858890608048,
      "loss": 12.1757,
      "step": 15849
    },
    {
      "epoch": 0.8630958458407356,
      "grad_norm": 0.5953776124197804,
      "learning_rate": 0.00012670009088733955,
      "loss": 12.2023,
      "step": 15850
    },
    {
      "epoch": 0.8631502998373187,
      "grad_norm": 0.5234957418215699,
      "learning_rate": 0.00012669159266097426,
      "loss": 12.1306,
      "step": 15851
    },
    {
      "epoch": 0.8632047538339017,
      "grad_norm": 0.6395342720574844,
      "learning_rate": 0.00012668309422705063,
      "loss": 12.3002,
      "step": 15852
    },
    {
      "epoch": 0.8632592078304847,
      "grad_norm": 0.5452036442524539,
      "learning_rate": 0.0001266745955856348,
      "loss": 12.1587,
      "step": 15853
    },
    {
      "epoch": 0.8633136618270677,
      "grad_norm": 0.5731158621347431,
      "learning_rate": 0.0001266660967367929,
      "loss": 12.2635,
      "step": 15854
    },
    {
      "epoch": 0.8633681158236507,
      "grad_norm": 0.5685607595502273,
      "learning_rate": 0.00012665759768059085,
      "loss": 12.2997,
      "step": 15855
    },
    {
      "epoch": 0.8634225698202338,
      "grad_norm": 0.578836813347422,
      "learning_rate": 0.00012664909841709495,
      "loss": 12.0963,
      "step": 15856
    },
    {
      "epoch": 0.8634770238168168,
      "grad_norm": 0.5947050777966181,
      "learning_rate": 0.00012664059894637112,
      "loss": 12.2036,
      "step": 15857
    },
    {
      "epoch": 0.8635314778133998,
      "grad_norm": 0.5946116539132492,
      "learning_rate": 0.00012663209926848555,
      "loss": 12.0137,
      "step": 15858
    },
    {
      "epoch": 0.8635859318099828,
      "grad_norm": 0.5032585307333325,
      "learning_rate": 0.00012662359938350433,
      "loss": 12.1626,
      "step": 15859
    },
    {
      "epoch": 0.8636403858065658,
      "grad_norm": 0.5593607639097642,
      "learning_rate": 0.00012661509929149352,
      "loss": 12.0341,
      "step": 15860
    },
    {
      "epoch": 0.8636948398031488,
      "grad_norm": 0.6007704384398124,
      "learning_rate": 0.00012660659899251924,
      "loss": 12.3132,
      "step": 15861
    },
    {
      "epoch": 0.8637492937997319,
      "grad_norm": 0.6161230813975721,
      "learning_rate": 0.00012659809848664756,
      "loss": 12.0831,
      "step": 15862
    },
    {
      "epoch": 0.8638037477963149,
      "grad_norm": 0.5722203566505186,
      "learning_rate": 0.00012658959777394463,
      "loss": 12.2513,
      "step": 15863
    },
    {
      "epoch": 0.8638582017928979,
      "grad_norm": 0.5794489421825233,
      "learning_rate": 0.00012658109685447652,
      "loss": 12.272,
      "step": 15864
    },
    {
      "epoch": 0.8639126557894808,
      "grad_norm": 0.6115055730524654,
      "learning_rate": 0.00012657259572830935,
      "loss": 12.2261,
      "step": 15865
    },
    {
      "epoch": 0.8639671097860638,
      "grad_norm": 0.5648488003671629,
      "learning_rate": 0.0001265640943955092,
      "loss": 12.2224,
      "step": 15866
    },
    {
      "epoch": 0.8640215637826468,
      "grad_norm": 0.5465945861616255,
      "learning_rate": 0.00012655559285614225,
      "loss": 12.189,
      "step": 15867
    },
    {
      "epoch": 0.8640760177792299,
      "grad_norm": 0.6274956086793052,
      "learning_rate": 0.00012654709111027448,
      "loss": 12.1032,
      "step": 15868
    },
    {
      "epoch": 0.8641304717758129,
      "grad_norm": 0.5903592400481897,
      "learning_rate": 0.00012653858915797212,
      "loss": 12.1017,
      "step": 15869
    },
    {
      "epoch": 0.8641849257723959,
      "grad_norm": 0.5421409523757088,
      "learning_rate": 0.00012653008699930123,
      "loss": 12.1491,
      "step": 15870
    },
    {
      "epoch": 0.8642393797689789,
      "grad_norm": 0.569210839807263,
      "learning_rate": 0.00012652158463432795,
      "loss": 12.2195,
      "step": 15871
    },
    {
      "epoch": 0.8642938337655619,
      "grad_norm": 0.5722602728817883,
      "learning_rate": 0.0001265130820631184,
      "loss": 12.2284,
      "step": 15872
    },
    {
      "epoch": 0.8643482877621449,
      "grad_norm": 0.6533167097267406,
      "learning_rate": 0.00012650457928573862,
      "loss": 12.2311,
      "step": 15873
    },
    {
      "epoch": 0.864402741758728,
      "grad_norm": 0.578072245143734,
      "learning_rate": 0.00012649607630225483,
      "loss": 12.245,
      "step": 15874
    },
    {
      "epoch": 0.864457195755311,
      "grad_norm": 0.5929833512597014,
      "learning_rate": 0.00012648757311273308,
      "loss": 12.24,
      "step": 15875
    },
    {
      "epoch": 0.864511649751894,
      "grad_norm": 0.6360942032083976,
      "learning_rate": 0.00012647906971723953,
      "loss": 12.1463,
      "step": 15876
    },
    {
      "epoch": 0.864566103748477,
      "grad_norm": 0.5783819182317385,
      "learning_rate": 0.00012647056611584032,
      "loss": 12.1567,
      "step": 15877
    },
    {
      "epoch": 0.86462055774506,
      "grad_norm": 0.6337107565036638,
      "learning_rate": 0.0001264620623086015,
      "loss": 12.234,
      "step": 15878
    },
    {
      "epoch": 0.864675011741643,
      "grad_norm": 0.5829480172215896,
      "learning_rate": 0.00012645355829558926,
      "loss": 12.1982,
      "step": 15879
    },
    {
      "epoch": 0.8647294657382261,
      "grad_norm": 0.5889035316620181,
      "learning_rate": 0.00012644505407686973,
      "loss": 12.2011,
      "step": 15880
    },
    {
      "epoch": 0.8647839197348091,
      "grad_norm": 0.5656346434970699,
      "learning_rate": 0.000126436549652509,
      "loss": 12.2205,
      "step": 15881
    },
    {
      "epoch": 0.864838373731392,
      "grad_norm": 0.6134097132972022,
      "learning_rate": 0.00012642804502257328,
      "loss": 12.0625,
      "step": 15882
    },
    {
      "epoch": 0.864892827727975,
      "grad_norm": 0.555717845222476,
      "learning_rate": 0.00012641954018712863,
      "loss": 12.1802,
      "step": 15883
    },
    {
      "epoch": 0.864947281724558,
      "grad_norm": 0.6775557458447009,
      "learning_rate": 0.00012641103514624116,
      "loss": 12.2947,
      "step": 15884
    },
    {
      "epoch": 0.865001735721141,
      "grad_norm": 0.6434186072760585,
      "learning_rate": 0.0001264025298999771,
      "loss": 12.3112,
      "step": 15885
    },
    {
      "epoch": 0.8650561897177241,
      "grad_norm": 0.5606701802164176,
      "learning_rate": 0.0001263940244484025,
      "loss": 12.1202,
      "step": 15886
    },
    {
      "epoch": 0.8651106437143071,
      "grad_norm": 0.5973302884811093,
      "learning_rate": 0.00012638551879158358,
      "loss": 12.264,
      "step": 15887
    },
    {
      "epoch": 0.8651650977108901,
      "grad_norm": 0.5938117907409824,
      "learning_rate": 0.00012637701292958644,
      "loss": 12.1346,
      "step": 15888
    },
    {
      "epoch": 0.8652195517074731,
      "grad_norm": 0.5702681890575382,
      "learning_rate": 0.0001263685068624772,
      "loss": 12.1939,
      "step": 15889
    },
    {
      "epoch": 0.8652740057040561,
      "grad_norm": 0.5664421468399776,
      "learning_rate": 0.000126360000590322,
      "loss": 12.1546,
      "step": 15890
    },
    {
      "epoch": 0.8653284597006392,
      "grad_norm": 0.5531560334769265,
      "learning_rate": 0.00012635149411318705,
      "loss": 12.2315,
      "step": 15891
    },
    {
      "epoch": 0.8653829136972222,
      "grad_norm": 0.5814574553270802,
      "learning_rate": 0.0001263429874311385,
      "loss": 12.0902,
      "step": 15892
    },
    {
      "epoch": 0.8654373676938052,
      "grad_norm": 0.5422698332386978,
      "learning_rate": 0.00012633448054424242,
      "loss": 12.0986,
      "step": 15893
    },
    {
      "epoch": 0.8654918216903882,
      "grad_norm": 0.5539154433284016,
      "learning_rate": 0.000126325973452565,
      "loss": 12.1193,
      "step": 15894
    },
    {
      "epoch": 0.8655462756869712,
      "grad_norm": 0.5322445342023846,
      "learning_rate": 0.0001263174661561724,
      "loss": 12.1351,
      "step": 15895
    },
    {
      "epoch": 0.8656007296835542,
      "grad_norm": 0.6011871815047078,
      "learning_rate": 0.00012630895865513075,
      "loss": 12.2558,
      "step": 15896
    },
    {
      "epoch": 0.8656551836801373,
      "grad_norm": 0.5421870278380945,
      "learning_rate": 0.00012630045094950626,
      "loss": 12.0785,
      "step": 15897
    },
    {
      "epoch": 0.8657096376767203,
      "grad_norm": 0.5153219346986837,
      "learning_rate": 0.00012629194303936508,
      "loss": 12.0322,
      "step": 15898
    },
    {
      "epoch": 0.8657640916733033,
      "grad_norm": 0.5698421826144286,
      "learning_rate": 0.0001262834349247733,
      "loss": 12.0598,
      "step": 15899
    },
    {
      "epoch": 0.8658185456698863,
      "grad_norm": 0.5698797197970296,
      "learning_rate": 0.00012627492660579711,
      "loss": 12.1255,
      "step": 15900
    },
    {
      "epoch": 0.8658729996664692,
      "grad_norm": 0.5740622128057058,
      "learning_rate": 0.0001262664180825027,
      "loss": 12.1623,
      "step": 15901
    },
    {
      "epoch": 0.8659274536630522,
      "grad_norm": 0.6295996044812211,
      "learning_rate": 0.0001262579093549562,
      "loss": 12.4251,
      "step": 15902
    },
    {
      "epoch": 0.8659819076596353,
      "grad_norm": 0.6189536561994822,
      "learning_rate": 0.0001262494004232238,
      "loss": 12.2374,
      "step": 15903
    },
    {
      "epoch": 0.8660363616562183,
      "grad_norm": 0.5553251154063896,
      "learning_rate": 0.0001262408912873717,
      "loss": 12.1977,
      "step": 15904
    },
    {
      "epoch": 0.8660908156528013,
      "grad_norm": 0.5551849578424735,
      "learning_rate": 0.000126232381947466,
      "loss": 12.112,
      "step": 15905
    },
    {
      "epoch": 0.8661452696493843,
      "grad_norm": 0.6238235130029314,
      "learning_rate": 0.0001262238724035729,
      "loss": 12.141,
      "step": 15906
    },
    {
      "epoch": 0.8661997236459673,
      "grad_norm": 0.49946598932732883,
      "learning_rate": 0.00012621536265575856,
      "loss": 12.1029,
      "step": 15907
    },
    {
      "epoch": 0.8662541776425503,
      "grad_norm": 0.645038326550859,
      "learning_rate": 0.00012620685270408916,
      "loss": 12.1514,
      "step": 15908
    },
    {
      "epoch": 0.8663086316391334,
      "grad_norm": 0.584876027405954,
      "learning_rate": 0.0001261983425486309,
      "loss": 12.2783,
      "step": 15909
    },
    {
      "epoch": 0.8663630856357164,
      "grad_norm": 0.5779854503830523,
      "learning_rate": 0.00012618983218944994,
      "loss": 12.2156,
      "step": 15910
    },
    {
      "epoch": 0.8664175396322994,
      "grad_norm": 0.5737317733374087,
      "learning_rate": 0.00012618132162661242,
      "loss": 12.2477,
      "step": 15911
    },
    {
      "epoch": 0.8664719936288824,
      "grad_norm": 0.5514906059361818,
      "learning_rate": 0.00012617281086018458,
      "loss": 12.2409,
      "step": 15912
    },
    {
      "epoch": 0.8665264476254654,
      "grad_norm": 0.6499006968948686,
      "learning_rate": 0.00012616429989023258,
      "loss": 12.3086,
      "step": 15913
    },
    {
      "epoch": 0.8665809016220484,
      "grad_norm": 0.5509850457484902,
      "learning_rate": 0.00012615578871682257,
      "loss": 12.1739,
      "step": 15914
    },
    {
      "epoch": 0.8666353556186315,
      "grad_norm": 0.6157756457540347,
      "learning_rate": 0.00012614727734002075,
      "loss": 11.9785,
      "step": 15915
    },
    {
      "epoch": 0.8666898096152145,
      "grad_norm": 0.5260273113757105,
      "learning_rate": 0.00012613876575989335,
      "loss": 12.1181,
      "step": 15916
    },
    {
      "epoch": 0.8667442636117975,
      "grad_norm": 0.5745111404782917,
      "learning_rate": 0.00012613025397650653,
      "loss": 12.21,
      "step": 15917
    },
    {
      "epoch": 0.8667987176083805,
      "grad_norm": 0.601576928480999,
      "learning_rate": 0.00012612174198992646,
      "loss": 12.3353,
      "step": 15918
    },
    {
      "epoch": 0.8668531716049634,
      "grad_norm": 0.7151657198245608,
      "learning_rate": 0.00012611322980021932,
      "loss": 12.2895,
      "step": 15919
    },
    {
      "epoch": 0.8669076256015464,
      "grad_norm": 0.5150769140549022,
      "learning_rate": 0.00012610471740745135,
      "loss": 12.1851,
      "step": 15920
    },
    {
      "epoch": 0.8669620795981295,
      "grad_norm": 0.6268954466952763,
      "learning_rate": 0.0001260962048116887,
      "loss": 12.1937,
      "step": 15921
    },
    {
      "epoch": 0.8670165335947125,
      "grad_norm": 0.5102438388337576,
      "learning_rate": 0.00012608769201299762,
      "loss": 12.1903,
      "step": 15922
    },
    {
      "epoch": 0.8670709875912955,
      "grad_norm": 0.6492952084312248,
      "learning_rate": 0.00012607917901144423,
      "loss": 12.3463,
      "step": 15923
    },
    {
      "epoch": 0.8671254415878785,
      "grad_norm": 0.5412306855807728,
      "learning_rate": 0.0001260706658070948,
      "loss": 12.2514,
      "step": 15924
    },
    {
      "epoch": 0.8671798955844615,
      "grad_norm": 0.5436226777635539,
      "learning_rate": 0.00012606215240001549,
      "loss": 12.1734,
      "step": 15925
    },
    {
      "epoch": 0.8672343495810446,
      "grad_norm": 0.5435772683271166,
      "learning_rate": 0.00012605363879027252,
      "loss": 12.1831,
      "step": 15926
    },
    {
      "epoch": 0.8672888035776276,
      "grad_norm": 0.5662856637085216,
      "learning_rate": 0.00012604512497793211,
      "loss": 12.2577,
      "step": 15927
    },
    {
      "epoch": 0.8673432575742106,
      "grad_norm": 0.557901457511073,
      "learning_rate": 0.0001260366109630604,
      "loss": 12.1894,
      "step": 15928
    },
    {
      "epoch": 0.8673977115707936,
      "grad_norm": 0.5852888269369317,
      "learning_rate": 0.00012602809674572364,
      "loss": 12.1945,
      "step": 15929
    },
    {
      "epoch": 0.8674521655673766,
      "grad_norm": 0.532930528537664,
      "learning_rate": 0.00012601958232598802,
      "loss": 12.1464,
      "step": 15930
    },
    {
      "epoch": 0.8675066195639596,
      "grad_norm": 0.5427560184485759,
      "learning_rate": 0.00012601106770391982,
      "loss": 12.1841,
      "step": 15931
    },
    {
      "epoch": 0.8675610735605427,
      "grad_norm": 0.593897701522849,
      "learning_rate": 0.0001260025528795852,
      "loss": 12.1475,
      "step": 15932
    },
    {
      "epoch": 0.8676155275571257,
      "grad_norm": 0.644566088488215,
      "learning_rate": 0.0001259940378530503,
      "loss": 12.2218,
      "step": 15933
    },
    {
      "epoch": 0.8676699815537087,
      "grad_norm": 0.6236033276293929,
      "learning_rate": 0.00012598552262438141,
      "loss": 12.2827,
      "step": 15934
    },
    {
      "epoch": 0.8677244355502917,
      "grad_norm": 0.5774315218311655,
      "learning_rate": 0.00012597700719364476,
      "loss": 12.2084,
      "step": 15935
    },
    {
      "epoch": 0.8677788895468747,
      "grad_norm": 0.5348287784866989,
      "learning_rate": 0.00012596849156090655,
      "loss": 12.1903,
      "step": 15936
    },
    {
      "epoch": 0.8678333435434576,
      "grad_norm": 0.5716130435917542,
      "learning_rate": 0.00012595997572623302,
      "loss": 12.1704,
      "step": 15937
    },
    {
      "epoch": 0.8678877975400408,
      "grad_norm": 0.8535427189945483,
      "learning_rate": 0.00012595145968969033,
      "loss": 12.1884,
      "step": 15938
    },
    {
      "epoch": 0.8679422515366237,
      "grad_norm": 0.5641547745325012,
      "learning_rate": 0.00012594294345134473,
      "loss": 12.2976,
      "step": 15939
    },
    {
      "epoch": 0.8679967055332067,
      "grad_norm": 0.5748166546920048,
      "learning_rate": 0.00012593442701126247,
      "loss": 12.0867,
      "step": 15940
    },
    {
      "epoch": 0.8680511595297897,
      "grad_norm": 0.5416784728172392,
      "learning_rate": 0.00012592591036950974,
      "loss": 12.2147,
      "step": 15941
    },
    {
      "epoch": 0.8681056135263727,
      "grad_norm": 0.5389782673423665,
      "learning_rate": 0.00012591739352615282,
      "loss": 12.2652,
      "step": 15942
    },
    {
      "epoch": 0.8681600675229557,
      "grad_norm": 0.5646252422136855,
      "learning_rate": 0.0001259088764812579,
      "loss": 12.2263,
      "step": 15943
    },
    {
      "epoch": 0.8682145215195388,
      "grad_norm": 0.5201360681378036,
      "learning_rate": 0.00012590035923489115,
      "loss": 12.0908,
      "step": 15944
    },
    {
      "epoch": 0.8682689755161218,
      "grad_norm": 0.5889451038920502,
      "learning_rate": 0.00012589184178711887,
      "loss": 12.0581,
      "step": 15945
    },
    {
      "epoch": 0.8683234295127048,
      "grad_norm": 0.5434486158643616,
      "learning_rate": 0.00012588332413800734,
      "loss": 12.3294,
      "step": 15946
    },
    {
      "epoch": 0.8683778835092878,
      "grad_norm": 0.5894799008113736,
      "learning_rate": 0.0001258748062876227,
      "loss": 12.3475,
      "step": 15947
    },
    {
      "epoch": 0.8684323375058708,
      "grad_norm": 0.5559143802835625,
      "learning_rate": 0.0001258662882360313,
      "loss": 12.1807,
      "step": 15948
    },
    {
      "epoch": 0.8684867915024538,
      "grad_norm": 0.5751187142046755,
      "learning_rate": 0.00012585776998329923,
      "loss": 12.2895,
      "step": 15949
    },
    {
      "epoch": 0.8685412454990369,
      "grad_norm": 0.583166156976266,
      "learning_rate": 0.0001258492515294928,
      "loss": 12.374,
      "step": 15950
    },
    {
      "epoch": 0.8685956994956199,
      "grad_norm": 0.5732070056996998,
      "learning_rate": 0.0001258407328746783,
      "loss": 12.1528,
      "step": 15951
    },
    {
      "epoch": 0.8686501534922029,
      "grad_norm": 0.5332503377591635,
      "learning_rate": 0.0001258322140189219,
      "loss": 12.0977,
      "step": 15952
    },
    {
      "epoch": 0.8687046074887859,
      "grad_norm": 0.5375694322894206,
      "learning_rate": 0.0001258236949622899,
      "loss": 12.1752,
      "step": 15953
    },
    {
      "epoch": 0.8687590614853689,
      "grad_norm": 0.5557590566613545,
      "learning_rate": 0.0001258151757048485,
      "loss": 12.177,
      "step": 15954
    },
    {
      "epoch": 0.868813515481952,
      "grad_norm": 0.5326035517730949,
      "learning_rate": 0.00012580665624666395,
      "loss": 12.1465,
      "step": 15955
    },
    {
      "epoch": 0.868867969478535,
      "grad_norm": 0.5769864726318715,
      "learning_rate": 0.0001257981365878025,
      "loss": 12.2239,
      "step": 15956
    },
    {
      "epoch": 0.8689224234751179,
      "grad_norm": 0.5476567244141276,
      "learning_rate": 0.00012578961672833044,
      "loss": 12.2517,
      "step": 15957
    },
    {
      "epoch": 0.8689768774717009,
      "grad_norm": 0.591401878428727,
      "learning_rate": 0.00012578109666831403,
      "loss": 12.1768,
      "step": 15958
    },
    {
      "epoch": 0.8690313314682839,
      "grad_norm": 0.5794924956390661,
      "learning_rate": 0.00012577257640781944,
      "loss": 12.2739,
      "step": 15959
    },
    {
      "epoch": 0.8690857854648669,
      "grad_norm": 0.5852044421285363,
      "learning_rate": 0.00012576405594691298,
      "loss": 12.1281,
      "step": 15960
    },
    {
      "epoch": 0.86914023946145,
      "grad_norm": 0.5417282944487868,
      "learning_rate": 0.00012575553528566092,
      "loss": 12.0221,
      "step": 15961
    },
    {
      "epoch": 0.869194693458033,
      "grad_norm": 0.5253617600990087,
      "learning_rate": 0.00012574701442412945,
      "loss": 12.2346,
      "step": 15962
    },
    {
      "epoch": 0.869249147454616,
      "grad_norm": 0.587901673765946,
      "learning_rate": 0.00012573849336238496,
      "loss": 12.1188,
      "step": 15963
    },
    {
      "epoch": 0.869303601451199,
      "grad_norm": 0.5501555510980765,
      "learning_rate": 0.00012572997210049354,
      "loss": 12.2694,
      "step": 15964
    },
    {
      "epoch": 0.869358055447782,
      "grad_norm": 0.5475981794367448,
      "learning_rate": 0.00012572145063852161,
      "loss": 12.2831,
      "step": 15965
    },
    {
      "epoch": 0.869412509444365,
      "grad_norm": 0.5385544190151177,
      "learning_rate": 0.00012571292897653534,
      "loss": 12.2199,
      "step": 15966
    },
    {
      "epoch": 0.8694669634409481,
      "grad_norm": 0.5752778166783943,
      "learning_rate": 0.000125704407114601,
      "loss": 12.2106,
      "step": 15967
    },
    {
      "epoch": 0.8695214174375311,
      "grad_norm": 0.5854288021900113,
      "learning_rate": 0.0001256958850527849,
      "loss": 12.2438,
      "step": 15968
    },
    {
      "epoch": 0.8695758714341141,
      "grad_norm": 0.5484390545530441,
      "learning_rate": 0.00012568736279115332,
      "loss": 12.211,
      "step": 15969
    },
    {
      "epoch": 0.8696303254306971,
      "grad_norm": 0.5633289860451813,
      "learning_rate": 0.00012567884032977245,
      "loss": 12.2505,
      "step": 15970
    },
    {
      "epoch": 0.8696847794272801,
      "grad_norm": 0.5497065293508597,
      "learning_rate": 0.00012567031766870865,
      "loss": 12.1282,
      "step": 15971
    },
    {
      "epoch": 0.8697392334238631,
      "grad_norm": 0.5536133855620301,
      "learning_rate": 0.00012566179480802812,
      "loss": 12.1886,
      "step": 15972
    },
    {
      "epoch": 0.8697936874204462,
      "grad_norm": 0.525991757190078,
      "learning_rate": 0.0001256532717477972,
      "loss": 12.1135,
      "step": 15973
    },
    {
      "epoch": 0.8698481414170292,
      "grad_norm": 0.5349464762038328,
      "learning_rate": 0.00012564474848808212,
      "loss": 12.248,
      "step": 15974
    },
    {
      "epoch": 0.8699025954136121,
      "grad_norm": 0.5385765964720809,
      "learning_rate": 0.00012563622502894918,
      "loss": 12.0118,
      "step": 15975
    },
    {
      "epoch": 0.8699570494101951,
      "grad_norm": 0.559178233160023,
      "learning_rate": 0.00012562770137046464,
      "loss": 12.1902,
      "step": 15976
    },
    {
      "epoch": 0.8700115034067781,
      "grad_norm": 0.5540507729228215,
      "learning_rate": 0.00012561917751269483,
      "loss": 12.1979,
      "step": 15977
    },
    {
      "epoch": 0.8700659574033611,
      "grad_norm": 0.5140207796261256,
      "learning_rate": 0.00012561065345570598,
      "loss": 12.0578,
      "step": 15978
    },
    {
      "epoch": 0.8701204113999442,
      "grad_norm": 0.5288952409117764,
      "learning_rate": 0.00012560212919956437,
      "loss": 12.2029,
      "step": 15979
    },
    {
      "epoch": 0.8701748653965272,
      "grad_norm": 0.5725414765013682,
      "learning_rate": 0.00012559360474433636,
      "loss": 12.1941,
      "step": 15980
    },
    {
      "epoch": 0.8702293193931102,
      "grad_norm": 0.5931801761743127,
      "learning_rate": 0.00012558508009008812,
      "loss": 12.2551,
      "step": 15981
    },
    {
      "epoch": 0.8702837733896932,
      "grad_norm": 0.5256172262315336,
      "learning_rate": 0.00012557655523688608,
      "loss": 12.123,
      "step": 15982
    },
    {
      "epoch": 0.8703382273862762,
      "grad_norm": 0.5238440893579909,
      "learning_rate": 0.00012556803018479642,
      "loss": 12.2338,
      "step": 15983
    },
    {
      "epoch": 0.8703926813828592,
      "grad_norm": 0.5536812282601996,
      "learning_rate": 0.00012555950493388547,
      "loss": 12.2798,
      "step": 15984
    },
    {
      "epoch": 0.8704471353794423,
      "grad_norm": 0.5504843169868991,
      "learning_rate": 0.00012555097948421952,
      "loss": 12.1371,
      "step": 15985
    },
    {
      "epoch": 0.8705015893760253,
      "grad_norm": 0.5987785474860935,
      "learning_rate": 0.00012554245383586488,
      "loss": 12.3446,
      "step": 15986
    },
    {
      "epoch": 0.8705560433726083,
      "grad_norm": 0.5325786338129379,
      "learning_rate": 0.00012553392798888785,
      "loss": 11.9497,
      "step": 15987
    },
    {
      "epoch": 0.8706104973691913,
      "grad_norm": 0.5434399515345769,
      "learning_rate": 0.00012552540194335466,
      "loss": 12.2137,
      "step": 15988
    },
    {
      "epoch": 0.8706649513657743,
      "grad_norm": 0.5620219933284507,
      "learning_rate": 0.00012551687569933173,
      "loss": 12.2886,
      "step": 15989
    },
    {
      "epoch": 0.8707194053623574,
      "grad_norm": 0.5546283490990715,
      "learning_rate": 0.00012550834925688525,
      "loss": 12.1043,
      "step": 15990
    },
    {
      "epoch": 0.8707738593589404,
      "grad_norm": 0.47488405181773846,
      "learning_rate": 0.0001254998226160816,
      "loss": 12.0254,
      "step": 15991
    },
    {
      "epoch": 0.8708283133555234,
      "grad_norm": 0.5528150439905408,
      "learning_rate": 0.000125491295776987,
      "loss": 12.0765,
      "step": 15992
    },
    {
      "epoch": 0.8708827673521063,
      "grad_norm": 0.5498831187915221,
      "learning_rate": 0.0001254827687396679,
      "loss": 12.0998,
      "step": 15993
    },
    {
      "epoch": 0.8709372213486893,
      "grad_norm": 0.543385726227821,
      "learning_rate": 0.00012547424150419044,
      "loss": 12.2286,
      "step": 15994
    },
    {
      "epoch": 0.8709916753452723,
      "grad_norm": 0.5621492505997553,
      "learning_rate": 0.00012546571407062103,
      "loss": 12.2143,
      "step": 15995
    },
    {
      "epoch": 0.8710461293418554,
      "grad_norm": 0.5433169361833831,
      "learning_rate": 0.00012545718643902594,
      "loss": 12.2198,
      "step": 15996
    },
    {
      "epoch": 0.8711005833384384,
      "grad_norm": 0.5223570553806087,
      "learning_rate": 0.0001254486586094715,
      "loss": 12.0896,
      "step": 15997
    },
    {
      "epoch": 0.8711550373350214,
      "grad_norm": 0.5249600347028466,
      "learning_rate": 0.00012544013058202405,
      "loss": 12.1902,
      "step": 15998
    },
    {
      "epoch": 0.8712094913316044,
      "grad_norm": 0.5305491138861335,
      "learning_rate": 0.00012543160235674986,
      "loss": 12.0778,
      "step": 15999
    },
    {
      "epoch": 0.8712639453281874,
      "grad_norm": 0.5129323620281291,
      "learning_rate": 0.00012542307393371526,
      "loss": 12.1419,
      "step": 16000
    },
    {
      "epoch": 0.8713183993247704,
      "grad_norm": 0.6005913292330536,
      "learning_rate": 0.0001254145453129866,
      "loss": 12.0793,
      "step": 16001
    },
    {
      "epoch": 0.8713728533213535,
      "grad_norm": 0.5728093693687083,
      "learning_rate": 0.00012540601649463015,
      "loss": 12.123,
      "step": 16002
    },
    {
      "epoch": 0.8714273073179365,
      "grad_norm": 0.5636253436070369,
      "learning_rate": 0.00012539748747871228,
      "loss": 12.1248,
      "step": 16003
    },
    {
      "epoch": 0.8714817613145195,
      "grad_norm": 0.5975695198443635,
      "learning_rate": 0.0001253889582652993,
      "loss": 12.1997,
      "step": 16004
    },
    {
      "epoch": 0.8715362153111025,
      "grad_norm": 0.5694835079399884,
      "learning_rate": 0.00012538042885445745,
      "loss": 12.2186,
      "step": 16005
    },
    {
      "epoch": 0.8715906693076855,
      "grad_norm": 0.533297799927533,
      "learning_rate": 0.00012537189924625316,
      "loss": 12.2206,
      "step": 16006
    },
    {
      "epoch": 0.8716451233042685,
      "grad_norm": 0.5422775699384693,
      "learning_rate": 0.00012536336944075276,
      "loss": 12.1786,
      "step": 16007
    },
    {
      "epoch": 0.8716995773008516,
      "grad_norm": 0.5741067704232433,
      "learning_rate": 0.00012535483943802253,
      "loss": 12.2374,
      "step": 16008
    },
    {
      "epoch": 0.8717540312974346,
      "grad_norm": 0.539661125867655,
      "learning_rate": 0.0001253463092381288,
      "loss": 12.1391,
      "step": 16009
    },
    {
      "epoch": 0.8718084852940176,
      "grad_norm": 0.5816111108862853,
      "learning_rate": 0.00012533777884113793,
      "loss": 12.1623,
      "step": 16010
    },
    {
      "epoch": 0.8718629392906005,
      "grad_norm": 0.5311156149067617,
      "learning_rate": 0.00012532924824711623,
      "loss": 12.2057,
      "step": 16011
    },
    {
      "epoch": 0.8719173932871835,
      "grad_norm": 0.5295481418916593,
      "learning_rate": 0.00012532071745613007,
      "loss": 12.1391,
      "step": 16012
    },
    {
      "epoch": 0.8719718472837665,
      "grad_norm": 0.6036077984497444,
      "learning_rate": 0.00012531218646824577,
      "loss": 12.1833,
      "step": 16013
    },
    {
      "epoch": 0.8720263012803496,
      "grad_norm": 0.5358288325689889,
      "learning_rate": 0.00012530365528352964,
      "loss": 12.1514,
      "step": 16014
    },
    {
      "epoch": 0.8720807552769326,
      "grad_norm": 0.6269518042813348,
      "learning_rate": 0.00012529512390204808,
      "loss": 12.2951,
      "step": 16015
    },
    {
      "epoch": 0.8721352092735156,
      "grad_norm": 0.6034229536886009,
      "learning_rate": 0.00012528659232386732,
      "loss": 12.2408,
      "step": 16016
    },
    {
      "epoch": 0.8721896632700986,
      "grad_norm": 0.5514941628900512,
      "learning_rate": 0.00012527806054905382,
      "loss": 12.0462,
      "step": 16017
    },
    {
      "epoch": 0.8722441172666816,
      "grad_norm": 0.6823523013440215,
      "learning_rate": 0.0001252695285776739,
      "loss": 12.3112,
      "step": 16018
    },
    {
      "epoch": 0.8722985712632646,
      "grad_norm": 0.6098942419908867,
      "learning_rate": 0.0001252609964097939,
      "loss": 12.175,
      "step": 16019
    },
    {
      "epoch": 0.8723530252598477,
      "grad_norm": 0.5719599376070134,
      "learning_rate": 0.00012525246404548011,
      "loss": 12.1688,
      "step": 16020
    },
    {
      "epoch": 0.8724074792564307,
      "grad_norm": 0.5623532728458146,
      "learning_rate": 0.00012524393148479897,
      "loss": 12.0794,
      "step": 16021
    },
    {
      "epoch": 0.8724619332530137,
      "grad_norm": 0.5240231925996959,
      "learning_rate": 0.0001252353987278167,
      "loss": 12.1055,
      "step": 16022
    },
    {
      "epoch": 0.8725163872495967,
      "grad_norm": 0.6506632285409836,
      "learning_rate": 0.00012522686577459984,
      "loss": 12.2739,
      "step": 16023
    },
    {
      "epoch": 0.8725708412461797,
      "grad_norm": 0.5723476929353505,
      "learning_rate": 0.00012521833262521458,
      "loss": 12.1607,
      "step": 16024
    },
    {
      "epoch": 0.8726252952427628,
      "grad_norm": 0.6015507472500625,
      "learning_rate": 0.00012520979927972738,
      "loss": 12.2483,
      "step": 16025
    },
    {
      "epoch": 0.8726797492393458,
      "grad_norm": 0.5956510012054886,
      "learning_rate": 0.00012520126573820452,
      "loss": 12.2262,
      "step": 16026
    },
    {
      "epoch": 0.8727342032359288,
      "grad_norm": 0.5526715767225209,
      "learning_rate": 0.00012519273200071237,
      "loss": 12.1711,
      "step": 16027
    },
    {
      "epoch": 0.8727886572325118,
      "grad_norm": 0.5974678518446006,
      "learning_rate": 0.00012518419806731735,
      "loss": 12.2869,
      "step": 16028
    },
    {
      "epoch": 0.8728431112290947,
      "grad_norm": 0.5520163993498713,
      "learning_rate": 0.0001251756639380858,
      "loss": 12.0081,
      "step": 16029
    },
    {
      "epoch": 0.8728975652256777,
      "grad_norm": 0.5639711796095425,
      "learning_rate": 0.00012516712961308402,
      "loss": 12.2842,
      "step": 16030
    },
    {
      "epoch": 0.8729520192222608,
      "grad_norm": 0.5580007563756508,
      "learning_rate": 0.00012515859509237845,
      "loss": 12.3199,
      "step": 16031
    },
    {
      "epoch": 0.8730064732188438,
      "grad_norm": 0.6335581879602564,
      "learning_rate": 0.0001251500603760354,
      "loss": 12.2911,
      "step": 16032
    },
    {
      "epoch": 0.8730609272154268,
      "grad_norm": 0.6078406349282465,
      "learning_rate": 0.00012514152546412127,
      "loss": 12.1284,
      "step": 16033
    },
    {
      "epoch": 0.8731153812120098,
      "grad_norm": 0.5154839706557784,
      "learning_rate": 0.00012513299035670246,
      "loss": 12.1873,
      "step": 16034
    },
    {
      "epoch": 0.8731698352085928,
      "grad_norm": 0.5187457942704194,
      "learning_rate": 0.00012512445505384525,
      "loss": 12.1551,
      "step": 16035
    },
    {
      "epoch": 0.8732242892051758,
      "grad_norm": 0.6452981420258556,
      "learning_rate": 0.00012511591955561608,
      "loss": 12.188,
      "step": 16036
    },
    {
      "epoch": 0.8732787432017589,
      "grad_norm": 0.6740558251184862,
      "learning_rate": 0.00012510738386208132,
      "loss": 12.3964,
      "step": 16037
    },
    {
      "epoch": 0.8733331971983419,
      "grad_norm": 0.5663356963926833,
      "learning_rate": 0.00012509884797330732,
      "loss": 12.0535,
      "step": 16038
    },
    {
      "epoch": 0.8733876511949249,
      "grad_norm": 0.5878177856957262,
      "learning_rate": 0.00012509031188936046,
      "loss": 12.0532,
      "step": 16039
    },
    {
      "epoch": 0.8734421051915079,
      "grad_norm": 0.5952682212176018,
      "learning_rate": 0.00012508177561030716,
      "loss": 12.1387,
      "step": 16040
    },
    {
      "epoch": 0.8734965591880909,
      "grad_norm": 0.5413794683777703,
      "learning_rate": 0.00012507323913621375,
      "loss": 12.1888,
      "step": 16041
    },
    {
      "epoch": 0.8735510131846739,
      "grad_norm": 0.5450852611443711,
      "learning_rate": 0.0001250647024671466,
      "loss": 12.1893,
      "step": 16042
    },
    {
      "epoch": 0.873605467181257,
      "grad_norm": 0.574623152064138,
      "learning_rate": 0.00012505616560317217,
      "loss": 12.2776,
      "step": 16043
    },
    {
      "epoch": 0.87365992117784,
      "grad_norm": 0.4820315833893964,
      "learning_rate": 0.00012504762854435676,
      "loss": 12.0415,
      "step": 16044
    },
    {
      "epoch": 0.873714375174423,
      "grad_norm": 0.5191268713251201,
      "learning_rate": 0.0001250390912907668,
      "loss": 12.2193,
      "step": 16045
    },
    {
      "epoch": 0.873768829171006,
      "grad_norm": 0.5708885380831076,
      "learning_rate": 0.00012503055384246867,
      "loss": 12.1649,
      "step": 16046
    },
    {
      "epoch": 0.873823283167589,
      "grad_norm": 0.626996981524449,
      "learning_rate": 0.00012502201619952875,
      "loss": 12.1362,
      "step": 16047
    },
    {
      "epoch": 0.8738777371641719,
      "grad_norm": 0.582192853950474,
      "learning_rate": 0.00012501347836201343,
      "loss": 12.1748,
      "step": 16048
    },
    {
      "epoch": 0.873932191160755,
      "grad_norm": 0.5500699566784814,
      "learning_rate": 0.0001250049403299891,
      "loss": 12.2157,
      "step": 16049
    },
    {
      "epoch": 0.873986645157338,
      "grad_norm": 0.5405585547116991,
      "learning_rate": 0.00012499640210352219,
      "loss": 12.212,
      "step": 16050
    },
    {
      "epoch": 0.874041099153921,
      "grad_norm": 0.5180377431941692,
      "learning_rate": 0.00012498786368267905,
      "loss": 12.1162,
      "step": 16051
    },
    {
      "epoch": 0.874095553150504,
      "grad_norm": 0.5348153153227052,
      "learning_rate": 0.0001249793250675261,
      "loss": 12.1576,
      "step": 16052
    },
    {
      "epoch": 0.874150007147087,
      "grad_norm": 0.604938726107424,
      "learning_rate": 0.00012497078625812975,
      "loss": 12.0198,
      "step": 16053
    },
    {
      "epoch": 0.87420446114367,
      "grad_norm": 0.5422674776228027,
      "learning_rate": 0.00012496224725455632,
      "loss": 12.1241,
      "step": 16054
    },
    {
      "epoch": 0.8742589151402531,
      "grad_norm": 0.5538596352364217,
      "learning_rate": 0.0001249537080568723,
      "loss": 12.0372,
      "step": 16055
    },
    {
      "epoch": 0.8743133691368361,
      "grad_norm": 0.5431686420588868,
      "learning_rate": 0.00012494516866514406,
      "loss": 12.187,
      "step": 16056
    },
    {
      "epoch": 0.8743678231334191,
      "grad_norm": 0.5386989318844304,
      "learning_rate": 0.000124936629079438,
      "loss": 12.2008,
      "step": 16057
    },
    {
      "epoch": 0.8744222771300021,
      "grad_norm": 0.544736835791952,
      "learning_rate": 0.00012492808929982056,
      "loss": 12.3372,
      "step": 16058
    },
    {
      "epoch": 0.8744767311265851,
      "grad_norm": 0.6666965240807943,
      "learning_rate": 0.0001249195493263581,
      "loss": 12.1557,
      "step": 16059
    },
    {
      "epoch": 0.8745311851231682,
      "grad_norm": 0.5583615651695085,
      "learning_rate": 0.00012491100915911702,
      "loss": 12.3522,
      "step": 16060
    },
    {
      "epoch": 0.8745856391197512,
      "grad_norm": 0.6220010193284105,
      "learning_rate": 0.00012490246879816376,
      "loss": 12.2049,
      "step": 16061
    },
    {
      "epoch": 0.8746400931163342,
      "grad_norm": 0.5007739341726704,
      "learning_rate": 0.00012489392824356475,
      "loss": 12.1438,
      "step": 16062
    },
    {
      "epoch": 0.8746945471129172,
      "grad_norm": 0.51788794478555,
      "learning_rate": 0.0001248853874953864,
      "loss": 12.1486,
      "step": 16063
    },
    {
      "epoch": 0.8747490011095002,
      "grad_norm": 0.5350124014753489,
      "learning_rate": 0.00012487684655369507,
      "loss": 12.1837,
      "step": 16064
    },
    {
      "epoch": 0.8748034551060831,
      "grad_norm": 0.5369649604490867,
      "learning_rate": 0.00012486830541855718,
      "loss": 12.2681,
      "step": 16065
    },
    {
      "epoch": 0.8748579091026663,
      "grad_norm": 0.536904150078264,
      "learning_rate": 0.0001248597640900392,
      "loss": 12.1093,
      "step": 16066
    },
    {
      "epoch": 0.8749123630992492,
      "grad_norm": 0.5389369443292675,
      "learning_rate": 0.00012485122256820756,
      "loss": 11.9681,
      "step": 16067
    },
    {
      "epoch": 0.8749668170958322,
      "grad_norm": 0.5705666993112773,
      "learning_rate": 0.00012484268085312863,
      "loss": 12.1541,
      "step": 16068
    },
    {
      "epoch": 0.8750212710924152,
      "grad_norm": 0.5060632652893851,
      "learning_rate": 0.00012483413894486884,
      "loss": 12.2575,
      "step": 16069
    },
    {
      "epoch": 0.8750757250889982,
      "grad_norm": 0.5912434449471482,
      "learning_rate": 0.00012482559684349461,
      "loss": 12.1034,
      "step": 16070
    },
    {
      "epoch": 0.8751301790855812,
      "grad_norm": 0.5795383894091256,
      "learning_rate": 0.00012481705454907237,
      "loss": 12.038,
      "step": 16071
    },
    {
      "epoch": 0.8751846330821643,
      "grad_norm": 0.5638125410718151,
      "learning_rate": 0.00012480851206166858,
      "loss": 12.1461,
      "step": 16072
    },
    {
      "epoch": 0.8752390870787473,
      "grad_norm": 0.5825779199294095,
      "learning_rate": 0.00012479996938134964,
      "loss": 12.0403,
      "step": 16073
    },
    {
      "epoch": 0.8752935410753303,
      "grad_norm": 0.5298141174132485,
      "learning_rate": 0.00012479142650818195,
      "loss": 12.0994,
      "step": 16074
    },
    {
      "epoch": 0.8753479950719133,
      "grad_norm": 0.6573123207917692,
      "learning_rate": 0.00012478288344223198,
      "loss": 12.3921,
      "step": 16075
    },
    {
      "epoch": 0.8754024490684963,
      "grad_norm": 0.6068568486815071,
      "learning_rate": 0.00012477434018356615,
      "loss": 12.1627,
      "step": 16076
    },
    {
      "epoch": 0.8754569030650793,
      "grad_norm": 0.609707243377987,
      "learning_rate": 0.0001247657967322509,
      "loss": 12.0332,
      "step": 16077
    },
    {
      "epoch": 0.8755113570616624,
      "grad_norm": 0.5680797285820538,
      "learning_rate": 0.00012475725308835268,
      "loss": 12.1601,
      "step": 16078
    },
    {
      "epoch": 0.8755658110582454,
      "grad_norm": 0.5526014221770983,
      "learning_rate": 0.0001247487092519379,
      "loss": 12.1961,
      "step": 16079
    },
    {
      "epoch": 0.8756202650548284,
      "grad_norm": 0.6115334683613696,
      "learning_rate": 0.00012474016522307302,
      "loss": 12.2713,
      "step": 16080
    },
    {
      "epoch": 0.8756747190514114,
      "grad_norm": 0.6858296789782814,
      "learning_rate": 0.00012473162100182442,
      "loss": 12.1809,
      "step": 16081
    },
    {
      "epoch": 0.8757291730479944,
      "grad_norm": 0.5441928771376101,
      "learning_rate": 0.0001247230765882586,
      "loss": 12.1297,
      "step": 16082
    },
    {
      "epoch": 0.8757836270445774,
      "grad_norm": 0.5270975788304015,
      "learning_rate": 0.00012471453198244204,
      "loss": 12.0124,
      "step": 16083
    },
    {
      "epoch": 0.8758380810411605,
      "grad_norm": 0.5791745606955657,
      "learning_rate": 0.0001247059871844411,
      "loss": 12.0933,
      "step": 16084
    },
    {
      "epoch": 0.8758925350377434,
      "grad_norm": 0.7093793546578929,
      "learning_rate": 0.00012469744219432226,
      "loss": 12.3378,
      "step": 16085
    },
    {
      "epoch": 0.8759469890343264,
      "grad_norm": 0.5754791802733077,
      "learning_rate": 0.00012468889701215197,
      "loss": 12.1667,
      "step": 16086
    },
    {
      "epoch": 0.8760014430309094,
      "grad_norm": 0.5710204967149822,
      "learning_rate": 0.00012468035163799667,
      "loss": 12.143,
      "step": 16087
    },
    {
      "epoch": 0.8760558970274924,
      "grad_norm": 0.5749549445115179,
      "learning_rate": 0.0001246718060719228,
      "loss": 12.1046,
      "step": 16088
    },
    {
      "epoch": 0.8761103510240754,
      "grad_norm": 0.5773812106423682,
      "learning_rate": 0.00012466326031399688,
      "loss": 12.2384,
      "step": 16089
    },
    {
      "epoch": 0.8761648050206585,
      "grad_norm": 0.5467160672068934,
      "learning_rate": 0.00012465471436428526,
      "loss": 12.1804,
      "step": 16090
    },
    {
      "epoch": 0.8762192590172415,
      "grad_norm": 0.5905830453636292,
      "learning_rate": 0.00012464616822285447,
      "loss": 12.0921,
      "step": 16091
    },
    {
      "epoch": 0.8762737130138245,
      "grad_norm": 0.48989163227231247,
      "learning_rate": 0.00012463762188977094,
      "loss": 11.9986,
      "step": 16092
    },
    {
      "epoch": 0.8763281670104075,
      "grad_norm": 0.6316447646107993,
      "learning_rate": 0.0001246290753651011,
      "loss": 12.1741,
      "step": 16093
    },
    {
      "epoch": 0.8763826210069905,
      "grad_norm": 0.5579716652402554,
      "learning_rate": 0.0001246205286489115,
      "loss": 12.1777,
      "step": 16094
    },
    {
      "epoch": 0.8764370750035736,
      "grad_norm": 0.5929747326262457,
      "learning_rate": 0.00012461198174126852,
      "loss": 12.3101,
      "step": 16095
    },
    {
      "epoch": 0.8764915290001566,
      "grad_norm": 0.5366363838568602,
      "learning_rate": 0.00012460343464223864,
      "loss": 12.1976,
      "step": 16096
    },
    {
      "epoch": 0.8765459829967396,
      "grad_norm": 0.5221075684624025,
      "learning_rate": 0.00012459488735188832,
      "loss": 11.9716,
      "step": 16097
    },
    {
      "epoch": 0.8766004369933226,
      "grad_norm": 0.5476321426506422,
      "learning_rate": 0.000124586339870284,
      "loss": 12.1252,
      "step": 16098
    },
    {
      "epoch": 0.8766548909899056,
      "grad_norm": 0.5792030541270509,
      "learning_rate": 0.0001245777921974922,
      "loss": 12.2472,
      "step": 16099
    },
    {
      "epoch": 0.8767093449864886,
      "grad_norm": 0.6985294307629022,
      "learning_rate": 0.0001245692443335794,
      "loss": 12.1955,
      "step": 16100
    },
    {
      "epoch": 0.8767637989830717,
      "grad_norm": 0.5577872331396814,
      "learning_rate": 0.000124560696278612,
      "loss": 12.0629,
      "step": 16101
    },
    {
      "epoch": 0.8768182529796547,
      "grad_norm": 0.5853442110348637,
      "learning_rate": 0.00012455214803265652,
      "loss": 12.1465,
      "step": 16102
    },
    {
      "epoch": 0.8768727069762376,
      "grad_norm": 0.5525518738121004,
      "learning_rate": 0.0001245435995957794,
      "loss": 12.1607,
      "step": 16103
    },
    {
      "epoch": 0.8769271609728206,
      "grad_norm": 0.5284200111715768,
      "learning_rate": 0.0001245350509680472,
      "loss": 12.204,
      "step": 16104
    },
    {
      "epoch": 0.8769816149694036,
      "grad_norm": 0.6105951133886501,
      "learning_rate": 0.00012452650214952624,
      "loss": 12.2329,
      "step": 16105
    },
    {
      "epoch": 0.8770360689659866,
      "grad_norm": 0.538846310148232,
      "learning_rate": 0.00012451795314028313,
      "loss": 11.8713,
      "step": 16106
    },
    {
      "epoch": 0.8770905229625697,
      "grad_norm": 0.5356062524521141,
      "learning_rate": 0.0001245094039403843,
      "loss": 12.1835,
      "step": 16107
    },
    {
      "epoch": 0.8771449769591527,
      "grad_norm": 0.5572700303346961,
      "learning_rate": 0.00012450085454989625,
      "loss": 12.0499,
      "step": 16108
    },
    {
      "epoch": 0.8771994309557357,
      "grad_norm": 0.6277722706686351,
      "learning_rate": 0.00012449230496888543,
      "loss": 12.2874,
      "step": 16109
    },
    {
      "epoch": 0.8772538849523187,
      "grad_norm": 0.5989562610312907,
      "learning_rate": 0.00012448375519741835,
      "loss": 12.2477,
      "step": 16110
    },
    {
      "epoch": 0.8773083389489017,
      "grad_norm": 0.591640548205753,
      "learning_rate": 0.00012447520523556146,
      "loss": 12.2729,
      "step": 16111
    },
    {
      "epoch": 0.8773627929454847,
      "grad_norm": 0.5043005990234394,
      "learning_rate": 0.00012446665508338128,
      "loss": 12.0538,
      "step": 16112
    },
    {
      "epoch": 0.8774172469420678,
      "grad_norm": 0.5474216805855566,
      "learning_rate": 0.0001244581047409443,
      "loss": 12.2178,
      "step": 16113
    },
    {
      "epoch": 0.8774717009386508,
      "grad_norm": 0.5857427999025753,
      "learning_rate": 0.000124449554208317,
      "loss": 12.162,
      "step": 16114
    },
    {
      "epoch": 0.8775261549352338,
      "grad_norm": 0.5805090016549473,
      "learning_rate": 0.00012444100348556585,
      "loss": 12.2238,
      "step": 16115
    },
    {
      "epoch": 0.8775806089318168,
      "grad_norm": 0.6105765736875828,
      "learning_rate": 0.00012443245257275735,
      "loss": 12.2629,
      "step": 16116
    },
    {
      "epoch": 0.8776350629283998,
      "grad_norm": 0.518552692809577,
      "learning_rate": 0.000124423901469958,
      "loss": 12.0606,
      "step": 16117
    },
    {
      "epoch": 0.8776895169249828,
      "grad_norm": 0.5763076452780304,
      "learning_rate": 0.00012441535017723433,
      "loss": 12.1436,
      "step": 16118
    },
    {
      "epoch": 0.8777439709215659,
      "grad_norm": 0.5341231124511039,
      "learning_rate": 0.0001244067986946528,
      "loss": 12.191,
      "step": 16119
    },
    {
      "epoch": 0.8777984249181489,
      "grad_norm": 0.5500435671005613,
      "learning_rate": 0.00012439824702227987,
      "loss": 12.2206,
      "step": 16120
    },
    {
      "epoch": 0.8778528789147318,
      "grad_norm": 0.5525902266037133,
      "learning_rate": 0.0001243896951601821,
      "loss": 12.1845,
      "step": 16121
    },
    {
      "epoch": 0.8779073329113148,
      "grad_norm": 0.6192969566452226,
      "learning_rate": 0.00012438114310842598,
      "loss": 12.2346,
      "step": 16122
    },
    {
      "epoch": 0.8779617869078978,
      "grad_norm": 0.5553910515782029,
      "learning_rate": 0.000124372590867078,
      "loss": 12.2288,
      "step": 16123
    },
    {
      "epoch": 0.8780162409044809,
      "grad_norm": 0.5477482552336438,
      "learning_rate": 0.00012436403843620468,
      "loss": 12.2031,
      "step": 16124
    },
    {
      "epoch": 0.8780706949010639,
      "grad_norm": 0.5097790965963074,
      "learning_rate": 0.00012435548581587246,
      "loss": 12.1158,
      "step": 16125
    },
    {
      "epoch": 0.8781251488976469,
      "grad_norm": 0.6096381870021689,
      "learning_rate": 0.00012434693300614793,
      "loss": 12.2922,
      "step": 16126
    },
    {
      "epoch": 0.8781796028942299,
      "grad_norm": 0.5502228426678689,
      "learning_rate": 0.00012433838000709755,
      "loss": 12.1405,
      "step": 16127
    },
    {
      "epoch": 0.8782340568908129,
      "grad_norm": 0.6006586891159624,
      "learning_rate": 0.0001243298268187879,
      "loss": 12.3248,
      "step": 16128
    },
    {
      "epoch": 0.8782885108873959,
      "grad_norm": 0.5581646276150829,
      "learning_rate": 0.0001243212734412854,
      "loss": 12.1715,
      "step": 16129
    },
    {
      "epoch": 0.878342964883979,
      "grad_norm": 0.5495172790231901,
      "learning_rate": 0.00012431271987465661,
      "loss": 12.2233,
      "step": 16130
    },
    {
      "epoch": 0.878397418880562,
      "grad_norm": 0.5375101576174629,
      "learning_rate": 0.00012430416611896797,
      "loss": 12.1081,
      "step": 16131
    },
    {
      "epoch": 0.878451872877145,
      "grad_norm": 0.5300128739993558,
      "learning_rate": 0.0001242956121742861,
      "loss": 12.2207,
      "step": 16132
    },
    {
      "epoch": 0.878506326873728,
      "grad_norm": 0.5687779995223394,
      "learning_rate": 0.00012428705804067751,
      "loss": 12.189,
      "step": 16133
    },
    {
      "epoch": 0.878560780870311,
      "grad_norm": 0.5836162326318499,
      "learning_rate": 0.00012427850371820868,
      "loss": 12.2623,
      "step": 16134
    },
    {
      "epoch": 0.878615234866894,
      "grad_norm": 0.5308718848591922,
      "learning_rate": 0.0001242699492069461,
      "loss": 12.2993,
      "step": 16135
    },
    {
      "epoch": 0.8786696888634771,
      "grad_norm": 0.49513660224252215,
      "learning_rate": 0.00012426139450695634,
      "loss": 12.0951,
      "step": 16136
    },
    {
      "epoch": 0.8787241428600601,
      "grad_norm": 0.5178325865330834,
      "learning_rate": 0.0001242528396183059,
      "loss": 11.9599,
      "step": 16137
    },
    {
      "epoch": 0.8787785968566431,
      "grad_norm": 0.5969490784962753,
      "learning_rate": 0.00012424428454106128,
      "loss": 12.2505,
      "step": 16138
    },
    {
      "epoch": 0.878833050853226,
      "grad_norm": 0.5901651262878582,
      "learning_rate": 0.0001242357292752891,
      "loss": 12.2431,
      "step": 16139
    },
    {
      "epoch": 0.878887504849809,
      "grad_norm": 0.5301687700694988,
      "learning_rate": 0.00012422717382105583,
      "loss": 12.1567,
      "step": 16140
    },
    {
      "epoch": 0.878941958846392,
      "grad_norm": 0.5653153214152838,
      "learning_rate": 0.00012421861817842796,
      "loss": 12.2456,
      "step": 16141
    },
    {
      "epoch": 0.8789964128429751,
      "grad_norm": 0.6632815726539283,
      "learning_rate": 0.000124210062347472,
      "loss": 12.4149,
      "step": 16142
    },
    {
      "epoch": 0.8790508668395581,
      "grad_norm": 0.5648761481147258,
      "learning_rate": 0.0001242015063282546,
      "loss": 12.2223,
      "step": 16143
    },
    {
      "epoch": 0.8791053208361411,
      "grad_norm": 0.5085759784843153,
      "learning_rate": 0.00012419295012084224,
      "loss": 12.2876,
      "step": 16144
    },
    {
      "epoch": 0.8791597748327241,
      "grad_norm": 0.5453487051823546,
      "learning_rate": 0.00012418439372530141,
      "loss": 12.1462,
      "step": 16145
    },
    {
      "epoch": 0.8792142288293071,
      "grad_norm": 0.5181223589736799,
      "learning_rate": 0.0001241758371416987,
      "loss": 12.0158,
      "step": 16146
    },
    {
      "epoch": 0.8792686828258901,
      "grad_norm": 0.677804920864788,
      "learning_rate": 0.00012416728037010062,
      "loss": 12.3514,
      "step": 16147
    },
    {
      "epoch": 0.8793231368224732,
      "grad_norm": 0.5147318760050673,
      "learning_rate": 0.00012415872341057369,
      "loss": 12.1635,
      "step": 16148
    },
    {
      "epoch": 0.8793775908190562,
      "grad_norm": 0.522595656999475,
      "learning_rate": 0.00012415016626318452,
      "loss": 12.0867,
      "step": 16149
    },
    {
      "epoch": 0.8794320448156392,
      "grad_norm": 0.6116936916835619,
      "learning_rate": 0.00012414160892799958,
      "loss": 12.1803,
      "step": 16150
    },
    {
      "epoch": 0.8794864988122222,
      "grad_norm": 0.5489823607717342,
      "learning_rate": 0.00012413305140508544,
      "loss": 12.1459,
      "step": 16151
    },
    {
      "epoch": 0.8795409528088052,
      "grad_norm": 0.5998587379305671,
      "learning_rate": 0.00012412449369450865,
      "loss": 12.0937,
      "step": 16152
    },
    {
      "epoch": 0.8795954068053882,
      "grad_norm": 0.6123606539421548,
      "learning_rate": 0.00012411593579633574,
      "loss": 12.1539,
      "step": 16153
    },
    {
      "epoch": 0.8796498608019713,
      "grad_norm": 0.6051188759788264,
      "learning_rate": 0.0001241073777106333,
      "loss": 12.173,
      "step": 16154
    },
    {
      "epoch": 0.8797043147985543,
      "grad_norm": 0.541487724905037,
      "learning_rate": 0.0001240988194374678,
      "loss": 12.0629,
      "step": 16155
    },
    {
      "epoch": 0.8797587687951373,
      "grad_norm": 0.5286488857519637,
      "learning_rate": 0.00012409026097690587,
      "loss": 12.237,
      "step": 16156
    },
    {
      "epoch": 0.8798132227917203,
      "grad_norm": 0.5222242290522313,
      "learning_rate": 0.00012408170232901404,
      "loss": 12.1247,
      "step": 16157
    },
    {
      "epoch": 0.8798676767883032,
      "grad_norm": 0.5605723823245377,
      "learning_rate": 0.00012407314349385885,
      "loss": 12.2575,
      "step": 16158
    },
    {
      "epoch": 0.8799221307848863,
      "grad_norm": 0.5311519997094444,
      "learning_rate": 0.00012406458447150685,
      "loss": 12.1567,
      "step": 16159
    },
    {
      "epoch": 0.8799765847814693,
      "grad_norm": 0.522106869753087,
      "learning_rate": 0.0001240560252620246,
      "loss": 12.1882,
      "step": 16160
    },
    {
      "epoch": 0.8800310387780523,
      "grad_norm": 0.5700146146601299,
      "learning_rate": 0.00012404746586547867,
      "loss": 11.8611,
      "step": 16161
    },
    {
      "epoch": 0.8800854927746353,
      "grad_norm": 0.5800248799485075,
      "learning_rate": 0.00012403890628193563,
      "loss": 12.0938,
      "step": 16162
    },
    {
      "epoch": 0.8801399467712183,
      "grad_norm": 0.5157443715077356,
      "learning_rate": 0.00012403034651146198,
      "loss": 12.1552,
      "step": 16163
    },
    {
      "epoch": 0.8801944007678013,
      "grad_norm": 0.5421930476489523,
      "learning_rate": 0.00012402178655412436,
      "loss": 12.1899,
      "step": 16164
    },
    {
      "epoch": 0.8802488547643844,
      "grad_norm": 0.6029714786139897,
      "learning_rate": 0.0001240132264099893,
      "loss": 12.1137,
      "step": 16165
    },
    {
      "epoch": 0.8803033087609674,
      "grad_norm": 0.5451626473064809,
      "learning_rate": 0.00012400466607912332,
      "loss": 12.1901,
      "step": 16166
    },
    {
      "epoch": 0.8803577627575504,
      "grad_norm": 0.5134452705606481,
      "learning_rate": 0.0001239961055615931,
      "loss": 12.1074,
      "step": 16167
    },
    {
      "epoch": 0.8804122167541334,
      "grad_norm": 0.5925524126284681,
      "learning_rate": 0.0001239875448574651,
      "loss": 12.3034,
      "step": 16168
    },
    {
      "epoch": 0.8804666707507164,
      "grad_norm": 0.5287324031552987,
      "learning_rate": 0.0001239789839668059,
      "loss": 12.2371,
      "step": 16169
    },
    {
      "epoch": 0.8805211247472994,
      "grad_norm": 0.5181531375923086,
      "learning_rate": 0.00012397042288968214,
      "loss": 12.079,
      "step": 16170
    },
    {
      "epoch": 0.8805755787438825,
      "grad_norm": 0.509508990530057,
      "learning_rate": 0.00012396186162616038,
      "loss": 11.9842,
      "step": 16171
    },
    {
      "epoch": 0.8806300327404655,
      "grad_norm": 0.5914236089719387,
      "learning_rate": 0.00012395330017630712,
      "loss": 12.2112,
      "step": 16172
    },
    {
      "epoch": 0.8806844867370485,
      "grad_norm": 0.5585576561169054,
      "learning_rate": 0.00012394473854018898,
      "loss": 12.0564,
      "step": 16173
    },
    {
      "epoch": 0.8807389407336315,
      "grad_norm": 0.5256569160661628,
      "learning_rate": 0.00012393617671787254,
      "loss": 12.1797,
      "step": 16174
    },
    {
      "epoch": 0.8807933947302145,
      "grad_norm": 0.5503036986398105,
      "learning_rate": 0.0001239276147094244,
      "loss": 12.1849,
      "step": 16175
    },
    {
      "epoch": 0.8808478487267974,
      "grad_norm": 0.5899224050988658,
      "learning_rate": 0.0001239190525149111,
      "loss": 12.2352,
      "step": 16176
    },
    {
      "epoch": 0.8809023027233805,
      "grad_norm": 0.5690699450700633,
      "learning_rate": 0.0001239104901343992,
      "loss": 12.114,
      "step": 16177
    },
    {
      "epoch": 0.8809567567199635,
      "grad_norm": 0.5769616797004105,
      "learning_rate": 0.00012390192756795538,
      "loss": 12.0492,
      "step": 16178
    },
    {
      "epoch": 0.8810112107165465,
      "grad_norm": 0.570383479686332,
      "learning_rate": 0.00012389336481564614,
      "loss": 12.1202,
      "step": 16179
    },
    {
      "epoch": 0.8810656647131295,
      "grad_norm": 0.5639287176449898,
      "learning_rate": 0.00012388480187753808,
      "loss": 12.0932,
      "step": 16180
    },
    {
      "epoch": 0.8811201187097125,
      "grad_norm": 0.5824002134140114,
      "learning_rate": 0.0001238762387536978,
      "loss": 12.2077,
      "step": 16181
    },
    {
      "epoch": 0.8811745727062955,
      "grad_norm": 0.5580492724272099,
      "learning_rate": 0.0001238676754441919,
      "loss": 12.1451,
      "step": 16182
    },
    {
      "epoch": 0.8812290267028786,
      "grad_norm": 0.5599035260414097,
      "learning_rate": 0.00012385911194908692,
      "loss": 12.1846,
      "step": 16183
    },
    {
      "epoch": 0.8812834806994616,
      "grad_norm": 0.5425513211680949,
      "learning_rate": 0.00012385054826844952,
      "loss": 11.9813,
      "step": 16184
    },
    {
      "epoch": 0.8813379346960446,
      "grad_norm": 0.5685357381558466,
      "learning_rate": 0.00012384198440234625,
      "loss": 12.0235,
      "step": 16185
    },
    {
      "epoch": 0.8813923886926276,
      "grad_norm": 0.5543533366808976,
      "learning_rate": 0.00012383342035084368,
      "loss": 12.1006,
      "step": 16186
    },
    {
      "epoch": 0.8814468426892106,
      "grad_norm": 0.5634053378083402,
      "learning_rate": 0.00012382485611400846,
      "loss": 12.1838,
      "step": 16187
    },
    {
      "epoch": 0.8815012966857936,
      "grad_norm": 0.6518959759414273,
      "learning_rate": 0.00012381629169190715,
      "loss": 12.2529,
      "step": 16188
    },
    {
      "epoch": 0.8815557506823767,
      "grad_norm": 0.6343989078058219,
      "learning_rate": 0.0001238077270846064,
      "loss": 12.1136,
      "step": 16189
    },
    {
      "epoch": 0.8816102046789597,
      "grad_norm": 0.5619803061551234,
      "learning_rate": 0.00012379916229217274,
      "loss": 12.0956,
      "step": 16190
    },
    {
      "epoch": 0.8816646586755427,
      "grad_norm": 0.5500800945417131,
      "learning_rate": 0.00012379059731467277,
      "loss": 12.1264,
      "step": 16191
    },
    {
      "epoch": 0.8817191126721257,
      "grad_norm": 0.6534795437272554,
      "learning_rate": 0.00012378203215217316,
      "loss": 12.1569,
      "step": 16192
    },
    {
      "epoch": 0.8817735666687087,
      "grad_norm": 0.5751586600273532,
      "learning_rate": 0.0001237734668047405,
      "loss": 12.1951,
      "step": 16193
    },
    {
      "epoch": 0.8818280206652918,
      "grad_norm": 0.5799723742902898,
      "learning_rate": 0.0001237649012724414,
      "loss": 12.2284,
      "step": 16194
    },
    {
      "epoch": 0.8818824746618747,
      "grad_norm": 0.5652758554051909,
      "learning_rate": 0.00012375633555534237,
      "loss": 12.078,
      "step": 16195
    },
    {
      "epoch": 0.8819369286584577,
      "grad_norm": 0.5470256921634873,
      "learning_rate": 0.00012374776965351012,
      "loss": 12.1292,
      "step": 16196
    },
    {
      "epoch": 0.8819913826550407,
      "grad_norm": 0.5500982540293085,
      "learning_rate": 0.00012373920356701122,
      "loss": 12.1825,
      "step": 16197
    },
    {
      "epoch": 0.8820458366516237,
      "grad_norm": 0.553900653584647,
      "learning_rate": 0.0001237306372959123,
      "loss": 12.1216,
      "step": 16198
    },
    {
      "epoch": 0.8821002906482067,
      "grad_norm": 0.5561992559435806,
      "learning_rate": 0.00012372207084027998,
      "loss": 12.2922,
      "step": 16199
    },
    {
      "epoch": 0.8821547446447898,
      "grad_norm": 0.5356433230478098,
      "learning_rate": 0.00012371350420018083,
      "loss": 12.1967,
      "step": 16200
    },
    {
      "epoch": 0.8822091986413728,
      "grad_norm": 0.5244114586518281,
      "learning_rate": 0.00012370493737568153,
      "loss": 12.2054,
      "step": 16201
    },
    {
      "epoch": 0.8822636526379558,
      "grad_norm": 0.6249332730815699,
      "learning_rate": 0.00012369637036684862,
      "loss": 12.1978,
      "step": 16202
    },
    {
      "epoch": 0.8823181066345388,
      "grad_norm": 0.5166215210519938,
      "learning_rate": 0.00012368780317374876,
      "loss": 12.172,
      "step": 16203
    },
    {
      "epoch": 0.8823725606311218,
      "grad_norm": 0.5655592203207452,
      "learning_rate": 0.00012367923579644863,
      "loss": 12.0116,
      "step": 16204
    },
    {
      "epoch": 0.8824270146277048,
      "grad_norm": 0.5342191190418638,
      "learning_rate": 0.00012367066823501475,
      "loss": 12.1477,
      "step": 16205
    },
    {
      "epoch": 0.8824814686242879,
      "grad_norm": 0.5122266335572258,
      "learning_rate": 0.0001236621004895138,
      "loss": 12.2533,
      "step": 16206
    },
    {
      "epoch": 0.8825359226208709,
      "grad_norm": 0.6474094156178377,
      "learning_rate": 0.00012365353256001238,
      "loss": 12.2779,
      "step": 16207
    },
    {
      "epoch": 0.8825903766174539,
      "grad_norm": 0.534846689296231,
      "learning_rate": 0.00012364496444657708,
      "loss": 12.1578,
      "step": 16208
    },
    {
      "epoch": 0.8826448306140369,
      "grad_norm": 0.5461817063501252,
      "learning_rate": 0.00012363639614927465,
      "loss": 12.2337,
      "step": 16209
    },
    {
      "epoch": 0.8826992846106199,
      "grad_norm": 0.5458809676123125,
      "learning_rate": 0.00012362782766817162,
      "loss": 12.2354,
      "step": 16210
    },
    {
      "epoch": 0.8827537386072029,
      "grad_norm": 0.49927285854458225,
      "learning_rate": 0.0001236192590033346,
      "loss": 12.107,
      "step": 16211
    },
    {
      "epoch": 0.882808192603786,
      "grad_norm": 0.5523903728560546,
      "learning_rate": 0.00012361069015483028,
      "loss": 12.1729,
      "step": 16212
    },
    {
      "epoch": 0.882862646600369,
      "grad_norm": 0.5428571481794514,
      "learning_rate": 0.00012360212112272527,
      "loss": 12.3167,
      "step": 16213
    },
    {
      "epoch": 0.8829171005969519,
      "grad_norm": 0.5616631045634104,
      "learning_rate": 0.00012359355190708622,
      "loss": 12.1382,
      "step": 16214
    },
    {
      "epoch": 0.8829715545935349,
      "grad_norm": 0.5735167151423458,
      "learning_rate": 0.00012358498250797975,
      "loss": 12.1902,
      "step": 16215
    },
    {
      "epoch": 0.8830260085901179,
      "grad_norm": 0.505655421612921,
      "learning_rate": 0.00012357641292547249,
      "loss": 12.1472,
      "step": 16216
    },
    {
      "epoch": 0.8830804625867009,
      "grad_norm": 0.5735619841639722,
      "learning_rate": 0.0001235678431596311,
      "loss": 12.1078,
      "step": 16217
    },
    {
      "epoch": 0.883134916583284,
      "grad_norm": 0.686628470289581,
      "learning_rate": 0.0001235592732105222,
      "loss": 12.2523,
      "step": 16218
    },
    {
      "epoch": 0.883189370579867,
      "grad_norm": 0.5126576113589593,
      "learning_rate": 0.00012355070307821245,
      "loss": 12.2123,
      "step": 16219
    },
    {
      "epoch": 0.88324382457645,
      "grad_norm": 0.6286287193684238,
      "learning_rate": 0.0001235421327627685,
      "loss": 12.2636,
      "step": 16220
    },
    {
      "epoch": 0.883298278573033,
      "grad_norm": 0.5610115455566963,
      "learning_rate": 0.00012353356226425693,
      "loss": 12.1751,
      "step": 16221
    },
    {
      "epoch": 0.883352732569616,
      "grad_norm": 0.5548192270939025,
      "learning_rate": 0.00012352499158274448,
      "loss": 12.2444,
      "step": 16222
    },
    {
      "epoch": 0.883407186566199,
      "grad_norm": 0.5182734831287024,
      "learning_rate": 0.0001235164207182977,
      "loss": 12.0495,
      "step": 16223
    },
    {
      "epoch": 0.8834616405627821,
      "grad_norm": 0.5282561495176271,
      "learning_rate": 0.00012350784967098333,
      "loss": 12.2166,
      "step": 16224
    },
    {
      "epoch": 0.8835160945593651,
      "grad_norm": 0.5400405941500195,
      "learning_rate": 0.000123499278440868,
      "loss": 12.1352,
      "step": 16225
    },
    {
      "epoch": 0.8835705485559481,
      "grad_norm": 0.5242033447130563,
      "learning_rate": 0.0001234907070280183,
      "loss": 12.1647,
      "step": 16226
    },
    {
      "epoch": 0.8836250025525311,
      "grad_norm": 0.5239109929409396,
      "learning_rate": 0.00012348213543250094,
      "loss": 12.1428,
      "step": 16227
    },
    {
      "epoch": 0.8836794565491141,
      "grad_norm": 0.5410506015404936,
      "learning_rate": 0.00012347356365438253,
      "loss": 12.1778,
      "step": 16228
    },
    {
      "epoch": 0.8837339105456972,
      "grad_norm": 0.6075298664685101,
      "learning_rate": 0.0001234649916937298,
      "loss": 12.1646,
      "step": 16229
    },
    {
      "epoch": 0.8837883645422802,
      "grad_norm": 0.5405876936004674,
      "learning_rate": 0.00012345641955060932,
      "loss": 12.2044,
      "step": 16230
    },
    {
      "epoch": 0.8838428185388632,
      "grad_norm": 0.5208606016764383,
      "learning_rate": 0.0001234478472250878,
      "loss": 12.0691,
      "step": 16231
    },
    {
      "epoch": 0.8838972725354461,
      "grad_norm": 0.5368623083267269,
      "learning_rate": 0.00012343927471723188,
      "loss": 12.2568,
      "step": 16232
    },
    {
      "epoch": 0.8839517265320291,
      "grad_norm": 0.5468750069143732,
      "learning_rate": 0.00012343070202710824,
      "loss": 12.0706,
      "step": 16233
    },
    {
      "epoch": 0.8840061805286121,
      "grad_norm": 0.5499608163084332,
      "learning_rate": 0.00012342212915478353,
      "loss": 12.2757,
      "step": 16234
    },
    {
      "epoch": 0.8840606345251952,
      "grad_norm": 0.5851504111453643,
      "learning_rate": 0.0001234135561003244,
      "loss": 12.2076,
      "step": 16235
    },
    {
      "epoch": 0.8841150885217782,
      "grad_norm": 0.5974130400887863,
      "learning_rate": 0.00012340498286379756,
      "loss": 12.1854,
      "step": 16236
    },
    {
      "epoch": 0.8841695425183612,
      "grad_norm": 0.5491304519499025,
      "learning_rate": 0.00012339640944526964,
      "loss": 12.2238,
      "step": 16237
    },
    {
      "epoch": 0.8842239965149442,
      "grad_norm": 0.5346073392197845,
      "learning_rate": 0.0001233878358448073,
      "loss": 12.202,
      "step": 16238
    },
    {
      "epoch": 0.8842784505115272,
      "grad_norm": 0.6736336593523841,
      "learning_rate": 0.00012337926206247723,
      "loss": 12.3677,
      "step": 16239
    },
    {
      "epoch": 0.8843329045081102,
      "grad_norm": 0.5083872491556979,
      "learning_rate": 0.0001233706880983461,
      "loss": 12.2147,
      "step": 16240
    },
    {
      "epoch": 0.8843873585046933,
      "grad_norm": 0.5800111367392793,
      "learning_rate": 0.00012336211395248058,
      "loss": 12.1883,
      "step": 16241
    },
    {
      "epoch": 0.8844418125012763,
      "grad_norm": 0.5869973957798797,
      "learning_rate": 0.00012335353962494736,
      "loss": 12.1675,
      "step": 16242
    },
    {
      "epoch": 0.8844962664978593,
      "grad_norm": 0.4909759782375525,
      "learning_rate": 0.0001233449651158131,
      "loss": 12.1226,
      "step": 16243
    },
    {
      "epoch": 0.8845507204944423,
      "grad_norm": 0.5963019301703091,
      "learning_rate": 0.00012333639042514446,
      "loss": 12.1958,
      "step": 16244
    },
    {
      "epoch": 0.8846051744910253,
      "grad_norm": 0.5483578119503779,
      "learning_rate": 0.00012332781555300816,
      "loss": 12.2078,
      "step": 16245
    },
    {
      "epoch": 0.8846596284876083,
      "grad_norm": 0.50012001467015,
      "learning_rate": 0.0001233192404994708,
      "loss": 12.1573,
      "step": 16246
    },
    {
      "epoch": 0.8847140824841914,
      "grad_norm": 0.5462820967201222,
      "learning_rate": 0.00012331066526459917,
      "loss": 12.1758,
      "step": 16247
    },
    {
      "epoch": 0.8847685364807744,
      "grad_norm": 0.5483925328240585,
      "learning_rate": 0.00012330208984845986,
      "loss": 12.1894,
      "step": 16248
    },
    {
      "epoch": 0.8848229904773574,
      "grad_norm": 0.5219144410997871,
      "learning_rate": 0.00012329351425111962,
      "loss": 12.1431,
      "step": 16249
    },
    {
      "epoch": 0.8848774444739403,
      "grad_norm": 0.5361374767689396,
      "learning_rate": 0.00012328493847264512,
      "loss": 12.1482,
      "step": 16250
    },
    {
      "epoch": 0.8849318984705233,
      "grad_norm": 0.5427429428242615,
      "learning_rate": 0.00012327636251310297,
      "loss": 12.258,
      "step": 16251
    },
    {
      "epoch": 0.8849863524671063,
      "grad_norm": 0.5912904179396163,
      "learning_rate": 0.00012326778637255996,
      "loss": 12.0507,
      "step": 16252
    },
    {
      "epoch": 0.8850408064636894,
      "grad_norm": 0.5378832184296238,
      "learning_rate": 0.00012325921005108275,
      "loss": 12.1293,
      "step": 16253
    },
    {
      "epoch": 0.8850952604602724,
      "grad_norm": 0.6089376138378523,
      "learning_rate": 0.000123250633548738,
      "loss": 12.1337,
      "step": 16254
    },
    {
      "epoch": 0.8851497144568554,
      "grad_norm": 0.5601034014331702,
      "learning_rate": 0.00012324205686559245,
      "loss": 12.2097,
      "step": 16255
    },
    {
      "epoch": 0.8852041684534384,
      "grad_norm": 0.5591657526431818,
      "learning_rate": 0.00012323348000171277,
      "loss": 12.0945,
      "step": 16256
    },
    {
      "epoch": 0.8852586224500214,
      "grad_norm": 0.5661957903672931,
      "learning_rate": 0.0001232249029571656,
      "loss": 12.1274,
      "step": 16257
    },
    {
      "epoch": 0.8853130764466045,
      "grad_norm": 0.5427624453408229,
      "learning_rate": 0.00012321632573201774,
      "loss": 12.1423,
      "step": 16258
    },
    {
      "epoch": 0.8853675304431875,
      "grad_norm": 0.6076214886273502,
      "learning_rate": 0.0001232077483263358,
      "loss": 12.1057,
      "step": 16259
    },
    {
      "epoch": 0.8854219844397705,
      "grad_norm": 0.5781501899130344,
      "learning_rate": 0.00012319917074018658,
      "loss": 12.1624,
      "step": 16260
    },
    {
      "epoch": 0.8854764384363535,
      "grad_norm": 0.5986244677879341,
      "learning_rate": 0.00012319059297363668,
      "loss": 12.2996,
      "step": 16261
    },
    {
      "epoch": 0.8855308924329365,
      "grad_norm": 0.6197660951665949,
      "learning_rate": 0.00012318201502675285,
      "loss": 12.154,
      "step": 16262
    },
    {
      "epoch": 0.8855853464295195,
      "grad_norm": 0.535054653933544,
      "learning_rate": 0.00012317343689960175,
      "loss": 12.0921,
      "step": 16263
    },
    {
      "epoch": 0.8856398004261026,
      "grad_norm": 0.565728316731146,
      "learning_rate": 0.00012316485859225016,
      "loss": 12.0625,
      "step": 16264
    },
    {
      "epoch": 0.8856942544226856,
      "grad_norm": 0.6148431799302818,
      "learning_rate": 0.0001231562801047647,
      "loss": 12.1517,
      "step": 16265
    },
    {
      "epoch": 0.8857487084192686,
      "grad_norm": 0.5625341975593025,
      "learning_rate": 0.00012314770143721218,
      "loss": 12.1513,
      "step": 16266
    },
    {
      "epoch": 0.8858031624158516,
      "grad_norm": 0.5596002114804742,
      "learning_rate": 0.00012313912258965924,
      "loss": 12.1375,
      "step": 16267
    },
    {
      "epoch": 0.8858576164124345,
      "grad_norm": 0.5337452297070369,
      "learning_rate": 0.00012313054356217256,
      "loss": 12.1935,
      "step": 16268
    },
    {
      "epoch": 0.8859120704090175,
      "grad_norm": 0.561215436744151,
      "learning_rate": 0.00012312196435481892,
      "loss": 12.2311,
      "step": 16269
    },
    {
      "epoch": 0.8859665244056006,
      "grad_norm": 0.701137105823549,
      "learning_rate": 0.000123113384967665,
      "loss": 12.1859,
      "step": 16270
    },
    {
      "epoch": 0.8860209784021836,
      "grad_norm": 0.6259806114999253,
      "learning_rate": 0.00012310480540077753,
      "loss": 12.2817,
      "step": 16271
    },
    {
      "epoch": 0.8860754323987666,
      "grad_norm": 0.575558700552531,
      "learning_rate": 0.00012309622565422323,
      "loss": 12.0026,
      "step": 16272
    },
    {
      "epoch": 0.8861298863953496,
      "grad_norm": 0.579891197993033,
      "learning_rate": 0.0001230876457280688,
      "loss": 12.1153,
      "step": 16273
    },
    {
      "epoch": 0.8861843403919326,
      "grad_norm": 0.5505349526446447,
      "learning_rate": 0.0001230790656223809,
      "loss": 11.9594,
      "step": 16274
    },
    {
      "epoch": 0.8862387943885156,
      "grad_norm": 0.6402945163461424,
      "learning_rate": 0.00012307048533722643,
      "loss": 12.2174,
      "step": 16275
    },
    {
      "epoch": 0.8862932483850987,
      "grad_norm": 0.5457796048605926,
      "learning_rate": 0.00012306190487267193,
      "loss": 12.0925,
      "step": 16276
    },
    {
      "epoch": 0.8863477023816817,
      "grad_norm": 0.5305821367555109,
      "learning_rate": 0.0001230533242287842,
      "loss": 12.1706,
      "step": 16277
    },
    {
      "epoch": 0.8864021563782647,
      "grad_norm": 0.5239564589926617,
      "learning_rate": 0.00012304474340562994,
      "loss": 12.1302,
      "step": 16278
    },
    {
      "epoch": 0.8864566103748477,
      "grad_norm": 0.5413941416062154,
      "learning_rate": 0.00012303616240327592,
      "loss": 12.0329,
      "step": 16279
    },
    {
      "epoch": 0.8865110643714307,
      "grad_norm": 0.5185014039488977,
      "learning_rate": 0.00012302758122178882,
      "loss": 12.0784,
      "step": 16280
    },
    {
      "epoch": 0.8865655183680137,
      "grad_norm": 0.4951657358478465,
      "learning_rate": 0.00012301899986123539,
      "loss": 12.1503,
      "step": 16281
    },
    {
      "epoch": 0.8866199723645968,
      "grad_norm": 0.5694149924321806,
      "learning_rate": 0.00012301041832168236,
      "loss": 12.2309,
      "step": 16282
    },
    {
      "epoch": 0.8866744263611798,
      "grad_norm": 0.5313585264494142,
      "learning_rate": 0.00012300183660319647,
      "loss": 12.1177,
      "step": 16283
    },
    {
      "epoch": 0.8867288803577628,
      "grad_norm": 0.6007296788229879,
      "learning_rate": 0.00012299325470584442,
      "loss": 12.1911,
      "step": 16284
    },
    {
      "epoch": 0.8867833343543458,
      "grad_norm": 0.5937648289150693,
      "learning_rate": 0.00012298467262969297,
      "loss": 12.3804,
      "step": 16285
    },
    {
      "epoch": 0.8868377883509287,
      "grad_norm": 0.5532089813326304,
      "learning_rate": 0.00012297609037480886,
      "loss": 12.0865,
      "step": 16286
    },
    {
      "epoch": 0.8868922423475117,
      "grad_norm": 0.6688215915836628,
      "learning_rate": 0.0001229675079412588,
      "loss": 12.1452,
      "step": 16287
    },
    {
      "epoch": 0.8869466963440948,
      "grad_norm": 0.5951680028150612,
      "learning_rate": 0.00012295892532910956,
      "loss": 12.3691,
      "step": 16288
    },
    {
      "epoch": 0.8870011503406778,
      "grad_norm": 0.5231733871097624,
      "learning_rate": 0.00012295034253842789,
      "loss": 12.0778,
      "step": 16289
    },
    {
      "epoch": 0.8870556043372608,
      "grad_norm": 0.5423677131696757,
      "learning_rate": 0.00012294175956928047,
      "loss": 12.1868,
      "step": 16290
    },
    {
      "epoch": 0.8871100583338438,
      "grad_norm": 0.4946320742927819,
      "learning_rate": 0.00012293317642173408,
      "loss": 12.0247,
      "step": 16291
    },
    {
      "epoch": 0.8871645123304268,
      "grad_norm": 0.5529772902355801,
      "learning_rate": 0.0001229245930958555,
      "loss": 12.1839,
      "step": 16292
    },
    {
      "epoch": 0.8872189663270099,
      "grad_norm": 0.6392097378935568,
      "learning_rate": 0.0001229160095917114,
      "loss": 12.286,
      "step": 16293
    },
    {
      "epoch": 0.8872734203235929,
      "grad_norm": 0.5291321917582374,
      "learning_rate": 0.00012290742590936857,
      "loss": 12.1872,
      "step": 16294
    },
    {
      "epoch": 0.8873278743201759,
      "grad_norm": 0.5831656257225017,
      "learning_rate": 0.00012289884204889378,
      "loss": 12.1548,
      "step": 16295
    },
    {
      "epoch": 0.8873823283167589,
      "grad_norm": 0.5499164854122962,
      "learning_rate": 0.00012289025801035373,
      "loss": 12.1475,
      "step": 16296
    },
    {
      "epoch": 0.8874367823133419,
      "grad_norm": 0.5528434084145767,
      "learning_rate": 0.0001228816737938152,
      "loss": 12.0544,
      "step": 16297
    },
    {
      "epoch": 0.8874912363099249,
      "grad_norm": 0.5137783709029952,
      "learning_rate": 0.00012287308939934496,
      "loss": 12.1277,
      "step": 16298
    },
    {
      "epoch": 0.887545690306508,
      "grad_norm": 0.5625471444950313,
      "learning_rate": 0.00012286450482700974,
      "loss": 12.2227,
      "step": 16299
    },
    {
      "epoch": 0.887600144303091,
      "grad_norm": 0.5936241104738309,
      "learning_rate": 0.00012285592007687626,
      "loss": 12.1372,
      "step": 16300
    },
    {
      "epoch": 0.887654598299674,
      "grad_norm": 0.5238975637094012,
      "learning_rate": 0.0001228473351490113,
      "loss": 12.0677,
      "step": 16301
    },
    {
      "epoch": 0.887709052296257,
      "grad_norm": 0.6606203666543881,
      "learning_rate": 0.00012283875004348167,
      "loss": 12.2815,
      "step": 16302
    },
    {
      "epoch": 0.88776350629284,
      "grad_norm": 0.5478035977552437,
      "learning_rate": 0.0001228301647603541,
      "loss": 12.1251,
      "step": 16303
    },
    {
      "epoch": 0.887817960289423,
      "grad_norm": 0.5400120488267081,
      "learning_rate": 0.00012282157929969533,
      "loss": 12.1598,
      "step": 16304
    },
    {
      "epoch": 0.887872414286006,
      "grad_norm": 0.6254885091083855,
      "learning_rate": 0.00012281299366157216,
      "loss": 12.2532,
      "step": 16305
    },
    {
      "epoch": 0.887926868282589,
      "grad_norm": 0.5666635884476116,
      "learning_rate": 0.00012280440784605124,
      "loss": 12.3314,
      "step": 16306
    },
    {
      "epoch": 0.887981322279172,
      "grad_norm": 0.5571324077998513,
      "learning_rate": 0.0001227958218531995,
      "loss": 12.2664,
      "step": 16307
    },
    {
      "epoch": 0.888035776275755,
      "grad_norm": 0.6383871738610409,
      "learning_rate": 0.00012278723568308358,
      "loss": 12.268,
      "step": 16308
    },
    {
      "epoch": 0.888090230272338,
      "grad_norm": 0.5489827954744986,
      "learning_rate": 0.00012277864933577033,
      "loss": 12.0518,
      "step": 16309
    },
    {
      "epoch": 0.888144684268921,
      "grad_norm": 0.5291341260448726,
      "learning_rate": 0.00012277006281132647,
      "loss": 12.1884,
      "step": 16310
    },
    {
      "epoch": 0.8881991382655041,
      "grad_norm": 0.6066827861748462,
      "learning_rate": 0.00012276147610981876,
      "loss": 12.1157,
      "step": 16311
    },
    {
      "epoch": 0.8882535922620871,
      "grad_norm": 0.5441557383002226,
      "learning_rate": 0.00012275288923131402,
      "loss": 12.0808,
      "step": 16312
    },
    {
      "epoch": 0.8883080462586701,
      "grad_norm": 0.4965751123135751,
      "learning_rate": 0.000122744302175879,
      "loss": 12.1255,
      "step": 16313
    },
    {
      "epoch": 0.8883625002552531,
      "grad_norm": 0.6936057137138693,
      "learning_rate": 0.00012273571494358045,
      "loss": 12.2344,
      "step": 16314
    },
    {
      "epoch": 0.8884169542518361,
      "grad_norm": 0.6014344061380988,
      "learning_rate": 0.00012272712753448522,
      "loss": 12.1617,
      "step": 16315
    },
    {
      "epoch": 0.8884714082484191,
      "grad_norm": 0.5859981341056422,
      "learning_rate": 0.00012271853994866,
      "loss": 12.0722,
      "step": 16316
    },
    {
      "epoch": 0.8885258622450022,
      "grad_norm": 0.573634206675812,
      "learning_rate": 0.00012270995218617155,
      "loss": 12.0755,
      "step": 16317
    },
    {
      "epoch": 0.8885803162415852,
      "grad_norm": 0.5927714899422986,
      "learning_rate": 0.00012270136424708675,
      "loss": 12.1773,
      "step": 16318
    },
    {
      "epoch": 0.8886347702381682,
      "grad_norm": 0.5807420605409117,
      "learning_rate": 0.00012269277613147233,
      "loss": 12.1732,
      "step": 16319
    },
    {
      "epoch": 0.8886892242347512,
      "grad_norm": 0.633388832174088,
      "learning_rate": 0.00012268418783939513,
      "loss": 12.1529,
      "step": 16320
    },
    {
      "epoch": 0.8887436782313342,
      "grad_norm": 0.6540598015292096,
      "learning_rate": 0.0001226755993709218,
      "loss": 11.977,
      "step": 16321
    },
    {
      "epoch": 0.8887981322279171,
      "grad_norm": 0.557404386843737,
      "learning_rate": 0.00012266701072611926,
      "loss": 12.097,
      "step": 16322
    },
    {
      "epoch": 0.8888525862245003,
      "grad_norm": 0.5960222979783917,
      "learning_rate": 0.0001226584219050542,
      "loss": 12.0737,
      "step": 16323
    },
    {
      "epoch": 0.8889070402210832,
      "grad_norm": 0.6314024877140787,
      "learning_rate": 0.00012264983290779347,
      "loss": 12.1702,
      "step": 16324
    },
    {
      "epoch": 0.8889614942176662,
      "grad_norm": 0.8419863945950672,
      "learning_rate": 0.00012264124373440388,
      "loss": 12.085,
      "step": 16325
    },
    {
      "epoch": 0.8890159482142492,
      "grad_norm": 0.5432509180119829,
      "learning_rate": 0.00012263265438495214,
      "loss": 12.069,
      "step": 16326
    },
    {
      "epoch": 0.8890704022108322,
      "grad_norm": 0.5902105609285478,
      "learning_rate": 0.0001226240648595051,
      "loss": 12.1508,
      "step": 16327
    },
    {
      "epoch": 0.8891248562074153,
      "grad_norm": 0.6650731727870468,
      "learning_rate": 0.00012261547515812952,
      "loss": 12.1691,
      "step": 16328
    },
    {
      "epoch": 0.8891793102039983,
      "grad_norm": 0.5477451446445849,
      "learning_rate": 0.00012260688528089222,
      "loss": 12.1464,
      "step": 16329
    },
    {
      "epoch": 0.8892337642005813,
      "grad_norm": 0.5608551356558081,
      "learning_rate": 0.00012259829522786003,
      "loss": 12.1764,
      "step": 16330
    },
    {
      "epoch": 0.8892882181971643,
      "grad_norm": 0.516292820888293,
      "learning_rate": 0.00012258970499909964,
      "loss": 12.1369,
      "step": 16331
    },
    {
      "epoch": 0.8893426721937473,
      "grad_norm": 0.558635622035302,
      "learning_rate": 0.00012258111459467796,
      "loss": 12.2017,
      "step": 16332
    },
    {
      "epoch": 0.8893971261903303,
      "grad_norm": 0.585917460390887,
      "learning_rate": 0.00012257252401466173,
      "loss": 12.2535,
      "step": 16333
    },
    {
      "epoch": 0.8894515801869134,
      "grad_norm": 0.5802831891033983,
      "learning_rate": 0.00012256393325911776,
      "loss": 12.1516,
      "step": 16334
    },
    {
      "epoch": 0.8895060341834964,
      "grad_norm": 0.5930622050620903,
      "learning_rate": 0.00012255534232811287,
      "loss": 12.1491,
      "step": 16335
    },
    {
      "epoch": 0.8895604881800794,
      "grad_norm": 0.5219656216979696,
      "learning_rate": 0.00012254675122171387,
      "loss": 12.1651,
      "step": 16336
    },
    {
      "epoch": 0.8896149421766624,
      "grad_norm": 0.5821970410221826,
      "learning_rate": 0.00012253815993998752,
      "loss": 12.1802,
      "step": 16337
    },
    {
      "epoch": 0.8896693961732454,
      "grad_norm": 0.5259181992579247,
      "learning_rate": 0.00012252956848300068,
      "loss": 12.2109,
      "step": 16338
    },
    {
      "epoch": 0.8897238501698284,
      "grad_norm": 0.562719113000839,
      "learning_rate": 0.0001225209768508201,
      "loss": 12.2265,
      "step": 16339
    },
    {
      "epoch": 0.8897783041664115,
      "grad_norm": 0.547578636302829,
      "learning_rate": 0.00012251238504351267,
      "loss": 12.1624,
      "step": 16340
    },
    {
      "epoch": 0.8898327581629945,
      "grad_norm": 0.5196405505890513,
      "learning_rate": 0.00012250379306114517,
      "loss": 12.0975,
      "step": 16341
    },
    {
      "epoch": 0.8898872121595774,
      "grad_norm": 0.6273714554787057,
      "learning_rate": 0.00012249520090378436,
      "loss": 12.3457,
      "step": 16342
    },
    {
      "epoch": 0.8899416661561604,
      "grad_norm": 0.5091159924147887,
      "learning_rate": 0.00012248660857149712,
      "loss": 12.2966,
      "step": 16343
    },
    {
      "epoch": 0.8899961201527434,
      "grad_norm": 0.5232684819703758,
      "learning_rate": 0.00012247801606435024,
      "loss": 12.1101,
      "step": 16344
    },
    {
      "epoch": 0.8900505741493264,
      "grad_norm": 0.5616859527095743,
      "learning_rate": 0.00012246942338241053,
      "loss": 12.1192,
      "step": 16345
    },
    {
      "epoch": 0.8901050281459095,
      "grad_norm": 0.6289801674171244,
      "learning_rate": 0.00012246083052574482,
      "loss": 12.1321,
      "step": 16346
    },
    {
      "epoch": 0.8901594821424925,
      "grad_norm": 0.5605368192195113,
      "learning_rate": 0.0001224522374944199,
      "loss": 12.1582,
      "step": 16347
    },
    {
      "epoch": 0.8902139361390755,
      "grad_norm": 0.59515441220397,
      "learning_rate": 0.00012244364428850267,
      "loss": 12.2295,
      "step": 16348
    },
    {
      "epoch": 0.8902683901356585,
      "grad_norm": 0.5613411178657675,
      "learning_rate": 0.00012243505090805986,
      "loss": 12.2211,
      "step": 16349
    },
    {
      "epoch": 0.8903228441322415,
      "grad_norm": 0.625794527188511,
      "learning_rate": 0.00012242645735315835,
      "loss": 12.0044,
      "step": 16350
    },
    {
      "epoch": 0.8903772981288245,
      "grad_norm": 0.6345129634846822,
      "learning_rate": 0.0001224178636238649,
      "loss": 12.1942,
      "step": 16351
    },
    {
      "epoch": 0.8904317521254076,
      "grad_norm": 0.5797800622091027,
      "learning_rate": 0.00012240926972024644,
      "loss": 12.0692,
      "step": 16352
    },
    {
      "epoch": 0.8904862061219906,
      "grad_norm": 0.6149532271097778,
      "learning_rate": 0.0001224006756423697,
      "loss": 12.163,
      "step": 16353
    },
    {
      "epoch": 0.8905406601185736,
      "grad_norm": 0.6390410226858315,
      "learning_rate": 0.0001223920813903016,
      "loss": 12.2444,
      "step": 16354
    },
    {
      "epoch": 0.8905951141151566,
      "grad_norm": 0.5575843300674095,
      "learning_rate": 0.00012238348696410887,
      "loss": 12.2189,
      "step": 16355
    },
    {
      "epoch": 0.8906495681117396,
      "grad_norm": 0.5934635885270143,
      "learning_rate": 0.00012237489236385842,
      "loss": 12.0775,
      "step": 16356
    },
    {
      "epoch": 0.8907040221083226,
      "grad_norm": 0.6940744268501108,
      "learning_rate": 0.00012236629758961704,
      "loss": 12.2922,
      "step": 16357
    },
    {
      "epoch": 0.8907584761049057,
      "grad_norm": 0.5486866213964949,
      "learning_rate": 0.00012235770264145158,
      "loss": 12.0936,
      "step": 16358
    },
    {
      "epoch": 0.8908129301014887,
      "grad_norm": 0.6384254601668586,
      "learning_rate": 0.00012234910751942888,
      "loss": 12.087,
      "step": 16359
    },
    {
      "epoch": 0.8908673840980716,
      "grad_norm": 0.5725247404303682,
      "learning_rate": 0.0001223405122236158,
      "loss": 12.1062,
      "step": 16360
    },
    {
      "epoch": 0.8909218380946546,
      "grad_norm": 0.6134408624917898,
      "learning_rate": 0.0001223319167540791,
      "loss": 12.2632,
      "step": 16361
    },
    {
      "epoch": 0.8909762920912376,
      "grad_norm": 0.5677903063661517,
      "learning_rate": 0.00012232332111088569,
      "loss": 12.1721,
      "step": 16362
    },
    {
      "epoch": 0.8910307460878207,
      "grad_norm": 0.5678554421060821,
      "learning_rate": 0.0001223147252941024,
      "loss": 12.2632,
      "step": 16363
    },
    {
      "epoch": 0.8910852000844037,
      "grad_norm": 0.5411360054176657,
      "learning_rate": 0.00012230612930379605,
      "loss": 12.1227,
      "step": 16364
    },
    {
      "epoch": 0.8911396540809867,
      "grad_norm": 0.5789415259530507,
      "learning_rate": 0.00012229753314003353,
      "loss": 12.0349,
      "step": 16365
    },
    {
      "epoch": 0.8911941080775697,
      "grad_norm": 0.6118033550791085,
      "learning_rate": 0.0001222889368028816,
      "loss": 12.3661,
      "step": 16366
    },
    {
      "epoch": 0.8912485620741527,
      "grad_norm": 0.5498114164510856,
      "learning_rate": 0.0001222803402924072,
      "loss": 12.0789,
      "step": 16367
    },
    {
      "epoch": 0.8913030160707357,
      "grad_norm": 0.5455619732312031,
      "learning_rate": 0.00012227174360867712,
      "loss": 12.1499,
      "step": 16368
    },
    {
      "epoch": 0.8913574700673188,
      "grad_norm": 0.5849315482521636,
      "learning_rate": 0.00012226314675175824,
      "loss": 12.123,
      "step": 16369
    },
    {
      "epoch": 0.8914119240639018,
      "grad_norm": 0.6706327880267889,
      "learning_rate": 0.00012225454972171742,
      "loss": 12.3123,
      "step": 16370
    },
    {
      "epoch": 0.8914663780604848,
      "grad_norm": 0.563472642508159,
      "learning_rate": 0.00012224595251862145,
      "loss": 12.1743,
      "step": 16371
    },
    {
      "epoch": 0.8915208320570678,
      "grad_norm": 0.5805514198217513,
      "learning_rate": 0.0001222373551425372,
      "loss": 12.2397,
      "step": 16372
    },
    {
      "epoch": 0.8915752860536508,
      "grad_norm": 0.5495186261985949,
      "learning_rate": 0.00012222875759353158,
      "loss": 12.2148,
      "step": 16373
    },
    {
      "epoch": 0.8916297400502338,
      "grad_norm": 0.5189203705272784,
      "learning_rate": 0.0001222201598716714,
      "loss": 12.1897,
      "step": 16374
    },
    {
      "epoch": 0.8916841940468169,
      "grad_norm": 0.5316500332118727,
      "learning_rate": 0.00012221156197702356,
      "loss": 12.1792,
      "step": 16375
    },
    {
      "epoch": 0.8917386480433999,
      "grad_norm": 0.5383019627872128,
      "learning_rate": 0.0001222029639096549,
      "loss": 12.1051,
      "step": 16376
    },
    {
      "epoch": 0.8917931020399829,
      "grad_norm": 0.5819361656303953,
      "learning_rate": 0.00012219436566963222,
      "loss": 12.3008,
      "step": 16377
    },
    {
      "epoch": 0.8918475560365658,
      "grad_norm": 0.6054698883628403,
      "learning_rate": 0.00012218576725702245,
      "loss": 12.2099,
      "step": 16378
    },
    {
      "epoch": 0.8919020100331488,
      "grad_norm": 0.56277415551667,
      "learning_rate": 0.00012217716867189243,
      "loss": 12.11,
      "step": 16379
    },
    {
      "epoch": 0.8919564640297318,
      "grad_norm": 0.5353622779874301,
      "learning_rate": 0.00012216856991430905,
      "loss": 12.1667,
      "step": 16380
    },
    {
      "epoch": 0.8920109180263149,
      "grad_norm": 0.5907613177814792,
      "learning_rate": 0.00012215997098433912,
      "loss": 12.2423,
      "step": 16381
    },
    {
      "epoch": 0.8920653720228979,
      "grad_norm": 0.5314572914567044,
      "learning_rate": 0.00012215137188204957,
      "loss": 12.2518,
      "step": 16382
    },
    {
      "epoch": 0.8921198260194809,
      "grad_norm": 0.5587918377791286,
      "learning_rate": 0.00012214277260750718,
      "loss": 12.1841,
      "step": 16383
    },
    {
      "epoch": 0.8921742800160639,
      "grad_norm": 0.5779357858442474,
      "learning_rate": 0.00012213417316077894,
      "loss": 12.0817,
      "step": 16384
    },
    {
      "epoch": 0.8922287340126469,
      "grad_norm": 0.5427932311031911,
      "learning_rate": 0.00012212557354193164,
      "loss": 12.1866,
      "step": 16385
    },
    {
      "epoch": 0.8922831880092299,
      "grad_norm": 0.5430726865693253,
      "learning_rate": 0.00012211697375103217,
      "loss": 12.0536,
      "step": 16386
    },
    {
      "epoch": 0.892337642005813,
      "grad_norm": 0.5788113619506856,
      "learning_rate": 0.0001221083737881474,
      "loss": 12.0162,
      "step": 16387
    },
    {
      "epoch": 0.892392096002396,
      "grad_norm": 0.5527416162532146,
      "learning_rate": 0.00012209977365334419,
      "loss": 12.1027,
      "step": 16388
    },
    {
      "epoch": 0.892446549998979,
      "grad_norm": 0.5431246939434727,
      "learning_rate": 0.00012209117334668944,
      "loss": 12.1254,
      "step": 16389
    },
    {
      "epoch": 0.892501003995562,
      "grad_norm": 0.5809144579873806,
      "learning_rate": 0.00012208257286825004,
      "loss": 12.0177,
      "step": 16390
    },
    {
      "epoch": 0.892555457992145,
      "grad_norm": 0.5657949703100597,
      "learning_rate": 0.00012207397221809286,
      "loss": 12.2054,
      "step": 16391
    },
    {
      "epoch": 0.8926099119887281,
      "grad_norm": 0.5635021259578552,
      "learning_rate": 0.00012206537139628476,
      "loss": 12.1643,
      "step": 16392
    },
    {
      "epoch": 0.8926643659853111,
      "grad_norm": 0.7856142703467077,
      "learning_rate": 0.00012205677040289263,
      "loss": 12.2693,
      "step": 16393
    },
    {
      "epoch": 0.8927188199818941,
      "grad_norm": 0.5431584101018093,
      "learning_rate": 0.00012204816923798332,
      "loss": 12.1436,
      "step": 16394
    },
    {
      "epoch": 0.892773273978477,
      "grad_norm": 0.5429192019043648,
      "learning_rate": 0.00012203956790162379,
      "loss": 12.2422,
      "step": 16395
    },
    {
      "epoch": 0.89282772797506,
      "grad_norm": 0.5613179884500105,
      "learning_rate": 0.00012203096639388088,
      "loss": 12.143,
      "step": 16396
    },
    {
      "epoch": 0.892882181971643,
      "grad_norm": 0.5946760079053818,
      "learning_rate": 0.00012202236471482147,
      "loss": 12.1774,
      "step": 16397
    },
    {
      "epoch": 0.8929366359682261,
      "grad_norm": 0.553657927886927,
      "learning_rate": 0.00012201376286451247,
      "loss": 12.0548,
      "step": 16398
    },
    {
      "epoch": 0.8929910899648091,
      "grad_norm": 0.5354058557561385,
      "learning_rate": 0.00012200516084302074,
      "loss": 12.2015,
      "step": 16399
    },
    {
      "epoch": 0.8930455439613921,
      "grad_norm": 0.5010026502091414,
      "learning_rate": 0.00012199655865041318,
      "loss": 12.102,
      "step": 16400
    },
    {
      "epoch": 0.8930999979579751,
      "grad_norm": 0.5683799680633952,
      "learning_rate": 0.00012198795628675673,
      "loss": 11.9626,
      "step": 16401
    },
    {
      "epoch": 0.8931544519545581,
      "grad_norm": 0.5685946238732208,
      "learning_rate": 0.00012197935375211822,
      "loss": 12.1334,
      "step": 16402
    },
    {
      "epoch": 0.8932089059511411,
      "grad_norm": 0.5321421061657889,
      "learning_rate": 0.00012197075104656457,
      "loss": 12.0824,
      "step": 16403
    },
    {
      "epoch": 0.8932633599477242,
      "grad_norm": 0.6369712125898802,
      "learning_rate": 0.00012196214817016267,
      "loss": 12.0904,
      "step": 16404
    },
    {
      "epoch": 0.8933178139443072,
      "grad_norm": 0.5264122019388323,
      "learning_rate": 0.0001219535451229794,
      "loss": 11.9427,
      "step": 16405
    },
    {
      "epoch": 0.8933722679408902,
      "grad_norm": 0.5250222597786064,
      "learning_rate": 0.00012194494190508175,
      "loss": 12.2276,
      "step": 16406
    },
    {
      "epoch": 0.8934267219374732,
      "grad_norm": 0.5850483257175525,
      "learning_rate": 0.00012193633851653652,
      "loss": 12.3033,
      "step": 16407
    },
    {
      "epoch": 0.8934811759340562,
      "grad_norm": 0.5621707395066748,
      "learning_rate": 0.00012192773495741063,
      "loss": 12.2006,
      "step": 16408
    },
    {
      "epoch": 0.8935356299306392,
      "grad_norm": 0.6184809514475227,
      "learning_rate": 0.00012191913122777098,
      "loss": 12.1792,
      "step": 16409
    },
    {
      "epoch": 0.8935900839272223,
      "grad_norm": 0.5997452134431591,
      "learning_rate": 0.00012191052732768453,
      "loss": 12.1659,
      "step": 16410
    },
    {
      "epoch": 0.8936445379238053,
      "grad_norm": 0.6570410268290692,
      "learning_rate": 0.00012190192325721812,
      "loss": 12.1779,
      "step": 16411
    },
    {
      "epoch": 0.8936989919203883,
      "grad_norm": 0.579003499709148,
      "learning_rate": 0.0001218933190164387,
      "loss": 12.2527,
      "step": 16412
    },
    {
      "epoch": 0.8937534459169713,
      "grad_norm": 0.5530522847933657,
      "learning_rate": 0.00012188471460541315,
      "loss": 12.0883,
      "step": 16413
    },
    {
      "epoch": 0.8938078999135542,
      "grad_norm": 0.6791854967291375,
      "learning_rate": 0.00012187611002420838,
      "loss": 12.1236,
      "step": 16414
    },
    {
      "epoch": 0.8938623539101372,
      "grad_norm": 0.5153430179909868,
      "learning_rate": 0.00012186750527289132,
      "loss": 12.1704,
      "step": 16415
    },
    {
      "epoch": 0.8939168079067203,
      "grad_norm": 0.5253876428987704,
      "learning_rate": 0.00012185890035152887,
      "loss": 12.2482,
      "step": 16416
    },
    {
      "epoch": 0.8939712619033033,
      "grad_norm": 0.5821672884226603,
      "learning_rate": 0.00012185029526018794,
      "loss": 12.2772,
      "step": 16417
    },
    {
      "epoch": 0.8940257158998863,
      "grad_norm": 0.5224363701557859,
      "learning_rate": 0.00012184168999893546,
      "loss": 12.1356,
      "step": 16418
    },
    {
      "epoch": 0.8940801698964693,
      "grad_norm": 0.588352730562737,
      "learning_rate": 0.00012183308456783832,
      "loss": 12.2662,
      "step": 16419
    },
    {
      "epoch": 0.8941346238930523,
      "grad_norm": 0.5155512892761352,
      "learning_rate": 0.00012182447896696347,
      "loss": 12.0925,
      "step": 16420
    },
    {
      "epoch": 0.8941890778896353,
      "grad_norm": 0.5575012062730131,
      "learning_rate": 0.00012181587319637782,
      "loss": 12.1494,
      "step": 16421
    },
    {
      "epoch": 0.8942435318862184,
      "grad_norm": 0.5537627331780988,
      "learning_rate": 0.00012180726725614826,
      "loss": 12.0568,
      "step": 16422
    },
    {
      "epoch": 0.8942979858828014,
      "grad_norm": 0.5203552197086324,
      "learning_rate": 0.00012179866114634174,
      "loss": 12.12,
      "step": 16423
    },
    {
      "epoch": 0.8943524398793844,
      "grad_norm": 0.5283891939212064,
      "learning_rate": 0.00012179005486702517,
      "loss": 12.0989,
      "step": 16424
    },
    {
      "epoch": 0.8944068938759674,
      "grad_norm": 0.5485087823080463,
      "learning_rate": 0.00012178144841826548,
      "loss": 12.159,
      "step": 16425
    },
    {
      "epoch": 0.8944613478725504,
      "grad_norm": 0.5690013527694032,
      "learning_rate": 0.0001217728418001296,
      "loss": 12.1885,
      "step": 16426
    },
    {
      "epoch": 0.8945158018691335,
      "grad_norm": 0.5463935145707699,
      "learning_rate": 0.00012176423501268445,
      "loss": 12.2017,
      "step": 16427
    },
    {
      "epoch": 0.8945702558657165,
      "grad_norm": 0.5334736599123772,
      "learning_rate": 0.00012175562805599696,
      "loss": 12.2179,
      "step": 16428
    },
    {
      "epoch": 0.8946247098622995,
      "grad_norm": 0.5351992571302758,
      "learning_rate": 0.00012174702093013403,
      "loss": 12.1498,
      "step": 16429
    },
    {
      "epoch": 0.8946791638588825,
      "grad_norm": 0.6364589176976457,
      "learning_rate": 0.00012173841363516265,
      "loss": 12.1802,
      "step": 16430
    },
    {
      "epoch": 0.8947336178554655,
      "grad_norm": 0.5253528278949113,
      "learning_rate": 0.00012172980617114975,
      "loss": 12.1021,
      "step": 16431
    },
    {
      "epoch": 0.8947880718520484,
      "grad_norm": 0.5145999047777803,
      "learning_rate": 0.00012172119853816217,
      "loss": 12.2855,
      "step": 16432
    },
    {
      "epoch": 0.8948425258486316,
      "grad_norm": 0.6613421138830665,
      "learning_rate": 0.00012171259073626693,
      "loss": 11.9993,
      "step": 16433
    },
    {
      "epoch": 0.8948969798452145,
      "grad_norm": 0.4970791425366797,
      "learning_rate": 0.00012170398276553094,
      "loss": 11.9895,
      "step": 16434
    },
    {
      "epoch": 0.8949514338417975,
      "grad_norm": 0.5340672710813823,
      "learning_rate": 0.00012169537462602117,
      "loss": 12.0471,
      "step": 16435
    },
    {
      "epoch": 0.8950058878383805,
      "grad_norm": 0.5608441757057984,
      "learning_rate": 0.00012168676631780451,
      "loss": 12.0705,
      "step": 16436
    },
    {
      "epoch": 0.8950603418349635,
      "grad_norm": 0.5554717285872464,
      "learning_rate": 0.0001216781578409479,
      "loss": 12.3042,
      "step": 16437
    },
    {
      "epoch": 0.8951147958315465,
      "grad_norm": 0.5135038200349453,
      "learning_rate": 0.00012166954919551832,
      "loss": 12.1329,
      "step": 16438
    },
    {
      "epoch": 0.8951692498281296,
      "grad_norm": 0.5576360282006322,
      "learning_rate": 0.00012166094038158267,
      "loss": 12.2105,
      "step": 16439
    },
    {
      "epoch": 0.8952237038247126,
      "grad_norm": 0.4971445563517799,
      "learning_rate": 0.00012165233139920793,
      "loss": 12.178,
      "step": 16440
    },
    {
      "epoch": 0.8952781578212956,
      "grad_norm": 0.5787494931942184,
      "learning_rate": 0.00012164372224846106,
      "loss": 12.145,
      "step": 16441
    },
    {
      "epoch": 0.8953326118178786,
      "grad_norm": 0.5199979770663736,
      "learning_rate": 0.00012163511292940894,
      "loss": 12.0949,
      "step": 16442
    },
    {
      "epoch": 0.8953870658144616,
      "grad_norm": 0.5910601885480786,
      "learning_rate": 0.00012162650344211855,
      "loss": 12.2081,
      "step": 16443
    },
    {
      "epoch": 0.8954415198110446,
      "grad_norm": 0.5637481454779644,
      "learning_rate": 0.00012161789378665684,
      "loss": 12.1374,
      "step": 16444
    },
    {
      "epoch": 0.8954959738076277,
      "grad_norm": 0.5433839584875032,
      "learning_rate": 0.00012160928396309077,
      "loss": 12.1934,
      "step": 16445
    },
    {
      "epoch": 0.8955504278042107,
      "grad_norm": 0.5625919676665406,
      "learning_rate": 0.00012160067397148732,
      "loss": 12.0736,
      "step": 16446
    },
    {
      "epoch": 0.8956048818007937,
      "grad_norm": 0.5686657784743229,
      "learning_rate": 0.00012159206381191337,
      "loss": 12.1472,
      "step": 16447
    },
    {
      "epoch": 0.8956593357973767,
      "grad_norm": 0.5481632261766926,
      "learning_rate": 0.00012158345348443592,
      "loss": 12.1238,
      "step": 16448
    },
    {
      "epoch": 0.8957137897939597,
      "grad_norm": 0.5579295242376465,
      "learning_rate": 0.00012157484298912189,
      "loss": 12.1607,
      "step": 16449
    },
    {
      "epoch": 0.8957682437905427,
      "grad_norm": 0.5298093247807992,
      "learning_rate": 0.0001215662323260383,
      "loss": 12.2074,
      "step": 16450
    },
    {
      "epoch": 0.8958226977871258,
      "grad_norm": 0.5798119261994258,
      "learning_rate": 0.00012155762149525207,
      "loss": 12.1358,
      "step": 16451
    },
    {
      "epoch": 0.8958771517837087,
      "grad_norm": 0.5433865315844972,
      "learning_rate": 0.00012154901049683014,
      "loss": 12.0943,
      "step": 16452
    },
    {
      "epoch": 0.8959316057802917,
      "grad_norm": 0.5994413735552185,
      "learning_rate": 0.00012154039933083949,
      "loss": 12.3357,
      "step": 16453
    },
    {
      "epoch": 0.8959860597768747,
      "grad_norm": 0.5811815042877232,
      "learning_rate": 0.00012153178799734707,
      "loss": 12.1058,
      "step": 16454
    },
    {
      "epoch": 0.8960405137734577,
      "grad_norm": 0.5344823959899129,
      "learning_rate": 0.00012152317649641989,
      "loss": 12.0799,
      "step": 16455
    },
    {
      "epoch": 0.8960949677700407,
      "grad_norm": 0.6342418638772556,
      "learning_rate": 0.0001215145648281249,
      "loss": 12.1704,
      "step": 16456
    },
    {
      "epoch": 0.8961494217666238,
      "grad_norm": 0.5936470227195616,
      "learning_rate": 0.00012150595299252898,
      "loss": 12.3105,
      "step": 16457
    },
    {
      "epoch": 0.8962038757632068,
      "grad_norm": 0.4930485342447415,
      "learning_rate": 0.00012149734098969921,
      "loss": 12.0914,
      "step": 16458
    },
    {
      "epoch": 0.8962583297597898,
      "grad_norm": 0.570412094111875,
      "learning_rate": 0.00012148872881970248,
      "loss": 12.2792,
      "step": 16459
    },
    {
      "epoch": 0.8963127837563728,
      "grad_norm": 0.5553123223184898,
      "learning_rate": 0.0001214801164826058,
      "loss": 12.1716,
      "step": 16460
    },
    {
      "epoch": 0.8963672377529558,
      "grad_norm": 0.5598916626976647,
      "learning_rate": 0.00012147150397847616,
      "loss": 12.1481,
      "step": 16461
    },
    {
      "epoch": 0.8964216917495389,
      "grad_norm": 0.573612473014463,
      "learning_rate": 0.00012146289130738046,
      "loss": 12.22,
      "step": 16462
    },
    {
      "epoch": 0.8964761457461219,
      "grad_norm": 0.5584110522872364,
      "learning_rate": 0.00012145427846938575,
      "loss": 11.9865,
      "step": 16463
    },
    {
      "epoch": 0.8965305997427049,
      "grad_norm": 0.5648022930877249,
      "learning_rate": 0.00012144566546455897,
      "loss": 12.1107,
      "step": 16464
    },
    {
      "epoch": 0.8965850537392879,
      "grad_norm": 0.4750639851872308,
      "learning_rate": 0.00012143705229296707,
      "loss": 11.9892,
      "step": 16465
    },
    {
      "epoch": 0.8966395077358709,
      "grad_norm": 0.5240865844768169,
      "learning_rate": 0.00012142843895467711,
      "loss": 12.1236,
      "step": 16466
    },
    {
      "epoch": 0.8966939617324539,
      "grad_norm": 0.5603857222852854,
      "learning_rate": 0.00012141982544975596,
      "loss": 12.2482,
      "step": 16467
    },
    {
      "epoch": 0.896748415729037,
      "grad_norm": 0.5896166433903733,
      "learning_rate": 0.00012141121177827068,
      "loss": 12.2365,
      "step": 16468
    },
    {
      "epoch": 0.89680286972562,
      "grad_norm": 0.5330245279845494,
      "learning_rate": 0.00012140259794028823,
      "loss": 12.08,
      "step": 16469
    },
    {
      "epoch": 0.896857323722203,
      "grad_norm": 0.5299275708835319,
      "learning_rate": 0.0001213939839358756,
      "loss": 12.1218,
      "step": 16470
    },
    {
      "epoch": 0.8969117777187859,
      "grad_norm": 0.5266691139680327,
      "learning_rate": 0.00012138536976509973,
      "loss": 12.244,
      "step": 16471
    },
    {
      "epoch": 0.8969662317153689,
      "grad_norm": 0.5208286285047266,
      "learning_rate": 0.00012137675542802767,
      "loss": 12.1085,
      "step": 16472
    },
    {
      "epoch": 0.8970206857119519,
      "grad_norm": 0.5754574343157628,
      "learning_rate": 0.00012136814092472635,
      "loss": 12.0592,
      "step": 16473
    },
    {
      "epoch": 0.897075139708535,
      "grad_norm": 0.584767279675821,
      "learning_rate": 0.00012135952625526278,
      "loss": 12.1781,
      "step": 16474
    },
    {
      "epoch": 0.897129593705118,
      "grad_norm": 0.5304841517778917,
      "learning_rate": 0.00012135091141970399,
      "loss": 12.1799,
      "step": 16475
    },
    {
      "epoch": 0.897184047701701,
      "grad_norm": 0.5576125404035996,
      "learning_rate": 0.00012134229641811689,
      "loss": 12.3244,
      "step": 16476
    },
    {
      "epoch": 0.897238501698284,
      "grad_norm": 0.5819146834555918,
      "learning_rate": 0.00012133368125056854,
      "loss": 12.2412,
      "step": 16477
    },
    {
      "epoch": 0.897292955694867,
      "grad_norm": 0.5235517952639437,
      "learning_rate": 0.00012132506591712592,
      "loss": 12.1088,
      "step": 16478
    },
    {
      "epoch": 0.89734740969145,
      "grad_norm": 0.5092297491141731,
      "learning_rate": 0.00012131645041785598,
      "loss": 12.0861,
      "step": 16479
    },
    {
      "epoch": 0.8974018636880331,
      "grad_norm": 0.5423376099011958,
      "learning_rate": 0.00012130783475282575,
      "loss": 12.3334,
      "step": 16480
    },
    {
      "epoch": 0.8974563176846161,
      "grad_norm": 0.6165992620351156,
      "learning_rate": 0.00012129921892210223,
      "loss": 12.1034,
      "step": 16481
    },
    {
      "epoch": 0.8975107716811991,
      "grad_norm": 0.560353220103204,
      "learning_rate": 0.0001212906029257524,
      "loss": 12.171,
      "step": 16482
    },
    {
      "epoch": 0.8975652256777821,
      "grad_norm": 0.6155562428640652,
      "learning_rate": 0.0001212819867638433,
      "loss": 12.198,
      "step": 16483
    },
    {
      "epoch": 0.8976196796743651,
      "grad_norm": 0.5437538950050923,
      "learning_rate": 0.00012127337043644189,
      "loss": 12.1642,
      "step": 16484
    },
    {
      "epoch": 0.8976741336709481,
      "grad_norm": 0.6687829324129224,
      "learning_rate": 0.00012126475394361518,
      "loss": 12.1437,
      "step": 16485
    },
    {
      "epoch": 0.8977285876675312,
      "grad_norm": 0.5454772058136936,
      "learning_rate": 0.00012125613728543017,
      "loss": 12.1478,
      "step": 16486
    },
    {
      "epoch": 0.8977830416641142,
      "grad_norm": 0.5472486153196883,
      "learning_rate": 0.00012124752046195386,
      "loss": 12.1001,
      "step": 16487
    },
    {
      "epoch": 0.8978374956606971,
      "grad_norm": 0.5995583406684463,
      "learning_rate": 0.0001212389034732533,
      "loss": 12.1316,
      "step": 16488
    },
    {
      "epoch": 0.8978919496572801,
      "grad_norm": 0.6390986154680263,
      "learning_rate": 0.00012123028631939546,
      "loss": 12.2673,
      "step": 16489
    },
    {
      "epoch": 0.8979464036538631,
      "grad_norm": 0.5580090457577993,
      "learning_rate": 0.00012122166900044734,
      "loss": 12.0893,
      "step": 16490
    },
    {
      "epoch": 0.8980008576504461,
      "grad_norm": 0.6298978408215506,
      "learning_rate": 0.000121213051516476,
      "loss": 12.0943,
      "step": 16491
    },
    {
      "epoch": 0.8980553116470292,
      "grad_norm": 0.5786218651642691,
      "learning_rate": 0.00012120443386754833,
      "loss": 12.2038,
      "step": 16492
    },
    {
      "epoch": 0.8981097656436122,
      "grad_norm": 0.6447769323897492,
      "learning_rate": 0.00012119581605373149,
      "loss": 12.324,
      "step": 16493
    },
    {
      "epoch": 0.8981642196401952,
      "grad_norm": 0.5846483339186045,
      "learning_rate": 0.00012118719807509242,
      "loss": 12.0203,
      "step": 16494
    },
    {
      "epoch": 0.8982186736367782,
      "grad_norm": 0.5444446996695285,
      "learning_rate": 0.00012117857993169815,
      "loss": 12.1582,
      "step": 16495
    },
    {
      "epoch": 0.8982731276333612,
      "grad_norm": 0.5039319594598647,
      "learning_rate": 0.00012116996162361569,
      "loss": 12.1209,
      "step": 16496
    },
    {
      "epoch": 0.8983275816299443,
      "grad_norm": 0.5367761049127899,
      "learning_rate": 0.00012116134315091205,
      "loss": 12.1595,
      "step": 16497
    },
    {
      "epoch": 0.8983820356265273,
      "grad_norm": 0.5986676385179812,
      "learning_rate": 0.00012115272451365425,
      "loss": 12.1886,
      "step": 16498
    },
    {
      "epoch": 0.8984364896231103,
      "grad_norm": 0.5056536004141298,
      "learning_rate": 0.00012114410571190932,
      "loss": 12.155,
      "step": 16499
    },
    {
      "epoch": 0.8984909436196933,
      "grad_norm": 0.5538299734606021,
      "learning_rate": 0.00012113548674574428,
      "loss": 12.1304,
      "step": 16500
    },
    {
      "epoch": 0.8985453976162763,
      "grad_norm": 0.6101289756036824,
      "learning_rate": 0.00012112686761522618,
      "loss": 12.2362,
      "step": 16501
    },
    {
      "epoch": 0.8985998516128593,
      "grad_norm": 0.5493581205779898,
      "learning_rate": 0.00012111824832042198,
      "loss": 12.1193,
      "step": 16502
    },
    {
      "epoch": 0.8986543056094424,
      "grad_norm": 0.5364252298008217,
      "learning_rate": 0.00012110962886139874,
      "loss": 12.0637,
      "step": 16503
    },
    {
      "epoch": 0.8987087596060254,
      "grad_norm": 0.6534881826446343,
      "learning_rate": 0.00012110100923822347,
      "loss": 12.0993,
      "step": 16504
    },
    {
      "epoch": 0.8987632136026084,
      "grad_norm": 0.5688817236521643,
      "learning_rate": 0.00012109238945096324,
      "loss": 12.1232,
      "step": 16505
    },
    {
      "epoch": 0.8988176675991914,
      "grad_norm": 0.571168880384876,
      "learning_rate": 0.00012108376949968507,
      "loss": 12.1747,
      "step": 16506
    },
    {
      "epoch": 0.8988721215957743,
      "grad_norm": 0.581402689275133,
      "learning_rate": 0.00012107514938445597,
      "loss": 12.0934,
      "step": 16507
    },
    {
      "epoch": 0.8989265755923573,
      "grad_norm": 0.565341996031247,
      "learning_rate": 0.00012106652910534295,
      "loss": 12.1374,
      "step": 16508
    },
    {
      "epoch": 0.8989810295889404,
      "grad_norm": 0.5806175343777127,
      "learning_rate": 0.00012105790866241304,
      "loss": 12.1423,
      "step": 16509
    },
    {
      "epoch": 0.8990354835855234,
      "grad_norm": 0.6088934935079416,
      "learning_rate": 0.00012104928805573335,
      "loss": 12.1107,
      "step": 16510
    },
    {
      "epoch": 0.8990899375821064,
      "grad_norm": 0.592673300247471,
      "learning_rate": 0.00012104066728537087,
      "loss": 12.221,
      "step": 16511
    },
    {
      "epoch": 0.8991443915786894,
      "grad_norm": 0.5911083965399428,
      "learning_rate": 0.0001210320463513926,
      "loss": 12.2857,
      "step": 16512
    },
    {
      "epoch": 0.8991988455752724,
      "grad_norm": 0.6696587962542493,
      "learning_rate": 0.00012102342525386563,
      "loss": 12.2196,
      "step": 16513
    },
    {
      "epoch": 0.8992532995718554,
      "grad_norm": 0.6126665467482124,
      "learning_rate": 0.00012101480399285694,
      "loss": 12.1568,
      "step": 16514
    },
    {
      "epoch": 0.8993077535684385,
      "grad_norm": 0.5412566276103402,
      "learning_rate": 0.00012100618256843365,
      "loss": 12.1299,
      "step": 16515
    },
    {
      "epoch": 0.8993622075650215,
      "grad_norm": 0.5894568191245585,
      "learning_rate": 0.00012099756098066277,
      "loss": 12.0663,
      "step": 16516
    },
    {
      "epoch": 0.8994166615616045,
      "grad_norm": 0.5975120390893832,
      "learning_rate": 0.00012098893922961132,
      "loss": 12.2112,
      "step": 16517
    },
    {
      "epoch": 0.8994711155581875,
      "grad_norm": 0.501871125377737,
      "learning_rate": 0.00012098031731534636,
      "loss": 12.1129,
      "step": 16518
    },
    {
      "epoch": 0.8995255695547705,
      "grad_norm": 0.5719878838629991,
      "learning_rate": 0.00012097169523793492,
      "loss": 12.2414,
      "step": 16519
    },
    {
      "epoch": 0.8995800235513535,
      "grad_norm": 0.5550650218828809,
      "learning_rate": 0.00012096307299744407,
      "loss": 12.1494,
      "step": 16520
    },
    {
      "epoch": 0.8996344775479366,
      "grad_norm": 0.5500582375834041,
      "learning_rate": 0.00012095445059394086,
      "loss": 12.2454,
      "step": 16521
    },
    {
      "epoch": 0.8996889315445196,
      "grad_norm": 0.5435338973403163,
      "learning_rate": 0.00012094582802749233,
      "loss": 12.09,
      "step": 16522
    },
    {
      "epoch": 0.8997433855411026,
      "grad_norm": 0.5817020199326123,
      "learning_rate": 0.0001209372052981655,
      "loss": 12.0923,
      "step": 16523
    },
    {
      "epoch": 0.8997978395376856,
      "grad_norm": 0.5539529391293029,
      "learning_rate": 0.00012092858240602746,
      "loss": 12.1604,
      "step": 16524
    },
    {
      "epoch": 0.8998522935342685,
      "grad_norm": 0.5991940563450556,
      "learning_rate": 0.00012091995935114529,
      "loss": 12.2028,
      "step": 16525
    },
    {
      "epoch": 0.8999067475308516,
      "grad_norm": 0.5861434759316408,
      "learning_rate": 0.00012091133613358594,
      "loss": 12.2572,
      "step": 16526
    },
    {
      "epoch": 0.8999612015274346,
      "grad_norm": 0.5538040311209945,
      "learning_rate": 0.0001209027127534166,
      "loss": 12.0785,
      "step": 16527
    },
    {
      "epoch": 0.9000156555240176,
      "grad_norm": 0.5297275052982207,
      "learning_rate": 0.00012089408921070424,
      "loss": 12.1511,
      "step": 16528
    },
    {
      "epoch": 0.9000701095206006,
      "grad_norm": 0.5395278733873242,
      "learning_rate": 0.00012088546550551592,
      "loss": 12.1181,
      "step": 16529
    },
    {
      "epoch": 0.9001245635171836,
      "grad_norm": 0.5456466831033283,
      "learning_rate": 0.00012087684163791873,
      "loss": 12.1539,
      "step": 16530
    },
    {
      "epoch": 0.9001790175137666,
      "grad_norm": 0.5656408884837834,
      "learning_rate": 0.0001208682176079797,
      "loss": 12.2807,
      "step": 16531
    },
    {
      "epoch": 0.9002334715103497,
      "grad_norm": 0.543053283049204,
      "learning_rate": 0.00012085959341576596,
      "loss": 12.1513,
      "step": 16532
    },
    {
      "epoch": 0.9002879255069327,
      "grad_norm": 0.5275488355315441,
      "learning_rate": 0.00012085096906134447,
      "loss": 12.167,
      "step": 16533
    },
    {
      "epoch": 0.9003423795035157,
      "grad_norm": 0.5525419018374172,
      "learning_rate": 0.00012084234454478239,
      "loss": 12.0559,
      "step": 16534
    },
    {
      "epoch": 0.9003968335000987,
      "grad_norm": 0.5121672343176237,
      "learning_rate": 0.00012083371986614671,
      "loss": 12.1287,
      "step": 16535
    },
    {
      "epoch": 0.9004512874966817,
      "grad_norm": 0.5679055100171616,
      "learning_rate": 0.00012082509502550454,
      "loss": 12.1174,
      "step": 16536
    },
    {
      "epoch": 0.9005057414932647,
      "grad_norm": 0.6259631770417831,
      "learning_rate": 0.00012081647002292296,
      "loss": 12.1103,
      "step": 16537
    },
    {
      "epoch": 0.9005601954898478,
      "grad_norm": 0.6373837573994012,
      "learning_rate": 0.00012080784485846899,
      "loss": 12.1739,
      "step": 16538
    },
    {
      "epoch": 0.9006146494864308,
      "grad_norm": 0.5627806904812449,
      "learning_rate": 0.00012079921953220975,
      "loss": 12.1975,
      "step": 16539
    },
    {
      "epoch": 0.9006691034830138,
      "grad_norm": 0.5666742303334366,
      "learning_rate": 0.00012079059404421227,
      "loss": 12.1863,
      "step": 16540
    },
    {
      "epoch": 0.9007235574795968,
      "grad_norm": 0.5884420986046116,
      "learning_rate": 0.00012078196839454365,
      "loss": 12.257,
      "step": 16541
    },
    {
      "epoch": 0.9007780114761798,
      "grad_norm": 0.6108872627903197,
      "learning_rate": 0.00012077334258327097,
      "loss": 12.1131,
      "step": 16542
    },
    {
      "epoch": 0.9008324654727627,
      "grad_norm": 0.6058293067479195,
      "learning_rate": 0.00012076471661046129,
      "loss": 12.146,
      "step": 16543
    },
    {
      "epoch": 0.9008869194693458,
      "grad_norm": 0.5609878745463391,
      "learning_rate": 0.00012075609047618169,
      "loss": 12.2037,
      "step": 16544
    },
    {
      "epoch": 0.9009413734659288,
      "grad_norm": 0.5445608032269247,
      "learning_rate": 0.00012074746418049924,
      "loss": 12.1039,
      "step": 16545
    },
    {
      "epoch": 0.9009958274625118,
      "grad_norm": 0.5625864225762172,
      "learning_rate": 0.00012073883772348105,
      "loss": 12.2862,
      "step": 16546
    },
    {
      "epoch": 0.9010502814590948,
      "grad_norm": 0.5848363906000628,
      "learning_rate": 0.00012073021110519416,
      "loss": 12.1611,
      "step": 16547
    },
    {
      "epoch": 0.9011047354556778,
      "grad_norm": 0.5851829608040879,
      "learning_rate": 0.00012072158432570569,
      "loss": 12.2337,
      "step": 16548
    },
    {
      "epoch": 0.9011591894522608,
      "grad_norm": 0.5248127918593556,
      "learning_rate": 0.00012071295738508268,
      "loss": 12.2178,
      "step": 16549
    },
    {
      "epoch": 0.9012136434488439,
      "grad_norm": 0.5568758893873234,
      "learning_rate": 0.00012070433028339226,
      "loss": 12.1396,
      "step": 16550
    },
    {
      "epoch": 0.9012680974454269,
      "grad_norm": 0.5732195807465644,
      "learning_rate": 0.00012069570302070148,
      "loss": 12.1719,
      "step": 16551
    },
    {
      "epoch": 0.9013225514420099,
      "grad_norm": 0.6810937493211286,
      "learning_rate": 0.00012068707559707746,
      "loss": 12.2129,
      "step": 16552
    },
    {
      "epoch": 0.9013770054385929,
      "grad_norm": 0.5264730181108157,
      "learning_rate": 0.00012067844801258726,
      "loss": 12.1977,
      "step": 16553
    },
    {
      "epoch": 0.9014314594351759,
      "grad_norm": 0.5334773683000122,
      "learning_rate": 0.00012066982026729798,
      "loss": 12.1088,
      "step": 16554
    },
    {
      "epoch": 0.9014859134317589,
      "grad_norm": 0.6875605989114848,
      "learning_rate": 0.0001206611923612767,
      "loss": 12.0489,
      "step": 16555
    },
    {
      "epoch": 0.901540367428342,
      "grad_norm": 0.6142806055177614,
      "learning_rate": 0.00012065256429459056,
      "loss": 12.2554,
      "step": 16556
    },
    {
      "epoch": 0.901594821424925,
      "grad_norm": 0.5594151040235987,
      "learning_rate": 0.00012064393606730662,
      "loss": 12.1569,
      "step": 16557
    },
    {
      "epoch": 0.901649275421508,
      "grad_norm": 0.5922801599066374,
      "learning_rate": 0.00012063530767949191,
      "loss": 12.1549,
      "step": 16558
    },
    {
      "epoch": 0.901703729418091,
      "grad_norm": 0.6089845067197045,
      "learning_rate": 0.00012062667913121363,
      "loss": 12.2396,
      "step": 16559
    },
    {
      "epoch": 0.901758183414674,
      "grad_norm": 0.5340371089060799,
      "learning_rate": 0.00012061805042253881,
      "loss": 11.9331,
      "step": 16560
    },
    {
      "epoch": 0.9018126374112571,
      "grad_norm": 0.55790354309246,
      "learning_rate": 0.0001206094215535346,
      "loss": 12.0415,
      "step": 16561
    },
    {
      "epoch": 0.90186709140784,
      "grad_norm": 0.6135537059450782,
      "learning_rate": 0.00012060079252426809,
      "loss": 12.1952,
      "step": 16562
    },
    {
      "epoch": 0.901921545404423,
      "grad_norm": 0.5821693916055684,
      "learning_rate": 0.00012059216333480632,
      "loss": 12.2118,
      "step": 16563
    },
    {
      "epoch": 0.901975999401006,
      "grad_norm": 0.5744206999331853,
      "learning_rate": 0.00012058353398521644,
      "loss": 12.2648,
      "step": 16564
    },
    {
      "epoch": 0.902030453397589,
      "grad_norm": 0.6871102465432373,
      "learning_rate": 0.00012057490447556556,
      "loss": 12.1974,
      "step": 16565
    },
    {
      "epoch": 0.902084907394172,
      "grad_norm": 0.5584545841478958,
      "learning_rate": 0.00012056627480592077,
      "loss": 12.1063,
      "step": 16566
    },
    {
      "epoch": 0.9021393613907551,
      "grad_norm": 0.5289270437003082,
      "learning_rate": 0.0001205576449763492,
      "loss": 12.0957,
      "step": 16567
    },
    {
      "epoch": 0.9021938153873381,
      "grad_norm": 0.613888501086125,
      "learning_rate": 0.0001205490149869179,
      "loss": 12.1563,
      "step": 16568
    },
    {
      "epoch": 0.9022482693839211,
      "grad_norm": 0.6974437227061876,
      "learning_rate": 0.00012054038483769401,
      "loss": 12.1972,
      "step": 16569
    },
    {
      "epoch": 0.9023027233805041,
      "grad_norm": 0.5782039366655223,
      "learning_rate": 0.00012053175452874466,
      "loss": 12.1505,
      "step": 16570
    },
    {
      "epoch": 0.9023571773770871,
      "grad_norm": 0.5986953500684845,
      "learning_rate": 0.00012052312406013694,
      "loss": 12.0572,
      "step": 16571
    },
    {
      "epoch": 0.9024116313736701,
      "grad_norm": 0.5781745381427984,
      "learning_rate": 0.00012051449343193799,
      "loss": 12.1791,
      "step": 16572
    },
    {
      "epoch": 0.9024660853702532,
      "grad_norm": 0.580489142635654,
      "learning_rate": 0.00012050586264421489,
      "loss": 12.1939,
      "step": 16573
    },
    {
      "epoch": 0.9025205393668362,
      "grad_norm": 0.5888350093249815,
      "learning_rate": 0.00012049723169703474,
      "loss": 12.1028,
      "step": 16574
    },
    {
      "epoch": 0.9025749933634192,
      "grad_norm": 0.6599516366929871,
      "learning_rate": 0.00012048860059046468,
      "loss": 12.33,
      "step": 16575
    },
    {
      "epoch": 0.9026294473600022,
      "grad_norm": 0.6568993104827798,
      "learning_rate": 0.00012047996932457182,
      "loss": 12.3188,
      "step": 16576
    },
    {
      "epoch": 0.9026839013565852,
      "grad_norm": 0.5526626692920116,
      "learning_rate": 0.0001204713378994233,
      "loss": 12.1452,
      "step": 16577
    },
    {
      "epoch": 0.9027383553531682,
      "grad_norm": 0.5411002258022113,
      "learning_rate": 0.00012046270631508623,
      "loss": 12.2017,
      "step": 16578
    },
    {
      "epoch": 0.9027928093497513,
      "grad_norm": 0.5304822759479126,
      "learning_rate": 0.00012045407457162772,
      "loss": 11.9785,
      "step": 16579
    },
    {
      "epoch": 0.9028472633463343,
      "grad_norm": 0.668688733482345,
      "learning_rate": 0.00012044544266911488,
      "loss": 12.1479,
      "step": 16580
    },
    {
      "epoch": 0.9029017173429172,
      "grad_norm": 0.5969611170805564,
      "learning_rate": 0.00012043681060761484,
      "loss": 12.1249,
      "step": 16581
    },
    {
      "epoch": 0.9029561713395002,
      "grad_norm": 0.6044686527675578,
      "learning_rate": 0.0001204281783871948,
      "loss": 12.2064,
      "step": 16582
    },
    {
      "epoch": 0.9030106253360832,
      "grad_norm": 0.577749237566279,
      "learning_rate": 0.00012041954600792175,
      "loss": 12.1072,
      "step": 16583
    },
    {
      "epoch": 0.9030650793326662,
      "grad_norm": 0.5766044759471065,
      "learning_rate": 0.00012041091346986292,
      "loss": 12.2084,
      "step": 16584
    },
    {
      "epoch": 0.9031195333292493,
      "grad_norm": 0.5806415623737022,
      "learning_rate": 0.0001204022807730854,
      "loss": 12.1572,
      "step": 16585
    },
    {
      "epoch": 0.9031739873258323,
      "grad_norm": 0.5656482861317721,
      "learning_rate": 0.0001203936479176563,
      "loss": 12.1701,
      "step": 16586
    },
    {
      "epoch": 0.9032284413224153,
      "grad_norm": 0.619887669845093,
      "learning_rate": 0.00012038501490364281,
      "loss": 12.19,
      "step": 16587
    },
    {
      "epoch": 0.9032828953189983,
      "grad_norm": 0.595453210340934,
      "learning_rate": 0.00012037638173111201,
      "loss": 12.0868,
      "step": 16588
    },
    {
      "epoch": 0.9033373493155813,
      "grad_norm": 0.5307560222648793,
      "learning_rate": 0.00012036774840013103,
      "loss": 12.0417,
      "step": 16589
    },
    {
      "epoch": 0.9033918033121643,
      "grad_norm": 0.5689723062921288,
      "learning_rate": 0.00012035911491076704,
      "loss": 12.0614,
      "step": 16590
    },
    {
      "epoch": 0.9034462573087474,
      "grad_norm": 0.5369195786912859,
      "learning_rate": 0.00012035048126308715,
      "loss": 12.1579,
      "step": 16591
    },
    {
      "epoch": 0.9035007113053304,
      "grad_norm": 0.528098688405729,
      "learning_rate": 0.00012034184745715853,
      "loss": 12.1516,
      "step": 16592
    },
    {
      "epoch": 0.9035551653019134,
      "grad_norm": 0.6048731746648017,
      "learning_rate": 0.00012033321349304827,
      "loss": 12.1478,
      "step": 16593
    },
    {
      "epoch": 0.9036096192984964,
      "grad_norm": 0.5574691218373621,
      "learning_rate": 0.00012032457937082353,
      "loss": 12.1007,
      "step": 16594
    },
    {
      "epoch": 0.9036640732950794,
      "grad_norm": 0.5729354710039366,
      "learning_rate": 0.00012031594509055146,
      "loss": 12.2509,
      "step": 16595
    },
    {
      "epoch": 0.9037185272916625,
      "grad_norm": 0.5898254632000199,
      "learning_rate": 0.00012030731065229918,
      "loss": 12.178,
      "step": 16596
    },
    {
      "epoch": 0.9037729812882455,
      "grad_norm": 0.5430476778389886,
      "learning_rate": 0.00012029867605613385,
      "loss": 12.1013,
      "step": 16597
    },
    {
      "epoch": 0.9038274352848285,
      "grad_norm": 0.5719532844173403,
      "learning_rate": 0.00012029004130212263,
      "loss": 12.2637,
      "step": 16598
    },
    {
      "epoch": 0.9038818892814114,
      "grad_norm": 0.5673084511445369,
      "learning_rate": 0.00012028140639033262,
      "loss": 12.2498,
      "step": 16599
    },
    {
      "epoch": 0.9039363432779944,
      "grad_norm": 0.5575708863107153,
      "learning_rate": 0.00012027277132083103,
      "loss": 12.0741,
      "step": 16600
    },
    {
      "epoch": 0.9039907972745774,
      "grad_norm": 0.5685739983488483,
      "learning_rate": 0.00012026413609368495,
      "loss": 12.1816,
      "step": 16601
    },
    {
      "epoch": 0.9040452512711605,
      "grad_norm": 0.581791608106945,
      "learning_rate": 0.00012025550070896155,
      "loss": 12.1209,
      "step": 16602
    },
    {
      "epoch": 0.9040997052677435,
      "grad_norm": 0.5168490029790749,
      "learning_rate": 0.00012024686516672796,
      "loss": 12.1303,
      "step": 16603
    },
    {
      "epoch": 0.9041541592643265,
      "grad_norm": 0.5735573094212707,
      "learning_rate": 0.0001202382294670514,
      "loss": 12.1718,
      "step": 16604
    },
    {
      "epoch": 0.9042086132609095,
      "grad_norm": 0.5851835687748101,
      "learning_rate": 0.00012022959360999893,
      "loss": 12.1299,
      "step": 16605
    },
    {
      "epoch": 0.9042630672574925,
      "grad_norm": 0.5561947463654558,
      "learning_rate": 0.00012022095759563777,
      "loss": 12.0567,
      "step": 16606
    },
    {
      "epoch": 0.9043175212540755,
      "grad_norm": 0.57096710325472,
      "learning_rate": 0.00012021232142403502,
      "loss": 12.1633,
      "step": 16607
    },
    {
      "epoch": 0.9043719752506586,
      "grad_norm": 0.6064023929898223,
      "learning_rate": 0.0001202036850952579,
      "loss": 12.195,
      "step": 16608
    },
    {
      "epoch": 0.9044264292472416,
      "grad_norm": 0.7204348761597378,
      "learning_rate": 0.00012019504860937352,
      "loss": 12.2921,
      "step": 16609
    },
    {
      "epoch": 0.9044808832438246,
      "grad_norm": 0.5646044232101393,
      "learning_rate": 0.00012018641196644906,
      "loss": 12.1701,
      "step": 16610
    },
    {
      "epoch": 0.9045353372404076,
      "grad_norm": 0.5274641380881213,
      "learning_rate": 0.00012017777516655169,
      "loss": 12.06,
      "step": 16611
    },
    {
      "epoch": 0.9045897912369906,
      "grad_norm": 0.5251427691853415,
      "learning_rate": 0.00012016913820974856,
      "loss": 12.2064,
      "step": 16612
    },
    {
      "epoch": 0.9046442452335736,
      "grad_norm": 0.6021069458024625,
      "learning_rate": 0.00012016050109610679,
      "loss": 12.111,
      "step": 16613
    },
    {
      "epoch": 0.9046986992301567,
      "grad_norm": 0.5338626495772549,
      "learning_rate": 0.00012015186382569362,
      "loss": 12.1724,
      "step": 16614
    },
    {
      "epoch": 0.9047531532267397,
      "grad_norm": 0.5322612991595553,
      "learning_rate": 0.00012014322639857616,
      "loss": 12.1126,
      "step": 16615
    },
    {
      "epoch": 0.9048076072233227,
      "grad_norm": 0.596317737541182,
      "learning_rate": 0.0001201345888148216,
      "loss": 12.1244,
      "step": 16616
    },
    {
      "epoch": 0.9048620612199056,
      "grad_norm": 0.5481072720850805,
      "learning_rate": 0.00012012595107449712,
      "loss": 12.2287,
      "step": 16617
    },
    {
      "epoch": 0.9049165152164886,
      "grad_norm": 0.5679762111084247,
      "learning_rate": 0.00012011731317766983,
      "loss": 12.1883,
      "step": 16618
    },
    {
      "epoch": 0.9049709692130716,
      "grad_norm": 0.5880208205737794,
      "learning_rate": 0.00012010867512440695,
      "loss": 12.1742,
      "step": 16619
    },
    {
      "epoch": 0.9050254232096547,
      "grad_norm": 0.6104970007982246,
      "learning_rate": 0.00012010003691477564,
      "loss": 12.2109,
      "step": 16620
    },
    {
      "epoch": 0.9050798772062377,
      "grad_norm": 0.5823536654745095,
      "learning_rate": 0.00012009139854884308,
      "loss": 12.1697,
      "step": 16621
    },
    {
      "epoch": 0.9051343312028207,
      "grad_norm": 0.6349717560007302,
      "learning_rate": 0.00012008276002667646,
      "loss": 12.0318,
      "step": 16622
    },
    {
      "epoch": 0.9051887851994037,
      "grad_norm": 0.5989307276431446,
      "learning_rate": 0.0001200741213483429,
      "loss": 12.0962,
      "step": 16623
    },
    {
      "epoch": 0.9052432391959867,
      "grad_norm": 0.6016610302270592,
      "learning_rate": 0.00012006548251390959,
      "loss": 12.1917,
      "step": 16624
    },
    {
      "epoch": 0.9052976931925697,
      "grad_norm": 0.5617629352693115,
      "learning_rate": 0.00012005684352344375,
      "loss": 12.0168,
      "step": 16625
    },
    {
      "epoch": 0.9053521471891528,
      "grad_norm": 0.7072618121372366,
      "learning_rate": 0.00012004820437701252,
      "loss": 12.1314,
      "step": 16626
    },
    {
      "epoch": 0.9054066011857358,
      "grad_norm": 0.8022261000813399,
      "learning_rate": 0.00012003956507468312,
      "loss": 12.1558,
      "step": 16627
    },
    {
      "epoch": 0.9054610551823188,
      "grad_norm": 0.6477531181079941,
      "learning_rate": 0.00012003092561652267,
      "loss": 12.184,
      "step": 16628
    },
    {
      "epoch": 0.9055155091789018,
      "grad_norm": 0.565053757093744,
      "learning_rate": 0.00012002228600259838,
      "loss": 12.2107,
      "step": 16629
    },
    {
      "epoch": 0.9055699631754848,
      "grad_norm": 0.6237518603632382,
      "learning_rate": 0.00012001364623297744,
      "loss": 12.045,
      "step": 16630
    },
    {
      "epoch": 0.9056244171720679,
      "grad_norm": 0.6334450869783368,
      "learning_rate": 0.00012000500630772705,
      "loss": 12.1567,
      "step": 16631
    },
    {
      "epoch": 0.9056788711686509,
      "grad_norm": 0.5331753187207506,
      "learning_rate": 0.00011999636622691438,
      "loss": 12.1466,
      "step": 16632
    },
    {
      "epoch": 0.9057333251652339,
      "grad_norm": 0.5556861753442944,
      "learning_rate": 0.0001199877259906066,
      "loss": 12.2399,
      "step": 16633
    },
    {
      "epoch": 0.9057877791618169,
      "grad_norm": 0.6053269076177362,
      "learning_rate": 0.0001199790855988709,
      "loss": 12.2552,
      "step": 16634
    },
    {
      "epoch": 0.9058422331583998,
      "grad_norm": 0.5511390999628655,
      "learning_rate": 0.00011997044505177446,
      "loss": 12.2137,
      "step": 16635
    },
    {
      "epoch": 0.9058966871549828,
      "grad_norm": 0.5596620142884977,
      "learning_rate": 0.00011996180434938451,
      "loss": 12.0874,
      "step": 16636
    },
    {
      "epoch": 0.9059511411515659,
      "grad_norm": 0.7251786263883778,
      "learning_rate": 0.00011995316349176825,
      "loss": 12.1136,
      "step": 16637
    },
    {
      "epoch": 0.9060055951481489,
      "grad_norm": 0.5735425418309598,
      "learning_rate": 0.00011994452247899284,
      "loss": 12.1922,
      "step": 16638
    },
    {
      "epoch": 0.9060600491447319,
      "grad_norm": 0.5538004555453727,
      "learning_rate": 0.00011993588131112545,
      "loss": 12.2168,
      "step": 16639
    },
    {
      "epoch": 0.9061145031413149,
      "grad_norm": 0.6079380700170678,
      "learning_rate": 0.0001199272399882333,
      "loss": 12.1329,
      "step": 16640
    },
    {
      "epoch": 0.9061689571378979,
      "grad_norm": 0.617454855656419,
      "learning_rate": 0.0001199185985103836,
      "loss": 12.144,
      "step": 16641
    },
    {
      "epoch": 0.9062234111344809,
      "grad_norm": 0.5169310498004772,
      "learning_rate": 0.00011990995687764357,
      "loss": 12.2623,
      "step": 16642
    },
    {
      "epoch": 0.906277865131064,
      "grad_norm": 0.6177406350488743,
      "learning_rate": 0.00011990131509008036,
      "loss": 12.12,
      "step": 16643
    },
    {
      "epoch": 0.906332319127647,
      "grad_norm": 0.6277145773760757,
      "learning_rate": 0.00011989267314776116,
      "loss": 12.2011,
      "step": 16644
    },
    {
      "epoch": 0.90638677312423,
      "grad_norm": 0.5425803391625272,
      "learning_rate": 0.00011988403105075323,
      "loss": 12.0853,
      "step": 16645
    },
    {
      "epoch": 0.906441227120813,
      "grad_norm": 0.5910161248310666,
      "learning_rate": 0.0001198753887991237,
      "loss": 12.2245,
      "step": 16646
    },
    {
      "epoch": 0.906495681117396,
      "grad_norm": 0.5450148777653727,
      "learning_rate": 0.00011986674639293984,
      "loss": 12.2561,
      "step": 16647
    },
    {
      "epoch": 0.906550135113979,
      "grad_norm": 0.6509899028924494,
      "learning_rate": 0.00011985810383226884,
      "loss": 12.3318,
      "step": 16648
    },
    {
      "epoch": 0.9066045891105621,
      "grad_norm": 0.5842785402708647,
      "learning_rate": 0.00011984946111717787,
      "loss": 12.1969,
      "step": 16649
    },
    {
      "epoch": 0.9066590431071451,
      "grad_norm": 0.5736884415395951,
      "learning_rate": 0.00011984081824773418,
      "loss": 12.2531,
      "step": 16650
    },
    {
      "epoch": 0.9067134971037281,
      "grad_norm": 0.6329788568815561,
      "learning_rate": 0.00011983217522400494,
      "loss": 12.2265,
      "step": 16651
    },
    {
      "epoch": 0.906767951100311,
      "grad_norm": 0.5200775966974495,
      "learning_rate": 0.0001198235320460574,
      "loss": 12.1019,
      "step": 16652
    },
    {
      "epoch": 0.906822405096894,
      "grad_norm": 0.6080449242599029,
      "learning_rate": 0.00011981488871395874,
      "loss": 12.2413,
      "step": 16653
    },
    {
      "epoch": 0.906876859093477,
      "grad_norm": 0.5179714179732703,
      "learning_rate": 0.0001198062452277762,
      "loss": 12.2128,
      "step": 16654
    },
    {
      "epoch": 0.9069313130900601,
      "grad_norm": 0.5315355453059781,
      "learning_rate": 0.00011979760158757693,
      "loss": 12.1813,
      "step": 16655
    },
    {
      "epoch": 0.9069857670866431,
      "grad_norm": 0.540844043152265,
      "learning_rate": 0.00011978895779342823,
      "loss": 12.0854,
      "step": 16656
    },
    {
      "epoch": 0.9070402210832261,
      "grad_norm": 0.5503422274480156,
      "learning_rate": 0.00011978031384539727,
      "loss": 12.2659,
      "step": 16657
    },
    {
      "epoch": 0.9070946750798091,
      "grad_norm": 0.5766267629462053,
      "learning_rate": 0.00011977166974355127,
      "loss": 12.252,
      "step": 16658
    },
    {
      "epoch": 0.9071491290763921,
      "grad_norm": 0.5981171855869646,
      "learning_rate": 0.00011976302548795746,
      "loss": 12.1501,
      "step": 16659
    },
    {
      "epoch": 0.9072035830729752,
      "grad_norm": 0.6071357970665403,
      "learning_rate": 0.00011975438107868302,
      "loss": 12.237,
      "step": 16660
    },
    {
      "epoch": 0.9072580370695582,
      "grad_norm": 0.5337519119985165,
      "learning_rate": 0.00011974573651579521,
      "loss": 12.0513,
      "step": 16661
    },
    {
      "epoch": 0.9073124910661412,
      "grad_norm": 0.6093536507145341,
      "learning_rate": 0.00011973709179936125,
      "loss": 12.2212,
      "step": 16662
    },
    {
      "epoch": 0.9073669450627242,
      "grad_norm": 0.5361183955849707,
      "learning_rate": 0.00011972844692944835,
      "loss": 11.9772,
      "step": 16663
    },
    {
      "epoch": 0.9074213990593072,
      "grad_norm": 0.6072014184975755,
      "learning_rate": 0.00011971980190612375,
      "loss": 12.0981,
      "step": 16664
    },
    {
      "epoch": 0.9074758530558902,
      "grad_norm": 0.5604623769710646,
      "learning_rate": 0.00011971115672945466,
      "loss": 12.2179,
      "step": 16665
    },
    {
      "epoch": 0.9075303070524733,
      "grad_norm": 0.6645669085703189,
      "learning_rate": 0.0001197025113995083,
      "loss": 12.3859,
      "step": 16666
    },
    {
      "epoch": 0.9075847610490563,
      "grad_norm": 0.6003976294910272,
      "learning_rate": 0.00011969386591635192,
      "loss": 12.164,
      "step": 16667
    },
    {
      "epoch": 0.9076392150456393,
      "grad_norm": 0.6585124722512923,
      "learning_rate": 0.00011968522028005273,
      "loss": 12.2127,
      "step": 16668
    },
    {
      "epoch": 0.9076936690422223,
      "grad_norm": 0.5535103951097059,
      "learning_rate": 0.00011967657449067795,
      "loss": 12.1997,
      "step": 16669
    },
    {
      "epoch": 0.9077481230388053,
      "grad_norm": 0.541805955353317,
      "learning_rate": 0.00011966792854829485,
      "loss": 12.0348,
      "step": 16670
    },
    {
      "epoch": 0.9078025770353882,
      "grad_norm": 0.5215776723442405,
      "learning_rate": 0.00011965928245297063,
      "loss": 12.1675,
      "step": 16671
    },
    {
      "epoch": 0.9078570310319714,
      "grad_norm": 0.5248796095810393,
      "learning_rate": 0.00011965063620477252,
      "loss": 12.0794,
      "step": 16672
    },
    {
      "epoch": 0.9079114850285543,
      "grad_norm": 0.48543629832874036,
      "learning_rate": 0.0001196419898037678,
      "loss": 12.0909,
      "step": 16673
    },
    {
      "epoch": 0.9079659390251373,
      "grad_norm": 0.5810116401522981,
      "learning_rate": 0.00011963334325002364,
      "loss": 12.1623,
      "step": 16674
    },
    {
      "epoch": 0.9080203930217203,
      "grad_norm": 0.5140735331995465,
      "learning_rate": 0.00011962469654360733,
      "loss": 12.0415,
      "step": 16675
    },
    {
      "epoch": 0.9080748470183033,
      "grad_norm": 0.5242609749151819,
      "learning_rate": 0.00011961604968458609,
      "loss": 12.1652,
      "step": 16676
    },
    {
      "epoch": 0.9081293010148863,
      "grad_norm": 0.5012690451280981,
      "learning_rate": 0.00011960740267302715,
      "loss": 12.0782,
      "step": 16677
    },
    {
      "epoch": 0.9081837550114694,
      "grad_norm": 0.5704474808309116,
      "learning_rate": 0.00011959875550899775,
      "loss": 12.2232,
      "step": 16678
    },
    {
      "epoch": 0.9082382090080524,
      "grad_norm": 0.5242457472681777,
      "learning_rate": 0.00011959010819256515,
      "loss": 12.1655,
      "step": 16679
    },
    {
      "epoch": 0.9082926630046354,
      "grad_norm": 0.5767583316134496,
      "learning_rate": 0.00011958146072379659,
      "loss": 12.0799,
      "step": 16680
    },
    {
      "epoch": 0.9083471170012184,
      "grad_norm": 0.612337674023645,
      "learning_rate": 0.00011957281310275929,
      "loss": 12.1277,
      "step": 16681
    },
    {
      "epoch": 0.9084015709978014,
      "grad_norm": 0.5639431431197535,
      "learning_rate": 0.00011956416532952052,
      "loss": 12.1048,
      "step": 16682
    },
    {
      "epoch": 0.9084560249943844,
      "grad_norm": 0.5781707248709169,
      "learning_rate": 0.00011955551740414754,
      "loss": 12.0985,
      "step": 16683
    },
    {
      "epoch": 0.9085104789909675,
      "grad_norm": 0.5415750537886928,
      "learning_rate": 0.00011954686932670755,
      "loss": 12.0249,
      "step": 16684
    },
    {
      "epoch": 0.9085649329875505,
      "grad_norm": 0.5771376377405151,
      "learning_rate": 0.00011953822109726785,
      "loss": 12.1427,
      "step": 16685
    },
    {
      "epoch": 0.9086193869841335,
      "grad_norm": 0.5812078814955836,
      "learning_rate": 0.00011952957271589565,
      "loss": 12.2247,
      "step": 16686
    },
    {
      "epoch": 0.9086738409807165,
      "grad_norm": 0.5473565554611068,
      "learning_rate": 0.00011952092418265821,
      "loss": 12.0588,
      "step": 16687
    },
    {
      "epoch": 0.9087282949772995,
      "grad_norm": 0.5833788408235893,
      "learning_rate": 0.00011951227549762283,
      "loss": 12.102,
      "step": 16688
    },
    {
      "epoch": 0.9087827489738824,
      "grad_norm": 0.5727748231134404,
      "learning_rate": 0.00011950362666085665,
      "loss": 12.1467,
      "step": 16689
    },
    {
      "epoch": 0.9088372029704656,
      "grad_norm": 0.5539800357546315,
      "learning_rate": 0.00011949497767242706,
      "loss": 12.1409,
      "step": 16690
    },
    {
      "epoch": 0.9088916569670485,
      "grad_norm": 0.5645437034529261,
      "learning_rate": 0.00011948632853240122,
      "loss": 12.0354,
      "step": 16691
    },
    {
      "epoch": 0.9089461109636315,
      "grad_norm": 0.605985359763697,
      "learning_rate": 0.00011947767924084645,
      "loss": 12.1695,
      "step": 16692
    },
    {
      "epoch": 0.9090005649602145,
      "grad_norm": 0.5683245753476472,
      "learning_rate": 0.00011946902979782999,
      "loss": 12.0926,
      "step": 16693
    },
    {
      "epoch": 0.9090550189567975,
      "grad_norm": 0.5298631423959123,
      "learning_rate": 0.00011946038020341905,
      "loss": 12.2171,
      "step": 16694
    },
    {
      "epoch": 0.9091094729533806,
      "grad_norm": 0.6042090785026276,
      "learning_rate": 0.00011945173045768095,
      "loss": 12.2209,
      "step": 16695
    },
    {
      "epoch": 0.9091639269499636,
      "grad_norm": 0.5521047723399338,
      "learning_rate": 0.00011944308056068292,
      "loss": 12.0592,
      "step": 16696
    },
    {
      "epoch": 0.9092183809465466,
      "grad_norm": 0.5184238358553434,
      "learning_rate": 0.00011943443051249224,
      "loss": 12.1804,
      "step": 16697
    },
    {
      "epoch": 0.9092728349431296,
      "grad_norm": 0.6132916575172702,
      "learning_rate": 0.00011942578031317619,
      "loss": 12.2734,
      "step": 16698
    },
    {
      "epoch": 0.9093272889397126,
      "grad_norm": 0.5377853963467617,
      "learning_rate": 0.00011941712996280201,
      "loss": 12.0987,
      "step": 16699
    },
    {
      "epoch": 0.9093817429362956,
      "grad_norm": 0.4954452948200757,
      "learning_rate": 0.00011940847946143696,
      "loss": 12.1779,
      "step": 16700
    },
    {
      "epoch": 0.9094361969328787,
      "grad_norm": 0.5452706709754639,
      "learning_rate": 0.00011939982880914828,
      "loss": 12.1383,
      "step": 16701
    },
    {
      "epoch": 0.9094906509294617,
      "grad_norm": 0.5960469894657686,
      "learning_rate": 0.00011939117800600333,
      "loss": 12.1946,
      "step": 16702
    },
    {
      "epoch": 0.9095451049260447,
      "grad_norm": 0.6250770426978768,
      "learning_rate": 0.00011938252705206934,
      "loss": 12.12,
      "step": 16703
    },
    {
      "epoch": 0.9095995589226277,
      "grad_norm": 0.5619249572154056,
      "learning_rate": 0.00011937387594741353,
      "loss": 12.1961,
      "step": 16704
    },
    {
      "epoch": 0.9096540129192107,
      "grad_norm": 0.5960458283295135,
      "learning_rate": 0.00011936522469210323,
      "loss": 12.1709,
      "step": 16705
    },
    {
      "epoch": 0.9097084669157937,
      "grad_norm": 0.545670326510443,
      "learning_rate": 0.00011935657328620566,
      "loss": 12.0751,
      "step": 16706
    },
    {
      "epoch": 0.9097629209123768,
      "grad_norm": 0.5305274367137507,
      "learning_rate": 0.00011934792172978815,
      "loss": 12.1436,
      "step": 16707
    },
    {
      "epoch": 0.9098173749089598,
      "grad_norm": 0.5447327452757997,
      "learning_rate": 0.00011933927002291801,
      "loss": 12.0792,
      "step": 16708
    },
    {
      "epoch": 0.9098718289055427,
      "grad_norm": 0.6511668231413018,
      "learning_rate": 0.0001193306181656624,
      "loss": 12.1077,
      "step": 16709
    },
    {
      "epoch": 0.9099262829021257,
      "grad_norm": 0.5451503446692807,
      "learning_rate": 0.00011932196615808868,
      "loss": 12.1108,
      "step": 16710
    },
    {
      "epoch": 0.9099807368987087,
      "grad_norm": 0.5807730649400389,
      "learning_rate": 0.0001193133140002641,
      "loss": 12.246,
      "step": 16711
    },
    {
      "epoch": 0.9100351908952917,
      "grad_norm": 0.5497627488136255,
      "learning_rate": 0.00011930466169225595,
      "loss": 12.1494,
      "step": 16712
    },
    {
      "epoch": 0.9100896448918748,
      "grad_norm": 0.568200531087904,
      "learning_rate": 0.00011929600923413156,
      "loss": 12.1344,
      "step": 16713
    },
    {
      "epoch": 0.9101440988884578,
      "grad_norm": 0.5289936895464843,
      "learning_rate": 0.00011928735662595812,
      "loss": 12.1177,
      "step": 16714
    },
    {
      "epoch": 0.9101985528850408,
      "grad_norm": 0.5598408108034993,
      "learning_rate": 0.00011927870386780298,
      "loss": 12.1175,
      "step": 16715
    },
    {
      "epoch": 0.9102530068816238,
      "grad_norm": 0.6177995714639638,
      "learning_rate": 0.00011927005095973341,
      "loss": 12.1444,
      "step": 16716
    },
    {
      "epoch": 0.9103074608782068,
      "grad_norm": 0.6274524773670057,
      "learning_rate": 0.00011926139790181663,
      "loss": 12.2214,
      "step": 16717
    },
    {
      "epoch": 0.9103619148747898,
      "grad_norm": 0.5085526453142428,
      "learning_rate": 0.00011925274469412007,
      "loss": 12.2215,
      "step": 16718
    },
    {
      "epoch": 0.9104163688713729,
      "grad_norm": 0.5974215900567931,
      "learning_rate": 0.00011924409133671091,
      "loss": 12.123,
      "step": 16719
    },
    {
      "epoch": 0.9104708228679559,
      "grad_norm": 0.5536194685243843,
      "learning_rate": 0.00011923543782965647,
      "loss": 12.1997,
      "step": 16720
    },
    {
      "epoch": 0.9105252768645389,
      "grad_norm": 0.548810186607552,
      "learning_rate": 0.00011922678417302404,
      "loss": 12.1594,
      "step": 16721
    },
    {
      "epoch": 0.9105797308611219,
      "grad_norm": 0.5358261416721035,
      "learning_rate": 0.0001192181303668809,
      "loss": 12.1352,
      "step": 16722
    },
    {
      "epoch": 0.9106341848577049,
      "grad_norm": 0.5869635801603279,
      "learning_rate": 0.00011920947641129437,
      "loss": 12.1825,
      "step": 16723
    },
    {
      "epoch": 0.9106886388542879,
      "grad_norm": 0.5178457255217145,
      "learning_rate": 0.00011920082230633172,
      "loss": 12.1535,
      "step": 16724
    },
    {
      "epoch": 0.910743092850871,
      "grad_norm": 0.5471474478485095,
      "learning_rate": 0.00011919216805206026,
      "loss": 12.0578,
      "step": 16725
    },
    {
      "epoch": 0.910797546847454,
      "grad_norm": 0.5618150486227647,
      "learning_rate": 0.00011918351364854728,
      "loss": 12.1135,
      "step": 16726
    },
    {
      "epoch": 0.910852000844037,
      "grad_norm": 0.5054061078826275,
      "learning_rate": 0.00011917485909586008,
      "loss": 12.1893,
      "step": 16727
    },
    {
      "epoch": 0.9109064548406199,
      "grad_norm": 0.5295583824315849,
      "learning_rate": 0.00011916620439406597,
      "loss": 12.131,
      "step": 16728
    },
    {
      "epoch": 0.9109609088372029,
      "grad_norm": 0.5810858529402816,
      "learning_rate": 0.00011915754954323222,
      "loss": 12.0111,
      "step": 16729
    },
    {
      "epoch": 0.911015362833786,
      "grad_norm": 0.581320298295613,
      "learning_rate": 0.00011914889454342617,
      "loss": 12.2068,
      "step": 16730
    },
    {
      "epoch": 0.911069816830369,
      "grad_norm": 0.5417895266138278,
      "learning_rate": 0.0001191402393947151,
      "loss": 12.1504,
      "step": 16731
    },
    {
      "epoch": 0.911124270826952,
      "grad_norm": 0.5312038085499952,
      "learning_rate": 0.00011913158409716631,
      "loss": 12.0605,
      "step": 16732
    },
    {
      "epoch": 0.911178724823535,
      "grad_norm": 0.5206094610344083,
      "learning_rate": 0.00011912292865084713,
      "loss": 12.1059,
      "step": 16733
    },
    {
      "epoch": 0.911233178820118,
      "grad_norm": 0.5489024764471662,
      "learning_rate": 0.00011911427305582486,
      "loss": 12.0937,
      "step": 16734
    },
    {
      "epoch": 0.911287632816701,
      "grad_norm": 0.5308736008490176,
      "learning_rate": 0.00011910561731216676,
      "loss": 12.2083,
      "step": 16735
    },
    {
      "epoch": 0.9113420868132841,
      "grad_norm": 0.5017524052748702,
      "learning_rate": 0.0001190969614199402,
      "loss": 12.079,
      "step": 16736
    },
    {
      "epoch": 0.9113965408098671,
      "grad_norm": 0.5330214212294354,
      "learning_rate": 0.00011908830537921247,
      "loss": 12.2847,
      "step": 16737
    },
    {
      "epoch": 0.9114509948064501,
      "grad_norm": 0.542780798730531,
      "learning_rate": 0.00011907964919005085,
      "loss": 12.253,
      "step": 16738
    },
    {
      "epoch": 0.9115054488030331,
      "grad_norm": 0.526216751731207,
      "learning_rate": 0.0001190709928525227,
      "loss": 12.1371,
      "step": 16739
    },
    {
      "epoch": 0.9115599027996161,
      "grad_norm": 0.5608380080254172,
      "learning_rate": 0.0001190623363666953,
      "loss": 12.0116,
      "step": 16740
    },
    {
      "epoch": 0.9116143567961991,
      "grad_norm": 0.5547421600370966,
      "learning_rate": 0.000119053679732636,
      "loss": 12.1079,
      "step": 16741
    },
    {
      "epoch": 0.9116688107927822,
      "grad_norm": 0.614869012298698,
      "learning_rate": 0.00011904502295041206,
      "loss": 12.2039,
      "step": 16742
    },
    {
      "epoch": 0.9117232647893652,
      "grad_norm": 0.5727502640962558,
      "learning_rate": 0.00011903636602009087,
      "loss": 12.1143,
      "step": 16743
    },
    {
      "epoch": 0.9117777187859482,
      "grad_norm": 0.5575695243558199,
      "learning_rate": 0.00011902770894173967,
      "loss": 12.1738,
      "step": 16744
    },
    {
      "epoch": 0.9118321727825311,
      "grad_norm": 0.5975922358915751,
      "learning_rate": 0.0001190190517154258,
      "loss": 12.277,
      "step": 16745
    },
    {
      "epoch": 0.9118866267791141,
      "grad_norm": 0.6040288247120295,
      "learning_rate": 0.00011901039434121661,
      "loss": 12.1355,
      "step": 16746
    },
    {
      "epoch": 0.9119410807756971,
      "grad_norm": 0.6450256711017941,
      "learning_rate": 0.00011900173681917944,
      "loss": 12.1493,
      "step": 16747
    },
    {
      "epoch": 0.9119955347722802,
      "grad_norm": 0.4947022240185635,
      "learning_rate": 0.00011899307914938157,
      "loss": 12.1176,
      "step": 16748
    },
    {
      "epoch": 0.9120499887688632,
      "grad_norm": 0.5670942648674901,
      "learning_rate": 0.0001189844213318903,
      "loss": 12.2005,
      "step": 16749
    },
    {
      "epoch": 0.9121044427654462,
      "grad_norm": 0.5765238273736617,
      "learning_rate": 0.00011897576336677297,
      "loss": 12.0534,
      "step": 16750
    },
    {
      "epoch": 0.9121588967620292,
      "grad_norm": 0.5754905418803878,
      "learning_rate": 0.00011896710525409696,
      "loss": 12.1506,
      "step": 16751
    },
    {
      "epoch": 0.9122133507586122,
      "grad_norm": 0.597427615061958,
      "learning_rate": 0.00011895844699392952,
      "loss": 12.132,
      "step": 16752
    },
    {
      "epoch": 0.9122678047551952,
      "grad_norm": 0.5888072462379587,
      "learning_rate": 0.00011894978858633808,
      "loss": 12.1797,
      "step": 16753
    },
    {
      "epoch": 0.9123222587517783,
      "grad_norm": 0.5648742161166406,
      "learning_rate": 0.00011894113003138987,
      "loss": 12.0295,
      "step": 16754
    },
    {
      "epoch": 0.9123767127483613,
      "grad_norm": 0.580604401098401,
      "learning_rate": 0.00011893247132915224,
      "loss": 12.0438,
      "step": 16755
    },
    {
      "epoch": 0.9124311667449443,
      "grad_norm": 0.6148933844315049,
      "learning_rate": 0.00011892381247969256,
      "loss": 12.1361,
      "step": 16756
    },
    {
      "epoch": 0.9124856207415273,
      "grad_norm": 0.5691730989122715,
      "learning_rate": 0.00011891515348307812,
      "loss": 12.0467,
      "step": 16757
    },
    {
      "epoch": 0.9125400747381103,
      "grad_norm": 0.5702840363937903,
      "learning_rate": 0.00011890649433937631,
      "loss": 12.1824,
      "step": 16758
    },
    {
      "epoch": 0.9125945287346933,
      "grad_norm": 0.7081968855598006,
      "learning_rate": 0.00011889783504865442,
      "loss": 12.1837,
      "step": 16759
    },
    {
      "epoch": 0.9126489827312764,
      "grad_norm": 0.5773421638162692,
      "learning_rate": 0.00011888917561097978,
      "loss": 12.2206,
      "step": 16760
    },
    {
      "epoch": 0.9127034367278594,
      "grad_norm": 0.6249849239293285,
      "learning_rate": 0.00011888051602641971,
      "loss": 12.2438,
      "step": 16761
    },
    {
      "epoch": 0.9127578907244424,
      "grad_norm": 0.5705795563291992,
      "learning_rate": 0.00011887185629504162,
      "loss": 12.1928,
      "step": 16762
    },
    {
      "epoch": 0.9128123447210253,
      "grad_norm": 0.565809674777921,
      "learning_rate": 0.00011886319641691284,
      "loss": 12.1484,
      "step": 16763
    },
    {
      "epoch": 0.9128667987176083,
      "grad_norm": 0.6011464088918341,
      "learning_rate": 0.00011885453639210064,
      "loss": 12.0994,
      "step": 16764
    },
    {
      "epoch": 0.9129212527141914,
      "grad_norm": 0.5115240947395492,
      "learning_rate": 0.00011884587622067243,
      "loss": 12.1974,
      "step": 16765
    },
    {
      "epoch": 0.9129757067107744,
      "grad_norm": 0.5979106644846061,
      "learning_rate": 0.00011883721590269548,
      "loss": 12.2424,
      "step": 16766
    },
    {
      "epoch": 0.9130301607073574,
      "grad_norm": 0.6040356674593268,
      "learning_rate": 0.00011882855543823721,
      "loss": 12.2643,
      "step": 16767
    },
    {
      "epoch": 0.9130846147039404,
      "grad_norm": 0.6333776562522606,
      "learning_rate": 0.00011881989482736496,
      "loss": 12.2181,
      "step": 16768
    },
    {
      "epoch": 0.9131390687005234,
      "grad_norm": 0.5974897631083352,
      "learning_rate": 0.00011881123407014602,
      "loss": 12.2313,
      "step": 16769
    },
    {
      "epoch": 0.9131935226971064,
      "grad_norm": 0.5542553523882653,
      "learning_rate": 0.00011880257316664778,
      "loss": 12.2712,
      "step": 16770
    },
    {
      "epoch": 0.9132479766936895,
      "grad_norm": 0.5402366430344611,
      "learning_rate": 0.00011879391211693757,
      "loss": 12.1131,
      "step": 16771
    },
    {
      "epoch": 0.9133024306902725,
      "grad_norm": 0.511598019376126,
      "learning_rate": 0.00011878525092108272,
      "loss": 12.0113,
      "step": 16772
    },
    {
      "epoch": 0.9133568846868555,
      "grad_norm": 0.5790887579476809,
      "learning_rate": 0.00011877658957915068,
      "loss": 12.3035,
      "step": 16773
    },
    {
      "epoch": 0.9134113386834385,
      "grad_norm": 0.5195244342633508,
      "learning_rate": 0.00011876792809120867,
      "loss": 12.1332,
      "step": 16774
    },
    {
      "epoch": 0.9134657926800215,
      "grad_norm": 0.569483171720159,
      "learning_rate": 0.00011875926645732413,
      "loss": 12.0297,
      "step": 16775
    },
    {
      "epoch": 0.9135202466766045,
      "grad_norm": 0.5953612424801111,
      "learning_rate": 0.00011875060467756438,
      "loss": 12.3428,
      "step": 16776
    },
    {
      "epoch": 0.9135747006731876,
      "grad_norm": 0.5306157216561193,
      "learning_rate": 0.0001187419427519968,
      "loss": 12.1489,
      "step": 16777
    },
    {
      "epoch": 0.9136291546697706,
      "grad_norm": 0.5446436430199308,
      "learning_rate": 0.0001187332806806887,
      "loss": 12.192,
      "step": 16778
    },
    {
      "epoch": 0.9136836086663536,
      "grad_norm": 0.5653427027897644,
      "learning_rate": 0.00011872461846370748,
      "loss": 12.184,
      "step": 16779
    },
    {
      "epoch": 0.9137380626629366,
      "grad_norm": 0.5296990011718393,
      "learning_rate": 0.0001187159561011205,
      "loss": 12.2265,
      "step": 16780
    },
    {
      "epoch": 0.9137925166595195,
      "grad_norm": 0.5344827860754364,
      "learning_rate": 0.0001187072935929951,
      "loss": 12.1615,
      "step": 16781
    },
    {
      "epoch": 0.9138469706561025,
      "grad_norm": 0.5383766982801177,
      "learning_rate": 0.00011869863093939864,
      "loss": 12.1102,
      "step": 16782
    },
    {
      "epoch": 0.9139014246526856,
      "grad_norm": 1.0582908680442866,
      "learning_rate": 0.0001186899681403985,
      "loss": 12.2392,
      "step": 16783
    },
    {
      "epoch": 0.9139558786492686,
      "grad_norm": 0.6430932939318632,
      "learning_rate": 0.00011868130519606202,
      "loss": 12.1362,
      "step": 16784
    },
    {
      "epoch": 0.9140103326458516,
      "grad_norm": 0.5464335146088309,
      "learning_rate": 0.00011867264210645659,
      "loss": 12.1514,
      "step": 16785
    },
    {
      "epoch": 0.9140647866424346,
      "grad_norm": 0.5557665961056553,
      "learning_rate": 0.00011866397887164958,
      "loss": 12.2397,
      "step": 16786
    },
    {
      "epoch": 0.9141192406390176,
      "grad_norm": 0.5393363904487715,
      "learning_rate": 0.0001186553154917083,
      "loss": 12.1043,
      "step": 16787
    },
    {
      "epoch": 0.9141736946356006,
      "grad_norm": 0.5548053206768346,
      "learning_rate": 0.00011864665196670018,
      "loss": 12.2035,
      "step": 16788
    },
    {
      "epoch": 0.9142281486321837,
      "grad_norm": 0.6222726367876813,
      "learning_rate": 0.00011863798829669257,
      "loss": 12.0901,
      "step": 16789
    },
    {
      "epoch": 0.9142826026287667,
      "grad_norm": 0.5679218291031215,
      "learning_rate": 0.00011862932448175283,
      "loss": 12.0993,
      "step": 16790
    },
    {
      "epoch": 0.9143370566253497,
      "grad_norm": 0.5486238511901713,
      "learning_rate": 0.00011862066052194833,
      "loss": 12.2531,
      "step": 16791
    },
    {
      "epoch": 0.9143915106219327,
      "grad_norm": 0.5321184478689914,
      "learning_rate": 0.00011861199641734648,
      "loss": 12.1077,
      "step": 16792
    },
    {
      "epoch": 0.9144459646185157,
      "grad_norm": 0.5435200884960917,
      "learning_rate": 0.0001186033321680146,
      "loss": 12.1614,
      "step": 16793
    },
    {
      "epoch": 0.9145004186150988,
      "grad_norm": 0.5414606235146378,
      "learning_rate": 0.00011859466777402009,
      "loss": 12.2113,
      "step": 16794
    },
    {
      "epoch": 0.9145548726116818,
      "grad_norm": 0.5413555138696808,
      "learning_rate": 0.00011858600323543035,
      "loss": 12.1002,
      "step": 16795
    },
    {
      "epoch": 0.9146093266082648,
      "grad_norm": 0.5328312176322992,
      "learning_rate": 0.0001185773385523127,
      "loss": 12.0829,
      "step": 16796
    },
    {
      "epoch": 0.9146637806048478,
      "grad_norm": 0.5645118336086263,
      "learning_rate": 0.00011856867372473456,
      "loss": 12.279,
      "step": 16797
    },
    {
      "epoch": 0.9147182346014308,
      "grad_norm": 0.594588448885202,
      "learning_rate": 0.00011856000875276332,
      "loss": 12.1769,
      "step": 16798
    },
    {
      "epoch": 0.9147726885980138,
      "grad_norm": 0.5671975738301612,
      "learning_rate": 0.00011855134363646631,
      "loss": 12.1391,
      "step": 16799
    },
    {
      "epoch": 0.9148271425945969,
      "grad_norm": 0.5324538805071298,
      "learning_rate": 0.00011854267837591095,
      "loss": 12.1197,
      "step": 16800
    },
    {
      "epoch": 0.9148815965911798,
      "grad_norm": 0.6474357680124087,
      "learning_rate": 0.00011853401297116462,
      "loss": 12.1605,
      "step": 16801
    },
    {
      "epoch": 0.9149360505877628,
      "grad_norm": 0.5367501737187403,
      "learning_rate": 0.00011852534742229469,
      "loss": 12.0937,
      "step": 16802
    },
    {
      "epoch": 0.9149905045843458,
      "grad_norm": 0.5720099048286376,
      "learning_rate": 0.00011851668172936858,
      "loss": 12.0828,
      "step": 16803
    },
    {
      "epoch": 0.9150449585809288,
      "grad_norm": 0.5583897433769571,
      "learning_rate": 0.0001185080158924536,
      "loss": 12.1997,
      "step": 16804
    },
    {
      "epoch": 0.9150994125775118,
      "grad_norm": 0.54360969188526,
      "learning_rate": 0.0001184993499116172,
      "loss": 12.1959,
      "step": 16805
    },
    {
      "epoch": 0.9151538665740949,
      "grad_norm": 0.522848415236117,
      "learning_rate": 0.00011849068378692676,
      "loss": 12.1708,
      "step": 16806
    },
    {
      "epoch": 0.9152083205706779,
      "grad_norm": 0.6759180594616391,
      "learning_rate": 0.00011848201751844967,
      "loss": 12.2251,
      "step": 16807
    },
    {
      "epoch": 0.9152627745672609,
      "grad_norm": 0.5675597634248547,
      "learning_rate": 0.00011847335110625333,
      "loss": 12.0897,
      "step": 16808
    },
    {
      "epoch": 0.9153172285638439,
      "grad_norm": 0.5468070558084621,
      "learning_rate": 0.00011846468455040509,
      "loss": 12.1422,
      "step": 16809
    },
    {
      "epoch": 0.9153716825604269,
      "grad_norm": 0.5786821751893668,
      "learning_rate": 0.00011845601785097233,
      "loss": 12.1366,
      "step": 16810
    },
    {
      "epoch": 0.9154261365570099,
      "grad_norm": 0.5456084060971332,
      "learning_rate": 0.00011844735100802253,
      "loss": 12.2959,
      "step": 16811
    },
    {
      "epoch": 0.915480590553593,
      "grad_norm": 0.5598519840243396,
      "learning_rate": 0.00011843868402162301,
      "loss": 12.076,
      "step": 16812
    },
    {
      "epoch": 0.915535044550176,
      "grad_norm": 0.5677753649414218,
      "learning_rate": 0.00011843001689184123,
      "loss": 12.2355,
      "step": 16813
    },
    {
      "epoch": 0.915589498546759,
      "grad_norm": 0.5350599288088638,
      "learning_rate": 0.00011842134961874451,
      "loss": 12.1107,
      "step": 16814
    },
    {
      "epoch": 0.915643952543342,
      "grad_norm": 0.5516091666079663,
      "learning_rate": 0.00011841268220240027,
      "loss": 12.145,
      "step": 16815
    },
    {
      "epoch": 0.915698406539925,
      "grad_norm": 0.5692472115070899,
      "learning_rate": 0.00011840401464287596,
      "loss": 11.952,
      "step": 16816
    },
    {
      "epoch": 0.915752860536508,
      "grad_norm": 0.598519974976252,
      "learning_rate": 0.00011839534694023893,
      "loss": 12.0693,
      "step": 16817
    },
    {
      "epoch": 0.915807314533091,
      "grad_norm": 0.5313337553609586,
      "learning_rate": 0.0001183866790945566,
      "loss": 12.1934,
      "step": 16818
    },
    {
      "epoch": 0.915861768529674,
      "grad_norm": 0.5734942775980757,
      "learning_rate": 0.00011837801110589639,
      "loss": 12.1456,
      "step": 16819
    },
    {
      "epoch": 0.915916222526257,
      "grad_norm": 0.5834459456631099,
      "learning_rate": 0.00011836934297432568,
      "loss": 12.1803,
      "step": 16820
    },
    {
      "epoch": 0.91597067652284,
      "grad_norm": 0.5465156658572613,
      "learning_rate": 0.00011836067469991184,
      "loss": 12.0439,
      "step": 16821
    },
    {
      "epoch": 0.916025130519423,
      "grad_norm": 0.5169095512192226,
      "learning_rate": 0.00011835200628272234,
      "loss": 12.0206,
      "step": 16822
    },
    {
      "epoch": 0.916079584516006,
      "grad_norm": 0.5766012605581224,
      "learning_rate": 0.00011834333772282455,
      "loss": 12.132,
      "step": 16823
    },
    {
      "epoch": 0.9161340385125891,
      "grad_norm": 0.5590255109156209,
      "learning_rate": 0.00011833466902028594,
      "loss": 12.1715,
      "step": 16824
    },
    {
      "epoch": 0.9161884925091721,
      "grad_norm": 0.5938053946101209,
      "learning_rate": 0.00011832600017517382,
      "loss": 12.171,
      "step": 16825
    },
    {
      "epoch": 0.9162429465057551,
      "grad_norm": 0.6132387898997561,
      "learning_rate": 0.00011831733118755566,
      "loss": 12.1305,
      "step": 16826
    },
    {
      "epoch": 0.9162974005023381,
      "grad_norm": 0.6982163513457541,
      "learning_rate": 0.00011830866205749884,
      "loss": 12.2049,
      "step": 16827
    },
    {
      "epoch": 0.9163518544989211,
      "grad_norm": 0.5599338143513132,
      "learning_rate": 0.00011829999278507083,
      "loss": 12.1475,
      "step": 16828
    },
    {
      "epoch": 0.9164063084955042,
      "grad_norm": 0.6149948822024341,
      "learning_rate": 0.00011829132337033899,
      "loss": 12.1869,
      "step": 16829
    },
    {
      "epoch": 0.9164607624920872,
      "grad_norm": 0.6304619061712248,
      "learning_rate": 0.00011828265381337076,
      "loss": 12.2142,
      "step": 16830
    },
    {
      "epoch": 0.9165152164886702,
      "grad_norm": 0.4988338920506397,
      "learning_rate": 0.00011827398411423354,
      "loss": 12.006,
      "step": 16831
    },
    {
      "epoch": 0.9165696704852532,
      "grad_norm": 0.6173633751853064,
      "learning_rate": 0.00011826531427299475,
      "loss": 12.1481,
      "step": 16832
    },
    {
      "epoch": 0.9166241244818362,
      "grad_norm": 0.5357110912482415,
      "learning_rate": 0.00011825664428972181,
      "loss": 12.0745,
      "step": 16833
    },
    {
      "epoch": 0.9166785784784192,
      "grad_norm": 0.6967866106962702,
      "learning_rate": 0.00011824797416448216,
      "loss": 12.2176,
      "step": 16834
    },
    {
      "epoch": 0.9167330324750023,
      "grad_norm": 0.6544247026014385,
      "learning_rate": 0.0001182393038973432,
      "loss": 12.287,
      "step": 16835
    },
    {
      "epoch": 0.9167874864715853,
      "grad_norm": 0.5644438850069057,
      "learning_rate": 0.00011823063348837235,
      "loss": 12.1593,
      "step": 16836
    },
    {
      "epoch": 0.9168419404681682,
      "grad_norm": 0.5911753838445318,
      "learning_rate": 0.00011822196293763704,
      "loss": 12.1714,
      "step": 16837
    },
    {
      "epoch": 0.9168963944647512,
      "grad_norm": 0.5883370221029104,
      "learning_rate": 0.00011821329224520465,
      "loss": 12.1827,
      "step": 16838
    },
    {
      "epoch": 0.9169508484613342,
      "grad_norm": 0.5699740769714241,
      "learning_rate": 0.0001182046214111427,
      "loss": 12.0986,
      "step": 16839
    },
    {
      "epoch": 0.9170053024579172,
      "grad_norm": 0.6434367700000917,
      "learning_rate": 0.00011819595043551854,
      "loss": 12.2088,
      "step": 16840
    },
    {
      "epoch": 0.9170597564545003,
      "grad_norm": 0.5810603582792637,
      "learning_rate": 0.00011818727931839964,
      "loss": 12.2819,
      "step": 16841
    },
    {
      "epoch": 0.9171142104510833,
      "grad_norm": 0.7095274139215517,
      "learning_rate": 0.00011817860805985337,
      "loss": 12.2811,
      "step": 16842
    },
    {
      "epoch": 0.9171686644476663,
      "grad_norm": 0.6104125487630834,
      "learning_rate": 0.0001181699366599472,
      "loss": 12.0946,
      "step": 16843
    },
    {
      "epoch": 0.9172231184442493,
      "grad_norm": 0.5246410991558379,
      "learning_rate": 0.00011816126511874859,
      "loss": 11.9659,
      "step": 16844
    },
    {
      "epoch": 0.9172775724408323,
      "grad_norm": 0.6089764921307709,
      "learning_rate": 0.0001181525934363249,
      "loss": 12.1241,
      "step": 16845
    },
    {
      "epoch": 0.9173320264374153,
      "grad_norm": 0.5785060702910032,
      "learning_rate": 0.00011814392161274361,
      "loss": 12.1418,
      "step": 16846
    },
    {
      "epoch": 0.9173864804339984,
      "grad_norm": 0.6352933490510204,
      "learning_rate": 0.00011813524964807215,
      "loss": 12.1555,
      "step": 16847
    },
    {
      "epoch": 0.9174409344305814,
      "grad_norm": 0.5131026586102728,
      "learning_rate": 0.00011812657754237795,
      "loss": 12.0899,
      "step": 16848
    },
    {
      "epoch": 0.9174953884271644,
      "grad_norm": 0.5821382588317344,
      "learning_rate": 0.00011811790529572842,
      "loss": 12.1434,
      "step": 16849
    },
    {
      "epoch": 0.9175498424237474,
      "grad_norm": 0.5447482139681347,
      "learning_rate": 0.00011810923290819104,
      "loss": 12.1217,
      "step": 16850
    },
    {
      "epoch": 0.9176042964203304,
      "grad_norm": 0.539484649326593,
      "learning_rate": 0.00011810056037983322,
      "loss": 12.1442,
      "step": 16851
    },
    {
      "epoch": 0.9176587504169134,
      "grad_norm": 0.6068708590455357,
      "learning_rate": 0.00011809188771072241,
      "loss": 12.0923,
      "step": 16852
    },
    {
      "epoch": 0.9177132044134965,
      "grad_norm": 0.5918613326342391,
      "learning_rate": 0.00011808321490092605,
      "loss": 12.2175,
      "step": 16853
    },
    {
      "epoch": 0.9177676584100795,
      "grad_norm": 0.49693951625138766,
      "learning_rate": 0.00011807454195051158,
      "loss": 12.0775,
      "step": 16854
    },
    {
      "epoch": 0.9178221124066624,
      "grad_norm": 0.585130212317906,
      "learning_rate": 0.00011806586885954642,
      "loss": 12.114,
      "step": 16855
    },
    {
      "epoch": 0.9178765664032454,
      "grad_norm": 0.5801909218668415,
      "learning_rate": 0.00011805719562809807,
      "loss": 12.2656,
      "step": 16856
    },
    {
      "epoch": 0.9179310203998284,
      "grad_norm": 0.5522355927824034,
      "learning_rate": 0.00011804852225623391,
      "loss": 12.2535,
      "step": 16857
    },
    {
      "epoch": 0.9179854743964114,
      "grad_norm": 0.5427828363518948,
      "learning_rate": 0.00011803984874402143,
      "loss": 12.0822,
      "step": 16858
    },
    {
      "epoch": 0.9180399283929945,
      "grad_norm": 0.5661936373007997,
      "learning_rate": 0.00011803117509152805,
      "loss": 12.0889,
      "step": 16859
    },
    {
      "epoch": 0.9180943823895775,
      "grad_norm": 0.637976219585058,
      "learning_rate": 0.00011802250129882124,
      "loss": 12.0885,
      "step": 16860
    },
    {
      "epoch": 0.9181488363861605,
      "grad_norm": 0.6508504526745578,
      "learning_rate": 0.00011801382736596842,
      "loss": 12.1481,
      "step": 16861
    },
    {
      "epoch": 0.9182032903827435,
      "grad_norm": 0.5508959358035357,
      "learning_rate": 0.00011800515329303707,
      "loss": 12.239,
      "step": 16862
    },
    {
      "epoch": 0.9182577443793265,
      "grad_norm": 0.6082919722422235,
      "learning_rate": 0.00011799647908009463,
      "loss": 12.2931,
      "step": 16863
    },
    {
      "epoch": 0.9183121983759096,
      "grad_norm": 0.5692475687106825,
      "learning_rate": 0.00011798780472720854,
      "loss": 12.0328,
      "step": 16864
    },
    {
      "epoch": 0.9183666523724926,
      "grad_norm": 0.544721000401607,
      "learning_rate": 0.00011797913023444626,
      "loss": 12.1171,
      "step": 16865
    },
    {
      "epoch": 0.9184211063690756,
      "grad_norm": 0.5488719223519434,
      "learning_rate": 0.00011797045560187527,
      "loss": 12.2137,
      "step": 16866
    },
    {
      "epoch": 0.9184755603656586,
      "grad_norm": 0.541763616208224,
      "learning_rate": 0.000117961780829563,
      "loss": 12.1395,
      "step": 16867
    },
    {
      "epoch": 0.9185300143622416,
      "grad_norm": 0.5545722866569928,
      "learning_rate": 0.00011795310591757691,
      "loss": 12.2065,
      "step": 16868
    },
    {
      "epoch": 0.9185844683588246,
      "grad_norm": 0.8852819561606639,
      "learning_rate": 0.00011794443086598446,
      "loss": 12.1008,
      "step": 16869
    },
    {
      "epoch": 0.9186389223554077,
      "grad_norm": 0.5838155507667105,
      "learning_rate": 0.00011793575567485308,
      "loss": 12.1819,
      "step": 16870
    },
    {
      "epoch": 0.9186933763519907,
      "grad_norm": 0.5629298051200554,
      "learning_rate": 0.00011792708034425027,
      "loss": 12.0794,
      "step": 16871
    },
    {
      "epoch": 0.9187478303485737,
      "grad_norm": 0.6035217143909042,
      "learning_rate": 0.00011791840487424348,
      "loss": 12.1327,
      "step": 16872
    },
    {
      "epoch": 0.9188022843451567,
      "grad_norm": 0.5309279145877762,
      "learning_rate": 0.00011790972926490018,
      "loss": 12.2339,
      "step": 16873
    },
    {
      "epoch": 0.9188567383417396,
      "grad_norm": 0.5765969595397223,
      "learning_rate": 0.00011790105351628782,
      "loss": 12.2462,
      "step": 16874
    },
    {
      "epoch": 0.9189111923383226,
      "grad_norm": 0.6436547861263355,
      "learning_rate": 0.00011789237762847385,
      "loss": 12.1905,
      "step": 16875
    },
    {
      "epoch": 0.9189656463349057,
      "grad_norm": 0.5637473395664372,
      "learning_rate": 0.00011788370160152575,
      "loss": 12.1323,
      "step": 16876
    },
    {
      "epoch": 0.9190201003314887,
      "grad_norm": 0.5530859550148169,
      "learning_rate": 0.00011787502543551099,
      "loss": 12.099,
      "step": 16877
    },
    {
      "epoch": 0.9190745543280717,
      "grad_norm": 0.6553952935885003,
      "learning_rate": 0.00011786634913049703,
      "loss": 12.2056,
      "step": 16878
    },
    {
      "epoch": 0.9191290083246547,
      "grad_norm": 0.6089985457195687,
      "learning_rate": 0.00011785767268655139,
      "loss": 12.1522,
      "step": 16879
    },
    {
      "epoch": 0.9191834623212377,
      "grad_norm": 0.5779445588832643,
      "learning_rate": 0.00011784899610374144,
      "loss": 12.223,
      "step": 16880
    },
    {
      "epoch": 0.9192379163178207,
      "grad_norm": 0.7227344842874233,
      "learning_rate": 0.00011784031938213471,
      "loss": 12.2078,
      "step": 16881
    },
    {
      "epoch": 0.9192923703144038,
      "grad_norm": 0.6875875596524184,
      "learning_rate": 0.00011783164252179866,
      "loss": 12.2143,
      "step": 16882
    },
    {
      "epoch": 0.9193468243109868,
      "grad_norm": 0.556742417679642,
      "learning_rate": 0.00011782296552280079,
      "loss": 12.2031,
      "step": 16883
    },
    {
      "epoch": 0.9194012783075698,
      "grad_norm": 0.6334430636052265,
      "learning_rate": 0.00011781428838520856,
      "loss": 12.1927,
      "step": 16884
    },
    {
      "epoch": 0.9194557323041528,
      "grad_norm": 0.6206500293231294,
      "learning_rate": 0.00011780561110908941,
      "loss": 12.326,
      "step": 16885
    },
    {
      "epoch": 0.9195101863007358,
      "grad_norm": 0.5766624897076231,
      "learning_rate": 0.00011779693369451086,
      "loss": 12.2196,
      "step": 16886
    },
    {
      "epoch": 0.9195646402973188,
      "grad_norm": 0.583325671430832,
      "learning_rate": 0.00011778825614154031,
      "loss": 12.1991,
      "step": 16887
    },
    {
      "epoch": 0.9196190942939019,
      "grad_norm": 0.5834547821471319,
      "learning_rate": 0.00011777957845024537,
      "loss": 12.1825,
      "step": 16888
    },
    {
      "epoch": 0.9196735482904849,
      "grad_norm": 0.5174570598618606,
      "learning_rate": 0.00011777090062069343,
      "loss": 12.1828,
      "step": 16889
    },
    {
      "epoch": 0.9197280022870679,
      "grad_norm": 0.622195757252222,
      "learning_rate": 0.00011776222265295199,
      "loss": 12.1731,
      "step": 16890
    },
    {
      "epoch": 0.9197824562836509,
      "grad_norm": 0.5755695132597045,
      "learning_rate": 0.00011775354454708851,
      "loss": 12.2292,
      "step": 16891
    },
    {
      "epoch": 0.9198369102802338,
      "grad_norm": 0.620146372251081,
      "learning_rate": 0.00011774486630317048,
      "loss": 12.159,
      "step": 16892
    },
    {
      "epoch": 0.9198913642768168,
      "grad_norm": 0.6189664842317881,
      "learning_rate": 0.00011773618792126542,
      "loss": 12.1616,
      "step": 16893
    },
    {
      "epoch": 0.9199458182733999,
      "grad_norm": 0.5461333527479741,
      "learning_rate": 0.0001177275094014408,
      "loss": 12.0693,
      "step": 16894
    },
    {
      "epoch": 0.9200002722699829,
      "grad_norm": 0.6395709052423875,
      "learning_rate": 0.00011771883074376406,
      "loss": 12.1698,
      "step": 16895
    },
    {
      "epoch": 0.9200547262665659,
      "grad_norm": 0.5707100063782455,
      "learning_rate": 0.00011771015194830273,
      "loss": 12.1584,
      "step": 16896
    },
    {
      "epoch": 0.9201091802631489,
      "grad_norm": 0.6704402001258953,
      "learning_rate": 0.00011770147301512429,
      "loss": 12.2118,
      "step": 16897
    },
    {
      "epoch": 0.9201636342597319,
      "grad_norm": 0.553766648058549,
      "learning_rate": 0.00011769279394429622,
      "loss": 12.1416,
      "step": 16898
    },
    {
      "epoch": 0.920218088256315,
      "grad_norm": 0.5445145089497211,
      "learning_rate": 0.00011768411473588603,
      "loss": 12.0892,
      "step": 16899
    },
    {
      "epoch": 0.920272542252898,
      "grad_norm": 0.5384689612754129,
      "learning_rate": 0.0001176754353899612,
      "loss": 12.1376,
      "step": 16900
    },
    {
      "epoch": 0.920326996249481,
      "grad_norm": 0.6064363198046522,
      "learning_rate": 0.00011766675590658922,
      "loss": 12.178,
      "step": 16901
    },
    {
      "epoch": 0.920381450246064,
      "grad_norm": 0.5880024864089798,
      "learning_rate": 0.00011765807628583758,
      "loss": 12.1358,
      "step": 16902
    },
    {
      "epoch": 0.920435904242647,
      "grad_norm": 0.534753270596174,
      "learning_rate": 0.00011764939652777376,
      "loss": 12.1993,
      "step": 16903
    },
    {
      "epoch": 0.92049035823923,
      "grad_norm": 0.6295164454306247,
      "learning_rate": 0.0001176407166324653,
      "loss": 12.1118,
      "step": 16904
    },
    {
      "epoch": 0.9205448122358131,
      "grad_norm": 0.6023643479493942,
      "learning_rate": 0.00011763203659997965,
      "loss": 12.2011,
      "step": 16905
    },
    {
      "epoch": 0.9205992662323961,
      "grad_norm": 0.5757732415578404,
      "learning_rate": 0.00011762335643038433,
      "loss": 12.1225,
      "step": 16906
    },
    {
      "epoch": 0.9206537202289791,
      "grad_norm": 0.5355212834199039,
      "learning_rate": 0.00011761467612374684,
      "loss": 12.1321,
      "step": 16907
    },
    {
      "epoch": 0.9207081742255621,
      "grad_norm": 0.5048507275544266,
      "learning_rate": 0.00011760599568013468,
      "loss": 12.0923,
      "step": 16908
    },
    {
      "epoch": 0.920762628222145,
      "grad_norm": 0.5551109810794038,
      "learning_rate": 0.00011759731509961534,
      "loss": 12.1753,
      "step": 16909
    },
    {
      "epoch": 0.920817082218728,
      "grad_norm": 0.5191144251279728,
      "learning_rate": 0.00011758863438225631,
      "loss": 12.0751,
      "step": 16910
    },
    {
      "epoch": 0.9208715362153111,
      "grad_norm": 0.5625247792112014,
      "learning_rate": 0.00011757995352812514,
      "loss": 12.1726,
      "step": 16911
    },
    {
      "epoch": 0.9209259902118941,
      "grad_norm": 0.5466597820845551,
      "learning_rate": 0.00011757127253728928,
      "loss": 12.1765,
      "step": 16912
    },
    {
      "epoch": 0.9209804442084771,
      "grad_norm": 0.5194500983946128,
      "learning_rate": 0.00011756259140981627,
      "loss": 12.0903,
      "step": 16913
    },
    {
      "epoch": 0.9210348982050601,
      "grad_norm": 0.541872524740386,
      "learning_rate": 0.0001175539101457736,
      "loss": 12.2169,
      "step": 16914
    },
    {
      "epoch": 0.9210893522016431,
      "grad_norm": 0.6645674759196384,
      "learning_rate": 0.00011754522874522877,
      "loss": 12.2635,
      "step": 16915
    },
    {
      "epoch": 0.9211438061982261,
      "grad_norm": 0.5777726404571027,
      "learning_rate": 0.00011753654720824932,
      "loss": 12.0605,
      "step": 16916
    },
    {
      "epoch": 0.9211982601948092,
      "grad_norm": 0.5564081110481262,
      "learning_rate": 0.00011752786553490272,
      "loss": 12.1363,
      "step": 16917
    },
    {
      "epoch": 0.9212527141913922,
      "grad_norm": 0.6655345929789,
      "learning_rate": 0.00011751918372525652,
      "loss": 12.1562,
      "step": 16918
    },
    {
      "epoch": 0.9213071681879752,
      "grad_norm": 0.5578526087544446,
      "learning_rate": 0.00011751050177937818,
      "loss": 12.1479,
      "step": 16919
    },
    {
      "epoch": 0.9213616221845582,
      "grad_norm": 0.5116884761127108,
      "learning_rate": 0.00011750181969733528,
      "loss": 12.0556,
      "step": 16920
    },
    {
      "epoch": 0.9214160761811412,
      "grad_norm": 0.5883180073812652,
      "learning_rate": 0.00011749313747919526,
      "loss": 12.152,
      "step": 16921
    },
    {
      "epoch": 0.9214705301777242,
      "grad_norm": 0.71928827180107,
      "learning_rate": 0.00011748445512502568,
      "loss": 12.1099,
      "step": 16922
    },
    {
      "epoch": 0.9215249841743073,
      "grad_norm": 0.5801052874839137,
      "learning_rate": 0.00011747577263489405,
      "loss": 12.1079,
      "step": 16923
    },
    {
      "epoch": 0.9215794381708903,
      "grad_norm": 0.5791300993972605,
      "learning_rate": 0.00011746709000886786,
      "loss": 12.0104,
      "step": 16924
    },
    {
      "epoch": 0.9216338921674733,
      "grad_norm": 0.5914497436093543,
      "learning_rate": 0.00011745840724701467,
      "loss": 12.1728,
      "step": 16925
    },
    {
      "epoch": 0.9216883461640563,
      "grad_norm": 0.6521740942280411,
      "learning_rate": 0.00011744972434940199,
      "loss": 12.2992,
      "step": 16926
    },
    {
      "epoch": 0.9217428001606393,
      "grad_norm": 0.6474119664235948,
      "learning_rate": 0.0001174410413160973,
      "loss": 12.3222,
      "step": 16927
    },
    {
      "epoch": 0.9217972541572224,
      "grad_norm": 0.6115920339132841,
      "learning_rate": 0.00011743235814716816,
      "loss": 12.1622,
      "step": 16928
    },
    {
      "epoch": 0.9218517081538053,
      "grad_norm": 0.571447806039776,
      "learning_rate": 0.0001174236748426821,
      "loss": 12.1173,
      "step": 16929
    },
    {
      "epoch": 0.9219061621503883,
      "grad_norm": 0.6527582706546112,
      "learning_rate": 0.00011741499140270658,
      "loss": 12.07,
      "step": 16930
    },
    {
      "epoch": 0.9219606161469713,
      "grad_norm": 0.5238781389090017,
      "learning_rate": 0.0001174063078273092,
      "loss": 12.1486,
      "step": 16931
    },
    {
      "epoch": 0.9220150701435543,
      "grad_norm": 0.53294174741349,
      "learning_rate": 0.00011739762411655741,
      "loss": 12.1161,
      "step": 16932
    },
    {
      "epoch": 0.9220695241401373,
      "grad_norm": 0.54549396412689,
      "learning_rate": 0.00011738894027051882,
      "loss": 12.0161,
      "step": 16933
    },
    {
      "epoch": 0.9221239781367204,
      "grad_norm": 0.5851738088661577,
      "learning_rate": 0.00011738025628926092,
      "loss": 12.0931,
      "step": 16934
    },
    {
      "epoch": 0.9221784321333034,
      "grad_norm": 0.5742586998918857,
      "learning_rate": 0.00011737157217285122,
      "loss": 11.9496,
      "step": 16935
    },
    {
      "epoch": 0.9222328861298864,
      "grad_norm": 0.5667530315524123,
      "learning_rate": 0.00011736288792135721,
      "loss": 12.1213,
      "step": 16936
    },
    {
      "epoch": 0.9222873401264694,
      "grad_norm": 0.5389019254871072,
      "learning_rate": 0.00011735420353484654,
      "loss": 12.1589,
      "step": 16937
    },
    {
      "epoch": 0.9223417941230524,
      "grad_norm": 0.5653999762083491,
      "learning_rate": 0.00011734551901338664,
      "loss": 12.1234,
      "step": 16938
    },
    {
      "epoch": 0.9223962481196354,
      "grad_norm": 0.5291458672313059,
      "learning_rate": 0.0001173368343570451,
      "loss": 11.9858,
      "step": 16939
    },
    {
      "epoch": 0.9224507021162185,
      "grad_norm": 0.5492650586281563,
      "learning_rate": 0.00011732814956588942,
      "loss": 12.217,
      "step": 16940
    },
    {
      "epoch": 0.9225051561128015,
      "grad_norm": 0.5499049350097684,
      "learning_rate": 0.00011731946463998711,
      "loss": 12.1631,
      "step": 16941
    },
    {
      "epoch": 0.9225596101093845,
      "grad_norm": 0.6615920540288064,
      "learning_rate": 0.00011731077957940578,
      "loss": 12.1106,
      "step": 16942
    },
    {
      "epoch": 0.9226140641059675,
      "grad_norm": 0.5381055966689792,
      "learning_rate": 0.00011730209438421288,
      "loss": 12.2094,
      "step": 16943
    },
    {
      "epoch": 0.9226685181025505,
      "grad_norm": 0.5609330805323823,
      "learning_rate": 0.00011729340905447606,
      "loss": 12.1898,
      "step": 16944
    },
    {
      "epoch": 0.9227229720991335,
      "grad_norm": 0.5939709915701091,
      "learning_rate": 0.00011728472359026275,
      "loss": 12.2755,
      "step": 16945
    },
    {
      "epoch": 0.9227774260957166,
      "grad_norm": 0.627128974055508,
      "learning_rate": 0.00011727603799164053,
      "loss": 12.1736,
      "step": 16946
    },
    {
      "epoch": 0.9228318800922996,
      "grad_norm": 0.5773341602635771,
      "learning_rate": 0.00011726735225867693,
      "loss": 12.0639,
      "step": 16947
    },
    {
      "epoch": 0.9228863340888825,
      "grad_norm": 0.6039647843673996,
      "learning_rate": 0.00011725866639143952,
      "loss": 12.178,
      "step": 16948
    },
    {
      "epoch": 0.9229407880854655,
      "grad_norm": 0.5758653069003834,
      "learning_rate": 0.00011724998038999585,
      "loss": 12.0593,
      "step": 16949
    },
    {
      "epoch": 0.9229952420820485,
      "grad_norm": 0.5899518105644326,
      "learning_rate": 0.0001172412942544134,
      "loss": 12.2028,
      "step": 16950
    },
    {
      "epoch": 0.9230496960786315,
      "grad_norm": 0.5652653779097088,
      "learning_rate": 0.00011723260798475976,
      "loss": 12.2616,
      "step": 16951
    },
    {
      "epoch": 0.9231041500752146,
      "grad_norm": 0.5888009417927444,
      "learning_rate": 0.00011722392158110249,
      "loss": 12.0104,
      "step": 16952
    },
    {
      "epoch": 0.9231586040717976,
      "grad_norm": 0.5669795658816074,
      "learning_rate": 0.00011721523504350909,
      "loss": 12.2135,
      "step": 16953
    },
    {
      "epoch": 0.9232130580683806,
      "grad_norm": 0.611730610878642,
      "learning_rate": 0.00011720654837204717,
      "loss": 12.2023,
      "step": 16954
    },
    {
      "epoch": 0.9232675120649636,
      "grad_norm": 0.51904773839831,
      "learning_rate": 0.00011719786156678423,
      "loss": 12.1355,
      "step": 16955
    },
    {
      "epoch": 0.9233219660615466,
      "grad_norm": 0.6915896744511061,
      "learning_rate": 0.00011718917462778782,
      "loss": 12.2847,
      "step": 16956
    },
    {
      "epoch": 0.9233764200581296,
      "grad_norm": 0.6096948831390273,
      "learning_rate": 0.00011718048755512552,
      "loss": 12.0716,
      "step": 16957
    },
    {
      "epoch": 0.9234308740547127,
      "grad_norm": 0.6273856058339771,
      "learning_rate": 0.00011717180034886484,
      "loss": 12.2011,
      "step": 16958
    },
    {
      "epoch": 0.9234853280512957,
      "grad_norm": 0.557686025784662,
      "learning_rate": 0.00011716311300907339,
      "loss": 12.2558,
      "step": 16959
    },
    {
      "epoch": 0.9235397820478787,
      "grad_norm": 0.5179974039851645,
      "learning_rate": 0.0001171544255358187,
      "loss": 12.1361,
      "step": 16960
    },
    {
      "epoch": 0.9235942360444617,
      "grad_norm": 0.6006390284439724,
      "learning_rate": 0.0001171457379291683,
      "loss": 12.1918,
      "step": 16961
    },
    {
      "epoch": 0.9236486900410447,
      "grad_norm": 0.5772358993306812,
      "learning_rate": 0.00011713705018918979,
      "loss": 12.0963,
      "step": 16962
    },
    {
      "epoch": 0.9237031440376278,
      "grad_norm": 0.5863016849465408,
      "learning_rate": 0.00011712836231595067,
      "loss": 12.1746,
      "step": 16963
    },
    {
      "epoch": 0.9237575980342108,
      "grad_norm": 0.5758399347318343,
      "learning_rate": 0.00011711967430951853,
      "loss": 12.1444,
      "step": 16964
    },
    {
      "epoch": 0.9238120520307938,
      "grad_norm": 0.5939292813079025,
      "learning_rate": 0.000117110986169961,
      "loss": 12.276,
      "step": 16965
    },
    {
      "epoch": 0.9238665060273767,
      "grad_norm": 0.541052639544623,
      "learning_rate": 0.00011710229789734551,
      "loss": 12.2945,
      "step": 16966
    },
    {
      "epoch": 0.9239209600239597,
      "grad_norm": 0.505180080963915,
      "learning_rate": 0.00011709360949173968,
      "loss": 11.9867,
      "step": 16967
    },
    {
      "epoch": 0.9239754140205427,
      "grad_norm": 0.5817509351335834,
      "learning_rate": 0.0001170849209532111,
      "loss": 12.055,
      "step": 16968
    },
    {
      "epoch": 0.9240298680171258,
      "grad_norm": 0.540099205163771,
      "learning_rate": 0.00011707623228182729,
      "loss": 12.1972,
      "step": 16969
    },
    {
      "epoch": 0.9240843220137088,
      "grad_norm": 0.5672608784576997,
      "learning_rate": 0.00011706754347765587,
      "loss": 12.2125,
      "step": 16970
    },
    {
      "epoch": 0.9241387760102918,
      "grad_norm": 0.5743230894713309,
      "learning_rate": 0.00011705885454076435,
      "loss": 12.1818,
      "step": 16971
    },
    {
      "epoch": 0.9241932300068748,
      "grad_norm": 0.5462976545022575,
      "learning_rate": 0.00011705016547122032,
      "loss": 12.209,
      "step": 16972
    },
    {
      "epoch": 0.9242476840034578,
      "grad_norm": 0.5832898955105185,
      "learning_rate": 0.00011704147626909134,
      "loss": 12.3018,
      "step": 16973
    },
    {
      "epoch": 0.9243021380000408,
      "grad_norm": 0.7573427014823444,
      "learning_rate": 0.00011703278693444498,
      "loss": 12.1651,
      "step": 16974
    },
    {
      "epoch": 0.9243565919966239,
      "grad_norm": 0.5762810005951213,
      "learning_rate": 0.00011702409746734882,
      "loss": 12.1974,
      "step": 16975
    },
    {
      "epoch": 0.9244110459932069,
      "grad_norm": 0.5916727994725474,
      "learning_rate": 0.0001170154078678704,
      "loss": 12.3085,
      "step": 16976
    },
    {
      "epoch": 0.9244654999897899,
      "grad_norm": 0.7111713426251668,
      "learning_rate": 0.00011700671813607734,
      "loss": 12.2125,
      "step": 16977
    },
    {
      "epoch": 0.9245199539863729,
      "grad_norm": 0.5577895221948833,
      "learning_rate": 0.00011699802827203718,
      "loss": 12.0943,
      "step": 16978
    },
    {
      "epoch": 0.9245744079829559,
      "grad_norm": 0.5312133239506819,
      "learning_rate": 0.0001169893382758175,
      "loss": 12.1762,
      "step": 16979
    },
    {
      "epoch": 0.9246288619795389,
      "grad_norm": 0.5249657690938024,
      "learning_rate": 0.00011698064814748586,
      "loss": 12.1692,
      "step": 16980
    },
    {
      "epoch": 0.924683315976122,
      "grad_norm": 0.5837377384106609,
      "learning_rate": 0.00011697195788710988,
      "loss": 11.9651,
      "step": 16981
    },
    {
      "epoch": 0.924737769972705,
      "grad_norm": 0.5471769020073902,
      "learning_rate": 0.0001169632674947571,
      "loss": 12.1227,
      "step": 16982
    },
    {
      "epoch": 0.924792223969288,
      "grad_norm": 0.6455960922545634,
      "learning_rate": 0.0001169545769704951,
      "loss": 12.1644,
      "step": 16983
    },
    {
      "epoch": 0.924846677965871,
      "grad_norm": 0.6383411558745579,
      "learning_rate": 0.00011694588631439147,
      "loss": 12.1891,
      "step": 16984
    },
    {
      "epoch": 0.9249011319624539,
      "grad_norm": 0.5272725937755892,
      "learning_rate": 0.00011693719552651378,
      "loss": 12.0553,
      "step": 16985
    },
    {
      "epoch": 0.9249555859590369,
      "grad_norm": 0.5199498291500081,
      "learning_rate": 0.00011692850460692964,
      "loss": 12.1509,
      "step": 16986
    },
    {
      "epoch": 0.92501003995562,
      "grad_norm": 0.5810983895908186,
      "learning_rate": 0.0001169198135557066,
      "loss": 12.1364,
      "step": 16987
    },
    {
      "epoch": 0.925064493952203,
      "grad_norm": 0.5786941192229852,
      "learning_rate": 0.00011691112237291224,
      "loss": 12.2021,
      "step": 16988
    },
    {
      "epoch": 0.925118947948786,
      "grad_norm": 0.5227920236417976,
      "learning_rate": 0.00011690243105861416,
      "loss": 12.0734,
      "step": 16989
    },
    {
      "epoch": 0.925173401945369,
      "grad_norm": 0.605307586156183,
      "learning_rate": 0.00011689373961287995,
      "loss": 12.1793,
      "step": 16990
    },
    {
      "epoch": 0.925227855941952,
      "grad_norm": 0.5932639111977884,
      "learning_rate": 0.00011688504803577718,
      "loss": 12.1466,
      "step": 16991
    },
    {
      "epoch": 0.925282309938535,
      "grad_norm": 0.6001483622846763,
      "learning_rate": 0.00011687635632737346,
      "loss": 12.2558,
      "step": 16992
    },
    {
      "epoch": 0.9253367639351181,
      "grad_norm": 0.6580245600331703,
      "learning_rate": 0.00011686766448773634,
      "loss": 12.2324,
      "step": 16993
    },
    {
      "epoch": 0.9253912179317011,
      "grad_norm": 0.6248616057460186,
      "learning_rate": 0.00011685897251693345,
      "loss": 12.0222,
      "step": 16994
    },
    {
      "epoch": 0.9254456719282841,
      "grad_norm": 0.6384201382347111,
      "learning_rate": 0.00011685028041503236,
      "loss": 12.2823,
      "step": 16995
    },
    {
      "epoch": 0.9255001259248671,
      "grad_norm": 0.5851656381722564,
      "learning_rate": 0.00011684158818210064,
      "loss": 12.1895,
      "step": 16996
    },
    {
      "epoch": 0.9255545799214501,
      "grad_norm": 0.6050196836611006,
      "learning_rate": 0.00011683289581820593,
      "loss": 12.1716,
      "step": 16997
    },
    {
      "epoch": 0.9256090339180332,
      "grad_norm": 0.5977317895680825,
      "learning_rate": 0.0001168242033234158,
      "loss": 12.1989,
      "step": 16998
    },
    {
      "epoch": 0.9256634879146162,
      "grad_norm": 0.5553430762164486,
      "learning_rate": 0.00011681551069779784,
      "loss": 12.099,
      "step": 16999
    },
    {
      "epoch": 0.9257179419111992,
      "grad_norm": 0.6662028747923707,
      "learning_rate": 0.00011680681794141965,
      "loss": 12.1246,
      "step": 17000
    },
    {
      "epoch": 0.9257723959077822,
      "grad_norm": 0.6443699801689361,
      "learning_rate": 0.00011679812505434882,
      "loss": 12.0578,
      "step": 17001
    },
    {
      "epoch": 0.9258268499043651,
      "grad_norm": 0.5643059761112635,
      "learning_rate": 0.00011678943203665294,
      "loss": 12.1783,
      "step": 17002
    },
    {
      "epoch": 0.9258813039009481,
      "grad_norm": 0.6073214020860899,
      "learning_rate": 0.00011678073888839965,
      "loss": 12.2232,
      "step": 17003
    },
    {
      "epoch": 0.9259357578975312,
      "grad_norm": 0.6071808298586896,
      "learning_rate": 0.00011677204560965649,
      "loss": 12.2522,
      "step": 17004
    },
    {
      "epoch": 0.9259902118941142,
      "grad_norm": 0.6351405987968302,
      "learning_rate": 0.00011676335220049112,
      "loss": 12.2171,
      "step": 17005
    },
    {
      "epoch": 0.9260446658906972,
      "grad_norm": 0.5831917980551805,
      "learning_rate": 0.0001167546586609711,
      "loss": 12.146,
      "step": 17006
    },
    {
      "epoch": 0.9260991198872802,
      "grad_norm": 0.6338100597883539,
      "learning_rate": 0.00011674596499116404,
      "loss": 12.169,
      "step": 17007
    },
    {
      "epoch": 0.9261535738838632,
      "grad_norm": 0.5556819623122808,
      "learning_rate": 0.00011673727119113756,
      "loss": 12.0856,
      "step": 17008
    },
    {
      "epoch": 0.9262080278804462,
      "grad_norm": 0.5452693802865192,
      "learning_rate": 0.00011672857726095923,
      "loss": 12.2743,
      "step": 17009
    },
    {
      "epoch": 0.9262624818770293,
      "grad_norm": 0.5845607427851699,
      "learning_rate": 0.0001167198832006967,
      "loss": 12.3197,
      "step": 17010
    },
    {
      "epoch": 0.9263169358736123,
      "grad_norm": 0.5174778972822843,
      "learning_rate": 0.00011671118901041755,
      "loss": 12.1127,
      "step": 17011
    },
    {
      "epoch": 0.9263713898701953,
      "grad_norm": 0.5765368900728236,
      "learning_rate": 0.00011670249469018939,
      "loss": 12.2682,
      "step": 17012
    },
    {
      "epoch": 0.9264258438667783,
      "grad_norm": 0.5312777638235685,
      "learning_rate": 0.00011669380024007981,
      "loss": 12.1433,
      "step": 17013
    },
    {
      "epoch": 0.9264802978633613,
      "grad_norm": 0.4942873016918579,
      "learning_rate": 0.00011668510566015644,
      "loss": 12.0418,
      "step": 17014
    },
    {
      "epoch": 0.9265347518599443,
      "grad_norm": 0.56684886519407,
      "learning_rate": 0.00011667641095048693,
      "loss": 12.1658,
      "step": 17015
    },
    {
      "epoch": 0.9265892058565274,
      "grad_norm": 0.5438479777725457,
      "learning_rate": 0.00011666771611113885,
      "loss": 12.3115,
      "step": 17016
    },
    {
      "epoch": 0.9266436598531104,
      "grad_norm": 0.5589051106902513,
      "learning_rate": 0.00011665902114217979,
      "loss": 11.9983,
      "step": 17017
    },
    {
      "epoch": 0.9266981138496934,
      "grad_norm": 0.55847418262368,
      "learning_rate": 0.00011665032604367736,
      "loss": 12.059,
      "step": 17018
    },
    {
      "epoch": 0.9267525678462764,
      "grad_norm": 0.5630048352703018,
      "learning_rate": 0.00011664163081569923,
      "loss": 12.1887,
      "step": 17019
    },
    {
      "epoch": 0.9268070218428593,
      "grad_norm": 0.5732274252668589,
      "learning_rate": 0.00011663293545831302,
      "loss": 12.3322,
      "step": 17020
    },
    {
      "epoch": 0.9268614758394423,
      "grad_norm": 0.5943019439752977,
      "learning_rate": 0.00011662423997158629,
      "loss": 12.1177,
      "step": 17021
    },
    {
      "epoch": 0.9269159298360254,
      "grad_norm": 0.5640282485806937,
      "learning_rate": 0.00011661554435558668,
      "loss": 12.1892,
      "step": 17022
    },
    {
      "epoch": 0.9269703838326084,
      "grad_norm": 0.5874228700734306,
      "learning_rate": 0.00011660684861038181,
      "loss": 12.1637,
      "step": 17023
    },
    {
      "epoch": 0.9270248378291914,
      "grad_norm": 0.5497658023687364,
      "learning_rate": 0.00011659815273603927,
      "loss": 12.1317,
      "step": 17024
    },
    {
      "epoch": 0.9270792918257744,
      "grad_norm": 0.7441199081125127,
      "learning_rate": 0.00011658945673262675,
      "loss": 12.3582,
      "step": 17025
    },
    {
      "epoch": 0.9271337458223574,
      "grad_norm": 0.6354938189067647,
      "learning_rate": 0.00011658076060021184,
      "loss": 12.2086,
      "step": 17026
    },
    {
      "epoch": 0.9271881998189404,
      "grad_norm": 0.5038329814281183,
      "learning_rate": 0.00011657206433886214,
      "loss": 12.1215,
      "step": 17027
    },
    {
      "epoch": 0.9272426538155235,
      "grad_norm": 0.5855466980777795,
      "learning_rate": 0.00011656336794864528,
      "loss": 12.2244,
      "step": 17028
    },
    {
      "epoch": 0.9272971078121065,
      "grad_norm": 0.6551742005373945,
      "learning_rate": 0.0001165546714296289,
      "loss": 12.1614,
      "step": 17029
    },
    {
      "epoch": 0.9273515618086895,
      "grad_norm": 0.6345041410268087,
      "learning_rate": 0.00011654597478188061,
      "loss": 12.115,
      "step": 17030
    },
    {
      "epoch": 0.9274060158052725,
      "grad_norm": 0.5470199496279786,
      "learning_rate": 0.00011653727800546805,
      "loss": 12.1861,
      "step": 17031
    },
    {
      "epoch": 0.9274604698018555,
      "grad_norm": 0.6602093942963679,
      "learning_rate": 0.00011652858110045886,
      "loss": 12.0375,
      "step": 17032
    },
    {
      "epoch": 0.9275149237984386,
      "grad_norm": 0.6575298540859775,
      "learning_rate": 0.00011651988406692062,
      "loss": 12.3605,
      "step": 17033
    },
    {
      "epoch": 0.9275693777950216,
      "grad_norm": 0.5075422298046308,
      "learning_rate": 0.00011651118690492102,
      "loss": 12.0472,
      "step": 17034
    },
    {
      "epoch": 0.9276238317916046,
      "grad_norm": 0.6874168942810434,
      "learning_rate": 0.00011650248961452765,
      "loss": 12.1397,
      "step": 17035
    },
    {
      "epoch": 0.9276782857881876,
      "grad_norm": 0.573716036356838,
      "learning_rate": 0.00011649379219580816,
      "loss": 12.1485,
      "step": 17036
    },
    {
      "epoch": 0.9277327397847706,
      "grad_norm": 0.594723876278983,
      "learning_rate": 0.00011648509464883018,
      "loss": 12.2291,
      "step": 17037
    },
    {
      "epoch": 0.9277871937813535,
      "grad_norm": 0.5567430749098153,
      "learning_rate": 0.0001164763969736613,
      "loss": 12.1369,
      "step": 17038
    },
    {
      "epoch": 0.9278416477779367,
      "grad_norm": 0.6402356541374973,
      "learning_rate": 0.00011646769917036923,
      "loss": 12.2883,
      "step": 17039
    },
    {
      "epoch": 0.9278961017745196,
      "grad_norm": 0.5948790511726406,
      "learning_rate": 0.00011645900123902156,
      "loss": 12.1557,
      "step": 17040
    },
    {
      "epoch": 0.9279505557711026,
      "grad_norm": 0.5748905277106298,
      "learning_rate": 0.00011645030317968594,
      "loss": 12.1694,
      "step": 17041
    },
    {
      "epoch": 0.9280050097676856,
      "grad_norm": 0.6725080394286576,
      "learning_rate": 0.00011644160499243002,
      "loss": 12.1196,
      "step": 17042
    },
    {
      "epoch": 0.9280594637642686,
      "grad_norm": 0.6136217550622626,
      "learning_rate": 0.00011643290667732141,
      "loss": 12.114,
      "step": 17043
    },
    {
      "epoch": 0.9281139177608516,
      "grad_norm": 0.5746529170486847,
      "learning_rate": 0.00011642420823442774,
      "loss": 12.1748,
      "step": 17044
    },
    {
      "epoch": 0.9281683717574347,
      "grad_norm": 0.7606424559623073,
      "learning_rate": 0.00011641550966381669,
      "loss": 12.1793,
      "step": 17045
    },
    {
      "epoch": 0.9282228257540177,
      "grad_norm": 0.5736704979817694,
      "learning_rate": 0.00011640681096555588,
      "loss": 12.2137,
      "step": 17046
    },
    {
      "epoch": 0.9282772797506007,
      "grad_norm": 0.5956818081165554,
      "learning_rate": 0.00011639811213971297,
      "loss": 12.2121,
      "step": 17047
    },
    {
      "epoch": 0.9283317337471837,
      "grad_norm": 0.5155022066522056,
      "learning_rate": 0.00011638941318635557,
      "loss": 12.2178,
      "step": 17048
    },
    {
      "epoch": 0.9283861877437667,
      "grad_norm": 0.6042941383500556,
      "learning_rate": 0.00011638071410555136,
      "loss": 12.1546,
      "step": 17049
    },
    {
      "epoch": 0.9284406417403497,
      "grad_norm": 0.6940539649814073,
      "learning_rate": 0.00011637201489736797,
      "loss": 12.1122,
      "step": 17050
    },
    {
      "epoch": 0.9284950957369328,
      "grad_norm": 0.5318145162753902,
      "learning_rate": 0.00011636331556187303,
      "loss": 12.1872,
      "step": 17051
    },
    {
      "epoch": 0.9285495497335158,
      "grad_norm": 0.6024309285582887,
      "learning_rate": 0.00011635461609913422,
      "loss": 12.2451,
      "step": 17052
    },
    {
      "epoch": 0.9286040037300988,
      "grad_norm": 0.5717953753086978,
      "learning_rate": 0.00011634591650921916,
      "loss": 12.1309,
      "step": 17053
    },
    {
      "epoch": 0.9286584577266818,
      "grad_norm": 0.6785739383229514,
      "learning_rate": 0.00011633721679219553,
      "loss": 12.0856,
      "step": 17054
    },
    {
      "epoch": 0.9287129117232648,
      "grad_norm": 0.5574867157515434,
      "learning_rate": 0.00011632851694813096,
      "loss": 12.2054,
      "step": 17055
    },
    {
      "epoch": 0.9287673657198477,
      "grad_norm": 0.5600461548071675,
      "learning_rate": 0.00011631981697709308,
      "loss": 12.2204,
      "step": 17056
    },
    {
      "epoch": 0.9288218197164309,
      "grad_norm": 0.5760992539546776,
      "learning_rate": 0.00011631111687914959,
      "loss": 12.1364,
      "step": 17057
    },
    {
      "epoch": 0.9288762737130138,
      "grad_norm": 0.5484661974854979,
      "learning_rate": 0.00011630241665436812,
      "loss": 12.2272,
      "step": 17058
    },
    {
      "epoch": 0.9289307277095968,
      "grad_norm": 0.6020621004749886,
      "learning_rate": 0.00011629371630281632,
      "loss": 12.1409,
      "step": 17059
    },
    {
      "epoch": 0.9289851817061798,
      "grad_norm": 0.554976863830771,
      "learning_rate": 0.00011628501582456186,
      "loss": 12.1648,
      "step": 17060
    },
    {
      "epoch": 0.9290396357027628,
      "grad_norm": 0.5700672634511531,
      "learning_rate": 0.00011627631521967237,
      "loss": 12.1624,
      "step": 17061
    },
    {
      "epoch": 0.9290940896993459,
      "grad_norm": 0.5677438278839031,
      "learning_rate": 0.00011626761448821551,
      "loss": 12.1885,
      "step": 17062
    },
    {
      "epoch": 0.9291485436959289,
      "grad_norm": 0.7112087422260562,
      "learning_rate": 0.00011625891363025896,
      "loss": 12.3993,
      "step": 17063
    },
    {
      "epoch": 0.9292029976925119,
      "grad_norm": 0.611967896965925,
      "learning_rate": 0.00011625021264587037,
      "loss": 12.1846,
      "step": 17064
    },
    {
      "epoch": 0.9292574516890949,
      "grad_norm": 0.5658316156390023,
      "learning_rate": 0.00011624151153511745,
      "loss": 12.1155,
      "step": 17065
    },
    {
      "epoch": 0.9293119056856779,
      "grad_norm": 0.5161574686800847,
      "learning_rate": 0.00011623281029806776,
      "loss": 12.102,
      "step": 17066
    },
    {
      "epoch": 0.9293663596822609,
      "grad_norm": 0.5298763617916129,
      "learning_rate": 0.000116224108934789,
      "loss": 12.1406,
      "step": 17067
    },
    {
      "epoch": 0.929420813678844,
      "grad_norm": 0.6023836352490733,
      "learning_rate": 0.00011621540744534886,
      "loss": 12.1412,
      "step": 17068
    },
    {
      "epoch": 0.929475267675427,
      "grad_norm": 0.6430555819292236,
      "learning_rate": 0.00011620670582981504,
      "loss": 12.17,
      "step": 17069
    },
    {
      "epoch": 0.92952972167201,
      "grad_norm": 0.5346111639127659,
      "learning_rate": 0.00011619800408825511,
      "loss": 12.2956,
      "step": 17070
    },
    {
      "epoch": 0.929584175668593,
      "grad_norm": 0.5934744123185334,
      "learning_rate": 0.00011618930222073681,
      "loss": 12.1488,
      "step": 17071
    },
    {
      "epoch": 0.929638629665176,
      "grad_norm": 0.5902132907553823,
      "learning_rate": 0.00011618060022732778,
      "loss": 12.137,
      "step": 17072
    },
    {
      "epoch": 0.929693083661759,
      "grad_norm": 0.5783018779464009,
      "learning_rate": 0.00011617189810809566,
      "loss": 12.1546,
      "step": 17073
    },
    {
      "epoch": 0.9297475376583421,
      "grad_norm": 0.5436054397773015,
      "learning_rate": 0.00011616319586310815,
      "loss": 12.0984,
      "step": 17074
    },
    {
      "epoch": 0.929801991654925,
      "grad_norm": 0.5234324656657,
      "learning_rate": 0.00011615449349243297,
      "loss": 12.1047,
      "step": 17075
    },
    {
      "epoch": 0.929856445651508,
      "grad_norm": 0.5774191004854853,
      "learning_rate": 0.0001161457909961377,
      "loss": 12.1284,
      "step": 17076
    },
    {
      "epoch": 0.929910899648091,
      "grad_norm": 0.5535469054706854,
      "learning_rate": 0.00011613708837429005,
      "loss": 12.1228,
      "step": 17077
    },
    {
      "epoch": 0.929965353644674,
      "grad_norm": 0.5883833818085753,
      "learning_rate": 0.00011612838562695772,
      "loss": 12.2282,
      "step": 17078
    },
    {
      "epoch": 0.930019807641257,
      "grad_norm": 0.5181596944895872,
      "learning_rate": 0.0001161196827542083,
      "loss": 12.1033,
      "step": 17079
    },
    {
      "epoch": 0.9300742616378401,
      "grad_norm": 0.5877125272973959,
      "learning_rate": 0.00011611097975610959,
      "loss": 12.1928,
      "step": 17080
    },
    {
      "epoch": 0.9301287156344231,
      "grad_norm": 0.5048991643199667,
      "learning_rate": 0.00011610227663272917,
      "loss": 12.1095,
      "step": 17081
    },
    {
      "epoch": 0.9301831696310061,
      "grad_norm": 0.6103487489415191,
      "learning_rate": 0.00011609357338413476,
      "loss": 11.9787,
      "step": 17082
    },
    {
      "epoch": 0.9302376236275891,
      "grad_norm": 0.5328646129693538,
      "learning_rate": 0.000116084870010394,
      "loss": 12.1755,
      "step": 17083
    },
    {
      "epoch": 0.9302920776241721,
      "grad_norm": 0.6728846970846543,
      "learning_rate": 0.00011607616651157461,
      "loss": 12.0957,
      "step": 17084
    },
    {
      "epoch": 0.9303465316207551,
      "grad_norm": 0.5457521547207682,
      "learning_rate": 0.00011606746288774426,
      "loss": 12.2521,
      "step": 17085
    },
    {
      "epoch": 0.9304009856173382,
      "grad_norm": 0.5770038977786668,
      "learning_rate": 0.00011605875913897062,
      "loss": 12.0996,
      "step": 17086
    },
    {
      "epoch": 0.9304554396139212,
      "grad_norm": 0.5414805298595619,
      "learning_rate": 0.00011605005526532136,
      "loss": 12.2006,
      "step": 17087
    },
    {
      "epoch": 0.9305098936105042,
      "grad_norm": 0.551139919852669,
      "learning_rate": 0.0001160413512668642,
      "loss": 12.1229,
      "step": 17088
    },
    {
      "epoch": 0.9305643476070872,
      "grad_norm": 0.5455866926700138,
      "learning_rate": 0.00011603264714366677,
      "loss": 12.082,
      "step": 17089
    },
    {
      "epoch": 0.9306188016036702,
      "grad_norm": 0.5882208100494172,
      "learning_rate": 0.00011602394289579681,
      "loss": 12.1895,
      "step": 17090
    },
    {
      "epoch": 0.9306732556002532,
      "grad_norm": 0.5537036790854779,
      "learning_rate": 0.00011601523852332199,
      "loss": 12.1104,
      "step": 17091
    },
    {
      "epoch": 0.9307277095968363,
      "grad_norm": 0.5617690974975039,
      "learning_rate": 0.00011600653402630999,
      "loss": 12.2329,
      "step": 17092
    },
    {
      "epoch": 0.9307821635934193,
      "grad_norm": 0.5464658860363083,
      "learning_rate": 0.00011599782940482849,
      "loss": 12.1214,
      "step": 17093
    },
    {
      "epoch": 0.9308366175900022,
      "grad_norm": 0.5451502554132573,
      "learning_rate": 0.00011598912465894518,
      "loss": 12.129,
      "step": 17094
    },
    {
      "epoch": 0.9308910715865852,
      "grad_norm": 0.5438190252079544,
      "learning_rate": 0.00011598041978872776,
      "loss": 12.102,
      "step": 17095
    },
    {
      "epoch": 0.9309455255831682,
      "grad_norm": 0.5555309407005641,
      "learning_rate": 0.00011597171479424391,
      "loss": 12.1562,
      "step": 17096
    },
    {
      "epoch": 0.9309999795797513,
      "grad_norm": 0.5117316313785095,
      "learning_rate": 0.00011596300967556132,
      "loss": 12.0907,
      "step": 17097
    },
    {
      "epoch": 0.9310544335763343,
      "grad_norm": 0.610060574719516,
      "learning_rate": 0.00011595430443274771,
      "loss": 12.1539,
      "step": 17098
    },
    {
      "epoch": 0.9311088875729173,
      "grad_norm": 0.674916055042679,
      "learning_rate": 0.00011594559906587075,
      "loss": 12.3322,
      "step": 17099
    },
    {
      "epoch": 0.9311633415695003,
      "grad_norm": 0.5116038869777758,
      "learning_rate": 0.00011593689357499813,
      "loss": 12.1525,
      "step": 17100
    },
    {
      "epoch": 0.9312177955660833,
      "grad_norm": 0.5640310944624323,
      "learning_rate": 0.00011592818796019756,
      "loss": 12.2007,
      "step": 17101
    },
    {
      "epoch": 0.9312722495626663,
      "grad_norm": 0.6681081203799246,
      "learning_rate": 0.00011591948222153672,
      "loss": 12.1388,
      "step": 17102
    },
    {
      "epoch": 0.9313267035592494,
      "grad_norm": 0.5026041518446537,
      "learning_rate": 0.00011591077635908332,
      "loss": 12.1392,
      "step": 17103
    },
    {
      "epoch": 0.9313811575558324,
      "grad_norm": 0.5451971935992687,
      "learning_rate": 0.00011590207037290506,
      "loss": 12.2014,
      "step": 17104
    },
    {
      "epoch": 0.9314356115524154,
      "grad_norm": 0.6192185932773268,
      "learning_rate": 0.00011589336426306963,
      "loss": 12.2781,
      "step": 17105
    },
    {
      "epoch": 0.9314900655489984,
      "grad_norm": 0.5562586554348979,
      "learning_rate": 0.00011588465802964472,
      "loss": 12.0791,
      "step": 17106
    },
    {
      "epoch": 0.9315445195455814,
      "grad_norm": 0.5452650641975968,
      "learning_rate": 0.00011587595167269807,
      "loss": 12.1655,
      "step": 17107
    },
    {
      "epoch": 0.9315989735421644,
      "grad_norm": 0.7624867306533909,
      "learning_rate": 0.00011586724519229734,
      "loss": 12.1808,
      "step": 17108
    },
    {
      "epoch": 0.9316534275387475,
      "grad_norm": 0.6006007027077219,
      "learning_rate": 0.00011585853858851026,
      "loss": 12.1602,
      "step": 17109
    },
    {
      "epoch": 0.9317078815353305,
      "grad_norm": 0.5315077069002945,
      "learning_rate": 0.00011584983186140453,
      "loss": 12.2366,
      "step": 17110
    },
    {
      "epoch": 0.9317623355319135,
      "grad_norm": 0.6186125620931717,
      "learning_rate": 0.00011584112501104785,
      "loss": 12.1332,
      "step": 17111
    },
    {
      "epoch": 0.9318167895284964,
      "grad_norm": 0.5338508517958055,
      "learning_rate": 0.0001158324180375079,
      "loss": 12.1224,
      "step": 17112
    },
    {
      "epoch": 0.9318712435250794,
      "grad_norm": 0.5366362161509145,
      "learning_rate": 0.00011582371094085243,
      "loss": 12.1756,
      "step": 17113
    },
    {
      "epoch": 0.9319256975216624,
      "grad_norm": 0.5225751437955348,
      "learning_rate": 0.00011581500372114912,
      "loss": 12.1414,
      "step": 17114
    },
    {
      "epoch": 0.9319801515182455,
      "grad_norm": 0.5488376445852008,
      "learning_rate": 0.00011580629637846572,
      "loss": 12.2341,
      "step": 17115
    },
    {
      "epoch": 0.9320346055148285,
      "grad_norm": 0.6082275491325198,
      "learning_rate": 0.00011579758891286988,
      "loss": 12.1232,
      "step": 17116
    },
    {
      "epoch": 0.9320890595114115,
      "grad_norm": 0.7465683112808241,
      "learning_rate": 0.00011578888132442935,
      "loss": 12.1643,
      "step": 17117
    },
    {
      "epoch": 0.9321435135079945,
      "grad_norm": 0.5606054589322913,
      "learning_rate": 0.00011578017361321183,
      "loss": 12.0169,
      "step": 17118
    },
    {
      "epoch": 0.9321979675045775,
      "grad_norm": 0.48457016303443323,
      "learning_rate": 0.00011577146577928501,
      "loss": 12.2196,
      "step": 17119
    },
    {
      "epoch": 0.9322524215011605,
      "grad_norm": 0.6522209736071727,
      "learning_rate": 0.00011576275782271666,
      "loss": 12.1335,
      "step": 17120
    },
    {
      "epoch": 0.9323068754977436,
      "grad_norm": 0.5705310071264384,
      "learning_rate": 0.00011575404974357447,
      "loss": 12.1468,
      "step": 17121
    },
    {
      "epoch": 0.9323613294943266,
      "grad_norm": 0.6997733089409353,
      "learning_rate": 0.00011574534154192611,
      "loss": 12.2644,
      "step": 17122
    },
    {
      "epoch": 0.9324157834909096,
      "grad_norm": 0.53727646261493,
      "learning_rate": 0.00011573663321783935,
      "loss": 12.2136,
      "step": 17123
    },
    {
      "epoch": 0.9324702374874926,
      "grad_norm": 0.5202735440742016,
      "learning_rate": 0.00011572792477138188,
      "loss": 12.0472,
      "step": 17124
    },
    {
      "epoch": 0.9325246914840756,
      "grad_norm": 0.5814386451768224,
      "learning_rate": 0.00011571921620262145,
      "loss": 12.1675,
      "step": 17125
    },
    {
      "epoch": 0.9325791454806586,
      "grad_norm": 0.597725688567498,
      "learning_rate": 0.00011571050751162577,
      "loss": 12.2002,
      "step": 17126
    },
    {
      "epoch": 0.9326335994772417,
      "grad_norm": 0.5559482868976506,
      "learning_rate": 0.00011570179869846252,
      "loss": 12.1005,
      "step": 17127
    },
    {
      "epoch": 0.9326880534738247,
      "grad_norm": 0.5902442834706637,
      "learning_rate": 0.00011569308976319946,
      "loss": 12.2999,
      "step": 17128
    },
    {
      "epoch": 0.9327425074704077,
      "grad_norm": 0.6024852485207571,
      "learning_rate": 0.00011568438070590429,
      "loss": 12.2589,
      "step": 17129
    },
    {
      "epoch": 0.9327969614669906,
      "grad_norm": 0.5433397289395396,
      "learning_rate": 0.00011567567152664476,
      "loss": 12.2047,
      "step": 17130
    },
    {
      "epoch": 0.9328514154635736,
      "grad_norm": 0.5315996546762495,
      "learning_rate": 0.00011566696222548858,
      "loss": 12.1595,
      "step": 17131
    },
    {
      "epoch": 0.9329058694601567,
      "grad_norm": 0.6238510583867447,
      "learning_rate": 0.00011565825280250348,
      "loss": 12.037,
      "step": 17132
    },
    {
      "epoch": 0.9329603234567397,
      "grad_norm": 0.5691266977985826,
      "learning_rate": 0.00011564954325775714,
      "loss": 12.2516,
      "step": 17133
    },
    {
      "epoch": 0.9330147774533227,
      "grad_norm": 0.5831989456009242,
      "learning_rate": 0.00011564083359131736,
      "loss": 12.2016,
      "step": 17134
    },
    {
      "epoch": 0.9330692314499057,
      "grad_norm": 0.5385068948884913,
      "learning_rate": 0.00011563212380325184,
      "loss": 12.0845,
      "step": 17135
    },
    {
      "epoch": 0.9331236854464887,
      "grad_norm": 0.5410471288548246,
      "learning_rate": 0.00011562341389362831,
      "loss": 12.16,
      "step": 17136
    },
    {
      "epoch": 0.9331781394430717,
      "grad_norm": 0.5636545198374513,
      "learning_rate": 0.00011561470386251449,
      "loss": 12.2888,
      "step": 17137
    },
    {
      "epoch": 0.9332325934396548,
      "grad_norm": 0.5218132981958773,
      "learning_rate": 0.00011560599370997812,
      "loss": 12.113,
      "step": 17138
    },
    {
      "epoch": 0.9332870474362378,
      "grad_norm": 0.5833455834743435,
      "learning_rate": 0.00011559728343608687,
      "loss": 12.0584,
      "step": 17139
    },
    {
      "epoch": 0.9333415014328208,
      "grad_norm": 0.5390141802259124,
      "learning_rate": 0.00011558857304090858,
      "loss": 12.0942,
      "step": 17140
    },
    {
      "epoch": 0.9333959554294038,
      "grad_norm": 0.5172155285648882,
      "learning_rate": 0.00011557986252451095,
      "loss": 12.1737,
      "step": 17141
    },
    {
      "epoch": 0.9334504094259868,
      "grad_norm": 0.5277277962681094,
      "learning_rate": 0.00011557115188696163,
      "loss": 11.991,
      "step": 17142
    },
    {
      "epoch": 0.9335048634225698,
      "grad_norm": 0.5706722403948381,
      "learning_rate": 0.00011556244112832848,
      "loss": 12.0166,
      "step": 17143
    },
    {
      "epoch": 0.9335593174191529,
      "grad_norm": 0.5892789767614426,
      "learning_rate": 0.00011555373024867912,
      "loss": 12.2115,
      "step": 17144
    },
    {
      "epoch": 0.9336137714157359,
      "grad_norm": 0.575780453957793,
      "learning_rate": 0.00011554501924808139,
      "loss": 12.0539,
      "step": 17145
    },
    {
      "epoch": 0.9336682254123189,
      "grad_norm": 0.5678263145783361,
      "learning_rate": 0.00011553630812660298,
      "loss": 12.1123,
      "step": 17146
    },
    {
      "epoch": 0.9337226794089019,
      "grad_norm": 0.5739674121461581,
      "learning_rate": 0.00011552759688431165,
      "loss": 12.143,
      "step": 17147
    },
    {
      "epoch": 0.9337771334054848,
      "grad_norm": 0.523763579277213,
      "learning_rate": 0.0001155188855212751,
      "loss": 12.0914,
      "step": 17148
    },
    {
      "epoch": 0.9338315874020678,
      "grad_norm": 0.5190423460737542,
      "learning_rate": 0.00011551017403756108,
      "loss": 11.9937,
      "step": 17149
    },
    {
      "epoch": 0.933886041398651,
      "grad_norm": 0.5127265184209078,
      "learning_rate": 0.00011550146243323734,
      "loss": 12.0303,
      "step": 17150
    },
    {
      "epoch": 0.9339404953952339,
      "grad_norm": 0.6086177874858811,
      "learning_rate": 0.00011549275070837167,
      "loss": 12.2432,
      "step": 17151
    },
    {
      "epoch": 0.9339949493918169,
      "grad_norm": 0.5150423244136452,
      "learning_rate": 0.00011548403886303176,
      "loss": 12.1052,
      "step": 17152
    },
    {
      "epoch": 0.9340494033883999,
      "grad_norm": 0.546345240118378,
      "learning_rate": 0.00011547532689728535,
      "loss": 12.1896,
      "step": 17153
    },
    {
      "epoch": 0.9341038573849829,
      "grad_norm": 0.6130336189200207,
      "learning_rate": 0.0001154666148112002,
      "loss": 12.0921,
      "step": 17154
    },
    {
      "epoch": 0.9341583113815659,
      "grad_norm": 0.5476637616569271,
      "learning_rate": 0.00011545790260484409,
      "loss": 12.0947,
      "step": 17155
    },
    {
      "epoch": 0.934212765378149,
      "grad_norm": 0.5547513808519101,
      "learning_rate": 0.00011544919027828472,
      "loss": 12.0596,
      "step": 17156
    },
    {
      "epoch": 0.934267219374732,
      "grad_norm": 0.5931607387938834,
      "learning_rate": 0.00011544047783158984,
      "loss": 12.1686,
      "step": 17157
    },
    {
      "epoch": 0.934321673371315,
      "grad_norm": 0.599223671621129,
      "learning_rate": 0.00011543176526482722,
      "loss": 12.0938,
      "step": 17158
    },
    {
      "epoch": 0.934376127367898,
      "grad_norm": 0.565651878471124,
      "learning_rate": 0.00011542305257806463,
      "loss": 12.1767,
      "step": 17159
    },
    {
      "epoch": 0.934430581364481,
      "grad_norm": 0.5577459724178409,
      "learning_rate": 0.00011541433977136977,
      "loss": 12.078,
      "step": 17160
    },
    {
      "epoch": 0.934485035361064,
      "grad_norm": 0.5095102241397735,
      "learning_rate": 0.00011540562684481042,
      "loss": 12.1707,
      "step": 17161
    },
    {
      "epoch": 0.9345394893576471,
      "grad_norm": 0.568478015192735,
      "learning_rate": 0.00011539691379845435,
      "loss": 12.0611,
      "step": 17162
    },
    {
      "epoch": 0.9345939433542301,
      "grad_norm": 0.5619916172790094,
      "learning_rate": 0.00011538820063236928,
      "loss": 12.2408,
      "step": 17163
    },
    {
      "epoch": 0.9346483973508131,
      "grad_norm": 0.5480443047733544,
      "learning_rate": 0.00011537948734662299,
      "loss": 12.025,
      "step": 17164
    },
    {
      "epoch": 0.9347028513473961,
      "grad_norm": 0.49224784515548436,
      "learning_rate": 0.00011537077394128321,
      "loss": 12.029,
      "step": 17165
    },
    {
      "epoch": 0.934757305343979,
      "grad_norm": 0.49048885080856797,
      "learning_rate": 0.00011536206041641776,
      "loss": 12.1366,
      "step": 17166
    },
    {
      "epoch": 0.9348117593405622,
      "grad_norm": 0.519980158503926,
      "learning_rate": 0.00011535334677209431,
      "loss": 12.0942,
      "step": 17167
    },
    {
      "epoch": 0.9348662133371451,
      "grad_norm": 0.5725900985707221,
      "learning_rate": 0.00011534463300838067,
      "loss": 12.2885,
      "step": 17168
    },
    {
      "epoch": 0.9349206673337281,
      "grad_norm": 0.5930581777586704,
      "learning_rate": 0.0001153359191253446,
      "loss": 12.1438,
      "step": 17169
    },
    {
      "epoch": 0.9349751213303111,
      "grad_norm": 0.592854074271066,
      "learning_rate": 0.00011532720512305384,
      "loss": 12.1531,
      "step": 17170
    },
    {
      "epoch": 0.9350295753268941,
      "grad_norm": 0.5424309855938475,
      "learning_rate": 0.00011531849100157617,
      "loss": 12.1337,
      "step": 17171
    },
    {
      "epoch": 0.9350840293234771,
      "grad_norm": 0.5206493011432844,
      "learning_rate": 0.00011530977676097934,
      "loss": 12.1362,
      "step": 17172
    },
    {
      "epoch": 0.9351384833200602,
      "grad_norm": 0.5947327816436266,
      "learning_rate": 0.00011530106240133112,
      "loss": 12.1783,
      "step": 17173
    },
    {
      "epoch": 0.9351929373166432,
      "grad_norm": 0.5273619577220753,
      "learning_rate": 0.00011529234792269927,
      "loss": 12.0998,
      "step": 17174
    },
    {
      "epoch": 0.9352473913132262,
      "grad_norm": 0.5809187372992236,
      "learning_rate": 0.00011528363332515155,
      "loss": 11.9999,
      "step": 17175
    },
    {
      "epoch": 0.9353018453098092,
      "grad_norm": 0.5708666708407913,
      "learning_rate": 0.00011527491860875575,
      "loss": 12.1564,
      "step": 17176
    },
    {
      "epoch": 0.9353562993063922,
      "grad_norm": 0.5071641418210248,
      "learning_rate": 0.00011526620377357961,
      "loss": 12.0695,
      "step": 17177
    },
    {
      "epoch": 0.9354107533029752,
      "grad_norm": 0.5717476277001297,
      "learning_rate": 0.00011525748881969091,
      "loss": 12.1292,
      "step": 17178
    },
    {
      "epoch": 0.9354652072995583,
      "grad_norm": 0.48412158610907385,
      "learning_rate": 0.00011524877374715743,
      "loss": 12.0474,
      "step": 17179
    },
    {
      "epoch": 0.9355196612961413,
      "grad_norm": 0.6004876501095229,
      "learning_rate": 0.00011524005855604692,
      "loss": 12.107,
      "step": 17180
    },
    {
      "epoch": 0.9355741152927243,
      "grad_norm": 0.5485097559377725,
      "learning_rate": 0.0001152313432464272,
      "loss": 11.9742,
      "step": 17181
    },
    {
      "epoch": 0.9356285692893073,
      "grad_norm": 0.5582121852771219,
      "learning_rate": 0.00011522262781836593,
      "loss": 12.0482,
      "step": 17182
    },
    {
      "epoch": 0.9356830232858903,
      "grad_norm": 0.6497945649112843,
      "learning_rate": 0.000115213912271931,
      "loss": 12.0386,
      "step": 17183
    },
    {
      "epoch": 0.9357374772824733,
      "grad_norm": 0.5430460869746485,
      "learning_rate": 0.00011520519660719009,
      "loss": 12.0862,
      "step": 17184
    },
    {
      "epoch": 0.9357919312790564,
      "grad_norm": 0.5625820920059277,
      "learning_rate": 0.00011519648082421107,
      "loss": 12.1591,
      "step": 17185
    },
    {
      "epoch": 0.9358463852756393,
      "grad_norm": 0.6288034785425524,
      "learning_rate": 0.00011518776492306167,
      "loss": 12.2041,
      "step": 17186
    },
    {
      "epoch": 0.9359008392722223,
      "grad_norm": 0.5300944084181654,
      "learning_rate": 0.00011517904890380963,
      "loss": 12.162,
      "step": 17187
    },
    {
      "epoch": 0.9359552932688053,
      "grad_norm": 0.5571033320051584,
      "learning_rate": 0.00011517033276652276,
      "loss": 11.9701,
      "step": 17188
    },
    {
      "epoch": 0.9360097472653883,
      "grad_norm": 0.6130325143516594,
      "learning_rate": 0.00011516161651126884,
      "loss": 12.077,
      "step": 17189
    },
    {
      "epoch": 0.9360642012619713,
      "grad_norm": 0.5379809855893798,
      "learning_rate": 0.00011515290013811565,
      "loss": 12.1274,
      "step": 17190
    },
    {
      "epoch": 0.9361186552585544,
      "grad_norm": 0.5765900320169567,
      "learning_rate": 0.00011514418364713102,
      "loss": 12.1006,
      "step": 17191
    },
    {
      "epoch": 0.9361731092551374,
      "grad_norm": 0.5541974507656546,
      "learning_rate": 0.00011513546703838263,
      "loss": 12.2088,
      "step": 17192
    },
    {
      "epoch": 0.9362275632517204,
      "grad_norm": 0.6294734378681487,
      "learning_rate": 0.0001151267503119383,
      "loss": 12.2842,
      "step": 17193
    },
    {
      "epoch": 0.9362820172483034,
      "grad_norm": 0.4899631038419695,
      "learning_rate": 0.00011511803346786583,
      "loss": 12.1487,
      "step": 17194
    },
    {
      "epoch": 0.9363364712448864,
      "grad_norm": 0.6317754888221799,
      "learning_rate": 0.000115109316506233,
      "loss": 12.2118,
      "step": 17195
    },
    {
      "epoch": 0.9363909252414695,
      "grad_norm": 0.577960628241482,
      "learning_rate": 0.00011510059942710762,
      "loss": 12.0954,
      "step": 17196
    },
    {
      "epoch": 0.9364453792380525,
      "grad_norm": 0.5164315592205132,
      "learning_rate": 0.00011509188223055742,
      "loss": 12.0404,
      "step": 17197
    },
    {
      "epoch": 0.9364998332346355,
      "grad_norm": 0.5667785249426822,
      "learning_rate": 0.00011508316491665021,
      "loss": 12.1055,
      "step": 17198
    },
    {
      "epoch": 0.9365542872312185,
      "grad_norm": 0.5659445918451027,
      "learning_rate": 0.00011507444748545375,
      "loss": 12.0473,
      "step": 17199
    },
    {
      "epoch": 0.9366087412278015,
      "grad_norm": 0.5705775571743129,
      "learning_rate": 0.0001150657299370359,
      "loss": 12.1798,
      "step": 17200
    },
    {
      "epoch": 0.9366631952243845,
      "grad_norm": 0.5987243004380423,
      "learning_rate": 0.00011505701227146441,
      "loss": 12.0715,
      "step": 17201
    },
    {
      "epoch": 0.9367176492209676,
      "grad_norm": 0.5876089105774565,
      "learning_rate": 0.00011504829448880708,
      "loss": 12.0864,
      "step": 17202
    },
    {
      "epoch": 0.9367721032175506,
      "grad_norm": 0.5385016581645973,
      "learning_rate": 0.00011503957658913165,
      "loss": 12.1301,
      "step": 17203
    },
    {
      "epoch": 0.9368265572141335,
      "grad_norm": 0.6261101427354301,
      "learning_rate": 0.00011503085857250595,
      "loss": 12.2408,
      "step": 17204
    },
    {
      "epoch": 0.9368810112107165,
      "grad_norm": 0.7358535512856772,
      "learning_rate": 0.00011502214043899779,
      "loss": 12.2206,
      "step": 17205
    },
    {
      "epoch": 0.9369354652072995,
      "grad_norm": 0.5778289403753424,
      "learning_rate": 0.00011501342218867496,
      "loss": 12.141,
      "step": 17206
    },
    {
      "epoch": 0.9369899192038825,
      "grad_norm": 0.5286401393506325,
      "learning_rate": 0.00011500470382160524,
      "loss": 12.1795,
      "step": 17207
    },
    {
      "epoch": 0.9370443732004656,
      "grad_norm": 0.5701706314532361,
      "learning_rate": 0.0001149959853378564,
      "loss": 12.1545,
      "step": 17208
    },
    {
      "epoch": 0.9370988271970486,
      "grad_norm": 0.5960299180323986,
      "learning_rate": 0.0001149872667374963,
      "loss": 12.1785,
      "step": 17209
    },
    {
      "epoch": 0.9371532811936316,
      "grad_norm": 0.5829854024390061,
      "learning_rate": 0.00011497854802059265,
      "loss": 12.1051,
      "step": 17210
    },
    {
      "epoch": 0.9372077351902146,
      "grad_norm": 0.5881724224735968,
      "learning_rate": 0.00011496982918721333,
      "loss": 12.2929,
      "step": 17211
    },
    {
      "epoch": 0.9372621891867976,
      "grad_norm": 0.5541811460465499,
      "learning_rate": 0.00011496111023742611,
      "loss": 12.2226,
      "step": 17212
    },
    {
      "epoch": 0.9373166431833806,
      "grad_norm": 0.5486564396126202,
      "learning_rate": 0.0001149523911712988,
      "loss": 12.1091,
      "step": 17213
    },
    {
      "epoch": 0.9373710971799637,
      "grad_norm": 0.6097862643791315,
      "learning_rate": 0.00011494367198889915,
      "loss": 12.155,
      "step": 17214
    },
    {
      "epoch": 0.9374255511765467,
      "grad_norm": 0.545545428506642,
      "learning_rate": 0.00011493495269029501,
      "loss": 12.0725,
      "step": 17215
    },
    {
      "epoch": 0.9374800051731297,
      "grad_norm": 0.609667211079781,
      "learning_rate": 0.0001149262332755542,
      "loss": 12.0824,
      "step": 17216
    },
    {
      "epoch": 0.9375344591697127,
      "grad_norm": 0.5271418269924881,
      "learning_rate": 0.00011491751374474447,
      "loss": 12.1602,
      "step": 17217
    },
    {
      "epoch": 0.9375889131662957,
      "grad_norm": 0.5761514603166165,
      "learning_rate": 0.00011490879409793367,
      "loss": 12.1334,
      "step": 17218
    },
    {
      "epoch": 0.9376433671628787,
      "grad_norm": 0.5680582296778267,
      "learning_rate": 0.00011490007433518956,
      "loss": 12.1941,
      "step": 17219
    },
    {
      "epoch": 0.9376978211594618,
      "grad_norm": 0.6078517328718738,
      "learning_rate": 0.00011489135445658001,
      "loss": 12.0645,
      "step": 17220
    },
    {
      "epoch": 0.9377522751560448,
      "grad_norm": 0.509081853414556,
      "learning_rate": 0.00011488263446217278,
      "loss": 12.0702,
      "step": 17221
    },
    {
      "epoch": 0.9378067291526277,
      "grad_norm": 0.5501102501809502,
      "learning_rate": 0.00011487391435203568,
      "loss": 12.0714,
      "step": 17222
    },
    {
      "epoch": 0.9378611831492107,
      "grad_norm": 0.5744126645045073,
      "learning_rate": 0.00011486519412623654,
      "loss": 12.1208,
      "step": 17223
    },
    {
      "epoch": 0.9379156371457937,
      "grad_norm": 0.5587439549384924,
      "learning_rate": 0.00011485647378484312,
      "loss": 12.1202,
      "step": 17224
    },
    {
      "epoch": 0.9379700911423767,
      "grad_norm": 0.561473849628231,
      "learning_rate": 0.0001148477533279233,
      "loss": 12.1321,
      "step": 17225
    },
    {
      "epoch": 0.9380245451389598,
      "grad_norm": 0.5659402032513542,
      "learning_rate": 0.00011483903275554486,
      "loss": 12.0077,
      "step": 17226
    },
    {
      "epoch": 0.9380789991355428,
      "grad_norm": 0.5377675953612329,
      "learning_rate": 0.00011483031206777562,
      "loss": 12.1217,
      "step": 17227
    },
    {
      "epoch": 0.9381334531321258,
      "grad_norm": 0.5242989901422854,
      "learning_rate": 0.00011482159126468338,
      "loss": 12.1191,
      "step": 17228
    },
    {
      "epoch": 0.9381879071287088,
      "grad_norm": 0.5297059636906761,
      "learning_rate": 0.00011481287034633595,
      "loss": 12.09,
      "step": 17229
    },
    {
      "epoch": 0.9382423611252918,
      "grad_norm": 0.5568369256020095,
      "learning_rate": 0.00011480414931280114,
      "loss": 12.2279,
      "step": 17230
    },
    {
      "epoch": 0.9382968151218749,
      "grad_norm": 0.5244586564878074,
      "learning_rate": 0.00011479542816414681,
      "loss": 12.1117,
      "step": 17231
    },
    {
      "epoch": 0.9383512691184579,
      "grad_norm": 0.5413030521735377,
      "learning_rate": 0.00011478670690044075,
      "loss": 12.3001,
      "step": 17232
    },
    {
      "epoch": 0.9384057231150409,
      "grad_norm": 0.5701559086699557,
      "learning_rate": 0.00011477798552175076,
      "loss": 12.208,
      "step": 17233
    },
    {
      "epoch": 0.9384601771116239,
      "grad_norm": 0.512633275163757,
      "learning_rate": 0.0001147692640281447,
      "loss": 12.1601,
      "step": 17234
    },
    {
      "epoch": 0.9385146311082069,
      "grad_norm": 0.5953874673114348,
      "learning_rate": 0.00011476054241969035,
      "loss": 12.2078,
      "step": 17235
    },
    {
      "epoch": 0.9385690851047899,
      "grad_norm": 0.5853808960841238,
      "learning_rate": 0.00011475182069645556,
      "loss": 12.1906,
      "step": 17236
    },
    {
      "epoch": 0.938623539101373,
      "grad_norm": 0.5601311855180856,
      "learning_rate": 0.00011474309885850811,
      "loss": 12.1287,
      "step": 17237
    },
    {
      "epoch": 0.938677993097956,
      "grad_norm": 0.5993874750692453,
      "learning_rate": 0.00011473437690591589,
      "loss": 12.3005,
      "step": 17238
    },
    {
      "epoch": 0.938732447094539,
      "grad_norm": 0.5524879737952889,
      "learning_rate": 0.00011472565483874665,
      "loss": 12.2526,
      "step": 17239
    },
    {
      "epoch": 0.938786901091122,
      "grad_norm": 0.6498326786926967,
      "learning_rate": 0.00011471693265706827,
      "loss": 12.1835,
      "step": 17240
    },
    {
      "epoch": 0.9388413550877049,
      "grad_norm": 0.547793442948194,
      "learning_rate": 0.00011470821036094856,
      "loss": 12.0828,
      "step": 17241
    },
    {
      "epoch": 0.9388958090842879,
      "grad_norm": 0.5926849825805017,
      "learning_rate": 0.0001146994879504553,
      "loss": 12.305,
      "step": 17242
    },
    {
      "epoch": 0.938950263080871,
      "grad_norm": 0.5139655451009593,
      "learning_rate": 0.00011469076542565637,
      "loss": 12.0733,
      "step": 17243
    },
    {
      "epoch": 0.939004717077454,
      "grad_norm": 0.5806839240526883,
      "learning_rate": 0.00011468204278661961,
      "loss": 12.0816,
      "step": 17244
    },
    {
      "epoch": 0.939059171074037,
      "grad_norm": 0.5360932660865105,
      "learning_rate": 0.00011467332003341281,
      "loss": 12.1387,
      "step": 17245
    },
    {
      "epoch": 0.93911362507062,
      "grad_norm": 0.5959651623575238,
      "learning_rate": 0.00011466459716610382,
      "loss": 12.1715,
      "step": 17246
    },
    {
      "epoch": 0.939168079067203,
      "grad_norm": 0.49148132757862095,
      "learning_rate": 0.00011465587418476047,
      "loss": 12.1486,
      "step": 17247
    },
    {
      "epoch": 0.939222533063786,
      "grad_norm": 0.5828766323037756,
      "learning_rate": 0.00011464715108945055,
      "loss": 12.1267,
      "step": 17248
    },
    {
      "epoch": 0.9392769870603691,
      "grad_norm": 0.556492706857238,
      "learning_rate": 0.00011463842788024192,
      "loss": 12.1077,
      "step": 17249
    },
    {
      "epoch": 0.9393314410569521,
      "grad_norm": 0.529851964618464,
      "learning_rate": 0.00011462970455720246,
      "loss": 12.0681,
      "step": 17250
    },
    {
      "epoch": 0.9393858950535351,
      "grad_norm": 0.5119082388277009,
      "learning_rate": 0.00011462098112039997,
      "loss": 12.1816,
      "step": 17251
    },
    {
      "epoch": 0.9394403490501181,
      "grad_norm": 0.5067629636196034,
      "learning_rate": 0.00011461225756990226,
      "loss": 12.1203,
      "step": 17252
    },
    {
      "epoch": 0.9394948030467011,
      "grad_norm": 0.5617860450760956,
      "learning_rate": 0.00011460353390577716,
      "loss": 12.2252,
      "step": 17253
    },
    {
      "epoch": 0.9395492570432841,
      "grad_norm": 0.4995096624139038,
      "learning_rate": 0.00011459481012809256,
      "loss": 12.1493,
      "step": 17254
    },
    {
      "epoch": 0.9396037110398672,
      "grad_norm": 0.5723783085188626,
      "learning_rate": 0.00011458608623691627,
      "loss": 12.1434,
      "step": 17255
    },
    {
      "epoch": 0.9396581650364502,
      "grad_norm": 0.6267075493003642,
      "learning_rate": 0.00011457736223231612,
      "loss": 12.1304,
      "step": 17256
    },
    {
      "epoch": 0.9397126190330332,
      "grad_norm": 0.5694948368007134,
      "learning_rate": 0.00011456863811435998,
      "loss": 12.0896,
      "step": 17257
    },
    {
      "epoch": 0.9397670730296162,
      "grad_norm": 0.5344020892925807,
      "learning_rate": 0.00011455991388311564,
      "loss": 12.2124,
      "step": 17258
    },
    {
      "epoch": 0.9398215270261991,
      "grad_norm": 0.542197920612981,
      "learning_rate": 0.00011455118953865096,
      "loss": 12.1603,
      "step": 17259
    },
    {
      "epoch": 0.9398759810227821,
      "grad_norm": 0.5471556562301783,
      "learning_rate": 0.00011454246508103379,
      "loss": 12.0228,
      "step": 17260
    },
    {
      "epoch": 0.9399304350193652,
      "grad_norm": 0.5374465446352593,
      "learning_rate": 0.00011453374051033199,
      "loss": 12.2189,
      "step": 17261
    },
    {
      "epoch": 0.9399848890159482,
      "grad_norm": 0.5364038580856191,
      "learning_rate": 0.00011452501582661341,
      "loss": 12.2213,
      "step": 17262
    },
    {
      "epoch": 0.9400393430125312,
      "grad_norm": 0.5654931347362897,
      "learning_rate": 0.00011451629102994583,
      "loss": 12.1549,
      "step": 17263
    },
    {
      "epoch": 0.9400937970091142,
      "grad_norm": 0.6023270565342111,
      "learning_rate": 0.00011450756612039715,
      "loss": 12.0658,
      "step": 17264
    },
    {
      "epoch": 0.9401482510056972,
      "grad_norm": 0.5695175187289669,
      "learning_rate": 0.00011449884109803519,
      "loss": 12.2456,
      "step": 17265
    },
    {
      "epoch": 0.9402027050022803,
      "grad_norm": 0.6502306398488362,
      "learning_rate": 0.00011449011596292783,
      "loss": 12.2209,
      "step": 17266
    },
    {
      "epoch": 0.9402571589988633,
      "grad_norm": 0.5582934648368669,
      "learning_rate": 0.0001144813907151429,
      "loss": 12.1111,
      "step": 17267
    },
    {
      "epoch": 0.9403116129954463,
      "grad_norm": 0.6445743452834216,
      "learning_rate": 0.00011447266535474824,
      "loss": 12.1412,
      "step": 17268
    },
    {
      "epoch": 0.9403660669920293,
      "grad_norm": 0.5357891514697972,
      "learning_rate": 0.0001144639398818117,
      "loss": 12.1874,
      "step": 17269
    },
    {
      "epoch": 0.9404205209886123,
      "grad_norm": 0.5512280444023175,
      "learning_rate": 0.00011445521429640114,
      "loss": 12.1655,
      "step": 17270
    },
    {
      "epoch": 0.9404749749851953,
      "grad_norm": 0.625777175265489,
      "learning_rate": 0.0001144464885985844,
      "loss": 12.0704,
      "step": 17271
    },
    {
      "epoch": 0.9405294289817784,
      "grad_norm": 0.5525918408539039,
      "learning_rate": 0.00011443776278842937,
      "loss": 11.9734,
      "step": 17272
    },
    {
      "epoch": 0.9405838829783614,
      "grad_norm": 0.5754948191262231,
      "learning_rate": 0.00011442903686600386,
      "loss": 12.1846,
      "step": 17273
    },
    {
      "epoch": 0.9406383369749444,
      "grad_norm": 0.5790135252150729,
      "learning_rate": 0.00011442031083137574,
      "loss": 12.223,
      "step": 17274
    },
    {
      "epoch": 0.9406927909715274,
      "grad_norm": 0.5600864123416874,
      "learning_rate": 0.00011441158468461286,
      "loss": 12.0566,
      "step": 17275
    },
    {
      "epoch": 0.9407472449681104,
      "grad_norm": 0.5736911960390423,
      "learning_rate": 0.00011440285842578306,
      "loss": 12.1032,
      "step": 17276
    },
    {
      "epoch": 0.9408016989646933,
      "grad_norm": 0.5543856612987784,
      "learning_rate": 0.00011439413205495428,
      "loss": 12.0983,
      "step": 17277
    },
    {
      "epoch": 0.9408561529612764,
      "grad_norm": 0.5984601213406878,
      "learning_rate": 0.00011438540557219429,
      "loss": 12.2118,
      "step": 17278
    },
    {
      "epoch": 0.9409106069578594,
      "grad_norm": 0.5530215820344018,
      "learning_rate": 0.00011437667897757093,
      "loss": 12.2162,
      "step": 17279
    },
    {
      "epoch": 0.9409650609544424,
      "grad_norm": 0.5248977660305737,
      "learning_rate": 0.00011436795227115216,
      "loss": 12.1515,
      "step": 17280
    },
    {
      "epoch": 0.9410195149510254,
      "grad_norm": 0.5153105000297468,
      "learning_rate": 0.00011435922545300572,
      "loss": 12.0293,
      "step": 17281
    },
    {
      "epoch": 0.9410739689476084,
      "grad_norm": 0.5206047845654533,
      "learning_rate": 0.00011435049852319961,
      "loss": 12.1381,
      "step": 17282
    },
    {
      "epoch": 0.9411284229441914,
      "grad_norm": 0.5185415528814564,
      "learning_rate": 0.00011434177148180159,
      "loss": 12.0153,
      "step": 17283
    },
    {
      "epoch": 0.9411828769407745,
      "grad_norm": 0.5520501755180541,
      "learning_rate": 0.00011433304432887952,
      "loss": 12.1841,
      "step": 17284
    },
    {
      "epoch": 0.9412373309373575,
      "grad_norm": 0.6256029817077796,
      "learning_rate": 0.00011432431706450133,
      "loss": 12.119,
      "step": 17285
    },
    {
      "epoch": 0.9412917849339405,
      "grad_norm": 0.5374855220323205,
      "learning_rate": 0.00011431558968873482,
      "loss": 12.198,
      "step": 17286
    },
    {
      "epoch": 0.9413462389305235,
      "grad_norm": 0.5433292626029383,
      "learning_rate": 0.0001143068622016479,
      "loss": 12.0907,
      "step": 17287
    },
    {
      "epoch": 0.9414006929271065,
      "grad_norm": 0.5521975390496797,
      "learning_rate": 0.00011429813460330841,
      "loss": 12.1391,
      "step": 17288
    },
    {
      "epoch": 0.9414551469236895,
      "grad_norm": 0.5429237677242953,
      "learning_rate": 0.00011428940689378423,
      "loss": 12.1754,
      "step": 17289
    },
    {
      "epoch": 0.9415096009202726,
      "grad_norm": 0.6030906254884438,
      "learning_rate": 0.00011428067907314324,
      "loss": 12.2256,
      "step": 17290
    },
    {
      "epoch": 0.9415640549168556,
      "grad_norm": 0.5743037190203135,
      "learning_rate": 0.00011427195114145328,
      "loss": 11.9909,
      "step": 17291
    },
    {
      "epoch": 0.9416185089134386,
      "grad_norm": 0.5468077782858237,
      "learning_rate": 0.00011426322309878223,
      "loss": 12.0723,
      "step": 17292
    },
    {
      "epoch": 0.9416729629100216,
      "grad_norm": 0.5599611855486325,
      "learning_rate": 0.00011425449494519798,
      "loss": 12.0496,
      "step": 17293
    },
    {
      "epoch": 0.9417274169066046,
      "grad_norm": 0.5863866721261957,
      "learning_rate": 0.00011424576668076838,
      "loss": 12.0658,
      "step": 17294
    },
    {
      "epoch": 0.9417818709031875,
      "grad_norm": 0.562459945614839,
      "learning_rate": 0.00011423703830556132,
      "loss": 12.1998,
      "step": 17295
    },
    {
      "epoch": 0.9418363248997706,
      "grad_norm": 0.5531156188848322,
      "learning_rate": 0.00011422830981964465,
      "loss": 12.1009,
      "step": 17296
    },
    {
      "epoch": 0.9418907788963536,
      "grad_norm": 0.5583643188921626,
      "learning_rate": 0.00011421958122308625,
      "loss": 12.1192,
      "step": 17297
    },
    {
      "epoch": 0.9419452328929366,
      "grad_norm": 0.568445649499285,
      "learning_rate": 0.00011421085251595402,
      "loss": 12.1539,
      "step": 17298
    },
    {
      "epoch": 0.9419996868895196,
      "grad_norm": 0.5543496850795123,
      "learning_rate": 0.00011420212369831579,
      "loss": 12.1915,
      "step": 17299
    },
    {
      "epoch": 0.9420541408861026,
      "grad_norm": 0.6451922077957235,
      "learning_rate": 0.0001141933947702395,
      "loss": 12.1941,
      "step": 17300
    },
    {
      "epoch": 0.9421085948826857,
      "grad_norm": 0.5803207346318053,
      "learning_rate": 0.00011418466573179297,
      "loss": 12.1449,
      "step": 17301
    },
    {
      "epoch": 0.9421630488792687,
      "grad_norm": 0.5794351050123866,
      "learning_rate": 0.00011417593658304411,
      "loss": 12.1409,
      "step": 17302
    },
    {
      "epoch": 0.9422175028758517,
      "grad_norm": 0.6212903231713002,
      "learning_rate": 0.00011416720732406078,
      "loss": 12.2715,
      "step": 17303
    },
    {
      "epoch": 0.9422719568724347,
      "grad_norm": 0.550008672363222,
      "learning_rate": 0.00011415847795491088,
      "loss": 11.8332,
      "step": 17304
    },
    {
      "epoch": 0.9423264108690177,
      "grad_norm": 0.5297503330256259,
      "learning_rate": 0.00011414974847566226,
      "loss": 11.9181,
      "step": 17305
    },
    {
      "epoch": 0.9423808648656007,
      "grad_norm": 0.5472197784649815,
      "learning_rate": 0.00011414101888638284,
      "loss": 12.0817,
      "step": 17306
    },
    {
      "epoch": 0.9424353188621838,
      "grad_norm": 0.561480153899593,
      "learning_rate": 0.00011413228918714051,
      "loss": 12.2081,
      "step": 17307
    },
    {
      "epoch": 0.9424897728587668,
      "grad_norm": 0.5529541879321098,
      "learning_rate": 0.0001141235593780031,
      "loss": 12.109,
      "step": 17308
    },
    {
      "epoch": 0.9425442268553498,
      "grad_norm": 0.6121092589831865,
      "learning_rate": 0.00011411482945903853,
      "loss": 12.1352,
      "step": 17309
    },
    {
      "epoch": 0.9425986808519328,
      "grad_norm": 0.5248546324420222,
      "learning_rate": 0.00011410609943031467,
      "loss": 12.1414,
      "step": 17310
    },
    {
      "epoch": 0.9426531348485158,
      "grad_norm": 0.5544470836079637,
      "learning_rate": 0.00011409736929189943,
      "loss": 12.1372,
      "step": 17311
    },
    {
      "epoch": 0.9427075888450988,
      "grad_norm": 0.6278726221777126,
      "learning_rate": 0.00011408863904386068,
      "loss": 12.2407,
      "step": 17312
    },
    {
      "epoch": 0.9427620428416819,
      "grad_norm": 0.5198308450201492,
      "learning_rate": 0.0001140799086862663,
      "loss": 12.0772,
      "step": 17313
    },
    {
      "epoch": 0.9428164968382649,
      "grad_norm": 0.6000937497366587,
      "learning_rate": 0.00011407117821918419,
      "loss": 12.1906,
      "step": 17314
    },
    {
      "epoch": 0.9428709508348478,
      "grad_norm": 0.504756532291079,
      "learning_rate": 0.00011406244764268223,
      "loss": 12.0159,
      "step": 17315
    },
    {
      "epoch": 0.9429254048314308,
      "grad_norm": 0.547540736074649,
      "learning_rate": 0.00011405371695682834,
      "loss": 12.2065,
      "step": 17316
    },
    {
      "epoch": 0.9429798588280138,
      "grad_norm": 0.5870598282344961,
      "learning_rate": 0.00011404498616169039,
      "loss": 12.247,
      "step": 17317
    },
    {
      "epoch": 0.9430343128245968,
      "grad_norm": 0.6027365008204192,
      "learning_rate": 0.00011403625525733628,
      "loss": 12.1875,
      "step": 17318
    },
    {
      "epoch": 0.9430887668211799,
      "grad_norm": 0.6294453674112667,
      "learning_rate": 0.00011402752424383385,
      "loss": 12.1007,
      "step": 17319
    },
    {
      "epoch": 0.9431432208177629,
      "grad_norm": 0.60501481424706,
      "learning_rate": 0.00011401879312125108,
      "loss": 12.2884,
      "step": 17320
    },
    {
      "epoch": 0.9431976748143459,
      "grad_norm": 0.5559023358331184,
      "learning_rate": 0.0001140100618896558,
      "loss": 12.1927,
      "step": 17321
    },
    {
      "epoch": 0.9432521288109289,
      "grad_norm": 0.49646878847966375,
      "learning_rate": 0.00011400133054911597,
      "loss": 12.0381,
      "step": 17322
    },
    {
      "epoch": 0.9433065828075119,
      "grad_norm": 0.534139219927877,
      "learning_rate": 0.00011399259909969942,
      "loss": 12.1031,
      "step": 17323
    },
    {
      "epoch": 0.9433610368040949,
      "grad_norm": 0.5331067913106019,
      "learning_rate": 0.00011398386754147405,
      "loss": 12.1819,
      "step": 17324
    },
    {
      "epoch": 0.943415490800678,
      "grad_norm": 0.5361743784298774,
      "learning_rate": 0.00011397513587450779,
      "loss": 12.0945,
      "step": 17325
    },
    {
      "epoch": 0.943469944797261,
      "grad_norm": 0.5337030188842238,
      "learning_rate": 0.00011396640409886854,
      "loss": 12.1484,
      "step": 17326
    },
    {
      "epoch": 0.943524398793844,
      "grad_norm": 0.6535158266766455,
      "learning_rate": 0.00011395767221462421,
      "loss": 12.2561,
      "step": 17327
    },
    {
      "epoch": 0.943578852790427,
      "grad_norm": 0.5496197226487632,
      "learning_rate": 0.00011394894022184266,
      "loss": 12.0781,
      "step": 17328
    },
    {
      "epoch": 0.94363330678701,
      "grad_norm": 0.5964472425795224,
      "learning_rate": 0.0001139402081205918,
      "loss": 12.1233,
      "step": 17329
    },
    {
      "epoch": 0.9436877607835931,
      "grad_norm": 0.5233459287429948,
      "learning_rate": 0.00011393147591093954,
      "loss": 12.2202,
      "step": 17330
    },
    {
      "epoch": 0.9437422147801761,
      "grad_norm": 0.5396707815865741,
      "learning_rate": 0.00011392274359295381,
      "loss": 12.0967,
      "step": 17331
    },
    {
      "epoch": 0.943796668776759,
      "grad_norm": 0.5324413065684591,
      "learning_rate": 0.00011391401116670248,
      "loss": 12.097,
      "step": 17332
    },
    {
      "epoch": 0.943851122773342,
      "grad_norm": 0.518466699604795,
      "learning_rate": 0.00011390527863225349,
      "loss": 12.0776,
      "step": 17333
    },
    {
      "epoch": 0.943905576769925,
      "grad_norm": 0.6130210757512741,
      "learning_rate": 0.00011389654598967469,
      "loss": 12.174,
      "step": 17334
    },
    {
      "epoch": 0.943960030766508,
      "grad_norm": 0.5183145927597441,
      "learning_rate": 0.00011388781323903403,
      "loss": 12.1552,
      "step": 17335
    },
    {
      "epoch": 0.9440144847630911,
      "grad_norm": 0.5357415610664379,
      "learning_rate": 0.00011387908038039938,
      "loss": 12.1089,
      "step": 17336
    },
    {
      "epoch": 0.9440689387596741,
      "grad_norm": 0.5192848791681334,
      "learning_rate": 0.00011387034741383872,
      "loss": 12.0486,
      "step": 17337
    },
    {
      "epoch": 0.9441233927562571,
      "grad_norm": 0.5211735670403752,
      "learning_rate": 0.00011386161433941988,
      "loss": 12.2151,
      "step": 17338
    },
    {
      "epoch": 0.9441778467528401,
      "grad_norm": 0.5316633016104322,
      "learning_rate": 0.00011385288115721082,
      "loss": 12.0402,
      "step": 17339
    },
    {
      "epoch": 0.9442323007494231,
      "grad_norm": 0.5882106176391751,
      "learning_rate": 0.00011384414786727942,
      "loss": 12.1058,
      "step": 17340
    },
    {
      "epoch": 0.9442867547460061,
      "grad_norm": 0.6081795735550459,
      "learning_rate": 0.00011383541446969362,
      "loss": 12.1262,
      "step": 17341
    },
    {
      "epoch": 0.9443412087425892,
      "grad_norm": 0.50259690448361,
      "learning_rate": 0.0001138266809645213,
      "loss": 12.0509,
      "step": 17342
    },
    {
      "epoch": 0.9443956627391722,
      "grad_norm": 0.5571861587041949,
      "learning_rate": 0.0001138179473518304,
      "loss": 12.1127,
      "step": 17343
    },
    {
      "epoch": 0.9444501167357552,
      "grad_norm": 0.5622518601050132,
      "learning_rate": 0.00011380921363168882,
      "loss": 12.1726,
      "step": 17344
    },
    {
      "epoch": 0.9445045707323382,
      "grad_norm": 0.5339367631474244,
      "learning_rate": 0.00011380047980416449,
      "loss": 12.1181,
      "step": 17345
    },
    {
      "epoch": 0.9445590247289212,
      "grad_norm": 0.5763643587724595,
      "learning_rate": 0.00011379174586932528,
      "loss": 12.0472,
      "step": 17346
    },
    {
      "epoch": 0.9446134787255042,
      "grad_norm": 0.53330973830854,
      "learning_rate": 0.00011378301182723918,
      "loss": 12.0635,
      "step": 17347
    },
    {
      "epoch": 0.9446679327220873,
      "grad_norm": 0.5257490232366746,
      "learning_rate": 0.00011377427767797404,
      "loss": 12.1081,
      "step": 17348
    },
    {
      "epoch": 0.9447223867186703,
      "grad_norm": 0.6136431341982396,
      "learning_rate": 0.00011376554342159785,
      "loss": 12.2335,
      "step": 17349
    },
    {
      "epoch": 0.9447768407152533,
      "grad_norm": 0.546794308497035,
      "learning_rate": 0.00011375680905817844,
      "loss": 12.1004,
      "step": 17350
    },
    {
      "epoch": 0.9448312947118362,
      "grad_norm": 0.6030840671502765,
      "learning_rate": 0.00011374807458778378,
      "loss": 12.1877,
      "step": 17351
    },
    {
      "epoch": 0.9448857487084192,
      "grad_norm": 0.5508296104098631,
      "learning_rate": 0.00011373934001048181,
      "loss": 12.237,
      "step": 17352
    },
    {
      "epoch": 0.9449402027050022,
      "grad_norm": 0.5684494836431603,
      "learning_rate": 0.00011373060532634041,
      "loss": 12.0696,
      "step": 17353
    },
    {
      "epoch": 0.9449946567015853,
      "grad_norm": 0.5099595884068093,
      "learning_rate": 0.00011372187053542753,
      "loss": 12.0227,
      "step": 17354
    },
    {
      "epoch": 0.9450491106981683,
      "grad_norm": 0.6102015865848511,
      "learning_rate": 0.00011371313563781107,
      "loss": 12.2132,
      "step": 17355
    },
    {
      "epoch": 0.9451035646947513,
      "grad_norm": 0.5640477794515227,
      "learning_rate": 0.00011370440063355898,
      "loss": 12.1198,
      "step": 17356
    },
    {
      "epoch": 0.9451580186913343,
      "grad_norm": 0.6693167233760415,
      "learning_rate": 0.00011369566552273919,
      "loss": 12.268,
      "step": 17357
    },
    {
      "epoch": 0.9452124726879173,
      "grad_norm": 0.5674765937082062,
      "learning_rate": 0.0001136869303054196,
      "loss": 12.066,
      "step": 17358
    },
    {
      "epoch": 0.9452669266845003,
      "grad_norm": 0.5607735624035676,
      "learning_rate": 0.00011367819498166812,
      "loss": 12.1589,
      "step": 17359
    },
    {
      "epoch": 0.9453213806810834,
      "grad_norm": 0.5681497183526665,
      "learning_rate": 0.00011366945955155271,
      "loss": 12.0458,
      "step": 17360
    },
    {
      "epoch": 0.9453758346776664,
      "grad_norm": 0.5625421382395502,
      "learning_rate": 0.00011366072401514129,
      "loss": 12.0679,
      "step": 17361
    },
    {
      "epoch": 0.9454302886742494,
      "grad_norm": 0.5649134208134503,
      "learning_rate": 0.00011365198837250182,
      "loss": 12.1959,
      "step": 17362
    },
    {
      "epoch": 0.9454847426708324,
      "grad_norm": 0.5822452562809511,
      "learning_rate": 0.00011364325262370215,
      "loss": 12.2539,
      "step": 17363
    },
    {
      "epoch": 0.9455391966674154,
      "grad_norm": 0.6101939104399868,
      "learning_rate": 0.0001136345167688103,
      "loss": 11.9465,
      "step": 17364
    },
    {
      "epoch": 0.9455936506639985,
      "grad_norm": 0.5451417122742256,
      "learning_rate": 0.00011362578080789413,
      "loss": 12.1123,
      "step": 17365
    },
    {
      "epoch": 0.9456481046605815,
      "grad_norm": 0.5445237279344438,
      "learning_rate": 0.00011361704474102162,
      "loss": 12.102,
      "step": 17366
    },
    {
      "epoch": 0.9457025586571645,
      "grad_norm": 0.5558014321065162,
      "learning_rate": 0.0001136083085682607,
      "loss": 12.1779,
      "step": 17367
    },
    {
      "epoch": 0.9457570126537475,
      "grad_norm": 0.5443084486631766,
      "learning_rate": 0.00011359957228967926,
      "loss": 12.0391,
      "step": 17368
    },
    {
      "epoch": 0.9458114666503304,
      "grad_norm": 0.5223724320434633,
      "learning_rate": 0.00011359083590534527,
      "loss": 12.0774,
      "step": 17369
    },
    {
      "epoch": 0.9458659206469134,
      "grad_norm": 0.6259468637705189,
      "learning_rate": 0.00011358209941532668,
      "loss": 12.2839,
      "step": 17370
    },
    {
      "epoch": 0.9459203746434965,
      "grad_norm": 0.501076660026277,
      "learning_rate": 0.0001135733628196914,
      "loss": 12.1208,
      "step": 17371
    },
    {
      "epoch": 0.9459748286400795,
      "grad_norm": 0.5180692728575292,
      "learning_rate": 0.0001135646261185074,
      "loss": 12.0093,
      "step": 17372
    },
    {
      "epoch": 0.9460292826366625,
      "grad_norm": 0.48719332354599426,
      "learning_rate": 0.00011355588931184256,
      "loss": 12.1396,
      "step": 17373
    },
    {
      "epoch": 0.9460837366332455,
      "grad_norm": 0.6126639995736392,
      "learning_rate": 0.00011354715239976483,
      "loss": 12.271,
      "step": 17374
    },
    {
      "epoch": 0.9461381906298285,
      "grad_norm": 0.5393699437816664,
      "learning_rate": 0.00011353841538234221,
      "loss": 12.1046,
      "step": 17375
    },
    {
      "epoch": 0.9461926446264115,
      "grad_norm": 0.5420248069570972,
      "learning_rate": 0.00011352967825964259,
      "loss": 12.1104,
      "step": 17376
    },
    {
      "epoch": 0.9462470986229946,
      "grad_norm": 0.5331460613546519,
      "learning_rate": 0.00011352094103173394,
      "loss": 12.0273,
      "step": 17377
    },
    {
      "epoch": 0.9463015526195776,
      "grad_norm": 0.4752978198092573,
      "learning_rate": 0.00011351220369868416,
      "loss": 12.2514,
      "step": 17378
    },
    {
      "epoch": 0.9463560066161606,
      "grad_norm": 0.6074988970316852,
      "learning_rate": 0.00011350346626056121,
      "loss": 12.2263,
      "step": 17379
    },
    {
      "epoch": 0.9464104606127436,
      "grad_norm": 0.5837326630588333,
      "learning_rate": 0.00011349472871743306,
      "loss": 12.1478,
      "step": 17380
    },
    {
      "epoch": 0.9464649146093266,
      "grad_norm": 0.7559202053192198,
      "learning_rate": 0.00011348599106936762,
      "loss": 12.2578,
      "step": 17381
    },
    {
      "epoch": 0.9465193686059096,
      "grad_norm": 0.6222093173873497,
      "learning_rate": 0.00011347725331643289,
      "loss": 12.2416,
      "step": 17382
    },
    {
      "epoch": 0.9465738226024927,
      "grad_norm": 0.5169883056107297,
      "learning_rate": 0.00011346851545869674,
      "loss": 12.0261,
      "step": 17383
    },
    {
      "epoch": 0.9466282765990757,
      "grad_norm": 0.5574210783211934,
      "learning_rate": 0.00011345977749622718,
      "loss": 12.1017,
      "step": 17384
    },
    {
      "epoch": 0.9466827305956587,
      "grad_norm": 0.6661342012506033,
      "learning_rate": 0.0001134510394290921,
      "loss": 12.1888,
      "step": 17385
    },
    {
      "epoch": 0.9467371845922417,
      "grad_norm": 0.5531060894825526,
      "learning_rate": 0.0001134423012573595,
      "loss": 12.1628,
      "step": 17386
    },
    {
      "epoch": 0.9467916385888246,
      "grad_norm": 1.0749222571419381,
      "learning_rate": 0.00011343356298109732,
      "loss": 12.1868,
      "step": 17387
    },
    {
      "epoch": 0.9468460925854076,
      "grad_norm": 0.5015559730252587,
      "learning_rate": 0.00011342482460037352,
      "loss": 12.1034,
      "step": 17388
    },
    {
      "epoch": 0.9469005465819907,
      "grad_norm": 0.5686342159403498,
      "learning_rate": 0.000113416086115256,
      "loss": 12.0834,
      "step": 17389
    },
    {
      "epoch": 0.9469550005785737,
      "grad_norm": 0.5218876135221702,
      "learning_rate": 0.00011340734752581274,
      "loss": 12.0357,
      "step": 17390
    },
    {
      "epoch": 0.9470094545751567,
      "grad_norm": 0.5554576526720056,
      "learning_rate": 0.00011339860883211171,
      "loss": 12.1212,
      "step": 17391
    },
    {
      "epoch": 0.9470639085717397,
      "grad_norm": 0.6075109268687903,
      "learning_rate": 0.00011338987003422086,
      "loss": 12.1669,
      "step": 17392
    },
    {
      "epoch": 0.9471183625683227,
      "grad_norm": 0.5674266591827817,
      "learning_rate": 0.00011338113113220814,
      "loss": 12.0762,
      "step": 17393
    },
    {
      "epoch": 0.9471728165649057,
      "grad_norm": 0.5507636381804221,
      "learning_rate": 0.00011337239212614148,
      "loss": 12.0855,
      "step": 17394
    },
    {
      "epoch": 0.9472272705614888,
      "grad_norm": 0.5326074009086318,
      "learning_rate": 0.00011336365301608887,
      "loss": 12.0757,
      "step": 17395
    },
    {
      "epoch": 0.9472817245580718,
      "grad_norm": 0.5373601830555752,
      "learning_rate": 0.00011335491380211823,
      "loss": 12.04,
      "step": 17396
    },
    {
      "epoch": 0.9473361785546548,
      "grad_norm": 0.5781711851409429,
      "learning_rate": 0.00011334617448429754,
      "loss": 12.1499,
      "step": 17397
    },
    {
      "epoch": 0.9473906325512378,
      "grad_norm": 0.62400690303693,
      "learning_rate": 0.00011333743506269479,
      "loss": 12.3313,
      "step": 17398
    },
    {
      "epoch": 0.9474450865478208,
      "grad_norm": 0.5670385056747921,
      "learning_rate": 0.0001133286955373779,
      "loss": 11.9934,
      "step": 17399
    },
    {
      "epoch": 0.9474995405444039,
      "grad_norm": 0.5790623640026985,
      "learning_rate": 0.00011331995590841482,
      "loss": 12.3538,
      "step": 17400
    },
    {
      "epoch": 0.9475539945409869,
      "grad_norm": 0.4869641044386059,
      "learning_rate": 0.00011331121617587355,
      "loss": 12.131,
      "step": 17401
    },
    {
      "epoch": 0.9476084485375699,
      "grad_norm": 0.605769205367832,
      "learning_rate": 0.000113302476339822,
      "loss": 12.2189,
      "step": 17402
    },
    {
      "epoch": 0.9476629025341529,
      "grad_norm": 0.5477340165700033,
      "learning_rate": 0.00011329373640032821,
      "loss": 12.2405,
      "step": 17403
    },
    {
      "epoch": 0.9477173565307359,
      "grad_norm": 0.5295438034670991,
      "learning_rate": 0.00011328499635746009,
      "loss": 12.1444,
      "step": 17404
    },
    {
      "epoch": 0.9477718105273188,
      "grad_norm": 0.5227246506383778,
      "learning_rate": 0.00011327625621128556,
      "loss": 12.128,
      "step": 17405
    },
    {
      "epoch": 0.947826264523902,
      "grad_norm": 0.5701971780820198,
      "learning_rate": 0.00011326751596187269,
      "loss": 12.0952,
      "step": 17406
    },
    {
      "epoch": 0.9478807185204849,
      "grad_norm": 0.5291297001215025,
      "learning_rate": 0.00011325877560928932,
      "loss": 12.0341,
      "step": 17407
    },
    {
      "epoch": 0.9479351725170679,
      "grad_norm": 0.561216245895998,
      "learning_rate": 0.00011325003515360357,
      "loss": 12.0441,
      "step": 17408
    },
    {
      "epoch": 0.9479896265136509,
      "grad_norm": 0.7248037630493238,
      "learning_rate": 0.00011324129459488329,
      "loss": 12.0241,
      "step": 17409
    },
    {
      "epoch": 0.9480440805102339,
      "grad_norm": 0.5478144736346154,
      "learning_rate": 0.00011323255393319647,
      "loss": 12.1368,
      "step": 17410
    },
    {
      "epoch": 0.9480985345068169,
      "grad_norm": 0.5832870659400218,
      "learning_rate": 0.00011322381316861112,
      "loss": 12.2738,
      "step": 17411
    },
    {
      "epoch": 0.9481529885034,
      "grad_norm": 0.5652275283387745,
      "learning_rate": 0.00011321507230119517,
      "loss": 12.1513,
      "step": 17412
    },
    {
      "epoch": 0.948207442499983,
      "grad_norm": 0.538845252796443,
      "learning_rate": 0.00011320633133101659,
      "loss": 12.0958,
      "step": 17413
    },
    {
      "epoch": 0.948261896496566,
      "grad_norm": 0.5822453906528845,
      "learning_rate": 0.00011319759025814335,
      "loss": 12.1057,
      "step": 17414
    },
    {
      "epoch": 0.948316350493149,
      "grad_norm": 0.6519578863689457,
      "learning_rate": 0.00011318884908264347,
      "loss": 12.3193,
      "step": 17415
    },
    {
      "epoch": 0.948370804489732,
      "grad_norm": 0.602744532893201,
      "learning_rate": 0.00011318010780458488,
      "loss": 12.1319,
      "step": 17416
    },
    {
      "epoch": 0.948425258486315,
      "grad_norm": 0.6262288472678893,
      "learning_rate": 0.00011317136642403554,
      "loss": 12.0864,
      "step": 17417
    },
    {
      "epoch": 0.9484797124828981,
      "grad_norm": 0.5392669614315889,
      "learning_rate": 0.00011316262494106347,
      "loss": 12.2444,
      "step": 17418
    },
    {
      "epoch": 0.9485341664794811,
      "grad_norm": 0.6239141160838986,
      "learning_rate": 0.0001131538833557366,
      "loss": 12.1257,
      "step": 17419
    },
    {
      "epoch": 0.9485886204760641,
      "grad_norm": 0.6145631515186998,
      "learning_rate": 0.00011314514166812295,
      "loss": 12.1822,
      "step": 17420
    },
    {
      "epoch": 0.9486430744726471,
      "grad_norm": 0.596281164024472,
      "learning_rate": 0.00011313639987829046,
      "loss": 12.1477,
      "step": 17421
    },
    {
      "epoch": 0.9486975284692301,
      "grad_norm": 0.5929985373760998,
      "learning_rate": 0.00011312765798630711,
      "loss": 12.0916,
      "step": 17422
    },
    {
      "epoch": 0.948751982465813,
      "grad_norm": 0.8604325299052797,
      "learning_rate": 0.00011311891599224092,
      "loss": 11.9498,
      "step": 17423
    },
    {
      "epoch": 0.9488064364623962,
      "grad_norm": 0.5983641398134796,
      "learning_rate": 0.00011311017389615981,
      "loss": 12.1452,
      "step": 17424
    },
    {
      "epoch": 0.9488608904589791,
      "grad_norm": 0.5306545391820514,
      "learning_rate": 0.0001131014316981318,
      "loss": 12.0887,
      "step": 17425
    },
    {
      "epoch": 0.9489153444555621,
      "grad_norm": 0.6019247104345119,
      "learning_rate": 0.00011309268939822486,
      "loss": 12.0831,
      "step": 17426
    },
    {
      "epoch": 0.9489697984521451,
      "grad_norm": 0.6030923956429187,
      "learning_rate": 0.00011308394699650697,
      "loss": 12.1672,
      "step": 17427
    },
    {
      "epoch": 0.9490242524487281,
      "grad_norm": 0.5375711174347375,
      "learning_rate": 0.00011307520449304614,
      "loss": 12.0472,
      "step": 17428
    },
    {
      "epoch": 0.9490787064453111,
      "grad_norm": 0.5398776364743896,
      "learning_rate": 0.0001130664618879103,
      "loss": 12.2488,
      "step": 17429
    },
    {
      "epoch": 0.9491331604418942,
      "grad_norm": 0.543823949854254,
      "learning_rate": 0.00011305771918116746,
      "loss": 12.1812,
      "step": 17430
    },
    {
      "epoch": 0.9491876144384772,
      "grad_norm": 0.5784847529335118,
      "learning_rate": 0.00011304897637288561,
      "loss": 12.0837,
      "step": 17431
    },
    {
      "epoch": 0.9492420684350602,
      "grad_norm": 0.5714020178424278,
      "learning_rate": 0.00011304023346313273,
      "loss": 12.1253,
      "step": 17432
    },
    {
      "epoch": 0.9492965224316432,
      "grad_norm": 0.4896225675110771,
      "learning_rate": 0.00011303149045197682,
      "loss": 12.0577,
      "step": 17433
    },
    {
      "epoch": 0.9493509764282262,
      "grad_norm": 0.5872673682035371,
      "learning_rate": 0.00011302274733948583,
      "loss": 12.1131,
      "step": 17434
    },
    {
      "epoch": 0.9494054304248093,
      "grad_norm": 0.5501561471508206,
      "learning_rate": 0.00011301400412572781,
      "loss": 12.1594,
      "step": 17435
    },
    {
      "epoch": 0.9494598844213923,
      "grad_norm": 0.6206510333411461,
      "learning_rate": 0.00011300526081077068,
      "loss": 12.3073,
      "step": 17436
    },
    {
      "epoch": 0.9495143384179753,
      "grad_norm": 0.5840100543707513,
      "learning_rate": 0.00011299651739468246,
      "loss": 12.17,
      "step": 17437
    },
    {
      "epoch": 0.9495687924145583,
      "grad_norm": 0.6196232425988139,
      "learning_rate": 0.00011298777387753118,
      "loss": 12.1427,
      "step": 17438
    },
    {
      "epoch": 0.9496232464111413,
      "grad_norm": 0.70709403486026,
      "learning_rate": 0.00011297903025938476,
      "loss": 12.1109,
      "step": 17439
    },
    {
      "epoch": 0.9496777004077243,
      "grad_norm": 0.6074900970206965,
      "learning_rate": 0.00011297028654031121,
      "loss": 12.0835,
      "step": 17440
    },
    {
      "epoch": 0.9497321544043074,
      "grad_norm": 0.6208250634797887,
      "learning_rate": 0.00011296154272037856,
      "loss": 12.1172,
      "step": 17441
    },
    {
      "epoch": 0.9497866084008904,
      "grad_norm": 0.5726205600853366,
      "learning_rate": 0.00011295279879965477,
      "loss": 12.1401,
      "step": 17442
    },
    {
      "epoch": 0.9498410623974733,
      "grad_norm": 0.5696277608532468,
      "learning_rate": 0.00011294405477820787,
      "loss": 12.2064,
      "step": 17443
    },
    {
      "epoch": 0.9498955163940563,
      "grad_norm": 0.5569385381906664,
      "learning_rate": 0.00011293531065610581,
      "loss": 12.1694,
      "step": 17444
    },
    {
      "epoch": 0.9499499703906393,
      "grad_norm": 0.6089017667986101,
      "learning_rate": 0.00011292656643341659,
      "loss": 12.1573,
      "step": 17445
    },
    {
      "epoch": 0.9500044243872223,
      "grad_norm": 0.5086641759852797,
      "learning_rate": 0.00011291782211020823,
      "loss": 12.098,
      "step": 17446
    },
    {
      "epoch": 0.9500588783838054,
      "grad_norm": 0.5903911370652851,
      "learning_rate": 0.00011290907768654872,
      "loss": 12.1332,
      "step": 17447
    },
    {
      "epoch": 0.9501133323803884,
      "grad_norm": 0.5118851465113656,
      "learning_rate": 0.00011290033316250608,
      "loss": 11.9755,
      "step": 17448
    },
    {
      "epoch": 0.9501677863769714,
      "grad_norm": 0.5796184135739372,
      "learning_rate": 0.00011289158853814827,
      "loss": 12.2098,
      "step": 17449
    },
    {
      "epoch": 0.9502222403735544,
      "grad_norm": 0.5154883143599293,
      "learning_rate": 0.0001128828438135433,
      "loss": 11.988,
      "step": 17450
    },
    {
      "epoch": 0.9502766943701374,
      "grad_norm": 0.5610163462632073,
      "learning_rate": 0.00011287409898875916,
      "loss": 12.2074,
      "step": 17451
    },
    {
      "epoch": 0.9503311483667204,
      "grad_norm": 0.5747716835570177,
      "learning_rate": 0.00011286535406386389,
      "loss": 12.079,
      "step": 17452
    },
    {
      "epoch": 0.9503856023633035,
      "grad_norm": 0.5182737197924299,
      "learning_rate": 0.0001128566090389255,
      "loss": 12.1829,
      "step": 17453
    },
    {
      "epoch": 0.9504400563598865,
      "grad_norm": 0.5359861646375332,
      "learning_rate": 0.00011284786391401191,
      "loss": 12.1085,
      "step": 17454
    },
    {
      "epoch": 0.9504945103564695,
      "grad_norm": 0.6373633744777849,
      "learning_rate": 0.00011283911868919119,
      "loss": 12.1897,
      "step": 17455
    },
    {
      "epoch": 0.9505489643530525,
      "grad_norm": 0.5455363959029207,
      "learning_rate": 0.00011283037336453132,
      "loss": 12.1452,
      "step": 17456
    },
    {
      "epoch": 0.9506034183496355,
      "grad_norm": 0.5436581687885378,
      "learning_rate": 0.00011282162794010034,
      "loss": 12.1088,
      "step": 17457
    },
    {
      "epoch": 0.9506578723462185,
      "grad_norm": 0.5684905876488265,
      "learning_rate": 0.00011281288241596624,
      "loss": 12.2331,
      "step": 17458
    },
    {
      "epoch": 0.9507123263428016,
      "grad_norm": 0.5930821930183624,
      "learning_rate": 0.000112804136792197,
      "loss": 11.9246,
      "step": 17459
    },
    {
      "epoch": 0.9507667803393846,
      "grad_norm": 0.5410520640706985,
      "learning_rate": 0.00011279539106886064,
      "loss": 12.033,
      "step": 17460
    },
    {
      "epoch": 0.9508212343359675,
      "grad_norm": 0.5541788671385539,
      "learning_rate": 0.00011278664524602516,
      "loss": 12.1253,
      "step": 17461
    },
    {
      "epoch": 0.9508756883325505,
      "grad_norm": 0.5638897814342899,
      "learning_rate": 0.00011277789932375858,
      "loss": 12.1153,
      "step": 17462
    },
    {
      "epoch": 0.9509301423291335,
      "grad_norm": 0.5427936828451158,
      "learning_rate": 0.00011276915330212894,
      "loss": 11.9585,
      "step": 17463
    },
    {
      "epoch": 0.9509845963257166,
      "grad_norm": 0.6050927126754538,
      "learning_rate": 0.00011276040718120422,
      "loss": 12.1493,
      "step": 17464
    },
    {
      "epoch": 0.9510390503222996,
      "grad_norm": 0.5121729638018014,
      "learning_rate": 0.00011275166096105243,
      "loss": 11.9161,
      "step": 17465
    },
    {
      "epoch": 0.9510935043188826,
      "grad_norm": 0.5349722848715646,
      "learning_rate": 0.00011274291464174158,
      "loss": 12.0515,
      "step": 17466
    },
    {
      "epoch": 0.9511479583154656,
      "grad_norm": 0.6259830681803067,
      "learning_rate": 0.00011273416822333969,
      "loss": 12.1932,
      "step": 17467
    },
    {
      "epoch": 0.9512024123120486,
      "grad_norm": 0.5447584965802076,
      "learning_rate": 0.00011272542170591478,
      "loss": 12.1151,
      "step": 17468
    },
    {
      "epoch": 0.9512568663086316,
      "grad_norm": 0.5172873978728029,
      "learning_rate": 0.00011271667508953485,
      "loss": 12.1788,
      "step": 17469
    },
    {
      "epoch": 0.9513113203052147,
      "grad_norm": 0.5173157791537659,
      "learning_rate": 0.00011270792837426791,
      "loss": 12.042,
      "step": 17470
    },
    {
      "epoch": 0.9513657743017977,
      "grad_norm": 0.6380131731191625,
      "learning_rate": 0.000112699181560182,
      "loss": 12.0384,
      "step": 17471
    },
    {
      "epoch": 0.9514202282983807,
      "grad_norm": 0.5753540757727378,
      "learning_rate": 0.00011269043464734513,
      "loss": 12.0551,
      "step": 17472
    },
    {
      "epoch": 0.9514746822949637,
      "grad_norm": 0.7165165108917677,
      "learning_rate": 0.00011268168763582529,
      "loss": 12.3405,
      "step": 17473
    },
    {
      "epoch": 0.9515291362915467,
      "grad_norm": 0.5157512469301627,
      "learning_rate": 0.00011267294052569055,
      "loss": 12.0039,
      "step": 17474
    },
    {
      "epoch": 0.9515835902881297,
      "grad_norm": 0.5786763748974254,
      "learning_rate": 0.00011266419331700888,
      "loss": 12.241,
      "step": 17475
    },
    {
      "epoch": 0.9516380442847128,
      "grad_norm": 0.5817474153048059,
      "learning_rate": 0.00011265544600984831,
      "loss": 12.1188,
      "step": 17476
    },
    {
      "epoch": 0.9516924982812958,
      "grad_norm": 0.5102441058380783,
      "learning_rate": 0.0001126466986042769,
      "loss": 12.0575,
      "step": 17477
    },
    {
      "epoch": 0.9517469522778788,
      "grad_norm": 0.6148685340058035,
      "learning_rate": 0.00011263795110036261,
      "loss": 12.2371,
      "step": 17478
    },
    {
      "epoch": 0.9518014062744617,
      "grad_norm": 0.6335387459693504,
      "learning_rate": 0.00011262920349817352,
      "loss": 12.0826,
      "step": 17479
    },
    {
      "epoch": 0.9518558602710447,
      "grad_norm": 0.5463841829279674,
      "learning_rate": 0.00011262045579777763,
      "loss": 12.0817,
      "step": 17480
    },
    {
      "epoch": 0.9519103142676277,
      "grad_norm": 0.5303469750614441,
      "learning_rate": 0.00011261170799924291,
      "loss": 12.1507,
      "step": 17481
    },
    {
      "epoch": 0.9519647682642108,
      "grad_norm": 0.5621517509390335,
      "learning_rate": 0.00011260296010263749,
      "loss": 12.1779,
      "step": 17482
    },
    {
      "epoch": 0.9520192222607938,
      "grad_norm": 0.5725064239488651,
      "learning_rate": 0.00011259421210802931,
      "loss": 12.147,
      "step": 17483
    },
    {
      "epoch": 0.9520736762573768,
      "grad_norm": 0.5288320670227035,
      "learning_rate": 0.00011258546401548641,
      "loss": 12.0917,
      "step": 17484
    },
    {
      "epoch": 0.9521281302539598,
      "grad_norm": 0.5175473878683832,
      "learning_rate": 0.00011257671582507687,
      "loss": 12.1715,
      "step": 17485
    },
    {
      "epoch": 0.9521825842505428,
      "grad_norm": 0.5596797458969293,
      "learning_rate": 0.00011256796753686867,
      "loss": 12.1307,
      "step": 17486
    },
    {
      "epoch": 0.9522370382471258,
      "grad_norm": 0.56719958493079,
      "learning_rate": 0.00011255921915092982,
      "loss": 12.1031,
      "step": 17487
    },
    {
      "epoch": 0.9522914922437089,
      "grad_norm": 0.5592444241773291,
      "learning_rate": 0.00011255047066732842,
      "loss": 12.0593,
      "step": 17488
    },
    {
      "epoch": 0.9523459462402919,
      "grad_norm": 0.5679003470701051,
      "learning_rate": 0.0001125417220861324,
      "loss": 12.2328,
      "step": 17489
    },
    {
      "epoch": 0.9524004002368749,
      "grad_norm": 0.5376761644488985,
      "learning_rate": 0.00011253297340740987,
      "loss": 12.1243,
      "step": 17490
    },
    {
      "epoch": 0.9524548542334579,
      "grad_norm": 0.5672837763664881,
      "learning_rate": 0.00011252422463122884,
      "loss": 12.0671,
      "step": 17491
    },
    {
      "epoch": 0.9525093082300409,
      "grad_norm": 0.6140860516736624,
      "learning_rate": 0.00011251547575765735,
      "loss": 12.2153,
      "step": 17492
    },
    {
      "epoch": 0.9525637622266239,
      "grad_norm": 0.5151714774041493,
      "learning_rate": 0.00011250672678676342,
      "loss": 12.0233,
      "step": 17493
    },
    {
      "epoch": 0.952618216223207,
      "grad_norm": 0.6192749628451423,
      "learning_rate": 0.00011249797771861506,
      "loss": 12.1945,
      "step": 17494
    },
    {
      "epoch": 0.95267267021979,
      "grad_norm": 0.5071903732072338,
      "learning_rate": 0.00011248922855328035,
      "loss": 12.1009,
      "step": 17495
    },
    {
      "epoch": 0.952727124216373,
      "grad_norm": 0.5047572016443033,
      "learning_rate": 0.0001124804792908273,
      "loss": 12.137,
      "step": 17496
    },
    {
      "epoch": 0.952781578212956,
      "grad_norm": 0.5803959738561252,
      "learning_rate": 0.00011247172993132394,
      "loss": 12.3401,
      "step": 17497
    },
    {
      "epoch": 0.9528360322095389,
      "grad_norm": 0.6369007115438952,
      "learning_rate": 0.00011246298047483834,
      "loss": 12.1584,
      "step": 17498
    },
    {
      "epoch": 0.952890486206122,
      "grad_norm": 0.6022015694890215,
      "learning_rate": 0.00011245423092143852,
      "loss": 12.2493,
      "step": 17499
    },
    {
      "epoch": 0.952944940202705,
      "grad_norm": 0.5495827166066689,
      "learning_rate": 0.00011244548127119245,
      "loss": 12.0801,
      "step": 17500
    },
    {
      "epoch": 0.952999394199288,
      "grad_norm": 0.5195826351260587,
      "learning_rate": 0.00011243673152416827,
      "loss": 12.0984,
      "step": 17501
    },
    {
      "epoch": 0.953053848195871,
      "grad_norm": 0.541215822448782,
      "learning_rate": 0.000112427981680434,
      "loss": 12.0597,
      "step": 17502
    },
    {
      "epoch": 0.953108302192454,
      "grad_norm": 0.546730479220331,
      "learning_rate": 0.00011241923174005767,
      "loss": 12.1773,
      "step": 17503
    },
    {
      "epoch": 0.953162756189037,
      "grad_norm": 0.5547994486952583,
      "learning_rate": 0.00011241048170310726,
      "loss": 12.1626,
      "step": 17504
    },
    {
      "epoch": 0.9532172101856201,
      "grad_norm": 0.546797868136595,
      "learning_rate": 0.00011240173156965088,
      "loss": 12.2097,
      "step": 17505
    },
    {
      "epoch": 0.9532716641822031,
      "grad_norm": 0.6532019259814937,
      "learning_rate": 0.00011239298133975656,
      "loss": 12.282,
      "step": 17506
    },
    {
      "epoch": 0.9533261181787861,
      "grad_norm": 0.5831149839815883,
      "learning_rate": 0.00011238423101349234,
      "loss": 12.2849,
      "step": 17507
    },
    {
      "epoch": 0.9533805721753691,
      "grad_norm": 0.5331895682878139,
      "learning_rate": 0.00011237548059092629,
      "loss": 12.2181,
      "step": 17508
    },
    {
      "epoch": 0.9534350261719521,
      "grad_norm": 0.5594185352289169,
      "learning_rate": 0.00011236673007212639,
      "loss": 12.0504,
      "step": 17509
    },
    {
      "epoch": 0.9534894801685351,
      "grad_norm": 0.6052484244390849,
      "learning_rate": 0.00011235797945716073,
      "loss": 12.2553,
      "step": 17510
    },
    {
      "epoch": 0.9535439341651182,
      "grad_norm": 0.5459851995298468,
      "learning_rate": 0.00011234922874609735,
      "loss": 12.2289,
      "step": 17511
    },
    {
      "epoch": 0.9535983881617012,
      "grad_norm": 0.5216090581938867,
      "learning_rate": 0.00011234047793900429,
      "loss": 12.0912,
      "step": 17512
    },
    {
      "epoch": 0.9536528421582842,
      "grad_norm": 0.6099670336744216,
      "learning_rate": 0.00011233172703594962,
      "loss": 12.2609,
      "step": 17513
    },
    {
      "epoch": 0.9537072961548672,
      "grad_norm": 0.5814475487547149,
      "learning_rate": 0.0001123229760370014,
      "loss": 12.1752,
      "step": 17514
    },
    {
      "epoch": 0.9537617501514501,
      "grad_norm": 0.597815180264054,
      "learning_rate": 0.00011231422494222761,
      "loss": 12.007,
      "step": 17515
    },
    {
      "epoch": 0.9538162041480331,
      "grad_norm": 0.49954782488132937,
      "learning_rate": 0.00011230547375169634,
      "loss": 12.0644,
      "step": 17516
    },
    {
      "epoch": 0.9538706581446162,
      "grad_norm": 0.5026829090145971,
      "learning_rate": 0.00011229672246547562,
      "loss": 12.1274,
      "step": 17517
    },
    {
      "epoch": 0.9539251121411992,
      "grad_norm": 0.5387715703406172,
      "learning_rate": 0.00011228797108363358,
      "loss": 12.0936,
      "step": 17518
    },
    {
      "epoch": 0.9539795661377822,
      "grad_norm": 0.5469127944330988,
      "learning_rate": 0.0001122792196062382,
      "loss": 12.1236,
      "step": 17519
    },
    {
      "epoch": 0.9540340201343652,
      "grad_norm": 0.506500734264548,
      "learning_rate": 0.00011227046803335755,
      "loss": 11.9995,
      "step": 17520
    },
    {
      "epoch": 0.9540884741309482,
      "grad_norm": 0.5343499369968282,
      "learning_rate": 0.00011226171636505967,
      "loss": 12.1079,
      "step": 17521
    },
    {
      "epoch": 0.9541429281275312,
      "grad_norm": 0.530386561431065,
      "learning_rate": 0.00011225296460141262,
      "loss": 12.0347,
      "step": 17522
    },
    {
      "epoch": 0.9541973821241143,
      "grad_norm": 0.5424374998579926,
      "learning_rate": 0.0001122442127424845,
      "loss": 12.1531,
      "step": 17523
    },
    {
      "epoch": 0.9542518361206973,
      "grad_norm": 0.558572783863881,
      "learning_rate": 0.00011223546078834328,
      "loss": 12.1718,
      "step": 17524
    },
    {
      "epoch": 0.9543062901172803,
      "grad_norm": 0.5112976002495363,
      "learning_rate": 0.0001122267087390571,
      "loss": 12.1203,
      "step": 17525
    },
    {
      "epoch": 0.9543607441138633,
      "grad_norm": 0.6139608398164536,
      "learning_rate": 0.00011221795659469396,
      "loss": 12.2115,
      "step": 17526
    },
    {
      "epoch": 0.9544151981104463,
      "grad_norm": 0.5590312583921028,
      "learning_rate": 0.00011220920435532197,
      "loss": 12.1637,
      "step": 17527
    },
    {
      "epoch": 0.9544696521070293,
      "grad_norm": 0.5220602531012433,
      "learning_rate": 0.00011220045202100913,
      "loss": 12.0965,
      "step": 17528
    },
    {
      "epoch": 0.9545241061036124,
      "grad_norm": 0.5146669588181602,
      "learning_rate": 0.00011219169959182354,
      "loss": 12.1581,
      "step": 17529
    },
    {
      "epoch": 0.9545785601001954,
      "grad_norm": 0.5103508856909138,
      "learning_rate": 0.00011218294706783323,
      "loss": 12.0123,
      "step": 17530
    },
    {
      "epoch": 0.9546330140967784,
      "grad_norm": 0.5394214752467305,
      "learning_rate": 0.00011217419444910631,
      "loss": 12.1238,
      "step": 17531
    },
    {
      "epoch": 0.9546874680933614,
      "grad_norm": 0.6403842059952954,
      "learning_rate": 0.0001121654417357108,
      "loss": 12.2112,
      "step": 17532
    },
    {
      "epoch": 0.9547419220899444,
      "grad_norm": 0.5181254678820698,
      "learning_rate": 0.00011215668892771478,
      "loss": 12.0037,
      "step": 17533
    },
    {
      "epoch": 0.9547963760865275,
      "grad_norm": 0.5086013058427543,
      "learning_rate": 0.0001121479360251863,
      "loss": 12.1228,
      "step": 17534
    },
    {
      "epoch": 0.9548508300831104,
      "grad_norm": 0.5257888747973913,
      "learning_rate": 0.00011213918302819344,
      "loss": 12.2236,
      "step": 17535
    },
    {
      "epoch": 0.9549052840796934,
      "grad_norm": 0.5328935148950803,
      "learning_rate": 0.00011213042993680424,
      "loss": 12.0387,
      "step": 17536
    },
    {
      "epoch": 0.9549597380762764,
      "grad_norm": 0.5784220847540573,
      "learning_rate": 0.00011212167675108683,
      "loss": 12.1117,
      "step": 17537
    },
    {
      "epoch": 0.9550141920728594,
      "grad_norm": 0.5529509688232136,
      "learning_rate": 0.00011211292347110918,
      "loss": 12.1224,
      "step": 17538
    },
    {
      "epoch": 0.9550686460694424,
      "grad_norm": 0.6476686014260221,
      "learning_rate": 0.0001121041700969394,
      "loss": 12.1743,
      "step": 17539
    },
    {
      "epoch": 0.9551231000660255,
      "grad_norm": 0.6194846217104752,
      "learning_rate": 0.0001120954166286456,
      "loss": 12.1517,
      "step": 17540
    },
    {
      "epoch": 0.9551775540626085,
      "grad_norm": 0.5867983675726799,
      "learning_rate": 0.00011208666306629581,
      "loss": 12.1768,
      "step": 17541
    },
    {
      "epoch": 0.9552320080591915,
      "grad_norm": 0.5464992669933516,
      "learning_rate": 0.00011207790940995808,
      "loss": 12.0183,
      "step": 17542
    },
    {
      "epoch": 0.9552864620557745,
      "grad_norm": 0.5798443116280011,
      "learning_rate": 0.0001120691556597005,
      "loss": 12.0348,
      "step": 17543
    },
    {
      "epoch": 0.9553409160523575,
      "grad_norm": 0.5732012006437913,
      "learning_rate": 0.00011206040181559117,
      "loss": 12.1493,
      "step": 17544
    },
    {
      "epoch": 0.9553953700489405,
      "grad_norm": 0.5767419723503235,
      "learning_rate": 0.0001120516478776981,
      "loss": 11.9773,
      "step": 17545
    },
    {
      "epoch": 0.9554498240455236,
      "grad_norm": 0.5359491833504862,
      "learning_rate": 0.00011204289384608941,
      "loss": 11.8632,
      "step": 17546
    },
    {
      "epoch": 0.9555042780421066,
      "grad_norm": 0.5848547547696398,
      "learning_rate": 0.00011203413972083315,
      "loss": 12.1047,
      "step": 17547
    },
    {
      "epoch": 0.9555587320386896,
      "grad_norm": 0.5465695726591704,
      "learning_rate": 0.00011202538550199742,
      "loss": 12.0137,
      "step": 17548
    },
    {
      "epoch": 0.9556131860352726,
      "grad_norm": 0.49996478425654056,
      "learning_rate": 0.00011201663118965025,
      "loss": 12.0775,
      "step": 17549
    },
    {
      "epoch": 0.9556676400318556,
      "grad_norm": 0.5747110527477752,
      "learning_rate": 0.00011200787678385975,
      "loss": 12.1007,
      "step": 17550
    },
    {
      "epoch": 0.9557220940284386,
      "grad_norm": 0.6505968349618484,
      "learning_rate": 0.000111999122284694,
      "loss": 12.2088,
      "step": 17551
    },
    {
      "epoch": 0.9557765480250217,
      "grad_norm": 0.5489434484718236,
      "learning_rate": 0.00011199036769222105,
      "loss": 12.26,
      "step": 17552
    },
    {
      "epoch": 0.9558310020216046,
      "grad_norm": 0.5103149364972873,
      "learning_rate": 0.000111981613006509,
      "loss": 12.174,
      "step": 17553
    },
    {
      "epoch": 0.9558854560181876,
      "grad_norm": 0.5294935254432681,
      "learning_rate": 0.0001119728582276259,
      "loss": 12.1472,
      "step": 17554
    },
    {
      "epoch": 0.9559399100147706,
      "grad_norm": 0.5714016666252779,
      "learning_rate": 0.00011196410335563984,
      "loss": 12.0593,
      "step": 17555
    },
    {
      "epoch": 0.9559943640113536,
      "grad_norm": 0.5641777737296872,
      "learning_rate": 0.00011195534839061895,
      "loss": 12.1639,
      "step": 17556
    },
    {
      "epoch": 0.9560488180079366,
      "grad_norm": 0.5434267121340497,
      "learning_rate": 0.00011194659333263122,
      "loss": 12.0999,
      "step": 17557
    },
    {
      "epoch": 0.9561032720045197,
      "grad_norm": 0.5531724406692369,
      "learning_rate": 0.00011193783818174482,
      "loss": 12.1736,
      "step": 17558
    },
    {
      "epoch": 0.9561577260011027,
      "grad_norm": 0.5639080263258313,
      "learning_rate": 0.00011192908293802778,
      "loss": 12.1458,
      "step": 17559
    },
    {
      "epoch": 0.9562121799976857,
      "grad_norm": 0.5885309555968149,
      "learning_rate": 0.00011192032760154814,
      "loss": 11.9894,
      "step": 17560
    },
    {
      "epoch": 0.9562666339942687,
      "grad_norm": 0.5576789959998956,
      "learning_rate": 0.00011191157217237406,
      "loss": 12.1119,
      "step": 17561
    },
    {
      "epoch": 0.9563210879908517,
      "grad_norm": 0.5315643033369111,
      "learning_rate": 0.00011190281665057362,
      "loss": 12.199,
      "step": 17562
    },
    {
      "epoch": 0.9563755419874347,
      "grad_norm": 0.5283050591210666,
      "learning_rate": 0.00011189406103621487,
      "loss": 12.2325,
      "step": 17563
    },
    {
      "epoch": 0.9564299959840178,
      "grad_norm": 0.5224092214354858,
      "learning_rate": 0.00011188530532936592,
      "loss": 12.1109,
      "step": 17564
    },
    {
      "epoch": 0.9564844499806008,
      "grad_norm": 0.6066525480426279,
      "learning_rate": 0.00011187654953009483,
      "loss": 12.1813,
      "step": 17565
    },
    {
      "epoch": 0.9565389039771838,
      "grad_norm": 0.649472746570281,
      "learning_rate": 0.00011186779363846966,
      "loss": 12.1454,
      "step": 17566
    },
    {
      "epoch": 0.9565933579737668,
      "grad_norm": 0.5786411478000614,
      "learning_rate": 0.00011185903765455859,
      "loss": 12.1886,
      "step": 17567
    },
    {
      "epoch": 0.9566478119703498,
      "grad_norm": 0.5527424783159517,
      "learning_rate": 0.00011185028157842962,
      "loss": 12.0374,
      "step": 17568
    },
    {
      "epoch": 0.9567022659669329,
      "grad_norm": 0.5821605386294587,
      "learning_rate": 0.00011184152541015092,
      "loss": 12.1065,
      "step": 17569
    },
    {
      "epoch": 0.9567567199635159,
      "grad_norm": 0.5925912060183305,
      "learning_rate": 0.00011183276914979051,
      "loss": 12.3158,
      "step": 17570
    },
    {
      "epoch": 0.9568111739600988,
      "grad_norm": 0.5634063289150791,
      "learning_rate": 0.00011182401279741648,
      "loss": 12.057,
      "step": 17571
    },
    {
      "epoch": 0.9568656279566818,
      "grad_norm": 0.5824811736735698,
      "learning_rate": 0.00011181525635309695,
      "loss": 12.2049,
      "step": 17572
    },
    {
      "epoch": 0.9569200819532648,
      "grad_norm": 0.5775520866209082,
      "learning_rate": 0.0001118064998169,
      "loss": 12.1704,
      "step": 17573
    },
    {
      "epoch": 0.9569745359498478,
      "grad_norm": 0.48993449407470774,
      "learning_rate": 0.00011179774318889378,
      "loss": 12.1007,
      "step": 17574
    },
    {
      "epoch": 0.9570289899464309,
      "grad_norm": 0.5439971950253318,
      "learning_rate": 0.00011178898646914629,
      "loss": 12.1628,
      "step": 17575
    },
    {
      "epoch": 0.9570834439430139,
      "grad_norm": 0.5620142288513571,
      "learning_rate": 0.00011178022965772566,
      "loss": 12.1764,
      "step": 17576
    },
    {
      "epoch": 0.9571378979395969,
      "grad_norm": 0.5586042403613316,
      "learning_rate": 0.00011177147275469997,
      "loss": 12.0713,
      "step": 17577
    },
    {
      "epoch": 0.9571923519361799,
      "grad_norm": 0.6908235593149696,
      "learning_rate": 0.00011176271576013738,
      "loss": 12.3033,
      "step": 17578
    },
    {
      "epoch": 0.9572468059327629,
      "grad_norm": 0.5475781376792936,
      "learning_rate": 0.00011175395867410592,
      "loss": 11.989,
      "step": 17579
    },
    {
      "epoch": 0.9573012599293459,
      "grad_norm": 0.6810380687100889,
      "learning_rate": 0.00011174520149667371,
      "loss": 12.0593,
      "step": 17580
    },
    {
      "epoch": 0.957355713925929,
      "grad_norm": 0.5340136630295348,
      "learning_rate": 0.00011173644422790883,
      "loss": 12.0288,
      "step": 17581
    },
    {
      "epoch": 0.957410167922512,
      "grad_norm": 0.5542940066010866,
      "learning_rate": 0.00011172768686787938,
      "loss": 12.1359,
      "step": 17582
    },
    {
      "epoch": 0.957464621919095,
      "grad_norm": 0.5892735554281785,
      "learning_rate": 0.00011171892941665349,
      "loss": 12.1703,
      "step": 17583
    },
    {
      "epoch": 0.957519075915678,
      "grad_norm": 0.544821778155144,
      "learning_rate": 0.00011171017187429926,
      "loss": 12.1065,
      "step": 17584
    },
    {
      "epoch": 0.957573529912261,
      "grad_norm": 0.5092524735271818,
      "learning_rate": 0.00011170141424088476,
      "loss": 12.0485,
      "step": 17585
    },
    {
      "epoch": 0.957627983908844,
      "grad_norm": 0.5919783181085536,
      "learning_rate": 0.00011169265651647809,
      "loss": 12.1461,
      "step": 17586
    },
    {
      "epoch": 0.9576824379054271,
      "grad_norm": 0.5603813725359015,
      "learning_rate": 0.00011168389870114735,
      "loss": 12.1747,
      "step": 17587
    },
    {
      "epoch": 0.9577368919020101,
      "grad_norm": 0.6012158867500161,
      "learning_rate": 0.00011167514079496064,
      "loss": 12.2645,
      "step": 17588
    },
    {
      "epoch": 0.957791345898593,
      "grad_norm": 0.5660345202872905,
      "learning_rate": 0.00011166638279798614,
      "loss": 12.1592,
      "step": 17589
    },
    {
      "epoch": 0.957845799895176,
      "grad_norm": 0.5603806205536941,
      "learning_rate": 0.00011165762471029184,
      "loss": 12.1503,
      "step": 17590
    },
    {
      "epoch": 0.957900253891759,
      "grad_norm": 0.5918656178193153,
      "learning_rate": 0.0001116488665319459,
      "loss": 12.2107,
      "step": 17591
    },
    {
      "epoch": 0.957954707888342,
      "grad_norm": 0.5341569004437059,
      "learning_rate": 0.00011164010826301645,
      "loss": 12.191,
      "step": 17592
    },
    {
      "epoch": 0.9580091618849251,
      "grad_norm": 0.6158463315800793,
      "learning_rate": 0.00011163134990357153,
      "loss": 12.2008,
      "step": 17593
    },
    {
      "epoch": 0.9580636158815081,
      "grad_norm": 0.5816977629065496,
      "learning_rate": 0.00011162259145367931,
      "loss": 12.1926,
      "step": 17594
    },
    {
      "epoch": 0.9581180698780911,
      "grad_norm": 0.5728226483155598,
      "learning_rate": 0.00011161383291340786,
      "loss": 12.0512,
      "step": 17595
    },
    {
      "epoch": 0.9581725238746741,
      "grad_norm": 0.5456983562201944,
      "learning_rate": 0.00011160507428282529,
      "loss": 12.1537,
      "step": 17596
    },
    {
      "epoch": 0.9582269778712571,
      "grad_norm": 0.5588182377531732,
      "learning_rate": 0.00011159631556199971,
      "loss": 12.2735,
      "step": 17597
    },
    {
      "epoch": 0.9582814318678402,
      "grad_norm": 0.5403151397688128,
      "learning_rate": 0.00011158755675099925,
      "loss": 12.0898,
      "step": 17598
    },
    {
      "epoch": 0.9583358858644232,
      "grad_norm": 0.6025445149650608,
      "learning_rate": 0.00011157879784989202,
      "loss": 12.077,
      "step": 17599
    },
    {
      "epoch": 0.9583903398610062,
      "grad_norm": 0.5381067905909196,
      "learning_rate": 0.00011157003885874609,
      "loss": 12.1695,
      "step": 17600
    },
    {
      "epoch": 0.9584447938575892,
      "grad_norm": 0.509999086899731,
      "learning_rate": 0.0001115612797776296,
      "loss": 12.0278,
      "step": 17601
    },
    {
      "epoch": 0.9584992478541722,
      "grad_norm": 0.5077297192498021,
      "learning_rate": 0.00011155252060661068,
      "loss": 12.1266,
      "step": 17602
    },
    {
      "epoch": 0.9585537018507552,
      "grad_norm": 0.5601415730476714,
      "learning_rate": 0.00011154376134575742,
      "loss": 12.07,
      "step": 17603
    },
    {
      "epoch": 0.9586081558473383,
      "grad_norm": 0.519053095973141,
      "learning_rate": 0.00011153500199513791,
      "loss": 12.1834,
      "step": 17604
    },
    {
      "epoch": 0.9586626098439213,
      "grad_norm": 0.5574930188491317,
      "learning_rate": 0.0001115262425548203,
      "loss": 12.1422,
      "step": 17605
    },
    {
      "epoch": 0.9587170638405043,
      "grad_norm": 0.5596730089308446,
      "learning_rate": 0.0001115174830248727,
      "loss": 12.1437,
      "step": 17606
    },
    {
      "epoch": 0.9587715178370873,
      "grad_norm": 0.5587333800127735,
      "learning_rate": 0.00011150872340536323,
      "loss": 12.1504,
      "step": 17607
    },
    {
      "epoch": 0.9588259718336702,
      "grad_norm": 0.5281109827424201,
      "learning_rate": 0.00011149996369635997,
      "loss": 12.083,
      "step": 17608
    },
    {
      "epoch": 0.9588804258302532,
      "grad_norm": 0.556475140686353,
      "learning_rate": 0.00011149120389793108,
      "loss": 12.0109,
      "step": 17609
    },
    {
      "epoch": 0.9589348798268363,
      "grad_norm": 0.6320131511446037,
      "learning_rate": 0.00011148244401014467,
      "loss": 11.9375,
      "step": 17610
    },
    {
      "epoch": 0.9589893338234193,
      "grad_norm": 0.5369064905067129,
      "learning_rate": 0.00011147368403306884,
      "loss": 12.1104,
      "step": 17611
    },
    {
      "epoch": 0.9590437878200023,
      "grad_norm": 0.5518176679425901,
      "learning_rate": 0.00011146492396677173,
      "loss": 12.0945,
      "step": 17612
    },
    {
      "epoch": 0.9590982418165853,
      "grad_norm": 0.5533819450011978,
      "learning_rate": 0.00011145616381132143,
      "loss": 12.0637,
      "step": 17613
    },
    {
      "epoch": 0.9591526958131683,
      "grad_norm": 0.5464334412896337,
      "learning_rate": 0.00011144740356678611,
      "loss": 12.1286,
      "step": 17614
    },
    {
      "epoch": 0.9592071498097513,
      "grad_norm": 0.5515032427606107,
      "learning_rate": 0.0001114386432332338,
      "loss": 12.0493,
      "step": 17615
    },
    {
      "epoch": 0.9592616038063344,
      "grad_norm": 0.48501813224335094,
      "learning_rate": 0.00011142988281073274,
      "loss": 12.1414,
      "step": 17616
    },
    {
      "epoch": 0.9593160578029174,
      "grad_norm": 0.508094754675668,
      "learning_rate": 0.00011142112229935097,
      "loss": 12.113,
      "step": 17617
    },
    {
      "epoch": 0.9593705117995004,
      "grad_norm": 0.5712777383401623,
      "learning_rate": 0.00011141236169915665,
      "loss": 12.2388,
      "step": 17618
    },
    {
      "epoch": 0.9594249657960834,
      "grad_norm": 0.49801658666359233,
      "learning_rate": 0.00011140360101021789,
      "loss": 12.1054,
      "step": 17619
    },
    {
      "epoch": 0.9594794197926664,
      "grad_norm": 0.5334850006632109,
      "learning_rate": 0.0001113948402326028,
      "loss": 12.1435,
      "step": 17620
    },
    {
      "epoch": 0.9595338737892494,
      "grad_norm": 0.5502584110738724,
      "learning_rate": 0.00011138607936637952,
      "loss": 12.0844,
      "step": 17621
    },
    {
      "epoch": 0.9595883277858325,
      "grad_norm": 0.6631027245293025,
      "learning_rate": 0.00011137731841161621,
      "loss": 12.1147,
      "step": 17622
    },
    {
      "epoch": 0.9596427817824155,
      "grad_norm": 0.6061199396871476,
      "learning_rate": 0.00011136855736838092,
      "loss": 12.1675,
      "step": 17623
    },
    {
      "epoch": 0.9596972357789985,
      "grad_norm": 0.5381270689582018,
      "learning_rate": 0.0001113597962367419,
      "loss": 12.2246,
      "step": 17624
    },
    {
      "epoch": 0.9597516897755815,
      "grad_norm": 0.5291846897679273,
      "learning_rate": 0.00011135103501676711,
      "loss": 12.0552,
      "step": 17625
    },
    {
      "epoch": 0.9598061437721644,
      "grad_norm": 0.5619720136638755,
      "learning_rate": 0.00011134227370852479,
      "loss": 12.2252,
      "step": 17626
    },
    {
      "epoch": 0.9598605977687474,
      "grad_norm": 0.5260468703160102,
      "learning_rate": 0.00011133351231208307,
      "loss": 12.093,
      "step": 17627
    },
    {
      "epoch": 0.9599150517653305,
      "grad_norm": 0.5527473910586487,
      "learning_rate": 0.00011132475082751004,
      "loss": 12.126,
      "step": 17628
    },
    {
      "epoch": 0.9599695057619135,
      "grad_norm": 0.5415193205337437,
      "learning_rate": 0.00011131598925487387,
      "loss": 11.9692,
      "step": 17629
    },
    {
      "epoch": 0.9600239597584965,
      "grad_norm": 0.5793518969808102,
      "learning_rate": 0.00011130722759424266,
      "loss": 12.0674,
      "step": 17630
    },
    {
      "epoch": 0.9600784137550795,
      "grad_norm": 0.6041437604961865,
      "learning_rate": 0.00011129846584568453,
      "loss": 12.1541,
      "step": 17631
    },
    {
      "epoch": 0.9601328677516625,
      "grad_norm": 0.5358846274478933,
      "learning_rate": 0.00011128970400926766,
      "loss": 12.1374,
      "step": 17632
    },
    {
      "epoch": 0.9601873217482456,
      "grad_norm": 0.5344201600769312,
      "learning_rate": 0.00011128094208506014,
      "loss": 12.138,
      "step": 17633
    },
    {
      "epoch": 0.9602417757448286,
      "grad_norm": 0.658089256867609,
      "learning_rate": 0.00011127218007313016,
      "loss": 12.1471,
      "step": 17634
    },
    {
      "epoch": 0.9602962297414116,
      "grad_norm": 0.5888500353636016,
      "learning_rate": 0.00011126341797354578,
      "loss": 12.1384,
      "step": 17635
    },
    {
      "epoch": 0.9603506837379946,
      "grad_norm": 0.5721624240589097,
      "learning_rate": 0.0001112546557863752,
      "loss": 12.1714,
      "step": 17636
    },
    {
      "epoch": 0.9604051377345776,
      "grad_norm": 0.5605762852581129,
      "learning_rate": 0.00011124589351168648,
      "loss": 12.0539,
      "step": 17637
    },
    {
      "epoch": 0.9604595917311606,
      "grad_norm": 0.6076806266827197,
      "learning_rate": 0.00011123713114954784,
      "loss": 12.1637,
      "step": 17638
    },
    {
      "epoch": 0.9605140457277437,
      "grad_norm": 0.5530730637761129,
      "learning_rate": 0.00011122836870002739,
      "loss": 12.0228,
      "step": 17639
    },
    {
      "epoch": 0.9605684997243267,
      "grad_norm": 0.5300881258613511,
      "learning_rate": 0.00011121960616319327,
      "loss": 12.1155,
      "step": 17640
    },
    {
      "epoch": 0.9606229537209097,
      "grad_norm": 0.5156942259719506,
      "learning_rate": 0.0001112108435391136,
      "loss": 12.1455,
      "step": 17641
    },
    {
      "epoch": 0.9606774077174927,
      "grad_norm": 0.5948524091580496,
      "learning_rate": 0.00011120208082785653,
      "loss": 12.1968,
      "step": 17642
    },
    {
      "epoch": 0.9607318617140757,
      "grad_norm": 0.5832431241578581,
      "learning_rate": 0.00011119331802949016,
      "loss": 12.1199,
      "step": 17643
    },
    {
      "epoch": 0.9607863157106586,
      "grad_norm": 0.5666121368406475,
      "learning_rate": 0.00011118455514408272,
      "loss": 12.0781,
      "step": 17644
    },
    {
      "epoch": 0.9608407697072417,
      "grad_norm": 0.534523597152492,
      "learning_rate": 0.0001111757921717023,
      "loss": 12.0978,
      "step": 17645
    },
    {
      "epoch": 0.9608952237038247,
      "grad_norm": 0.5605840777661149,
      "learning_rate": 0.00011116702911241703,
      "loss": 12.1463,
      "step": 17646
    },
    {
      "epoch": 0.9609496777004077,
      "grad_norm": 0.5344915870698154,
      "learning_rate": 0.00011115826596629508,
      "loss": 12.169,
      "step": 17647
    },
    {
      "epoch": 0.9610041316969907,
      "grad_norm": 0.593752192813583,
      "learning_rate": 0.00011114950273340456,
      "loss": 12.1766,
      "step": 17648
    },
    {
      "epoch": 0.9610585856935737,
      "grad_norm": 0.5331963208485385,
      "learning_rate": 0.00011114073941381369,
      "loss": 12.1211,
      "step": 17649
    },
    {
      "epoch": 0.9611130396901567,
      "grad_norm": 0.6414026744605359,
      "learning_rate": 0.00011113197600759053,
      "loss": 12.1183,
      "step": 17650
    },
    {
      "epoch": 0.9611674936867398,
      "grad_norm": 0.6785769352508368,
      "learning_rate": 0.00011112321251480324,
      "loss": 12.2273,
      "step": 17651
    },
    {
      "epoch": 0.9612219476833228,
      "grad_norm": 0.5554691315348811,
      "learning_rate": 0.00011111444893552,
      "loss": 11.9135,
      "step": 17652
    },
    {
      "epoch": 0.9612764016799058,
      "grad_norm": 0.5031722199806385,
      "learning_rate": 0.00011110568526980896,
      "loss": 12.1629,
      "step": 17653
    },
    {
      "epoch": 0.9613308556764888,
      "grad_norm": 0.5353251694854835,
      "learning_rate": 0.00011109692151773822,
      "loss": 12.1008,
      "step": 17654
    },
    {
      "epoch": 0.9613853096730718,
      "grad_norm": 0.5602726696502285,
      "learning_rate": 0.00011108815767937598,
      "loss": 12.0703,
      "step": 17655
    },
    {
      "epoch": 0.9614397636696548,
      "grad_norm": 0.5902007713398969,
      "learning_rate": 0.00011107939375479035,
      "loss": 12.1335,
      "step": 17656
    },
    {
      "epoch": 0.9614942176662379,
      "grad_norm": 0.5433913493252555,
      "learning_rate": 0.00011107062974404949,
      "loss": 12.0951,
      "step": 17657
    },
    {
      "epoch": 0.9615486716628209,
      "grad_norm": 0.5543159795777511,
      "learning_rate": 0.00011106186564722156,
      "loss": 12.1574,
      "step": 17658
    },
    {
      "epoch": 0.9616031256594039,
      "grad_norm": 0.5747791802624055,
      "learning_rate": 0.00011105310146437473,
      "loss": 12.0994,
      "step": 17659
    },
    {
      "epoch": 0.9616575796559869,
      "grad_norm": 0.5090408070972073,
      "learning_rate": 0.00011104433719557711,
      "loss": 12.1205,
      "step": 17660
    },
    {
      "epoch": 0.9617120336525699,
      "grad_norm": 0.508055808689014,
      "learning_rate": 0.00011103557284089688,
      "loss": 12.1905,
      "step": 17661
    },
    {
      "epoch": 0.9617664876491528,
      "grad_norm": 0.5597201386370542,
      "learning_rate": 0.00011102680840040218,
      "loss": 12.0787,
      "step": 17662
    },
    {
      "epoch": 0.961820941645736,
      "grad_norm": 0.49863322307253,
      "learning_rate": 0.00011101804387416117,
      "loss": 12.151,
      "step": 17663
    },
    {
      "epoch": 0.9618753956423189,
      "grad_norm": 0.6389963740600546,
      "learning_rate": 0.000111009279262242,
      "loss": 12.2213,
      "step": 17664
    },
    {
      "epoch": 0.9619298496389019,
      "grad_norm": 0.6343941955728799,
      "learning_rate": 0.00011100051456471283,
      "loss": 12.0429,
      "step": 17665
    },
    {
      "epoch": 0.9619843036354849,
      "grad_norm": 0.5353282170232694,
      "learning_rate": 0.00011099174978164182,
      "loss": 12.1827,
      "step": 17666
    },
    {
      "epoch": 0.9620387576320679,
      "grad_norm": 0.5617569530792962,
      "learning_rate": 0.00011098298491309711,
      "loss": 12.2126,
      "step": 17667
    },
    {
      "epoch": 0.962093211628651,
      "grad_norm": 0.5868797748808696,
      "learning_rate": 0.00011097421995914687,
      "loss": 12.2281,
      "step": 17668
    },
    {
      "epoch": 0.962147665625234,
      "grad_norm": 0.534053789297622,
      "learning_rate": 0.00011096545491985926,
      "loss": 12.1078,
      "step": 17669
    },
    {
      "epoch": 0.962202119621817,
      "grad_norm": 0.722243932219114,
      "learning_rate": 0.00011095668979530242,
      "loss": 12.3244,
      "step": 17670
    },
    {
      "epoch": 0.9622565736184,
      "grad_norm": 0.5627001055793047,
      "learning_rate": 0.00011094792458554455,
      "loss": 12.09,
      "step": 17671
    },
    {
      "epoch": 0.962311027614983,
      "grad_norm": 0.5470272643302989,
      "learning_rate": 0.00011093915929065378,
      "loss": 12.0246,
      "step": 17672
    },
    {
      "epoch": 0.962365481611566,
      "grad_norm": 0.5584494205081291,
      "learning_rate": 0.00011093039391069823,
      "loss": 12.1595,
      "step": 17673
    },
    {
      "epoch": 0.9624199356081491,
      "grad_norm": 0.5521649001758041,
      "learning_rate": 0.00011092162844574616,
      "loss": 12.0636,
      "step": 17674
    },
    {
      "epoch": 0.9624743896047321,
      "grad_norm": 0.5869416157238587,
      "learning_rate": 0.00011091286289586564,
      "loss": 12.2035,
      "step": 17675
    },
    {
      "epoch": 0.9625288436013151,
      "grad_norm": 0.6040009460125166,
      "learning_rate": 0.00011090409726112487,
      "loss": 12.246,
      "step": 17676
    },
    {
      "epoch": 0.9625832975978981,
      "grad_norm": 0.525820810259676,
      "learning_rate": 0.00011089533154159202,
      "loss": 12.1115,
      "step": 17677
    },
    {
      "epoch": 0.9626377515944811,
      "grad_norm": 0.550318637173561,
      "learning_rate": 0.00011088656573733524,
      "loss": 12.0287,
      "step": 17678
    },
    {
      "epoch": 0.962692205591064,
      "grad_norm": 0.5773091445174735,
      "learning_rate": 0.00011087779984842273,
      "loss": 12.0927,
      "step": 17679
    },
    {
      "epoch": 0.9627466595876472,
      "grad_norm": 0.5159794268401052,
      "learning_rate": 0.00011086903387492257,
      "loss": 12.1421,
      "step": 17680
    },
    {
      "epoch": 0.9628011135842302,
      "grad_norm": 0.5845435845603085,
      "learning_rate": 0.00011086026781690299,
      "loss": 12.1661,
      "step": 17681
    },
    {
      "epoch": 0.9628555675808131,
      "grad_norm": 0.570819557846207,
      "learning_rate": 0.00011085150167443217,
      "loss": 12.1378,
      "step": 17682
    },
    {
      "epoch": 0.9629100215773961,
      "grad_norm": 0.5376903021291405,
      "learning_rate": 0.00011084273544757826,
      "loss": 12.0382,
      "step": 17683
    },
    {
      "epoch": 0.9629644755739791,
      "grad_norm": 0.6044262125617864,
      "learning_rate": 0.00011083396913640942,
      "loss": 12.2581,
      "step": 17684
    },
    {
      "epoch": 0.9630189295705621,
      "grad_norm": 0.5417333473630804,
      "learning_rate": 0.00011082520274099382,
      "loss": 12.1245,
      "step": 17685
    },
    {
      "epoch": 0.9630733835671452,
      "grad_norm": 0.5730865039814494,
      "learning_rate": 0.00011081643626139957,
      "loss": 12.2003,
      "step": 17686
    },
    {
      "epoch": 0.9631278375637282,
      "grad_norm": 0.5888580206213043,
      "learning_rate": 0.00011080766969769493,
      "loss": 12.1222,
      "step": 17687
    },
    {
      "epoch": 0.9631822915603112,
      "grad_norm": 0.5706176479690456,
      "learning_rate": 0.00011079890304994807,
      "loss": 12.0993,
      "step": 17688
    },
    {
      "epoch": 0.9632367455568942,
      "grad_norm": 0.6091264510310652,
      "learning_rate": 0.0001107901363182271,
      "loss": 12.1513,
      "step": 17689
    },
    {
      "epoch": 0.9632911995534772,
      "grad_norm": 0.5900429395195937,
      "learning_rate": 0.00011078136950260025,
      "loss": 12.1448,
      "step": 17690
    },
    {
      "epoch": 0.9633456535500602,
      "grad_norm": 0.5261933964563386,
      "learning_rate": 0.00011077260260313565,
      "loss": 12.1517,
      "step": 17691
    },
    {
      "epoch": 0.9634001075466433,
      "grad_norm": 0.5556032226876118,
      "learning_rate": 0.00011076383561990145,
      "loss": 12.1654,
      "step": 17692
    },
    {
      "epoch": 0.9634545615432263,
      "grad_norm": 0.6023480256814611,
      "learning_rate": 0.0001107550685529659,
      "loss": 12.1451,
      "step": 17693
    },
    {
      "epoch": 0.9635090155398093,
      "grad_norm": 0.5415615996758417,
      "learning_rate": 0.0001107463014023971,
      "loss": 12.1361,
      "step": 17694
    },
    {
      "epoch": 0.9635634695363923,
      "grad_norm": 0.5984563325807083,
      "learning_rate": 0.00011073753416826331,
      "loss": 12.2119,
      "step": 17695
    },
    {
      "epoch": 0.9636179235329753,
      "grad_norm": 0.522196686225686,
      "learning_rate": 0.00011072876685063262,
      "loss": 12.1152,
      "step": 17696
    },
    {
      "epoch": 0.9636723775295583,
      "grad_norm": 0.5544319855034716,
      "learning_rate": 0.00011071999944957321,
      "loss": 12.0433,
      "step": 17697
    },
    {
      "epoch": 0.9637268315261414,
      "grad_norm": 0.6373008070139872,
      "learning_rate": 0.00011071123196515332,
      "loss": 12.2079,
      "step": 17698
    },
    {
      "epoch": 0.9637812855227244,
      "grad_norm": 0.5717749510728661,
      "learning_rate": 0.0001107024643974411,
      "loss": 12.0661,
      "step": 17699
    },
    {
      "epoch": 0.9638357395193073,
      "grad_norm": 0.5143182302457973,
      "learning_rate": 0.00011069369674650474,
      "loss": 12.0557,
      "step": 17700
    },
    {
      "epoch": 0.9638901935158903,
      "grad_norm": 0.6181546577870921,
      "learning_rate": 0.00011068492901241237,
      "loss": 12.3,
      "step": 17701
    },
    {
      "epoch": 0.9639446475124733,
      "grad_norm": 0.5729675225960451,
      "learning_rate": 0.0001106761611952322,
      "loss": 12.2037,
      "step": 17702
    },
    {
      "epoch": 0.9639991015090564,
      "grad_norm": 0.6016097035970154,
      "learning_rate": 0.0001106673932950324,
      "loss": 12.2443,
      "step": 17703
    },
    {
      "epoch": 0.9640535555056394,
      "grad_norm": 0.5869604428655072,
      "learning_rate": 0.00011065862531188116,
      "loss": 12.1124,
      "step": 17704
    },
    {
      "epoch": 0.9641080095022224,
      "grad_norm": 0.4965795860117334,
      "learning_rate": 0.00011064985724584671,
      "loss": 12.0727,
      "step": 17705
    },
    {
      "epoch": 0.9641624634988054,
      "grad_norm": 0.6585504075663329,
      "learning_rate": 0.00011064108909699715,
      "loss": 12.2583,
      "step": 17706
    },
    {
      "epoch": 0.9642169174953884,
      "grad_norm": 0.5649950048506434,
      "learning_rate": 0.00011063232086540069,
      "loss": 11.9266,
      "step": 17707
    },
    {
      "epoch": 0.9642713714919714,
      "grad_norm": 0.6744596465749182,
      "learning_rate": 0.00011062355255112552,
      "loss": 12.1738,
      "step": 17708
    },
    {
      "epoch": 0.9643258254885545,
      "grad_norm": 0.5533997239712947,
      "learning_rate": 0.00011061478415423983,
      "loss": 12.1556,
      "step": 17709
    },
    {
      "epoch": 0.9643802794851375,
      "grad_norm": 0.5526966187371531,
      "learning_rate": 0.00011060601567481181,
      "loss": 12.1112,
      "step": 17710
    },
    {
      "epoch": 0.9644347334817205,
      "grad_norm": 0.5610632617639335,
      "learning_rate": 0.00011059724711290961,
      "loss": 12.0933,
      "step": 17711
    },
    {
      "epoch": 0.9644891874783035,
      "grad_norm": 0.6103466403065875,
      "learning_rate": 0.00011058847846860147,
      "loss": 12.2399,
      "step": 17712
    },
    {
      "epoch": 0.9645436414748865,
      "grad_norm": 0.6296615986685131,
      "learning_rate": 0.00011057970974195553,
      "loss": 12.0778,
      "step": 17713
    },
    {
      "epoch": 0.9645980954714695,
      "grad_norm": 0.5589011964809298,
      "learning_rate": 0.00011057094093303997,
      "loss": 12.2017,
      "step": 17714
    },
    {
      "epoch": 0.9646525494680526,
      "grad_norm": 0.5706032467020418,
      "learning_rate": 0.00011056217204192306,
      "loss": 12.1385,
      "step": 17715
    },
    {
      "epoch": 0.9647070034646356,
      "grad_norm": 0.5380067185680935,
      "learning_rate": 0.00011055340306867288,
      "loss": 12.1828,
      "step": 17716
    },
    {
      "epoch": 0.9647614574612186,
      "grad_norm": 0.5678778439882624,
      "learning_rate": 0.00011054463401335769,
      "loss": 12.009,
      "step": 17717
    },
    {
      "epoch": 0.9648159114578015,
      "grad_norm": 0.5380345678007299,
      "learning_rate": 0.00011053586487604563,
      "loss": 12.0443,
      "step": 17718
    },
    {
      "epoch": 0.9648703654543845,
      "grad_norm": 0.5474811307620698,
      "learning_rate": 0.00011052709565680493,
      "loss": 12.1248,
      "step": 17719
    },
    {
      "epoch": 0.9649248194509675,
      "grad_norm": 0.5442985197543375,
      "learning_rate": 0.00011051832635570379,
      "loss": 12.0413,
      "step": 17720
    },
    {
      "epoch": 0.9649792734475506,
      "grad_norm": 0.5285841645711664,
      "learning_rate": 0.00011050955697281036,
      "loss": 12.0744,
      "step": 17721
    },
    {
      "epoch": 0.9650337274441336,
      "grad_norm": 0.5890287986663996,
      "learning_rate": 0.00011050078750819284,
      "loss": 12.0981,
      "step": 17722
    },
    {
      "epoch": 0.9650881814407166,
      "grad_norm": 0.5926662553098975,
      "learning_rate": 0.00011049201796191945,
      "loss": 12.135,
      "step": 17723
    },
    {
      "epoch": 0.9651426354372996,
      "grad_norm": 0.556356239472455,
      "learning_rate": 0.00011048324833405839,
      "loss": 12.0762,
      "step": 17724
    },
    {
      "epoch": 0.9651970894338826,
      "grad_norm": 0.5723507110626933,
      "learning_rate": 0.00011047447862467781,
      "loss": 12.095,
      "step": 17725
    },
    {
      "epoch": 0.9652515434304656,
      "grad_norm": 0.6284135079318599,
      "learning_rate": 0.00011046570883384593,
      "loss": 12.2335,
      "step": 17726
    },
    {
      "epoch": 0.9653059974270487,
      "grad_norm": 0.5506230573164945,
      "learning_rate": 0.00011045693896163094,
      "loss": 12.1929,
      "step": 17727
    },
    {
      "epoch": 0.9653604514236317,
      "grad_norm": 0.5633949960312391,
      "learning_rate": 0.00011044816900810105,
      "loss": 12.2633,
      "step": 17728
    },
    {
      "epoch": 0.9654149054202147,
      "grad_norm": 0.5426691026278101,
      "learning_rate": 0.00011043939897332442,
      "loss": 12.0683,
      "step": 17729
    },
    {
      "epoch": 0.9654693594167977,
      "grad_norm": 0.556582190461888,
      "learning_rate": 0.0001104306288573693,
      "loss": 12.0986,
      "step": 17730
    },
    {
      "epoch": 0.9655238134133807,
      "grad_norm": 0.5444972117144539,
      "learning_rate": 0.00011042185866030386,
      "loss": 12.2007,
      "step": 17731
    },
    {
      "epoch": 0.9655782674099638,
      "grad_norm": 0.6540297175499002,
      "learning_rate": 0.00011041308838219628,
      "loss": 12.009,
      "step": 17732
    },
    {
      "epoch": 0.9656327214065468,
      "grad_norm": 0.5411306170642678,
      "learning_rate": 0.0001104043180231148,
      "loss": 12.095,
      "step": 17733
    },
    {
      "epoch": 0.9656871754031298,
      "grad_norm": 0.5715791532516885,
      "learning_rate": 0.00011039554758312758,
      "loss": 12.2714,
      "step": 17734
    },
    {
      "epoch": 0.9657416293997128,
      "grad_norm": 0.48470411648455586,
      "learning_rate": 0.00011038677706230285,
      "loss": 11.9713,
      "step": 17735
    },
    {
      "epoch": 0.9657960833962957,
      "grad_norm": 0.5726443792608469,
      "learning_rate": 0.00011037800646070879,
      "loss": 12.2113,
      "step": 17736
    },
    {
      "epoch": 0.9658505373928787,
      "grad_norm": 0.5876872602975912,
      "learning_rate": 0.00011036923577841363,
      "loss": 12.1221,
      "step": 17737
    },
    {
      "epoch": 0.9659049913894618,
      "grad_norm": 0.6080695032280878,
      "learning_rate": 0.00011036046501548554,
      "loss": 12.1666,
      "step": 17738
    },
    {
      "epoch": 0.9659594453860448,
      "grad_norm": 0.5340274869133574,
      "learning_rate": 0.00011035169417199274,
      "loss": 12.2262,
      "step": 17739
    },
    {
      "epoch": 0.9660138993826278,
      "grad_norm": 0.5118138027530789,
      "learning_rate": 0.00011034292324800342,
      "loss": 12.0464,
      "step": 17740
    },
    {
      "epoch": 0.9660683533792108,
      "grad_norm": 0.557631660530052,
      "learning_rate": 0.00011033415224358581,
      "loss": 12.0623,
      "step": 17741
    },
    {
      "epoch": 0.9661228073757938,
      "grad_norm": 0.5622665514054274,
      "learning_rate": 0.00011032538115880809,
      "loss": 12.1165,
      "step": 17742
    },
    {
      "epoch": 0.9661772613723768,
      "grad_norm": 0.6304248670208875,
      "learning_rate": 0.00011031660999373847,
      "loss": 12.2801,
      "step": 17743
    },
    {
      "epoch": 0.9662317153689599,
      "grad_norm": 0.5487926451839776,
      "learning_rate": 0.00011030783874844517,
      "loss": 12.2362,
      "step": 17744
    },
    {
      "epoch": 0.9662861693655429,
      "grad_norm": 0.5422608903886796,
      "learning_rate": 0.00011029906742299641,
      "loss": 12.1801,
      "step": 17745
    },
    {
      "epoch": 0.9663406233621259,
      "grad_norm": 0.5263104983244903,
      "learning_rate": 0.00011029029601746033,
      "loss": 12.0961,
      "step": 17746
    },
    {
      "epoch": 0.9663950773587089,
      "grad_norm": 0.5613791572161544,
      "learning_rate": 0.00011028152453190518,
      "loss": 12.1236,
      "step": 17747
    },
    {
      "epoch": 0.9664495313552919,
      "grad_norm": 0.5716835210741285,
      "learning_rate": 0.00011027275296639921,
      "loss": 12.1599,
      "step": 17748
    },
    {
      "epoch": 0.9665039853518749,
      "grad_norm": 0.4813681714826752,
      "learning_rate": 0.00011026398132101057,
      "loss": 12.0847,
      "step": 17749
    },
    {
      "epoch": 0.966558439348458,
      "grad_norm": 0.5561675500158483,
      "learning_rate": 0.0001102552095958075,
      "loss": 12.069,
      "step": 17750
    },
    {
      "epoch": 0.966612893345041,
      "grad_norm": 0.5888608590341523,
      "learning_rate": 0.00011024643779085819,
      "loss": 12.1539,
      "step": 17751
    },
    {
      "epoch": 0.966667347341624,
      "grad_norm": 0.5710489765909422,
      "learning_rate": 0.00011023766590623085,
      "loss": 12.1051,
      "step": 17752
    },
    {
      "epoch": 0.966721801338207,
      "grad_norm": 0.5384051393906014,
      "learning_rate": 0.00011022889394199371,
      "loss": 12.1687,
      "step": 17753
    },
    {
      "epoch": 0.96677625533479,
      "grad_norm": 0.6134717562884549,
      "learning_rate": 0.000110220121898215,
      "loss": 12.2179,
      "step": 17754
    },
    {
      "epoch": 0.9668307093313729,
      "grad_norm": 0.6099856991129531,
      "learning_rate": 0.0001102113497749629,
      "loss": 12.1484,
      "step": 17755
    },
    {
      "epoch": 0.966885163327956,
      "grad_norm": 0.5933025472236174,
      "learning_rate": 0.00011020257757230563,
      "loss": 12.0995,
      "step": 17756
    },
    {
      "epoch": 0.966939617324539,
      "grad_norm": 0.6946683299584384,
      "learning_rate": 0.00011019380529031138,
      "loss": 12.1113,
      "step": 17757
    },
    {
      "epoch": 0.966994071321122,
      "grad_norm": 0.5606703202864911,
      "learning_rate": 0.00011018503292904841,
      "loss": 12.0408,
      "step": 17758
    },
    {
      "epoch": 0.967048525317705,
      "grad_norm": 0.5298869291861762,
      "learning_rate": 0.00011017626048858491,
      "loss": 12.2027,
      "step": 17759
    },
    {
      "epoch": 0.967102979314288,
      "grad_norm": 0.5606826260571348,
      "learning_rate": 0.00011016748796898913,
      "loss": 12.0425,
      "step": 17760
    },
    {
      "epoch": 0.967157433310871,
      "grad_norm": 0.5834963687793829,
      "learning_rate": 0.00011015871537032923,
      "loss": 12.1075,
      "step": 17761
    },
    {
      "epoch": 0.9672118873074541,
      "grad_norm": 0.5592877249862209,
      "learning_rate": 0.00011014994269267347,
      "loss": 12.0068,
      "step": 17762
    },
    {
      "epoch": 0.9672663413040371,
      "grad_norm": 0.5764920098079752,
      "learning_rate": 0.00011014116993609001,
      "loss": 12.1481,
      "step": 17763
    },
    {
      "epoch": 0.9673207953006201,
      "grad_norm": 0.5605968035673906,
      "learning_rate": 0.00011013239710064716,
      "loss": 12.1694,
      "step": 17764
    },
    {
      "epoch": 0.9673752492972031,
      "grad_norm": 0.5566859412922541,
      "learning_rate": 0.00011012362418641309,
      "loss": 12.2004,
      "step": 17765
    },
    {
      "epoch": 0.9674297032937861,
      "grad_norm": 0.5714501816840976,
      "learning_rate": 0.00011011485119345602,
      "loss": 12.1422,
      "step": 17766
    },
    {
      "epoch": 0.9674841572903692,
      "grad_norm": 0.5339414451837563,
      "learning_rate": 0.00011010607812184415,
      "loss": 12.0454,
      "step": 17767
    },
    {
      "epoch": 0.9675386112869522,
      "grad_norm": 0.5231871867158163,
      "learning_rate": 0.00011009730497164572,
      "loss": 12.0991,
      "step": 17768
    },
    {
      "epoch": 0.9675930652835352,
      "grad_norm": 0.5128397437189953,
      "learning_rate": 0.00011008853174292895,
      "loss": 12.0688,
      "step": 17769
    },
    {
      "epoch": 0.9676475192801182,
      "grad_norm": 0.5893411496827528,
      "learning_rate": 0.0001100797584357621,
      "loss": 12.1022,
      "step": 17770
    },
    {
      "epoch": 0.9677019732767012,
      "grad_norm": 0.541464055289501,
      "learning_rate": 0.00011007098505021334,
      "loss": 12.0411,
      "step": 17771
    },
    {
      "epoch": 0.9677564272732841,
      "grad_norm": 0.5888980133927575,
      "learning_rate": 0.0001100622115863509,
      "loss": 12.1288,
      "step": 17772
    },
    {
      "epoch": 0.9678108812698673,
      "grad_norm": 0.6788846626525558,
      "learning_rate": 0.00011005343804424302,
      "loss": 12.1538,
      "step": 17773
    },
    {
      "epoch": 0.9678653352664502,
      "grad_norm": 0.5944965815213048,
      "learning_rate": 0.00011004466442395792,
      "loss": 12.0351,
      "step": 17774
    },
    {
      "epoch": 0.9679197892630332,
      "grad_norm": 0.5203669199642835,
      "learning_rate": 0.00011003589072556384,
      "loss": 12.1308,
      "step": 17775
    },
    {
      "epoch": 0.9679742432596162,
      "grad_norm": 0.5225541564802234,
      "learning_rate": 0.00011002711694912898,
      "loss": 12.1828,
      "step": 17776
    },
    {
      "epoch": 0.9680286972561992,
      "grad_norm": 0.5809015328095049,
      "learning_rate": 0.00011001834309472157,
      "loss": 12.094,
      "step": 17777
    },
    {
      "epoch": 0.9680831512527822,
      "grad_norm": 0.5858089071552731,
      "learning_rate": 0.00011000956916240985,
      "loss": 12.1914,
      "step": 17778
    },
    {
      "epoch": 0.9681376052493653,
      "grad_norm": 0.5900559454762778,
      "learning_rate": 0.00011000079515226204,
      "loss": 12.1834,
      "step": 17779
    },
    {
      "epoch": 0.9681920592459483,
      "grad_norm": 0.5748059397943471,
      "learning_rate": 0.00010999202106434637,
      "loss": 12.1808,
      "step": 17780
    },
    {
      "epoch": 0.9682465132425313,
      "grad_norm": 0.5127534384417431,
      "learning_rate": 0.00010998324689873107,
      "loss": 11.8834,
      "step": 17781
    },
    {
      "epoch": 0.9683009672391143,
      "grad_norm": 0.5320361643919194,
      "learning_rate": 0.00010997447265548437,
      "loss": 12.0902,
      "step": 17782
    },
    {
      "epoch": 0.9683554212356973,
      "grad_norm": 0.5377705507502049,
      "learning_rate": 0.00010996569833467449,
      "loss": 12.1417,
      "step": 17783
    },
    {
      "epoch": 0.9684098752322803,
      "grad_norm": 0.5265742666871742,
      "learning_rate": 0.00010995692393636968,
      "loss": 12.1022,
      "step": 17784
    },
    {
      "epoch": 0.9684643292288634,
      "grad_norm": 0.5190329800999794,
      "learning_rate": 0.00010994814946063816,
      "loss": 11.987,
      "step": 17785
    },
    {
      "epoch": 0.9685187832254464,
      "grad_norm": 0.5031755079473749,
      "learning_rate": 0.00010993937490754815,
      "loss": 12.1866,
      "step": 17786
    },
    {
      "epoch": 0.9685732372220294,
      "grad_norm": 0.5982588842106145,
      "learning_rate": 0.00010993060027716791,
      "loss": 12.0101,
      "step": 17787
    },
    {
      "epoch": 0.9686276912186124,
      "grad_norm": 0.550467381176369,
      "learning_rate": 0.00010992182556956562,
      "loss": 12.1632,
      "step": 17788
    },
    {
      "epoch": 0.9686821452151954,
      "grad_norm": 0.6039951623824136,
      "learning_rate": 0.00010991305078480957,
      "loss": 12.1521,
      "step": 17789
    },
    {
      "epoch": 0.9687365992117783,
      "grad_norm": 0.5398755963835501,
      "learning_rate": 0.000109904275922968,
      "loss": 12.1698,
      "step": 17790
    },
    {
      "epoch": 0.9687910532083615,
      "grad_norm": 0.5309708003642307,
      "learning_rate": 0.0001098955009841091,
      "loss": 12.0762,
      "step": 17791
    },
    {
      "epoch": 0.9688455072049444,
      "grad_norm": 0.5860085116735837,
      "learning_rate": 0.00010988672596830112,
      "loss": 12.3387,
      "step": 17792
    },
    {
      "epoch": 0.9688999612015274,
      "grad_norm": 0.5844423560245651,
      "learning_rate": 0.00010987795087561232,
      "loss": 12.1434,
      "step": 17793
    },
    {
      "epoch": 0.9689544151981104,
      "grad_norm": 0.6662312473550055,
      "learning_rate": 0.0001098691757061109,
      "loss": 12.0429,
      "step": 17794
    },
    {
      "epoch": 0.9690088691946934,
      "grad_norm": 0.5643006113425973,
      "learning_rate": 0.00010986040045986512,
      "loss": 12.1875,
      "step": 17795
    },
    {
      "epoch": 0.9690633231912764,
      "grad_norm": 0.5834323785455291,
      "learning_rate": 0.0001098516251369432,
      "loss": 12.1732,
      "step": 17796
    },
    {
      "epoch": 0.9691177771878595,
      "grad_norm": 0.5963028530585607,
      "learning_rate": 0.0001098428497374134,
      "loss": 12.0946,
      "step": 17797
    },
    {
      "epoch": 0.9691722311844425,
      "grad_norm": 0.6125634195562416,
      "learning_rate": 0.00010983407426134396,
      "loss": 12.1698,
      "step": 17798
    },
    {
      "epoch": 0.9692266851810255,
      "grad_norm": 0.615414593265336,
      "learning_rate": 0.0001098252987088031,
      "loss": 12.107,
      "step": 17799
    },
    {
      "epoch": 0.9692811391776085,
      "grad_norm": 0.533551237794284,
      "learning_rate": 0.0001098165230798591,
      "loss": 12.0986,
      "step": 17800
    },
    {
      "epoch": 0.9693355931741915,
      "grad_norm": 0.6055228077578817,
      "learning_rate": 0.00010980774737458011,
      "loss": 12.2149,
      "step": 17801
    },
    {
      "epoch": 0.9693900471707746,
      "grad_norm": 0.571576525970899,
      "learning_rate": 0.00010979897159303447,
      "loss": 12.1945,
      "step": 17802
    },
    {
      "epoch": 0.9694445011673576,
      "grad_norm": 0.6362828666064082,
      "learning_rate": 0.00010979019573529037,
      "loss": 12.0863,
      "step": 17803
    },
    {
      "epoch": 0.9694989551639406,
      "grad_norm": 0.5302447488075314,
      "learning_rate": 0.00010978141980141608,
      "loss": 12.0461,
      "step": 17804
    },
    {
      "epoch": 0.9695534091605236,
      "grad_norm": 0.4924079645027928,
      "learning_rate": 0.00010977264379147985,
      "loss": 12.0715,
      "step": 17805
    },
    {
      "epoch": 0.9696078631571066,
      "grad_norm": 0.5327589830801305,
      "learning_rate": 0.00010976386770554983,
      "loss": 12.1169,
      "step": 17806
    },
    {
      "epoch": 0.9696623171536896,
      "grad_norm": 0.5716778845807983,
      "learning_rate": 0.00010975509154369439,
      "loss": 12.0048,
      "step": 17807
    },
    {
      "epoch": 0.9697167711502727,
      "grad_norm": 0.5491710740186433,
      "learning_rate": 0.00010974631530598171,
      "loss": 12.0912,
      "step": 17808
    },
    {
      "epoch": 0.9697712251468557,
      "grad_norm": 0.5898653267786542,
      "learning_rate": 0.00010973753899248005,
      "loss": 12.2711,
      "step": 17809
    },
    {
      "epoch": 0.9698256791434386,
      "grad_norm": 0.6055769095504616,
      "learning_rate": 0.00010972876260325769,
      "loss": 12.2608,
      "step": 17810
    },
    {
      "epoch": 0.9698801331400216,
      "grad_norm": 0.6172966019298604,
      "learning_rate": 0.0001097199861383828,
      "loss": 12.1219,
      "step": 17811
    },
    {
      "epoch": 0.9699345871366046,
      "grad_norm": 0.5145952516661499,
      "learning_rate": 0.00010971120959792365,
      "loss": 12.1383,
      "step": 17812
    },
    {
      "epoch": 0.9699890411331876,
      "grad_norm": 0.5068824144878783,
      "learning_rate": 0.00010970243298194853,
      "loss": 12.098,
      "step": 17813
    },
    {
      "epoch": 0.9700434951297707,
      "grad_norm": 0.5364138632936271,
      "learning_rate": 0.00010969365629052566,
      "loss": 12.0551,
      "step": 17814
    },
    {
      "epoch": 0.9700979491263537,
      "grad_norm": 0.5586029030784768,
      "learning_rate": 0.00010968487952372333,
      "loss": 12.1015,
      "step": 17815
    },
    {
      "epoch": 0.9701524031229367,
      "grad_norm": 0.5385237767372034,
      "learning_rate": 0.0001096761026816097,
      "loss": 12.119,
      "step": 17816
    },
    {
      "epoch": 0.9702068571195197,
      "grad_norm": 0.5140552714449329,
      "learning_rate": 0.00010966732576425309,
      "loss": 12.1271,
      "step": 17817
    },
    {
      "epoch": 0.9702613111161027,
      "grad_norm": 0.6726861153897186,
      "learning_rate": 0.00010965854877172172,
      "loss": 12.008,
      "step": 17818
    },
    {
      "epoch": 0.9703157651126857,
      "grad_norm": 0.519146769147053,
      "learning_rate": 0.00010964977170408387,
      "loss": 12.1725,
      "step": 17819
    },
    {
      "epoch": 0.9703702191092688,
      "grad_norm": 0.53818594048876,
      "learning_rate": 0.00010964099456140781,
      "loss": 12.1311,
      "step": 17820
    },
    {
      "epoch": 0.9704246731058518,
      "grad_norm": 0.5203950796474744,
      "learning_rate": 0.00010963221734376172,
      "loss": 12.0239,
      "step": 17821
    },
    {
      "epoch": 0.9704791271024348,
      "grad_norm": 0.5622449476707068,
      "learning_rate": 0.0001096234400512139,
      "loss": 12.2025,
      "step": 17822
    },
    {
      "epoch": 0.9705335810990178,
      "grad_norm": 0.5555495555370853,
      "learning_rate": 0.00010961466268383258,
      "loss": 12.0297,
      "step": 17823
    },
    {
      "epoch": 0.9705880350956008,
      "grad_norm": 0.5917465536259531,
      "learning_rate": 0.00010960588524168604,
      "loss": 12.1552,
      "step": 17824
    },
    {
      "epoch": 0.9706424890921838,
      "grad_norm": 0.6715298222792764,
      "learning_rate": 0.00010959710772484256,
      "loss": 12.1018,
      "step": 17825
    },
    {
      "epoch": 0.9706969430887669,
      "grad_norm": 0.5380566954643132,
      "learning_rate": 0.00010958833013337033,
      "loss": 11.9396,
      "step": 17826
    },
    {
      "epoch": 0.9707513970853499,
      "grad_norm": 0.5373949699354345,
      "learning_rate": 0.00010957955246733764,
      "loss": 12.1633,
      "step": 17827
    },
    {
      "epoch": 0.9708058510819328,
      "grad_norm": 0.5680912823299027,
      "learning_rate": 0.00010957077472681274,
      "loss": 12.1199,
      "step": 17828
    },
    {
      "epoch": 0.9708603050785158,
      "grad_norm": 0.5263261907686924,
      "learning_rate": 0.00010956199691186388,
      "loss": 12.1293,
      "step": 17829
    },
    {
      "epoch": 0.9709147590750988,
      "grad_norm": 0.581323079354999,
      "learning_rate": 0.00010955321902255935,
      "loss": 12.2449,
      "step": 17830
    },
    {
      "epoch": 0.9709692130716818,
      "grad_norm": 0.5268762044732026,
      "learning_rate": 0.00010954444105896739,
      "loss": 12.1004,
      "step": 17831
    },
    {
      "epoch": 0.9710236670682649,
      "grad_norm": 0.5384281904412787,
      "learning_rate": 0.00010953566302115625,
      "loss": 12.1106,
      "step": 17832
    },
    {
      "epoch": 0.9710781210648479,
      "grad_norm": 0.5466667623989441,
      "learning_rate": 0.00010952688490919419,
      "loss": 12.1187,
      "step": 17833
    },
    {
      "epoch": 0.9711325750614309,
      "grad_norm": 0.7741320210031013,
      "learning_rate": 0.00010951810672314946,
      "loss": 12.0845,
      "step": 17834
    },
    {
      "epoch": 0.9711870290580139,
      "grad_norm": 0.5379145355908516,
      "learning_rate": 0.00010950932846309034,
      "loss": 12.1771,
      "step": 17835
    },
    {
      "epoch": 0.9712414830545969,
      "grad_norm": 0.5433365394397129,
      "learning_rate": 0.00010950055012908513,
      "loss": 12.1766,
      "step": 17836
    },
    {
      "epoch": 0.97129593705118,
      "grad_norm": 0.6430230755614863,
      "learning_rate": 0.00010949177172120202,
      "loss": 12.0224,
      "step": 17837
    },
    {
      "epoch": 0.971350391047763,
      "grad_norm": 0.5650864017654597,
      "learning_rate": 0.00010948299323950928,
      "loss": 11.9911,
      "step": 17838
    },
    {
      "epoch": 0.971404845044346,
      "grad_norm": 0.5924520159138199,
      "learning_rate": 0.00010947421468407522,
      "loss": 12.0785,
      "step": 17839
    },
    {
      "epoch": 0.971459299040929,
      "grad_norm": 0.5629261714071007,
      "learning_rate": 0.00010946543605496806,
      "loss": 12.0858,
      "step": 17840
    },
    {
      "epoch": 0.971513753037512,
      "grad_norm": 0.6190367299550407,
      "learning_rate": 0.0001094566573522561,
      "loss": 12.1683,
      "step": 17841
    },
    {
      "epoch": 0.971568207034095,
      "grad_norm": 0.5187158542715369,
      "learning_rate": 0.00010944787857600758,
      "loss": 12.137,
      "step": 17842
    },
    {
      "epoch": 0.9716226610306781,
      "grad_norm": 0.5514563108516193,
      "learning_rate": 0.00010943909972629078,
      "loss": 11.932,
      "step": 17843
    },
    {
      "epoch": 0.9716771150272611,
      "grad_norm": 0.5467067275674893,
      "learning_rate": 0.00010943032080317394,
      "loss": 12.0348,
      "step": 17844
    },
    {
      "epoch": 0.971731569023844,
      "grad_norm": 0.9845368637661265,
      "learning_rate": 0.00010942154180672535,
      "loss": 12.1329,
      "step": 17845
    },
    {
      "epoch": 0.971786023020427,
      "grad_norm": 0.5554885818409444,
      "learning_rate": 0.00010941276273701328,
      "loss": 12.0405,
      "step": 17846
    },
    {
      "epoch": 0.97184047701701,
      "grad_norm": 0.57104996085094,
      "learning_rate": 0.00010940398359410598,
      "loss": 12.111,
      "step": 17847
    },
    {
      "epoch": 0.971894931013593,
      "grad_norm": 0.5571519314162232,
      "learning_rate": 0.00010939520437807174,
      "loss": 12.133,
      "step": 17848
    },
    {
      "epoch": 0.9719493850101761,
      "grad_norm": 0.5739596263599989,
      "learning_rate": 0.0001093864250889788,
      "loss": 12.0966,
      "step": 17849
    },
    {
      "epoch": 0.9720038390067591,
      "grad_norm": 0.5896902198048481,
      "learning_rate": 0.00010937764572689544,
      "loss": 12.0106,
      "step": 17850
    },
    {
      "epoch": 0.9720582930033421,
      "grad_norm": 0.5185716274990738,
      "learning_rate": 0.00010936886629188993,
      "loss": 12.0609,
      "step": 17851
    },
    {
      "epoch": 0.9721127469999251,
      "grad_norm": 0.5247556641643809,
      "learning_rate": 0.00010936008678403057,
      "loss": 12.214,
      "step": 17852
    },
    {
      "epoch": 0.9721672009965081,
      "grad_norm": 0.5813388227502093,
      "learning_rate": 0.0001093513072033856,
      "loss": 12.1157,
      "step": 17853
    },
    {
      "epoch": 0.9722216549930911,
      "grad_norm": 0.6682536334487343,
      "learning_rate": 0.00010934252755002328,
      "loss": 12.2233,
      "step": 17854
    },
    {
      "epoch": 0.9722761089896742,
      "grad_norm": 0.8240527314908686,
      "learning_rate": 0.00010933374782401191,
      "loss": 12.259,
      "step": 17855
    },
    {
      "epoch": 0.9723305629862572,
      "grad_norm": 0.5808292742415025,
      "learning_rate": 0.00010932496802541976,
      "loss": 12.0409,
      "step": 17856
    },
    {
      "epoch": 0.9723850169828402,
      "grad_norm": 0.6116237421436368,
      "learning_rate": 0.00010931618815431508,
      "loss": 12.0902,
      "step": 17857
    },
    {
      "epoch": 0.9724394709794232,
      "grad_norm": 0.5712556415067888,
      "learning_rate": 0.00010930740821076618,
      "loss": 12.1183,
      "step": 17858
    },
    {
      "epoch": 0.9724939249760062,
      "grad_norm": 0.5495703744822813,
      "learning_rate": 0.00010929862819484129,
      "loss": 12.2186,
      "step": 17859
    },
    {
      "epoch": 0.9725483789725892,
      "grad_norm": 0.53985474865941,
      "learning_rate": 0.0001092898481066087,
      "loss": 12.0988,
      "step": 17860
    },
    {
      "epoch": 0.9726028329691723,
      "grad_norm": 0.567346809786572,
      "learning_rate": 0.0001092810679461367,
      "loss": 11.9864,
      "step": 17861
    },
    {
      "epoch": 0.9726572869657553,
      "grad_norm": 0.5352723508895281,
      "learning_rate": 0.00010927228771349358,
      "loss": 12.0967,
      "step": 17862
    },
    {
      "epoch": 0.9727117409623383,
      "grad_norm": 0.5864110588587573,
      "learning_rate": 0.00010926350740874757,
      "loss": 12.35,
      "step": 17863
    },
    {
      "epoch": 0.9727661949589212,
      "grad_norm": 0.5903631071486342,
      "learning_rate": 0.000109254727031967,
      "loss": 12.145,
      "step": 17864
    },
    {
      "epoch": 0.9728206489555042,
      "grad_norm": 0.6475830505747637,
      "learning_rate": 0.0001092459465832201,
      "loss": 12.252,
      "step": 17865
    },
    {
      "epoch": 0.9728751029520873,
      "grad_norm": 0.5593561081555694,
      "learning_rate": 0.00010923716606257517,
      "loss": 12.1333,
      "step": 17866
    },
    {
      "epoch": 0.9729295569486703,
      "grad_norm": 0.6374503415738403,
      "learning_rate": 0.0001092283854701005,
      "loss": 12.0877,
      "step": 17867
    },
    {
      "epoch": 0.9729840109452533,
      "grad_norm": 0.5556889201637897,
      "learning_rate": 0.00010921960480586435,
      "loss": 12.1325,
      "step": 17868
    },
    {
      "epoch": 0.9730384649418363,
      "grad_norm": 0.6601778675459046,
      "learning_rate": 0.00010921082406993502,
      "loss": 12.1835,
      "step": 17869
    },
    {
      "epoch": 0.9730929189384193,
      "grad_norm": 0.6017919349112363,
      "learning_rate": 0.00010920204326238075,
      "loss": 12.1647,
      "step": 17870
    },
    {
      "epoch": 0.9731473729350023,
      "grad_norm": 0.5306760234908892,
      "learning_rate": 0.00010919326238326988,
      "loss": 12.0787,
      "step": 17871
    },
    {
      "epoch": 0.9732018269315854,
      "grad_norm": 0.5448361215848412,
      "learning_rate": 0.00010918448143267062,
      "loss": 12.1605,
      "step": 17872
    },
    {
      "epoch": 0.9732562809281684,
      "grad_norm": 0.5346807218112846,
      "learning_rate": 0.00010917570041065132,
      "loss": 12.1521,
      "step": 17873
    },
    {
      "epoch": 0.9733107349247514,
      "grad_norm": 0.5584559000184194,
      "learning_rate": 0.00010916691931728022,
      "loss": 12.1005,
      "step": 17874
    },
    {
      "epoch": 0.9733651889213344,
      "grad_norm": 0.5442613405654735,
      "learning_rate": 0.00010915813815262564,
      "loss": 12.1057,
      "step": 17875
    },
    {
      "epoch": 0.9734196429179174,
      "grad_norm": 0.5360045078582204,
      "learning_rate": 0.00010914935691675586,
      "loss": 12.136,
      "step": 17876
    },
    {
      "epoch": 0.9734740969145004,
      "grad_norm": 0.5755332612033531,
      "learning_rate": 0.00010914057560973909,
      "loss": 12.297,
      "step": 17877
    },
    {
      "epoch": 0.9735285509110835,
      "grad_norm": 0.5595959670494073,
      "learning_rate": 0.00010913179423164368,
      "loss": 12.2594,
      "step": 17878
    },
    {
      "epoch": 0.9735830049076665,
      "grad_norm": 0.47179394787793116,
      "learning_rate": 0.00010912301278253793,
      "loss": 12.0613,
      "step": 17879
    },
    {
      "epoch": 0.9736374589042495,
      "grad_norm": 0.6288161960743753,
      "learning_rate": 0.0001091142312624901,
      "loss": 12.1996,
      "step": 17880
    },
    {
      "epoch": 0.9736919129008325,
      "grad_norm": 0.5473793202384529,
      "learning_rate": 0.00010910544967156849,
      "loss": 12.1201,
      "step": 17881
    },
    {
      "epoch": 0.9737463668974154,
      "grad_norm": 0.5471473747021309,
      "learning_rate": 0.00010909666800984136,
      "loss": 12.1405,
      "step": 17882
    },
    {
      "epoch": 0.9738008208939984,
      "grad_norm": 0.5209045787765305,
      "learning_rate": 0.000109087886277377,
      "loss": 12.0672,
      "step": 17883
    },
    {
      "epoch": 0.9738552748905815,
      "grad_norm": 0.5390163506174233,
      "learning_rate": 0.00010907910447424373,
      "loss": 12.0384,
      "step": 17884
    },
    {
      "epoch": 0.9739097288871645,
      "grad_norm": 0.5157419432595162,
      "learning_rate": 0.00010907032260050982,
      "loss": 12.0893,
      "step": 17885
    },
    {
      "epoch": 0.9739641828837475,
      "grad_norm": 0.56821905247156,
      "learning_rate": 0.00010906154065624356,
      "loss": 12.1278,
      "step": 17886
    },
    {
      "epoch": 0.9740186368803305,
      "grad_norm": 0.5938279214721289,
      "learning_rate": 0.00010905275864151326,
      "loss": 12.1691,
      "step": 17887
    },
    {
      "epoch": 0.9740730908769135,
      "grad_norm": 0.599030570834302,
      "learning_rate": 0.00010904397655638717,
      "loss": 12.1988,
      "step": 17888
    },
    {
      "epoch": 0.9741275448734965,
      "grad_norm": 0.49742525863148496,
      "learning_rate": 0.00010903519440093357,
      "loss": 12.0655,
      "step": 17889
    },
    {
      "epoch": 0.9741819988700796,
      "grad_norm": 0.5680825248237323,
      "learning_rate": 0.00010902641217522083,
      "loss": 12.1229,
      "step": 17890
    },
    {
      "epoch": 0.9742364528666626,
      "grad_norm": 0.5945795928096985,
      "learning_rate": 0.00010901762987931718,
      "loss": 12.1609,
      "step": 17891
    },
    {
      "epoch": 0.9742909068632456,
      "grad_norm": 0.5741389455881379,
      "learning_rate": 0.00010900884751329095,
      "loss": 12.1743,
      "step": 17892
    },
    {
      "epoch": 0.9743453608598286,
      "grad_norm": 0.6398993803084989,
      "learning_rate": 0.00010900006507721036,
      "loss": 12.2676,
      "step": 17893
    },
    {
      "epoch": 0.9743998148564116,
      "grad_norm": 0.6475032989074506,
      "learning_rate": 0.00010899128257114377,
      "loss": 12.2419,
      "step": 17894
    },
    {
      "epoch": 0.9744542688529946,
      "grad_norm": 0.5941851830064357,
      "learning_rate": 0.00010898249999515949,
      "loss": 12.2465,
      "step": 17895
    },
    {
      "epoch": 0.9745087228495777,
      "grad_norm": 0.5181394519052931,
      "learning_rate": 0.00010897371734932578,
      "loss": 12.0279,
      "step": 17896
    },
    {
      "epoch": 0.9745631768461607,
      "grad_norm": 0.5359386713850717,
      "learning_rate": 0.00010896493463371092,
      "loss": 12.0701,
      "step": 17897
    },
    {
      "epoch": 0.9746176308427437,
      "grad_norm": 0.5777851464649929,
      "learning_rate": 0.00010895615184838324,
      "loss": 12.2015,
      "step": 17898
    },
    {
      "epoch": 0.9746720848393267,
      "grad_norm": 0.5159998961382526,
      "learning_rate": 0.00010894736899341101,
      "loss": 11.9164,
      "step": 17899
    },
    {
      "epoch": 0.9747265388359097,
      "grad_norm": 0.5260505807175065,
      "learning_rate": 0.00010893858606886251,
      "loss": 12.1511,
      "step": 17900
    },
    {
      "epoch": 0.9747809928324928,
      "grad_norm": 0.5636914625494629,
      "learning_rate": 0.00010892980307480612,
      "loss": 12.0698,
      "step": 17901
    },
    {
      "epoch": 0.9748354468290757,
      "grad_norm": 0.6476801604471305,
      "learning_rate": 0.00010892102001131006,
      "loss": 12.2996,
      "step": 17902
    },
    {
      "epoch": 0.9748899008256587,
      "grad_norm": 0.5250071769126085,
      "learning_rate": 0.00010891223687844266,
      "loss": 12.0254,
      "step": 17903
    },
    {
      "epoch": 0.9749443548222417,
      "grad_norm": 0.631637222132833,
      "learning_rate": 0.00010890345367627219,
      "loss": 12.2131,
      "step": 17904
    },
    {
      "epoch": 0.9749988088188247,
      "grad_norm": 0.6668428762840949,
      "learning_rate": 0.00010889467040486699,
      "loss": 12.0375,
      "step": 17905
    },
    {
      "epoch": 0.9750532628154077,
      "grad_norm": 0.550951569104823,
      "learning_rate": 0.00010888588706429532,
      "loss": 12.1446,
      "step": 17906
    },
    {
      "epoch": 0.9751077168119908,
      "grad_norm": 0.5678601252962668,
      "learning_rate": 0.00010887710365462554,
      "loss": 12.1614,
      "step": 17907
    },
    {
      "epoch": 0.9751621708085738,
      "grad_norm": 0.5540205006570619,
      "learning_rate": 0.00010886832017592588,
      "loss": 12.1959,
      "step": 17908
    },
    {
      "epoch": 0.9752166248051568,
      "grad_norm": 0.5464565421988227,
      "learning_rate": 0.00010885953662826467,
      "loss": 12.0261,
      "step": 17909
    },
    {
      "epoch": 0.9752710788017398,
      "grad_norm": 0.575540090418497,
      "learning_rate": 0.00010885075301171024,
      "loss": 12.0915,
      "step": 17910
    },
    {
      "epoch": 0.9753255327983228,
      "grad_norm": 0.6135515535481806,
      "learning_rate": 0.00010884196932633086,
      "loss": 12.1961,
      "step": 17911
    },
    {
      "epoch": 0.9753799867949058,
      "grad_norm": 0.6282773902043612,
      "learning_rate": 0.00010883318557219486,
      "loss": 12.0977,
      "step": 17912
    },
    {
      "epoch": 0.9754344407914889,
      "grad_norm": 0.5265484255082639,
      "learning_rate": 0.00010882440174937052,
      "loss": 12.0893,
      "step": 17913
    },
    {
      "epoch": 0.9754888947880719,
      "grad_norm": 0.5622450247447242,
      "learning_rate": 0.00010881561785792614,
      "loss": 12.1413,
      "step": 17914
    },
    {
      "epoch": 0.9755433487846549,
      "grad_norm": 0.6854546710007255,
      "learning_rate": 0.00010880683389793005,
      "loss": 12.108,
      "step": 17915
    },
    {
      "epoch": 0.9755978027812379,
      "grad_norm": 0.6047047159213987,
      "learning_rate": 0.00010879804986945053,
      "loss": 12.0257,
      "step": 17916
    },
    {
      "epoch": 0.9756522567778209,
      "grad_norm": 0.6069312642192195,
      "learning_rate": 0.0001087892657725559,
      "loss": 12.1501,
      "step": 17917
    },
    {
      "epoch": 0.9757067107744039,
      "grad_norm": 0.6288459884045596,
      "learning_rate": 0.00010878048160731447,
      "loss": 12.0399,
      "step": 17918
    },
    {
      "epoch": 0.975761164770987,
      "grad_norm": 0.548933130048583,
      "learning_rate": 0.00010877169737379454,
      "loss": 12.2037,
      "step": 17919
    },
    {
      "epoch": 0.97581561876757,
      "grad_norm": 0.5325861854351972,
      "learning_rate": 0.00010876291307206444,
      "loss": 12.1012,
      "step": 17920
    },
    {
      "epoch": 0.9758700727641529,
      "grad_norm": 0.5778575483593685,
      "learning_rate": 0.00010875412870219244,
      "loss": 12.1451,
      "step": 17921
    },
    {
      "epoch": 0.9759245267607359,
      "grad_norm": 0.7127994538232065,
      "learning_rate": 0.00010874534426424685,
      "loss": 12.2466,
      "step": 17922
    },
    {
      "epoch": 0.9759789807573189,
      "grad_norm": 0.6218779145716515,
      "learning_rate": 0.00010873655975829601,
      "loss": 12.0478,
      "step": 17923
    },
    {
      "epoch": 0.9760334347539019,
      "grad_norm": 0.6401297530367082,
      "learning_rate": 0.00010872777518440825,
      "loss": 12.1656,
      "step": 17924
    },
    {
      "epoch": 0.976087888750485,
      "grad_norm": 0.5689038896878923,
      "learning_rate": 0.00010871899054265179,
      "loss": 12.1548,
      "step": 17925
    },
    {
      "epoch": 0.976142342747068,
      "grad_norm": 0.6836915922090367,
      "learning_rate": 0.00010871020583309506,
      "loss": 12.1765,
      "step": 17926
    },
    {
      "epoch": 0.976196796743651,
      "grad_norm": 0.5045814336424027,
      "learning_rate": 0.00010870142105580626,
      "loss": 12.05,
      "step": 17927
    },
    {
      "epoch": 0.976251250740234,
      "grad_norm": 0.5209395539749949,
      "learning_rate": 0.00010869263621085374,
      "loss": 12.1797,
      "step": 17928
    },
    {
      "epoch": 0.976305704736817,
      "grad_norm": 0.6326383548968321,
      "learning_rate": 0.00010868385129830587,
      "loss": 12.1397,
      "step": 17929
    },
    {
      "epoch": 0.9763601587334,
      "grad_norm": 0.5497160922094432,
      "learning_rate": 0.0001086750663182309,
      "loss": 12.0658,
      "step": 17930
    },
    {
      "epoch": 0.9764146127299831,
      "grad_norm": 0.6505528718449232,
      "learning_rate": 0.00010866628127069716,
      "loss": 12.0582,
      "step": 17931
    },
    {
      "epoch": 0.9764690667265661,
      "grad_norm": 0.5561586077183442,
      "learning_rate": 0.00010865749615577295,
      "loss": 12.0352,
      "step": 17932
    },
    {
      "epoch": 0.9765235207231491,
      "grad_norm": 0.5374969245163094,
      "learning_rate": 0.0001086487109735266,
      "loss": 12.1648,
      "step": 17933
    },
    {
      "epoch": 0.9765779747197321,
      "grad_norm": 0.5859100683304503,
      "learning_rate": 0.00010863992572402642,
      "loss": 12.1309,
      "step": 17934
    },
    {
      "epoch": 0.9766324287163151,
      "grad_norm": 0.5627313103501189,
      "learning_rate": 0.00010863114040734075,
      "loss": 12.1343,
      "step": 17935
    },
    {
      "epoch": 0.9766868827128982,
      "grad_norm": 0.527947713424401,
      "learning_rate": 0.00010862235502353788,
      "loss": 11.9982,
      "step": 17936
    },
    {
      "epoch": 0.9767413367094812,
      "grad_norm": 0.5412487045573909,
      "learning_rate": 0.00010861356957268613,
      "loss": 12.1632,
      "step": 17937
    },
    {
      "epoch": 0.9767957907060641,
      "grad_norm": 0.5426660693000421,
      "learning_rate": 0.00010860478405485379,
      "loss": 12.1363,
      "step": 17938
    },
    {
      "epoch": 0.9768502447026471,
      "grad_norm": 0.5833797058817659,
      "learning_rate": 0.00010859599847010922,
      "loss": 12.0881,
      "step": 17939
    },
    {
      "epoch": 0.9769046986992301,
      "grad_norm": 0.6033501398873252,
      "learning_rate": 0.00010858721281852074,
      "loss": 12.2206,
      "step": 17940
    },
    {
      "epoch": 0.9769591526958131,
      "grad_norm": 0.5812379565297366,
      "learning_rate": 0.00010857842710015666,
      "loss": 12.1085,
      "step": 17941
    },
    {
      "epoch": 0.9770136066923962,
      "grad_norm": 0.575416761901028,
      "learning_rate": 0.00010856964131508526,
      "loss": 12.0669,
      "step": 17942
    },
    {
      "epoch": 0.9770680606889792,
      "grad_norm": 0.5377194087122438,
      "learning_rate": 0.0001085608554633749,
      "loss": 12.1132,
      "step": 17943
    },
    {
      "epoch": 0.9771225146855622,
      "grad_norm": 0.6368059974770356,
      "learning_rate": 0.00010855206954509391,
      "loss": 12.1264,
      "step": 17944
    },
    {
      "epoch": 0.9771769686821452,
      "grad_norm": 0.5726633456620404,
      "learning_rate": 0.00010854328356031059,
      "loss": 12.1398,
      "step": 17945
    },
    {
      "epoch": 0.9772314226787282,
      "grad_norm": 0.5546276963249545,
      "learning_rate": 0.00010853449750909328,
      "loss": 11.9444,
      "step": 17946
    },
    {
      "epoch": 0.9772858766753112,
      "grad_norm": 0.5278076544779617,
      "learning_rate": 0.00010852571139151027,
      "loss": 12.1496,
      "step": 17947
    },
    {
      "epoch": 0.9773403306718943,
      "grad_norm": 0.5866875037092192,
      "learning_rate": 0.00010851692520762989,
      "loss": 11.8878,
      "step": 17948
    },
    {
      "epoch": 0.9773947846684773,
      "grad_norm": 0.6536364797697218,
      "learning_rate": 0.00010850813895752043,
      "loss": 12.1626,
      "step": 17949
    },
    {
      "epoch": 0.9774492386650603,
      "grad_norm": 0.500705799452207,
      "learning_rate": 0.00010849935264125031,
      "loss": 12.1102,
      "step": 17950
    },
    {
      "epoch": 0.9775036926616433,
      "grad_norm": 0.6214715212457284,
      "learning_rate": 0.0001084905662588878,
      "loss": 12.2064,
      "step": 17951
    },
    {
      "epoch": 0.9775581466582263,
      "grad_norm": 0.5569672393862045,
      "learning_rate": 0.00010848177981050123,
      "loss": 12.099,
      "step": 17952
    },
    {
      "epoch": 0.9776126006548093,
      "grad_norm": 0.5271052084321294,
      "learning_rate": 0.0001084729932961589,
      "loss": 12.134,
      "step": 17953
    },
    {
      "epoch": 0.9776670546513924,
      "grad_norm": 0.5813167820322008,
      "learning_rate": 0.00010846420671592913,
      "loss": 11.932,
      "step": 17954
    },
    {
      "epoch": 0.9777215086479754,
      "grad_norm": 0.5652022496642765,
      "learning_rate": 0.00010845542006988028,
      "loss": 12.084,
      "step": 17955
    },
    {
      "epoch": 0.9777759626445583,
      "grad_norm": 0.6489469752678734,
      "learning_rate": 0.0001084466333580807,
      "loss": 12.1296,
      "step": 17956
    },
    {
      "epoch": 0.9778304166411413,
      "grad_norm": 0.5975893786025105,
      "learning_rate": 0.00010843784658059865,
      "loss": 12.0141,
      "step": 17957
    },
    {
      "epoch": 0.9778848706377243,
      "grad_norm": 0.5519125805054584,
      "learning_rate": 0.0001084290597375025,
      "loss": 12.2335,
      "step": 17958
    },
    {
      "epoch": 0.9779393246343073,
      "grad_norm": 0.5395481619309116,
      "learning_rate": 0.00010842027282886059,
      "loss": 11.9542,
      "step": 17959
    },
    {
      "epoch": 0.9779937786308904,
      "grad_norm": 0.6353045346707804,
      "learning_rate": 0.00010841148585474117,
      "loss": 12.1194,
      "step": 17960
    },
    {
      "epoch": 0.9780482326274734,
      "grad_norm": 0.5823209191961672,
      "learning_rate": 0.00010840269881521268,
      "loss": 12.1265,
      "step": 17961
    },
    {
      "epoch": 0.9781026866240564,
      "grad_norm": 0.5776965927558751,
      "learning_rate": 0.00010839391171034336,
      "loss": 12.0926,
      "step": 17962
    },
    {
      "epoch": 0.9781571406206394,
      "grad_norm": 0.6333194385576171,
      "learning_rate": 0.0001083851245402016,
      "loss": 12.1832,
      "step": 17963
    },
    {
      "epoch": 0.9782115946172224,
      "grad_norm": 0.6352767740368886,
      "learning_rate": 0.0001083763373048557,
      "loss": 12.0946,
      "step": 17964
    },
    {
      "epoch": 0.9782660486138054,
      "grad_norm": 0.5645207598118537,
      "learning_rate": 0.00010836755000437397,
      "loss": 12.0763,
      "step": 17965
    },
    {
      "epoch": 0.9783205026103885,
      "grad_norm": 0.5937879276484013,
      "learning_rate": 0.00010835876263882481,
      "loss": 12.0289,
      "step": 17966
    },
    {
      "epoch": 0.9783749566069715,
      "grad_norm": 0.6074090833865735,
      "learning_rate": 0.00010834997520827648,
      "loss": 12.1025,
      "step": 17967
    },
    {
      "epoch": 0.9784294106035545,
      "grad_norm": 0.5633156870660759,
      "learning_rate": 0.00010834118771279736,
      "loss": 12.1154,
      "step": 17968
    },
    {
      "epoch": 0.9784838646001375,
      "grad_norm": 0.5320985531607659,
      "learning_rate": 0.00010833240015245573,
      "loss": 12.1585,
      "step": 17969
    },
    {
      "epoch": 0.9785383185967205,
      "grad_norm": 0.6191167257525989,
      "learning_rate": 0.00010832361252731999,
      "loss": 11.9955,
      "step": 17970
    },
    {
      "epoch": 0.9785927725933036,
      "grad_norm": 0.5422379859682265,
      "learning_rate": 0.00010831482483745845,
      "loss": 12.1218,
      "step": 17971
    },
    {
      "epoch": 0.9786472265898866,
      "grad_norm": 0.5304028876755441,
      "learning_rate": 0.00010830603708293942,
      "loss": 12.2471,
      "step": 17972
    },
    {
      "epoch": 0.9787016805864696,
      "grad_norm": 0.6449485704565627,
      "learning_rate": 0.00010829724926383127,
      "loss": 12.225,
      "step": 17973
    },
    {
      "epoch": 0.9787561345830526,
      "grad_norm": 0.5306371704220962,
      "learning_rate": 0.0001082884613802023,
      "loss": 12.1296,
      "step": 17974
    },
    {
      "epoch": 0.9788105885796355,
      "grad_norm": 0.5167911108564022,
      "learning_rate": 0.00010827967343212087,
      "loss": 12.0481,
      "step": 17975
    },
    {
      "epoch": 0.9788650425762185,
      "grad_norm": 0.5274856214728664,
      "learning_rate": 0.00010827088541965531,
      "loss": 12.1757,
      "step": 17976
    },
    {
      "epoch": 0.9789194965728016,
      "grad_norm": 0.5521135327126813,
      "learning_rate": 0.00010826209734287396,
      "loss": 12.0129,
      "step": 17977
    },
    {
      "epoch": 0.9789739505693846,
      "grad_norm": 0.5761098151717915,
      "learning_rate": 0.00010825330920184515,
      "loss": 12.2367,
      "step": 17978
    },
    {
      "epoch": 0.9790284045659676,
      "grad_norm": 0.5076363892495853,
      "learning_rate": 0.00010824452099663725,
      "loss": 12.1218,
      "step": 17979
    },
    {
      "epoch": 0.9790828585625506,
      "grad_norm": 0.5273708890063069,
      "learning_rate": 0.00010823573272731855,
      "loss": 12.0445,
      "step": 17980
    },
    {
      "epoch": 0.9791373125591336,
      "grad_norm": 0.546803219978149,
      "learning_rate": 0.0001082269443939574,
      "loss": 12.1183,
      "step": 17981
    },
    {
      "epoch": 0.9791917665557166,
      "grad_norm": 0.5666503969234457,
      "learning_rate": 0.00010821815599662217,
      "loss": 12.1089,
      "step": 17982
    },
    {
      "epoch": 0.9792462205522997,
      "grad_norm": 0.5976456319310408,
      "learning_rate": 0.00010820936753538118,
      "loss": 12.2222,
      "step": 17983
    },
    {
      "epoch": 0.9793006745488827,
      "grad_norm": 0.5807534805493791,
      "learning_rate": 0.00010820057901030276,
      "loss": 12.1417,
      "step": 17984
    },
    {
      "epoch": 0.9793551285454657,
      "grad_norm": 0.5618407017610094,
      "learning_rate": 0.00010819179042145527,
      "loss": 12.1541,
      "step": 17985
    },
    {
      "epoch": 0.9794095825420487,
      "grad_norm": 0.5399095579670065,
      "learning_rate": 0.00010818300176890703,
      "loss": 12.165,
      "step": 17986
    },
    {
      "epoch": 0.9794640365386317,
      "grad_norm": 0.5595505093425492,
      "learning_rate": 0.00010817421305272642,
      "loss": 11.9565,
      "step": 17987
    },
    {
      "epoch": 0.9795184905352147,
      "grad_norm": 0.6173433296027873,
      "learning_rate": 0.00010816542427298173,
      "loss": 12.1378,
      "step": 17988
    },
    {
      "epoch": 0.9795729445317978,
      "grad_norm": 0.5635028286575519,
      "learning_rate": 0.00010815663542974135,
      "loss": 12.0596,
      "step": 17989
    },
    {
      "epoch": 0.9796273985283808,
      "grad_norm": 0.5128177697315897,
      "learning_rate": 0.00010814784652307361,
      "loss": 12.1665,
      "step": 17990
    },
    {
      "epoch": 0.9796818525249638,
      "grad_norm": 0.5451806256854534,
      "learning_rate": 0.00010813905755304686,
      "loss": 12.0466,
      "step": 17991
    },
    {
      "epoch": 0.9797363065215468,
      "grad_norm": 0.546160671865186,
      "learning_rate": 0.00010813026851972941,
      "loss": 12.2315,
      "step": 17992
    },
    {
      "epoch": 0.9797907605181297,
      "grad_norm": 0.6423773110763001,
      "learning_rate": 0.00010812147942318961,
      "loss": 12.0426,
      "step": 17993
    },
    {
      "epoch": 0.9798452145147127,
      "grad_norm": 0.49260296577336354,
      "learning_rate": 0.00010811269026349587,
      "loss": 12.0204,
      "step": 17994
    },
    {
      "epoch": 0.9798996685112958,
      "grad_norm": 0.5733819918469207,
      "learning_rate": 0.00010810390104071646,
      "loss": 12.1513,
      "step": 17995
    },
    {
      "epoch": 0.9799541225078788,
      "grad_norm": 0.5574227799978274,
      "learning_rate": 0.00010809511175491976,
      "loss": 12.0072,
      "step": 17996
    },
    {
      "epoch": 0.9800085765044618,
      "grad_norm": 0.5219117221008631,
      "learning_rate": 0.00010808632240617414,
      "loss": 12.1725,
      "step": 17997
    },
    {
      "epoch": 0.9800630305010448,
      "grad_norm": 0.5741168714520937,
      "learning_rate": 0.00010807753299454789,
      "loss": 11.9589,
      "step": 17998
    },
    {
      "epoch": 0.9801174844976278,
      "grad_norm": 0.5677294057678615,
      "learning_rate": 0.00010806874352010938,
      "loss": 12.0291,
      "step": 17999
    },
    {
      "epoch": 0.9801719384942109,
      "grad_norm": 0.5318485484630929,
      "learning_rate": 0.00010805995398292699,
      "loss": 12.1372,
      "step": 18000
    },
    {
      "epoch": 0.9802263924907939,
      "grad_norm": 0.5543616701944382,
      "learning_rate": 0.00010805116438306903,
      "loss": 12.1638,
      "step": 18001
    },
    {
      "epoch": 0.9802808464873769,
      "grad_norm": 0.5477869601113502,
      "learning_rate": 0.0001080423747206039,
      "loss": 12.0025,
      "step": 18002
    },
    {
      "epoch": 0.9803353004839599,
      "grad_norm": 0.5230337758187195,
      "learning_rate": 0.0001080335849955999,
      "loss": 12.0703,
      "step": 18003
    },
    {
      "epoch": 0.9803897544805429,
      "grad_norm": 0.5306094902096865,
      "learning_rate": 0.00010802479520812536,
      "loss": 11.9167,
      "step": 18004
    },
    {
      "epoch": 0.9804442084771259,
      "grad_norm": 0.6366613017358728,
      "learning_rate": 0.00010801600535824869,
      "loss": 12.1444,
      "step": 18005
    },
    {
      "epoch": 0.980498662473709,
      "grad_norm": 0.5727482572658238,
      "learning_rate": 0.00010800721544603822,
      "loss": 12.0977,
      "step": 18006
    },
    {
      "epoch": 0.980553116470292,
      "grad_norm": 0.5089284317859405,
      "learning_rate": 0.00010799842547156234,
      "loss": 12.0473,
      "step": 18007
    },
    {
      "epoch": 0.980607570466875,
      "grad_norm": 0.5667934645021488,
      "learning_rate": 0.00010798963543488932,
      "loss": 12.137,
      "step": 18008
    },
    {
      "epoch": 0.980662024463458,
      "grad_norm": 0.6196388965212127,
      "learning_rate": 0.00010798084533608754,
      "loss": 12.2894,
      "step": 18009
    },
    {
      "epoch": 0.980716478460041,
      "grad_norm": 0.562464564711028,
      "learning_rate": 0.0001079720551752254,
      "loss": 12.1624,
      "step": 18010
    },
    {
      "epoch": 0.980770932456624,
      "grad_norm": 0.5348173101555526,
      "learning_rate": 0.0001079632649523712,
      "loss": 12.2295,
      "step": 18011
    },
    {
      "epoch": 0.980825386453207,
      "grad_norm": 0.5570937732739201,
      "learning_rate": 0.00010795447466759335,
      "loss": 12.1079,
      "step": 18012
    },
    {
      "epoch": 0.98087984044979,
      "grad_norm": 0.5647120952256477,
      "learning_rate": 0.00010794568432096015,
      "loss": 12.1854,
      "step": 18013
    },
    {
      "epoch": 0.980934294446373,
      "grad_norm": 0.5815748623492203,
      "learning_rate": 0.00010793689391253996,
      "loss": 12.1952,
      "step": 18014
    },
    {
      "epoch": 0.980988748442956,
      "grad_norm": 0.5537602408110703,
      "learning_rate": 0.00010792810344240114,
      "loss": 12.1408,
      "step": 18015
    },
    {
      "epoch": 0.981043202439539,
      "grad_norm": 0.540086712463153,
      "learning_rate": 0.0001079193129106121,
      "loss": 12.0253,
      "step": 18016
    },
    {
      "epoch": 0.981097656436122,
      "grad_norm": 0.7435614062524526,
      "learning_rate": 0.00010791052231724115,
      "loss": 12.1708,
      "step": 18017
    },
    {
      "epoch": 0.9811521104327051,
      "grad_norm": 0.5266529019604518,
      "learning_rate": 0.00010790173166235665,
      "loss": 12.1274,
      "step": 18018
    },
    {
      "epoch": 0.9812065644292881,
      "grad_norm": 0.5536468972179033,
      "learning_rate": 0.00010789294094602694,
      "loss": 12.1007,
      "step": 18019
    },
    {
      "epoch": 0.9812610184258711,
      "grad_norm": 0.644183661388461,
      "learning_rate": 0.00010788415016832039,
      "loss": 12.0889,
      "step": 18020
    },
    {
      "epoch": 0.9813154724224541,
      "grad_norm": 0.5300747203805085,
      "learning_rate": 0.00010787535932930537,
      "loss": 12.1142,
      "step": 18021
    },
    {
      "epoch": 0.9813699264190371,
      "grad_norm": 0.5795795154313257,
      "learning_rate": 0.00010786656842905028,
      "loss": 12.0046,
      "step": 18022
    },
    {
      "epoch": 0.9814243804156201,
      "grad_norm": 0.5725988801129808,
      "learning_rate": 0.0001078577774676234,
      "loss": 12.2399,
      "step": 18023
    },
    {
      "epoch": 0.9814788344122032,
      "grad_norm": 0.5818950412111472,
      "learning_rate": 0.00010784898644509313,
      "loss": 12.0231,
      "step": 18024
    },
    {
      "epoch": 0.9815332884087862,
      "grad_norm": 0.5411298785123068,
      "learning_rate": 0.00010784019536152783,
      "loss": 11.8646,
      "step": 18025
    },
    {
      "epoch": 0.9815877424053692,
      "grad_norm": 0.5472834620427034,
      "learning_rate": 0.00010783140421699582,
      "loss": 11.995,
      "step": 18026
    },
    {
      "epoch": 0.9816421964019522,
      "grad_norm": 0.5295784041111941,
      "learning_rate": 0.00010782261301156555,
      "loss": 12.1112,
      "step": 18027
    },
    {
      "epoch": 0.9816966503985352,
      "grad_norm": 0.5634509356491804,
      "learning_rate": 0.0001078138217453053,
      "loss": 12.0938,
      "step": 18028
    },
    {
      "epoch": 0.9817511043951181,
      "grad_norm": 0.5739913712880079,
      "learning_rate": 0.00010780503041828347,
      "loss": 12.1133,
      "step": 18029
    },
    {
      "epoch": 0.9818055583917012,
      "grad_norm": 0.5719679045022055,
      "learning_rate": 0.00010779623903056842,
      "loss": 12.183,
      "step": 18030
    },
    {
      "epoch": 0.9818600123882842,
      "grad_norm": 0.5433390891457947,
      "learning_rate": 0.0001077874475822285,
      "loss": 12.072,
      "step": 18031
    },
    {
      "epoch": 0.9819144663848672,
      "grad_norm": 0.596004488459137,
      "learning_rate": 0.00010777865607333208,
      "loss": 12.0495,
      "step": 18032
    },
    {
      "epoch": 0.9819689203814502,
      "grad_norm": 0.5773870752684507,
      "learning_rate": 0.00010776986450394753,
      "loss": 12.2028,
      "step": 18033
    },
    {
      "epoch": 0.9820233743780332,
      "grad_norm": 0.5502718478168531,
      "learning_rate": 0.0001077610728741432,
      "loss": 12.0119,
      "step": 18034
    },
    {
      "epoch": 0.9820778283746163,
      "grad_norm": 0.6144172766414152,
      "learning_rate": 0.00010775228118398748,
      "loss": 12.1112,
      "step": 18035
    },
    {
      "epoch": 0.9821322823711993,
      "grad_norm": 0.5632127743324946,
      "learning_rate": 0.0001077434894335487,
      "loss": 12.0601,
      "step": 18036
    },
    {
      "epoch": 0.9821867363677823,
      "grad_norm": 0.5986532670411304,
      "learning_rate": 0.00010773469762289528,
      "loss": 12.1375,
      "step": 18037
    },
    {
      "epoch": 0.9822411903643653,
      "grad_norm": 0.5348121444254492,
      "learning_rate": 0.00010772590575209553,
      "loss": 12.0448,
      "step": 18038
    },
    {
      "epoch": 0.9822956443609483,
      "grad_norm": 0.6977366636723346,
      "learning_rate": 0.00010771711382121786,
      "loss": 12.0596,
      "step": 18039
    },
    {
      "epoch": 0.9823500983575313,
      "grad_norm": 0.5281573876140327,
      "learning_rate": 0.0001077083218303306,
      "loss": 12.1564,
      "step": 18040
    },
    {
      "epoch": 0.9824045523541144,
      "grad_norm": 0.6582946415270958,
      "learning_rate": 0.00010769952977950215,
      "loss": 12.2643,
      "step": 18041
    },
    {
      "epoch": 0.9824590063506974,
      "grad_norm": 0.5929653534678702,
      "learning_rate": 0.00010769073766880085,
      "loss": 12.0594,
      "step": 18042
    },
    {
      "epoch": 0.9825134603472804,
      "grad_norm": 0.5457434530855422,
      "learning_rate": 0.00010768194549829508,
      "loss": 12.164,
      "step": 18043
    },
    {
      "epoch": 0.9825679143438634,
      "grad_norm": 0.5500436007280676,
      "learning_rate": 0.00010767315326805323,
      "loss": 12.2298,
      "step": 18044
    },
    {
      "epoch": 0.9826223683404464,
      "grad_norm": 0.5224746472815626,
      "learning_rate": 0.00010766436097814365,
      "loss": 12.134,
      "step": 18045
    },
    {
      "epoch": 0.9826768223370294,
      "grad_norm": 0.5619769872720628,
      "learning_rate": 0.00010765556862863472,
      "loss": 11.9188,
      "step": 18046
    },
    {
      "epoch": 0.9827312763336125,
      "grad_norm": 0.5357558992994437,
      "learning_rate": 0.0001076467762195948,
      "loss": 12.1788,
      "step": 18047
    },
    {
      "epoch": 0.9827857303301955,
      "grad_norm": 0.5352705390795878,
      "learning_rate": 0.00010763798375109226,
      "loss": 11.9371,
      "step": 18048
    },
    {
      "epoch": 0.9828401843267784,
      "grad_norm": 0.5181102203185849,
      "learning_rate": 0.00010762919122319548,
      "loss": 12.072,
      "step": 18049
    },
    {
      "epoch": 0.9828946383233614,
      "grad_norm": 0.5860233741332485,
      "learning_rate": 0.00010762039863597284,
      "loss": 12.1642,
      "step": 18050
    },
    {
      "epoch": 0.9829490923199444,
      "grad_norm": 0.5729014679436384,
      "learning_rate": 0.00010761160598949269,
      "loss": 12.2005,
      "step": 18051
    },
    {
      "epoch": 0.9830035463165274,
      "grad_norm": 0.5920741834598858,
      "learning_rate": 0.00010760281328382344,
      "loss": 12.2143,
      "step": 18052
    },
    {
      "epoch": 0.9830580003131105,
      "grad_norm": 0.5143756884138256,
      "learning_rate": 0.0001075940205190334,
      "loss": 12.152,
      "step": 18053
    },
    {
      "epoch": 0.9831124543096935,
      "grad_norm": 0.565118066991634,
      "learning_rate": 0.000107585227695191,
      "loss": 12.0064,
      "step": 18054
    },
    {
      "epoch": 0.9831669083062765,
      "grad_norm": 0.5744740854247854,
      "learning_rate": 0.0001075764348123646,
      "loss": 11.9919,
      "step": 18055
    },
    {
      "epoch": 0.9832213623028595,
      "grad_norm": 0.6331700601523153,
      "learning_rate": 0.00010756764187062257,
      "loss": 12.0929,
      "step": 18056
    },
    {
      "epoch": 0.9832758162994425,
      "grad_norm": 0.6835017108603204,
      "learning_rate": 0.00010755884887003331,
      "loss": 12.2208,
      "step": 18057
    },
    {
      "epoch": 0.9833302702960255,
      "grad_norm": 0.5656699014455033,
      "learning_rate": 0.00010755005581066513,
      "loss": 12.1017,
      "step": 18058
    },
    {
      "epoch": 0.9833847242926086,
      "grad_norm": 0.5311723015634895,
      "learning_rate": 0.00010754126269258647,
      "loss": 11.9106,
      "step": 18059
    },
    {
      "epoch": 0.9834391782891916,
      "grad_norm": 0.5720387434249207,
      "learning_rate": 0.00010753246951586568,
      "loss": 12.0141,
      "step": 18060
    },
    {
      "epoch": 0.9834936322857746,
      "grad_norm": 0.6936595517459607,
      "learning_rate": 0.00010752367628057116,
      "loss": 12.189,
      "step": 18061
    },
    {
      "epoch": 0.9835480862823576,
      "grad_norm": 0.4982052031486083,
      "learning_rate": 0.00010751488298677128,
      "loss": 12.0908,
      "step": 18062
    },
    {
      "epoch": 0.9836025402789406,
      "grad_norm": 0.5599579395059051,
      "learning_rate": 0.00010750608963453438,
      "loss": 12.1435,
      "step": 18063
    },
    {
      "epoch": 0.9836569942755236,
      "grad_norm": 0.5998340544163681,
      "learning_rate": 0.00010749729622392888,
      "loss": 12.075,
      "step": 18064
    },
    {
      "epoch": 0.9837114482721067,
      "grad_norm": 0.5876829349332333,
      "learning_rate": 0.00010748850275502316,
      "loss": 12.0845,
      "step": 18065
    },
    {
      "epoch": 0.9837659022686897,
      "grad_norm": 0.5161903199408583,
      "learning_rate": 0.00010747970922788557,
      "loss": 11.9695,
      "step": 18066
    },
    {
      "epoch": 0.9838203562652726,
      "grad_norm": 0.6270574736815702,
      "learning_rate": 0.00010747091564258454,
      "loss": 12.1766,
      "step": 18067
    },
    {
      "epoch": 0.9838748102618556,
      "grad_norm": 0.5369537586885043,
      "learning_rate": 0.00010746212199918838,
      "loss": 12.0638,
      "step": 18068
    },
    {
      "epoch": 0.9839292642584386,
      "grad_norm": 0.5853144281322175,
      "learning_rate": 0.00010745332829776548,
      "loss": 11.9787,
      "step": 18069
    },
    {
      "epoch": 0.9839837182550217,
      "grad_norm": 0.60185097776034,
      "learning_rate": 0.0001074445345383843,
      "loss": 12.009,
      "step": 18070
    },
    {
      "epoch": 0.9840381722516047,
      "grad_norm": 0.5639261433739271,
      "learning_rate": 0.00010743574072111315,
      "loss": 12.104,
      "step": 18071
    },
    {
      "epoch": 0.9840926262481877,
      "grad_norm": 0.5537166577957787,
      "learning_rate": 0.00010742694684602046,
      "loss": 11.9331,
      "step": 18072
    },
    {
      "epoch": 0.9841470802447707,
      "grad_norm": 0.6591769633023632,
      "learning_rate": 0.00010741815291317459,
      "loss": 12.3092,
      "step": 18073
    },
    {
      "epoch": 0.9842015342413537,
      "grad_norm": 0.5947384769187573,
      "learning_rate": 0.00010740935892264387,
      "loss": 12.2152,
      "step": 18074
    },
    {
      "epoch": 0.9842559882379367,
      "grad_norm": 0.5599849847542595,
      "learning_rate": 0.00010740056487449674,
      "loss": 12.1216,
      "step": 18075
    },
    {
      "epoch": 0.9843104422345198,
      "grad_norm": 0.5517022014958264,
      "learning_rate": 0.0001073917707688016,
      "loss": 12.0721,
      "step": 18076
    },
    {
      "epoch": 0.9843648962311028,
      "grad_norm": 0.618082533686037,
      "learning_rate": 0.00010738297660562682,
      "loss": 12.0813,
      "step": 18077
    },
    {
      "epoch": 0.9844193502276858,
      "grad_norm": 0.5772571839574674,
      "learning_rate": 0.00010737418238504078,
      "loss": 12.195,
      "step": 18078
    },
    {
      "epoch": 0.9844738042242688,
      "grad_norm": 0.5673878961561942,
      "learning_rate": 0.00010736538810711184,
      "loss": 12.0617,
      "step": 18079
    },
    {
      "epoch": 0.9845282582208518,
      "grad_norm": 0.5546780893853501,
      "learning_rate": 0.00010735659377190841,
      "loss": 12.0691,
      "step": 18080
    },
    {
      "epoch": 0.9845827122174348,
      "grad_norm": 0.5853216934400048,
      "learning_rate": 0.00010734779937949883,
      "loss": 12.1178,
      "step": 18081
    },
    {
      "epoch": 0.9846371662140179,
      "grad_norm": 0.5358952268618732,
      "learning_rate": 0.0001073390049299516,
      "loss": 11.9943,
      "step": 18082
    },
    {
      "epoch": 0.9846916202106009,
      "grad_norm": 0.5755377660157842,
      "learning_rate": 0.00010733021042333502,
      "loss": 12.1944,
      "step": 18083
    },
    {
      "epoch": 0.9847460742071839,
      "grad_norm": 0.566788817104964,
      "learning_rate": 0.00010732141585971747,
      "loss": 12.1504,
      "step": 18084
    },
    {
      "epoch": 0.9848005282037668,
      "grad_norm": 0.6196705651253785,
      "learning_rate": 0.00010731262123916739,
      "loss": 12.1013,
      "step": 18085
    },
    {
      "epoch": 0.9848549822003498,
      "grad_norm": 0.5980150147717634,
      "learning_rate": 0.00010730382656175311,
      "loss": 11.9677,
      "step": 18086
    },
    {
      "epoch": 0.9849094361969328,
      "grad_norm": 0.5577501627316889,
      "learning_rate": 0.00010729503182754308,
      "loss": 12.1253,
      "step": 18087
    },
    {
      "epoch": 0.9849638901935159,
      "grad_norm": 0.4919465787672597,
      "learning_rate": 0.00010728623703660562,
      "loss": 12.0263,
      "step": 18088
    },
    {
      "epoch": 0.9850183441900989,
      "grad_norm": 0.5853269404716969,
      "learning_rate": 0.00010727744218900921,
      "loss": 12.1506,
      "step": 18089
    },
    {
      "epoch": 0.9850727981866819,
      "grad_norm": 0.6055052020375156,
      "learning_rate": 0.00010726864728482215,
      "loss": 12.0832,
      "step": 18090
    },
    {
      "epoch": 0.9851272521832649,
      "grad_norm": 0.625309898276744,
      "learning_rate": 0.00010725985232411288,
      "loss": 12.15,
      "step": 18091
    },
    {
      "epoch": 0.9851817061798479,
      "grad_norm": 0.5676098635808333,
      "learning_rate": 0.0001072510573069498,
      "loss": 12.0437,
      "step": 18092
    },
    {
      "epoch": 0.9852361601764309,
      "grad_norm": 0.5124787306749896,
      "learning_rate": 0.00010724226223340125,
      "loss": 12.1212,
      "step": 18093
    },
    {
      "epoch": 0.985290614173014,
      "grad_norm": 0.5933495437373781,
      "learning_rate": 0.00010723346710353568,
      "loss": 11.9825,
      "step": 18094
    },
    {
      "epoch": 0.985345068169597,
      "grad_norm": 0.6130369637735575,
      "learning_rate": 0.00010722467191742144,
      "loss": 12.0794,
      "step": 18095
    },
    {
      "epoch": 0.98539952216618,
      "grad_norm": 0.5928905745980033,
      "learning_rate": 0.00010721587667512695,
      "loss": 12.1596,
      "step": 18096
    },
    {
      "epoch": 0.985453976162763,
      "grad_norm": 0.5369197231079916,
      "learning_rate": 0.0001072070813767206,
      "loss": 11.9061,
      "step": 18097
    },
    {
      "epoch": 0.985508430159346,
      "grad_norm": 0.520335964142194,
      "learning_rate": 0.00010719828602227075,
      "loss": 12.0001,
      "step": 18098
    },
    {
      "epoch": 0.985562884155929,
      "grad_norm": 0.5918418697364416,
      "learning_rate": 0.00010718949061184585,
      "loss": 12.091,
      "step": 18099
    },
    {
      "epoch": 0.9856173381525121,
      "grad_norm": 1.0030124548635482,
      "learning_rate": 0.00010718069514551426,
      "loss": 12.1456,
      "step": 18100
    },
    {
      "epoch": 0.9856717921490951,
      "grad_norm": 0.6393789302062024,
      "learning_rate": 0.00010717189962334437,
      "loss": 12.1362,
      "step": 18101
    },
    {
      "epoch": 0.985726246145678,
      "grad_norm": 0.5601102862060541,
      "learning_rate": 0.00010716310404540459,
      "loss": 12.0762,
      "step": 18102
    },
    {
      "epoch": 0.985780700142261,
      "grad_norm": 0.5786649803546119,
      "learning_rate": 0.0001071543084117633,
      "loss": 12.0578,
      "step": 18103
    },
    {
      "epoch": 0.985835154138844,
      "grad_norm": 0.5872084322909208,
      "learning_rate": 0.00010714551272248891,
      "loss": 12.2205,
      "step": 18104
    },
    {
      "epoch": 0.9858896081354271,
      "grad_norm": 0.5060151325768041,
      "learning_rate": 0.00010713671697764984,
      "loss": 12.0591,
      "step": 18105
    },
    {
      "epoch": 0.9859440621320101,
      "grad_norm": 0.5528163577261255,
      "learning_rate": 0.00010712792117731445,
      "loss": 12.0899,
      "step": 18106
    },
    {
      "epoch": 0.9859985161285931,
      "grad_norm": 0.565991675505829,
      "learning_rate": 0.00010711912532155115,
      "loss": 12.1375,
      "step": 18107
    },
    {
      "epoch": 0.9860529701251761,
      "grad_norm": 0.5447176746423495,
      "learning_rate": 0.0001071103294104283,
      "loss": 12.0478,
      "step": 18108
    },
    {
      "epoch": 0.9861074241217591,
      "grad_norm": 0.696030140774074,
      "learning_rate": 0.00010710153344401439,
      "loss": 12.3183,
      "step": 18109
    },
    {
      "epoch": 0.9861618781183421,
      "grad_norm": 0.5880121858085362,
      "learning_rate": 0.00010709273742237776,
      "loss": 12.0449,
      "step": 18110
    },
    {
      "epoch": 0.9862163321149252,
      "grad_norm": 0.49293080395869104,
      "learning_rate": 0.00010708394134558678,
      "loss": 11.9812,
      "step": 18111
    },
    {
      "epoch": 0.9862707861115082,
      "grad_norm": 0.6154783409642931,
      "learning_rate": 0.00010707514521370994,
      "loss": 12.0409,
      "step": 18112
    },
    {
      "epoch": 0.9863252401080912,
      "grad_norm": 0.5396766468657854,
      "learning_rate": 0.00010706634902681551,
      "loss": 12.022,
      "step": 18113
    },
    {
      "epoch": 0.9863796941046742,
      "grad_norm": 0.5503026621440306,
      "learning_rate": 0.000107057552784972,
      "loss": 12.1035,
      "step": 18114
    },
    {
      "epoch": 0.9864341481012572,
      "grad_norm": 0.5962588816072459,
      "learning_rate": 0.00010704875648824777,
      "loss": 11.9612,
      "step": 18115
    },
    {
      "epoch": 0.9864886020978402,
      "grad_norm": 0.631680261499398,
      "learning_rate": 0.00010703996013671124,
      "loss": 12.0747,
      "step": 18116
    },
    {
      "epoch": 0.9865430560944233,
      "grad_norm": 0.542234954428847,
      "learning_rate": 0.00010703116373043082,
      "loss": 12.1221,
      "step": 18117
    },
    {
      "epoch": 0.9865975100910063,
      "grad_norm": 0.6474710621477797,
      "learning_rate": 0.00010702236726947485,
      "loss": 12.2001,
      "step": 18118
    },
    {
      "epoch": 0.9866519640875893,
      "grad_norm": 0.5821040606293187,
      "learning_rate": 0.00010701357075391178,
      "loss": 12.099,
      "step": 18119
    },
    {
      "epoch": 0.9867064180841723,
      "grad_norm": 0.6329945116586946,
      "learning_rate": 0.00010700477418381003,
      "loss": 12.2078,
      "step": 18120
    },
    {
      "epoch": 0.9867608720807552,
      "grad_norm": 0.5319668920455514,
      "learning_rate": 0.00010699597755923796,
      "loss": 12.0948,
      "step": 18121
    },
    {
      "epoch": 0.9868153260773382,
      "grad_norm": 0.5900748869649325,
      "learning_rate": 0.000106987180880264,
      "loss": 12.1813,
      "step": 18122
    },
    {
      "epoch": 0.9868697800739213,
      "grad_norm": 0.6149433502827112,
      "learning_rate": 0.00010697838414695657,
      "loss": 12.2179,
      "step": 18123
    },
    {
      "epoch": 0.9869242340705043,
      "grad_norm": 0.562985381858628,
      "learning_rate": 0.00010696958735938403,
      "loss": 12.1519,
      "step": 18124
    },
    {
      "epoch": 0.9869786880670873,
      "grad_norm": 0.5658279954443907,
      "learning_rate": 0.00010696079051761483,
      "loss": 12.091,
      "step": 18125
    },
    {
      "epoch": 0.9870331420636703,
      "grad_norm": 0.6312905273552373,
      "learning_rate": 0.00010695199362171733,
      "loss": 12.1555,
      "step": 18126
    },
    {
      "epoch": 0.9870875960602533,
      "grad_norm": 0.5886784688030845,
      "learning_rate": 0.00010694319667175998,
      "loss": 12.205,
      "step": 18127
    },
    {
      "epoch": 0.9871420500568363,
      "grad_norm": 0.4950797462419129,
      "learning_rate": 0.00010693439966781118,
      "loss": 11.8538,
      "step": 18128
    },
    {
      "epoch": 0.9871965040534194,
      "grad_norm": 0.5522936025725195,
      "learning_rate": 0.00010692560260993931,
      "loss": 12.1399,
      "step": 18129
    },
    {
      "epoch": 0.9872509580500024,
      "grad_norm": 0.6427225968953367,
      "learning_rate": 0.00010691680549821277,
      "loss": 12.07,
      "step": 18130
    },
    {
      "epoch": 0.9873054120465854,
      "grad_norm": 0.5154353236629954,
      "learning_rate": 0.00010690800833270002,
      "loss": 12.0826,
      "step": 18131
    },
    {
      "epoch": 0.9873598660431684,
      "grad_norm": 0.6143806323295387,
      "learning_rate": 0.00010689921111346943,
      "loss": 12.1745,
      "step": 18132
    },
    {
      "epoch": 0.9874143200397514,
      "grad_norm": 0.6140008287528899,
      "learning_rate": 0.00010689041384058944,
      "loss": 12.3005,
      "step": 18133
    },
    {
      "epoch": 0.9874687740363345,
      "grad_norm": 0.5372926978927672,
      "learning_rate": 0.00010688161651412843,
      "loss": 12.0937,
      "step": 18134
    },
    {
      "epoch": 0.9875232280329175,
      "grad_norm": 0.6016048547083367,
      "learning_rate": 0.00010687281913415477,
      "loss": 12.0777,
      "step": 18135
    },
    {
      "epoch": 0.9875776820295005,
      "grad_norm": 0.5706361887179785,
      "learning_rate": 0.00010686402170073695,
      "loss": 12.2219,
      "step": 18136
    },
    {
      "epoch": 0.9876321360260835,
      "grad_norm": 0.5111482874190844,
      "learning_rate": 0.00010685522421394334,
      "loss": 12.0913,
      "step": 18137
    },
    {
      "epoch": 0.9876865900226665,
      "grad_norm": 0.5407132927064728,
      "learning_rate": 0.00010684642667384239,
      "loss": 12.2046,
      "step": 18138
    },
    {
      "epoch": 0.9877410440192494,
      "grad_norm": 0.5551703354684117,
      "learning_rate": 0.00010683762908050243,
      "loss": 12.0568,
      "step": 18139
    },
    {
      "epoch": 0.9877954980158326,
      "grad_norm": 0.556144476717898,
      "learning_rate": 0.00010682883143399194,
      "loss": 12.1016,
      "step": 18140
    },
    {
      "epoch": 0.9878499520124155,
      "grad_norm": 0.52610775539807,
      "learning_rate": 0.0001068200337343793,
      "loss": 12.0702,
      "step": 18141
    },
    {
      "epoch": 0.9879044060089985,
      "grad_norm": 0.5839966922936672,
      "learning_rate": 0.00010681123598173295,
      "loss": 12.2517,
      "step": 18142
    },
    {
      "epoch": 0.9879588600055815,
      "grad_norm": 0.5289322739960025,
      "learning_rate": 0.0001068024381761213,
      "loss": 12.1082,
      "step": 18143
    },
    {
      "epoch": 0.9880133140021645,
      "grad_norm": 0.5662415515881658,
      "learning_rate": 0.00010679364031761273,
      "loss": 11.9966,
      "step": 18144
    },
    {
      "epoch": 0.9880677679987475,
      "grad_norm": 0.595066729336303,
      "learning_rate": 0.00010678484240627566,
      "loss": 12.0954,
      "step": 18145
    },
    {
      "epoch": 0.9881222219953306,
      "grad_norm": 0.756702540809058,
      "learning_rate": 0.00010677604444217853,
      "loss": 12.0219,
      "step": 18146
    },
    {
      "epoch": 0.9881766759919136,
      "grad_norm": 0.584760628442418,
      "learning_rate": 0.00010676724642538973,
      "loss": 12.1959,
      "step": 18147
    },
    {
      "epoch": 0.9882311299884966,
      "grad_norm": 0.5955713714862642,
      "learning_rate": 0.00010675844835597772,
      "loss": 12.1417,
      "step": 18148
    },
    {
      "epoch": 0.9882855839850796,
      "grad_norm": 0.545147343129215,
      "learning_rate": 0.00010674965023401087,
      "loss": 12.1022,
      "step": 18149
    },
    {
      "epoch": 0.9883400379816626,
      "grad_norm": 0.5710069833773501,
      "learning_rate": 0.00010674085205955759,
      "loss": 12.1044,
      "step": 18150
    },
    {
      "epoch": 0.9883944919782456,
      "grad_norm": 0.5329622577089048,
      "learning_rate": 0.00010673205383268632,
      "loss": 12.0673,
      "step": 18151
    },
    {
      "epoch": 0.9884489459748287,
      "grad_norm": 0.5487041818413038,
      "learning_rate": 0.00010672325555346545,
      "loss": 12.1085,
      "step": 18152
    },
    {
      "epoch": 0.9885033999714117,
      "grad_norm": 0.5415157448620397,
      "learning_rate": 0.00010671445722196346,
      "loss": 12.1328,
      "step": 18153
    },
    {
      "epoch": 0.9885578539679947,
      "grad_norm": 0.5624921937625975,
      "learning_rate": 0.00010670565883824872,
      "loss": 12.1548,
      "step": 18154
    },
    {
      "epoch": 0.9886123079645777,
      "grad_norm": 0.6212541456400029,
      "learning_rate": 0.00010669686040238964,
      "loss": 12.1634,
      "step": 18155
    },
    {
      "epoch": 0.9886667619611607,
      "grad_norm": 0.5194000136149667,
      "learning_rate": 0.00010668806191445466,
      "loss": 11.9366,
      "step": 18156
    },
    {
      "epoch": 0.9887212159577436,
      "grad_norm": 0.5155042191634405,
      "learning_rate": 0.00010667926337451217,
      "loss": 12.1547,
      "step": 18157
    },
    {
      "epoch": 0.9887756699543268,
      "grad_norm": 0.5246555161359897,
      "learning_rate": 0.0001066704647826306,
      "loss": 12.0819,
      "step": 18158
    },
    {
      "epoch": 0.9888301239509097,
      "grad_norm": 0.5078959519576165,
      "learning_rate": 0.0001066616661388784,
      "loss": 12.1861,
      "step": 18159
    },
    {
      "epoch": 0.9888845779474927,
      "grad_norm": 0.6152159808564477,
      "learning_rate": 0.00010665286744332397,
      "loss": 12.2136,
      "step": 18160
    },
    {
      "epoch": 0.9889390319440757,
      "grad_norm": 0.5111939428258314,
      "learning_rate": 0.00010664406869603572,
      "loss": 12.0874,
      "step": 18161
    },
    {
      "epoch": 0.9889934859406587,
      "grad_norm": 0.5119214118859521,
      "learning_rate": 0.00010663526989708209,
      "loss": 12.0325,
      "step": 18162
    },
    {
      "epoch": 0.9890479399372417,
      "grad_norm": 0.5219377048092392,
      "learning_rate": 0.00010662647104653146,
      "loss": 11.8204,
      "step": 18163
    },
    {
      "epoch": 0.9891023939338248,
      "grad_norm": 0.5443243507868772,
      "learning_rate": 0.00010661767214445229,
      "loss": 12.1334,
      "step": 18164
    },
    {
      "epoch": 0.9891568479304078,
      "grad_norm": 0.5892688335860076,
      "learning_rate": 0.000106608873190913,
      "loss": 12.0278,
      "step": 18165
    },
    {
      "epoch": 0.9892113019269908,
      "grad_norm": 0.5884705795429934,
      "learning_rate": 0.00010660007418598199,
      "loss": 12.082,
      "step": 18166
    },
    {
      "epoch": 0.9892657559235738,
      "grad_norm": 0.5458190704317927,
      "learning_rate": 0.00010659127512972771,
      "loss": 12.1603,
      "step": 18167
    },
    {
      "epoch": 0.9893202099201568,
      "grad_norm": 0.6312072324698195,
      "learning_rate": 0.00010658247602221855,
      "loss": 12.1664,
      "step": 18168
    },
    {
      "epoch": 0.9893746639167399,
      "grad_norm": 0.580064155012794,
      "learning_rate": 0.00010657367686352298,
      "loss": 12.0865,
      "step": 18169
    },
    {
      "epoch": 0.9894291179133229,
      "grad_norm": 0.5564038444545667,
      "learning_rate": 0.0001065648776537094,
      "loss": 12.2043,
      "step": 18170
    },
    {
      "epoch": 0.9894835719099059,
      "grad_norm": 0.5270218157556361,
      "learning_rate": 0.0001065560783928462,
      "loss": 12.0985,
      "step": 18171
    },
    {
      "epoch": 0.9895380259064889,
      "grad_norm": 0.5923197475586851,
      "learning_rate": 0.00010654727908100183,
      "loss": 12.0379,
      "step": 18172
    },
    {
      "epoch": 0.9895924799030719,
      "grad_norm": 0.5829247033872216,
      "learning_rate": 0.00010653847971824476,
      "loss": 12.0139,
      "step": 18173
    },
    {
      "epoch": 0.9896469338996549,
      "grad_norm": 0.5516682687822828,
      "learning_rate": 0.00010652968030464334,
      "loss": 12.0776,
      "step": 18174
    },
    {
      "epoch": 0.989701387896238,
      "grad_norm": 0.5639592853444282,
      "learning_rate": 0.00010652088084026606,
      "loss": 12.1019,
      "step": 18175
    },
    {
      "epoch": 0.989755841892821,
      "grad_norm": 0.5384906436819519,
      "learning_rate": 0.00010651208132518129,
      "loss": 11.9862,
      "step": 18176
    },
    {
      "epoch": 0.989810295889404,
      "grad_norm": 0.6070968546473495,
      "learning_rate": 0.0001065032817594575,
      "loss": 12.0275,
      "step": 18177
    },
    {
      "epoch": 0.9898647498859869,
      "grad_norm": 0.5808379207583481,
      "learning_rate": 0.00010649448214316314,
      "loss": 12.1359,
      "step": 18178
    },
    {
      "epoch": 0.9899192038825699,
      "grad_norm": 0.4988094591452025,
      "learning_rate": 0.00010648568247636653,
      "loss": 12.0005,
      "step": 18179
    },
    {
      "epoch": 0.9899736578791529,
      "grad_norm": 0.5838907201723186,
      "learning_rate": 0.00010647688275913616,
      "loss": 12.172,
      "step": 18180
    },
    {
      "epoch": 0.990028111875736,
      "grad_norm": 0.6488156091959372,
      "learning_rate": 0.00010646808299154049,
      "loss": 12.1838,
      "step": 18181
    },
    {
      "epoch": 0.990082565872319,
      "grad_norm": 0.53625159665213,
      "learning_rate": 0.00010645928317364795,
      "loss": 12.1533,
      "step": 18182
    },
    {
      "epoch": 0.990137019868902,
      "grad_norm": 0.5081836936590858,
      "learning_rate": 0.00010645048330552692,
      "loss": 12.0303,
      "step": 18183
    },
    {
      "epoch": 0.990191473865485,
      "grad_norm": 0.5390404382885886,
      "learning_rate": 0.00010644168338724583,
      "loss": 12.1266,
      "step": 18184
    },
    {
      "epoch": 0.990245927862068,
      "grad_norm": 0.5996580270093308,
      "learning_rate": 0.00010643288341887314,
      "loss": 11.9662,
      "step": 18185
    },
    {
      "epoch": 0.990300381858651,
      "grad_norm": 0.4910316126595332,
      "learning_rate": 0.00010642408340047728,
      "loss": 12.1151,
      "step": 18186
    },
    {
      "epoch": 0.9903548358552341,
      "grad_norm": 0.5270594120462088,
      "learning_rate": 0.00010641528333212667,
      "loss": 12.1097,
      "step": 18187
    },
    {
      "epoch": 0.9904092898518171,
      "grad_norm": 0.5193287254087474,
      "learning_rate": 0.00010640648321388976,
      "loss": 12.0473,
      "step": 18188
    },
    {
      "epoch": 0.9904637438484001,
      "grad_norm": 0.5753604760633858,
      "learning_rate": 0.00010639768304583493,
      "loss": 12.1701,
      "step": 18189
    },
    {
      "epoch": 0.9905181978449831,
      "grad_norm": 0.5739489111380263,
      "learning_rate": 0.00010638888282803064,
      "loss": 12.1724,
      "step": 18190
    },
    {
      "epoch": 0.9905726518415661,
      "grad_norm": 0.5255039712114434,
      "learning_rate": 0.00010638008256054532,
      "loss": 11.9917,
      "step": 18191
    },
    {
      "epoch": 0.9906271058381491,
      "grad_norm": 0.5672165055545463,
      "learning_rate": 0.00010637128224344743,
      "loss": 12.1203,
      "step": 18192
    },
    {
      "epoch": 0.9906815598347322,
      "grad_norm": 0.5776361743452018,
      "learning_rate": 0.0001063624818768054,
      "loss": 12.0376,
      "step": 18193
    },
    {
      "epoch": 0.9907360138313152,
      "grad_norm": 0.510166317980021,
      "learning_rate": 0.00010635368146068763,
      "loss": 11.9939,
      "step": 18194
    },
    {
      "epoch": 0.9907904678278981,
      "grad_norm": 0.5041406687352171,
      "learning_rate": 0.00010634488099516253,
      "loss": 11.9797,
      "step": 18195
    },
    {
      "epoch": 0.9908449218244811,
      "grad_norm": 0.5805858677346144,
      "learning_rate": 0.00010633608048029859,
      "loss": 11.9986,
      "step": 18196
    },
    {
      "epoch": 0.9908993758210641,
      "grad_norm": 0.5529385195231652,
      "learning_rate": 0.00010632727991616425,
      "loss": 12.0752,
      "step": 18197
    },
    {
      "epoch": 0.9909538298176471,
      "grad_norm": 0.5922940201271124,
      "learning_rate": 0.00010631847930282792,
      "loss": 12.1256,
      "step": 18198
    },
    {
      "epoch": 0.9910082838142302,
      "grad_norm": 0.5176892169025592,
      "learning_rate": 0.00010630967864035801,
      "loss": 12.1186,
      "step": 18199
    },
    {
      "epoch": 0.9910627378108132,
      "grad_norm": 0.5392799211225574,
      "learning_rate": 0.00010630087792882299,
      "loss": 12.0985,
      "step": 18200
    },
    {
      "epoch": 0.9911171918073962,
      "grad_norm": 0.5741306146407585,
      "learning_rate": 0.00010629207716829128,
      "loss": 12.2897,
      "step": 18201
    },
    {
      "epoch": 0.9911716458039792,
      "grad_norm": 0.6588728596304212,
      "learning_rate": 0.00010628327635883133,
      "loss": 12.1349,
      "step": 18202
    },
    {
      "epoch": 0.9912260998005622,
      "grad_norm": 0.5016770517506026,
      "learning_rate": 0.0001062744755005116,
      "loss": 12.1715,
      "step": 18203
    },
    {
      "epoch": 0.9912805537971453,
      "grad_norm": 0.6018119933793505,
      "learning_rate": 0.00010626567459340047,
      "loss": 12.0232,
      "step": 18204
    },
    {
      "epoch": 0.9913350077937283,
      "grad_norm": 0.5748866193437475,
      "learning_rate": 0.00010625687363756638,
      "loss": 12.0883,
      "step": 18205
    },
    {
      "epoch": 0.9913894617903113,
      "grad_norm": 0.5671183525950302,
      "learning_rate": 0.0001062480726330778,
      "loss": 12.1136,
      "step": 18206
    },
    {
      "epoch": 0.9914439157868943,
      "grad_norm": 0.5512127695258885,
      "learning_rate": 0.00010623927158000318,
      "loss": 12.1586,
      "step": 18207
    },
    {
      "epoch": 0.9914983697834773,
      "grad_norm": 0.5306245357987395,
      "learning_rate": 0.00010623047047841095,
      "loss": 12.0574,
      "step": 18208
    },
    {
      "epoch": 0.9915528237800603,
      "grad_norm": 0.7030287098542315,
      "learning_rate": 0.0001062216693283695,
      "loss": 12.1599,
      "step": 18209
    },
    {
      "epoch": 0.9916072777766434,
      "grad_norm": 0.5550678850782707,
      "learning_rate": 0.00010621286812994733,
      "loss": 12.0858,
      "step": 18210
    },
    {
      "epoch": 0.9916617317732264,
      "grad_norm": 0.614911713623915,
      "learning_rate": 0.00010620406688321285,
      "loss": 12.0705,
      "step": 18211
    },
    {
      "epoch": 0.9917161857698094,
      "grad_norm": 0.5689402836679145,
      "learning_rate": 0.00010619526558823447,
      "loss": 12.0661,
      "step": 18212
    },
    {
      "epoch": 0.9917706397663923,
      "grad_norm": 0.601418914315275,
      "learning_rate": 0.00010618646424508072,
      "loss": 12.0915,
      "step": 18213
    },
    {
      "epoch": 0.9918250937629753,
      "grad_norm": 0.5290614927342719,
      "learning_rate": 0.00010617766285381997,
      "loss": 12.0135,
      "step": 18214
    },
    {
      "epoch": 0.9918795477595583,
      "grad_norm": 0.6662437714409607,
      "learning_rate": 0.00010616886141452066,
      "loss": 12.1376,
      "step": 18215
    },
    {
      "epoch": 0.9919340017561414,
      "grad_norm": 0.5554799207212264,
      "learning_rate": 0.00010616005992725127,
      "loss": 12.0578,
      "step": 18216
    },
    {
      "epoch": 0.9919884557527244,
      "grad_norm": 0.5894407616829986,
      "learning_rate": 0.0001061512583920802,
      "loss": 12.1633,
      "step": 18217
    },
    {
      "epoch": 0.9920429097493074,
      "grad_norm": 0.6647002522545795,
      "learning_rate": 0.00010614245680907593,
      "loss": 12.0707,
      "step": 18218
    },
    {
      "epoch": 0.9920973637458904,
      "grad_norm": 0.5547368028447971,
      "learning_rate": 0.0001061336551783069,
      "loss": 12.1337,
      "step": 18219
    },
    {
      "epoch": 0.9921518177424734,
      "grad_norm": 0.5780326738983894,
      "learning_rate": 0.00010612485349984151,
      "loss": 12.1128,
      "step": 18220
    },
    {
      "epoch": 0.9922062717390564,
      "grad_norm": 0.6561917013664642,
      "learning_rate": 0.00010611605177374824,
      "loss": 12.2249,
      "step": 18221
    },
    {
      "epoch": 0.9922607257356395,
      "grad_norm": 0.5224015433566719,
      "learning_rate": 0.00010610725000009551,
      "loss": 12.0916,
      "step": 18222
    },
    {
      "epoch": 0.9923151797322225,
      "grad_norm": 0.7322683937666953,
      "learning_rate": 0.00010609844817895181,
      "loss": 12.0303,
      "step": 18223
    },
    {
      "epoch": 0.9923696337288055,
      "grad_norm": 0.5827006291050971,
      "learning_rate": 0.00010608964631038553,
      "loss": 12.1149,
      "step": 18224
    },
    {
      "epoch": 0.9924240877253885,
      "grad_norm": 0.6341172939816072,
      "learning_rate": 0.00010608084439446517,
      "loss": 12.2159,
      "step": 18225
    },
    {
      "epoch": 0.9924785417219715,
      "grad_norm": 0.5055789506140991,
      "learning_rate": 0.00010607204243125912,
      "loss": 12.0186,
      "step": 18226
    },
    {
      "epoch": 0.9925329957185545,
      "grad_norm": 0.5937480331497839,
      "learning_rate": 0.00010606324042083586,
      "loss": 12.2173,
      "step": 18227
    },
    {
      "epoch": 0.9925874497151376,
      "grad_norm": 0.5765533212067099,
      "learning_rate": 0.0001060544383632638,
      "loss": 12.1039,
      "step": 18228
    },
    {
      "epoch": 0.9926419037117206,
      "grad_norm": 0.5856380091206774,
      "learning_rate": 0.00010604563625861146,
      "loss": 12.1383,
      "step": 18229
    },
    {
      "epoch": 0.9926963577083036,
      "grad_norm": 0.5355559798841263,
      "learning_rate": 0.00010603683410694721,
      "loss": 12.0303,
      "step": 18230
    },
    {
      "epoch": 0.9927508117048865,
      "grad_norm": 0.5242059372670049,
      "learning_rate": 0.00010602803190833952,
      "loss": 12.0785,
      "step": 18231
    },
    {
      "epoch": 0.9928052657014695,
      "grad_norm": 0.5529125861572977,
      "learning_rate": 0.00010601922966285685,
      "loss": 12.1323,
      "step": 18232
    },
    {
      "epoch": 0.9928597196980525,
      "grad_norm": 0.579570952487471,
      "learning_rate": 0.00010601042737056763,
      "loss": 12.1193,
      "step": 18233
    },
    {
      "epoch": 0.9929141736946356,
      "grad_norm": 0.5615623403966941,
      "learning_rate": 0.00010600162503154034,
      "loss": 12.063,
      "step": 18234
    },
    {
      "epoch": 0.9929686276912186,
      "grad_norm": 0.5263312129752604,
      "learning_rate": 0.00010599282264584338,
      "loss": 12.0795,
      "step": 18235
    },
    {
      "epoch": 0.9930230816878016,
      "grad_norm": 0.5741760977770437,
      "learning_rate": 0.00010598402021354525,
      "loss": 12.1506,
      "step": 18236
    },
    {
      "epoch": 0.9930775356843846,
      "grad_norm": 0.5453528232582568,
      "learning_rate": 0.00010597521773471437,
      "loss": 12.1111,
      "step": 18237
    },
    {
      "epoch": 0.9931319896809676,
      "grad_norm": 0.5848245007509505,
      "learning_rate": 0.00010596641520941921,
      "loss": 12.0653,
      "step": 18238
    },
    {
      "epoch": 0.9931864436775507,
      "grad_norm": 0.5734984559056954,
      "learning_rate": 0.00010595761263772816,
      "loss": 12.0619,
      "step": 18239
    },
    {
      "epoch": 0.9932408976741337,
      "grad_norm": 0.5634072989729753,
      "learning_rate": 0.00010594881001970975,
      "loss": 11.9756,
      "step": 18240
    },
    {
      "epoch": 0.9932953516707167,
      "grad_norm": 0.5894397628042625,
      "learning_rate": 0.00010594000735543239,
      "loss": 11.9566,
      "step": 18241
    },
    {
      "epoch": 0.9933498056672997,
      "grad_norm": 0.5321180965263821,
      "learning_rate": 0.00010593120464496453,
      "loss": 12.0611,
      "step": 18242
    },
    {
      "epoch": 0.9934042596638827,
      "grad_norm": 0.5994931595328526,
      "learning_rate": 0.00010592240188837465,
      "loss": 12.105,
      "step": 18243
    },
    {
      "epoch": 0.9934587136604657,
      "grad_norm": 0.5939063125313075,
      "learning_rate": 0.00010591359908573115,
      "loss": 12.0165,
      "step": 18244
    },
    {
      "epoch": 0.9935131676570488,
      "grad_norm": 0.5686234838614362,
      "learning_rate": 0.00010590479623710252,
      "loss": 12.0928,
      "step": 18245
    },
    {
      "epoch": 0.9935676216536318,
      "grad_norm": 0.5747729490831621,
      "learning_rate": 0.00010589599334255722,
      "loss": 12.141,
      "step": 18246
    },
    {
      "epoch": 0.9936220756502148,
      "grad_norm": 0.5933524183797961,
      "learning_rate": 0.00010588719040216366,
      "loss": 12.0861,
      "step": 18247
    },
    {
      "epoch": 0.9936765296467978,
      "grad_norm": 0.5578922147546961,
      "learning_rate": 0.00010587838741599037,
      "loss": 12.107,
      "step": 18248
    },
    {
      "epoch": 0.9937309836433807,
      "grad_norm": 0.5309388163475087,
      "learning_rate": 0.00010586958438410572,
      "loss": 12.052,
      "step": 18249
    },
    {
      "epoch": 0.9937854376399637,
      "grad_norm": 0.5308965616404674,
      "learning_rate": 0.00010586078130657817,
      "loss": 12.0533,
      "step": 18250
    },
    {
      "epoch": 0.9938398916365468,
      "grad_norm": 0.5162989002287177,
      "learning_rate": 0.00010585197818347624,
      "loss": 11.9956,
      "step": 18251
    },
    {
      "epoch": 0.9938943456331298,
      "grad_norm": 0.5915427121879998,
      "learning_rate": 0.00010584317501486833,
      "loss": 12.0785,
      "step": 18252
    },
    {
      "epoch": 0.9939487996297128,
      "grad_norm": 0.5787166950197778,
      "learning_rate": 0.00010583437180082293,
      "loss": 12.0896,
      "step": 18253
    },
    {
      "epoch": 0.9940032536262958,
      "grad_norm": 0.5140441644242902,
      "learning_rate": 0.00010582556854140846,
      "loss": 12.0014,
      "step": 18254
    },
    {
      "epoch": 0.9940577076228788,
      "grad_norm": 0.5961023277506549,
      "learning_rate": 0.00010581676523669339,
      "loss": 12.1662,
      "step": 18255
    },
    {
      "epoch": 0.9941121616194618,
      "grad_norm": 0.5563828940257974,
      "learning_rate": 0.00010580796188674618,
      "loss": 12.09,
      "step": 18256
    },
    {
      "epoch": 0.9941666156160449,
      "grad_norm": 0.527189994076118,
      "learning_rate": 0.00010579915849163528,
      "loss": 11.9866,
      "step": 18257
    },
    {
      "epoch": 0.9942210696126279,
      "grad_norm": 0.6035817247554295,
      "learning_rate": 0.00010579035505142918,
      "loss": 12.2226,
      "step": 18258
    },
    {
      "epoch": 0.9942755236092109,
      "grad_norm": 0.5345851780364875,
      "learning_rate": 0.00010578155156619629,
      "loss": 12.1269,
      "step": 18259
    },
    {
      "epoch": 0.9943299776057939,
      "grad_norm": 0.7405022779870131,
      "learning_rate": 0.00010577274803600508,
      "loss": 12.0807,
      "step": 18260
    },
    {
      "epoch": 0.9943844316023769,
      "grad_norm": 0.5171137351541494,
      "learning_rate": 0.00010576394446092399,
      "loss": 12.0452,
      "step": 18261
    },
    {
      "epoch": 0.9944388855989599,
      "grad_norm": 0.5749189436715378,
      "learning_rate": 0.00010575514084102151,
      "loss": 12.1096,
      "step": 18262
    },
    {
      "epoch": 0.994493339595543,
      "grad_norm": 0.7449553375505649,
      "learning_rate": 0.00010574633717636615,
      "loss": 12.1099,
      "step": 18263
    },
    {
      "epoch": 0.994547793592126,
      "grad_norm": 0.6323078374538844,
      "learning_rate": 0.00010573753346702626,
      "loss": 12.2157,
      "step": 18264
    },
    {
      "epoch": 0.994602247588709,
      "grad_norm": 0.5005928734970015,
      "learning_rate": 0.00010572872971307035,
      "loss": 11.8993,
      "step": 18265
    },
    {
      "epoch": 0.994656701585292,
      "grad_norm": 0.5487925587482815,
      "learning_rate": 0.00010571992591456687,
      "loss": 12.1289,
      "step": 18266
    },
    {
      "epoch": 0.994711155581875,
      "grad_norm": 0.5645526783870632,
      "learning_rate": 0.00010571112207158429,
      "loss": 12.1062,
      "step": 18267
    },
    {
      "epoch": 0.994765609578458,
      "grad_norm": 0.5660600084508958,
      "learning_rate": 0.00010570231818419106,
      "loss": 12.0204,
      "step": 18268
    },
    {
      "epoch": 0.994820063575041,
      "grad_norm": 0.5105849321119167,
      "learning_rate": 0.00010569351425245569,
      "loss": 12.1643,
      "step": 18269
    },
    {
      "epoch": 0.994874517571624,
      "grad_norm": 0.5828619360755012,
      "learning_rate": 0.00010568471027644656,
      "loss": 12.1449,
      "step": 18270
    },
    {
      "epoch": 0.994928971568207,
      "grad_norm": 0.5563200390594023,
      "learning_rate": 0.00010567590625623219,
      "loss": 12.0673,
      "step": 18271
    },
    {
      "epoch": 0.99498342556479,
      "grad_norm": 0.5429151672763892,
      "learning_rate": 0.00010566710219188098,
      "loss": 12.1319,
      "step": 18272
    },
    {
      "epoch": 0.995037879561373,
      "grad_norm": 0.5584770559140745,
      "learning_rate": 0.00010565829808346146,
      "loss": 12.1062,
      "step": 18273
    },
    {
      "epoch": 0.9950923335579561,
      "grad_norm": 0.6520903283099112,
      "learning_rate": 0.00010564949393104208,
      "loss": 12.1104,
      "step": 18274
    },
    {
      "epoch": 0.9951467875545391,
      "grad_norm": 0.5716805206843671,
      "learning_rate": 0.00010564068973469128,
      "loss": 12.1117,
      "step": 18275
    },
    {
      "epoch": 0.9952012415511221,
      "grad_norm": 1.2849329251298771,
      "learning_rate": 0.00010563188549447754,
      "loss": 12.2314,
      "step": 18276
    },
    {
      "epoch": 0.9952556955477051,
      "grad_norm": 0.571927857807694,
      "learning_rate": 0.00010562308121046929,
      "loss": 12.098,
      "step": 18277
    },
    {
      "epoch": 0.9953101495442881,
      "grad_norm": 0.6030794560881759,
      "learning_rate": 0.000105614276882735,
      "loss": 11.971,
      "step": 18278
    },
    {
      "epoch": 0.9953646035408711,
      "grad_norm": 0.5171068645879546,
      "learning_rate": 0.0001056054725113432,
      "loss": 12.124,
      "step": 18279
    },
    {
      "epoch": 0.9954190575374542,
      "grad_norm": 0.5842618670819876,
      "learning_rate": 0.00010559666809636229,
      "loss": 12.0035,
      "step": 18280
    },
    {
      "epoch": 0.9954735115340372,
      "grad_norm": 0.5434906527951233,
      "learning_rate": 0.00010558786363786075,
      "loss": 12.1142,
      "step": 18281
    },
    {
      "epoch": 0.9955279655306202,
      "grad_norm": 0.49536206412147676,
      "learning_rate": 0.00010557905913590704,
      "loss": 12.0497,
      "step": 18282
    },
    {
      "epoch": 0.9955824195272032,
      "grad_norm": 0.585149941217784,
      "learning_rate": 0.00010557025459056962,
      "loss": 12.1341,
      "step": 18283
    },
    {
      "epoch": 0.9956368735237862,
      "grad_norm": 0.631751785017183,
      "learning_rate": 0.00010556145000191697,
      "loss": 12.0925,
      "step": 18284
    },
    {
      "epoch": 0.9956913275203692,
      "grad_norm": 0.5918976363038114,
      "learning_rate": 0.00010555264537001757,
      "loss": 12.293,
      "step": 18285
    },
    {
      "epoch": 0.9957457815169523,
      "grad_norm": 0.6808319682134729,
      "learning_rate": 0.00010554384069493985,
      "loss": 12.1553,
      "step": 18286
    },
    {
      "epoch": 0.9958002355135352,
      "grad_norm": 0.5531333942652283,
      "learning_rate": 0.00010553503597675231,
      "loss": 12.0735,
      "step": 18287
    },
    {
      "epoch": 0.9958546895101182,
      "grad_norm": 0.5628731217928776,
      "learning_rate": 0.00010552623121552339,
      "loss": 12.106,
      "step": 18288
    },
    {
      "epoch": 0.9959091435067012,
      "grad_norm": 0.5609251496824058,
      "learning_rate": 0.00010551742641132159,
      "loss": 12.0956,
      "step": 18289
    },
    {
      "epoch": 0.9959635975032842,
      "grad_norm": 0.568234454533744,
      "learning_rate": 0.00010550862156421532,
      "loss": 12.0748,
      "step": 18290
    },
    {
      "epoch": 0.9960180514998672,
      "grad_norm": 0.580822952517534,
      "learning_rate": 0.00010549981667427312,
      "loss": 12.0833,
      "step": 18291
    },
    {
      "epoch": 0.9960725054964503,
      "grad_norm": 0.5632507217341308,
      "learning_rate": 0.0001054910117415634,
      "loss": 12.1432,
      "step": 18292
    },
    {
      "epoch": 0.9961269594930333,
      "grad_norm": 0.6988838802306254,
      "learning_rate": 0.00010548220676615466,
      "loss": 12.3317,
      "step": 18293
    },
    {
      "epoch": 0.9961814134896163,
      "grad_norm": 0.6437110816396651,
      "learning_rate": 0.00010547340174811538,
      "loss": 11.9467,
      "step": 18294
    },
    {
      "epoch": 0.9962358674861993,
      "grad_norm": 0.5156467182743514,
      "learning_rate": 0.00010546459668751398,
      "loss": 11.8644,
      "step": 18295
    },
    {
      "epoch": 0.9962903214827823,
      "grad_norm": 0.5651016554038113,
      "learning_rate": 0.00010545579158441896,
      "loss": 12.0975,
      "step": 18296
    },
    {
      "epoch": 0.9963447754793653,
      "grad_norm": 0.547738898561608,
      "learning_rate": 0.0001054469864388988,
      "loss": 12.0015,
      "step": 18297
    },
    {
      "epoch": 0.9963992294759484,
      "grad_norm": 0.5631191104540555,
      "learning_rate": 0.00010543818125102197,
      "loss": 12.1064,
      "step": 18298
    },
    {
      "epoch": 0.9964536834725314,
      "grad_norm": 0.5402298000145086,
      "learning_rate": 0.00010542937602085692,
      "loss": 12.0611,
      "step": 18299
    },
    {
      "epoch": 0.9965081374691144,
      "grad_norm": 0.54779694513099,
      "learning_rate": 0.00010542057074847214,
      "loss": 12.1614,
      "step": 18300
    },
    {
      "epoch": 0.9965625914656974,
      "grad_norm": 0.5866882211911401,
      "learning_rate": 0.00010541176543393607,
      "loss": 12.1319,
      "step": 18301
    },
    {
      "epoch": 0.9966170454622804,
      "grad_norm": 0.5580272799539157,
      "learning_rate": 0.00010540296007731723,
      "loss": 12.1469,
      "step": 18302
    },
    {
      "epoch": 0.9966714994588635,
      "grad_norm": 0.604361212709162,
      "learning_rate": 0.00010539415467868406,
      "loss": 12.1564,
      "step": 18303
    },
    {
      "epoch": 0.9967259534554465,
      "grad_norm": 0.569177199760398,
      "learning_rate": 0.00010538534923810506,
      "loss": 12.1793,
      "step": 18304
    },
    {
      "epoch": 0.9967804074520294,
      "grad_norm": 0.5338852867265473,
      "learning_rate": 0.00010537654375564862,
      "loss": 12.0523,
      "step": 18305
    },
    {
      "epoch": 0.9968348614486124,
      "grad_norm": 0.5389476611207479,
      "learning_rate": 0.00010536773823138333,
      "loss": 12.1469,
      "step": 18306
    },
    {
      "epoch": 0.9968893154451954,
      "grad_norm": 0.5520982290746783,
      "learning_rate": 0.00010535893266537758,
      "loss": 12.0407,
      "step": 18307
    },
    {
      "epoch": 0.9969437694417784,
      "grad_norm": 0.5762786107442041,
      "learning_rate": 0.00010535012705769989,
      "loss": 11.9832,
      "step": 18308
    },
    {
      "epoch": 0.9969982234383615,
      "grad_norm": 0.5447738058091682,
      "learning_rate": 0.00010534132140841873,
      "loss": 12.2752,
      "step": 18309
    },
    {
      "epoch": 0.9970526774349445,
      "grad_norm": 0.48809500004019324,
      "learning_rate": 0.0001053325157176025,
      "loss": 12.0917,
      "step": 18310
    },
    {
      "epoch": 0.9971071314315275,
      "grad_norm": 0.5557349364671941,
      "learning_rate": 0.00010532370998531976,
      "loss": 12.106,
      "step": 18311
    },
    {
      "epoch": 0.9971615854281105,
      "grad_norm": 0.5191673464952495,
      "learning_rate": 0.00010531490421163897,
      "loss": 12.1023,
      "step": 18312
    },
    {
      "epoch": 0.9972160394246935,
      "grad_norm": 0.5318529763885843,
      "learning_rate": 0.00010530609839662857,
      "loss": 11.9875,
      "step": 18313
    },
    {
      "epoch": 0.9972704934212765,
      "grad_norm": 0.5989152240444041,
      "learning_rate": 0.00010529729254035712,
      "loss": 12.1512,
      "step": 18314
    },
    {
      "epoch": 0.9973249474178596,
      "grad_norm": 0.5312593750250028,
      "learning_rate": 0.00010528848664289299,
      "loss": 12.023,
      "step": 18315
    },
    {
      "epoch": 0.9973794014144426,
      "grad_norm": 0.6016810621628746,
      "learning_rate": 0.00010527968070430467,
      "loss": 12.1681,
      "step": 18316
    },
    {
      "epoch": 0.9974338554110256,
      "grad_norm": 0.5905204779838882,
      "learning_rate": 0.0001052708747246607,
      "loss": 11.9806,
      "step": 18317
    },
    {
      "epoch": 0.9974883094076086,
      "grad_norm": 0.5281817931669279,
      "learning_rate": 0.00010526206870402952,
      "loss": 12.0172,
      "step": 18318
    },
    {
      "epoch": 0.9975427634041916,
      "grad_norm": 0.5710793478121307,
      "learning_rate": 0.00010525326264247965,
      "loss": 12.0452,
      "step": 18319
    },
    {
      "epoch": 0.9975972174007746,
      "grad_norm": 0.5690343274241164,
      "learning_rate": 0.00010524445654007949,
      "loss": 12.1181,
      "step": 18320
    },
    {
      "epoch": 0.9976516713973577,
      "grad_norm": 0.6045292433790345,
      "learning_rate": 0.00010523565039689753,
      "loss": 12.1459,
      "step": 18321
    },
    {
      "epoch": 0.9977061253939407,
      "grad_norm": 0.5120165329316765,
      "learning_rate": 0.0001052268442130023,
      "loss": 12.0556,
      "step": 18322
    },
    {
      "epoch": 0.9977605793905236,
      "grad_norm": 0.5256668955838202,
      "learning_rate": 0.00010521803798846228,
      "loss": 12.1167,
      "step": 18323
    },
    {
      "epoch": 0.9978150333871066,
      "grad_norm": 0.5949637649102439,
      "learning_rate": 0.00010520923172334592,
      "loss": 12.2264,
      "step": 18324
    },
    {
      "epoch": 0.9978694873836896,
      "grad_norm": 0.49523772762042173,
      "learning_rate": 0.00010520042541772168,
      "loss": 12.122,
      "step": 18325
    },
    {
      "epoch": 0.9979239413802726,
      "grad_norm": 0.5778216801940866,
      "learning_rate": 0.00010519161907165806,
      "loss": 12.1547,
      "step": 18326
    },
    {
      "epoch": 0.9979783953768557,
      "grad_norm": 0.5770705735686229,
      "learning_rate": 0.00010518281268522352,
      "loss": 12.0834,
      "step": 18327
    },
    {
      "epoch": 0.9980328493734387,
      "grad_norm": 0.5596684489682436,
      "learning_rate": 0.00010517400625848657,
      "loss": 12.2114,
      "step": 18328
    },
    {
      "epoch": 0.9980873033700217,
      "grad_norm": 0.5670008931842118,
      "learning_rate": 0.00010516519979151572,
      "loss": 12.1024,
      "step": 18329
    },
    {
      "epoch": 0.9981417573666047,
      "grad_norm": 0.5357328828073912,
      "learning_rate": 0.00010515639328437938,
      "loss": 12.0188,
      "step": 18330
    },
    {
      "epoch": 0.9981962113631877,
      "grad_norm": 0.515364442144104,
      "learning_rate": 0.00010514758673714607,
      "loss": 12.1352,
      "step": 18331
    },
    {
      "epoch": 0.9982506653597707,
      "grad_norm": 0.5324150196422278,
      "learning_rate": 0.00010513878014988422,
      "loss": 12.0376,
      "step": 18332
    },
    {
      "epoch": 0.9983051193563538,
      "grad_norm": 0.7198694397915397,
      "learning_rate": 0.00010512997352266239,
      "loss": 12.2379,
      "step": 18333
    },
    {
      "epoch": 0.9983595733529368,
      "grad_norm": 0.7596864730164228,
      "learning_rate": 0.00010512116685554904,
      "loss": 12.2487,
      "step": 18334
    },
    {
      "epoch": 0.9984140273495198,
      "grad_norm": 0.5273094451393162,
      "learning_rate": 0.00010511236014861261,
      "loss": 12.0907,
      "step": 18335
    },
    {
      "epoch": 0.9984684813461028,
      "grad_norm": 0.6108561774057857,
      "learning_rate": 0.00010510355340192162,
      "loss": 12.0449,
      "step": 18336
    },
    {
      "epoch": 0.9985229353426858,
      "grad_norm": 0.5945808666942504,
      "learning_rate": 0.00010509474661554454,
      "loss": 12.0426,
      "step": 18337
    },
    {
      "epoch": 0.9985773893392689,
      "grad_norm": 0.6075217557210957,
      "learning_rate": 0.00010508593978954984,
      "loss": 12.0566,
      "step": 18338
    },
    {
      "epoch": 0.9986318433358519,
      "grad_norm": 0.5271617568745623,
      "learning_rate": 0.00010507713292400607,
      "loss": 12.1659,
      "step": 18339
    },
    {
      "epoch": 0.9986862973324349,
      "grad_norm": 0.6434501091325255,
      "learning_rate": 0.00010506832601898162,
      "loss": 12.1409,
      "step": 18340
    },
    {
      "epoch": 0.9987407513290179,
      "grad_norm": 0.601408971484078,
      "learning_rate": 0.00010505951907454504,
      "loss": 12.2454,
      "step": 18341
    },
    {
      "epoch": 0.9987952053256008,
      "grad_norm": 0.5887687221593112,
      "learning_rate": 0.00010505071209076478,
      "loss": 12.1978,
      "step": 18342
    },
    {
      "epoch": 0.9988496593221838,
      "grad_norm": 0.5492925724499284,
      "learning_rate": 0.00010504190506770932,
      "loss": 12.0751,
      "step": 18343
    },
    {
      "epoch": 0.9989041133187669,
      "grad_norm": 0.5594962061512018,
      "learning_rate": 0.00010503309800544718,
      "loss": 12.0993,
      "step": 18344
    },
    {
      "epoch": 0.9989585673153499,
      "grad_norm": 0.537376461163246,
      "learning_rate": 0.00010502429090404681,
      "loss": 12.1009,
      "step": 18345
    },
    {
      "epoch": 0.9990130213119329,
      "grad_norm": 0.5483466664034863,
      "learning_rate": 0.00010501548376357673,
      "loss": 12.0368,
      "step": 18346
    },
    {
      "epoch": 0.9990674753085159,
      "grad_norm": 0.5815370462057136,
      "learning_rate": 0.00010500667658410541,
      "loss": 12.2623,
      "step": 18347
    },
    {
      "epoch": 0.9991219293050989,
      "grad_norm": 0.5212947122431155,
      "learning_rate": 0.00010499786936570133,
      "loss": 12.0358,
      "step": 18348
    },
    {
      "epoch": 0.9991763833016819,
      "grad_norm": 0.543887849801361,
      "learning_rate": 0.00010498906210843296,
      "loss": 12.0294,
      "step": 18349
    },
    {
      "epoch": 0.999230837298265,
      "grad_norm": 0.5967475053340418,
      "learning_rate": 0.00010498025481236881,
      "loss": 12.0394,
      "step": 18350
    },
    {
      "epoch": 0.999285291294848,
      "grad_norm": 0.6079438630113693,
      "learning_rate": 0.0001049714474775774,
      "loss": 12.112,
      "step": 18351
    },
    {
      "epoch": 0.999339745291431,
      "grad_norm": 0.5832927223163846,
      "learning_rate": 0.00010496264010412714,
      "loss": 12.1601,
      "step": 18352
    },
    {
      "epoch": 0.999394199288014,
      "grad_norm": 0.5199728927111259,
      "learning_rate": 0.00010495383269208656,
      "loss": 12.1181,
      "step": 18353
    },
    {
      "epoch": 0.999448653284597,
      "grad_norm": 0.5941050390356781,
      "learning_rate": 0.00010494502524152417,
      "loss": 12.2261,
      "step": 18354
    },
    {
      "epoch": 0.99950310728118,
      "grad_norm": 0.6533739538334977,
      "learning_rate": 0.00010493621775250842,
      "loss": 12.1423,
      "step": 18355
    },
    {
      "epoch": 0.9995575612777631,
      "grad_norm": 0.6397750626797618,
      "learning_rate": 0.00010492741022510781,
      "loss": 11.9695,
      "step": 18356
    },
    {
      "epoch": 0.9996120152743461,
      "grad_norm": 0.615535482329696,
      "learning_rate": 0.00010491860265939084,
      "loss": 11.9663,
      "step": 18357
    },
    {
      "epoch": 0.9996664692709291,
      "grad_norm": 0.5402599506044298,
      "learning_rate": 0.000104909795055426,
      "loss": 11.8675,
      "step": 18358
    },
    {
      "epoch": 0.999720923267512,
      "grad_norm": 0.5627393779909873,
      "learning_rate": 0.00010490098741328174,
      "loss": 12.0418,
      "step": 18359
    },
    {
      "epoch": 0.999775377264095,
      "grad_norm": 0.5455322046325075,
      "learning_rate": 0.00010489217973302661,
      "loss": 12.0932,
      "step": 18360
    },
    {
      "epoch": 0.999829831260678,
      "grad_norm": 0.5342616361648647,
      "learning_rate": 0.00010488337201472905,
      "loss": 11.9299,
      "step": 18361
    },
    {
      "epoch": 0.9998842852572611,
      "grad_norm": 0.5799942188983298,
      "learning_rate": 0.00010487456425845758,
      "loss": 11.969,
      "step": 18362
    },
    {
      "epoch": 0.9999387392538441,
      "grad_norm": 0.583225236752225,
      "learning_rate": 0.00010486575646428067,
      "loss": 12.1362,
      "step": 18363
    },
    {
      "epoch": 0.9999931932504271,
      "grad_norm": 0.5327575410878634,
      "learning_rate": 0.00010485694863226687,
      "loss": 12.0343,
      "step": 18364
    },
    {
      "epoch": 1.0000476472470101,
      "grad_norm": 0.6505595880771092,
      "learning_rate": 0.00010484814076248455,
      "loss": 12.085,
      "step": 18365
    },
    {
      "epoch": 1.0001021012435931,
      "grad_norm": 0.5183267779782971,
      "learning_rate": 0.00010483933285500231,
      "loss": 11.9666,
      "step": 18366
    },
    {
      "epoch": 1.000156555240176,
      "grad_norm": 0.6106950132925038,
      "learning_rate": 0.00010483052490988861,
      "loss": 11.988,
      "step": 18367
    },
    {
      "epoch": 1.000211009236759,
      "grad_norm": 0.5487647852412104,
      "learning_rate": 0.00010482171692721192,
      "loss": 12.043,
      "step": 18368
    },
    {
      "epoch": 1.000265463233342,
      "grad_norm": 0.547568729580315,
      "learning_rate": 0.00010481290890704079,
      "loss": 12.086,
      "step": 18369
    },
    {
      "epoch": 1.0003199172299253,
      "grad_norm": 0.5356534198690128,
      "learning_rate": 0.00010480410084944363,
      "loss": 12.1115,
      "step": 18370
    },
    {
      "epoch": 1.0003743712265083,
      "grad_norm": 0.56201989986331,
      "learning_rate": 0.00010479529275448899,
      "loss": 11.7968,
      "step": 18371
    },
    {
      "epoch": 1.0004288252230913,
      "grad_norm": 0.5862078200930146,
      "learning_rate": 0.00010478648462224536,
      "loss": 12.098,
      "step": 18372
    },
    {
      "epoch": 1.0004832792196743,
      "grad_norm": 0.5426265159006913,
      "learning_rate": 0.0001047776764527812,
      "loss": 12.15,
      "step": 18373
    },
    {
      "epoch": 1.0005377332162573,
      "grad_norm": 0.6361834878239455,
      "learning_rate": 0.00010476886824616508,
      "loss": 12.1512,
      "step": 18374
    },
    {
      "epoch": 1.0005921872128403,
      "grad_norm": 0.643370746183964,
      "learning_rate": 0.00010476006000246539,
      "loss": 12.2027,
      "step": 18375
    },
    {
      "epoch": 1.0006466412094233,
      "grad_norm": 0.5831748434881041,
      "learning_rate": 0.00010475125172175066,
      "loss": 12.0607,
      "step": 18376
    },
    {
      "epoch": 1.0007010952060063,
      "grad_norm": 0.5546440422407036,
      "learning_rate": 0.00010474244340408943,
      "loss": 12.0811,
      "step": 18377
    },
    {
      "epoch": 1.0007555492025892,
      "grad_norm": 0.5409337616191884,
      "learning_rate": 0.00010473363504955017,
      "loss": 11.9853,
      "step": 18378
    },
    {
      "epoch": 1.0008100031991722,
      "grad_norm": 0.5619493544416494,
      "learning_rate": 0.00010472482665820139,
      "loss": 12.109,
      "step": 18379
    },
    {
      "epoch": 1.0008644571957552,
      "grad_norm": 0.4948552519989432,
      "learning_rate": 0.00010471601823011152,
      "loss": 12.121,
      "step": 18380
    },
    {
      "epoch": 1.0009189111923382,
      "grad_norm": 0.6170624624020086,
      "learning_rate": 0.00010470720976534913,
      "loss": 12.2022,
      "step": 18381
    },
    {
      "epoch": 1.0009733651889214,
      "grad_norm": 0.5417170613625275,
      "learning_rate": 0.00010469840126398265,
      "loss": 12.1306,
      "step": 18382
    },
    {
      "epoch": 1.0010278191855044,
      "grad_norm": 0.4905086632246415,
      "learning_rate": 0.00010468959272608063,
      "loss": 12.0468,
      "step": 18383
    },
    {
      "epoch": 1.0010822731820874,
      "grad_norm": 0.6156457670311729,
      "learning_rate": 0.00010468078415171159,
      "loss": 12.1791,
      "step": 18384
    },
    {
      "epoch": 1.0011367271786704,
      "grad_norm": 0.5629794884599503,
      "learning_rate": 0.00010467197554094397,
      "loss": 12.2719,
      "step": 18385
    },
    {
      "epoch": 1.0011911811752534,
      "grad_norm": 0.5730873010065706,
      "learning_rate": 0.00010466316689384625,
      "loss": 12.0505,
      "step": 18386
    },
    {
      "epoch": 1.0012456351718364,
      "grad_norm": 0.5959155320820246,
      "learning_rate": 0.00010465435821048697,
      "loss": 12.1176,
      "step": 18387
    },
    {
      "epoch": 1.0013000891684194,
      "grad_norm": 0.5252390152725271,
      "learning_rate": 0.00010464554949093463,
      "loss": 12.0478,
      "step": 18388
    },
    {
      "epoch": 1.0013545431650024,
      "grad_norm": 0.5594494652330558,
      "learning_rate": 0.00010463674073525775,
      "loss": 12.0372,
      "step": 18389
    },
    {
      "epoch": 1.0014089971615854,
      "grad_norm": 0.5531813407378738,
      "learning_rate": 0.00010462793194352478,
      "loss": 12.0106,
      "step": 18390
    },
    {
      "epoch": 1.0014634511581684,
      "grad_norm": 0.6363738001080059,
      "learning_rate": 0.00010461912311580422,
      "loss": 12.1426,
      "step": 18391
    },
    {
      "epoch": 1.0015179051547514,
      "grad_norm": 0.5505332682820571,
      "learning_rate": 0.00010461031425216459,
      "loss": 12.1038,
      "step": 18392
    },
    {
      "epoch": 1.0015723591513344,
      "grad_norm": 0.5347841437757687,
      "learning_rate": 0.00010460150535267436,
      "loss": 11.9116,
      "step": 18393
    },
    {
      "epoch": 1.0016268131479176,
      "grad_norm": 0.6161254610542731,
      "learning_rate": 0.0001045926964174021,
      "loss": 12.1878,
      "step": 18394
    },
    {
      "epoch": 1.0016812671445006,
      "grad_norm": 0.5473230783030792,
      "learning_rate": 0.00010458388744641622,
      "loss": 11.9945,
      "step": 18395
    },
    {
      "epoch": 1.0017357211410836,
      "grad_norm": 0.5361957484183014,
      "learning_rate": 0.00010457507843978529,
      "loss": 12.0654,
      "step": 18396
    },
    {
      "epoch": 1.0017901751376665,
      "grad_norm": 0.564843165650648,
      "learning_rate": 0.00010456626939757779,
      "loss": 12.0611,
      "step": 18397
    },
    {
      "epoch": 1.0018446291342495,
      "grad_norm": 0.49246724425305344,
      "learning_rate": 0.00010455746031986215,
      "loss": 12.0085,
      "step": 18398
    },
    {
      "epoch": 1.0018990831308325,
      "grad_norm": 0.517952796975952,
      "learning_rate": 0.00010454865120670701,
      "loss": 11.9902,
      "step": 18399
    },
    {
      "epoch": 1.0019535371274155,
      "grad_norm": 0.5715093319395536,
      "learning_rate": 0.00010453984205818078,
      "loss": 12.0314,
      "step": 18400
    },
    {
      "epoch": 1.0020079911239985,
      "grad_norm": 0.6381694403156342,
      "learning_rate": 0.00010453103287435196,
      "loss": 12.1184,
      "step": 18401
    },
    {
      "epoch": 1.0020624451205815,
      "grad_norm": 0.5189112439951066,
      "learning_rate": 0.00010452222365528906,
      "loss": 12.1006,
      "step": 18402
    },
    {
      "epoch": 1.0021168991171645,
      "grad_norm": 0.5427153553870643,
      "learning_rate": 0.00010451341440106059,
      "loss": 12.1646,
      "step": 18403
    },
    {
      "epoch": 1.0021713531137475,
      "grad_norm": 0.5508173660075053,
      "learning_rate": 0.00010450460511173508,
      "loss": 11.9587,
      "step": 18404
    },
    {
      "epoch": 1.0022258071103307,
      "grad_norm": 0.5125258189425049,
      "learning_rate": 0.00010449579578738097,
      "loss": 12.0517,
      "step": 18405
    },
    {
      "epoch": 1.0022802611069137,
      "grad_norm": 0.5640314476131568,
      "learning_rate": 0.00010448698642806682,
      "loss": 11.9638,
      "step": 18406
    },
    {
      "epoch": 1.0023347151034967,
      "grad_norm": 0.5247936949494446,
      "learning_rate": 0.0001044781770338611,
      "loss": 12.1338,
      "step": 18407
    },
    {
      "epoch": 1.0023891691000797,
      "grad_norm": 0.5699248662268073,
      "learning_rate": 0.00010446936760483235,
      "loss": 12.0554,
      "step": 18408
    },
    {
      "epoch": 1.0024436230966627,
      "grad_norm": 0.5832846353753555,
      "learning_rate": 0.00010446055814104903,
      "loss": 12.1558,
      "step": 18409
    },
    {
      "epoch": 1.0024980770932457,
      "grad_norm": 0.5888257898048711,
      "learning_rate": 0.00010445174864257967,
      "loss": 11.9725,
      "step": 18410
    },
    {
      "epoch": 1.0025525310898287,
      "grad_norm": 0.5342655938104394,
      "learning_rate": 0.00010444293910949277,
      "loss": 12.1343,
      "step": 18411
    },
    {
      "epoch": 1.0026069850864117,
      "grad_norm": 0.5222292673682406,
      "learning_rate": 0.00010443412954185681,
      "loss": 12.01,
      "step": 18412
    },
    {
      "epoch": 1.0026614390829947,
      "grad_norm": 0.5605641546372595,
      "learning_rate": 0.00010442531993974031,
      "loss": 12.1621,
      "step": 18413
    },
    {
      "epoch": 1.0027158930795776,
      "grad_norm": 0.5506957085219449,
      "learning_rate": 0.0001044165103032118,
      "loss": 12.017,
      "step": 18414
    },
    {
      "epoch": 1.0027703470761606,
      "grad_norm": 0.5362068520464105,
      "learning_rate": 0.00010440770063233976,
      "loss": 12.0291,
      "step": 18415
    },
    {
      "epoch": 1.0028248010727436,
      "grad_norm": 0.6547642284338772,
      "learning_rate": 0.00010439889092719271,
      "loss": 12.3063,
      "step": 18416
    },
    {
      "epoch": 1.0028792550693268,
      "grad_norm": 0.5953052514900333,
      "learning_rate": 0.00010439008118783913,
      "loss": 12.0355,
      "step": 18417
    },
    {
      "epoch": 1.0029337090659098,
      "grad_norm": 0.5622270585981601,
      "learning_rate": 0.00010438127141434754,
      "loss": 12.1138,
      "step": 18418
    },
    {
      "epoch": 1.0029881630624928,
      "grad_norm": 0.6166438221046772,
      "learning_rate": 0.00010437246160678647,
      "loss": 12.173,
      "step": 18419
    },
    {
      "epoch": 1.0030426170590758,
      "grad_norm": 0.6443790331998598,
      "learning_rate": 0.00010436365176522438,
      "loss": 12.1626,
      "step": 18420
    },
    {
      "epoch": 1.0030970710556588,
      "grad_norm": 0.5064232929602658,
      "learning_rate": 0.00010435484188972982,
      "loss": 12.036,
      "step": 18421
    },
    {
      "epoch": 1.0031515250522418,
      "grad_norm": 0.6420988468737352,
      "learning_rate": 0.00010434603198037127,
      "loss": 12.2683,
      "step": 18422
    },
    {
      "epoch": 1.0032059790488248,
      "grad_norm": 0.5796531182415882,
      "learning_rate": 0.00010433722203721725,
      "loss": 12.0797,
      "step": 18423
    },
    {
      "epoch": 1.0032604330454078,
      "grad_norm": 0.7296537883602282,
      "learning_rate": 0.00010432841206033627,
      "loss": 12.3333,
      "step": 18424
    },
    {
      "epoch": 1.0033148870419908,
      "grad_norm": 0.5213973256119997,
      "learning_rate": 0.00010431960204979683,
      "loss": 12.0892,
      "step": 18425
    },
    {
      "epoch": 1.0033693410385738,
      "grad_norm": 0.5898352075567528,
      "learning_rate": 0.00010431079200566745,
      "loss": 11.9569,
      "step": 18426
    },
    {
      "epoch": 1.0034237950351568,
      "grad_norm": 0.6532708033340328,
      "learning_rate": 0.00010430198192801662,
      "loss": 12.0374,
      "step": 18427
    },
    {
      "epoch": 1.0034782490317398,
      "grad_norm": 0.5499115261605702,
      "learning_rate": 0.00010429317181691283,
      "loss": 11.9641,
      "step": 18428
    },
    {
      "epoch": 1.003532703028323,
      "grad_norm": 0.6449371239041375,
      "learning_rate": 0.00010428436167242464,
      "loss": 12.0602,
      "step": 18429
    },
    {
      "epoch": 1.003587157024906,
      "grad_norm": 0.7144262127444471,
      "learning_rate": 0.00010427555149462053,
      "loss": 12.108,
      "step": 18430
    },
    {
      "epoch": 1.003641611021489,
      "grad_norm": 0.577597272843995,
      "learning_rate": 0.00010426674128356904,
      "loss": 12.0951,
      "step": 18431
    },
    {
      "epoch": 1.003696065018072,
      "grad_norm": 0.5852692927340701,
      "learning_rate": 0.0001042579310393386,
      "loss": 12.0938,
      "step": 18432
    },
    {
      "epoch": 1.003750519014655,
      "grad_norm": 0.7541386181334564,
      "learning_rate": 0.00010424912076199782,
      "loss": 12.1857,
      "step": 18433
    },
    {
      "epoch": 1.003804973011238,
      "grad_norm": 0.5903454039399789,
      "learning_rate": 0.00010424031045161516,
      "loss": 12.0776,
      "step": 18434
    },
    {
      "epoch": 1.003859427007821,
      "grad_norm": 0.5015008043754426,
      "learning_rate": 0.00010423150010825915,
      "loss": 12.0856,
      "step": 18435
    },
    {
      "epoch": 1.003913881004404,
      "grad_norm": 0.6864607542844916,
      "learning_rate": 0.00010422268973199822,
      "loss": 12.1423,
      "step": 18436
    },
    {
      "epoch": 1.003968335000987,
      "grad_norm": 0.576314090636686,
      "learning_rate": 0.000104213879322901,
      "loss": 12.1007,
      "step": 18437
    },
    {
      "epoch": 1.00402278899757,
      "grad_norm": 0.5434246964817178,
      "learning_rate": 0.00010420506888103593,
      "loss": 11.9121,
      "step": 18438
    },
    {
      "epoch": 1.004077242994153,
      "grad_norm": 0.6407559040439097,
      "learning_rate": 0.00010419625840647156,
      "loss": 12.1781,
      "step": 18439
    },
    {
      "epoch": 1.0041316969907361,
      "grad_norm": 0.6164183897597848,
      "learning_rate": 0.00010418744789927637,
      "loss": 12.1224,
      "step": 18440
    },
    {
      "epoch": 1.0041861509873191,
      "grad_norm": 0.57202267820432,
      "learning_rate": 0.0001041786373595189,
      "loss": 12.081,
      "step": 18441
    },
    {
      "epoch": 1.004240604983902,
      "grad_norm": 0.5641499945102079,
      "learning_rate": 0.0001041698267872676,
      "loss": 12.1598,
      "step": 18442
    },
    {
      "epoch": 1.004295058980485,
      "grad_norm": 0.6088130538427488,
      "learning_rate": 0.00010416101618259104,
      "loss": 12.1856,
      "step": 18443
    },
    {
      "epoch": 1.004349512977068,
      "grad_norm": 0.8869373652399809,
      "learning_rate": 0.00010415220554555774,
      "loss": 12.061,
      "step": 18444
    },
    {
      "epoch": 1.004403966973651,
      "grad_norm": 0.5206873460568164,
      "learning_rate": 0.00010414339487623618,
      "loss": 12.047,
      "step": 18445
    },
    {
      "epoch": 1.004458420970234,
      "grad_norm": 0.525736502612232,
      "learning_rate": 0.00010413458417469491,
      "loss": 12.0695,
      "step": 18446
    },
    {
      "epoch": 1.004512874966817,
      "grad_norm": 0.5385043026808839,
      "learning_rate": 0.00010412577344100239,
      "loss": 12.1558,
      "step": 18447
    },
    {
      "epoch": 1.0045673289634,
      "grad_norm": 0.5321986529369179,
      "learning_rate": 0.00010411696267522718,
      "loss": 12.0883,
      "step": 18448
    },
    {
      "epoch": 1.004621782959983,
      "grad_norm": 0.5432885165476395,
      "learning_rate": 0.00010410815187743776,
      "loss": 12.0963,
      "step": 18449
    },
    {
      "epoch": 1.004676236956566,
      "grad_norm": 0.5153007684808616,
      "learning_rate": 0.00010409934104770269,
      "loss": 12.1426,
      "step": 18450
    },
    {
      "epoch": 1.004730690953149,
      "grad_norm": 0.7251781770255962,
      "learning_rate": 0.00010409053018609045,
      "loss": 12.091,
      "step": 18451
    },
    {
      "epoch": 1.0047851449497323,
      "grad_norm": 0.5546233647379865,
      "learning_rate": 0.00010408171929266954,
      "loss": 12.0329,
      "step": 18452
    },
    {
      "epoch": 1.0048395989463152,
      "grad_norm": 0.5691353512399754,
      "learning_rate": 0.00010407290836750849,
      "loss": 12.0922,
      "step": 18453
    },
    {
      "epoch": 1.0048940529428982,
      "grad_norm": 0.5097120315588252,
      "learning_rate": 0.00010406409741067584,
      "loss": 12.0595,
      "step": 18454
    },
    {
      "epoch": 1.0049485069394812,
      "grad_norm": 0.5588728055036118,
      "learning_rate": 0.00010405528642224011,
      "loss": 12.2999,
      "step": 18455
    },
    {
      "epoch": 1.0050029609360642,
      "grad_norm": 0.5455730220948464,
      "learning_rate": 0.00010404647540226977,
      "loss": 12.0505,
      "step": 18456
    },
    {
      "epoch": 1.0050574149326472,
      "grad_norm": 0.4835546737187294,
      "learning_rate": 0.00010403766435083337,
      "loss": 12.0044,
      "step": 18457
    },
    {
      "epoch": 1.0051118689292302,
      "grad_norm": 0.5266537781447859,
      "learning_rate": 0.0001040288532679994,
      "loss": 12.0936,
      "step": 18458
    },
    {
      "epoch": 1.0051663229258132,
      "grad_norm": 0.7211166245977295,
      "learning_rate": 0.00010402004215383638,
      "loss": 12.2525,
      "step": 18459
    },
    {
      "epoch": 1.0052207769223962,
      "grad_norm": 0.5973374180082196,
      "learning_rate": 0.00010401123100841288,
      "loss": 12.1424,
      "step": 18460
    },
    {
      "epoch": 1.0052752309189792,
      "grad_norm": 0.5170532918173177,
      "learning_rate": 0.00010400241983179735,
      "loss": 12.0298,
      "step": 18461
    },
    {
      "epoch": 1.0053296849155622,
      "grad_norm": 0.5375349396750642,
      "learning_rate": 0.00010399360862405832,
      "loss": 12.1485,
      "step": 18462
    },
    {
      "epoch": 1.0053841389121452,
      "grad_norm": 0.520035242682927,
      "learning_rate": 0.00010398479738526434,
      "loss": 12.1356,
      "step": 18463
    },
    {
      "epoch": 1.0054385929087284,
      "grad_norm": 0.5432332946684317,
      "learning_rate": 0.00010397598611548387,
      "loss": 11.8634,
      "step": 18464
    },
    {
      "epoch": 1.0054930469053114,
      "grad_norm": 0.5893141249318277,
      "learning_rate": 0.00010396717481478551,
      "loss": 12.1584,
      "step": 18465
    },
    {
      "epoch": 1.0055475009018944,
      "grad_norm": 0.5408766358330444,
      "learning_rate": 0.00010395836348323771,
      "loss": 12.0764,
      "step": 18466
    },
    {
      "epoch": 1.0056019548984774,
      "grad_norm": 0.5598581105351644,
      "learning_rate": 0.00010394955212090903,
      "loss": 11.7836,
      "step": 18467
    },
    {
      "epoch": 1.0056564088950604,
      "grad_norm": 0.538544943045943,
      "learning_rate": 0.00010394074072786794,
      "loss": 11.8081,
      "step": 18468
    },
    {
      "epoch": 1.0057108628916434,
      "grad_norm": 0.5679894467120316,
      "learning_rate": 0.00010393192930418302,
      "loss": 11.9564,
      "step": 18469
    },
    {
      "epoch": 1.0057653168882263,
      "grad_norm": 0.6323405377769352,
      "learning_rate": 0.00010392311784992275,
      "loss": 12.0496,
      "step": 18470
    },
    {
      "epoch": 1.0058197708848093,
      "grad_norm": 0.5766824622562673,
      "learning_rate": 0.00010391430636515565,
      "loss": 12.0748,
      "step": 18471
    },
    {
      "epoch": 1.0058742248813923,
      "grad_norm": 0.5564550229807947,
      "learning_rate": 0.00010390549484995024,
      "loss": 11.993,
      "step": 18472
    },
    {
      "epoch": 1.0059286788779753,
      "grad_norm": 0.6228642070002696,
      "learning_rate": 0.00010389668330437507,
      "loss": 11.996,
      "step": 18473
    },
    {
      "epoch": 1.0059831328745583,
      "grad_norm": 0.5529858077664811,
      "learning_rate": 0.00010388787172849863,
      "loss": 12.076,
      "step": 18474
    },
    {
      "epoch": 1.0060375868711415,
      "grad_norm": 0.5923726170221378,
      "learning_rate": 0.00010387906012238943,
      "loss": 12.1028,
      "step": 18475
    },
    {
      "epoch": 1.0060920408677245,
      "grad_norm": 0.5525567253092625,
      "learning_rate": 0.00010387024848611604,
      "loss": 11.9582,
      "step": 18476
    },
    {
      "epoch": 1.0061464948643075,
      "grad_norm": 0.5672762174972753,
      "learning_rate": 0.00010386143681974692,
      "loss": 11.9318,
      "step": 18477
    },
    {
      "epoch": 1.0062009488608905,
      "grad_norm": 0.5536384723037703,
      "learning_rate": 0.00010385262512335063,
      "loss": 12.0877,
      "step": 18478
    },
    {
      "epoch": 1.0062554028574735,
      "grad_norm": 0.5384526911923556,
      "learning_rate": 0.00010384381339699567,
      "loss": 12.0977,
      "step": 18479
    },
    {
      "epoch": 1.0063098568540565,
      "grad_norm": 0.5593960088882626,
      "learning_rate": 0.00010383500164075059,
      "loss": 12.0818,
      "step": 18480
    },
    {
      "epoch": 1.0063643108506395,
      "grad_norm": 0.6027909589961592,
      "learning_rate": 0.00010382618985468389,
      "loss": 12.0996,
      "step": 18481
    },
    {
      "epoch": 1.0064187648472225,
      "grad_norm": 0.6295374486803064,
      "learning_rate": 0.00010381737803886409,
      "loss": 12.3242,
      "step": 18482
    },
    {
      "epoch": 1.0064732188438055,
      "grad_norm": 0.5972378296093238,
      "learning_rate": 0.00010380856619335973,
      "loss": 12.1621,
      "step": 18483
    },
    {
      "epoch": 1.0065276728403885,
      "grad_norm": 0.6392123771019983,
      "learning_rate": 0.0001037997543182393,
      "loss": 12.1677,
      "step": 18484
    },
    {
      "epoch": 1.0065821268369715,
      "grad_norm": 0.5159467258469377,
      "learning_rate": 0.00010379094241357134,
      "loss": 12.0815,
      "step": 18485
    },
    {
      "epoch": 1.0066365808335545,
      "grad_norm": 0.5290614198308999,
      "learning_rate": 0.0001037821304794244,
      "loss": 12.0834,
      "step": 18486
    },
    {
      "epoch": 1.0066910348301377,
      "grad_norm": 0.5618735098496637,
      "learning_rate": 0.00010377331851586699,
      "loss": 12.0976,
      "step": 18487
    },
    {
      "epoch": 1.0067454888267207,
      "grad_norm": 0.5436998168386749,
      "learning_rate": 0.00010376450652296759,
      "loss": 12.0281,
      "step": 18488
    },
    {
      "epoch": 1.0067999428233037,
      "grad_norm": 0.5783917074189285,
      "learning_rate": 0.00010375569450079476,
      "loss": 12.0415,
      "step": 18489
    },
    {
      "epoch": 1.0068543968198866,
      "grad_norm": 0.5613412995931357,
      "learning_rate": 0.00010374688244941707,
      "loss": 12.0988,
      "step": 18490
    },
    {
      "epoch": 1.0069088508164696,
      "grad_norm": 0.5748374790527079,
      "learning_rate": 0.00010373807036890291,
      "loss": 11.9973,
      "step": 18491
    },
    {
      "epoch": 1.0069633048130526,
      "grad_norm": 0.512362412059344,
      "learning_rate": 0.00010372925825932093,
      "loss": 11.956,
      "step": 18492
    },
    {
      "epoch": 1.0070177588096356,
      "grad_norm": 0.6758198926200745,
      "learning_rate": 0.00010372044612073961,
      "loss": 12.0425,
      "step": 18493
    },
    {
      "epoch": 1.0070722128062186,
      "grad_norm": 0.5579764231238112,
      "learning_rate": 0.00010371163395322749,
      "loss": 12.06,
      "step": 18494
    },
    {
      "epoch": 1.0071266668028016,
      "grad_norm": 0.5418374459029155,
      "learning_rate": 0.00010370282175685308,
      "loss": 12.021,
      "step": 18495
    },
    {
      "epoch": 1.0071811207993846,
      "grad_norm": 0.6048706028105221,
      "learning_rate": 0.00010369400953168489,
      "loss": 12.1112,
      "step": 18496
    },
    {
      "epoch": 1.0072355747959676,
      "grad_norm": 0.5569148287664402,
      "learning_rate": 0.00010368519727779147,
      "loss": 12.1173,
      "step": 18497
    },
    {
      "epoch": 1.0072900287925506,
      "grad_norm": 0.527313019239876,
      "learning_rate": 0.00010367638499524132,
      "loss": 12.1017,
      "step": 18498
    },
    {
      "epoch": 1.0073444827891338,
      "grad_norm": 0.5977487624522466,
      "learning_rate": 0.00010366757268410302,
      "loss": 12.112,
      "step": 18499
    },
    {
      "epoch": 1.0073989367857168,
      "grad_norm": 0.5721359655878242,
      "learning_rate": 0.00010365876034444506,
      "loss": 12.0808,
      "step": 18500
    },
    {
      "epoch": 1.0074533907822998,
      "grad_norm": 0.5107380598881754,
      "learning_rate": 0.00010364994797633594,
      "loss": 11.9403,
      "step": 18501
    },
    {
      "epoch": 1.0075078447788828,
      "grad_norm": 0.6051001091539683,
      "learning_rate": 0.0001036411355798442,
      "loss": 12.0673,
      "step": 18502
    },
    {
      "epoch": 1.0075622987754658,
      "grad_norm": 0.6471543229360646,
      "learning_rate": 0.00010363232315503841,
      "loss": 12.1068,
      "step": 18503
    },
    {
      "epoch": 1.0076167527720488,
      "grad_norm": 0.564946503043798,
      "learning_rate": 0.00010362351070198705,
      "loss": 11.9981,
      "step": 18504
    },
    {
      "epoch": 1.0076712067686318,
      "grad_norm": 0.5337812757312993,
      "learning_rate": 0.00010361469822075869,
      "loss": 12.1306,
      "step": 18505
    },
    {
      "epoch": 1.0077256607652147,
      "grad_norm": 0.6579308245165749,
      "learning_rate": 0.0001036058857114218,
      "loss": 12.0056,
      "step": 18506
    },
    {
      "epoch": 1.0077801147617977,
      "grad_norm": 0.5874164251985708,
      "learning_rate": 0.00010359707317404494,
      "loss": 12.1955,
      "step": 18507
    },
    {
      "epoch": 1.0078345687583807,
      "grad_norm": 0.5816155901461761,
      "learning_rate": 0.00010358826060869664,
      "loss": 12.006,
      "step": 18508
    },
    {
      "epoch": 1.0078890227549637,
      "grad_norm": 0.6768919139281958,
      "learning_rate": 0.00010357944801544541,
      "loss": 11.9988,
      "step": 18509
    },
    {
      "epoch": 1.007943476751547,
      "grad_norm": 0.6709855862271281,
      "learning_rate": 0.00010357063539435985,
      "loss": 12.206,
      "step": 18510
    },
    {
      "epoch": 1.00799793074813,
      "grad_norm": 0.5917606577492475,
      "learning_rate": 0.00010356182274550838,
      "loss": 12.0946,
      "step": 18511
    },
    {
      "epoch": 1.008052384744713,
      "grad_norm": 0.5531900919101767,
      "learning_rate": 0.00010355301006895958,
      "loss": 12.1235,
      "step": 18512
    },
    {
      "epoch": 1.008106838741296,
      "grad_norm": 0.5723874950368045,
      "learning_rate": 0.00010354419736478198,
      "loss": 12.0916,
      "step": 18513
    },
    {
      "epoch": 1.008161292737879,
      "grad_norm": 0.6002824915363437,
      "learning_rate": 0.00010353538463304411,
      "loss": 12.1344,
      "step": 18514
    },
    {
      "epoch": 1.008215746734462,
      "grad_norm": 0.5513697032116269,
      "learning_rate": 0.00010352657187381451,
      "loss": 12.0326,
      "step": 18515
    },
    {
      "epoch": 1.008270200731045,
      "grad_norm": 0.5574745845879414,
      "learning_rate": 0.0001035177590871617,
      "loss": 12.0332,
      "step": 18516
    },
    {
      "epoch": 1.008324654727628,
      "grad_norm": 0.5538227245458262,
      "learning_rate": 0.00010350894627315417,
      "loss": 12.177,
      "step": 18517
    },
    {
      "epoch": 1.0083791087242109,
      "grad_norm": 0.6256207524064054,
      "learning_rate": 0.00010350013343186051,
      "loss": 12.1336,
      "step": 18518
    },
    {
      "epoch": 1.0084335627207939,
      "grad_norm": 0.544668977596636,
      "learning_rate": 0.0001034913205633492,
      "loss": 12.1373,
      "step": 18519
    },
    {
      "epoch": 1.0084880167173769,
      "grad_norm": 0.5971507286826666,
      "learning_rate": 0.00010348250766768885,
      "loss": 12.0926,
      "step": 18520
    },
    {
      "epoch": 1.0085424707139599,
      "grad_norm": 0.5554077923661265,
      "learning_rate": 0.00010347369474494791,
      "loss": 12.0171,
      "step": 18521
    },
    {
      "epoch": 1.008596924710543,
      "grad_norm": 0.6661562125947798,
      "learning_rate": 0.00010346488179519491,
      "loss": 12.0987,
      "step": 18522
    },
    {
      "epoch": 1.008651378707126,
      "grad_norm": 0.5218887040607149,
      "learning_rate": 0.00010345606881849843,
      "loss": 12.1033,
      "step": 18523
    },
    {
      "epoch": 1.008705832703709,
      "grad_norm": 0.5641752186109523,
      "learning_rate": 0.00010344725581492695,
      "loss": 12.2255,
      "step": 18524
    },
    {
      "epoch": 1.008760286700292,
      "grad_norm": 0.5770632410257892,
      "learning_rate": 0.00010343844278454908,
      "loss": 12.1212,
      "step": 18525
    },
    {
      "epoch": 1.008814740696875,
      "grad_norm": 0.5254078477638074,
      "learning_rate": 0.00010342962972743329,
      "loss": 11.8766,
      "step": 18526
    },
    {
      "epoch": 1.008869194693458,
      "grad_norm": 0.5831145330438583,
      "learning_rate": 0.00010342081664364811,
      "loss": 11.9643,
      "step": 18527
    },
    {
      "epoch": 1.008923648690041,
      "grad_norm": 0.556557972163656,
      "learning_rate": 0.00010341200353326211,
      "loss": 12.1719,
      "step": 18528
    },
    {
      "epoch": 1.008978102686624,
      "grad_norm": 0.5579480008859713,
      "learning_rate": 0.00010340319039634376,
      "loss": 12.1567,
      "step": 18529
    },
    {
      "epoch": 1.009032556683207,
      "grad_norm": 0.5204330978413887,
      "learning_rate": 0.00010339437723296166,
      "loss": 12.0736,
      "step": 18530
    },
    {
      "epoch": 1.00908701067979,
      "grad_norm": 0.5512177356789631,
      "learning_rate": 0.0001033855640431843,
      "loss": 12.0599,
      "step": 18531
    },
    {
      "epoch": 1.009141464676373,
      "grad_norm": 0.5358028471647213,
      "learning_rate": 0.00010337675082708023,
      "loss": 11.9859,
      "step": 18532
    },
    {
      "epoch": 1.009195918672956,
      "grad_norm": 0.5478092408227072,
      "learning_rate": 0.00010336793758471797,
      "loss": 11.9727,
      "step": 18533
    },
    {
      "epoch": 1.0092503726695392,
      "grad_norm": 0.5514440248075692,
      "learning_rate": 0.00010335912431616608,
      "loss": 12.0721,
      "step": 18534
    },
    {
      "epoch": 1.0093048266661222,
      "grad_norm": 0.659187372167803,
      "learning_rate": 0.00010335031102149306,
      "loss": 12.1512,
      "step": 18535
    },
    {
      "epoch": 1.0093592806627052,
      "grad_norm": 0.539932060291416,
      "learning_rate": 0.00010334149770076747,
      "loss": 12.1408,
      "step": 18536
    },
    {
      "epoch": 1.0094137346592882,
      "grad_norm": 0.6274236836397442,
      "learning_rate": 0.00010333268435405783,
      "loss": 12.2776,
      "step": 18537
    },
    {
      "epoch": 1.0094681886558712,
      "grad_norm": 0.6185688062471368,
      "learning_rate": 0.00010332387098143267,
      "loss": 12.1801,
      "step": 18538
    },
    {
      "epoch": 1.0095226426524542,
      "grad_norm": 0.5242210851813076,
      "learning_rate": 0.00010331505758296054,
      "loss": 12.0495,
      "step": 18539
    },
    {
      "epoch": 1.0095770966490372,
      "grad_norm": 0.5033737270364694,
      "learning_rate": 0.00010330624415870998,
      "loss": 11.9722,
      "step": 18540
    },
    {
      "epoch": 1.0096315506456202,
      "grad_norm": 0.5950935511479065,
      "learning_rate": 0.00010329743070874949,
      "loss": 12.1449,
      "step": 18541
    },
    {
      "epoch": 1.0096860046422031,
      "grad_norm": 0.5280528480410401,
      "learning_rate": 0.00010328861723314763,
      "loss": 11.964,
      "step": 18542
    },
    {
      "epoch": 1.0097404586387861,
      "grad_norm": 0.5309463025005629,
      "learning_rate": 0.00010327980373197294,
      "loss": 12.0875,
      "step": 18543
    },
    {
      "epoch": 1.0097949126353691,
      "grad_norm": 0.5320408860846423,
      "learning_rate": 0.00010327099020529393,
      "loss": 12.0683,
      "step": 18544
    },
    {
      "epoch": 1.0098493666319523,
      "grad_norm": 0.6038898483784096,
      "learning_rate": 0.00010326217665317916,
      "loss": 11.9718,
      "step": 18545
    },
    {
      "epoch": 1.0099038206285353,
      "grad_norm": 0.6014477134507331,
      "learning_rate": 0.00010325336307569717,
      "loss": 12.13,
      "step": 18546
    },
    {
      "epoch": 1.0099582746251183,
      "grad_norm": 0.5495599267243306,
      "learning_rate": 0.00010324454947291647,
      "loss": 12.045,
      "step": 18547
    },
    {
      "epoch": 1.0100127286217013,
      "grad_norm": 0.5280405649469392,
      "learning_rate": 0.00010323573584490561,
      "loss": 12.0402,
      "step": 18548
    },
    {
      "epoch": 1.0100671826182843,
      "grad_norm": 0.5590605689991598,
      "learning_rate": 0.00010322692219173314,
      "loss": 12.1249,
      "step": 18549
    },
    {
      "epoch": 1.0101216366148673,
      "grad_norm": 0.6570155279799302,
      "learning_rate": 0.00010321810851346758,
      "loss": 12.0431,
      "step": 18550
    },
    {
      "epoch": 1.0101760906114503,
      "grad_norm": 0.5111924942415713,
      "learning_rate": 0.00010320929481017743,
      "loss": 12.0129,
      "step": 18551
    },
    {
      "epoch": 1.0102305446080333,
      "grad_norm": 0.5239726843256711,
      "learning_rate": 0.0001032004810819313,
      "loss": 12.0665,
      "step": 18552
    },
    {
      "epoch": 1.0102849986046163,
      "grad_norm": 0.5345809725825313,
      "learning_rate": 0.00010319166732879768,
      "loss": 12.0353,
      "step": 18553
    },
    {
      "epoch": 1.0103394526011993,
      "grad_norm": 0.630011898885438,
      "learning_rate": 0.00010318285355084512,
      "loss": 12.1699,
      "step": 18554
    },
    {
      "epoch": 1.0103939065977823,
      "grad_norm": 0.5191116007882293,
      "learning_rate": 0.00010317403974814217,
      "loss": 11.9667,
      "step": 18555
    },
    {
      "epoch": 1.0104483605943653,
      "grad_norm": 0.4831786103221667,
      "learning_rate": 0.00010316522592075734,
      "loss": 12.1,
      "step": 18556
    },
    {
      "epoch": 1.0105028145909485,
      "grad_norm": 0.5561397048347925,
      "learning_rate": 0.00010315641206875919,
      "loss": 12.0128,
      "step": 18557
    },
    {
      "epoch": 1.0105572685875315,
      "grad_norm": 0.6440097086515711,
      "learning_rate": 0.00010314759819221624,
      "loss": 12.2757,
      "step": 18558
    },
    {
      "epoch": 1.0106117225841145,
      "grad_norm": 0.656110245218873,
      "learning_rate": 0.00010313878429119705,
      "loss": 12.1454,
      "step": 18559
    },
    {
      "epoch": 1.0106661765806975,
      "grad_norm": 0.5903269479289363,
      "learning_rate": 0.00010312997036577014,
      "loss": 11.9893,
      "step": 18560
    },
    {
      "epoch": 1.0107206305772805,
      "grad_norm": 0.6022232729710705,
      "learning_rate": 0.00010312115641600408,
      "loss": 12.0534,
      "step": 18561
    },
    {
      "epoch": 1.0107750845738634,
      "grad_norm": 0.5060077336455366,
      "learning_rate": 0.00010311234244196735,
      "loss": 11.9423,
      "step": 18562
    },
    {
      "epoch": 1.0108295385704464,
      "grad_norm": 0.5435731159339288,
      "learning_rate": 0.00010310352844372855,
      "loss": 12.089,
      "step": 18563
    },
    {
      "epoch": 1.0108839925670294,
      "grad_norm": 0.5674002940175599,
      "learning_rate": 0.00010309471442135617,
      "loss": 12.0693,
      "step": 18564
    },
    {
      "epoch": 1.0109384465636124,
      "grad_norm": 0.5404684656688796,
      "learning_rate": 0.0001030859003749188,
      "loss": 12.1342,
      "step": 18565
    },
    {
      "epoch": 1.0109929005601954,
      "grad_norm": 0.5961366588591839,
      "learning_rate": 0.00010307708630448494,
      "loss": 11.9892,
      "step": 18566
    },
    {
      "epoch": 1.0110473545567784,
      "grad_norm": 0.5454753540434794,
      "learning_rate": 0.00010306827221012312,
      "loss": 12.0481,
      "step": 18567
    },
    {
      "epoch": 1.0111018085533616,
      "grad_norm": 0.566266637646021,
      "learning_rate": 0.0001030594580919019,
      "loss": 12.1063,
      "step": 18568
    },
    {
      "epoch": 1.0111562625499446,
      "grad_norm": 0.5533092897930931,
      "learning_rate": 0.00010305064394988984,
      "loss": 12.0457,
      "step": 18569
    },
    {
      "epoch": 1.0112107165465276,
      "grad_norm": 0.5386943447678215,
      "learning_rate": 0.00010304182978415544,
      "loss": 11.8717,
      "step": 18570
    },
    {
      "epoch": 1.0112651705431106,
      "grad_norm": 0.5470324120297019,
      "learning_rate": 0.0001030330155947673,
      "loss": 12.0962,
      "step": 18571
    },
    {
      "epoch": 1.0113196245396936,
      "grad_norm": 0.5911071622271485,
      "learning_rate": 0.00010302420138179391,
      "loss": 12.2112,
      "step": 18572
    },
    {
      "epoch": 1.0113740785362766,
      "grad_norm": 0.5854015202959052,
      "learning_rate": 0.00010301538714530379,
      "loss": 12.0793,
      "step": 18573
    },
    {
      "epoch": 1.0114285325328596,
      "grad_norm": 0.5698593275883355,
      "learning_rate": 0.00010300657288536553,
      "loss": 12.0904,
      "step": 18574
    },
    {
      "epoch": 1.0114829865294426,
      "grad_norm": 0.5832187817899874,
      "learning_rate": 0.00010299775860204768,
      "loss": 12.1363,
      "step": 18575
    },
    {
      "epoch": 1.0115374405260256,
      "grad_norm": 0.5679029416512514,
      "learning_rate": 0.00010298894429541874,
      "loss": 12.1225,
      "step": 18576
    },
    {
      "epoch": 1.0115918945226086,
      "grad_norm": 0.5673043538062682,
      "learning_rate": 0.00010298012996554727,
      "loss": 11.9131,
      "step": 18577
    },
    {
      "epoch": 1.0116463485191916,
      "grad_norm": 0.5366736456338547,
      "learning_rate": 0.00010297131561250182,
      "loss": 11.7861,
      "step": 18578
    },
    {
      "epoch": 1.0117008025157745,
      "grad_norm": 0.6059117847345783,
      "learning_rate": 0.00010296250123635087,
      "loss": 12.0755,
      "step": 18579
    },
    {
      "epoch": 1.0117552565123578,
      "grad_norm": 0.5766746731730145,
      "learning_rate": 0.00010295368683716305,
      "loss": 12.1941,
      "step": 18580
    },
    {
      "epoch": 1.0118097105089408,
      "grad_norm": 0.5975853286218374,
      "learning_rate": 0.0001029448724150069,
      "loss": 12.2667,
      "step": 18581
    },
    {
      "epoch": 1.0118641645055237,
      "grad_norm": 0.5813723381540811,
      "learning_rate": 0.0001029360579699509,
      "loss": 12.0985,
      "step": 18582
    },
    {
      "epoch": 1.0119186185021067,
      "grad_norm": 0.5144480498398242,
      "learning_rate": 0.00010292724350206359,
      "loss": 12.1034,
      "step": 18583
    },
    {
      "epoch": 1.0119730724986897,
      "grad_norm": 0.5496274656727532,
      "learning_rate": 0.00010291842901141357,
      "loss": 11.8125,
      "step": 18584
    },
    {
      "epoch": 1.0120275264952727,
      "grad_norm": 0.5870109527864531,
      "learning_rate": 0.00010290961449806935,
      "loss": 12.1074,
      "step": 18585
    },
    {
      "epoch": 1.0120819804918557,
      "grad_norm": 0.5517661081173563,
      "learning_rate": 0.00010290079996209949,
      "loss": 12.1656,
      "step": 18586
    },
    {
      "epoch": 1.0121364344884387,
      "grad_norm": 0.5204688912020631,
      "learning_rate": 0.00010289198540357252,
      "loss": 12.1468,
      "step": 18587
    },
    {
      "epoch": 1.0121908884850217,
      "grad_norm": 0.6109212134761495,
      "learning_rate": 0.00010288317082255698,
      "loss": 12.0229,
      "step": 18588
    },
    {
      "epoch": 1.0122453424816047,
      "grad_norm": 0.50168078275934,
      "learning_rate": 0.00010287435621912145,
      "loss": 12.0959,
      "step": 18589
    },
    {
      "epoch": 1.0122997964781877,
      "grad_norm": 0.5870154397372992,
      "learning_rate": 0.00010286554159333439,
      "loss": 12.1773,
      "step": 18590
    },
    {
      "epoch": 1.0123542504747707,
      "grad_norm": 0.5661339588060216,
      "learning_rate": 0.00010285672694526445,
      "loss": 12.1347,
      "step": 18591
    },
    {
      "epoch": 1.012408704471354,
      "grad_norm": 0.5685607270196777,
      "learning_rate": 0.0001028479122749801,
      "loss": 11.9809,
      "step": 18592
    },
    {
      "epoch": 1.0124631584679369,
      "grad_norm": 0.586910468464697,
      "learning_rate": 0.0001028390975825499,
      "loss": 12.1174,
      "step": 18593
    },
    {
      "epoch": 1.0125176124645199,
      "grad_norm": 0.5493072224736333,
      "learning_rate": 0.00010283028286804241,
      "loss": 12.0765,
      "step": 18594
    },
    {
      "epoch": 1.0125720664611029,
      "grad_norm": 0.6860285133901204,
      "learning_rate": 0.00010282146813152616,
      "loss": 12.1214,
      "step": 18595
    },
    {
      "epoch": 1.0126265204576859,
      "grad_norm": 0.5578553443266105,
      "learning_rate": 0.00010281265337306971,
      "loss": 12.1132,
      "step": 18596
    },
    {
      "epoch": 1.0126809744542689,
      "grad_norm": 0.760650444662143,
      "learning_rate": 0.00010280383859274159,
      "loss": 12.1466,
      "step": 18597
    },
    {
      "epoch": 1.0127354284508518,
      "grad_norm": 0.5341634657264303,
      "learning_rate": 0.00010279502379061035,
      "loss": 12.0617,
      "step": 18598
    },
    {
      "epoch": 1.0127898824474348,
      "grad_norm": 0.598461962481969,
      "learning_rate": 0.00010278620896674453,
      "loss": 12.0937,
      "step": 18599
    },
    {
      "epoch": 1.0128443364440178,
      "grad_norm": 0.6275053519413482,
      "learning_rate": 0.00010277739412121267,
      "loss": 12.0502,
      "step": 18600
    },
    {
      "epoch": 1.0128987904406008,
      "grad_norm": 0.5627921254857454,
      "learning_rate": 0.00010276857925408337,
      "loss": 12.0195,
      "step": 18601
    },
    {
      "epoch": 1.0129532444371838,
      "grad_norm": 0.6277134432781751,
      "learning_rate": 0.00010275976436542509,
      "loss": 12.0657,
      "step": 18602
    },
    {
      "epoch": 1.0130076984337668,
      "grad_norm": 0.552236707841139,
      "learning_rate": 0.00010275094945530645,
      "loss": 12.087,
      "step": 18603
    },
    {
      "epoch": 1.01306215243035,
      "grad_norm": 0.5339410393649697,
      "learning_rate": 0.00010274213452379595,
      "loss": 12.0573,
      "step": 18604
    },
    {
      "epoch": 1.013116606426933,
      "grad_norm": 0.5199770355858867,
      "learning_rate": 0.00010273331957096215,
      "loss": 12.0424,
      "step": 18605
    },
    {
      "epoch": 1.013171060423516,
      "grad_norm": 0.6059916556226567,
      "learning_rate": 0.00010272450459687362,
      "loss": 12.1755,
      "step": 18606
    },
    {
      "epoch": 1.013225514420099,
      "grad_norm": 0.6439764951819686,
      "learning_rate": 0.00010271568960159887,
      "loss": 11.9937,
      "step": 18607
    },
    {
      "epoch": 1.013279968416682,
      "grad_norm": 0.5616246688610501,
      "learning_rate": 0.00010270687458520645,
      "loss": 12.2604,
      "step": 18608
    },
    {
      "epoch": 1.013334422413265,
      "grad_norm": 0.5976513162085014,
      "learning_rate": 0.00010269805954776495,
      "loss": 12.0994,
      "step": 18609
    },
    {
      "epoch": 1.013388876409848,
      "grad_norm": 0.5814974973020262,
      "learning_rate": 0.00010268924448934285,
      "loss": 12.225,
      "step": 18610
    },
    {
      "epoch": 1.013443330406431,
      "grad_norm": 0.5557409579034941,
      "learning_rate": 0.00010268042941000874,
      "loss": 12.1234,
      "step": 18611
    },
    {
      "epoch": 1.013497784403014,
      "grad_norm": 0.5961948506538783,
      "learning_rate": 0.00010267161430983119,
      "loss": 12.1489,
      "step": 18612
    },
    {
      "epoch": 1.013552238399597,
      "grad_norm": 0.5275476176618449,
      "learning_rate": 0.00010266279918887872,
      "loss": 12.0462,
      "step": 18613
    },
    {
      "epoch": 1.01360669239618,
      "grad_norm": 0.46270763533338205,
      "learning_rate": 0.00010265398404721984,
      "loss": 12.0466,
      "step": 18614
    },
    {
      "epoch": 1.0136611463927632,
      "grad_norm": 0.5396062804897083,
      "learning_rate": 0.00010264516888492315,
      "loss": 12.1478,
      "step": 18615
    },
    {
      "epoch": 1.0137156003893462,
      "grad_norm": 0.5603888801196434,
      "learning_rate": 0.00010263635370205722,
      "loss": 12.0761,
      "step": 18616
    },
    {
      "epoch": 1.0137700543859292,
      "grad_norm": 0.5949006213362621,
      "learning_rate": 0.00010262753849869051,
      "loss": 12.1037,
      "step": 18617
    },
    {
      "epoch": 1.0138245083825121,
      "grad_norm": 0.5478553417772944,
      "learning_rate": 0.00010261872327489164,
      "loss": 12.0866,
      "step": 18618
    },
    {
      "epoch": 1.0138789623790951,
      "grad_norm": 0.5663167078664715,
      "learning_rate": 0.00010260990803072915,
      "loss": 12.1534,
      "step": 18619
    },
    {
      "epoch": 1.0139334163756781,
      "grad_norm": 0.5931662652583294,
      "learning_rate": 0.00010260109276627154,
      "loss": 12.0215,
      "step": 18620
    },
    {
      "epoch": 1.0139878703722611,
      "grad_norm": 0.5409318674949393,
      "learning_rate": 0.00010259227748158747,
      "loss": 12.1054,
      "step": 18621
    },
    {
      "epoch": 1.0140423243688441,
      "grad_norm": 0.5896010595760806,
      "learning_rate": 0.00010258346217674532,
      "loss": 12.1124,
      "step": 18622
    },
    {
      "epoch": 1.014096778365427,
      "grad_norm": 0.5106183988735412,
      "learning_rate": 0.0001025746468518138,
      "loss": 12.1734,
      "step": 18623
    },
    {
      "epoch": 1.01415123236201,
      "grad_norm": 0.5312656153615478,
      "learning_rate": 0.00010256583150686136,
      "loss": 11.9229,
      "step": 18624
    },
    {
      "epoch": 1.014205686358593,
      "grad_norm": 0.5194900895636492,
      "learning_rate": 0.00010255701614195661,
      "loss": 12.1093,
      "step": 18625
    },
    {
      "epoch": 1.014260140355176,
      "grad_norm": 0.5859684979836088,
      "learning_rate": 0.00010254820075716809,
      "loss": 12.1531,
      "step": 18626
    },
    {
      "epoch": 1.0143145943517593,
      "grad_norm": 0.5551928335979637,
      "learning_rate": 0.00010253938535256431,
      "loss": 12.1345,
      "step": 18627
    },
    {
      "epoch": 1.0143690483483423,
      "grad_norm": 0.5775510465028477,
      "learning_rate": 0.00010253056992821382,
      "loss": 12.1252,
      "step": 18628
    },
    {
      "epoch": 1.0144235023449253,
      "grad_norm": 0.5427232453071151,
      "learning_rate": 0.0001025217544841852,
      "loss": 12.0609,
      "step": 18629
    },
    {
      "epoch": 1.0144779563415083,
      "grad_norm": 0.6115163583493086,
      "learning_rate": 0.00010251293902054701,
      "loss": 12.0571,
      "step": 18630
    },
    {
      "epoch": 1.0145324103380913,
      "grad_norm": 0.5386722267689895,
      "learning_rate": 0.0001025041235373678,
      "loss": 12.0379,
      "step": 18631
    },
    {
      "epoch": 1.0145868643346743,
      "grad_norm": 0.5506371185172241,
      "learning_rate": 0.0001024953080347161,
      "loss": 12.0624,
      "step": 18632
    },
    {
      "epoch": 1.0146413183312573,
      "grad_norm": 0.5827098744078735,
      "learning_rate": 0.00010248649251266042,
      "loss": 11.9455,
      "step": 18633
    },
    {
      "epoch": 1.0146957723278403,
      "grad_norm": 0.592762604673711,
      "learning_rate": 0.0001024776769712694,
      "loss": 12.078,
      "step": 18634
    },
    {
      "epoch": 1.0147502263244232,
      "grad_norm": 0.5656728108497389,
      "learning_rate": 0.00010246886141061154,
      "loss": 12.1246,
      "step": 18635
    },
    {
      "epoch": 1.0148046803210062,
      "grad_norm": 0.5517144152095224,
      "learning_rate": 0.00010246004583075544,
      "loss": 12.0933,
      "step": 18636
    },
    {
      "epoch": 1.0148591343175892,
      "grad_norm": 0.5596194562044612,
      "learning_rate": 0.00010245123023176957,
      "loss": 12.0172,
      "step": 18637
    },
    {
      "epoch": 1.0149135883141724,
      "grad_norm": 0.5771751091706115,
      "learning_rate": 0.00010244241461372252,
      "loss": 12.2037,
      "step": 18638
    },
    {
      "epoch": 1.0149680423107554,
      "grad_norm": 0.5211343834215023,
      "learning_rate": 0.00010243359897668283,
      "loss": 12.0722,
      "step": 18639
    },
    {
      "epoch": 1.0150224963073384,
      "grad_norm": 0.5699645617576364,
      "learning_rate": 0.00010242478332071907,
      "loss": 12.1202,
      "step": 18640
    },
    {
      "epoch": 1.0150769503039214,
      "grad_norm": 0.5044244164856309,
      "learning_rate": 0.00010241596764589985,
      "loss": 12.0559,
      "step": 18641
    },
    {
      "epoch": 1.0151314043005044,
      "grad_norm": 0.534380669527365,
      "learning_rate": 0.0001024071519522936,
      "loss": 11.8102,
      "step": 18642
    },
    {
      "epoch": 1.0151858582970874,
      "grad_norm": 0.5889098771521126,
      "learning_rate": 0.00010239833623996895,
      "loss": 12.0705,
      "step": 18643
    },
    {
      "epoch": 1.0152403122936704,
      "grad_norm": 0.541823910174673,
      "learning_rate": 0.00010238952050899442,
      "loss": 12.1835,
      "step": 18644
    },
    {
      "epoch": 1.0152947662902534,
      "grad_norm": 0.6260545936225633,
      "learning_rate": 0.00010238070475943857,
      "loss": 12.0786,
      "step": 18645
    },
    {
      "epoch": 1.0153492202868364,
      "grad_norm": 0.51878762670923,
      "learning_rate": 0.00010237188899137,
      "loss": 11.8656,
      "step": 18646
    },
    {
      "epoch": 1.0154036742834194,
      "grad_norm": 0.5707399005959253,
      "learning_rate": 0.00010236307320485721,
      "loss": 12.0872,
      "step": 18647
    },
    {
      "epoch": 1.0154581282800024,
      "grad_norm": 0.5309342136202445,
      "learning_rate": 0.00010235425739996876,
      "loss": 12.2071,
      "step": 18648
    },
    {
      "epoch": 1.0155125822765854,
      "grad_norm": 0.5721763583756236,
      "learning_rate": 0.00010234544157677322,
      "loss": 12.201,
      "step": 18649
    },
    {
      "epoch": 1.0155670362731686,
      "grad_norm": 0.5475461071889699,
      "learning_rate": 0.00010233662573533909,
      "loss": 12.103,
      "step": 18650
    },
    {
      "epoch": 1.0156214902697516,
      "grad_norm": 0.5219852763402135,
      "learning_rate": 0.00010232780987573502,
      "loss": 12.017,
      "step": 18651
    },
    {
      "epoch": 1.0156759442663346,
      "grad_norm": 0.5092900645153045,
      "learning_rate": 0.0001023189939980295,
      "loss": 12.0555,
      "step": 18652
    },
    {
      "epoch": 1.0157303982629176,
      "grad_norm": 0.5388537015061275,
      "learning_rate": 0.00010231017810229108,
      "loss": 12.0605,
      "step": 18653
    },
    {
      "epoch": 1.0157848522595005,
      "grad_norm": 0.588907815625058,
      "learning_rate": 0.00010230136218858832,
      "loss": 12.0845,
      "step": 18654
    },
    {
      "epoch": 1.0158393062560835,
      "grad_norm": 0.5230904598736458,
      "learning_rate": 0.00010229254625698981,
      "loss": 12.1564,
      "step": 18655
    },
    {
      "epoch": 1.0158937602526665,
      "grad_norm": 0.5401076147941122,
      "learning_rate": 0.00010228373030756404,
      "loss": 12.0952,
      "step": 18656
    },
    {
      "epoch": 1.0159482142492495,
      "grad_norm": 0.5646841759125969,
      "learning_rate": 0.00010227491434037963,
      "loss": 12.0907,
      "step": 18657
    },
    {
      "epoch": 1.0160026682458325,
      "grad_norm": 0.5290137046555882,
      "learning_rate": 0.0001022660983555051,
      "loss": 12.0885,
      "step": 18658
    },
    {
      "epoch": 1.0160571222424155,
      "grad_norm": 0.5215509143252841,
      "learning_rate": 0.000102257282353009,
      "loss": 12.0678,
      "step": 18659
    },
    {
      "epoch": 1.0161115762389985,
      "grad_norm": 0.5440424781375254,
      "learning_rate": 0.00010224846633295988,
      "loss": 12.1164,
      "step": 18660
    },
    {
      "epoch": 1.0161660302355815,
      "grad_norm": 0.6125029465965208,
      "learning_rate": 0.00010223965029542632,
      "loss": 12.1944,
      "step": 18661
    },
    {
      "epoch": 1.0162204842321647,
      "grad_norm": 0.5361094072719735,
      "learning_rate": 0.00010223083424047689,
      "loss": 11.9944,
      "step": 18662
    },
    {
      "epoch": 1.0162749382287477,
      "grad_norm": 0.5382271270974258,
      "learning_rate": 0.00010222201816818009,
      "loss": 12.0014,
      "step": 18663
    },
    {
      "epoch": 1.0163293922253307,
      "grad_norm": 0.5677736707067792,
      "learning_rate": 0.00010221320207860452,
      "loss": 11.8384,
      "step": 18664
    },
    {
      "epoch": 1.0163838462219137,
      "grad_norm": 0.5571622510638461,
      "learning_rate": 0.0001022043859718187,
      "loss": 12.0366,
      "step": 18665
    },
    {
      "epoch": 1.0164383002184967,
      "grad_norm": 0.562284739796698,
      "learning_rate": 0.00010219556984789123,
      "loss": 12.062,
      "step": 18666
    },
    {
      "epoch": 1.0164927542150797,
      "grad_norm": 0.5145086992467447,
      "learning_rate": 0.00010218675370689061,
      "loss": 11.8804,
      "step": 18667
    },
    {
      "epoch": 1.0165472082116627,
      "grad_norm": 0.6198656528706378,
      "learning_rate": 0.00010217793754888544,
      "loss": 12.1277,
      "step": 18668
    },
    {
      "epoch": 1.0166016622082457,
      "grad_norm": 0.5516678091316137,
      "learning_rate": 0.00010216912137394428,
      "loss": 12.085,
      "step": 18669
    },
    {
      "epoch": 1.0166561162048287,
      "grad_norm": 0.576048071907269,
      "learning_rate": 0.00010216030518213564,
      "loss": 12.0683,
      "step": 18670
    },
    {
      "epoch": 1.0167105702014116,
      "grad_norm": 0.5530427825086842,
      "learning_rate": 0.00010215148897352814,
      "loss": 12.06,
      "step": 18671
    },
    {
      "epoch": 1.0167650241979946,
      "grad_norm": 0.5391895765062983,
      "learning_rate": 0.00010214267274819027,
      "loss": 12.0372,
      "step": 18672
    },
    {
      "epoch": 1.0168194781945776,
      "grad_norm": 0.5363936955345292,
      "learning_rate": 0.00010213385650619063,
      "loss": 11.9683,
      "step": 18673
    },
    {
      "epoch": 1.0168739321911608,
      "grad_norm": 0.5527397413340591,
      "learning_rate": 0.00010212504024759775,
      "loss": 12.1035,
      "step": 18674
    },
    {
      "epoch": 1.0169283861877438,
      "grad_norm": 0.5391156205400923,
      "learning_rate": 0.00010211622397248022,
      "loss": 12.0273,
      "step": 18675
    },
    {
      "epoch": 1.0169828401843268,
      "grad_norm": 0.5325203017680954,
      "learning_rate": 0.00010210740768090659,
      "loss": 12.134,
      "step": 18676
    },
    {
      "epoch": 1.0170372941809098,
      "grad_norm": 0.5224397247187977,
      "learning_rate": 0.00010209859137294535,
      "loss": 12.087,
      "step": 18677
    },
    {
      "epoch": 1.0170917481774928,
      "grad_norm": 0.508210641030574,
      "learning_rate": 0.00010208977504866514,
      "loss": 12.1013,
      "step": 18678
    },
    {
      "epoch": 1.0171462021740758,
      "grad_norm": 0.5594443000847564,
      "learning_rate": 0.0001020809587081345,
      "loss": 12.2038,
      "step": 18679
    },
    {
      "epoch": 1.0172006561706588,
      "grad_norm": 0.5325988145437173,
      "learning_rate": 0.00010207214235142197,
      "loss": 12.0305,
      "step": 18680
    },
    {
      "epoch": 1.0172551101672418,
      "grad_norm": 0.46681354969671174,
      "learning_rate": 0.00010206332597859614,
      "loss": 12.0177,
      "step": 18681
    },
    {
      "epoch": 1.0173095641638248,
      "grad_norm": 0.55466047261685,
      "learning_rate": 0.00010205450958972549,
      "loss": 12.1903,
      "step": 18682
    },
    {
      "epoch": 1.0173640181604078,
      "grad_norm": 0.5211680837345727,
      "learning_rate": 0.00010204569318487867,
      "loss": 12.0376,
      "step": 18683
    },
    {
      "epoch": 1.0174184721569908,
      "grad_norm": 0.531284352544976,
      "learning_rate": 0.00010203687676412416,
      "loss": 12.1672,
      "step": 18684
    },
    {
      "epoch": 1.017472926153574,
      "grad_norm": 0.5212183307590827,
      "learning_rate": 0.00010202806032753059,
      "loss": 12.1094,
      "step": 18685
    },
    {
      "epoch": 1.017527380150157,
      "grad_norm": 0.5192248511028463,
      "learning_rate": 0.00010201924387516648,
      "loss": 11.8798,
      "step": 18686
    },
    {
      "epoch": 1.01758183414674,
      "grad_norm": 0.5739008702696496,
      "learning_rate": 0.00010201042740710039,
      "loss": 12.0077,
      "step": 18687
    },
    {
      "epoch": 1.017636288143323,
      "grad_norm": 0.5127739004057067,
      "learning_rate": 0.00010200161092340083,
      "loss": 12.0929,
      "step": 18688
    },
    {
      "epoch": 1.017690742139906,
      "grad_norm": 0.5732873555817004,
      "learning_rate": 0.00010199279442413645,
      "loss": 12.0487,
      "step": 18689
    },
    {
      "epoch": 1.017745196136489,
      "grad_norm": 0.609518732694126,
      "learning_rate": 0.00010198397790937577,
      "loss": 12.1473,
      "step": 18690
    },
    {
      "epoch": 1.017799650133072,
      "grad_norm": 0.5678661229413832,
      "learning_rate": 0.00010197516137918734,
      "loss": 11.9684,
      "step": 18691
    },
    {
      "epoch": 1.017854104129655,
      "grad_norm": 0.6481102039488043,
      "learning_rate": 0.00010196634483363974,
      "loss": 12.0967,
      "step": 18692
    },
    {
      "epoch": 1.017908558126238,
      "grad_norm": 0.6070736689133404,
      "learning_rate": 0.00010195752827280149,
      "loss": 12.0106,
      "step": 18693
    },
    {
      "epoch": 1.017963012122821,
      "grad_norm": 0.5671805438971573,
      "learning_rate": 0.00010194871169674117,
      "loss": 12.1093,
      "step": 18694
    },
    {
      "epoch": 1.018017466119404,
      "grad_norm": 0.5596878343691645,
      "learning_rate": 0.00010193989510552732,
      "loss": 12.1828,
      "step": 18695
    },
    {
      "epoch": 1.018071920115987,
      "grad_norm": 0.5731959237727028,
      "learning_rate": 0.00010193107849922859,
      "loss": 12.1028,
      "step": 18696
    },
    {
      "epoch": 1.0181263741125701,
      "grad_norm": 0.557951890413181,
      "learning_rate": 0.0001019222618779134,
      "loss": 12.0865,
      "step": 18697
    },
    {
      "epoch": 1.0181808281091531,
      "grad_norm": 0.5547917029109849,
      "learning_rate": 0.00010191344524165043,
      "loss": 12.0514,
      "step": 18698
    },
    {
      "epoch": 1.018235282105736,
      "grad_norm": 0.536152548658443,
      "learning_rate": 0.00010190462859050813,
      "loss": 12.0461,
      "step": 18699
    },
    {
      "epoch": 1.018289736102319,
      "grad_norm": 0.5658531788317116,
      "learning_rate": 0.00010189581192455515,
      "loss": 12.0915,
      "step": 18700
    },
    {
      "epoch": 1.018344190098902,
      "grad_norm": 0.5396720468549974,
      "learning_rate": 0.00010188699524386003,
      "loss": 12.14,
      "step": 18701
    },
    {
      "epoch": 1.018398644095485,
      "grad_norm": 0.5452850946768681,
      "learning_rate": 0.00010187817854849133,
      "loss": 12.179,
      "step": 18702
    },
    {
      "epoch": 1.018453098092068,
      "grad_norm": 0.5293811185081714,
      "learning_rate": 0.00010186936183851759,
      "loss": 12.0484,
      "step": 18703
    },
    {
      "epoch": 1.018507552088651,
      "grad_norm": 0.5231718692384336,
      "learning_rate": 0.00010186054511400735,
      "loss": 11.9543,
      "step": 18704
    },
    {
      "epoch": 1.018562006085234,
      "grad_norm": 0.537399970978767,
      "learning_rate": 0.00010185172837502921,
      "loss": 12.0999,
      "step": 18705
    },
    {
      "epoch": 1.018616460081817,
      "grad_norm": 0.5760471061648441,
      "learning_rate": 0.00010184291162165172,
      "loss": 12.0845,
      "step": 18706
    },
    {
      "epoch": 1.0186709140784,
      "grad_norm": 0.5970680416849219,
      "learning_rate": 0.00010183409485394348,
      "loss": 12.0448,
      "step": 18707
    },
    {
      "epoch": 1.0187253680749833,
      "grad_norm": 0.5310220685462248,
      "learning_rate": 0.00010182527807197297,
      "loss": 12.1472,
      "step": 18708
    },
    {
      "epoch": 1.0187798220715663,
      "grad_norm": 0.4984097673992301,
      "learning_rate": 0.0001018164612758088,
      "loss": 12.0098,
      "step": 18709
    },
    {
      "epoch": 1.0188342760681492,
      "grad_norm": 0.5165077518009882,
      "learning_rate": 0.0001018076444655195,
      "loss": 12.1112,
      "step": 18710
    },
    {
      "epoch": 1.0188887300647322,
      "grad_norm": 0.48983985957564,
      "learning_rate": 0.00010179882764117368,
      "loss": 12.1085,
      "step": 18711
    },
    {
      "epoch": 1.0189431840613152,
      "grad_norm": 0.5714553058215357,
      "learning_rate": 0.00010179001080283989,
      "loss": 12.0102,
      "step": 18712
    },
    {
      "epoch": 1.0189976380578982,
      "grad_norm": 0.5127940910981188,
      "learning_rate": 0.00010178119395058665,
      "loss": 12.0016,
      "step": 18713
    },
    {
      "epoch": 1.0190520920544812,
      "grad_norm": 0.5399543294573546,
      "learning_rate": 0.00010177237708448255,
      "loss": 12.0651,
      "step": 18714
    },
    {
      "epoch": 1.0191065460510642,
      "grad_norm": 0.5470640580069777,
      "learning_rate": 0.00010176356020459617,
      "loss": 12.1846,
      "step": 18715
    },
    {
      "epoch": 1.0191610000476472,
      "grad_norm": 0.6234479197958068,
      "learning_rate": 0.000101754743310996,
      "loss": 11.93,
      "step": 18716
    },
    {
      "epoch": 1.0192154540442302,
      "grad_norm": 0.5405988025274492,
      "learning_rate": 0.00010174592640375072,
      "loss": 12.1231,
      "step": 18717
    },
    {
      "epoch": 1.0192699080408132,
      "grad_norm": 0.5078125142188712,
      "learning_rate": 0.00010173710948292878,
      "loss": 12.0405,
      "step": 18718
    },
    {
      "epoch": 1.0193243620373962,
      "grad_norm": 0.6023465117020487,
      "learning_rate": 0.00010172829254859879,
      "loss": 11.965,
      "step": 18719
    },
    {
      "epoch": 1.0193788160339794,
      "grad_norm": 0.5971529182156567,
      "learning_rate": 0.0001017194756008293,
      "loss": 12.27,
      "step": 18720
    },
    {
      "epoch": 1.0194332700305624,
      "grad_norm": 0.5185990422077906,
      "learning_rate": 0.00010171065863968889,
      "loss": 12.0221,
      "step": 18721
    },
    {
      "epoch": 1.0194877240271454,
      "grad_norm": 0.5241390327715381,
      "learning_rate": 0.00010170184166524612,
      "loss": 12.1332,
      "step": 18722
    },
    {
      "epoch": 1.0195421780237284,
      "grad_norm": 0.5010651696477758,
      "learning_rate": 0.00010169302467756953,
      "loss": 12.0712,
      "step": 18723
    },
    {
      "epoch": 1.0195966320203114,
      "grad_norm": 0.49437860750760554,
      "learning_rate": 0.0001016842076767277,
      "loss": 12.1145,
      "step": 18724
    },
    {
      "epoch": 1.0196510860168944,
      "grad_norm": 0.6015474759820658,
      "learning_rate": 0.00010167539066278919,
      "loss": 12.1264,
      "step": 18725
    },
    {
      "epoch": 1.0197055400134774,
      "grad_norm": 0.5337388508705605,
      "learning_rate": 0.00010166657363582257,
      "loss": 12.0994,
      "step": 18726
    },
    {
      "epoch": 1.0197599940100603,
      "grad_norm": 0.5894897986921385,
      "learning_rate": 0.00010165775659589639,
      "loss": 12.1271,
      "step": 18727
    },
    {
      "epoch": 1.0198144480066433,
      "grad_norm": 0.5502165812525377,
      "learning_rate": 0.00010164893954307919,
      "loss": 12.0668,
      "step": 18728
    },
    {
      "epoch": 1.0198689020032263,
      "grad_norm": 0.5448571962174019,
      "learning_rate": 0.00010164012247743959,
      "loss": 12.0644,
      "step": 18729
    },
    {
      "epoch": 1.0199233559998093,
      "grad_norm": 0.590115771224414,
      "learning_rate": 0.0001016313053990461,
      "loss": 12.1188,
      "step": 18730
    },
    {
      "epoch": 1.0199778099963923,
      "grad_norm": 0.5500449590839104,
      "learning_rate": 0.00010162248830796733,
      "loss": 12.0466,
      "step": 18731
    },
    {
      "epoch": 1.0200322639929755,
      "grad_norm": 0.5412151787779529,
      "learning_rate": 0.00010161367120427181,
      "loss": 12.033,
      "step": 18732
    },
    {
      "epoch": 1.0200867179895585,
      "grad_norm": 0.6254430498979263,
      "learning_rate": 0.00010160485408802811,
      "loss": 12.2435,
      "step": 18733
    },
    {
      "epoch": 1.0201411719861415,
      "grad_norm": 0.5336524786885203,
      "learning_rate": 0.00010159603695930479,
      "loss": 12.0075,
      "step": 18734
    },
    {
      "epoch": 1.0201956259827245,
      "grad_norm": 0.5837363686218118,
      "learning_rate": 0.00010158721981817044,
      "loss": 12.0391,
      "step": 18735
    },
    {
      "epoch": 1.0202500799793075,
      "grad_norm": 0.6079896229766106,
      "learning_rate": 0.00010157840266469359,
      "loss": 12.0356,
      "step": 18736
    },
    {
      "epoch": 1.0203045339758905,
      "grad_norm": 0.5245603346550712,
      "learning_rate": 0.0001015695854989428,
      "loss": 12.1001,
      "step": 18737
    },
    {
      "epoch": 1.0203589879724735,
      "grad_norm": 0.5537582378652263,
      "learning_rate": 0.00010156076832098666,
      "loss": 11.9315,
      "step": 18738
    },
    {
      "epoch": 1.0204134419690565,
      "grad_norm": 0.5868965655185666,
      "learning_rate": 0.00010155195113089373,
      "loss": 12.0819,
      "step": 18739
    },
    {
      "epoch": 1.0204678959656395,
      "grad_norm": 0.5751484799267724,
      "learning_rate": 0.00010154313392873257,
      "loss": 12.0688,
      "step": 18740
    },
    {
      "epoch": 1.0205223499622225,
      "grad_norm": 0.5558567163544844,
      "learning_rate": 0.00010153431671457174,
      "loss": 12.0395,
      "step": 18741
    },
    {
      "epoch": 1.0205768039588055,
      "grad_norm": 0.5417134858944368,
      "learning_rate": 0.00010152549948847982,
      "loss": 11.9737,
      "step": 18742
    },
    {
      "epoch": 1.0206312579553887,
      "grad_norm": 0.5436549049238402,
      "learning_rate": 0.0001015166822505253,
      "loss": 12.0564,
      "step": 18743
    },
    {
      "epoch": 1.0206857119519717,
      "grad_norm": 0.5613973847463412,
      "learning_rate": 0.00010150786500077687,
      "loss": 12.0953,
      "step": 18744
    },
    {
      "epoch": 1.0207401659485547,
      "grad_norm": 0.6256121230482002,
      "learning_rate": 0.00010149904773930301,
      "loss": 12.0651,
      "step": 18745
    },
    {
      "epoch": 1.0207946199451376,
      "grad_norm": 0.563324983194387,
      "learning_rate": 0.0001014902304661723,
      "loss": 11.9725,
      "step": 18746
    },
    {
      "epoch": 1.0208490739417206,
      "grad_norm": 0.5449957795585058,
      "learning_rate": 0.00010148141318145333,
      "loss": 12.0534,
      "step": 18747
    },
    {
      "epoch": 1.0209035279383036,
      "grad_norm": 0.5355158319275114,
      "learning_rate": 0.00010147259588521458,
      "loss": 12.0776,
      "step": 18748
    },
    {
      "epoch": 1.0209579819348866,
      "grad_norm": 0.5237347803738321,
      "learning_rate": 0.00010146377857752472,
      "loss": 12.1556,
      "step": 18749
    },
    {
      "epoch": 1.0210124359314696,
      "grad_norm": 0.6142868047989184,
      "learning_rate": 0.00010145496125845227,
      "loss": 12.0608,
      "step": 18750
    },
    {
      "epoch": 1.0210668899280526,
      "grad_norm": 0.5820946703122036,
      "learning_rate": 0.0001014461439280658,
      "loss": 11.9877,
      "step": 18751
    },
    {
      "epoch": 1.0211213439246356,
      "grad_norm": 0.5053531335963286,
      "learning_rate": 0.0001014373265864339,
      "loss": 11.9562,
      "step": 18752
    },
    {
      "epoch": 1.0211757979212186,
      "grad_norm": 0.6387350861190165,
      "learning_rate": 0.00010142850923362505,
      "loss": 12.2674,
      "step": 18753
    },
    {
      "epoch": 1.0212302519178016,
      "grad_norm": 0.555291955522916,
      "learning_rate": 0.0001014196918697079,
      "loss": 12.0657,
      "step": 18754
    },
    {
      "epoch": 1.0212847059143848,
      "grad_norm": 0.5306074304395978,
      "learning_rate": 0.00010141087449475098,
      "loss": 11.9807,
      "step": 18755
    },
    {
      "epoch": 1.0213391599109678,
      "grad_norm": 0.6337599413581059,
      "learning_rate": 0.00010140205710882287,
      "loss": 12.1726,
      "step": 18756
    },
    {
      "epoch": 1.0213936139075508,
      "grad_norm": 0.5208889541758751,
      "learning_rate": 0.00010139323971199216,
      "loss": 11.9897,
      "step": 18757
    },
    {
      "epoch": 1.0214480679041338,
      "grad_norm": 0.5865241119324267,
      "learning_rate": 0.00010138442230432735,
      "loss": 11.9329,
      "step": 18758
    },
    {
      "epoch": 1.0215025219007168,
      "grad_norm": 0.5366785199446498,
      "learning_rate": 0.00010137560488589701,
      "loss": 12.0962,
      "step": 18759
    },
    {
      "epoch": 1.0215569758972998,
      "grad_norm": 0.5554973337117057,
      "learning_rate": 0.00010136678745676977,
      "loss": 12.098,
      "step": 18760
    },
    {
      "epoch": 1.0216114298938828,
      "grad_norm": 0.5298973012350927,
      "learning_rate": 0.00010135797001701417,
      "loss": 12.1539,
      "step": 18761
    },
    {
      "epoch": 1.0216658838904658,
      "grad_norm": 0.589512390220508,
      "learning_rate": 0.00010134915256669878,
      "loss": 12.0498,
      "step": 18762
    },
    {
      "epoch": 1.0217203378870487,
      "grad_norm": 0.5424917531098438,
      "learning_rate": 0.00010134033510589213,
      "loss": 11.9644,
      "step": 18763
    },
    {
      "epoch": 1.0217747918836317,
      "grad_norm": 0.5703553316374649,
      "learning_rate": 0.00010133151763466282,
      "loss": 12.0504,
      "step": 18764
    },
    {
      "epoch": 1.0218292458802147,
      "grad_norm": 0.4877443637565837,
      "learning_rate": 0.00010132270015307937,
      "loss": 12.0059,
      "step": 18765
    },
    {
      "epoch": 1.0218836998767977,
      "grad_norm": 0.5437625770251306,
      "learning_rate": 0.00010131388266121041,
      "loss": 12.0672,
      "step": 18766
    },
    {
      "epoch": 1.021938153873381,
      "grad_norm": 0.5800226829041949,
      "learning_rate": 0.0001013050651591245,
      "loss": 12.0847,
      "step": 18767
    },
    {
      "epoch": 1.021992607869964,
      "grad_norm": 0.5554761728053199,
      "learning_rate": 0.00010129624764689016,
      "loss": 12.0638,
      "step": 18768
    },
    {
      "epoch": 1.022047061866547,
      "grad_norm": 0.5443212101981026,
      "learning_rate": 0.00010128743012457598,
      "loss": 11.9379,
      "step": 18769
    },
    {
      "epoch": 1.02210151586313,
      "grad_norm": 0.5293439211571511,
      "learning_rate": 0.00010127861259225053,
      "loss": 12.105,
      "step": 18770
    },
    {
      "epoch": 1.022155969859713,
      "grad_norm": 0.5387399325359294,
      "learning_rate": 0.00010126979504998235,
      "loss": 12.1094,
      "step": 18771
    },
    {
      "epoch": 1.022210423856296,
      "grad_norm": 0.5950910700409785,
      "learning_rate": 0.00010126097749784007,
      "loss": 12.0754,
      "step": 18772
    },
    {
      "epoch": 1.022264877852879,
      "grad_norm": 0.57269072448735,
      "learning_rate": 0.0001012521599358922,
      "loss": 11.9847,
      "step": 18773
    },
    {
      "epoch": 1.0223193318494619,
      "grad_norm": 0.4811092820120146,
      "learning_rate": 0.00010124334236420734,
      "loss": 11.9244,
      "step": 18774
    },
    {
      "epoch": 1.0223737858460449,
      "grad_norm": 0.5758684597581273,
      "learning_rate": 0.00010123452478285403,
      "loss": 12.1405,
      "step": 18775
    },
    {
      "epoch": 1.0224282398426279,
      "grad_norm": 0.5886928417106222,
      "learning_rate": 0.0001012257071919008,
      "loss": 11.9732,
      "step": 18776
    },
    {
      "epoch": 1.0224826938392109,
      "grad_norm": 0.5781547324240595,
      "learning_rate": 0.00010121688959141635,
      "loss": 12.1145,
      "step": 18777
    },
    {
      "epoch": 1.022537147835794,
      "grad_norm": 0.6275044757931225,
      "learning_rate": 0.00010120807198146914,
      "loss": 12.2266,
      "step": 18778
    },
    {
      "epoch": 1.022591601832377,
      "grad_norm": 0.494586053572155,
      "learning_rate": 0.00010119925436212772,
      "loss": 12.0316,
      "step": 18779
    },
    {
      "epoch": 1.02264605582896,
      "grad_norm": 0.587922190618471,
      "learning_rate": 0.00010119043673346071,
      "loss": 12.0634,
      "step": 18780
    },
    {
      "epoch": 1.022700509825543,
      "grad_norm": 0.5239084659528589,
      "learning_rate": 0.0001011816190955367,
      "loss": 12.0887,
      "step": 18781
    },
    {
      "epoch": 1.022754963822126,
      "grad_norm": 0.516350361404884,
      "learning_rate": 0.00010117280144842419,
      "loss": 12.0924,
      "step": 18782
    },
    {
      "epoch": 1.022809417818709,
      "grad_norm": 0.552523659378862,
      "learning_rate": 0.00010116398379219179,
      "loss": 12.1637,
      "step": 18783
    },
    {
      "epoch": 1.022863871815292,
      "grad_norm": 0.5703124789098535,
      "learning_rate": 0.00010115516612690805,
      "loss": 12.0717,
      "step": 18784
    },
    {
      "epoch": 1.022918325811875,
      "grad_norm": 0.5249871002893897,
      "learning_rate": 0.00010114634845264155,
      "loss": 12.1681,
      "step": 18785
    },
    {
      "epoch": 1.022972779808458,
      "grad_norm": 0.5586287033768093,
      "learning_rate": 0.00010113753076946084,
      "loss": 12.0204,
      "step": 18786
    },
    {
      "epoch": 1.023027233805041,
      "grad_norm": 0.5640718775388597,
      "learning_rate": 0.0001011287130774345,
      "loss": 12.1506,
      "step": 18787
    },
    {
      "epoch": 1.023081687801624,
      "grad_norm": 0.6064181594035376,
      "learning_rate": 0.0001011198953766311,
      "loss": 12.2754,
      "step": 18788
    },
    {
      "epoch": 1.023136141798207,
      "grad_norm": 0.6162582884740045,
      "learning_rate": 0.00010111107766711922,
      "loss": 12.0792,
      "step": 18789
    },
    {
      "epoch": 1.0231905957947902,
      "grad_norm": 0.5912703805811506,
      "learning_rate": 0.0001011022599489674,
      "loss": 12.0633,
      "step": 18790
    },
    {
      "epoch": 1.0232450497913732,
      "grad_norm": 0.5679337295473877,
      "learning_rate": 0.00010109344222224425,
      "loss": 12.0239,
      "step": 18791
    },
    {
      "epoch": 1.0232995037879562,
      "grad_norm": 0.661333811706787,
      "learning_rate": 0.00010108462448701827,
      "loss": 12.0685,
      "step": 18792
    },
    {
      "epoch": 1.0233539577845392,
      "grad_norm": 0.5829655380196324,
      "learning_rate": 0.0001010758067433581,
      "loss": 12.16,
      "step": 18793
    },
    {
      "epoch": 1.0234084117811222,
      "grad_norm": 0.5264206204636439,
      "learning_rate": 0.00010106698899133227,
      "loss": 12.02,
      "step": 18794
    },
    {
      "epoch": 1.0234628657777052,
      "grad_norm": 0.5469072976668753,
      "learning_rate": 0.00010105817123100933,
      "loss": 11.9848,
      "step": 18795
    },
    {
      "epoch": 1.0235173197742882,
      "grad_norm": 0.5762653358726504,
      "learning_rate": 0.0001010493534624579,
      "loss": 12.1183,
      "step": 18796
    },
    {
      "epoch": 1.0235717737708712,
      "grad_norm": 0.5336393034489133,
      "learning_rate": 0.0001010405356857465,
      "loss": 11.9188,
      "step": 18797
    },
    {
      "epoch": 1.0236262277674542,
      "grad_norm": 0.5976938336576505,
      "learning_rate": 0.00010103171790094375,
      "loss": 12.0602,
      "step": 18798
    },
    {
      "epoch": 1.0236806817640371,
      "grad_norm": 0.5245667416753186,
      "learning_rate": 0.00010102290010811816,
      "loss": 12.0888,
      "step": 18799
    },
    {
      "epoch": 1.0237351357606201,
      "grad_norm": 0.5591741725535495,
      "learning_rate": 0.00010101408230733833,
      "loss": 12.0078,
      "step": 18800
    },
    {
      "epoch": 1.0237895897572034,
      "grad_norm": 0.555771938556028,
      "learning_rate": 0.00010100526449867284,
      "loss": 12.2226,
      "step": 18801
    },
    {
      "epoch": 1.0238440437537863,
      "grad_norm": 0.5595947083970808,
      "learning_rate": 0.00010099644668219027,
      "loss": 12.1012,
      "step": 18802
    },
    {
      "epoch": 1.0238984977503693,
      "grad_norm": 0.604175891713645,
      "learning_rate": 0.00010098762885795909,
      "loss": 12.1059,
      "step": 18803
    },
    {
      "epoch": 1.0239529517469523,
      "grad_norm": 0.5114344969036151,
      "learning_rate": 0.00010097881102604798,
      "loss": 12.0845,
      "step": 18804
    },
    {
      "epoch": 1.0240074057435353,
      "grad_norm": 0.5097780349521573,
      "learning_rate": 0.0001009699931865255,
      "loss": 12.03,
      "step": 18805
    },
    {
      "epoch": 1.0240618597401183,
      "grad_norm": 0.530155162660626,
      "learning_rate": 0.00010096117533946014,
      "loss": 12.0368,
      "step": 18806
    },
    {
      "epoch": 1.0241163137367013,
      "grad_norm": 0.589263072886528,
      "learning_rate": 0.00010095235748492058,
      "loss": 11.9318,
      "step": 18807
    },
    {
      "epoch": 1.0241707677332843,
      "grad_norm": 0.5603419836453702,
      "learning_rate": 0.00010094353962297526,
      "loss": 12.0558,
      "step": 18808
    },
    {
      "epoch": 1.0242252217298673,
      "grad_norm": 0.5514549651704579,
      "learning_rate": 0.00010093472175369286,
      "loss": 12.0309,
      "step": 18809
    },
    {
      "epoch": 1.0242796757264503,
      "grad_norm": 0.5324272182449006,
      "learning_rate": 0.00010092590387714189,
      "loss": 12.1358,
      "step": 18810
    },
    {
      "epoch": 1.0243341297230333,
      "grad_norm": 0.5853805115201112,
      "learning_rate": 0.00010091708599339095,
      "loss": 11.9852,
      "step": 18811
    },
    {
      "epoch": 1.0243885837196163,
      "grad_norm": 0.4729042685800001,
      "learning_rate": 0.00010090826810250862,
      "loss": 11.9887,
      "step": 18812
    },
    {
      "epoch": 1.0244430377161995,
      "grad_norm": 0.6148014443636458,
      "learning_rate": 0.00010089945020456342,
      "loss": 12.0331,
      "step": 18813
    },
    {
      "epoch": 1.0244974917127825,
      "grad_norm": 0.635639106078519,
      "learning_rate": 0.0001008906322996239,
      "loss": 12.1934,
      "step": 18814
    },
    {
      "epoch": 1.0245519457093655,
      "grad_norm": 0.5633292596161977,
      "learning_rate": 0.00010088181438775873,
      "loss": 12.0455,
      "step": 18815
    },
    {
      "epoch": 1.0246063997059485,
      "grad_norm": 0.5560741591155846,
      "learning_rate": 0.00010087299646903639,
      "loss": 12.0702,
      "step": 18816
    },
    {
      "epoch": 1.0246608537025315,
      "grad_norm": 0.5651794701755701,
      "learning_rate": 0.00010086417854352552,
      "loss": 11.9584,
      "step": 18817
    },
    {
      "epoch": 1.0247153076991145,
      "grad_norm": 0.7044712598019117,
      "learning_rate": 0.00010085536061129463,
      "loss": 12.2901,
      "step": 18818
    },
    {
      "epoch": 1.0247697616956974,
      "grad_norm": 0.5552101311602395,
      "learning_rate": 0.00010084654267241231,
      "loss": 12.1134,
      "step": 18819
    },
    {
      "epoch": 1.0248242156922804,
      "grad_norm": 0.5905702158058971,
      "learning_rate": 0.00010083772472694713,
      "loss": 12.0672,
      "step": 18820
    },
    {
      "epoch": 1.0248786696888634,
      "grad_norm": 0.5573521983669069,
      "learning_rate": 0.00010082890677496766,
      "loss": 12.0905,
      "step": 18821
    },
    {
      "epoch": 1.0249331236854464,
      "grad_norm": 0.6824700731125648,
      "learning_rate": 0.0001008200888165425,
      "loss": 12.2232,
      "step": 18822
    },
    {
      "epoch": 1.0249875776820294,
      "grad_norm": 0.5409647664250958,
      "learning_rate": 0.00010081127085174019,
      "loss": 12.1013,
      "step": 18823
    },
    {
      "epoch": 1.0250420316786124,
      "grad_norm": 0.5736042445755989,
      "learning_rate": 0.00010080245288062928,
      "loss": 12.0641,
      "step": 18824
    },
    {
      "epoch": 1.0250964856751956,
      "grad_norm": 0.6756140383975451,
      "learning_rate": 0.00010079363490327833,
      "loss": 12.0934,
      "step": 18825
    },
    {
      "epoch": 1.0251509396717786,
      "grad_norm": 0.536914119748739,
      "learning_rate": 0.00010078481691975599,
      "loss": 11.9342,
      "step": 18826
    },
    {
      "epoch": 1.0252053936683616,
      "grad_norm": 0.5894137031872616,
      "learning_rate": 0.00010077599893013079,
      "loss": 12.0131,
      "step": 18827
    },
    {
      "epoch": 1.0252598476649446,
      "grad_norm": 0.5786068936267161,
      "learning_rate": 0.00010076718093447126,
      "loss": 12.047,
      "step": 18828
    },
    {
      "epoch": 1.0253143016615276,
      "grad_norm": 0.5762402343191538,
      "learning_rate": 0.00010075836293284602,
      "loss": 12.0729,
      "step": 18829
    },
    {
      "epoch": 1.0253687556581106,
      "grad_norm": 0.6076074032551696,
      "learning_rate": 0.00010074954492532362,
      "loss": 12.1409,
      "step": 18830
    },
    {
      "epoch": 1.0254232096546936,
      "grad_norm": 0.5985679684566773,
      "learning_rate": 0.0001007407269119726,
      "loss": 12.1766,
      "step": 18831
    },
    {
      "epoch": 1.0254776636512766,
      "grad_norm": 0.5925485044810468,
      "learning_rate": 0.00010073190889286164,
      "loss": 12.2808,
      "step": 18832
    },
    {
      "epoch": 1.0255321176478596,
      "grad_norm": 0.5589669324917635,
      "learning_rate": 0.00010072309086805918,
      "loss": 12.131,
      "step": 18833
    },
    {
      "epoch": 1.0255865716444426,
      "grad_norm": 0.5891691454120911,
      "learning_rate": 0.00010071427283763385,
      "loss": 12.1005,
      "step": 18834
    },
    {
      "epoch": 1.0256410256410255,
      "grad_norm": 0.5507771665067509,
      "learning_rate": 0.00010070545480165422,
      "loss": 12.1302,
      "step": 18835
    },
    {
      "epoch": 1.0256954796376085,
      "grad_norm": 0.5988299559898301,
      "learning_rate": 0.0001006966367601888,
      "loss": 12.0119,
      "step": 18836
    },
    {
      "epoch": 1.0257499336341918,
      "grad_norm": 0.6153467427206969,
      "learning_rate": 0.0001006878187133063,
      "loss": 12.1839,
      "step": 18837
    },
    {
      "epoch": 1.0258043876307747,
      "grad_norm": 0.5110678266933942,
      "learning_rate": 0.00010067900066107519,
      "loss": 12.0238,
      "step": 18838
    },
    {
      "epoch": 1.0258588416273577,
      "grad_norm": 0.5701816809723554,
      "learning_rate": 0.00010067018260356402,
      "loss": 12.0576,
      "step": 18839
    },
    {
      "epoch": 1.0259132956239407,
      "grad_norm": 0.5255673439636622,
      "learning_rate": 0.00010066136454084143,
      "loss": 11.9858,
      "step": 18840
    },
    {
      "epoch": 1.0259677496205237,
      "grad_norm": 0.5161129919900771,
      "learning_rate": 0.00010065254647297595,
      "loss": 11.9873,
      "step": 18841
    },
    {
      "epoch": 1.0260222036171067,
      "grad_norm": 0.5421377360540655,
      "learning_rate": 0.00010064372840003615,
      "loss": 12.1362,
      "step": 18842
    },
    {
      "epoch": 1.0260766576136897,
      "grad_norm": 0.5521988833133786,
      "learning_rate": 0.00010063491032209063,
      "loss": 12.0913,
      "step": 18843
    },
    {
      "epoch": 1.0261311116102727,
      "grad_norm": 0.5272891891303668,
      "learning_rate": 0.00010062609223920791,
      "loss": 12.1025,
      "step": 18844
    },
    {
      "epoch": 1.0261855656068557,
      "grad_norm": 0.5198128569718347,
      "learning_rate": 0.00010061727415145661,
      "loss": 12.0384,
      "step": 18845
    },
    {
      "epoch": 1.0262400196034387,
      "grad_norm": 0.6355999518648248,
      "learning_rate": 0.00010060845605890528,
      "loss": 12.0885,
      "step": 18846
    },
    {
      "epoch": 1.0262944736000217,
      "grad_norm": 0.5310977995229271,
      "learning_rate": 0.00010059963796162251,
      "loss": 12.136,
      "step": 18847
    },
    {
      "epoch": 1.026348927596605,
      "grad_norm": 0.5305271542221838,
      "learning_rate": 0.00010059081985967682,
      "loss": 12.1,
      "step": 18848
    },
    {
      "epoch": 1.026403381593188,
      "grad_norm": 0.5428545717662631,
      "learning_rate": 0.00010058200175313684,
      "loss": 12.0657,
      "step": 18849
    },
    {
      "epoch": 1.0264578355897709,
      "grad_norm": 0.5361102279776829,
      "learning_rate": 0.00010057318364207111,
      "loss": 12.0446,
      "step": 18850
    },
    {
      "epoch": 1.0265122895863539,
      "grad_norm": 0.5577618963549263,
      "learning_rate": 0.00010056436552654822,
      "loss": 12.1281,
      "step": 18851
    },
    {
      "epoch": 1.0265667435829369,
      "grad_norm": 0.5283476355851188,
      "learning_rate": 0.0001005555474066367,
      "loss": 12.0621,
      "step": 18852
    },
    {
      "epoch": 1.0266211975795199,
      "grad_norm": 0.6257579941479259,
      "learning_rate": 0.00010054672928240519,
      "loss": 12.1508,
      "step": 18853
    },
    {
      "epoch": 1.0266756515761029,
      "grad_norm": 0.603240107097262,
      "learning_rate": 0.0001005379111539222,
      "loss": 12.156,
      "step": 18854
    },
    {
      "epoch": 1.0267301055726858,
      "grad_norm": 0.5314831448689977,
      "learning_rate": 0.0001005290930212563,
      "loss": 12.087,
      "step": 18855
    },
    {
      "epoch": 1.0267845595692688,
      "grad_norm": 0.5783292631175788,
      "learning_rate": 0.00010052027488447612,
      "loss": 12.0822,
      "step": 18856
    },
    {
      "epoch": 1.0268390135658518,
      "grad_norm": 0.5412065536058184,
      "learning_rate": 0.00010051145674365019,
      "loss": 12.103,
      "step": 18857
    },
    {
      "epoch": 1.0268934675624348,
      "grad_norm": 0.5276334499808119,
      "learning_rate": 0.00010050263859884708,
      "loss": 12.1075,
      "step": 18858
    },
    {
      "epoch": 1.0269479215590178,
      "grad_norm": 0.5480083335352859,
      "learning_rate": 0.00010049382045013536,
      "loss": 12.0958,
      "step": 18859
    },
    {
      "epoch": 1.027002375555601,
      "grad_norm": 0.5588973912889202,
      "learning_rate": 0.00010048500229758362,
      "loss": 12.0568,
      "step": 18860
    },
    {
      "epoch": 1.027056829552184,
      "grad_norm": 0.6231732824018359,
      "learning_rate": 0.0001004761841412604,
      "loss": 12.0312,
      "step": 18861
    },
    {
      "epoch": 1.027111283548767,
      "grad_norm": 0.5221917786578817,
      "learning_rate": 0.00010046736598123432,
      "loss": 11.9448,
      "step": 18862
    },
    {
      "epoch": 1.02716573754535,
      "grad_norm": 0.585005556567255,
      "learning_rate": 0.00010045854781757392,
      "loss": 12.1332,
      "step": 18863
    },
    {
      "epoch": 1.027220191541933,
      "grad_norm": 0.5552817641655828,
      "learning_rate": 0.00010044972965034775,
      "loss": 11.9784,
      "step": 18864
    },
    {
      "epoch": 1.027274645538516,
      "grad_norm": 0.5862540149650539,
      "learning_rate": 0.00010044091147962442,
      "loss": 12.1105,
      "step": 18865
    },
    {
      "epoch": 1.027329099535099,
      "grad_norm": 0.5233113635404432,
      "learning_rate": 0.0001004320933054725,
      "loss": 12.0754,
      "step": 18866
    },
    {
      "epoch": 1.027383553531682,
      "grad_norm": 0.6066314094699042,
      "learning_rate": 0.00010042327512796055,
      "loss": 12.1417,
      "step": 18867
    },
    {
      "epoch": 1.027438007528265,
      "grad_norm": 0.5895423082417945,
      "learning_rate": 0.00010041445694715716,
      "loss": 11.9777,
      "step": 18868
    },
    {
      "epoch": 1.027492461524848,
      "grad_norm": 0.5582136229584224,
      "learning_rate": 0.00010040563876313082,
      "loss": 12.0938,
      "step": 18869
    },
    {
      "epoch": 1.027546915521431,
      "grad_norm": 0.746659076517137,
      "learning_rate": 0.00010039682057595021,
      "loss": 12.1253,
      "step": 18870
    },
    {
      "epoch": 1.0276013695180142,
      "grad_norm": 0.5547433612075412,
      "learning_rate": 0.00010038800238568384,
      "loss": 12.044,
      "step": 18871
    },
    {
      "epoch": 1.0276558235145972,
      "grad_norm": 0.6296806052435052,
      "learning_rate": 0.00010037918419240033,
      "loss": 12.1935,
      "step": 18872
    },
    {
      "epoch": 1.0277102775111802,
      "grad_norm": 0.5703231434418233,
      "learning_rate": 0.0001003703659961682,
      "loss": 12.1033,
      "step": 18873
    },
    {
      "epoch": 1.0277647315077632,
      "grad_norm": 0.6215786874718152,
      "learning_rate": 0.00010036154779705602,
      "loss": 12.0414,
      "step": 18874
    },
    {
      "epoch": 1.0278191855043461,
      "grad_norm": 0.5927630134487893,
      "learning_rate": 0.00010035272959513243,
      "loss": 12.06,
      "step": 18875
    },
    {
      "epoch": 1.0278736395009291,
      "grad_norm": 0.6420508848971314,
      "learning_rate": 0.0001003439113904659,
      "loss": 12.0871,
      "step": 18876
    },
    {
      "epoch": 1.0279280934975121,
      "grad_norm": 0.5103978927575662,
      "learning_rate": 0.00010033509318312511,
      "loss": 12.0255,
      "step": 18877
    },
    {
      "epoch": 1.0279825474940951,
      "grad_norm": 0.552259339180842,
      "learning_rate": 0.00010032627497317857,
      "loss": 12.0814,
      "step": 18878
    },
    {
      "epoch": 1.0280370014906781,
      "grad_norm": 0.5748723134775683,
      "learning_rate": 0.00010031745676069484,
      "loss": 12.0785,
      "step": 18879
    },
    {
      "epoch": 1.028091455487261,
      "grad_norm": 0.6033855987766538,
      "learning_rate": 0.0001003086385457425,
      "loss": 12.0556,
      "step": 18880
    },
    {
      "epoch": 1.028145909483844,
      "grad_norm": 0.5400001704991296,
      "learning_rate": 0.00010029982032839016,
      "loss": 12.1273,
      "step": 18881
    },
    {
      "epoch": 1.028200363480427,
      "grad_norm": 0.5946102420376946,
      "learning_rate": 0.00010029100210870636,
      "loss": 12.0104,
      "step": 18882
    },
    {
      "epoch": 1.0282548174770103,
      "grad_norm": 0.5710579850850007,
      "learning_rate": 0.0001002821838867597,
      "loss": 12.0413,
      "step": 18883
    },
    {
      "epoch": 1.0283092714735933,
      "grad_norm": 0.5396354673695882,
      "learning_rate": 0.00010027336566261871,
      "loss": 11.9725,
      "step": 18884
    },
    {
      "epoch": 1.0283637254701763,
      "grad_norm": 0.5580340497124632,
      "learning_rate": 0.00010026454743635196,
      "loss": 12.0463,
      "step": 18885
    },
    {
      "epoch": 1.0284181794667593,
      "grad_norm": 0.5317955365880455,
      "learning_rate": 0.00010025572920802808,
      "loss": 11.9057,
      "step": 18886
    },
    {
      "epoch": 1.0284726334633423,
      "grad_norm": 0.5387542121345305,
      "learning_rate": 0.00010024691097771559,
      "loss": 12.178,
      "step": 18887
    },
    {
      "epoch": 1.0285270874599253,
      "grad_norm": 0.5915490733558898,
      "learning_rate": 0.0001002380927454831,
      "loss": 12.0586,
      "step": 18888
    },
    {
      "epoch": 1.0285815414565083,
      "grad_norm": 0.5333617342298406,
      "learning_rate": 0.00010022927451139915,
      "loss": 12.0333,
      "step": 18889
    },
    {
      "epoch": 1.0286359954530913,
      "grad_norm": 0.5935844742153724,
      "learning_rate": 0.00010022045627553232,
      "loss": 12.1081,
      "step": 18890
    },
    {
      "epoch": 1.0286904494496742,
      "grad_norm": 0.6277507616477012,
      "learning_rate": 0.00010021163803795115,
      "loss": 12.0982,
      "step": 18891
    },
    {
      "epoch": 1.0287449034462572,
      "grad_norm": 0.5572110632178765,
      "learning_rate": 0.00010020281979872431,
      "loss": 12.1265,
      "step": 18892
    },
    {
      "epoch": 1.0287993574428402,
      "grad_norm": 0.5242220196092897,
      "learning_rate": 0.00010019400155792029,
      "loss": 12.0825,
      "step": 18893
    },
    {
      "epoch": 1.0288538114394232,
      "grad_norm": 0.5328799708456232,
      "learning_rate": 0.00010018518331560767,
      "loss": 11.9993,
      "step": 18894
    },
    {
      "epoch": 1.0289082654360064,
      "grad_norm": 0.5325890831838296,
      "learning_rate": 0.00010017636507185504,
      "loss": 12.1233,
      "step": 18895
    },
    {
      "epoch": 1.0289627194325894,
      "grad_norm": 0.5734016280637351,
      "learning_rate": 0.00010016754682673096,
      "loss": 11.9025,
      "step": 18896
    },
    {
      "epoch": 1.0290171734291724,
      "grad_norm": 0.6176216690609465,
      "learning_rate": 0.00010015872858030397,
      "loss": 12.2049,
      "step": 18897
    },
    {
      "epoch": 1.0290716274257554,
      "grad_norm": 0.6092051065623323,
      "learning_rate": 0.00010014991033264274,
      "loss": 12.2688,
      "step": 18898
    },
    {
      "epoch": 1.0291260814223384,
      "grad_norm": 0.508883902672158,
      "learning_rate": 0.00010014109208381577,
      "loss": 12.005,
      "step": 18899
    },
    {
      "epoch": 1.0291805354189214,
      "grad_norm": 0.5493388982078555,
      "learning_rate": 0.00010013227383389163,
      "loss": 12.0163,
      "step": 18900
    },
    {
      "epoch": 1.0292349894155044,
      "grad_norm": 0.5805459862865998,
      "learning_rate": 0.00010012345558293892,
      "loss": 12.0681,
      "step": 18901
    },
    {
      "epoch": 1.0292894434120874,
      "grad_norm": 0.5456664730677401,
      "learning_rate": 0.00010011463733102615,
      "loss": 11.9971,
      "step": 18902
    },
    {
      "epoch": 1.0293438974086704,
      "grad_norm": 0.5602136876295947,
      "learning_rate": 0.00010010581907822201,
      "loss": 12.0039,
      "step": 18903
    },
    {
      "epoch": 1.0293983514052534,
      "grad_norm": 0.5162878149882081,
      "learning_rate": 0.00010009700082459496,
      "loss": 12.0213,
      "step": 18904
    },
    {
      "epoch": 1.0294528054018364,
      "grad_norm": 0.5679069808387381,
      "learning_rate": 0.00010008818257021363,
      "loss": 12.2285,
      "step": 18905
    },
    {
      "epoch": 1.0295072593984194,
      "grad_norm": 0.5750738974008015,
      "learning_rate": 0.0001000793643151466,
      "loss": 12.1621,
      "step": 18906
    },
    {
      "epoch": 1.0295617133950026,
      "grad_norm": 0.5284897092297695,
      "learning_rate": 0.00010007054605946237,
      "loss": 11.9923,
      "step": 18907
    },
    {
      "epoch": 1.0296161673915856,
      "grad_norm": 0.5325368772512918,
      "learning_rate": 0.0001000617278032296,
      "loss": 12.012,
      "step": 18908
    },
    {
      "epoch": 1.0296706213881686,
      "grad_norm": 0.6267462450949988,
      "learning_rate": 0.00010005290954651681,
      "loss": 11.9871,
      "step": 18909
    },
    {
      "epoch": 1.0297250753847516,
      "grad_norm": 0.632236511368181,
      "learning_rate": 0.00010004409128939258,
      "loss": 12.2269,
      "step": 18910
    },
    {
      "epoch": 1.0297795293813345,
      "grad_norm": 0.6192054578441146,
      "learning_rate": 0.0001000352730319255,
      "loss": 12.0524,
      "step": 18911
    },
    {
      "epoch": 1.0298339833779175,
      "grad_norm": 0.5384892591463919,
      "learning_rate": 0.00010002645477418413,
      "loss": 11.9744,
      "step": 18912
    },
    {
      "epoch": 1.0298884373745005,
      "grad_norm": 0.5472096567294882,
      "learning_rate": 0.00010001763651623706,
      "loss": 11.8775,
      "step": 18913
    },
    {
      "epoch": 1.0299428913710835,
      "grad_norm": 0.5666239111993764,
      "learning_rate": 0.00010000881825815283,
      "loss": 11.9846,
      "step": 18914
    },
    {
      "epoch": 1.0299973453676665,
      "grad_norm": 0.5516408627753739,
      "learning_rate": 0.0001,
      "loss": 12.0715,
      "step": 18915
    },
    {
      "epoch": 1.0300517993642495,
      "grad_norm": 0.5649090063726684,
      "learning_rate": 9.999118174184721e-05,
      "loss": 12.0079,
      "step": 18916
    },
    {
      "epoch": 1.0301062533608325,
      "grad_norm": 0.6835071626704871,
      "learning_rate": 9.998236348376298e-05,
      "loss": 12.1264,
      "step": 18917
    },
    {
      "epoch": 1.0301607073574157,
      "grad_norm": 0.6434222873091657,
      "learning_rate": 9.997354522581588e-05,
      "loss": 11.988,
      "step": 18918
    },
    {
      "epoch": 1.0302151613539987,
      "grad_norm": 0.52579603385314,
      "learning_rate": 9.996472696807449e-05,
      "loss": 12.023,
      "step": 18919
    },
    {
      "epoch": 1.0302696153505817,
      "grad_norm": 0.5374831986440027,
      "learning_rate": 9.99559087106074e-05,
      "loss": 11.9577,
      "step": 18920
    },
    {
      "epoch": 1.0303240693471647,
      "grad_norm": 0.6623159749593205,
      "learning_rate": 9.994709045348323e-05,
      "loss": 12.0303,
      "step": 18921
    },
    {
      "epoch": 1.0303785233437477,
      "grad_norm": 0.5375802379687675,
      "learning_rate": 9.993827219677044e-05,
      "loss": 12.0215,
      "step": 18922
    },
    {
      "epoch": 1.0304329773403307,
      "grad_norm": 0.5363274067085633,
      "learning_rate": 9.992945394053764e-05,
      "loss": 12.1377,
      "step": 18923
    },
    {
      "epoch": 1.0304874313369137,
      "grad_norm": 0.5387464707574885,
      "learning_rate": 9.992063568485344e-05,
      "loss": 12.1371,
      "step": 18924
    },
    {
      "epoch": 1.0305418853334967,
      "grad_norm": 0.5193837127488434,
      "learning_rate": 9.991181742978638e-05,
      "loss": 12.1071,
      "step": 18925
    },
    {
      "epoch": 1.0305963393300797,
      "grad_norm": 0.5706090335186389,
      "learning_rate": 9.990299917540506e-05,
      "loss": 12.0525,
      "step": 18926
    },
    {
      "epoch": 1.0306507933266627,
      "grad_norm": 0.5737329560214532,
      "learning_rate": 9.989418092177801e-05,
      "loss": 12.0354,
      "step": 18927
    },
    {
      "epoch": 1.0307052473232456,
      "grad_norm": 0.5472470873820268,
      "learning_rate": 9.988536266897384e-05,
      "loss": 11.9458,
      "step": 18928
    },
    {
      "epoch": 1.0307597013198286,
      "grad_norm": 0.5789577964659437,
      "learning_rate": 9.987654441706112e-05,
      "loss": 12.2022,
      "step": 18929
    },
    {
      "epoch": 1.0308141553164119,
      "grad_norm": 0.6027020474997213,
      "learning_rate": 9.986772616610838e-05,
      "loss": 12.0335,
      "step": 18930
    },
    {
      "epoch": 1.0308686093129948,
      "grad_norm": 0.559245253812309,
      "learning_rate": 9.985890791618428e-05,
      "loss": 11.9799,
      "step": 18931
    },
    {
      "epoch": 1.0309230633095778,
      "grad_norm": 0.5994855825821289,
      "learning_rate": 9.985008966735731e-05,
      "loss": 12.1266,
      "step": 18932
    },
    {
      "epoch": 1.0309775173061608,
      "grad_norm": 0.5604127136129978,
      "learning_rate": 9.984127141969605e-05,
      "loss": 12.0725,
      "step": 18933
    },
    {
      "epoch": 1.0310319713027438,
      "grad_norm": 0.584477924604607,
      "learning_rate": 9.983245317326907e-05,
      "loss": 12.0621,
      "step": 18934
    },
    {
      "epoch": 1.0310864252993268,
      "grad_norm": 0.543314089318705,
      "learning_rate": 9.9823634928145e-05,
      "loss": 12.0443,
      "step": 18935
    },
    {
      "epoch": 1.0311408792959098,
      "grad_norm": 0.6018410254304223,
      "learning_rate": 9.981481668439236e-05,
      "loss": 12.1504,
      "step": 18936
    },
    {
      "epoch": 1.0311953332924928,
      "grad_norm": 0.6363326342899834,
      "learning_rate": 9.980599844207974e-05,
      "loss": 11.9964,
      "step": 18937
    },
    {
      "epoch": 1.0312497872890758,
      "grad_norm": 0.5265196045905317,
      "learning_rate": 9.97971802012757e-05,
      "loss": 12.1074,
      "step": 18938
    },
    {
      "epoch": 1.0313042412856588,
      "grad_norm": 0.577653919376557,
      "learning_rate": 9.978836196204884e-05,
      "loss": 12.0779,
      "step": 18939
    },
    {
      "epoch": 1.0313586952822418,
      "grad_norm": 0.6271410352203819,
      "learning_rate": 9.97795437244677e-05,
      "loss": 12.024,
      "step": 18940
    },
    {
      "epoch": 1.031413149278825,
      "grad_norm": 0.545593795326789,
      "learning_rate": 9.977072548860089e-05,
      "loss": 12.2078,
      "step": 18941
    },
    {
      "epoch": 1.031467603275408,
      "grad_norm": 0.7223263909651635,
      "learning_rate": 9.976190725451694e-05,
      "loss": 12.1842,
      "step": 18942
    },
    {
      "epoch": 1.031522057271991,
      "grad_norm": 0.5921282937729481,
      "learning_rate": 9.975308902228444e-05,
      "loss": 11.9853,
      "step": 18943
    },
    {
      "epoch": 1.031576511268574,
      "grad_norm": 0.5496087592341936,
      "learning_rate": 9.974427079197195e-05,
      "loss": 12.0496,
      "step": 18944
    },
    {
      "epoch": 1.031630965265157,
      "grad_norm": 0.5592203365582061,
      "learning_rate": 9.973545256364806e-05,
      "loss": 12.1286,
      "step": 18945
    },
    {
      "epoch": 1.03168541926174,
      "grad_norm": 0.4917921990182803,
      "learning_rate": 9.972663433738132e-05,
      "loss": 12.021,
      "step": 18946
    },
    {
      "epoch": 1.031739873258323,
      "grad_norm": 0.6193108216780243,
      "learning_rate": 9.971781611324031e-05,
      "loss": 12.0472,
      "step": 18947
    },
    {
      "epoch": 1.031794327254906,
      "grad_norm": 0.6255273614029856,
      "learning_rate": 9.970899789129365e-05,
      "loss": 12.0162,
      "step": 18948
    },
    {
      "epoch": 1.031848781251489,
      "grad_norm": 0.5431678806221594,
      "learning_rate": 9.970017967160984e-05,
      "loss": 12.0859,
      "step": 18949
    },
    {
      "epoch": 1.031903235248072,
      "grad_norm": 0.5535537159302271,
      "learning_rate": 9.96913614542575e-05,
      "loss": 12.0325,
      "step": 18950
    },
    {
      "epoch": 1.031957689244655,
      "grad_norm": 0.5821790594525964,
      "learning_rate": 9.968254323930521e-05,
      "loss": 12.0404,
      "step": 18951
    },
    {
      "epoch": 1.032012143241238,
      "grad_norm": 0.5588029126194511,
      "learning_rate": 9.967372502682148e-05,
      "loss": 12.0284,
      "step": 18952
    },
    {
      "epoch": 1.0320665972378211,
      "grad_norm": 0.5647439710683829,
      "learning_rate": 9.966490681687493e-05,
      "loss": 12.0481,
      "step": 18953
    },
    {
      "epoch": 1.0321210512344041,
      "grad_norm": 0.5671428492368455,
      "learning_rate": 9.965608860953411e-05,
      "loss": 12.0792,
      "step": 18954
    },
    {
      "epoch": 1.0321755052309871,
      "grad_norm": 0.7456889785914095,
      "learning_rate": 9.964727040486761e-05,
      "loss": 12.1077,
      "step": 18955
    },
    {
      "epoch": 1.03222995922757,
      "grad_norm": 0.6149926470951985,
      "learning_rate": 9.963845220294399e-05,
      "loss": 12.1788,
      "step": 18956
    },
    {
      "epoch": 1.032284413224153,
      "grad_norm": 0.5729471131619598,
      "learning_rate": 9.96296340038318e-05,
      "loss": 12.0537,
      "step": 18957
    },
    {
      "epoch": 1.032338867220736,
      "grad_norm": 0.5583344837668531,
      "learning_rate": 9.962081580759968e-05,
      "loss": 12.2213,
      "step": 18958
    },
    {
      "epoch": 1.032393321217319,
      "grad_norm": 0.5723927892301177,
      "learning_rate": 9.961199761431614e-05,
      "loss": 12.0922,
      "step": 18959
    },
    {
      "epoch": 1.032447775213902,
      "grad_norm": 0.5593927752480166,
      "learning_rate": 9.960317942404978e-05,
      "loss": 12.1068,
      "step": 18960
    },
    {
      "epoch": 1.032502229210485,
      "grad_norm": 0.5554236215804222,
      "learning_rate": 9.959436123686919e-05,
      "loss": 12.114,
      "step": 18961
    },
    {
      "epoch": 1.032556683207068,
      "grad_norm": 0.5022972277314655,
      "learning_rate": 9.958554305284289e-05,
      "loss": 12.1123,
      "step": 18962
    },
    {
      "epoch": 1.032611137203651,
      "grad_norm": 0.5686257319727192,
      "learning_rate": 9.957672487203948e-05,
      "loss": 12.0303,
      "step": 18963
    },
    {
      "epoch": 1.032665591200234,
      "grad_norm": 0.5288932357435517,
      "learning_rate": 9.956790669452751e-05,
      "loss": 12.1181,
      "step": 18964
    },
    {
      "epoch": 1.0327200451968173,
      "grad_norm": 0.5488837938781982,
      "learning_rate": 9.955908852037559e-05,
      "loss": 12.1645,
      "step": 18965
    },
    {
      "epoch": 1.0327744991934003,
      "grad_norm": 0.5616232895228722,
      "learning_rate": 9.955027034965226e-05,
      "loss": 12.1661,
      "step": 18966
    },
    {
      "epoch": 1.0328289531899832,
      "grad_norm": 0.7209270444507411,
      "learning_rate": 9.954145218242611e-05,
      "loss": 12.1788,
      "step": 18967
    },
    {
      "epoch": 1.0328834071865662,
      "grad_norm": 0.5679915308447578,
      "learning_rate": 9.953263401876568e-05,
      "loss": 11.8683,
      "step": 18968
    },
    {
      "epoch": 1.0329378611831492,
      "grad_norm": 0.5261119171360865,
      "learning_rate": 9.952381585873959e-05,
      "loss": 12.0068,
      "step": 18969
    },
    {
      "epoch": 1.0329923151797322,
      "grad_norm": 0.5270703981807149,
      "learning_rate": 9.95149977024164e-05,
      "loss": 12.1285,
      "step": 18970
    },
    {
      "epoch": 1.0330467691763152,
      "grad_norm": 0.5303365022994876,
      "learning_rate": 9.950617954986469e-05,
      "loss": 12.0673,
      "step": 18971
    },
    {
      "epoch": 1.0331012231728982,
      "grad_norm": 0.5403425930205259,
      "learning_rate": 9.949736140115295e-05,
      "loss": 12.0178,
      "step": 18972
    },
    {
      "epoch": 1.0331556771694812,
      "grad_norm": 0.5136572749686279,
      "learning_rate": 9.948854325634984e-05,
      "loss": 12.0827,
      "step": 18973
    },
    {
      "epoch": 1.0332101311660642,
      "grad_norm": 0.5512873047640114,
      "learning_rate": 9.94797251155239e-05,
      "loss": 11.9257,
      "step": 18974
    },
    {
      "epoch": 1.0332645851626472,
      "grad_norm": 0.5574323799652663,
      "learning_rate": 9.94709069787437e-05,
      "loss": 12.092,
      "step": 18975
    },
    {
      "epoch": 1.0333190391592302,
      "grad_norm": 0.576200540979803,
      "learning_rate": 9.946208884607782e-05,
      "loss": 12.2042,
      "step": 18976
    },
    {
      "epoch": 1.0333734931558134,
      "grad_norm": 0.6078327664055824,
      "learning_rate": 9.945327071759484e-05,
      "loss": 12.1118,
      "step": 18977
    },
    {
      "epoch": 1.0334279471523964,
      "grad_norm": 0.5972974214912442,
      "learning_rate": 9.944445259336332e-05,
      "loss": 12.0486,
      "step": 18978
    },
    {
      "epoch": 1.0334824011489794,
      "grad_norm": 0.5406042584852266,
      "learning_rate": 9.943563447345179e-05,
      "loss": 12.1173,
      "step": 18979
    },
    {
      "epoch": 1.0335368551455624,
      "grad_norm": 0.5917761282289588,
      "learning_rate": 9.942681635792888e-05,
      "loss": 12.2366,
      "step": 18980
    },
    {
      "epoch": 1.0335913091421454,
      "grad_norm": 0.5517644699994737,
      "learning_rate": 9.94179982468632e-05,
      "loss": 12.1436,
      "step": 18981
    },
    {
      "epoch": 1.0336457631387284,
      "grad_norm": 0.6317805706296912,
      "learning_rate": 9.940918014032323e-05,
      "loss": 12.1843,
      "step": 18982
    },
    {
      "epoch": 1.0337002171353113,
      "grad_norm": 0.5568818214034049,
      "learning_rate": 9.940036203837753e-05,
      "loss": 12.0803,
      "step": 18983
    },
    {
      "epoch": 1.0337546711318943,
      "grad_norm": 0.5646030978093064,
      "learning_rate": 9.939154394109473e-05,
      "loss": 12.0041,
      "step": 18984
    },
    {
      "epoch": 1.0338091251284773,
      "grad_norm": 0.5578569716580203,
      "learning_rate": 9.938272584854341e-05,
      "loss": 12.1064,
      "step": 18985
    },
    {
      "epoch": 1.0338635791250603,
      "grad_norm": 0.4875324723148824,
      "learning_rate": 9.937390776079211e-05,
      "loss": 11.9227,
      "step": 18986
    },
    {
      "epoch": 1.0339180331216433,
      "grad_norm": 0.5457733707355378,
      "learning_rate": 9.936508967790941e-05,
      "loss": 12.1069,
      "step": 18987
    },
    {
      "epoch": 1.0339724871182265,
      "grad_norm": 0.5743029530179137,
      "learning_rate": 9.935627159996386e-05,
      "loss": 12.0204,
      "step": 18988
    },
    {
      "epoch": 1.0340269411148095,
      "grad_norm": 0.5679480383613924,
      "learning_rate": 9.934745352702406e-05,
      "loss": 12.0522,
      "step": 18989
    },
    {
      "epoch": 1.0340813951113925,
      "grad_norm": 0.6754449658587147,
      "learning_rate": 9.933863545915856e-05,
      "loss": 12.2621,
      "step": 18990
    },
    {
      "epoch": 1.0341358491079755,
      "grad_norm": 0.5515893171453582,
      "learning_rate": 9.9329817396436e-05,
      "loss": 12.139,
      "step": 18991
    },
    {
      "epoch": 1.0341903031045585,
      "grad_norm": 0.544087538047309,
      "learning_rate": 9.932099933892486e-05,
      "loss": 11.9348,
      "step": 18992
    },
    {
      "epoch": 1.0342447571011415,
      "grad_norm": 0.5359449039020331,
      "learning_rate": 9.931218128669373e-05,
      "loss": 12.0435,
      "step": 18993
    },
    {
      "epoch": 1.0342992110977245,
      "grad_norm": 0.5110707846928103,
      "learning_rate": 9.93033632398112e-05,
      "loss": 12.0414,
      "step": 18994
    },
    {
      "epoch": 1.0343536650943075,
      "grad_norm": 0.4865654922492864,
      "learning_rate": 9.929454519834581e-05,
      "loss": 12.0509,
      "step": 18995
    },
    {
      "epoch": 1.0344081190908905,
      "grad_norm": 0.5841905001779324,
      "learning_rate": 9.928572716236616e-05,
      "loss": 12.0051,
      "step": 18996
    },
    {
      "epoch": 1.0344625730874735,
      "grad_norm": 0.5291241370961548,
      "learning_rate": 9.927690913194085e-05,
      "loss": 12.0463,
      "step": 18997
    },
    {
      "epoch": 1.0345170270840565,
      "grad_norm": 0.5640243804897475,
      "learning_rate": 9.92680911071384e-05,
      "loss": 11.8798,
      "step": 18998
    },
    {
      "epoch": 1.0345714810806395,
      "grad_norm": 0.542482064559819,
      "learning_rate": 9.925927308802738e-05,
      "loss": 12.117,
      "step": 18999
    },
    {
      "epoch": 1.0346259350772227,
      "grad_norm": 0.5547007371518616,
      "learning_rate": 9.92504550746764e-05,
      "loss": 12.1063,
      "step": 19000
    },
    {
      "epoch": 1.0346803890738057,
      "grad_norm": 0.5291276435062658,
      "learning_rate": 9.924163706715402e-05,
      "loss": 12.0121,
      "step": 19001
    },
    {
      "epoch": 1.0347348430703887,
      "grad_norm": 0.54012404073616,
      "learning_rate": 9.923281906552877e-05,
      "loss": 12.0397,
      "step": 19002
    },
    {
      "epoch": 1.0347892970669716,
      "grad_norm": 0.562145421207624,
      "learning_rate": 9.922400106986926e-05,
      "loss": 12.1357,
      "step": 19003
    },
    {
      "epoch": 1.0348437510635546,
      "grad_norm": 0.5854892169410031,
      "learning_rate": 9.921518308024403e-05,
      "loss": 12.1699,
      "step": 19004
    },
    {
      "epoch": 1.0348982050601376,
      "grad_norm": 0.5723611719763151,
      "learning_rate": 9.920636509672168e-05,
      "loss": 12.0045,
      "step": 19005
    },
    {
      "epoch": 1.0349526590567206,
      "grad_norm": 0.5125566985009139,
      "learning_rate": 9.919754711937074e-05,
      "loss": 12.0007,
      "step": 19006
    },
    {
      "epoch": 1.0350071130533036,
      "grad_norm": 0.500166941294284,
      "learning_rate": 9.918872914825984e-05,
      "loss": 11.9961,
      "step": 19007
    },
    {
      "epoch": 1.0350615670498866,
      "grad_norm": 0.5550410627066128,
      "learning_rate": 9.917991118345751e-05,
      "loss": 11.9788,
      "step": 19008
    },
    {
      "epoch": 1.0351160210464696,
      "grad_norm": 0.5116309465811975,
      "learning_rate": 9.917109322503232e-05,
      "loss": 12.0869,
      "step": 19009
    },
    {
      "epoch": 1.0351704750430526,
      "grad_norm": 0.5259939680912251,
      "learning_rate": 9.916227527305287e-05,
      "loss": 12.066,
      "step": 19010
    },
    {
      "epoch": 1.0352249290396358,
      "grad_norm": 0.5247178368056492,
      "learning_rate": 9.915345732758772e-05,
      "loss": 12.0876,
      "step": 19011
    },
    {
      "epoch": 1.0352793830362188,
      "grad_norm": 0.5336971277089876,
      "learning_rate": 9.91446393887054e-05,
      "loss": 12.0686,
      "step": 19012
    },
    {
      "epoch": 1.0353338370328018,
      "grad_norm": 0.5337462776018592,
      "learning_rate": 9.913582145647452e-05,
      "loss": 11.9633,
      "step": 19013
    },
    {
      "epoch": 1.0353882910293848,
      "grad_norm": 0.5420632483238385,
      "learning_rate": 9.912700353096362e-05,
      "loss": 12.1343,
      "step": 19014
    },
    {
      "epoch": 1.0354427450259678,
      "grad_norm": 0.5160256288543138,
      "learning_rate": 9.91181856122413e-05,
      "loss": 12.128,
      "step": 19015
    },
    {
      "epoch": 1.0354971990225508,
      "grad_norm": 0.478344528213043,
      "learning_rate": 9.91093677003761e-05,
      "loss": 12.0558,
      "step": 19016
    },
    {
      "epoch": 1.0355516530191338,
      "grad_norm": 0.4993093672597959,
      "learning_rate": 9.910054979543662e-05,
      "loss": 12.1166,
      "step": 19017
    },
    {
      "epoch": 1.0356061070157168,
      "grad_norm": 0.6508014470582698,
      "learning_rate": 9.90917318974914e-05,
      "loss": 12.1159,
      "step": 19018
    },
    {
      "epoch": 1.0356605610122998,
      "grad_norm": 0.532951712132628,
      "learning_rate": 9.908291400660906e-05,
      "loss": 12.1033,
      "step": 19019
    },
    {
      "epoch": 1.0357150150088827,
      "grad_norm": 0.5157335528993278,
      "learning_rate": 9.90740961228581e-05,
      "loss": 11.9439,
      "step": 19020
    },
    {
      "epoch": 1.0357694690054657,
      "grad_norm": 0.5211195795040291,
      "learning_rate": 9.906527824630715e-05,
      "loss": 12.0271,
      "step": 19021
    },
    {
      "epoch": 1.0358239230020487,
      "grad_norm": 0.4964343356487457,
      "learning_rate": 9.905646037702475e-05,
      "loss": 12.0784,
      "step": 19022
    },
    {
      "epoch": 1.035878376998632,
      "grad_norm": 0.5121292915204849,
      "learning_rate": 9.904764251507946e-05,
      "loss": 11.8588,
      "step": 19023
    },
    {
      "epoch": 1.035932830995215,
      "grad_norm": 0.5395009713858713,
      "learning_rate": 9.903882466053987e-05,
      "loss": 12.078,
      "step": 19024
    },
    {
      "epoch": 1.035987284991798,
      "grad_norm": 0.5724794808973698,
      "learning_rate": 9.903000681347453e-05,
      "loss": 12.1624,
      "step": 19025
    },
    {
      "epoch": 1.036041738988381,
      "grad_norm": 0.528610169588521,
      "learning_rate": 9.902118897395203e-05,
      "loss": 12.1195,
      "step": 19026
    },
    {
      "epoch": 1.036096192984964,
      "grad_norm": 0.5874990058259184,
      "learning_rate": 9.901237114204092e-05,
      "loss": 12.0297,
      "step": 19027
    },
    {
      "epoch": 1.036150646981547,
      "grad_norm": 0.5577128810521385,
      "learning_rate": 9.900355331780976e-05,
      "loss": 12.1688,
      "step": 19028
    },
    {
      "epoch": 1.03620510097813,
      "grad_norm": 0.5359707995224262,
      "learning_rate": 9.899473550132717e-05,
      "loss": 12.0934,
      "step": 19029
    },
    {
      "epoch": 1.036259554974713,
      "grad_norm": 0.5320585447586443,
      "learning_rate": 9.898591769266166e-05,
      "loss": 12.0178,
      "step": 19030
    },
    {
      "epoch": 1.0363140089712959,
      "grad_norm": 0.5996530740109022,
      "learning_rate": 9.897709989188189e-05,
      "loss": 12.1333,
      "step": 19031
    },
    {
      "epoch": 1.0363684629678789,
      "grad_norm": 0.530592100026717,
      "learning_rate": 9.896828209905629e-05,
      "loss": 11.8981,
      "step": 19032
    },
    {
      "epoch": 1.0364229169644619,
      "grad_norm": 0.5071047991797685,
      "learning_rate": 9.895946431425351e-05,
      "loss": 12.0599,
      "step": 19033
    },
    {
      "epoch": 1.036477370961045,
      "grad_norm": 0.5045948386650949,
      "learning_rate": 9.895064653754212e-05,
      "loss": 12.0018,
      "step": 19034
    },
    {
      "epoch": 1.036531824957628,
      "grad_norm": 0.5357652354755498,
      "learning_rate": 9.894182876899069e-05,
      "loss": 12.106,
      "step": 19035
    },
    {
      "epoch": 1.036586278954211,
      "grad_norm": 0.49056228823431847,
      "learning_rate": 9.893301100866777e-05,
      "loss": 12.0361,
      "step": 19036
    },
    {
      "epoch": 1.036640732950794,
      "grad_norm": 0.514162932443665,
      "learning_rate": 9.892419325664193e-05,
      "loss": 11.9886,
      "step": 19037
    },
    {
      "epoch": 1.036695186947377,
      "grad_norm": 0.48106338028908796,
      "learning_rate": 9.891537551298175e-05,
      "loss": 12.0299,
      "step": 19038
    },
    {
      "epoch": 1.03674964094396,
      "grad_norm": 0.5403447290973654,
      "learning_rate": 9.890655777775576e-05,
      "loss": 11.9933,
      "step": 19039
    },
    {
      "epoch": 1.036804094940543,
      "grad_norm": 0.5711878205368841,
      "learning_rate": 9.889774005103258e-05,
      "loss": 12.0667,
      "step": 19040
    },
    {
      "epoch": 1.036858548937126,
      "grad_norm": 0.5227090801811834,
      "learning_rate": 9.888892233288081e-05,
      "loss": 11.9956,
      "step": 19041
    },
    {
      "epoch": 1.036913002933709,
      "grad_norm": 0.5509038475517687,
      "learning_rate": 9.888010462336893e-05,
      "loss": 12.0979,
      "step": 19042
    },
    {
      "epoch": 1.036967456930292,
      "grad_norm": 0.5564037998168841,
      "learning_rate": 9.887128692256554e-05,
      "loss": 12.0903,
      "step": 19043
    },
    {
      "epoch": 1.037021910926875,
      "grad_norm": 0.5489284125232368,
      "learning_rate": 9.886246923053918e-05,
      "loss": 12.124,
      "step": 19044
    },
    {
      "epoch": 1.037076364923458,
      "grad_norm": 0.48695931999911213,
      "learning_rate": 9.885365154735849e-05,
      "loss": 11.9784,
      "step": 19045
    },
    {
      "epoch": 1.0371308189200412,
      "grad_norm": 0.5119116175650932,
      "learning_rate": 9.884483387309197e-05,
      "loss": 12.0831,
      "step": 19046
    },
    {
      "epoch": 1.0371852729166242,
      "grad_norm": 0.5402343458051497,
      "learning_rate": 9.883601620780825e-05,
      "loss": 12.0994,
      "step": 19047
    },
    {
      "epoch": 1.0372397269132072,
      "grad_norm": 0.5443656623920647,
      "learning_rate": 9.882719855157584e-05,
      "loss": 12.0989,
      "step": 19048
    },
    {
      "epoch": 1.0372941809097902,
      "grad_norm": 0.5613854849086186,
      "learning_rate": 9.881838090446334e-05,
      "loss": 11.9348,
      "step": 19049
    },
    {
      "epoch": 1.0373486349063732,
      "grad_norm": 0.5716948056406087,
      "learning_rate": 9.88095632665393e-05,
      "loss": 11.9324,
      "step": 19050
    },
    {
      "epoch": 1.0374030889029562,
      "grad_norm": 0.5615163943071505,
      "learning_rate": 9.880074563787232e-05,
      "loss": 12.079,
      "step": 19051
    },
    {
      "epoch": 1.0374575428995392,
      "grad_norm": 0.5416536736159172,
      "learning_rate": 9.879192801853093e-05,
      "loss": 11.9152,
      "step": 19052
    },
    {
      "epoch": 1.0375119968961222,
      "grad_norm": 0.5257042329245045,
      "learning_rate": 9.87831104085837e-05,
      "loss": 11.9454,
      "step": 19053
    },
    {
      "epoch": 1.0375664508927052,
      "grad_norm": 0.5153521266220805,
      "learning_rate": 9.87742928080992e-05,
      "loss": 12.022,
      "step": 19054
    },
    {
      "epoch": 1.0376209048892882,
      "grad_norm": 0.532221667724164,
      "learning_rate": 9.8765475217146e-05,
      "loss": 12.061,
      "step": 19055
    },
    {
      "epoch": 1.0376753588858711,
      "grad_norm": 0.5538100715675632,
      "learning_rate": 9.875665763579269e-05,
      "loss": 12.1585,
      "step": 19056
    },
    {
      "epoch": 1.0377298128824541,
      "grad_norm": 0.5054629450823357,
      "learning_rate": 9.874784006410782e-05,
      "loss": 11.7964,
      "step": 19057
    },
    {
      "epoch": 1.0377842668790374,
      "grad_norm": 0.5457479810413797,
      "learning_rate": 9.873902250215994e-05,
      "loss": 11.8406,
      "step": 19058
    },
    {
      "epoch": 1.0378387208756203,
      "grad_norm": 0.5415794466342914,
      "learning_rate": 9.873020495001765e-05,
      "loss": 12.1593,
      "step": 19059
    },
    {
      "epoch": 1.0378931748722033,
      "grad_norm": 0.5725519633500826,
      "learning_rate": 9.87213874077495e-05,
      "loss": 12.055,
      "step": 19060
    },
    {
      "epoch": 1.0379476288687863,
      "grad_norm": 0.5763961304813658,
      "learning_rate": 9.871256987542404e-05,
      "loss": 12.1032,
      "step": 19061
    },
    {
      "epoch": 1.0380020828653693,
      "grad_norm": 0.5393060127149656,
      "learning_rate": 9.870375235310989e-05,
      "loss": 11.9136,
      "step": 19062
    },
    {
      "epoch": 1.0380565368619523,
      "grad_norm": 0.5573715002208469,
      "learning_rate": 9.869493484087556e-05,
      "loss": 12.0489,
      "step": 19063
    },
    {
      "epoch": 1.0381109908585353,
      "grad_norm": 0.5628184000735703,
      "learning_rate": 9.868611733878961e-05,
      "loss": 12.0815,
      "step": 19064
    },
    {
      "epoch": 1.0381654448551183,
      "grad_norm": 0.6216920499794681,
      "learning_rate": 9.867729984692065e-05,
      "loss": 12.0082,
      "step": 19065
    },
    {
      "epoch": 1.0382198988517013,
      "grad_norm": 0.5540984827577488,
      "learning_rate": 9.86684823653372e-05,
      "loss": 12.0654,
      "step": 19066
    },
    {
      "epoch": 1.0382743528482843,
      "grad_norm": 0.5134707959928382,
      "learning_rate": 9.865966489410789e-05,
      "loss": 12.0314,
      "step": 19067
    },
    {
      "epoch": 1.0383288068448673,
      "grad_norm": 0.5443577726995507,
      "learning_rate": 9.865084743330124e-05,
      "loss": 12.0544,
      "step": 19068
    },
    {
      "epoch": 1.0383832608414503,
      "grad_norm": 0.5897961064460123,
      "learning_rate": 9.864202998298583e-05,
      "loss": 12.0468,
      "step": 19069
    },
    {
      "epoch": 1.0384377148380335,
      "grad_norm": 0.5367005604858965,
      "learning_rate": 9.863321254323022e-05,
      "loss": 11.9909,
      "step": 19070
    },
    {
      "epoch": 1.0384921688346165,
      "grad_norm": 0.571728376343309,
      "learning_rate": 9.862439511410297e-05,
      "loss": 11.984,
      "step": 19071
    },
    {
      "epoch": 1.0385466228311995,
      "grad_norm": 0.5030556817084691,
      "learning_rate": 9.86155776956727e-05,
      "loss": 12.0831,
      "step": 19072
    },
    {
      "epoch": 1.0386010768277825,
      "grad_norm": 0.5938392851995647,
      "learning_rate": 9.86067602880079e-05,
      "loss": 12.0718,
      "step": 19073
    },
    {
      "epoch": 1.0386555308243655,
      "grad_norm": 0.5332649327413911,
      "learning_rate": 9.859794289117716e-05,
      "loss": 11.9327,
      "step": 19074
    },
    {
      "epoch": 1.0387099848209485,
      "grad_norm": 0.5293914447218716,
      "learning_rate": 9.858912550524903e-05,
      "loss": 12.0487,
      "step": 19075
    },
    {
      "epoch": 1.0387644388175314,
      "grad_norm": 0.5238701352832031,
      "learning_rate": 9.858030813029214e-05,
      "loss": 12.1218,
      "step": 19076
    },
    {
      "epoch": 1.0388188928141144,
      "grad_norm": 0.6132977437798783,
      "learning_rate": 9.857149076637496e-05,
      "loss": 12.1199,
      "step": 19077
    },
    {
      "epoch": 1.0388733468106974,
      "grad_norm": 0.5851041481376358,
      "learning_rate": 9.856267341356613e-05,
      "loss": 11.8652,
      "step": 19078
    },
    {
      "epoch": 1.0389278008072804,
      "grad_norm": 0.5249538310080266,
      "learning_rate": 9.85538560719342e-05,
      "loss": 11.9839,
      "step": 19079
    },
    {
      "epoch": 1.0389822548038634,
      "grad_norm": 0.5606023303923444,
      "learning_rate": 9.854503874154773e-05,
      "loss": 11.9515,
      "step": 19080
    },
    {
      "epoch": 1.0390367088004466,
      "grad_norm": 0.5288047906674928,
      "learning_rate": 9.853622142247528e-05,
      "loss": 12.0651,
      "step": 19081
    },
    {
      "epoch": 1.0390911627970296,
      "grad_norm": 0.5558862569835413,
      "learning_rate": 9.852740411478543e-05,
      "loss": 12.0179,
      "step": 19082
    },
    {
      "epoch": 1.0391456167936126,
      "grad_norm": 0.605587366926134,
      "learning_rate": 9.851858681854673e-05,
      "loss": 12.1148,
      "step": 19083
    },
    {
      "epoch": 1.0392000707901956,
      "grad_norm": 0.6323635269478186,
      "learning_rate": 9.850976953382773e-05,
      "loss": 12.0581,
      "step": 19084
    },
    {
      "epoch": 1.0392545247867786,
      "grad_norm": 0.5735279356919927,
      "learning_rate": 9.850095226069702e-05,
      "loss": 12.2183,
      "step": 19085
    },
    {
      "epoch": 1.0393089787833616,
      "grad_norm": 0.5211586953071619,
      "learning_rate": 9.849213499922316e-05,
      "loss": 11.8532,
      "step": 19086
    },
    {
      "epoch": 1.0393634327799446,
      "grad_norm": 0.5557305775491236,
      "learning_rate": 9.848331774947471e-05,
      "loss": 11.9333,
      "step": 19087
    },
    {
      "epoch": 1.0394178867765276,
      "grad_norm": 0.5457468580264888,
      "learning_rate": 9.84745005115202e-05,
      "loss": 12.0708,
      "step": 19088
    },
    {
      "epoch": 1.0394723407731106,
      "grad_norm": 0.5179863842829507,
      "learning_rate": 9.846568328542827e-05,
      "loss": 12.0847,
      "step": 19089
    },
    {
      "epoch": 1.0395267947696936,
      "grad_norm": 0.5237475254373459,
      "learning_rate": 9.845686607126744e-05,
      "loss": 12.0979,
      "step": 19090
    },
    {
      "epoch": 1.0395812487662766,
      "grad_norm": 0.6168798677338446,
      "learning_rate": 9.844804886910627e-05,
      "loss": 12.062,
      "step": 19091
    },
    {
      "epoch": 1.0396357027628595,
      "grad_norm": 0.5484336017930553,
      "learning_rate": 9.843923167901336e-05,
      "loss": 12.0649,
      "step": 19092
    },
    {
      "epoch": 1.0396901567594428,
      "grad_norm": 0.5006446798431007,
      "learning_rate": 9.843041450105722e-05,
      "loss": 12.0172,
      "step": 19093
    },
    {
      "epoch": 1.0397446107560258,
      "grad_norm": 0.5387552497608356,
      "learning_rate": 9.842159733530645e-05,
      "loss": 12.0331,
      "step": 19094
    },
    {
      "epoch": 1.0397990647526087,
      "grad_norm": 0.5992092342972867,
      "learning_rate": 9.841278018182959e-05,
      "loss": 12.042,
      "step": 19095
    },
    {
      "epoch": 1.0398535187491917,
      "grad_norm": 0.5518016802549429,
      "learning_rate": 9.840396304069522e-05,
      "loss": 12.1501,
      "step": 19096
    },
    {
      "epoch": 1.0399079727457747,
      "grad_norm": 0.5231586092041085,
      "learning_rate": 9.839514591197191e-05,
      "loss": 12.1477,
      "step": 19097
    },
    {
      "epoch": 1.0399624267423577,
      "grad_norm": 0.5250451595601056,
      "learning_rate": 9.838632879572821e-05,
      "loss": 11.9877,
      "step": 19098
    },
    {
      "epoch": 1.0400168807389407,
      "grad_norm": 0.5751355650493677,
      "learning_rate": 9.837751169203268e-05,
      "loss": 12.0432,
      "step": 19099
    },
    {
      "epoch": 1.0400713347355237,
      "grad_norm": 0.5475822383731445,
      "learning_rate": 9.836869460095388e-05,
      "loss": 12.1132,
      "step": 19100
    },
    {
      "epoch": 1.0401257887321067,
      "grad_norm": 0.5095680052981686,
      "learning_rate": 9.83598775225604e-05,
      "loss": 12.0877,
      "step": 19101
    },
    {
      "epoch": 1.0401802427286897,
      "grad_norm": 0.5203026601536467,
      "learning_rate": 9.835106045692084e-05,
      "loss": 12.1754,
      "step": 19102
    },
    {
      "epoch": 1.0402346967252727,
      "grad_norm": 0.5393357415466559,
      "learning_rate": 9.834224340410366e-05,
      "loss": 12.0501,
      "step": 19103
    },
    {
      "epoch": 1.040289150721856,
      "grad_norm": 0.5147226504600757,
      "learning_rate": 9.833342636417747e-05,
      "loss": 12.2127,
      "step": 19104
    },
    {
      "epoch": 1.040343604718439,
      "grad_norm": 0.5151805541245005,
      "learning_rate": 9.832460933721083e-05,
      "loss": 12.0432,
      "step": 19105
    },
    {
      "epoch": 1.040398058715022,
      "grad_norm": 0.5298712966251693,
      "learning_rate": 9.831579232327231e-05,
      "loss": 11.9219,
      "step": 19106
    },
    {
      "epoch": 1.0404525127116049,
      "grad_norm": 0.5028227447111524,
      "learning_rate": 9.830697532243049e-05,
      "loss": 12.0594,
      "step": 19107
    },
    {
      "epoch": 1.0405069667081879,
      "grad_norm": 0.5437150776845209,
      "learning_rate": 9.82981583347539e-05,
      "loss": 12.0402,
      "step": 19108
    },
    {
      "epoch": 1.0405614207047709,
      "grad_norm": 0.5714209694702647,
      "learning_rate": 9.828934136031113e-05,
      "loss": 12.0347,
      "step": 19109
    },
    {
      "epoch": 1.0406158747013539,
      "grad_norm": 0.5090023205310441,
      "learning_rate": 9.828052439917071e-05,
      "loss": 11.9786,
      "step": 19110
    },
    {
      "epoch": 1.0406703286979369,
      "grad_norm": 0.5052478240697585,
      "learning_rate": 9.827170745140121e-05,
      "loss": 11.9894,
      "step": 19111
    },
    {
      "epoch": 1.0407247826945198,
      "grad_norm": 0.520875318006523,
      "learning_rate": 9.826289051707127e-05,
      "loss": 12.0789,
      "step": 19112
    },
    {
      "epoch": 1.0407792366911028,
      "grad_norm": 0.5776766893725211,
      "learning_rate": 9.825407359624935e-05,
      "loss": 11.9286,
      "step": 19113
    },
    {
      "epoch": 1.0408336906876858,
      "grad_norm": 0.5587302106616825,
      "learning_rate": 9.824525668900402e-05,
      "loss": 11.995,
      "step": 19114
    },
    {
      "epoch": 1.0408881446842688,
      "grad_norm": 0.5382811136105777,
      "learning_rate": 9.823643979540386e-05,
      "loss": 11.9325,
      "step": 19115
    },
    {
      "epoch": 1.040942598680852,
      "grad_norm": 0.5148025263395594,
      "learning_rate": 9.822762291551746e-05,
      "loss": 11.9106,
      "step": 19116
    },
    {
      "epoch": 1.040997052677435,
      "grad_norm": 0.5838594177462733,
      "learning_rate": 9.821880604941337e-05,
      "loss": 12.2477,
      "step": 19117
    },
    {
      "epoch": 1.041051506674018,
      "grad_norm": 0.5123221411956282,
      "learning_rate": 9.820998919716013e-05,
      "loss": 12.0669,
      "step": 19118
    },
    {
      "epoch": 1.041105960670601,
      "grad_norm": 0.565955948169352,
      "learning_rate": 9.820117235882633e-05,
      "loss": 12.0751,
      "step": 19119
    },
    {
      "epoch": 1.041160414667184,
      "grad_norm": 0.5621022639430323,
      "learning_rate": 9.81923555344805e-05,
      "loss": 12.0712,
      "step": 19120
    },
    {
      "epoch": 1.041214868663767,
      "grad_norm": 0.5499786944827505,
      "learning_rate": 9.818353872419121e-05,
      "loss": 12.1968,
      "step": 19121
    },
    {
      "epoch": 1.04126932266035,
      "grad_norm": 0.5583103952439287,
      "learning_rate": 9.817472192802707e-05,
      "loss": 12.1014,
      "step": 19122
    },
    {
      "epoch": 1.041323776656933,
      "grad_norm": 0.5163468250550807,
      "learning_rate": 9.816590514605657e-05,
      "loss": 11.9431,
      "step": 19123
    },
    {
      "epoch": 1.041378230653516,
      "grad_norm": 0.5665414309500235,
      "learning_rate": 9.815708837834829e-05,
      "loss": 12.0765,
      "step": 19124
    },
    {
      "epoch": 1.041432684650099,
      "grad_norm": 0.5285343488866202,
      "learning_rate": 9.814827162497082e-05,
      "loss": 12.055,
      "step": 19125
    },
    {
      "epoch": 1.041487138646682,
      "grad_norm": 0.5315934905705026,
      "learning_rate": 9.813945488599266e-05,
      "loss": 12.0311,
      "step": 19126
    },
    {
      "epoch": 1.041541592643265,
      "grad_norm": 0.6469743773818489,
      "learning_rate": 9.813063816148244e-05,
      "loss": 12.1042,
      "step": 19127
    },
    {
      "epoch": 1.0415960466398482,
      "grad_norm": 0.5832980168406593,
      "learning_rate": 9.81218214515087e-05,
      "loss": 12.1441,
      "step": 19128
    },
    {
      "epoch": 1.0416505006364312,
      "grad_norm": 0.5807553796234487,
      "learning_rate": 9.811300475613997e-05,
      "loss": 12.1559,
      "step": 19129
    },
    {
      "epoch": 1.0417049546330142,
      "grad_norm": 0.5623617827311373,
      "learning_rate": 9.810418807544483e-05,
      "loss": 12.1247,
      "step": 19130
    },
    {
      "epoch": 1.0417594086295972,
      "grad_norm": 0.5565142904968856,
      "learning_rate": 9.809537140949187e-05,
      "loss": 12.0773,
      "step": 19131
    },
    {
      "epoch": 1.0418138626261801,
      "grad_norm": 0.48345586368369114,
      "learning_rate": 9.808655475834962e-05,
      "loss": 12.0419,
      "step": 19132
    },
    {
      "epoch": 1.0418683166227631,
      "grad_norm": 0.6031622305737758,
      "learning_rate": 9.807773812208662e-05,
      "loss": 12.1528,
      "step": 19133
    },
    {
      "epoch": 1.0419227706193461,
      "grad_norm": 0.5163447645287528,
      "learning_rate": 9.806892150077147e-05,
      "loss": 12.037,
      "step": 19134
    },
    {
      "epoch": 1.0419772246159291,
      "grad_norm": 0.5209695862658643,
      "learning_rate": 9.80601048944727e-05,
      "loss": 12.0789,
      "step": 19135
    },
    {
      "epoch": 1.0420316786125121,
      "grad_norm": 0.5494787522018224,
      "learning_rate": 9.805128830325887e-05,
      "loss": 12.1595,
      "step": 19136
    },
    {
      "epoch": 1.042086132609095,
      "grad_norm": 0.5891751228617196,
      "learning_rate": 9.804247172719854e-05,
      "loss": 12.1231,
      "step": 19137
    },
    {
      "epoch": 1.042140586605678,
      "grad_norm": 0.5817300733494469,
      "learning_rate": 9.803365516636028e-05,
      "loss": 12.1236,
      "step": 19138
    },
    {
      "epoch": 1.042195040602261,
      "grad_norm": 0.5814731823831777,
      "learning_rate": 9.802483862081267e-05,
      "loss": 12.1709,
      "step": 19139
    },
    {
      "epoch": 1.0422494945988443,
      "grad_norm": 0.5579691487335727,
      "learning_rate": 9.801602209062424e-05,
      "loss": 12.1035,
      "step": 19140
    },
    {
      "epoch": 1.0423039485954273,
      "grad_norm": 0.5740579702293666,
      "learning_rate": 9.800720557586354e-05,
      "loss": 11.904,
      "step": 19141
    },
    {
      "epoch": 1.0423584025920103,
      "grad_norm": 0.5928448752546925,
      "learning_rate": 9.799838907659918e-05,
      "loss": 12.1609,
      "step": 19142
    },
    {
      "epoch": 1.0424128565885933,
      "grad_norm": 0.5212588888353008,
      "learning_rate": 9.798957259289966e-05,
      "loss": 12.1121,
      "step": 19143
    },
    {
      "epoch": 1.0424673105851763,
      "grad_norm": 0.5754309059415892,
      "learning_rate": 9.798075612483356e-05,
      "loss": 12.0057,
      "step": 19144
    },
    {
      "epoch": 1.0425217645817593,
      "grad_norm": 0.544196322848585,
      "learning_rate": 9.797193967246943e-05,
      "loss": 12.0195,
      "step": 19145
    },
    {
      "epoch": 1.0425762185783423,
      "grad_norm": 0.5588146386938192,
      "learning_rate": 9.796312323587585e-05,
      "loss": 12.0907,
      "step": 19146
    },
    {
      "epoch": 1.0426306725749253,
      "grad_norm": 0.567915320166216,
      "learning_rate": 9.795430681512137e-05,
      "loss": 12.0485,
      "step": 19147
    },
    {
      "epoch": 1.0426851265715082,
      "grad_norm": 0.5346818621860815,
      "learning_rate": 9.794549041027454e-05,
      "loss": 12.1685,
      "step": 19148
    },
    {
      "epoch": 1.0427395805680912,
      "grad_norm": 0.5617888190170663,
      "learning_rate": 9.793667402140388e-05,
      "loss": 12.1387,
      "step": 19149
    },
    {
      "epoch": 1.0427940345646742,
      "grad_norm": 0.524658144114567,
      "learning_rate": 9.792785764857802e-05,
      "loss": 12.0566,
      "step": 19150
    },
    {
      "epoch": 1.0428484885612574,
      "grad_norm": 0.5713117601184278,
      "learning_rate": 9.79190412918655e-05,
      "loss": 12.0968,
      "step": 19151
    },
    {
      "epoch": 1.0429029425578404,
      "grad_norm": 0.49936954976258013,
      "learning_rate": 9.791022495133489e-05,
      "loss": 12.0575,
      "step": 19152
    },
    {
      "epoch": 1.0429573965544234,
      "grad_norm": 0.6129223975829453,
      "learning_rate": 9.790140862705468e-05,
      "loss": 12.193,
      "step": 19153
    },
    {
      "epoch": 1.0430118505510064,
      "grad_norm": 0.6142847898671409,
      "learning_rate": 9.789259231909346e-05,
      "loss": 12.1012,
      "step": 19154
    },
    {
      "epoch": 1.0430663045475894,
      "grad_norm": 0.5700271812475715,
      "learning_rate": 9.788377602751982e-05,
      "loss": 12.015,
      "step": 19155
    },
    {
      "epoch": 1.0431207585441724,
      "grad_norm": 0.4888156128269216,
      "learning_rate": 9.787495975240227e-05,
      "loss": 11.9991,
      "step": 19156
    },
    {
      "epoch": 1.0431752125407554,
      "grad_norm": 0.5083256571510111,
      "learning_rate": 9.78661434938094e-05,
      "loss": 11.995,
      "step": 19157
    },
    {
      "epoch": 1.0432296665373384,
      "grad_norm": 0.6171669719782117,
      "learning_rate": 9.785732725180977e-05,
      "loss": 12.0293,
      "step": 19158
    },
    {
      "epoch": 1.0432841205339214,
      "grad_norm": 0.585761777581188,
      "learning_rate": 9.78485110264719e-05,
      "loss": 11.9625,
      "step": 19159
    },
    {
      "epoch": 1.0433385745305044,
      "grad_norm": 0.6041261370127157,
      "learning_rate": 9.783969481786435e-05,
      "loss": 12.0765,
      "step": 19160
    },
    {
      "epoch": 1.0433930285270874,
      "grad_norm": 0.6467978619240164,
      "learning_rate": 9.783087862605572e-05,
      "loss": 12.1038,
      "step": 19161
    },
    {
      "epoch": 1.0434474825236704,
      "grad_norm": 0.5517345887780721,
      "learning_rate": 9.782206245111459e-05,
      "loss": 12.184,
      "step": 19162
    },
    {
      "epoch": 1.0435019365202536,
      "grad_norm": 0.5585662895914788,
      "learning_rate": 9.781324629310942e-05,
      "loss": 12.1267,
      "step": 19163
    },
    {
      "epoch": 1.0435563905168366,
      "grad_norm": 0.5072476663086728,
      "learning_rate": 9.780443015210881e-05,
      "loss": 12.1028,
      "step": 19164
    },
    {
      "epoch": 1.0436108445134196,
      "grad_norm": 0.5149472102231276,
      "learning_rate": 9.779561402818131e-05,
      "loss": 12.1073,
      "step": 19165
    },
    {
      "epoch": 1.0436652985100026,
      "grad_norm": 0.5470801062193369,
      "learning_rate": 9.778679792139552e-05,
      "loss": 12.0336,
      "step": 19166
    },
    {
      "epoch": 1.0437197525065856,
      "grad_norm": 0.4981163730176472,
      "learning_rate": 9.777798183181993e-05,
      "loss": 12.0814,
      "step": 19167
    },
    {
      "epoch": 1.0437742065031685,
      "grad_norm": 0.5134029700065394,
      "learning_rate": 9.776916575952314e-05,
      "loss": 12.08,
      "step": 19168
    },
    {
      "epoch": 1.0438286604997515,
      "grad_norm": 0.5832556991628526,
      "learning_rate": 9.776034970457369e-05,
      "loss": 12.0783,
      "step": 19169
    },
    {
      "epoch": 1.0438831144963345,
      "grad_norm": 0.5377942104959453,
      "learning_rate": 9.775153366704013e-05,
      "loss": 12.1176,
      "step": 19170
    },
    {
      "epoch": 1.0439375684929175,
      "grad_norm": 0.6218772734868826,
      "learning_rate": 9.774271764699101e-05,
      "loss": 11.987,
      "step": 19171
    },
    {
      "epoch": 1.0439920224895005,
      "grad_norm": 0.5466535332473306,
      "learning_rate": 9.773390164449495e-05,
      "loss": 11.9507,
      "step": 19172
    },
    {
      "epoch": 1.0440464764860835,
      "grad_norm": 0.5297762922881801,
      "learning_rate": 9.772508565962042e-05,
      "loss": 11.9809,
      "step": 19173
    },
    {
      "epoch": 1.0441009304826667,
      "grad_norm": 0.5582817533417825,
      "learning_rate": 9.7716269692436e-05,
      "loss": 12.1064,
      "step": 19174
    },
    {
      "epoch": 1.0441553844792497,
      "grad_norm": 0.5852479088644624,
      "learning_rate": 9.770745374301022e-05,
      "loss": 12.076,
      "step": 19175
    },
    {
      "epoch": 1.0442098384758327,
      "grad_norm": 0.5511628923438155,
      "learning_rate": 9.76986378114117e-05,
      "loss": 12.1843,
      "step": 19176
    },
    {
      "epoch": 1.0442642924724157,
      "grad_norm": 0.5528512384448054,
      "learning_rate": 9.768982189770894e-05,
      "loss": 12.042,
      "step": 19177
    },
    {
      "epoch": 1.0443187464689987,
      "grad_norm": 0.5579554993906317,
      "learning_rate": 9.768100600197053e-05,
      "loss": 12.0521,
      "step": 19178
    },
    {
      "epoch": 1.0443732004655817,
      "grad_norm": 0.5920345224342375,
      "learning_rate": 9.7672190124265e-05,
      "loss": 12.1227,
      "step": 19179
    },
    {
      "epoch": 1.0444276544621647,
      "grad_norm": 0.5402265066864046,
      "learning_rate": 9.76633742646609e-05,
      "loss": 12.1896,
      "step": 19180
    },
    {
      "epoch": 1.0444821084587477,
      "grad_norm": 0.5230294482354321,
      "learning_rate": 9.76545584232268e-05,
      "loss": 11.9813,
      "step": 19181
    },
    {
      "epoch": 1.0445365624553307,
      "grad_norm": 0.5513612958958378,
      "learning_rate": 9.764574260003128e-05,
      "loss": 12.1628,
      "step": 19182
    },
    {
      "epoch": 1.0445910164519137,
      "grad_norm": 0.5647767986230641,
      "learning_rate": 9.763692679514284e-05,
      "loss": 11.8818,
      "step": 19183
    },
    {
      "epoch": 1.0446454704484966,
      "grad_norm": 0.5521198979994846,
      "learning_rate": 9.762811100863003e-05,
      "loss": 12.1333,
      "step": 19184
    },
    {
      "epoch": 1.0446999244450796,
      "grad_norm": 0.5230827615522323,
      "learning_rate": 9.761929524056145e-05,
      "loss": 12.1484,
      "step": 19185
    },
    {
      "epoch": 1.0447543784416629,
      "grad_norm": 0.5349816199825447,
      "learning_rate": 9.761047949100558e-05,
      "loss": 12.0065,
      "step": 19186
    },
    {
      "epoch": 1.0448088324382458,
      "grad_norm": 0.5682286337147697,
      "learning_rate": 9.760166376003107e-05,
      "loss": 12.1066,
      "step": 19187
    },
    {
      "epoch": 1.0448632864348288,
      "grad_norm": 0.533617795780643,
      "learning_rate": 9.759284804770642e-05,
      "loss": 12.0198,
      "step": 19188
    },
    {
      "epoch": 1.0449177404314118,
      "grad_norm": 0.6171098731477254,
      "learning_rate": 9.758403235410019e-05,
      "loss": 12.1067,
      "step": 19189
    },
    {
      "epoch": 1.0449721944279948,
      "grad_norm": 0.5329021938234568,
      "learning_rate": 9.757521667928092e-05,
      "loss": 11.9268,
      "step": 19190
    },
    {
      "epoch": 1.0450266484245778,
      "grad_norm": 0.5366953787561417,
      "learning_rate": 9.756640102331718e-05,
      "loss": 12.0649,
      "step": 19191
    },
    {
      "epoch": 1.0450811024211608,
      "grad_norm": 0.5661101853878987,
      "learning_rate": 9.755758538627753e-05,
      "loss": 12.1772,
      "step": 19192
    },
    {
      "epoch": 1.0451355564177438,
      "grad_norm": 0.6285858479349945,
      "learning_rate": 9.754876976823049e-05,
      "loss": 12.073,
      "step": 19193
    },
    {
      "epoch": 1.0451900104143268,
      "grad_norm": 0.5891952748044051,
      "learning_rate": 9.753995416924462e-05,
      "loss": 11.997,
      "step": 19194
    },
    {
      "epoch": 1.0452444644109098,
      "grad_norm": 0.5354460728775402,
      "learning_rate": 9.753113858938847e-05,
      "loss": 12.0828,
      "step": 19195
    },
    {
      "epoch": 1.0452989184074928,
      "grad_norm": 0.6363409217279798,
      "learning_rate": 9.752232302873061e-05,
      "loss": 12.0971,
      "step": 19196
    },
    {
      "epoch": 1.0453533724040758,
      "grad_norm": 0.6408417171455034,
      "learning_rate": 9.751350748733959e-05,
      "loss": 12.0337,
      "step": 19197
    },
    {
      "epoch": 1.045407826400659,
      "grad_norm": 0.5498651446757598,
      "learning_rate": 9.750469196528392e-05,
      "loss": 12.0722,
      "step": 19198
    },
    {
      "epoch": 1.045462280397242,
      "grad_norm": 0.6273062786380214,
      "learning_rate": 9.749587646263221e-05,
      "loss": 11.9887,
      "step": 19199
    },
    {
      "epoch": 1.045516734393825,
      "grad_norm": 0.631159313500319,
      "learning_rate": 9.748706097945298e-05,
      "loss": 12.0544,
      "step": 19200
    },
    {
      "epoch": 1.045571188390408,
      "grad_norm": 0.5390729842486164,
      "learning_rate": 9.74782455158148e-05,
      "loss": 12.036,
      "step": 19201
    },
    {
      "epoch": 1.045625642386991,
      "grad_norm": 0.5481233693458728,
      "learning_rate": 9.746943007178622e-05,
      "loss": 12.1109,
      "step": 19202
    },
    {
      "epoch": 1.045680096383574,
      "grad_norm": 0.6065527736780751,
      "learning_rate": 9.746061464743575e-05,
      "loss": 12.1451,
      "step": 19203
    },
    {
      "epoch": 1.045734550380157,
      "grad_norm": 0.573328558430312,
      "learning_rate": 9.745179924283196e-05,
      "loss": 12.1437,
      "step": 19204
    },
    {
      "epoch": 1.04578900437674,
      "grad_norm": 0.5542725518096085,
      "learning_rate": 9.744298385804341e-05,
      "loss": 11.9381,
      "step": 19205
    },
    {
      "epoch": 1.045843458373323,
      "grad_norm": 0.5887718192429499,
      "learning_rate": 9.743416849313866e-05,
      "loss": 12.1066,
      "step": 19206
    },
    {
      "epoch": 1.045897912369906,
      "grad_norm": 0.5810414666499816,
      "learning_rate": 9.742535314818624e-05,
      "loss": 12.1456,
      "step": 19207
    },
    {
      "epoch": 1.045952366366489,
      "grad_norm": 0.5934364387216299,
      "learning_rate": 9.74165378232547e-05,
      "loss": 12.07,
      "step": 19208
    },
    {
      "epoch": 1.046006820363072,
      "grad_norm": 0.544803947709447,
      "learning_rate": 9.740772251841257e-05,
      "loss": 11.9487,
      "step": 19209
    },
    {
      "epoch": 1.0460612743596551,
      "grad_norm": 0.6212074752348654,
      "learning_rate": 9.739890723372845e-05,
      "loss": 12.0921,
      "step": 19210
    },
    {
      "epoch": 1.0461157283562381,
      "grad_norm": 0.6278018299328106,
      "learning_rate": 9.739009196927086e-05,
      "loss": 12.0741,
      "step": 19211
    },
    {
      "epoch": 1.046170182352821,
      "grad_norm": 0.5454639466797658,
      "learning_rate": 9.738127672510836e-05,
      "loss": 12.0252,
      "step": 19212
    },
    {
      "epoch": 1.046224636349404,
      "grad_norm": 0.5713008151891188,
      "learning_rate": 9.737246150130951e-05,
      "loss": 11.949,
      "step": 19213
    },
    {
      "epoch": 1.046279090345987,
      "grad_norm": 0.5813303234196573,
      "learning_rate": 9.736364629794283e-05,
      "loss": 12.1801,
      "step": 19214
    },
    {
      "epoch": 1.04633354434257,
      "grad_norm": 0.563521297845805,
      "learning_rate": 9.735483111507686e-05,
      "loss": 11.8658,
      "step": 19215
    },
    {
      "epoch": 1.046387998339153,
      "grad_norm": 0.5257352010378813,
      "learning_rate": 9.734601595278018e-05,
      "loss": 12.0067,
      "step": 19216
    },
    {
      "epoch": 1.046442452335736,
      "grad_norm": 0.5098668625026491,
      "learning_rate": 9.733720081112132e-05,
      "loss": 12.052,
      "step": 19217
    },
    {
      "epoch": 1.046496906332319,
      "grad_norm": 0.5296134836923844,
      "learning_rate": 9.732838569016884e-05,
      "loss": 12.1223,
      "step": 19218
    },
    {
      "epoch": 1.046551360328902,
      "grad_norm": 0.5534850423144859,
      "learning_rate": 9.731957058999127e-05,
      "loss": 12.0633,
      "step": 19219
    },
    {
      "epoch": 1.046605814325485,
      "grad_norm": 0.5026348269791794,
      "learning_rate": 9.731075551065714e-05,
      "loss": 12.0089,
      "step": 19220
    },
    {
      "epoch": 1.0466602683220683,
      "grad_norm": 0.5766132676570002,
      "learning_rate": 9.730194045223506e-05,
      "loss": 12.0963,
      "step": 19221
    },
    {
      "epoch": 1.0467147223186513,
      "grad_norm": 0.5168019245496915,
      "learning_rate": 9.729312541479355e-05,
      "loss": 12.0861,
      "step": 19222
    },
    {
      "epoch": 1.0467691763152343,
      "grad_norm": 0.6158430966438531,
      "learning_rate": 9.728431039840118e-05,
      "loss": 12.0681,
      "step": 19223
    },
    {
      "epoch": 1.0468236303118172,
      "grad_norm": 0.5121067618505272,
      "learning_rate": 9.72754954031264e-05,
      "loss": 11.8811,
      "step": 19224
    },
    {
      "epoch": 1.0468780843084002,
      "grad_norm": 0.5995318782549505,
      "learning_rate": 9.726668042903786e-05,
      "loss": 12.1476,
      "step": 19225
    },
    {
      "epoch": 1.0469325383049832,
      "grad_norm": 0.5251005441098441,
      "learning_rate": 9.725786547620407e-05,
      "loss": 12.0328,
      "step": 19226
    },
    {
      "epoch": 1.0469869923015662,
      "grad_norm": 0.5925345936070446,
      "learning_rate": 9.724905054469357e-05,
      "loss": 12.2029,
      "step": 19227
    },
    {
      "epoch": 1.0470414462981492,
      "grad_norm": 0.5644629534226009,
      "learning_rate": 9.724023563457492e-05,
      "loss": 12.0922,
      "step": 19228
    },
    {
      "epoch": 1.0470959002947322,
      "grad_norm": 0.5145965892903692,
      "learning_rate": 9.723142074591665e-05,
      "loss": 12.0152,
      "step": 19229
    },
    {
      "epoch": 1.0471503542913152,
      "grad_norm": 0.617468919722614,
      "learning_rate": 9.722260587878734e-05,
      "loss": 11.9832,
      "step": 19230
    },
    {
      "epoch": 1.0472048082878982,
      "grad_norm": 0.5774638227556544,
      "learning_rate": 9.721379103325548e-05,
      "loss": 12.1204,
      "step": 19231
    },
    {
      "epoch": 1.0472592622844812,
      "grad_norm": 0.48206586144768443,
      "learning_rate": 9.720497620938965e-05,
      "loss": 12.013,
      "step": 19232
    },
    {
      "epoch": 1.0473137162810644,
      "grad_norm": 0.6548010509130806,
      "learning_rate": 9.719616140725846e-05,
      "loss": 12.1303,
      "step": 19233
    },
    {
      "epoch": 1.0473681702776474,
      "grad_norm": 0.698867251761497,
      "learning_rate": 9.718734662693034e-05,
      "loss": 12.2486,
      "step": 19234
    },
    {
      "epoch": 1.0474226242742304,
      "grad_norm": 0.5256828253879882,
      "learning_rate": 9.717853186847386e-05,
      "loss": 12.0381,
      "step": 19235
    },
    {
      "epoch": 1.0474770782708134,
      "grad_norm": 0.6207702632861498,
      "learning_rate": 9.716971713195762e-05,
      "loss": 12.1398,
      "step": 19236
    },
    {
      "epoch": 1.0475315322673964,
      "grad_norm": 0.5712375661066159,
      "learning_rate": 9.716090241745012e-05,
      "loss": 12.0401,
      "step": 19237
    },
    {
      "epoch": 1.0475859862639794,
      "grad_norm": 0.4935113312443913,
      "learning_rate": 9.715208772501992e-05,
      "loss": 12.0417,
      "step": 19238
    },
    {
      "epoch": 1.0476404402605624,
      "grad_norm": 0.6217792785204247,
      "learning_rate": 9.714327305473558e-05,
      "loss": 12.1523,
      "step": 19239
    },
    {
      "epoch": 1.0476948942571453,
      "grad_norm": 0.5723809438569106,
      "learning_rate": 9.713445840666562e-05,
      "loss": 12.0664,
      "step": 19240
    },
    {
      "epoch": 1.0477493482537283,
      "grad_norm": 0.5677739931138212,
      "learning_rate": 9.712564378087858e-05,
      "loss": 12.0922,
      "step": 19241
    },
    {
      "epoch": 1.0478038022503113,
      "grad_norm": 0.5640126414664229,
      "learning_rate": 9.7116829177443e-05,
      "loss": 12.131,
      "step": 19242
    },
    {
      "epoch": 1.0478582562468943,
      "grad_norm": 0.6640791811526993,
      "learning_rate": 9.710801459642751e-05,
      "loss": 11.9364,
      "step": 19243
    },
    {
      "epoch": 1.0479127102434775,
      "grad_norm": 0.5250140126089315,
      "learning_rate": 9.709920003790054e-05,
      "loss": 12.0516,
      "step": 19244
    },
    {
      "epoch": 1.0479671642400605,
      "grad_norm": 0.6142053809536904,
      "learning_rate": 9.709038550193068e-05,
      "loss": 12.0713,
      "step": 19245
    },
    {
      "epoch": 1.0480216182366435,
      "grad_norm": 0.5986476016795224,
      "learning_rate": 9.708157098858645e-05,
      "loss": 12.1929,
      "step": 19246
    },
    {
      "epoch": 1.0480760722332265,
      "grad_norm": 0.6491176720117351,
      "learning_rate": 9.707275649793642e-05,
      "loss": 12.0925,
      "step": 19247
    },
    {
      "epoch": 1.0481305262298095,
      "grad_norm": 0.6034636611147448,
      "learning_rate": 9.706394203004914e-05,
      "loss": 11.9917,
      "step": 19248
    },
    {
      "epoch": 1.0481849802263925,
      "grad_norm": 0.5886294219976741,
      "learning_rate": 9.705512758499313e-05,
      "loss": 12.1267,
      "step": 19249
    },
    {
      "epoch": 1.0482394342229755,
      "grad_norm": 0.6028214189036069,
      "learning_rate": 9.704631316283695e-05,
      "loss": 11.9418,
      "step": 19250
    },
    {
      "epoch": 1.0482938882195585,
      "grad_norm": 0.5716779343276686,
      "learning_rate": 9.703749876364913e-05,
      "loss": 12.0819,
      "step": 19251
    },
    {
      "epoch": 1.0483483422161415,
      "grad_norm": 0.5664468715715333,
      "learning_rate": 9.702868438749822e-05,
      "loss": 12.1577,
      "step": 19252
    },
    {
      "epoch": 1.0484027962127245,
      "grad_norm": 0.6202810352599408,
      "learning_rate": 9.701987003445278e-05,
      "loss": 11.9958,
      "step": 19253
    },
    {
      "epoch": 1.0484572502093075,
      "grad_norm": 0.6287441056132557,
      "learning_rate": 9.70110557045813e-05,
      "loss": 12.074,
      "step": 19254
    },
    {
      "epoch": 1.0485117042058905,
      "grad_norm": 0.6081310291659571,
      "learning_rate": 9.700224139795236e-05,
      "loss": 12.0567,
      "step": 19255
    },
    {
      "epoch": 1.0485661582024737,
      "grad_norm": 0.6344891453275684,
      "learning_rate": 9.699342711463448e-05,
      "loss": 12.0007,
      "step": 19256
    },
    {
      "epoch": 1.0486206121990567,
      "grad_norm": 0.5084448846445512,
      "learning_rate": 9.698461285469624e-05,
      "loss": 12.0539,
      "step": 19257
    },
    {
      "epoch": 1.0486750661956397,
      "grad_norm": 0.5475980659381137,
      "learning_rate": 9.697579861820611e-05,
      "loss": 12.1256,
      "step": 19258
    },
    {
      "epoch": 1.0487295201922227,
      "grad_norm": 0.523600751764549,
      "learning_rate": 9.696698440523271e-05,
      "loss": 12.1141,
      "step": 19259
    },
    {
      "epoch": 1.0487839741888056,
      "grad_norm": 0.6069424425625963,
      "learning_rate": 9.695817021584454e-05,
      "loss": 12.0484,
      "step": 19260
    },
    {
      "epoch": 1.0488384281853886,
      "grad_norm": 0.5742100551630694,
      "learning_rate": 9.694935605011017e-05,
      "loss": 12.0797,
      "step": 19261
    },
    {
      "epoch": 1.0488928821819716,
      "grad_norm": 0.5941016295730495,
      "learning_rate": 9.69405419080981e-05,
      "loss": 12.1006,
      "step": 19262
    },
    {
      "epoch": 1.0489473361785546,
      "grad_norm": 0.554492287310076,
      "learning_rate": 9.693172778987692e-05,
      "loss": 12.0692,
      "step": 19263
    },
    {
      "epoch": 1.0490017901751376,
      "grad_norm": 0.6211903334798932,
      "learning_rate": 9.69229136955151e-05,
      "loss": 11.9765,
      "step": 19264
    },
    {
      "epoch": 1.0490562441717206,
      "grad_norm": 0.6354330720514915,
      "learning_rate": 9.691409962508124e-05,
      "loss": 12.0981,
      "step": 19265
    },
    {
      "epoch": 1.0491106981683036,
      "grad_norm": 0.581561168575633,
      "learning_rate": 9.690528557864386e-05,
      "loss": 12.0893,
      "step": 19266
    },
    {
      "epoch": 1.0491651521648866,
      "grad_norm": 0.5602093649184966,
      "learning_rate": 9.689647155627149e-05,
      "loss": 11.992,
      "step": 19267
    },
    {
      "epoch": 1.0492196061614698,
      "grad_norm": 0.541758484464551,
      "learning_rate": 9.688765755803268e-05,
      "loss": 11.9762,
      "step": 19268
    },
    {
      "epoch": 1.0492740601580528,
      "grad_norm": 0.5567906003713667,
      "learning_rate": 9.687884358399594e-05,
      "loss": 12.0616,
      "step": 19269
    },
    {
      "epoch": 1.0493285141546358,
      "grad_norm": 0.5468525484934957,
      "learning_rate": 9.687002963422986e-05,
      "loss": 12.0693,
      "step": 19270
    },
    {
      "epoch": 1.0493829681512188,
      "grad_norm": 0.6748688401673388,
      "learning_rate": 9.686121570880294e-05,
      "loss": 12.0306,
      "step": 19271
    },
    {
      "epoch": 1.0494374221478018,
      "grad_norm": 0.6129731420684097,
      "learning_rate": 9.685240180778376e-05,
      "loss": 12.0475,
      "step": 19272
    },
    {
      "epoch": 1.0494918761443848,
      "grad_norm": 0.5224925834604115,
      "learning_rate": 9.684358793124084e-05,
      "loss": 12.1006,
      "step": 19273
    },
    {
      "epoch": 1.0495463301409678,
      "grad_norm": 0.525354978290664,
      "learning_rate": 9.683477407924268e-05,
      "loss": 12.0718,
      "step": 19274
    },
    {
      "epoch": 1.0496007841375508,
      "grad_norm": 0.5501231338286172,
      "learning_rate": 9.682596025185786e-05,
      "loss": 12.037,
      "step": 19275
    },
    {
      "epoch": 1.0496552381341338,
      "grad_norm": 0.5525824228493883,
      "learning_rate": 9.68171464491549e-05,
      "loss": 11.8471,
      "step": 19276
    },
    {
      "epoch": 1.0497096921307167,
      "grad_norm": 0.5237595322292766,
      "learning_rate": 9.680833267120234e-05,
      "loss": 12.1528,
      "step": 19277
    },
    {
      "epoch": 1.0497641461272997,
      "grad_norm": 0.547492121864182,
      "learning_rate": 9.679951891806873e-05,
      "loss": 12.1126,
      "step": 19278
    },
    {
      "epoch": 1.0498186001238827,
      "grad_norm": 0.5423175265745502,
      "learning_rate": 9.679070518982259e-05,
      "loss": 12.0847,
      "step": 19279
    },
    {
      "epoch": 1.049873054120466,
      "grad_norm": 0.5409640984493687,
      "learning_rate": 9.678189148653246e-05,
      "loss": 12.018,
      "step": 19280
    },
    {
      "epoch": 1.049927508117049,
      "grad_norm": 0.5342685429477202,
      "learning_rate": 9.677307780826687e-05,
      "loss": 12.0387,
      "step": 19281
    },
    {
      "epoch": 1.049981962113632,
      "grad_norm": 0.5299288139375267,
      "learning_rate": 9.676426415509439e-05,
      "loss": 12.0441,
      "step": 19282
    },
    {
      "epoch": 1.050036416110215,
      "grad_norm": 0.664810509725584,
      "learning_rate": 9.675545052708358e-05,
      "loss": 12.095,
      "step": 19283
    },
    {
      "epoch": 1.050090870106798,
      "grad_norm": 0.5478869296306361,
      "learning_rate": 9.674663692430286e-05,
      "loss": 12.0243,
      "step": 19284
    },
    {
      "epoch": 1.050145324103381,
      "grad_norm": 0.5575254940100597,
      "learning_rate": 9.673782334682085e-05,
      "loss": 12.0398,
      "step": 19285
    },
    {
      "epoch": 1.050199778099964,
      "grad_norm": 0.5672195448002826,
      "learning_rate": 9.672900979470608e-05,
      "loss": 12.0406,
      "step": 19286
    },
    {
      "epoch": 1.050254232096547,
      "grad_norm": 0.5550502418984661,
      "learning_rate": 9.672019626802708e-05,
      "loss": 12.0973,
      "step": 19287
    },
    {
      "epoch": 1.0503086860931299,
      "grad_norm": 0.5527939947632422,
      "learning_rate": 9.671138276685238e-05,
      "loss": 12.1604,
      "step": 19288
    },
    {
      "epoch": 1.0503631400897129,
      "grad_norm": 0.524206438490076,
      "learning_rate": 9.670256929125053e-05,
      "loss": 12.1217,
      "step": 19289
    },
    {
      "epoch": 1.0504175940862959,
      "grad_norm": 0.56160257257029,
      "learning_rate": 9.669375584129005e-05,
      "loss": 12.0312,
      "step": 19290
    },
    {
      "epoch": 1.050472048082879,
      "grad_norm": 0.5204986254117848,
      "learning_rate": 9.668494241703945e-05,
      "loss": 12.0738,
      "step": 19291
    },
    {
      "epoch": 1.050526502079462,
      "grad_norm": 0.5324262785130692,
      "learning_rate": 9.667612901856732e-05,
      "loss": 11.7358,
      "step": 19292
    },
    {
      "epoch": 1.050580956076045,
      "grad_norm": 0.5772614596923383,
      "learning_rate": 9.666731564594222e-05,
      "loss": 12.0383,
      "step": 19293
    },
    {
      "epoch": 1.050635410072628,
      "grad_norm": 0.5660892623770573,
      "learning_rate": 9.665850229923258e-05,
      "loss": 12.0707,
      "step": 19294
    },
    {
      "epoch": 1.050689864069211,
      "grad_norm": 0.5317718569733886,
      "learning_rate": 9.664968897850695e-05,
      "loss": 12.0076,
      "step": 19295
    },
    {
      "epoch": 1.050744318065794,
      "grad_norm": 0.5517253525771574,
      "learning_rate": 9.664087568383394e-05,
      "loss": 12.115,
      "step": 19296
    },
    {
      "epoch": 1.050798772062377,
      "grad_norm": 0.5468316584293257,
      "learning_rate": 9.663206241528204e-05,
      "loss": 12.1009,
      "step": 19297
    },
    {
      "epoch": 1.05085322605896,
      "grad_norm": 0.539461242898282,
      "learning_rate": 9.662324917291979e-05,
      "loss": 12.1092,
      "step": 19298
    },
    {
      "epoch": 1.050907680055543,
      "grad_norm": 0.4899975306952434,
      "learning_rate": 9.661443595681573e-05,
      "loss": 11.9413,
      "step": 19299
    },
    {
      "epoch": 1.050962134052126,
      "grad_norm": 0.5716632962528633,
      "learning_rate": 9.660562276703838e-05,
      "loss": 12.0186,
      "step": 19300
    },
    {
      "epoch": 1.051016588048709,
      "grad_norm": 0.531144636239472,
      "learning_rate": 9.659680960365626e-05,
      "loss": 11.9862,
      "step": 19301
    },
    {
      "epoch": 1.051071042045292,
      "grad_norm": 0.5140450260765325,
      "learning_rate": 9.658799646673793e-05,
      "loss": 12.1339,
      "step": 19302
    },
    {
      "epoch": 1.0511254960418752,
      "grad_norm": 0.5405131973007933,
      "learning_rate": 9.657918335635194e-05,
      "loss": 12.002,
      "step": 19303
    },
    {
      "epoch": 1.0511799500384582,
      "grad_norm": 0.5421925293191863,
      "learning_rate": 9.657037027256676e-05,
      "loss": 12.0472,
      "step": 19304
    },
    {
      "epoch": 1.0512344040350412,
      "grad_norm": 0.5010776098536464,
      "learning_rate": 9.656155721545094e-05,
      "loss": 12.0539,
      "step": 19305
    },
    {
      "epoch": 1.0512888580316242,
      "grad_norm": 0.5373765483696279,
      "learning_rate": 9.655274418507307e-05,
      "loss": 12.0694,
      "step": 19306
    },
    {
      "epoch": 1.0513433120282072,
      "grad_norm": 0.5229203708756343,
      "learning_rate": 9.654393118150159e-05,
      "loss": 11.9288,
      "step": 19307
    },
    {
      "epoch": 1.0513977660247902,
      "grad_norm": 0.523050323587943,
      "learning_rate": 9.653511820480511e-05,
      "loss": 12.0839,
      "step": 19308
    },
    {
      "epoch": 1.0514522200213732,
      "grad_norm": 0.52587795112077,
      "learning_rate": 9.652630525505213e-05,
      "loss": 12.0395,
      "step": 19309
    },
    {
      "epoch": 1.0515066740179562,
      "grad_norm": 0.5440393969364299,
      "learning_rate": 9.651749233231117e-05,
      "loss": 12.0127,
      "step": 19310
    },
    {
      "epoch": 1.0515611280145392,
      "grad_norm": 0.5877520952516384,
      "learning_rate": 9.65086794366508e-05,
      "loss": 12.1713,
      "step": 19311
    },
    {
      "epoch": 1.0516155820111222,
      "grad_norm": 0.5626253843201025,
      "learning_rate": 9.649986656813951e-05,
      "loss": 12.0419,
      "step": 19312
    },
    {
      "epoch": 1.0516700360077051,
      "grad_norm": 0.560972630053483,
      "learning_rate": 9.649105372684586e-05,
      "loss": 11.9527,
      "step": 19313
    },
    {
      "epoch": 1.0517244900042884,
      "grad_norm": 0.5201536054810402,
      "learning_rate": 9.648224091283835e-05,
      "loss": 11.8398,
      "step": 19314
    },
    {
      "epoch": 1.0517789440008714,
      "grad_norm": 0.5708413242075481,
      "learning_rate": 9.647342812618553e-05,
      "loss": 12.0981,
      "step": 19315
    },
    {
      "epoch": 1.0518333979974543,
      "grad_norm": 0.538158698770974,
      "learning_rate": 9.646461536695591e-05,
      "loss": 11.9265,
      "step": 19316
    },
    {
      "epoch": 1.0518878519940373,
      "grad_norm": 0.5379127553390638,
      "learning_rate": 9.645580263521805e-05,
      "loss": 12.1438,
      "step": 19317
    },
    {
      "epoch": 1.0519423059906203,
      "grad_norm": 0.5904144811653973,
      "learning_rate": 9.644698993104044e-05,
      "loss": 11.9552,
      "step": 19318
    },
    {
      "epoch": 1.0519967599872033,
      "grad_norm": 0.5484578526745109,
      "learning_rate": 9.643817725449163e-05,
      "loss": 12.0061,
      "step": 19319
    },
    {
      "epoch": 1.0520512139837863,
      "grad_norm": 0.5452909707857968,
      "learning_rate": 9.642936460564019e-05,
      "loss": 12.0431,
      "step": 19320
    },
    {
      "epoch": 1.0521056679803693,
      "grad_norm": 0.5896543186751604,
      "learning_rate": 9.642055198455457e-05,
      "loss": 12.1793,
      "step": 19321
    },
    {
      "epoch": 1.0521601219769523,
      "grad_norm": 0.5424352016462574,
      "learning_rate": 9.641173939130337e-05,
      "loss": 11.9724,
      "step": 19322
    },
    {
      "epoch": 1.0522145759735353,
      "grad_norm": 0.548154158162294,
      "learning_rate": 9.640292682595508e-05,
      "loss": 11.9925,
      "step": 19323
    },
    {
      "epoch": 1.0522690299701183,
      "grad_norm": 0.5113758994292995,
      "learning_rate": 9.639411428857823e-05,
      "loss": 12.145,
      "step": 19324
    },
    {
      "epoch": 1.0523234839667013,
      "grad_norm": 0.5444394344892941,
      "learning_rate": 9.638530177924136e-05,
      "loss": 12.011,
      "step": 19325
    },
    {
      "epoch": 1.0523779379632845,
      "grad_norm": 0.5504481687682757,
      "learning_rate": 9.637648929801297e-05,
      "loss": 11.9842,
      "step": 19326
    },
    {
      "epoch": 1.0524323919598675,
      "grad_norm": 0.5950719535763734,
      "learning_rate": 9.636767684496162e-05,
      "loss": 12.0425,
      "step": 19327
    },
    {
      "epoch": 1.0524868459564505,
      "grad_norm": 0.5807610266139246,
      "learning_rate": 9.635886442015582e-05,
      "loss": 12.0858,
      "step": 19328
    },
    {
      "epoch": 1.0525412999530335,
      "grad_norm": 0.5517142506957727,
      "learning_rate": 9.635005202366407e-05,
      "loss": 11.974,
      "step": 19329
    },
    {
      "epoch": 1.0525957539496165,
      "grad_norm": 0.6235125416785774,
      "learning_rate": 9.634123965555495e-05,
      "loss": 12.0126,
      "step": 19330
    },
    {
      "epoch": 1.0526502079461995,
      "grad_norm": 0.5510346855094569,
      "learning_rate": 9.633242731589698e-05,
      "loss": 12.0586,
      "step": 19331
    },
    {
      "epoch": 1.0527046619427824,
      "grad_norm": 0.5484858157775377,
      "learning_rate": 9.632361500475866e-05,
      "loss": 12.1546,
      "step": 19332
    },
    {
      "epoch": 1.0527591159393654,
      "grad_norm": 0.5569662093016531,
      "learning_rate": 9.631480272220855e-05,
      "loss": 12.0479,
      "step": 19333
    },
    {
      "epoch": 1.0528135699359484,
      "grad_norm": 0.5913187816809812,
      "learning_rate": 9.630599046831513e-05,
      "loss": 12.048,
      "step": 19334
    },
    {
      "epoch": 1.0528680239325314,
      "grad_norm": 0.5415003726459723,
      "learning_rate": 9.629717824314696e-05,
      "loss": 12.1657,
      "step": 19335
    },
    {
      "epoch": 1.0529224779291144,
      "grad_norm": 0.5880368853026926,
      "learning_rate": 9.628836604677253e-05,
      "loss": 11.9169,
      "step": 19336
    },
    {
      "epoch": 1.0529769319256976,
      "grad_norm": 0.6214821322491036,
      "learning_rate": 9.627955387926041e-05,
      "loss": 11.9572,
      "step": 19337
    },
    {
      "epoch": 1.0530313859222806,
      "grad_norm": 0.5374898237994633,
      "learning_rate": 9.627074174067909e-05,
      "loss": 12.0442,
      "step": 19338
    },
    {
      "epoch": 1.0530858399188636,
      "grad_norm": 0.5723645363983014,
      "learning_rate": 9.62619296310971e-05,
      "loss": 12.2039,
      "step": 19339
    },
    {
      "epoch": 1.0531402939154466,
      "grad_norm": 0.5839935285049328,
      "learning_rate": 9.625311755058296e-05,
      "loss": 11.9861,
      "step": 19340
    },
    {
      "epoch": 1.0531947479120296,
      "grad_norm": 0.5062275376127757,
      "learning_rate": 9.624430549920523e-05,
      "loss": 11.994,
      "step": 19341
    },
    {
      "epoch": 1.0532492019086126,
      "grad_norm": 0.5318769916007122,
      "learning_rate": 9.62354934770324e-05,
      "loss": 12.105,
      "step": 19342
    },
    {
      "epoch": 1.0533036559051956,
      "grad_norm": 0.5984575688586973,
      "learning_rate": 9.622668148413306e-05,
      "loss": 12.076,
      "step": 19343
    },
    {
      "epoch": 1.0533581099017786,
      "grad_norm": 0.5588173460010458,
      "learning_rate": 9.621786952057561e-05,
      "loss": 11.9369,
      "step": 19344
    },
    {
      "epoch": 1.0534125638983616,
      "grad_norm": 0.5773820580009359,
      "learning_rate": 9.620905758642867e-05,
      "loss": 12.0063,
      "step": 19345
    },
    {
      "epoch": 1.0534670178949446,
      "grad_norm": 0.5589950202886801,
      "learning_rate": 9.620024568176071e-05,
      "loss": 12.0991,
      "step": 19346
    },
    {
      "epoch": 1.0535214718915276,
      "grad_norm": 0.6144528426187391,
      "learning_rate": 9.61914338066403e-05,
      "loss": 11.9937,
      "step": 19347
    },
    {
      "epoch": 1.0535759258881106,
      "grad_norm": 0.5543121445619258,
      "learning_rate": 9.618262196113594e-05,
      "loss": 12.1332,
      "step": 19348
    },
    {
      "epoch": 1.0536303798846938,
      "grad_norm": 0.5227719779637813,
      "learning_rate": 9.617381014531614e-05,
      "loss": 12.1155,
      "step": 19349
    },
    {
      "epoch": 1.0536848338812768,
      "grad_norm": 0.47408434348923856,
      "learning_rate": 9.616499835924943e-05,
      "loss": 12.0586,
      "step": 19350
    },
    {
      "epoch": 1.0537392878778598,
      "grad_norm": 0.5539712618336022,
      "learning_rate": 9.615618660300434e-05,
      "loss": 12.0447,
      "step": 19351
    },
    {
      "epoch": 1.0537937418744427,
      "grad_norm": 0.5532620334707203,
      "learning_rate": 9.614737487664938e-05,
      "loss": 12.1215,
      "step": 19352
    },
    {
      "epoch": 1.0538481958710257,
      "grad_norm": 0.4965213600162703,
      "learning_rate": 9.613856318025308e-05,
      "loss": 11.9894,
      "step": 19353
    },
    {
      "epoch": 1.0539026498676087,
      "grad_norm": 0.5440742933537539,
      "learning_rate": 9.612975151388401e-05,
      "loss": 12.0088,
      "step": 19354
    },
    {
      "epoch": 1.0539571038641917,
      "grad_norm": 0.6165878924064676,
      "learning_rate": 9.61209398776106e-05,
      "loss": 11.9843,
      "step": 19355
    },
    {
      "epoch": 1.0540115578607747,
      "grad_norm": 0.5577108598869567,
      "learning_rate": 9.61121282715014e-05,
      "loss": 12.0244,
      "step": 19356
    },
    {
      "epoch": 1.0540660118573577,
      "grad_norm": 0.5275445677344449,
      "learning_rate": 9.610331669562495e-05,
      "loss": 12.0602,
      "step": 19357
    },
    {
      "epoch": 1.0541204658539407,
      "grad_norm": 0.5336427825182262,
      "learning_rate": 9.609450515004977e-05,
      "loss": 12.1088,
      "step": 19358
    },
    {
      "epoch": 1.0541749198505237,
      "grad_norm": 0.5544962996409228,
      "learning_rate": 9.608569363484436e-05,
      "loss": 12.0743,
      "step": 19359
    },
    {
      "epoch": 1.0542293738471067,
      "grad_norm": 0.5712553238000525,
      "learning_rate": 9.607688215007728e-05,
      "loss": 12.0085,
      "step": 19360
    },
    {
      "epoch": 1.05428382784369,
      "grad_norm": 0.5872483044221241,
      "learning_rate": 9.6068070695817e-05,
      "loss": 12.0825,
      "step": 19361
    },
    {
      "epoch": 1.054338281840273,
      "grad_norm": 0.5301262680488378,
      "learning_rate": 9.605925927213207e-05,
      "loss": 12.0546,
      "step": 19362
    },
    {
      "epoch": 1.0543927358368559,
      "grad_norm": 0.5941182422544317,
      "learning_rate": 9.605044787909098e-05,
      "loss": 11.9446,
      "step": 19363
    },
    {
      "epoch": 1.0544471898334389,
      "grad_norm": 0.5981957575214851,
      "learning_rate": 9.604163651676232e-05,
      "loss": 11.9136,
      "step": 19364
    },
    {
      "epoch": 1.0545016438300219,
      "grad_norm": 0.5661166063853473,
      "learning_rate": 9.603282518521453e-05,
      "loss": 12.0079,
      "step": 19365
    },
    {
      "epoch": 1.0545560978266049,
      "grad_norm": 0.5799740773054626,
      "learning_rate": 9.602401388451615e-05,
      "loss": 11.96,
      "step": 19366
    },
    {
      "epoch": 1.0546105518231879,
      "grad_norm": 0.5895977707861102,
      "learning_rate": 9.60152026147357e-05,
      "loss": 12.0868,
      "step": 19367
    },
    {
      "epoch": 1.0546650058197709,
      "grad_norm": 0.5138707350489121,
      "learning_rate": 9.60063913759417e-05,
      "loss": 12.1091,
      "step": 19368
    },
    {
      "epoch": 1.0547194598163538,
      "grad_norm": 0.5883943166266282,
      "learning_rate": 9.599758016820269e-05,
      "loss": 11.9937,
      "step": 19369
    },
    {
      "epoch": 1.0547739138129368,
      "grad_norm": 0.5673055198369863,
      "learning_rate": 9.598876899158715e-05,
      "loss": 12.0495,
      "step": 19370
    },
    {
      "epoch": 1.0548283678095198,
      "grad_norm": 0.5766062835107026,
      "learning_rate": 9.597995784616363e-05,
      "loss": 12.0353,
      "step": 19371
    },
    {
      "epoch": 1.0548828218061028,
      "grad_norm": 0.6058495704831871,
      "learning_rate": 9.597114673200062e-05,
      "loss": 12.0471,
      "step": 19372
    },
    {
      "epoch": 1.054937275802686,
      "grad_norm": 0.5255673156686416,
      "learning_rate": 9.596233564916665e-05,
      "loss": 12.0796,
      "step": 19373
    },
    {
      "epoch": 1.054991729799269,
      "grad_norm": 0.6016153150932537,
      "learning_rate": 9.595352459773025e-05,
      "loss": 12.1089,
      "step": 19374
    },
    {
      "epoch": 1.055046183795852,
      "grad_norm": 0.556265804799959,
      "learning_rate": 9.594471357775993e-05,
      "loss": 12.0073,
      "step": 19375
    },
    {
      "epoch": 1.055100637792435,
      "grad_norm": 0.5608704283257167,
      "learning_rate": 9.593590258932417e-05,
      "loss": 11.8563,
      "step": 19376
    },
    {
      "epoch": 1.055155091789018,
      "grad_norm": 0.5781513393423386,
      "learning_rate": 9.592709163249153e-05,
      "loss": 12.1517,
      "step": 19377
    },
    {
      "epoch": 1.055209545785601,
      "grad_norm": 0.5657138345497488,
      "learning_rate": 9.591828070733047e-05,
      "loss": 12.0803,
      "step": 19378
    },
    {
      "epoch": 1.055263999782184,
      "grad_norm": 0.4916113789165632,
      "learning_rate": 9.590946981390958e-05,
      "loss": 12.1075,
      "step": 19379
    },
    {
      "epoch": 1.055318453778767,
      "grad_norm": 0.5017788759661065,
      "learning_rate": 9.590065895229732e-05,
      "loss": 12.0225,
      "step": 19380
    },
    {
      "epoch": 1.05537290777535,
      "grad_norm": 0.6423824315364451,
      "learning_rate": 9.589184812256225e-05,
      "loss": 12.0609,
      "step": 19381
    },
    {
      "epoch": 1.055427361771933,
      "grad_norm": 0.5300731974853954,
      "learning_rate": 9.588303732477283e-05,
      "loss": 12.0407,
      "step": 19382
    },
    {
      "epoch": 1.055481815768516,
      "grad_norm": 0.533408299576154,
      "learning_rate": 9.587422655899762e-05,
      "loss": 12.114,
      "step": 19383
    },
    {
      "epoch": 1.0555362697650992,
      "grad_norm": 0.565065946695608,
      "learning_rate": 9.586541582530514e-05,
      "loss": 12.0338,
      "step": 19384
    },
    {
      "epoch": 1.0555907237616822,
      "grad_norm": 0.5949280397479372,
      "learning_rate": 9.585660512376384e-05,
      "loss": 11.9603,
      "step": 19385
    },
    {
      "epoch": 1.0556451777582652,
      "grad_norm": 0.5447417047295078,
      "learning_rate": 9.58477944544423e-05,
      "loss": 11.9043,
      "step": 19386
    },
    {
      "epoch": 1.0556996317548482,
      "grad_norm": 0.6134685780708565,
      "learning_rate": 9.583898381740898e-05,
      "loss": 12.0248,
      "step": 19387
    },
    {
      "epoch": 1.0557540857514311,
      "grad_norm": 0.5944890783600187,
      "learning_rate": 9.583017321273243e-05,
      "loss": 11.9876,
      "step": 19388
    },
    {
      "epoch": 1.0558085397480141,
      "grad_norm": 0.5340596628731121,
      "learning_rate": 9.582136264048114e-05,
      "loss": 12.1036,
      "step": 19389
    },
    {
      "epoch": 1.0558629937445971,
      "grad_norm": 0.6800523990437837,
      "learning_rate": 9.581255210072364e-05,
      "loss": 12.0343,
      "step": 19390
    },
    {
      "epoch": 1.0559174477411801,
      "grad_norm": 0.5345874595919592,
      "learning_rate": 9.580374159352845e-05,
      "loss": 12.071,
      "step": 19391
    },
    {
      "epoch": 1.0559719017377631,
      "grad_norm": 0.48246863244610083,
      "learning_rate": 9.579493111896406e-05,
      "loss": 11.9095,
      "step": 19392
    },
    {
      "epoch": 1.0560263557343461,
      "grad_norm": 0.564840799839841,
      "learning_rate": 9.5786120677099e-05,
      "loss": 11.8749,
      "step": 19393
    },
    {
      "epoch": 1.056080809730929,
      "grad_norm": 0.5859494509249186,
      "learning_rate": 9.577731026800179e-05,
      "loss": 12.0816,
      "step": 19394
    },
    {
      "epoch": 1.056135263727512,
      "grad_norm": 0.5356116104743887,
      "learning_rate": 9.57684998917409e-05,
      "loss": 11.9958,
      "step": 19395
    },
    {
      "epoch": 1.0561897177240953,
      "grad_norm": 0.602266790571894,
      "learning_rate": 9.575968954838487e-05,
      "loss": 12.1512,
      "step": 19396
    },
    {
      "epoch": 1.0562441717206783,
      "grad_norm": 0.6157757269371643,
      "learning_rate": 9.57508792380022e-05,
      "loss": 12.1429,
      "step": 19397
    },
    {
      "epoch": 1.0562986257172613,
      "grad_norm": 0.5980005309795724,
      "learning_rate": 9.57420689606614e-05,
      "loss": 12.0587,
      "step": 19398
    },
    {
      "epoch": 1.0563530797138443,
      "grad_norm": 0.5908034979075625,
      "learning_rate": 9.5733258716431e-05,
      "loss": 12.1337,
      "step": 19399
    },
    {
      "epoch": 1.0564075337104273,
      "grad_norm": 0.5343902534124586,
      "learning_rate": 9.572444850537948e-05,
      "loss": 11.9634,
      "step": 19400
    },
    {
      "epoch": 1.0564619877070103,
      "grad_norm": 0.5842238118117028,
      "learning_rate": 9.571563832757536e-05,
      "loss": 12.0509,
      "step": 19401
    },
    {
      "epoch": 1.0565164417035933,
      "grad_norm": 0.61275874507477,
      "learning_rate": 9.570682818308715e-05,
      "loss": 12.0451,
      "step": 19402
    },
    {
      "epoch": 1.0565708957001763,
      "grad_norm": 0.5382671440730098,
      "learning_rate": 9.56980180719834e-05,
      "loss": 12.1103,
      "step": 19403
    },
    {
      "epoch": 1.0566253496967593,
      "grad_norm": 0.5133640065547999,
      "learning_rate": 9.568920799433261e-05,
      "loss": 11.9877,
      "step": 19404
    },
    {
      "epoch": 1.0566798036933422,
      "grad_norm": 0.5625235678126909,
      "learning_rate": 9.568039795020319e-05,
      "loss": 12.07,
      "step": 19405
    },
    {
      "epoch": 1.0567342576899252,
      "grad_norm": 0.6284570348095425,
      "learning_rate": 9.567158793966374e-05,
      "loss": 12.1412,
      "step": 19406
    },
    {
      "epoch": 1.0567887116865085,
      "grad_norm": 0.576328893838415,
      "learning_rate": 9.566277796278276e-05,
      "loss": 11.9983,
      "step": 19407
    },
    {
      "epoch": 1.0568431656830914,
      "grad_norm": 0.5531351910635469,
      "learning_rate": 9.565396801962874e-05,
      "loss": 12.0894,
      "step": 19408
    },
    {
      "epoch": 1.0568976196796744,
      "grad_norm": 0.5862286076121059,
      "learning_rate": 9.56451581102702e-05,
      "loss": 12.0713,
      "step": 19409
    },
    {
      "epoch": 1.0569520736762574,
      "grad_norm": 0.49995967750380066,
      "learning_rate": 9.563634823477563e-05,
      "loss": 12.0698,
      "step": 19410
    },
    {
      "epoch": 1.0570065276728404,
      "grad_norm": 0.5108491936131714,
      "learning_rate": 9.562753839321355e-05,
      "loss": 12.0247,
      "step": 19411
    },
    {
      "epoch": 1.0570609816694234,
      "grad_norm": 0.5810846071878304,
      "learning_rate": 9.561872858565245e-05,
      "loss": 12.1781,
      "step": 19412
    },
    {
      "epoch": 1.0571154356660064,
      "grad_norm": 0.5720799743612933,
      "learning_rate": 9.560991881216088e-05,
      "loss": 11.911,
      "step": 19413
    },
    {
      "epoch": 1.0571698896625894,
      "grad_norm": 0.5087805100179055,
      "learning_rate": 9.560110907280734e-05,
      "loss": 11.9394,
      "step": 19414
    },
    {
      "epoch": 1.0572243436591724,
      "grad_norm": 0.5364730959626036,
      "learning_rate": 9.559229936766028e-05,
      "loss": 12.0287,
      "step": 19415
    },
    {
      "epoch": 1.0572787976557554,
      "grad_norm": 0.5284840458321186,
      "learning_rate": 9.558348969678822e-05,
      "loss": 11.9812,
      "step": 19416
    },
    {
      "epoch": 1.0573332516523384,
      "grad_norm": 0.5570402363204487,
      "learning_rate": 9.55746800602597e-05,
      "loss": 12.0449,
      "step": 19417
    },
    {
      "epoch": 1.0573877056489214,
      "grad_norm": 0.6044717380725996,
      "learning_rate": 9.556587045814321e-05,
      "loss": 12.1534,
      "step": 19418
    },
    {
      "epoch": 1.0574421596455046,
      "grad_norm": 0.5526742651145066,
      "learning_rate": 9.555706089050727e-05,
      "loss": 12.1023,
      "step": 19419
    },
    {
      "epoch": 1.0574966136420876,
      "grad_norm": 0.5613785787803098,
      "learning_rate": 9.554825135742037e-05,
      "loss": 12.0306,
      "step": 19420
    },
    {
      "epoch": 1.0575510676386706,
      "grad_norm": 0.5217897384186181,
      "learning_rate": 9.553944185895098e-05,
      "loss": 11.9021,
      "step": 19421
    },
    {
      "epoch": 1.0576055216352536,
      "grad_norm": 0.5383082673278936,
      "learning_rate": 9.553063239516766e-05,
      "loss": 12.214,
      "step": 19422
    },
    {
      "epoch": 1.0576599756318366,
      "grad_norm": 0.5102142305765428,
      "learning_rate": 9.552182296613888e-05,
      "loss": 12.0834,
      "step": 19423
    },
    {
      "epoch": 1.0577144296284196,
      "grad_norm": 0.5830261124845366,
      "learning_rate": 9.55130135719332e-05,
      "loss": 12.0077,
      "step": 19424
    },
    {
      "epoch": 1.0577688836250025,
      "grad_norm": 0.5591125259697896,
      "learning_rate": 9.550420421261905e-05,
      "loss": 12.0461,
      "step": 19425
    },
    {
      "epoch": 1.0578233376215855,
      "grad_norm": 0.5584932430347865,
      "learning_rate": 9.549539488826497e-05,
      "loss": 12.0933,
      "step": 19426
    },
    {
      "epoch": 1.0578777916181685,
      "grad_norm": 0.606281460145651,
      "learning_rate": 9.548658559893942e-05,
      "loss": 12.1265,
      "step": 19427
    },
    {
      "epoch": 1.0579322456147515,
      "grad_norm": 0.5216271627094825,
      "learning_rate": 9.547777634471095e-05,
      "loss": 12.0587,
      "step": 19428
    },
    {
      "epoch": 1.0579866996113345,
      "grad_norm": 0.6148456316966616,
      "learning_rate": 9.546896712564807e-05,
      "loss": 12.3049,
      "step": 19429
    },
    {
      "epoch": 1.0580411536079175,
      "grad_norm": 0.5183933982480431,
      "learning_rate": 9.546015794181925e-05,
      "loss": 12.1366,
      "step": 19430
    },
    {
      "epoch": 1.0580956076045007,
      "grad_norm": 0.5372059927978043,
      "learning_rate": 9.5451348793293e-05,
      "loss": 11.9844,
      "step": 19431
    },
    {
      "epoch": 1.0581500616010837,
      "grad_norm": 0.5360406190489272,
      "learning_rate": 9.544253968013784e-05,
      "loss": 11.9618,
      "step": 19432
    },
    {
      "epoch": 1.0582045155976667,
      "grad_norm": 0.5348052474035279,
      "learning_rate": 9.543373060242225e-05,
      "loss": 12.1427,
      "step": 19433
    },
    {
      "epoch": 1.0582589695942497,
      "grad_norm": 0.5252043494377767,
      "learning_rate": 9.542492156021475e-05,
      "loss": 12.0432,
      "step": 19434
    },
    {
      "epoch": 1.0583134235908327,
      "grad_norm": 0.5148590361233946,
      "learning_rate": 9.541611255358381e-05,
      "loss": 11.8975,
      "step": 19435
    },
    {
      "epoch": 1.0583678775874157,
      "grad_norm": 0.5501657885838557,
      "learning_rate": 9.540730358259795e-05,
      "loss": 12.0395,
      "step": 19436
    },
    {
      "epoch": 1.0584223315839987,
      "grad_norm": 0.5489019089616707,
      "learning_rate": 9.539849464732566e-05,
      "loss": 12.0335,
      "step": 19437
    },
    {
      "epoch": 1.0584767855805817,
      "grad_norm": 0.6431949386899316,
      "learning_rate": 9.538968574783543e-05,
      "loss": 12.0754,
      "step": 19438
    },
    {
      "epoch": 1.0585312395771647,
      "grad_norm": 0.6179053784918723,
      "learning_rate": 9.53808768841958e-05,
      "loss": 12.1651,
      "step": 19439
    },
    {
      "epoch": 1.0585856935737477,
      "grad_norm": 0.529575356971183,
      "learning_rate": 9.537206805647524e-05,
      "loss": 11.9781,
      "step": 19440
    },
    {
      "epoch": 1.0586401475703306,
      "grad_norm": 0.5669109622144446,
      "learning_rate": 9.536325926474227e-05,
      "loss": 12.092,
      "step": 19441
    },
    {
      "epoch": 1.0586946015669136,
      "grad_norm": 0.5637886966334247,
      "learning_rate": 9.535445050906536e-05,
      "loss": 12.0313,
      "step": 19442
    },
    {
      "epoch": 1.0587490555634969,
      "grad_norm": 0.5354476191036248,
      "learning_rate": 9.534564178951302e-05,
      "loss": 11.9178,
      "step": 19443
    },
    {
      "epoch": 1.0588035095600798,
      "grad_norm": 0.5809895908083332,
      "learning_rate": 9.533683310615378e-05,
      "loss": 12.1999,
      "step": 19444
    },
    {
      "epoch": 1.0588579635566628,
      "grad_norm": 0.5156859117911462,
      "learning_rate": 9.532802445905608e-05,
      "loss": 12.0573,
      "step": 19445
    },
    {
      "epoch": 1.0589124175532458,
      "grad_norm": 0.524969301738632,
      "learning_rate": 9.531921584828845e-05,
      "loss": 12.1403,
      "step": 19446
    },
    {
      "epoch": 1.0589668715498288,
      "grad_norm": 0.5506294772032159,
      "learning_rate": 9.531040727391938e-05,
      "loss": 11.8896,
      "step": 19447
    },
    {
      "epoch": 1.0590213255464118,
      "grad_norm": 0.515800456877427,
      "learning_rate": 9.530159873601738e-05,
      "loss": 12.0135,
      "step": 19448
    },
    {
      "epoch": 1.0590757795429948,
      "grad_norm": 0.5434180280161715,
      "learning_rate": 9.529279023465089e-05,
      "loss": 12.0846,
      "step": 19449
    },
    {
      "epoch": 1.0591302335395778,
      "grad_norm": 0.49115390638875817,
      "learning_rate": 9.528398176988849e-05,
      "loss": 11.9628,
      "step": 19450
    },
    {
      "epoch": 1.0591846875361608,
      "grad_norm": 0.5960538189920022,
      "learning_rate": 9.527517334179864e-05,
      "loss": 12.1663,
      "step": 19451
    },
    {
      "epoch": 1.0592391415327438,
      "grad_norm": 0.6018235777881578,
      "learning_rate": 9.526636495044984e-05,
      "loss": 12.0909,
      "step": 19452
    },
    {
      "epoch": 1.0592935955293268,
      "grad_norm": 0.5492354738252602,
      "learning_rate": 9.525755659591057e-05,
      "loss": 12.0349,
      "step": 19453
    },
    {
      "epoch": 1.05934804952591,
      "grad_norm": 0.5318644500366964,
      "learning_rate": 9.524874827824936e-05,
      "loss": 12.1293,
      "step": 19454
    },
    {
      "epoch": 1.059402503522493,
      "grad_norm": 0.5377986794027186,
      "learning_rate": 9.523993999753466e-05,
      "loss": 12.0167,
      "step": 19455
    },
    {
      "epoch": 1.059456957519076,
      "grad_norm": 0.5683297293895401,
      "learning_rate": 9.523113175383498e-05,
      "loss": 12.1201,
      "step": 19456
    },
    {
      "epoch": 1.059511411515659,
      "grad_norm": 0.5281439523592487,
      "learning_rate": 9.522232354721882e-05,
      "loss": 12.0985,
      "step": 19457
    },
    {
      "epoch": 1.059565865512242,
      "grad_norm": 0.5574099817183141,
      "learning_rate": 9.521351537775467e-05,
      "loss": 12.0668,
      "step": 19458
    },
    {
      "epoch": 1.059620319508825,
      "grad_norm": 0.5481013815660997,
      "learning_rate": 9.520470724551104e-05,
      "loss": 12.0165,
      "step": 19459
    },
    {
      "epoch": 1.059674773505408,
      "grad_norm": 0.5258555008954695,
      "learning_rate": 9.51958991505564e-05,
      "loss": 12.1447,
      "step": 19460
    },
    {
      "epoch": 1.059729227501991,
      "grad_norm": 0.5913586523355205,
      "learning_rate": 9.518709109295922e-05,
      "loss": 11.9954,
      "step": 19461
    },
    {
      "epoch": 1.059783681498574,
      "grad_norm": 0.5426523361113232,
      "learning_rate": 9.517828307278807e-05,
      "loss": 12.1235,
      "step": 19462
    },
    {
      "epoch": 1.059838135495157,
      "grad_norm": 0.6265948071629064,
      "learning_rate": 9.51694750901114e-05,
      "loss": 12.0963,
      "step": 19463
    },
    {
      "epoch": 1.05989258949174,
      "grad_norm": 0.5890047993985151,
      "learning_rate": 9.516066714499772e-05,
      "loss": 12.1072,
      "step": 19464
    },
    {
      "epoch": 1.059947043488323,
      "grad_norm": 0.5556535862287739,
      "learning_rate": 9.515185923751547e-05,
      "loss": 12.0839,
      "step": 19465
    },
    {
      "epoch": 1.0600014974849061,
      "grad_norm": 0.5251302373355244,
      "learning_rate": 9.514305136773318e-05,
      "loss": 12.1077,
      "step": 19466
    },
    {
      "epoch": 1.0600559514814891,
      "grad_norm": 0.5924405189573182,
      "learning_rate": 9.513424353571934e-05,
      "loss": 12.1408,
      "step": 19467
    },
    {
      "epoch": 1.0601104054780721,
      "grad_norm": 0.533802556666249,
      "learning_rate": 9.512543574154245e-05,
      "loss": 11.9806,
      "step": 19468
    },
    {
      "epoch": 1.060164859474655,
      "grad_norm": 0.5166721009870676,
      "learning_rate": 9.511662798527096e-05,
      "loss": 12.1354,
      "step": 19469
    },
    {
      "epoch": 1.060219313471238,
      "grad_norm": 0.5407636437450513,
      "learning_rate": 9.510782026697343e-05,
      "loss": 11.9844,
      "step": 19470
    },
    {
      "epoch": 1.060273767467821,
      "grad_norm": 0.5331792045706529,
      "learning_rate": 9.509901258671827e-05,
      "loss": 12.0391,
      "step": 19471
    },
    {
      "epoch": 1.060328221464404,
      "grad_norm": 0.5508183535166898,
      "learning_rate": 9.5090204944574e-05,
      "loss": 12.0949,
      "step": 19472
    },
    {
      "epoch": 1.060382675460987,
      "grad_norm": 0.5688078405882937,
      "learning_rate": 9.508139734060915e-05,
      "loss": 12.086,
      "step": 19473
    },
    {
      "epoch": 1.06043712945757,
      "grad_norm": 0.5733624994938668,
      "learning_rate": 9.507258977489223e-05,
      "loss": 12.0644,
      "step": 19474
    },
    {
      "epoch": 1.060491583454153,
      "grad_norm": 0.5443580440120191,
      "learning_rate": 9.506378224749163e-05,
      "loss": 12.0725,
      "step": 19475
    },
    {
      "epoch": 1.060546037450736,
      "grad_norm": 0.5061060765819786,
      "learning_rate": 9.505497475847586e-05,
      "loss": 12.0051,
      "step": 19476
    },
    {
      "epoch": 1.0606004914473193,
      "grad_norm": 0.571105398310208,
      "learning_rate": 9.504616730791345e-05,
      "loss": 12.1534,
      "step": 19477
    },
    {
      "epoch": 1.0606549454439023,
      "grad_norm": 0.5546744851328956,
      "learning_rate": 9.503735989587289e-05,
      "loss": 12.0144,
      "step": 19478
    },
    {
      "epoch": 1.0607093994404853,
      "grad_norm": 0.5460818596781322,
      "learning_rate": 9.502855252242264e-05,
      "loss": 12.0555,
      "step": 19479
    },
    {
      "epoch": 1.0607638534370682,
      "grad_norm": 0.545806618163049,
      "learning_rate": 9.50197451876312e-05,
      "loss": 11.9989,
      "step": 19480
    },
    {
      "epoch": 1.0608183074336512,
      "grad_norm": 0.5690009421971667,
      "learning_rate": 9.501093789156706e-05,
      "loss": 12.1087,
      "step": 19481
    },
    {
      "epoch": 1.0608727614302342,
      "grad_norm": 0.5247933770362788,
      "learning_rate": 9.50021306342987e-05,
      "loss": 12.0453,
      "step": 19482
    },
    {
      "epoch": 1.0609272154268172,
      "grad_norm": 0.5369592685857938,
      "learning_rate": 9.49933234158946e-05,
      "loss": 11.8929,
      "step": 19483
    },
    {
      "epoch": 1.0609816694234002,
      "grad_norm": 0.5544839201165852,
      "learning_rate": 9.49845162364233e-05,
      "loss": 12.0165,
      "step": 19484
    },
    {
      "epoch": 1.0610361234199832,
      "grad_norm": 0.5717511692715508,
      "learning_rate": 9.497570909595322e-05,
      "loss": 12.03,
      "step": 19485
    },
    {
      "epoch": 1.0610905774165662,
      "grad_norm": 0.5651487710115926,
      "learning_rate": 9.496690199455286e-05,
      "loss": 12.047,
      "step": 19486
    },
    {
      "epoch": 1.0611450314131492,
      "grad_norm": 0.5318155068592443,
      "learning_rate": 9.49580949322907e-05,
      "loss": 12.0402,
      "step": 19487
    },
    {
      "epoch": 1.0611994854097322,
      "grad_norm": 0.5704015631935876,
      "learning_rate": 9.494928790923525e-05,
      "loss": 12.0697,
      "step": 19488
    },
    {
      "epoch": 1.0612539394063154,
      "grad_norm": 0.513769491953973,
      "learning_rate": 9.494048092545499e-05,
      "loss": 11.8972,
      "step": 19489
    },
    {
      "epoch": 1.0613083934028984,
      "grad_norm": 0.5308279098688354,
      "learning_rate": 9.49316739810184e-05,
      "loss": 11.9496,
      "step": 19490
    },
    {
      "epoch": 1.0613628473994814,
      "grad_norm": 0.5652421659166454,
      "learning_rate": 9.492286707599395e-05,
      "loss": 12.0402,
      "step": 19491
    },
    {
      "epoch": 1.0614173013960644,
      "grad_norm": 0.6012514551528869,
      "learning_rate": 9.491406021045016e-05,
      "loss": 12.0058,
      "step": 19492
    },
    {
      "epoch": 1.0614717553926474,
      "grad_norm": 0.6147753858557873,
      "learning_rate": 9.490525338445547e-05,
      "loss": 12.1222,
      "step": 19493
    },
    {
      "epoch": 1.0615262093892304,
      "grad_norm": 0.4778765331953231,
      "learning_rate": 9.489644659807842e-05,
      "loss": 12.003,
      "step": 19494
    },
    {
      "epoch": 1.0615806633858134,
      "grad_norm": 0.5490182527436342,
      "learning_rate": 9.488763985138742e-05,
      "loss": 12.0958,
      "step": 19495
    },
    {
      "epoch": 1.0616351173823964,
      "grad_norm": 0.6310603682268788,
      "learning_rate": 9.487883314445101e-05,
      "loss": 12.0279,
      "step": 19496
    },
    {
      "epoch": 1.0616895713789793,
      "grad_norm": 0.5287781068237357,
      "learning_rate": 9.487002647733763e-05,
      "loss": 11.9563,
      "step": 19497
    },
    {
      "epoch": 1.0617440253755623,
      "grad_norm": 0.5139150948204092,
      "learning_rate": 9.486121985011579e-05,
      "loss": 12.152,
      "step": 19498
    },
    {
      "epoch": 1.0617984793721453,
      "grad_norm": 0.537475871099728,
      "learning_rate": 9.485241326285397e-05,
      "loss": 11.9787,
      "step": 19499
    },
    {
      "epoch": 1.0618529333687283,
      "grad_norm": 0.5714777270599907,
      "learning_rate": 9.484360671562064e-05,
      "loss": 11.9879,
      "step": 19500
    },
    {
      "epoch": 1.0619073873653115,
      "grad_norm": 0.6072144152856551,
      "learning_rate": 9.483480020848431e-05,
      "loss": 12.0891,
      "step": 19501
    },
    {
      "epoch": 1.0619618413618945,
      "grad_norm": 0.5306079632416387,
      "learning_rate": 9.482599374151342e-05,
      "loss": 12.0473,
      "step": 19502
    },
    {
      "epoch": 1.0620162953584775,
      "grad_norm": 0.571602416194433,
      "learning_rate": 9.481718731477648e-05,
      "loss": 12.0993,
      "step": 19503
    },
    {
      "epoch": 1.0620707493550605,
      "grad_norm": 0.5274166829278563,
      "learning_rate": 9.480838092834196e-05,
      "loss": 12.0607,
      "step": 19504
    },
    {
      "epoch": 1.0621252033516435,
      "grad_norm": 0.5308149700387995,
      "learning_rate": 9.479957458227836e-05,
      "loss": 12.0445,
      "step": 19505
    },
    {
      "epoch": 1.0621796573482265,
      "grad_norm": 0.5719164937398165,
      "learning_rate": 9.479076827665413e-05,
      "loss": 12.2446,
      "step": 19506
    },
    {
      "epoch": 1.0622341113448095,
      "grad_norm": 0.5484840373867881,
      "learning_rate": 9.478196201153775e-05,
      "loss": 12.0868,
      "step": 19507
    },
    {
      "epoch": 1.0622885653413925,
      "grad_norm": 0.5174545724860756,
      "learning_rate": 9.47731557869977e-05,
      "loss": 12.0515,
      "step": 19508
    },
    {
      "epoch": 1.0623430193379755,
      "grad_norm": 0.5325860096749648,
      "learning_rate": 9.476434960310248e-05,
      "loss": 12.0681,
      "step": 19509
    },
    {
      "epoch": 1.0623974733345585,
      "grad_norm": 0.5766542534429969,
      "learning_rate": 9.475554345992052e-05,
      "loss": 12.1705,
      "step": 19510
    },
    {
      "epoch": 1.0624519273311415,
      "grad_norm": 0.6170470995755548,
      "learning_rate": 9.474673735752038e-05,
      "loss": 12.1246,
      "step": 19511
    },
    {
      "epoch": 1.0625063813277245,
      "grad_norm": 0.5433823077381589,
      "learning_rate": 9.473793129597047e-05,
      "loss": 12.1055,
      "step": 19512
    },
    {
      "epoch": 1.0625608353243077,
      "grad_norm": 0.5693997414772987,
      "learning_rate": 9.472912527533929e-05,
      "loss": 12.0046,
      "step": 19513
    },
    {
      "epoch": 1.0626152893208907,
      "grad_norm": 0.5397460210874416,
      "learning_rate": 9.472031929569533e-05,
      "loss": 12.1269,
      "step": 19514
    },
    {
      "epoch": 1.0626697433174737,
      "grad_norm": 0.6686676208401305,
      "learning_rate": 9.471151335710706e-05,
      "loss": 12.0209,
      "step": 19515
    },
    {
      "epoch": 1.0627241973140567,
      "grad_norm": 0.590749113523507,
      "learning_rate": 9.470270745964294e-05,
      "loss": 12.106,
      "step": 19516
    },
    {
      "epoch": 1.0627786513106396,
      "grad_norm": 0.5271313705140186,
      "learning_rate": 9.469390160337144e-05,
      "loss": 12.0992,
      "step": 19517
    },
    {
      "epoch": 1.0628331053072226,
      "grad_norm": 0.623477010706432,
      "learning_rate": 9.468509578836105e-05,
      "loss": 12.0731,
      "step": 19518
    },
    {
      "epoch": 1.0628875593038056,
      "grad_norm": 0.6143950276182006,
      "learning_rate": 9.467629001468025e-05,
      "loss": 12.1269,
      "step": 19519
    },
    {
      "epoch": 1.0629420133003886,
      "grad_norm": 0.5382089392313393,
      "learning_rate": 9.466748428239753e-05,
      "loss": 12.0799,
      "step": 19520
    },
    {
      "epoch": 1.0629964672969716,
      "grad_norm": 0.5141725536001891,
      "learning_rate": 9.465867859158131e-05,
      "loss": 12.1073,
      "step": 19521
    },
    {
      "epoch": 1.0630509212935546,
      "grad_norm": 0.5260665309132682,
      "learning_rate": 9.464987294230012e-05,
      "loss": 11.9465,
      "step": 19522
    },
    {
      "epoch": 1.0631053752901376,
      "grad_norm": 0.5607901580338629,
      "learning_rate": 9.464106733462242e-05,
      "loss": 11.952,
      "step": 19523
    },
    {
      "epoch": 1.0631598292867208,
      "grad_norm": 0.5205969784510579,
      "learning_rate": 9.463226176861668e-05,
      "loss": 12.0901,
      "step": 19524
    },
    {
      "epoch": 1.0632142832833038,
      "grad_norm": 0.5739424072770057,
      "learning_rate": 9.46234562443514e-05,
      "loss": 12.0407,
      "step": 19525
    },
    {
      "epoch": 1.0632687372798868,
      "grad_norm": 0.6787278281315261,
      "learning_rate": 9.461465076189499e-05,
      "loss": 12.1001,
      "step": 19526
    },
    {
      "epoch": 1.0633231912764698,
      "grad_norm": 0.541509223022238,
      "learning_rate": 9.460584532131596e-05,
      "loss": 12.0837,
      "step": 19527
    },
    {
      "epoch": 1.0633776452730528,
      "grad_norm": 0.5348501043810973,
      "learning_rate": 9.45970399226828e-05,
      "loss": 12.1429,
      "step": 19528
    },
    {
      "epoch": 1.0634320992696358,
      "grad_norm": 0.5568295755065166,
      "learning_rate": 9.458823456606394e-05,
      "loss": 11.994,
      "step": 19529
    },
    {
      "epoch": 1.0634865532662188,
      "grad_norm": 0.5496086327124223,
      "learning_rate": 9.457942925152788e-05,
      "loss": 12.0708,
      "step": 19530
    },
    {
      "epoch": 1.0635410072628018,
      "grad_norm": 0.4676056507372799,
      "learning_rate": 9.45706239791431e-05,
      "loss": 11.9249,
      "step": 19531
    },
    {
      "epoch": 1.0635954612593848,
      "grad_norm": 0.6022564596587495,
      "learning_rate": 9.456181874897803e-05,
      "loss": 12.0366,
      "step": 19532
    },
    {
      "epoch": 1.0636499152559677,
      "grad_norm": 0.5478683764357231,
      "learning_rate": 9.455301356110119e-05,
      "loss": 12.0423,
      "step": 19533
    },
    {
      "epoch": 1.0637043692525507,
      "grad_norm": 0.6722146474969767,
      "learning_rate": 9.454420841558103e-05,
      "loss": 12.206,
      "step": 19534
    },
    {
      "epoch": 1.0637588232491337,
      "grad_norm": 0.6034418312173684,
      "learning_rate": 9.453540331248607e-05,
      "loss": 12.155,
      "step": 19535
    },
    {
      "epoch": 1.063813277245717,
      "grad_norm": 0.5906543027720849,
      "learning_rate": 9.452659825188466e-05,
      "loss": 12.0325,
      "step": 19536
    },
    {
      "epoch": 1.0638677312423,
      "grad_norm": 0.6082275877099078,
      "learning_rate": 9.451779323384535e-05,
      "loss": 12.0325,
      "step": 19537
    },
    {
      "epoch": 1.063922185238883,
      "grad_norm": 0.5894290320323842,
      "learning_rate": 9.450898825843662e-05,
      "loss": 12.0888,
      "step": 19538
    },
    {
      "epoch": 1.063976639235466,
      "grad_norm": 0.5957966361615813,
      "learning_rate": 9.45001833257269e-05,
      "loss": 12.115,
      "step": 19539
    },
    {
      "epoch": 1.064031093232049,
      "grad_norm": 0.5244965725214917,
      "learning_rate": 9.449137843578469e-05,
      "loss": 12.0886,
      "step": 19540
    },
    {
      "epoch": 1.064085547228632,
      "grad_norm": 0.5694397091625615,
      "learning_rate": 9.448257358867845e-05,
      "loss": 12.0047,
      "step": 19541
    },
    {
      "epoch": 1.064140001225215,
      "grad_norm": 0.5498986514812435,
      "learning_rate": 9.447376878447662e-05,
      "loss": 12.0668,
      "step": 19542
    },
    {
      "epoch": 1.064194455221798,
      "grad_norm": 0.5324227496696069,
      "learning_rate": 9.44649640232477e-05,
      "loss": 12.0771,
      "step": 19543
    },
    {
      "epoch": 1.064248909218381,
      "grad_norm": 0.5567550837917793,
      "learning_rate": 9.445615930506014e-05,
      "loss": 12.1119,
      "step": 19544
    },
    {
      "epoch": 1.0643033632149639,
      "grad_norm": 0.4977052343332378,
      "learning_rate": 9.444735462998248e-05,
      "loss": 12.0982,
      "step": 19545
    },
    {
      "epoch": 1.0643578172115469,
      "grad_norm": 0.5370782475408601,
      "learning_rate": 9.443854999808305e-05,
      "loss": 12.0679,
      "step": 19546
    },
    {
      "epoch": 1.06441227120813,
      "grad_norm": 0.5201187509950764,
      "learning_rate": 9.442974540943039e-05,
      "loss": 12.0481,
      "step": 19547
    },
    {
      "epoch": 1.064466725204713,
      "grad_norm": 0.5978019918197744,
      "learning_rate": 9.442094086409298e-05,
      "loss": 12.0072,
      "step": 19548
    },
    {
      "epoch": 1.064521179201296,
      "grad_norm": 0.491584250591522,
      "learning_rate": 9.441213636213928e-05,
      "loss": 12.0287,
      "step": 19549
    },
    {
      "epoch": 1.064575633197879,
      "grad_norm": 0.5829670045537616,
      "learning_rate": 9.440333190363772e-05,
      "loss": 12.1143,
      "step": 19550
    },
    {
      "epoch": 1.064630087194462,
      "grad_norm": 0.4979406577192298,
      "learning_rate": 9.439452748865682e-05,
      "loss": 11.9809,
      "step": 19551
    },
    {
      "epoch": 1.064684541191045,
      "grad_norm": 0.5641474829559183,
      "learning_rate": 9.438572311726499e-05,
      "loss": 11.993,
      "step": 19552
    },
    {
      "epoch": 1.064738995187628,
      "grad_norm": 0.509601444512635,
      "learning_rate": 9.437691878953074e-05,
      "loss": 12.0681,
      "step": 19553
    },
    {
      "epoch": 1.064793449184211,
      "grad_norm": 0.5973256500028669,
      "learning_rate": 9.436811450552248e-05,
      "loss": 12.0173,
      "step": 19554
    },
    {
      "epoch": 1.064847903180794,
      "grad_norm": 0.48301332542954895,
      "learning_rate": 9.435931026530876e-05,
      "loss": 12.0674,
      "step": 19555
    },
    {
      "epoch": 1.064902357177377,
      "grad_norm": 0.6157513310070719,
      "learning_rate": 9.435050606895796e-05,
      "loss": 12.0662,
      "step": 19556
    },
    {
      "epoch": 1.06495681117396,
      "grad_norm": 0.5259273499843691,
      "learning_rate": 9.434170191653856e-05,
      "loss": 12.1298,
      "step": 19557
    },
    {
      "epoch": 1.065011265170543,
      "grad_norm": 0.533012420826072,
      "learning_rate": 9.433289780811905e-05,
      "loss": 11.9811,
      "step": 19558
    },
    {
      "epoch": 1.0650657191671262,
      "grad_norm": 0.5566711277744718,
      "learning_rate": 9.432409374376783e-05,
      "loss": 12.0475,
      "step": 19559
    },
    {
      "epoch": 1.0651201731637092,
      "grad_norm": 0.5622517247526756,
      "learning_rate": 9.431528972355345e-05,
      "loss": 12.0318,
      "step": 19560
    },
    {
      "epoch": 1.0651746271602922,
      "grad_norm": 0.5727603606850494,
      "learning_rate": 9.430648574754433e-05,
      "loss": 11.9535,
      "step": 19561
    },
    {
      "epoch": 1.0652290811568752,
      "grad_norm": 0.5337964597823531,
      "learning_rate": 9.429768181580894e-05,
      "loss": 11.9784,
      "step": 19562
    },
    {
      "epoch": 1.0652835351534582,
      "grad_norm": 0.598288206977741,
      "learning_rate": 9.428887792841572e-05,
      "loss": 12.0688,
      "step": 19563
    },
    {
      "epoch": 1.0653379891500412,
      "grad_norm": 0.5484710879790106,
      "learning_rate": 9.428007408543314e-05,
      "loss": 12.0379,
      "step": 19564
    },
    {
      "epoch": 1.0653924431466242,
      "grad_norm": 0.6066171274342483,
      "learning_rate": 9.42712702869297e-05,
      "loss": 12.0907,
      "step": 19565
    },
    {
      "epoch": 1.0654468971432072,
      "grad_norm": 0.5275822158642726,
      "learning_rate": 9.426246653297379e-05,
      "loss": 12.0258,
      "step": 19566
    },
    {
      "epoch": 1.0655013511397902,
      "grad_norm": 0.5733642999382029,
      "learning_rate": 9.42536628236339e-05,
      "loss": 12.0058,
      "step": 19567
    },
    {
      "epoch": 1.0655558051363732,
      "grad_norm": 0.47405334283595374,
      "learning_rate": 9.42448591589785e-05,
      "loss": 12.043,
      "step": 19568
    },
    {
      "epoch": 1.0656102591329562,
      "grad_norm": 0.6533793324086261,
      "learning_rate": 9.423605553907603e-05,
      "loss": 12.2868,
      "step": 19569
    },
    {
      "epoch": 1.0656647131295394,
      "grad_norm": 0.5733888302247439,
      "learning_rate": 9.422725196399495e-05,
      "loss": 12.0235,
      "step": 19570
    },
    {
      "epoch": 1.0657191671261224,
      "grad_norm": 0.5723354138462762,
      "learning_rate": 9.421844843380375e-05,
      "loss": 12.1644,
      "step": 19571
    },
    {
      "epoch": 1.0657736211227054,
      "grad_norm": 0.5991701179514328,
      "learning_rate": 9.420964494857085e-05,
      "loss": 12.0841,
      "step": 19572
    },
    {
      "epoch": 1.0658280751192883,
      "grad_norm": 0.558710862217451,
      "learning_rate": 9.420084150836473e-05,
      "loss": 12.0514,
      "step": 19573
    },
    {
      "epoch": 1.0658825291158713,
      "grad_norm": 0.5469144969444083,
      "learning_rate": 9.419203811325383e-05,
      "loss": 12.2402,
      "step": 19574
    },
    {
      "epoch": 1.0659369831124543,
      "grad_norm": 0.5620895209488639,
      "learning_rate": 9.418323476330664e-05,
      "loss": 11.9929,
      "step": 19575
    },
    {
      "epoch": 1.0659914371090373,
      "grad_norm": 0.5949008351746472,
      "learning_rate": 9.417443145859159e-05,
      "loss": 12.1162,
      "step": 19576
    },
    {
      "epoch": 1.0660458911056203,
      "grad_norm": 0.6035968787730213,
      "learning_rate": 9.416562819917712e-05,
      "loss": 12.0362,
      "step": 19577
    },
    {
      "epoch": 1.0661003451022033,
      "grad_norm": 0.5390471506547698,
      "learning_rate": 9.415682498513169e-05,
      "loss": 12.1288,
      "step": 19578
    },
    {
      "epoch": 1.0661547990987863,
      "grad_norm": 0.5114014464809856,
      "learning_rate": 9.414802181652379e-05,
      "loss": 12.0689,
      "step": 19579
    },
    {
      "epoch": 1.0662092530953693,
      "grad_norm": 0.5633583605680194,
      "learning_rate": 9.413921869342185e-05,
      "loss": 12.0915,
      "step": 19580
    },
    {
      "epoch": 1.0662637070919523,
      "grad_norm": 0.5469194789493036,
      "learning_rate": 9.41304156158943e-05,
      "loss": 12.1314,
      "step": 19581
    },
    {
      "epoch": 1.0663181610885353,
      "grad_norm": 0.5300734025899072,
      "learning_rate": 9.412161258400966e-05,
      "loss": 12.055,
      "step": 19582
    },
    {
      "epoch": 1.0663726150851185,
      "grad_norm": 0.5632702110680303,
      "learning_rate": 9.411280959783633e-05,
      "loss": 12.0251,
      "step": 19583
    },
    {
      "epoch": 1.0664270690817015,
      "grad_norm": 0.5470144743292025,
      "learning_rate": 9.410400665744279e-05,
      "loss": 12.028,
      "step": 19584
    },
    {
      "epoch": 1.0664815230782845,
      "grad_norm": 0.6196978189107037,
      "learning_rate": 9.40952037628975e-05,
      "loss": 12.0663,
      "step": 19585
    },
    {
      "epoch": 1.0665359770748675,
      "grad_norm": 0.5865783159570598,
      "learning_rate": 9.408640091426887e-05,
      "loss": 12.1339,
      "step": 19586
    },
    {
      "epoch": 1.0665904310714505,
      "grad_norm": 0.5862795288555418,
      "learning_rate": 9.407759811162539e-05,
      "loss": 12.0186,
      "step": 19587
    },
    {
      "epoch": 1.0666448850680335,
      "grad_norm": 0.619000559758314,
      "learning_rate": 9.406879535503549e-05,
      "loss": 12.2063,
      "step": 19588
    },
    {
      "epoch": 1.0666993390646164,
      "grad_norm": 0.5430701053644917,
      "learning_rate": 9.405999264456763e-05,
      "loss": 12.0267,
      "step": 19589
    },
    {
      "epoch": 1.0667537930611994,
      "grad_norm": 0.5631622351629713,
      "learning_rate": 9.405118998029027e-05,
      "loss": 12.1148,
      "step": 19590
    },
    {
      "epoch": 1.0668082470577824,
      "grad_norm": 0.5454245133367549,
      "learning_rate": 9.404238736227186e-05,
      "loss": 12.12,
      "step": 19591
    },
    {
      "epoch": 1.0668627010543654,
      "grad_norm": 0.5889022062111674,
      "learning_rate": 9.403358479058081e-05,
      "loss": 12.0937,
      "step": 19592
    },
    {
      "epoch": 1.0669171550509484,
      "grad_norm": 0.5065637392389071,
      "learning_rate": 9.402478226528562e-05,
      "loss": 12.0159,
      "step": 19593
    },
    {
      "epoch": 1.0669716090475316,
      "grad_norm": 0.5709809758938452,
      "learning_rate": 9.401597978645474e-05,
      "loss": 12.0441,
      "step": 19594
    },
    {
      "epoch": 1.0670260630441146,
      "grad_norm": 0.5713043581230659,
      "learning_rate": 9.400717735415665e-05,
      "loss": 12.1235,
      "step": 19595
    },
    {
      "epoch": 1.0670805170406976,
      "grad_norm": 0.5593388188584824,
      "learning_rate": 9.399837496845969e-05,
      "loss": 12.026,
      "step": 19596
    },
    {
      "epoch": 1.0671349710372806,
      "grad_norm": 0.5719142697400577,
      "learning_rate": 9.398957262943238e-05,
      "loss": 12.0984,
      "step": 19597
    },
    {
      "epoch": 1.0671894250338636,
      "grad_norm": 0.557749710207554,
      "learning_rate": 9.398077033714317e-05,
      "loss": 12.1424,
      "step": 19598
    },
    {
      "epoch": 1.0672438790304466,
      "grad_norm": 0.5740652249237118,
      "learning_rate": 9.397196809166052e-05,
      "loss": 11.9395,
      "step": 19599
    },
    {
      "epoch": 1.0672983330270296,
      "grad_norm": 0.5598491479754933,
      "learning_rate": 9.396316589305283e-05,
      "loss": 12.0671,
      "step": 19600
    },
    {
      "epoch": 1.0673527870236126,
      "grad_norm": 0.5444547240942728,
      "learning_rate": 9.395436374138857e-05,
      "loss": 12.0203,
      "step": 19601
    },
    {
      "epoch": 1.0674072410201956,
      "grad_norm": 0.5634520180772729,
      "learning_rate": 9.39455616367362e-05,
      "loss": 12.0253,
      "step": 19602
    },
    {
      "epoch": 1.0674616950167786,
      "grad_norm": 0.5152166816141512,
      "learning_rate": 9.393675957916415e-05,
      "loss": 12.0868,
      "step": 19603
    },
    {
      "epoch": 1.0675161490133616,
      "grad_norm": 0.5936970924799284,
      "learning_rate": 9.392795756874088e-05,
      "loss": 12.1438,
      "step": 19604
    },
    {
      "epoch": 1.0675706030099446,
      "grad_norm": 0.6169697993657006,
      "learning_rate": 9.391915560553488e-05,
      "loss": 12.1226,
      "step": 19605
    },
    {
      "epoch": 1.0676250570065278,
      "grad_norm": 0.5925611242588401,
      "learning_rate": 9.391035368961449e-05,
      "loss": 11.9379,
      "step": 19606
    },
    {
      "epoch": 1.0676795110031108,
      "grad_norm": 0.53909222039229,
      "learning_rate": 9.390155182104823e-05,
      "loss": 11.9986,
      "step": 19607
    },
    {
      "epoch": 1.0677339649996938,
      "grad_norm": 0.532438590720537,
      "learning_rate": 9.38927499999045e-05,
      "loss": 11.9909,
      "step": 19608
    },
    {
      "epoch": 1.0677884189962767,
      "grad_norm": 0.4970209959955968,
      "learning_rate": 9.388394822625179e-05,
      "loss": 12.0488,
      "step": 19609
    },
    {
      "epoch": 1.0678428729928597,
      "grad_norm": 0.5998770171857966,
      "learning_rate": 9.387514650015851e-05,
      "loss": 12.19,
      "step": 19610
    },
    {
      "epoch": 1.0678973269894427,
      "grad_norm": 0.552797692326681,
      "learning_rate": 9.386634482169313e-05,
      "loss": 12.1439,
      "step": 19611
    },
    {
      "epoch": 1.0679517809860257,
      "grad_norm": 0.501143711536902,
      "learning_rate": 9.385754319092409e-05,
      "loss": 11.9222,
      "step": 19612
    },
    {
      "epoch": 1.0680062349826087,
      "grad_norm": 0.524867903996969,
      "learning_rate": 9.384874160791981e-05,
      "loss": 11.9817,
      "step": 19613
    },
    {
      "epoch": 1.0680606889791917,
      "grad_norm": 0.4905934833221118,
      "learning_rate": 9.383994007274875e-05,
      "loss": 12.0082,
      "step": 19614
    },
    {
      "epoch": 1.0681151429757747,
      "grad_norm": 0.5459065117734525,
      "learning_rate": 9.383113858547939e-05,
      "loss": 12.0647,
      "step": 19615
    },
    {
      "epoch": 1.0681695969723577,
      "grad_norm": 0.5203237520036577,
      "learning_rate": 9.382233714618008e-05,
      "loss": 11.8745,
      "step": 19616
    },
    {
      "epoch": 1.068224050968941,
      "grad_norm": 0.5411184863716206,
      "learning_rate": 9.381353575491933e-05,
      "loss": 12.0235,
      "step": 19617
    },
    {
      "epoch": 1.068278504965524,
      "grad_norm": 0.517806378865651,
      "learning_rate": 9.380473441176554e-05,
      "loss": 12.0839,
      "step": 19618
    },
    {
      "epoch": 1.068332958962107,
      "grad_norm": 0.5317189778110568,
      "learning_rate": 9.379593311678719e-05,
      "loss": 12.0236,
      "step": 19619
    },
    {
      "epoch": 1.0683874129586899,
      "grad_norm": 0.6585334192721659,
      "learning_rate": 9.378713187005271e-05,
      "loss": 12.1009,
      "step": 19620
    },
    {
      "epoch": 1.0684418669552729,
      "grad_norm": 0.5453745591164424,
      "learning_rate": 9.377833067163052e-05,
      "loss": 12.1599,
      "step": 19621
    },
    {
      "epoch": 1.0684963209518559,
      "grad_norm": 0.5446659967144174,
      "learning_rate": 9.376952952158908e-05,
      "loss": 12.0764,
      "step": 19622
    },
    {
      "epoch": 1.0685507749484389,
      "grad_norm": 0.6447871707801308,
      "learning_rate": 9.376072841999683e-05,
      "loss": 12.1909,
      "step": 19623
    },
    {
      "epoch": 1.0686052289450219,
      "grad_norm": 0.5668415135101286,
      "learning_rate": 9.37519273669222e-05,
      "loss": 11.9781,
      "step": 19624
    },
    {
      "epoch": 1.0686596829416048,
      "grad_norm": 0.5501112084761463,
      "learning_rate": 9.374312636243366e-05,
      "loss": 12.0072,
      "step": 19625
    },
    {
      "epoch": 1.0687141369381878,
      "grad_norm": 0.5165365292243039,
      "learning_rate": 9.373432540659958e-05,
      "loss": 11.9746,
      "step": 19626
    },
    {
      "epoch": 1.0687685909347708,
      "grad_norm": 0.5392483614694529,
      "learning_rate": 9.372552449948845e-05,
      "loss": 12.0356,
      "step": 19627
    },
    {
      "epoch": 1.0688230449313538,
      "grad_norm": 0.5047342672491422,
      "learning_rate": 9.37167236411687e-05,
      "loss": 12.137,
      "step": 19628
    },
    {
      "epoch": 1.068877498927937,
      "grad_norm": 0.511943413415665,
      "learning_rate": 9.370792283170874e-05,
      "loss": 12.0181,
      "step": 19629
    },
    {
      "epoch": 1.06893195292452,
      "grad_norm": 0.5675433984161167,
      "learning_rate": 9.369912207117702e-05,
      "loss": 12.0602,
      "step": 19630
    },
    {
      "epoch": 1.068986406921103,
      "grad_norm": 0.5173814895325556,
      "learning_rate": 9.3690321359642e-05,
      "loss": 12.0295,
      "step": 19631
    },
    {
      "epoch": 1.069040860917686,
      "grad_norm": 0.5028811051182804,
      "learning_rate": 9.368152069717209e-05,
      "loss": 11.9977,
      "step": 19632
    },
    {
      "epoch": 1.069095314914269,
      "grad_norm": 0.5410129930454582,
      "learning_rate": 9.367272008383575e-05,
      "loss": 12.0648,
      "step": 19633
    },
    {
      "epoch": 1.069149768910852,
      "grad_norm": 0.525174263316461,
      "learning_rate": 9.36639195197014e-05,
      "loss": 12.0833,
      "step": 19634
    },
    {
      "epoch": 1.069204222907435,
      "grad_norm": 0.5199187033738576,
      "learning_rate": 9.365511900483749e-05,
      "loss": 11.9703,
      "step": 19635
    },
    {
      "epoch": 1.069258676904018,
      "grad_norm": 0.5663549514936985,
      "learning_rate": 9.364631853931242e-05,
      "loss": 12.0634,
      "step": 19636
    },
    {
      "epoch": 1.069313130900601,
      "grad_norm": 0.539456448316862,
      "learning_rate": 9.363751812319463e-05,
      "loss": 12.0389,
      "step": 19637
    },
    {
      "epoch": 1.069367584897184,
      "grad_norm": 0.6009446430801567,
      "learning_rate": 9.362871775655259e-05,
      "loss": 12.1115,
      "step": 19638
    },
    {
      "epoch": 1.069422038893767,
      "grad_norm": 0.5332866736126232,
      "learning_rate": 9.36199174394547e-05,
      "loss": 12.0605,
      "step": 19639
    },
    {
      "epoch": 1.0694764928903502,
      "grad_norm": 0.4817335547024529,
      "learning_rate": 9.361111717196939e-05,
      "loss": 11.9135,
      "step": 19640
    },
    {
      "epoch": 1.0695309468869332,
      "grad_norm": 0.6048174574996448,
      "learning_rate": 9.360231695416509e-05,
      "loss": 11.8708,
      "step": 19641
    },
    {
      "epoch": 1.0695854008835162,
      "grad_norm": 0.6008411408181712,
      "learning_rate": 9.359351678611027e-05,
      "loss": 12.1587,
      "step": 19642
    },
    {
      "epoch": 1.0696398548800992,
      "grad_norm": 0.5591678637215574,
      "learning_rate": 9.358471666787334e-05,
      "loss": 12.0818,
      "step": 19643
    },
    {
      "epoch": 1.0696943088766822,
      "grad_norm": 0.5587664758147319,
      "learning_rate": 9.357591659952272e-05,
      "loss": 12.1245,
      "step": 19644
    },
    {
      "epoch": 1.0697487628732651,
      "grad_norm": 0.5643472269024349,
      "learning_rate": 9.356711658112685e-05,
      "loss": 12.2275,
      "step": 19645
    },
    {
      "epoch": 1.0698032168698481,
      "grad_norm": 0.5681908835197387,
      "learning_rate": 9.355831661275419e-05,
      "loss": 12.1445,
      "step": 19646
    },
    {
      "epoch": 1.0698576708664311,
      "grad_norm": 0.5659950063808108,
      "learning_rate": 9.354951669447312e-05,
      "loss": 12.0002,
      "step": 19647
    },
    {
      "epoch": 1.0699121248630141,
      "grad_norm": 0.5655190560505993,
      "learning_rate": 9.354071682635208e-05,
      "loss": 12.124,
      "step": 19648
    },
    {
      "epoch": 1.0699665788595971,
      "grad_norm": 0.5745424362482439,
      "learning_rate": 9.353191700845952e-05,
      "loss": 12.0382,
      "step": 19649
    },
    {
      "epoch": 1.07002103285618,
      "grad_norm": 0.559601469047529,
      "learning_rate": 9.352311724086385e-05,
      "loss": 12.0086,
      "step": 19650
    },
    {
      "epoch": 1.070075486852763,
      "grad_norm": 0.6435323007808978,
      "learning_rate": 9.351431752363351e-05,
      "loss": 12.0784,
      "step": 19651
    },
    {
      "epoch": 1.070129940849346,
      "grad_norm": 0.5525163137115525,
      "learning_rate": 9.35055178568369e-05,
      "loss": 12.0001,
      "step": 19652
    },
    {
      "epoch": 1.0701843948459293,
      "grad_norm": 0.6056573547042106,
      "learning_rate": 9.34967182405425e-05,
      "loss": 12.1013,
      "step": 19653
    },
    {
      "epoch": 1.0702388488425123,
      "grad_norm": 0.5763798225266978,
      "learning_rate": 9.34879186748187e-05,
      "loss": 11.9068,
      "step": 19654
    },
    {
      "epoch": 1.0702933028390953,
      "grad_norm": 0.5364143506435711,
      "learning_rate": 9.347911915973394e-05,
      "loss": 11.9713,
      "step": 19655
    },
    {
      "epoch": 1.0703477568356783,
      "grad_norm": 0.5208839549174432,
      "learning_rate": 9.347031969535669e-05,
      "loss": 12.0338,
      "step": 19656
    },
    {
      "epoch": 1.0704022108322613,
      "grad_norm": 0.5959918016455371,
      "learning_rate": 9.346152028175527e-05,
      "loss": 12.0198,
      "step": 19657
    },
    {
      "epoch": 1.0704566648288443,
      "grad_norm": 0.5705087335878948,
      "learning_rate": 9.345272091899818e-05,
      "loss": 11.9041,
      "step": 19658
    },
    {
      "epoch": 1.0705111188254273,
      "grad_norm": 0.5065772424078389,
      "learning_rate": 9.344392160715383e-05,
      "loss": 12.0587,
      "step": 19659
    },
    {
      "epoch": 1.0705655728220103,
      "grad_norm": 0.54577926969694,
      "learning_rate": 9.343512234629064e-05,
      "loss": 12.167,
      "step": 19660
    },
    {
      "epoch": 1.0706200268185933,
      "grad_norm": 0.7992702878849013,
      "learning_rate": 9.342632313647703e-05,
      "loss": 12.2066,
      "step": 19661
    },
    {
      "epoch": 1.0706744808151762,
      "grad_norm": 0.5313366529744018,
      "learning_rate": 9.341752397778146e-05,
      "loss": 11.997,
      "step": 19662
    },
    {
      "epoch": 1.0707289348117595,
      "grad_norm": 0.5810677183782393,
      "learning_rate": 9.340872487027231e-05,
      "loss": 12.1089,
      "step": 19663
    },
    {
      "epoch": 1.0707833888083425,
      "grad_norm": 0.5265076902015842,
      "learning_rate": 9.339992581401801e-05,
      "loss": 12.012,
      "step": 19664
    },
    {
      "epoch": 1.0708378428049254,
      "grad_norm": 0.5335293416359094,
      "learning_rate": 9.339112680908701e-05,
      "loss": 12.0478,
      "step": 19665
    },
    {
      "epoch": 1.0708922968015084,
      "grad_norm": 0.6252614505693662,
      "learning_rate": 9.338232785554776e-05,
      "loss": 12.0812,
      "step": 19666
    },
    {
      "epoch": 1.0709467507980914,
      "grad_norm": 0.5053589716009665,
      "learning_rate": 9.337352895346858e-05,
      "loss": 12.057,
      "step": 19667
    },
    {
      "epoch": 1.0710012047946744,
      "grad_norm": 0.5472130868707699,
      "learning_rate": 9.336473010291795e-05,
      "loss": 12.1641,
      "step": 19668
    },
    {
      "epoch": 1.0710556587912574,
      "grad_norm": 0.6263777805974757,
      "learning_rate": 9.33559313039643e-05,
      "loss": 11.9064,
      "step": 19669
    },
    {
      "epoch": 1.0711101127878404,
      "grad_norm": 0.6581749372028808,
      "learning_rate": 9.334713255667606e-05,
      "loss": 12.1449,
      "step": 19670
    },
    {
      "epoch": 1.0711645667844234,
      "grad_norm": 0.5375481306366297,
      "learning_rate": 9.33383338611216e-05,
      "loss": 12.1026,
      "step": 19671
    },
    {
      "epoch": 1.0712190207810064,
      "grad_norm": 0.5738863975371193,
      "learning_rate": 9.332953521736941e-05,
      "loss": 12.0464,
      "step": 19672
    },
    {
      "epoch": 1.0712734747775894,
      "grad_norm": 0.5699091332528314,
      "learning_rate": 9.332073662548784e-05,
      "loss": 12.1012,
      "step": 19673
    },
    {
      "epoch": 1.0713279287741724,
      "grad_norm": 0.5551052554943668,
      "learning_rate": 9.331193808554538e-05,
      "loss": 12.0756,
      "step": 19674
    },
    {
      "epoch": 1.0713823827707554,
      "grad_norm": 0.5517571887946263,
      "learning_rate": 9.330313959761035e-05,
      "loss": 12.1347,
      "step": 19675
    },
    {
      "epoch": 1.0714368367673386,
      "grad_norm": 0.5738099283189709,
      "learning_rate": 9.329434116175132e-05,
      "loss": 12.0,
      "step": 19676
    },
    {
      "epoch": 1.0714912907639216,
      "grad_norm": 0.6123625032427639,
      "learning_rate": 9.328554277803657e-05,
      "loss": 12.1277,
      "step": 19677
    },
    {
      "epoch": 1.0715457447605046,
      "grad_norm": 0.555374077530613,
      "learning_rate": 9.327674444653456e-05,
      "loss": 12.1037,
      "step": 19678
    },
    {
      "epoch": 1.0716001987570876,
      "grad_norm": 0.5859849512211,
      "learning_rate": 9.326794616731369e-05,
      "loss": 12.2179,
      "step": 19679
    },
    {
      "epoch": 1.0716546527536706,
      "grad_norm": 0.5651225046719894,
      "learning_rate": 9.325914794044243e-05,
      "loss": 12.0778,
      "step": 19680
    },
    {
      "epoch": 1.0717091067502535,
      "grad_norm": 0.5450966905251793,
      "learning_rate": 9.325034976598916e-05,
      "loss": 12.0184,
      "step": 19681
    },
    {
      "epoch": 1.0717635607468365,
      "grad_norm": 0.5909705718224748,
      "learning_rate": 9.32415516440223e-05,
      "loss": 11.9641,
      "step": 19682
    },
    {
      "epoch": 1.0718180147434195,
      "grad_norm": 0.5573469872729605,
      "learning_rate": 9.323275357461028e-05,
      "loss": 11.9704,
      "step": 19683
    },
    {
      "epoch": 1.0718724687400025,
      "grad_norm": 0.5780723863034695,
      "learning_rate": 9.322395555782148e-05,
      "loss": 12.0434,
      "step": 19684
    },
    {
      "epoch": 1.0719269227365855,
      "grad_norm": 0.5458575677248596,
      "learning_rate": 9.321515759372436e-05,
      "loss": 12.0696,
      "step": 19685
    },
    {
      "epoch": 1.0719813767331685,
      "grad_norm": 0.5381775132127713,
      "learning_rate": 9.320635968238732e-05,
      "loss": 11.8941,
      "step": 19686
    },
    {
      "epoch": 1.0720358307297517,
      "grad_norm": 0.5852552369318467,
      "learning_rate": 9.319756182387876e-05,
      "loss": 12.1267,
      "step": 19687
    },
    {
      "epoch": 1.0720902847263347,
      "grad_norm": 0.495255603248781,
      "learning_rate": 9.318876401826708e-05,
      "loss": 12.0333,
      "step": 19688
    },
    {
      "epoch": 1.0721447387229177,
      "grad_norm": 0.6171089323945832,
      "learning_rate": 9.317996626562074e-05,
      "loss": 11.9859,
      "step": 19689
    },
    {
      "epoch": 1.0721991927195007,
      "grad_norm": 0.5408401519883336,
      "learning_rate": 9.317116856600807e-05,
      "loss": 11.9493,
      "step": 19690
    },
    {
      "epoch": 1.0722536467160837,
      "grad_norm": 0.5010606561302176,
      "learning_rate": 9.316237091949758e-05,
      "loss": 11.9302,
      "step": 19691
    },
    {
      "epoch": 1.0723081007126667,
      "grad_norm": 0.5705628538375778,
      "learning_rate": 9.315357332615763e-05,
      "loss": 12.0055,
      "step": 19692
    },
    {
      "epoch": 1.0723625547092497,
      "grad_norm": 0.5563152377068212,
      "learning_rate": 9.314477578605665e-05,
      "loss": 12.1444,
      "step": 19693
    },
    {
      "epoch": 1.0724170087058327,
      "grad_norm": 0.5762005531402592,
      "learning_rate": 9.313597829926306e-05,
      "loss": 12.1609,
      "step": 19694
    },
    {
      "epoch": 1.0724714627024157,
      "grad_norm": 0.5241867847401332,
      "learning_rate": 9.312718086584523e-05,
      "loss": 11.9909,
      "step": 19695
    },
    {
      "epoch": 1.0725259166989987,
      "grad_norm": 0.5294744951317413,
      "learning_rate": 9.311838348587162e-05,
      "loss": 12.113,
      "step": 19696
    },
    {
      "epoch": 1.0725803706955817,
      "grad_norm": 0.48081367114560963,
      "learning_rate": 9.31095861594106e-05,
      "loss": 12.0423,
      "step": 19697
    },
    {
      "epoch": 1.0726348246921646,
      "grad_norm": 0.529069169475715,
      "learning_rate": 9.310078888653059e-05,
      "loss": 12.0427,
      "step": 19698
    },
    {
      "epoch": 1.0726892786887479,
      "grad_norm": 0.5197802975156901,
      "learning_rate": 9.30919916673e-05,
      "loss": 12.0114,
      "step": 19699
    },
    {
      "epoch": 1.0727437326853309,
      "grad_norm": 0.5195023067243941,
      "learning_rate": 9.308319450178724e-05,
      "loss": 11.9279,
      "step": 19700
    },
    {
      "epoch": 1.0727981866819138,
      "grad_norm": 0.5350785743310953,
      "learning_rate": 9.30743973900607e-05,
      "loss": 12.0655,
      "step": 19701
    },
    {
      "epoch": 1.0728526406784968,
      "grad_norm": 0.54928519418668,
      "learning_rate": 9.306560033218883e-05,
      "loss": 12.0851,
      "step": 19702
    },
    {
      "epoch": 1.0729070946750798,
      "grad_norm": 0.5802036630071491,
      "learning_rate": 9.305680332824001e-05,
      "loss": 12.1587,
      "step": 19703
    },
    {
      "epoch": 1.0729615486716628,
      "grad_norm": 0.6146438445621326,
      "learning_rate": 9.304800637828266e-05,
      "loss": 11.8956,
      "step": 19704
    },
    {
      "epoch": 1.0730160026682458,
      "grad_norm": 0.519056383374783,
      "learning_rate": 9.303920948238518e-05,
      "loss": 11.9627,
      "step": 19705
    },
    {
      "epoch": 1.0730704566648288,
      "grad_norm": 0.5166818684415007,
      "learning_rate": 9.3030412640616e-05,
      "loss": 12.0785,
      "step": 19706
    },
    {
      "epoch": 1.0731249106614118,
      "grad_norm": 0.550748553582585,
      "learning_rate": 9.302161585304345e-05,
      "loss": 11.9352,
      "step": 19707
    },
    {
      "epoch": 1.0731793646579948,
      "grad_norm": 0.5569639363473886,
      "learning_rate": 9.301281911973601e-05,
      "loss": 11.9928,
      "step": 19708
    },
    {
      "epoch": 1.0732338186545778,
      "grad_norm": 0.565818920261101,
      "learning_rate": 9.300402244076206e-05,
      "loss": 12.042,
      "step": 19709
    },
    {
      "epoch": 1.073288272651161,
      "grad_norm": 0.5144223908841118,
      "learning_rate": 9.299522581619e-05,
      "loss": 11.9998,
      "step": 19710
    },
    {
      "epoch": 1.073342726647744,
      "grad_norm": 0.5705112569405207,
      "learning_rate": 9.298642924608824e-05,
      "loss": 12.1002,
      "step": 19711
    },
    {
      "epoch": 1.073397180644327,
      "grad_norm": 0.5158525970327664,
      "learning_rate": 9.297763273052517e-05,
      "loss": 12.0533,
      "step": 19712
    },
    {
      "epoch": 1.07345163464091,
      "grad_norm": 0.5697894517000133,
      "learning_rate": 9.29688362695692e-05,
      "loss": 11.9829,
      "step": 19713
    },
    {
      "epoch": 1.073506088637493,
      "grad_norm": 0.5632348411887524,
      "learning_rate": 9.296003986328875e-05,
      "loss": 12.1041,
      "step": 19714
    },
    {
      "epoch": 1.073560542634076,
      "grad_norm": 0.6554950222699993,
      "learning_rate": 9.295124351175222e-05,
      "loss": 12.1818,
      "step": 19715
    },
    {
      "epoch": 1.073614996630659,
      "grad_norm": 0.5770111055480366,
      "learning_rate": 9.294244721502804e-05,
      "loss": 11.9797,
      "step": 19716
    },
    {
      "epoch": 1.073669450627242,
      "grad_norm": 0.528664251870247,
      "learning_rate": 9.293365097318452e-05,
      "loss": 12.0613,
      "step": 19717
    },
    {
      "epoch": 1.073723904623825,
      "grad_norm": 0.6015470481259096,
      "learning_rate": 9.292485478629011e-05,
      "loss": 12.034,
      "step": 19718
    },
    {
      "epoch": 1.073778358620408,
      "grad_norm": 0.5302372269520177,
      "learning_rate": 9.291605865441324e-05,
      "loss": 12.0271,
      "step": 19719
    },
    {
      "epoch": 1.073832812616991,
      "grad_norm": 0.5323550202865295,
      "learning_rate": 9.290726257762228e-05,
      "loss": 12.022,
      "step": 19720
    },
    {
      "epoch": 1.073887266613574,
      "grad_norm": 0.5080055405797005,
      "learning_rate": 9.289846655598564e-05,
      "loss": 11.9097,
      "step": 19721
    },
    {
      "epoch": 1.0739417206101571,
      "grad_norm": 0.5355357619919747,
      "learning_rate": 9.28896705895717e-05,
      "loss": 12.0039,
      "step": 19722
    },
    {
      "epoch": 1.0739961746067401,
      "grad_norm": 0.5721031071639637,
      "learning_rate": 9.288087467844888e-05,
      "loss": 11.9958,
      "step": 19723
    },
    {
      "epoch": 1.0740506286033231,
      "grad_norm": 0.5302677196758343,
      "learning_rate": 9.287207882268556e-05,
      "loss": 11.977,
      "step": 19724
    },
    {
      "epoch": 1.0741050825999061,
      "grad_norm": 0.513418394157247,
      "learning_rate": 9.286328302235015e-05,
      "loss": 12.0055,
      "step": 19725
    },
    {
      "epoch": 1.074159536596489,
      "grad_norm": 0.5404659909582347,
      "learning_rate": 9.285448727751111e-05,
      "loss": 12.1113,
      "step": 19726
    },
    {
      "epoch": 1.074213990593072,
      "grad_norm": 0.5523398496251313,
      "learning_rate": 9.284569158823673e-05,
      "loss": 11.9773,
      "step": 19727
    },
    {
      "epoch": 1.074268444589655,
      "grad_norm": 0.5637373287061666,
      "learning_rate": 9.283689595459544e-05,
      "loss": 12.1073,
      "step": 19728
    },
    {
      "epoch": 1.074322898586238,
      "grad_norm": 0.4880685056636543,
      "learning_rate": 9.282810037665566e-05,
      "loss": 12.0374,
      "step": 19729
    },
    {
      "epoch": 1.074377352582821,
      "grad_norm": 0.5277581048098522,
      "learning_rate": 9.281930485448576e-05,
      "loss": 12.1272,
      "step": 19730
    },
    {
      "epoch": 1.074431806579404,
      "grad_norm": 0.5349540016588795,
      "learning_rate": 9.281050938815416e-05,
      "loss": 11.9587,
      "step": 19731
    },
    {
      "epoch": 1.074486260575987,
      "grad_norm": 0.5629341828892994,
      "learning_rate": 9.280171397772926e-05,
      "loss": 12.0396,
      "step": 19732
    },
    {
      "epoch": 1.0745407145725703,
      "grad_norm": 0.513798232372738,
      "learning_rate": 9.279291862327943e-05,
      "loss": 12.0225,
      "step": 19733
    },
    {
      "epoch": 1.0745951685691533,
      "grad_norm": 0.5353011570995032,
      "learning_rate": 9.278412332487306e-05,
      "loss": 12.1895,
      "step": 19734
    },
    {
      "epoch": 1.0746496225657363,
      "grad_norm": 0.5271662973407337,
      "learning_rate": 9.277532808257855e-05,
      "loss": 12.1193,
      "step": 19735
    },
    {
      "epoch": 1.0747040765623193,
      "grad_norm": 0.5634901604870521,
      "learning_rate": 9.276653289646437e-05,
      "loss": 12.0464,
      "step": 19736
    },
    {
      "epoch": 1.0747585305589022,
      "grad_norm": 0.5648214953375595,
      "learning_rate": 9.275773776659878e-05,
      "loss": 12.1405,
      "step": 19737
    },
    {
      "epoch": 1.0748129845554852,
      "grad_norm": 0.5593975129303271,
      "learning_rate": 9.274894269305025e-05,
      "loss": 11.97,
      "step": 19738
    },
    {
      "epoch": 1.0748674385520682,
      "grad_norm": 0.5080976573341622,
      "learning_rate": 9.274014767588714e-05,
      "loss": 11.9913,
      "step": 19739
    },
    {
      "epoch": 1.0749218925486512,
      "grad_norm": 0.5214783380460148,
      "learning_rate": 9.273135271517787e-05,
      "loss": 12.1178,
      "step": 19740
    },
    {
      "epoch": 1.0749763465452342,
      "grad_norm": 0.5308172619512298,
      "learning_rate": 9.272255781099082e-05,
      "loss": 12.0744,
      "step": 19741
    },
    {
      "epoch": 1.0750308005418172,
      "grad_norm": 0.6115011038726479,
      "learning_rate": 9.271376296339439e-05,
      "loss": 12.1452,
      "step": 19742
    },
    {
      "epoch": 1.0750852545384002,
      "grad_norm": 0.5792122936650925,
      "learning_rate": 9.270496817245696e-05,
      "loss": 11.9555,
      "step": 19743
    },
    {
      "epoch": 1.0751397085349832,
      "grad_norm": 0.5638891886366614,
      "learning_rate": 9.26961734382469e-05,
      "loss": 12.0211,
      "step": 19744
    },
    {
      "epoch": 1.0751941625315662,
      "grad_norm": 0.5014330119294012,
      "learning_rate": 9.268737876083265e-05,
      "loss": 11.9881,
      "step": 19745
    },
    {
      "epoch": 1.0752486165281494,
      "grad_norm": 0.5313886809459771,
      "learning_rate": 9.267858414028257e-05,
      "loss": 12.0741,
      "step": 19746
    },
    {
      "epoch": 1.0753030705247324,
      "grad_norm": 0.5412319133024731,
      "learning_rate": 9.266978957666504e-05,
      "loss": 11.9375,
      "step": 19747
    },
    {
      "epoch": 1.0753575245213154,
      "grad_norm": 0.560446843590823,
      "learning_rate": 9.266099507004844e-05,
      "loss": 12.0861,
      "step": 19748
    },
    {
      "epoch": 1.0754119785178984,
      "grad_norm": 0.5805220017073658,
      "learning_rate": 9.265220062050119e-05,
      "loss": 12.1195,
      "step": 19749
    },
    {
      "epoch": 1.0754664325144814,
      "grad_norm": 0.504645492060845,
      "learning_rate": 9.264340622809163e-05,
      "loss": 11.993,
      "step": 19750
    },
    {
      "epoch": 1.0755208865110644,
      "grad_norm": 0.5617676971852157,
      "learning_rate": 9.263461189288819e-05,
      "loss": 12.0088,
      "step": 19751
    },
    {
      "epoch": 1.0755753405076474,
      "grad_norm": 0.6004664230819657,
      "learning_rate": 9.262581761495926e-05,
      "loss": 12.0947,
      "step": 19752
    },
    {
      "epoch": 1.0756297945042304,
      "grad_norm": 0.5362974399259468,
      "learning_rate": 9.261702339437319e-05,
      "loss": 11.9674,
      "step": 19753
    },
    {
      "epoch": 1.0756842485008133,
      "grad_norm": 0.536708197993625,
      "learning_rate": 9.26082292311984e-05,
      "loss": 12.1122,
      "step": 19754
    },
    {
      "epoch": 1.0757387024973963,
      "grad_norm": 0.6283518034842255,
      "learning_rate": 9.259943512550325e-05,
      "loss": 12.0162,
      "step": 19755
    },
    {
      "epoch": 1.0757931564939793,
      "grad_norm": 0.6175713448932344,
      "learning_rate": 9.259064107735616e-05,
      "loss": 12.0244,
      "step": 19756
    },
    {
      "epoch": 1.0758476104905625,
      "grad_norm": 0.5444838556408125,
      "learning_rate": 9.258184708682546e-05,
      "loss": 12.0978,
      "step": 19757
    },
    {
      "epoch": 1.0759020644871455,
      "grad_norm": 0.5624415326003953,
      "learning_rate": 9.257305315397957e-05,
      "loss": 11.9287,
      "step": 19758
    },
    {
      "epoch": 1.0759565184837285,
      "grad_norm": 0.734909947442244,
      "learning_rate": 9.256425927888687e-05,
      "loss": 12.1097,
      "step": 19759
    },
    {
      "epoch": 1.0760109724803115,
      "grad_norm": 0.52136453544778,
      "learning_rate": 9.255546546161573e-05,
      "loss": 12.0123,
      "step": 19760
    },
    {
      "epoch": 1.0760654264768945,
      "grad_norm": 0.5185993467141862,
      "learning_rate": 9.254667170223453e-05,
      "loss": 12.0284,
      "step": 19761
    },
    {
      "epoch": 1.0761198804734775,
      "grad_norm": 0.5748138736745205,
      "learning_rate": 9.253787800081163e-05,
      "loss": 12.0021,
      "step": 19762
    },
    {
      "epoch": 1.0761743344700605,
      "grad_norm": 0.6109488906780046,
      "learning_rate": 9.25290843574155e-05,
      "loss": 12.1826,
      "step": 19763
    },
    {
      "epoch": 1.0762287884666435,
      "grad_norm": 0.6420538097109626,
      "learning_rate": 9.252029077211444e-05,
      "loss": 12.1369,
      "step": 19764
    },
    {
      "epoch": 1.0762832424632265,
      "grad_norm": 0.8507627928391415,
      "learning_rate": 9.251149724497685e-05,
      "loss": 12.0548,
      "step": 19765
    },
    {
      "epoch": 1.0763376964598095,
      "grad_norm": 0.6035701910113489,
      "learning_rate": 9.250270377607114e-05,
      "loss": 12.0962,
      "step": 19766
    },
    {
      "epoch": 1.0763921504563925,
      "grad_norm": 0.6450246044835068,
      "learning_rate": 9.249391036546564e-05,
      "loss": 12.0165,
      "step": 19767
    },
    {
      "epoch": 1.0764466044529755,
      "grad_norm": 0.697372254910513,
      "learning_rate": 9.248511701322876e-05,
      "loss": 11.9212,
      "step": 19768
    },
    {
      "epoch": 1.0765010584495587,
      "grad_norm": 0.5828957964868062,
      "learning_rate": 9.247632371942885e-05,
      "loss": 12.078,
      "step": 19769
    },
    {
      "epoch": 1.0765555124461417,
      "grad_norm": 0.6521946979037693,
      "learning_rate": 9.246753048413433e-05,
      "loss": 11.9964,
      "step": 19770
    },
    {
      "epoch": 1.0766099664427247,
      "grad_norm": 0.6143374194357522,
      "learning_rate": 9.245873730741356e-05,
      "loss": 11.9747,
      "step": 19771
    },
    {
      "epoch": 1.0766644204393077,
      "grad_norm": 0.5431569085936767,
      "learning_rate": 9.24499441893349e-05,
      "loss": 12.0608,
      "step": 19772
    },
    {
      "epoch": 1.0767188744358906,
      "grad_norm": 0.54729633266687,
      "learning_rate": 9.244115112996671e-05,
      "loss": 12.0335,
      "step": 19773
    },
    {
      "epoch": 1.0767733284324736,
      "grad_norm": 0.5452416566310941,
      "learning_rate": 9.243235812937743e-05,
      "loss": 12.1001,
      "step": 19774
    },
    {
      "epoch": 1.0768277824290566,
      "grad_norm": 0.5694708937075351,
      "learning_rate": 9.24235651876354e-05,
      "loss": 12.0374,
      "step": 19775
    },
    {
      "epoch": 1.0768822364256396,
      "grad_norm": 0.611623532142073,
      "learning_rate": 9.241477230480904e-05,
      "loss": 12.0619,
      "step": 19776
    },
    {
      "epoch": 1.0769366904222226,
      "grad_norm": 0.6025778412048199,
      "learning_rate": 9.240597948096662e-05,
      "loss": 12.1174,
      "step": 19777
    },
    {
      "epoch": 1.0769911444188056,
      "grad_norm": 0.5934774686655594,
      "learning_rate": 9.23971867161766e-05,
      "loss": 11.984,
      "step": 19778
    },
    {
      "epoch": 1.0770455984153886,
      "grad_norm": 0.5202670150688955,
      "learning_rate": 9.238839401050732e-05,
      "loss": 12.087,
      "step": 19779
    },
    {
      "epoch": 1.0771000524119718,
      "grad_norm": 0.5626755625868136,
      "learning_rate": 9.237960136402718e-05,
      "loss": 12.0306,
      "step": 19780
    },
    {
      "epoch": 1.0771545064085548,
      "grad_norm": 0.5285808303053771,
      "learning_rate": 9.237080877680453e-05,
      "loss": 12.0357,
      "step": 19781
    },
    {
      "epoch": 1.0772089604051378,
      "grad_norm": 0.4884744313035158,
      "learning_rate": 9.236201624890776e-05,
      "loss": 11.9146,
      "step": 19782
    },
    {
      "epoch": 1.0772634144017208,
      "grad_norm": 0.5199914902622657,
      "learning_rate": 9.235322378040522e-05,
      "loss": 12.0246,
      "step": 19783
    },
    {
      "epoch": 1.0773178683983038,
      "grad_norm": 0.5091585427844985,
      "learning_rate": 9.234443137136529e-05,
      "loss": 11.8571,
      "step": 19784
    },
    {
      "epoch": 1.0773723223948868,
      "grad_norm": 0.5342246407419041,
      "learning_rate": 9.233563902185633e-05,
      "loss": 12.1116,
      "step": 19785
    },
    {
      "epoch": 1.0774267763914698,
      "grad_norm": 0.5571358933957491,
      "learning_rate": 9.232684673194676e-05,
      "loss": 12.1752,
      "step": 19786
    },
    {
      "epoch": 1.0774812303880528,
      "grad_norm": 0.5312270871426307,
      "learning_rate": 9.231805450170495e-05,
      "loss": 12.0943,
      "step": 19787
    },
    {
      "epoch": 1.0775356843846358,
      "grad_norm": 0.5564105200114353,
      "learning_rate": 9.230926233119918e-05,
      "loss": 12.0351,
      "step": 19788
    },
    {
      "epoch": 1.0775901383812188,
      "grad_norm": 0.5238471065822412,
      "learning_rate": 9.230047022049788e-05,
      "loss": 11.9625,
      "step": 19789
    },
    {
      "epoch": 1.0776445923778017,
      "grad_norm": 0.6448027773800971,
      "learning_rate": 9.229167816966943e-05,
      "loss": 11.8557,
      "step": 19790
    },
    {
      "epoch": 1.0776990463743847,
      "grad_norm": 0.5475912973962153,
      "learning_rate": 9.228288617878217e-05,
      "loss": 11.902,
      "step": 19791
    },
    {
      "epoch": 1.077753500370968,
      "grad_norm": 0.48479578615231733,
      "learning_rate": 9.227409424790449e-05,
      "loss": 11.7981,
      "step": 19792
    },
    {
      "epoch": 1.077807954367551,
      "grad_norm": 0.4969317117219396,
      "learning_rate": 9.226530237710474e-05,
      "loss": 11.9856,
      "step": 19793
    },
    {
      "epoch": 1.077862408364134,
      "grad_norm": 0.5843041316087823,
      "learning_rate": 9.22565105664513e-05,
      "loss": 11.9749,
      "step": 19794
    },
    {
      "epoch": 1.077916862360717,
      "grad_norm": 0.4976117563615809,
      "learning_rate": 9.224771881601252e-05,
      "loss": 12.003,
      "step": 19795
    },
    {
      "epoch": 1.0779713163573,
      "grad_norm": 0.5603380367776644,
      "learning_rate": 9.22389271258568e-05,
      "loss": 11.9834,
      "step": 19796
    },
    {
      "epoch": 1.078025770353883,
      "grad_norm": 0.5859491969938188,
      "learning_rate": 9.223013549605252e-05,
      "loss": 12.2125,
      "step": 19797
    },
    {
      "epoch": 1.078080224350466,
      "grad_norm": 0.5718941113825561,
      "learning_rate": 9.222134392666797e-05,
      "loss": 12.162,
      "step": 19798
    },
    {
      "epoch": 1.078134678347049,
      "grad_norm": 0.4832940013238779,
      "learning_rate": 9.221255241777152e-05,
      "loss": 12.0352,
      "step": 19799
    },
    {
      "epoch": 1.078189132343632,
      "grad_norm": 0.5301172166326132,
      "learning_rate": 9.22037609694316e-05,
      "loss": 12.0537,
      "step": 19800
    },
    {
      "epoch": 1.0782435863402149,
      "grad_norm": 0.5360919272697344,
      "learning_rate": 9.219496958171656e-05,
      "loss": 12.0751,
      "step": 19801
    },
    {
      "epoch": 1.0782980403367979,
      "grad_norm": 0.5142437128335051,
      "learning_rate": 9.218617825469471e-05,
      "loss": 12.0602,
      "step": 19802
    },
    {
      "epoch": 1.078352494333381,
      "grad_norm": 0.573524430103686,
      "learning_rate": 9.217738698843447e-05,
      "loss": 12.0742,
      "step": 19803
    },
    {
      "epoch": 1.078406948329964,
      "grad_norm": 0.5571919577366389,
      "learning_rate": 9.216859578300418e-05,
      "loss": 11.9993,
      "step": 19804
    },
    {
      "epoch": 1.078461402326547,
      "grad_norm": 0.5056902483771932,
      "learning_rate": 9.215980463847221e-05,
      "loss": 12.0203,
      "step": 19805
    },
    {
      "epoch": 1.07851585632313,
      "grad_norm": 0.5503370149624551,
      "learning_rate": 9.215101355490688e-05,
      "loss": 12.0564,
      "step": 19806
    },
    {
      "epoch": 1.078570310319713,
      "grad_norm": 0.613107159969414,
      "learning_rate": 9.214222253237664e-05,
      "loss": 11.9812,
      "step": 19807
    },
    {
      "epoch": 1.078624764316296,
      "grad_norm": 0.5653136291345044,
      "learning_rate": 9.213343157094976e-05,
      "loss": 12.0614,
      "step": 19808
    },
    {
      "epoch": 1.078679218312879,
      "grad_norm": 0.5599890854147888,
      "learning_rate": 9.212464067069464e-05,
      "loss": 12.0691,
      "step": 19809
    },
    {
      "epoch": 1.078733672309462,
      "grad_norm": 0.5443108479194124,
      "learning_rate": 9.211584983167963e-05,
      "loss": 12.139,
      "step": 19810
    },
    {
      "epoch": 1.078788126306045,
      "grad_norm": 0.5315181193854013,
      "learning_rate": 9.210705905397307e-05,
      "loss": 12.0453,
      "step": 19811
    },
    {
      "epoch": 1.078842580302628,
      "grad_norm": 0.5482503737684693,
      "learning_rate": 9.209826833764338e-05,
      "loss": 12.0683,
      "step": 19812
    },
    {
      "epoch": 1.078897034299211,
      "grad_norm": 0.513803567111454,
      "learning_rate": 9.208947768275886e-05,
      "loss": 12.0799,
      "step": 19813
    },
    {
      "epoch": 1.078951488295794,
      "grad_norm": 0.4985326492421896,
      "learning_rate": 9.208068708938791e-05,
      "loss": 12.0466,
      "step": 19814
    },
    {
      "epoch": 1.079005942292377,
      "grad_norm": 0.525518283907291,
      "learning_rate": 9.207189655759885e-05,
      "loss": 11.9293,
      "step": 19815
    },
    {
      "epoch": 1.0790603962889602,
      "grad_norm": 0.5482022188079608,
      "learning_rate": 9.206310608746005e-05,
      "loss": 12.0235,
      "step": 19816
    },
    {
      "epoch": 1.0791148502855432,
      "grad_norm": 0.51705141705295,
      "learning_rate": 9.20543156790399e-05,
      "loss": 12.1233,
      "step": 19817
    },
    {
      "epoch": 1.0791693042821262,
      "grad_norm": 0.5997363988760009,
      "learning_rate": 9.20455253324067e-05,
      "loss": 12.0288,
      "step": 19818
    },
    {
      "epoch": 1.0792237582787092,
      "grad_norm": 0.5943740067115404,
      "learning_rate": 9.203673504762881e-05,
      "loss": 12.0617,
      "step": 19819
    },
    {
      "epoch": 1.0792782122752922,
      "grad_norm": 0.5372814630032169,
      "learning_rate": 9.202794482477464e-05,
      "loss": 11.9594,
      "step": 19820
    },
    {
      "epoch": 1.0793326662718752,
      "grad_norm": 0.5385223976676541,
      "learning_rate": 9.201915466391248e-05,
      "loss": 11.9173,
      "step": 19821
    },
    {
      "epoch": 1.0793871202684582,
      "grad_norm": 0.630652030358931,
      "learning_rate": 9.201036456511069e-05,
      "loss": 12.0707,
      "step": 19822
    },
    {
      "epoch": 1.0794415742650412,
      "grad_norm": 0.5256875959058562,
      "learning_rate": 9.200157452843768e-05,
      "loss": 12.0294,
      "step": 19823
    },
    {
      "epoch": 1.0794960282616242,
      "grad_norm": 0.5809380055850841,
      "learning_rate": 9.199278455396176e-05,
      "loss": 12.1731,
      "step": 19824
    },
    {
      "epoch": 1.0795504822582072,
      "grad_norm": 0.4861395432745022,
      "learning_rate": 9.19839946417513e-05,
      "loss": 11.9302,
      "step": 19825
    },
    {
      "epoch": 1.0796049362547901,
      "grad_norm": 0.5560515010457268,
      "learning_rate": 9.197520479187463e-05,
      "loss": 12.0977,
      "step": 19826
    },
    {
      "epoch": 1.0796593902513734,
      "grad_norm": 0.5504295389034873,
      "learning_rate": 9.196641500440014e-05,
      "loss": 11.9997,
      "step": 19827
    },
    {
      "epoch": 1.0797138442479564,
      "grad_norm": 0.5612610027101926,
      "learning_rate": 9.195762527939613e-05,
      "loss": 12.1131,
      "step": 19828
    },
    {
      "epoch": 1.0797682982445393,
      "grad_norm": 0.5119241540906272,
      "learning_rate": 9.194883561693098e-05,
      "loss": 11.9476,
      "step": 19829
    },
    {
      "epoch": 1.0798227522411223,
      "grad_norm": 0.5424708040339065,
      "learning_rate": 9.194004601707304e-05,
      "loss": 11.9638,
      "step": 19830
    },
    {
      "epoch": 1.0798772062377053,
      "grad_norm": 0.5738978063293178,
      "learning_rate": 9.193125647989064e-05,
      "loss": 12.1109,
      "step": 19831
    },
    {
      "epoch": 1.0799316602342883,
      "grad_norm": 0.6388819616332108,
      "learning_rate": 9.192246700545215e-05,
      "loss": 12.0414,
      "step": 19832
    },
    {
      "epoch": 1.0799861142308713,
      "grad_norm": 0.5863079235322,
      "learning_rate": 9.191367759382587e-05,
      "loss": 12.1259,
      "step": 19833
    },
    {
      "epoch": 1.0800405682274543,
      "grad_norm": 0.5865267660978265,
      "learning_rate": 9.190488824508024e-05,
      "loss": 11.9783,
      "step": 19834
    },
    {
      "epoch": 1.0800950222240373,
      "grad_norm": 0.5942721031349165,
      "learning_rate": 9.189609895928353e-05,
      "loss": 12.07,
      "step": 19835
    },
    {
      "epoch": 1.0801494762206203,
      "grad_norm": 0.5580497812786545,
      "learning_rate": 9.188730973650414e-05,
      "loss": 12.0397,
      "step": 19836
    },
    {
      "epoch": 1.0802039302172033,
      "grad_norm": 0.5574042050126845,
      "learning_rate": 9.18785205768104e-05,
      "loss": 12.1398,
      "step": 19837
    },
    {
      "epoch": 1.0802583842137863,
      "grad_norm": 0.5082332192849419,
      "learning_rate": 9.186973148027063e-05,
      "loss": 11.935,
      "step": 19838
    },
    {
      "epoch": 1.0803128382103695,
      "grad_norm": 0.506636418655922,
      "learning_rate": 9.186094244695318e-05,
      "loss": 12.0906,
      "step": 19839
    },
    {
      "epoch": 1.0803672922069525,
      "grad_norm": 0.5550175858574048,
      "learning_rate": 9.185215347692641e-05,
      "loss": 12.0529,
      "step": 19840
    },
    {
      "epoch": 1.0804217462035355,
      "grad_norm": 0.5306856622699136,
      "learning_rate": 9.184336457025866e-05,
      "loss": 12.0439,
      "step": 19841
    },
    {
      "epoch": 1.0804762002001185,
      "grad_norm": 0.5751168937192283,
      "learning_rate": 9.183457572701828e-05,
      "loss": 12.0151,
      "step": 19842
    },
    {
      "epoch": 1.0805306541967015,
      "grad_norm": 0.5028561230990091,
      "learning_rate": 9.18257869472736e-05,
      "loss": 11.926,
      "step": 19843
    },
    {
      "epoch": 1.0805851081932845,
      "grad_norm": 0.5415314966630953,
      "learning_rate": 9.181699823109296e-05,
      "loss": 12.0274,
      "step": 19844
    },
    {
      "epoch": 1.0806395621898675,
      "grad_norm": 0.5355038008075436,
      "learning_rate": 9.180820957854473e-05,
      "loss": 12.0452,
      "step": 19845
    },
    {
      "epoch": 1.0806940161864504,
      "grad_norm": 0.5102761512889469,
      "learning_rate": 9.179942098969725e-05,
      "loss": 12.0176,
      "step": 19846
    },
    {
      "epoch": 1.0807484701830334,
      "grad_norm": 0.5606942033044104,
      "learning_rate": 9.179063246461887e-05,
      "loss": 12.0285,
      "step": 19847
    },
    {
      "epoch": 1.0808029241796164,
      "grad_norm": 0.5991012448887261,
      "learning_rate": 9.178184400337786e-05,
      "loss": 12.1098,
      "step": 19848
    },
    {
      "epoch": 1.0808573781761994,
      "grad_norm": 0.5367535375112495,
      "learning_rate": 9.177305560604261e-05,
      "loss": 11.9596,
      "step": 19849
    },
    {
      "epoch": 1.0809118321727826,
      "grad_norm": 0.5004396697476243,
      "learning_rate": 9.176426727268148e-05,
      "loss": 12.0197,
      "step": 19850
    },
    {
      "epoch": 1.0809662861693656,
      "grad_norm": 0.5799104998920007,
      "learning_rate": 9.175547900336279e-05,
      "loss": 12.2113,
      "step": 19851
    },
    {
      "epoch": 1.0810207401659486,
      "grad_norm": 0.5749851243210843,
      "learning_rate": 9.174669079815486e-05,
      "loss": 12.086,
      "step": 19852
    },
    {
      "epoch": 1.0810751941625316,
      "grad_norm": 0.5086492814233681,
      "learning_rate": 9.173790265712605e-05,
      "loss": 11.9306,
      "step": 19853
    },
    {
      "epoch": 1.0811296481591146,
      "grad_norm": 0.5366057011087302,
      "learning_rate": 9.172911458034471e-05,
      "loss": 12.0747,
      "step": 19854
    },
    {
      "epoch": 1.0811841021556976,
      "grad_norm": 0.5403876564087411,
      "learning_rate": 9.172032656787913e-05,
      "loss": 11.9863,
      "step": 19855
    },
    {
      "epoch": 1.0812385561522806,
      "grad_norm": 0.5950587307869056,
      "learning_rate": 9.171153861979769e-05,
      "loss": 12.1185,
      "step": 19856
    },
    {
      "epoch": 1.0812930101488636,
      "grad_norm": 0.6135176279796881,
      "learning_rate": 9.170275073616877e-05,
      "loss": 12.0065,
      "step": 19857
    },
    {
      "epoch": 1.0813474641454466,
      "grad_norm": 0.5083830167400131,
      "learning_rate": 9.169396291706061e-05,
      "loss": 11.8831,
      "step": 19858
    },
    {
      "epoch": 1.0814019181420296,
      "grad_norm": 0.5703483834579192,
      "learning_rate": 9.168517516254158e-05,
      "loss": 12.0295,
      "step": 19859
    },
    {
      "epoch": 1.0814563721386126,
      "grad_norm": 0.6074469083028137,
      "learning_rate": 9.167638747268002e-05,
      "loss": 12.0883,
      "step": 19860
    },
    {
      "epoch": 1.0815108261351956,
      "grad_norm": 0.5191785966022419,
      "learning_rate": 9.166759984754428e-05,
      "loss": 12.0195,
      "step": 19861
    },
    {
      "epoch": 1.0815652801317788,
      "grad_norm": 0.5605617016454646,
      "learning_rate": 9.165881228720267e-05,
      "loss": 12.1311,
      "step": 19862
    },
    {
      "epoch": 1.0816197341283618,
      "grad_norm": 0.5390198212079476,
      "learning_rate": 9.165002479172354e-05,
      "loss": 12.1054,
      "step": 19863
    },
    {
      "epoch": 1.0816741881249448,
      "grad_norm": 0.5659962448785403,
      "learning_rate": 9.164123736117523e-05,
      "loss": 12.1215,
      "step": 19864
    },
    {
      "epoch": 1.0817286421215278,
      "grad_norm": 0.5921829218396504,
      "learning_rate": 9.163244999562604e-05,
      "loss": 12.0794,
      "step": 19865
    },
    {
      "epoch": 1.0817830961181107,
      "grad_norm": 0.5304361683476481,
      "learning_rate": 9.162366269514432e-05,
      "loss": 12.1065,
      "step": 19866
    },
    {
      "epoch": 1.0818375501146937,
      "grad_norm": 0.5807102749268369,
      "learning_rate": 9.161487545979844e-05,
      "loss": 12.0703,
      "step": 19867
    },
    {
      "epoch": 1.0818920041112767,
      "grad_norm": 0.5276436081169286,
      "learning_rate": 9.160608828965666e-05,
      "loss": 11.9785,
      "step": 19868
    },
    {
      "epoch": 1.0819464581078597,
      "grad_norm": 0.5661969343941075,
      "learning_rate": 9.159730118478737e-05,
      "loss": 12.0878,
      "step": 19869
    },
    {
      "epoch": 1.0820009121044427,
      "grad_norm": 0.5362217126007136,
      "learning_rate": 9.158851414525886e-05,
      "loss": 11.8735,
      "step": 19870
    },
    {
      "epoch": 1.0820553661010257,
      "grad_norm": 0.5508481037378804,
      "learning_rate": 9.157972717113945e-05,
      "loss": 12.1429,
      "step": 19871
    },
    {
      "epoch": 1.0821098200976087,
      "grad_norm": 0.562342947545884,
      "learning_rate": 9.15709402624975e-05,
      "loss": 12.1656,
      "step": 19872
    },
    {
      "epoch": 1.082164274094192,
      "grad_norm": 0.500928640737876,
      "learning_rate": 9.156215341940136e-05,
      "loss": 12.081,
      "step": 19873
    },
    {
      "epoch": 1.082218728090775,
      "grad_norm": 0.4966378820375146,
      "learning_rate": 9.155336664191932e-05,
      "loss": 12.1059,
      "step": 19874
    },
    {
      "epoch": 1.082273182087358,
      "grad_norm": 0.5528438165037192,
      "learning_rate": 9.154457993011972e-05,
      "loss": 11.9159,
      "step": 19875
    },
    {
      "epoch": 1.082327636083941,
      "grad_norm": 0.5484776550801352,
      "learning_rate": 9.153579328407088e-05,
      "loss": 11.9934,
      "step": 19876
    },
    {
      "epoch": 1.0823820900805239,
      "grad_norm": 0.5575050542309941,
      "learning_rate": 9.152700670384116e-05,
      "loss": 12.0534,
      "step": 19877
    },
    {
      "epoch": 1.0824365440771069,
      "grad_norm": 0.5426585602757926,
      "learning_rate": 9.151822018949881e-05,
      "loss": 12.0549,
      "step": 19878
    },
    {
      "epoch": 1.0824909980736899,
      "grad_norm": 0.6123470415390558,
      "learning_rate": 9.150943374111222e-05,
      "loss": 11.9764,
      "step": 19879
    },
    {
      "epoch": 1.0825454520702729,
      "grad_norm": 0.5821696164344822,
      "learning_rate": 9.15006473587497e-05,
      "loss": 12.0439,
      "step": 19880
    },
    {
      "epoch": 1.0825999060668559,
      "grad_norm": 0.5430943652819135,
      "learning_rate": 9.149186104247958e-05,
      "loss": 12.1013,
      "step": 19881
    },
    {
      "epoch": 1.0826543600634388,
      "grad_norm": 0.6327773300958239,
      "learning_rate": 9.148307479237014e-05,
      "loss": 12.0013,
      "step": 19882
    },
    {
      "epoch": 1.0827088140600218,
      "grad_norm": 0.518913600156964,
      "learning_rate": 9.147428860848977e-05,
      "loss": 12.094,
      "step": 19883
    },
    {
      "epoch": 1.0827632680566048,
      "grad_norm": 0.5645919083115407,
      "learning_rate": 9.146550249090675e-05,
      "loss": 11.8218,
      "step": 19884
    },
    {
      "epoch": 1.0828177220531878,
      "grad_norm": 0.5441439766884042,
      "learning_rate": 9.145671643968942e-05,
      "loss": 12.0403,
      "step": 19885
    },
    {
      "epoch": 1.082872176049771,
      "grad_norm": 0.5940575936371498,
      "learning_rate": 9.14479304549061e-05,
      "loss": 12.0329,
      "step": 19886
    },
    {
      "epoch": 1.082926630046354,
      "grad_norm": 0.6043211414787795,
      "learning_rate": 9.143914453662512e-05,
      "loss": 12.0626,
      "step": 19887
    },
    {
      "epoch": 1.082981084042937,
      "grad_norm": 0.541390862693401,
      "learning_rate": 9.143035868491476e-05,
      "loss": 12.048,
      "step": 19888
    },
    {
      "epoch": 1.08303553803952,
      "grad_norm": 0.531445579848505,
      "learning_rate": 9.142157289984337e-05,
      "loss": 12.0976,
      "step": 19889
    },
    {
      "epoch": 1.083089992036103,
      "grad_norm": 0.513369597722478,
      "learning_rate": 9.141278718147929e-05,
      "loss": 11.9035,
      "step": 19890
    },
    {
      "epoch": 1.083144446032686,
      "grad_norm": 0.5891037711902334,
      "learning_rate": 9.140400152989079e-05,
      "loss": 12.0648,
      "step": 19891
    },
    {
      "epoch": 1.083198900029269,
      "grad_norm": 0.5543595431977636,
      "learning_rate": 9.139521594514623e-05,
      "loss": 12.0995,
      "step": 19892
    },
    {
      "epoch": 1.083253354025852,
      "grad_norm": 0.5438529356820038,
      "learning_rate": 9.138643042731389e-05,
      "loss": 12.0272,
      "step": 19893
    },
    {
      "epoch": 1.083307808022435,
      "grad_norm": 0.5467179470834783,
      "learning_rate": 9.137764497646213e-05,
      "loss": 12.1245,
      "step": 19894
    },
    {
      "epoch": 1.083362262019018,
      "grad_norm": 0.4989757185609027,
      "learning_rate": 9.136885959265926e-05,
      "loss": 12.0709,
      "step": 19895
    },
    {
      "epoch": 1.083416716015601,
      "grad_norm": 0.54996556335291,
      "learning_rate": 9.136007427597358e-05,
      "loss": 11.9735,
      "step": 19896
    },
    {
      "epoch": 1.0834711700121842,
      "grad_norm": 0.5670794449956671,
      "learning_rate": 9.135128902647344e-05,
      "loss": 12.0571,
      "step": 19897
    },
    {
      "epoch": 1.0835256240087672,
      "grad_norm": 0.5164709777065184,
      "learning_rate": 9.134250384422708e-05,
      "loss": 12.0582,
      "step": 19898
    },
    {
      "epoch": 1.0835800780053502,
      "grad_norm": 0.5613349135182275,
      "learning_rate": 9.133371872930288e-05,
      "loss": 12.0784,
      "step": 19899
    },
    {
      "epoch": 1.0836345320019332,
      "grad_norm": 0.5563800171494543,
      "learning_rate": 9.132493368176913e-05,
      "loss": 12.08,
      "step": 19900
    },
    {
      "epoch": 1.0836889859985162,
      "grad_norm": 0.5206282421692374,
      "learning_rate": 9.131614870169416e-05,
      "loss": 12.1176,
      "step": 19901
    },
    {
      "epoch": 1.0837434399950991,
      "grad_norm": 0.5123339744790743,
      "learning_rate": 9.130736378914627e-05,
      "loss": 12.1279,
      "step": 19902
    },
    {
      "epoch": 1.0837978939916821,
      "grad_norm": 0.5556191526608115,
      "learning_rate": 9.129857894419377e-05,
      "loss": 11.9749,
      "step": 19903
    },
    {
      "epoch": 1.0838523479882651,
      "grad_norm": 0.5499176741013869,
      "learning_rate": 9.128979416690497e-05,
      "loss": 11.9974,
      "step": 19904
    },
    {
      "epoch": 1.0839068019848481,
      "grad_norm": 0.49136687224716824,
      "learning_rate": 9.12810094573482e-05,
      "loss": 11.9636,
      "step": 19905
    },
    {
      "epoch": 1.0839612559814311,
      "grad_norm": 0.5667764010346874,
      "learning_rate": 9.127222481559176e-05,
      "loss": 12.1023,
      "step": 19906
    },
    {
      "epoch": 1.084015709978014,
      "grad_norm": 0.5326825165723065,
      "learning_rate": 9.126344024170402e-05,
      "loss": 12.0809,
      "step": 19907
    },
    {
      "epoch": 1.084070163974597,
      "grad_norm": 0.5370302873191101,
      "learning_rate": 9.125465573575316e-05,
      "loss": 12.0619,
      "step": 19908
    },
    {
      "epoch": 1.0841246179711803,
      "grad_norm": 0.5560791563957217,
      "learning_rate": 9.12458712978076e-05,
      "loss": 12.0018,
      "step": 19909
    },
    {
      "epoch": 1.0841790719677633,
      "grad_norm": 0.5058743738074365,
      "learning_rate": 9.12370869279356e-05,
      "loss": 12.0217,
      "step": 19910
    },
    {
      "epoch": 1.0842335259643463,
      "grad_norm": 0.5990810630139142,
      "learning_rate": 9.122830262620547e-05,
      "loss": 12.1044,
      "step": 19911
    },
    {
      "epoch": 1.0842879799609293,
      "grad_norm": 0.5410580811988184,
      "learning_rate": 9.121951839268556e-05,
      "loss": 11.9739,
      "step": 19912
    },
    {
      "epoch": 1.0843424339575123,
      "grad_norm": 0.5568019602049695,
      "learning_rate": 9.121073422744413e-05,
      "loss": 12.1421,
      "step": 19913
    },
    {
      "epoch": 1.0843968879540953,
      "grad_norm": 0.5569345235690699,
      "learning_rate": 9.12019501305495e-05,
      "loss": 11.8679,
      "step": 19914
    },
    {
      "epoch": 1.0844513419506783,
      "grad_norm": 0.5183950269777624,
      "learning_rate": 9.119316610206998e-05,
      "loss": 12.0276,
      "step": 19915
    },
    {
      "epoch": 1.0845057959472613,
      "grad_norm": 0.5064129400906593,
      "learning_rate": 9.118438214207387e-05,
      "loss": 12.1255,
      "step": 19916
    },
    {
      "epoch": 1.0845602499438443,
      "grad_norm": 0.6021752218151646,
      "learning_rate": 9.117559825062953e-05,
      "loss": 12.0236,
      "step": 19917
    },
    {
      "epoch": 1.0846147039404272,
      "grad_norm": 0.5157551802055335,
      "learning_rate": 9.116681442780519e-05,
      "loss": 11.9216,
      "step": 19918
    },
    {
      "epoch": 1.0846691579370102,
      "grad_norm": 0.5436451505997936,
      "learning_rate": 9.115803067366918e-05,
      "loss": 12.1423,
      "step": 19919
    },
    {
      "epoch": 1.0847236119335935,
      "grad_norm": 0.5973381170371647,
      "learning_rate": 9.114924698828978e-05,
      "loss": 12.0881,
      "step": 19920
    },
    {
      "epoch": 1.0847780659301764,
      "grad_norm": 0.5206950377698941,
      "learning_rate": 9.114046337173534e-05,
      "loss": 12.1696,
      "step": 19921
    },
    {
      "epoch": 1.0848325199267594,
      "grad_norm": 0.5343122491422189,
      "learning_rate": 9.113167982407414e-05,
      "loss": 12.0945,
      "step": 19922
    },
    {
      "epoch": 1.0848869739233424,
      "grad_norm": 0.5405889694086997,
      "learning_rate": 9.11228963453745e-05,
      "loss": 12.1126,
      "step": 19923
    },
    {
      "epoch": 1.0849414279199254,
      "grad_norm": 0.6113488144750007,
      "learning_rate": 9.11141129357047e-05,
      "loss": 11.9141,
      "step": 19924
    },
    {
      "epoch": 1.0849958819165084,
      "grad_norm": 0.5452989191660556,
      "learning_rate": 9.110532959513304e-05,
      "loss": 12.0923,
      "step": 19925
    },
    {
      "epoch": 1.0850503359130914,
      "grad_norm": 0.5544969239297456,
      "learning_rate": 9.109654632372784e-05,
      "loss": 12.088,
      "step": 19926
    },
    {
      "epoch": 1.0851047899096744,
      "grad_norm": 0.6536920844173818,
      "learning_rate": 9.10877631215574e-05,
      "loss": 12.079,
      "step": 19927
    },
    {
      "epoch": 1.0851592439062574,
      "grad_norm": 0.5271331003375122,
      "learning_rate": 9.107897998869e-05,
      "loss": 12.1115,
      "step": 19928
    },
    {
      "epoch": 1.0852136979028404,
      "grad_norm": 0.49071808736334793,
      "learning_rate": 9.107019692519393e-05,
      "loss": 11.9722,
      "step": 19929
    },
    {
      "epoch": 1.0852681518994234,
      "grad_norm": 0.6372190248928601,
      "learning_rate": 9.106141393113752e-05,
      "loss": 12.0689,
      "step": 19930
    },
    {
      "epoch": 1.0853226058960064,
      "grad_norm": 0.5980830574240124,
      "learning_rate": 9.105263100658902e-05,
      "loss": 12.0512,
      "step": 19931
    },
    {
      "epoch": 1.0853770598925896,
      "grad_norm": 0.5493450308605987,
      "learning_rate": 9.10438481516168e-05,
      "loss": 12.0894,
      "step": 19932
    },
    {
      "epoch": 1.0854315138891726,
      "grad_norm": 0.6556303700259074,
      "learning_rate": 9.10350653662891e-05,
      "loss": 12.1051,
      "step": 19933
    },
    {
      "epoch": 1.0854859678857556,
      "grad_norm": 0.5177009660696416,
      "learning_rate": 9.102628265067425e-05,
      "loss": 11.9703,
      "step": 19934
    },
    {
      "epoch": 1.0855404218823386,
      "grad_norm": 0.5046612680720217,
      "learning_rate": 9.101750000484052e-05,
      "loss": 12.0555,
      "step": 19935
    },
    {
      "epoch": 1.0855948758789216,
      "grad_norm": 0.5705460610848567,
      "learning_rate": 9.100871742885622e-05,
      "loss": 11.8679,
      "step": 19936
    },
    {
      "epoch": 1.0856493298755046,
      "grad_norm": 0.5619400076838802,
      "learning_rate": 9.099993492278965e-05,
      "loss": 12.1102,
      "step": 19937
    },
    {
      "epoch": 1.0857037838720875,
      "grad_norm": 0.5665782307003279,
      "learning_rate": 9.09911524867091e-05,
      "loss": 12.0628,
      "step": 19938
    },
    {
      "epoch": 1.0857582378686705,
      "grad_norm": 0.5035529583625047,
      "learning_rate": 9.098237012068286e-05,
      "loss": 11.9725,
      "step": 19939
    },
    {
      "epoch": 1.0858126918652535,
      "grad_norm": 0.5110637395954961,
      "learning_rate": 9.09735878247792e-05,
      "loss": 11.9253,
      "step": 19940
    },
    {
      "epoch": 1.0858671458618365,
      "grad_norm": 0.539311907986121,
      "learning_rate": 9.096480559906645e-05,
      "loss": 12.0985,
      "step": 19941
    },
    {
      "epoch": 1.0859215998584195,
      "grad_norm": 0.6409624825959144,
      "learning_rate": 9.095602344361286e-05,
      "loss": 12.023,
      "step": 19942
    },
    {
      "epoch": 1.0859760538550027,
      "grad_norm": 0.5882799322296413,
      "learning_rate": 9.094724135848677e-05,
      "loss": 12.0073,
      "step": 19943
    },
    {
      "epoch": 1.0860305078515857,
      "grad_norm": 0.546907140635376,
      "learning_rate": 9.093845934375645e-05,
      "loss": 11.8597,
      "step": 19944
    },
    {
      "epoch": 1.0860849618481687,
      "grad_norm": 0.5879607208007526,
      "learning_rate": 9.092967739949019e-05,
      "loss": 11.9852,
      "step": 19945
    },
    {
      "epoch": 1.0861394158447517,
      "grad_norm": 0.5450063255763744,
      "learning_rate": 9.092089552575628e-05,
      "loss": 12.0052,
      "step": 19946
    },
    {
      "epoch": 1.0861938698413347,
      "grad_norm": 0.5314778428648255,
      "learning_rate": 9.0912113722623e-05,
      "loss": 12.0105,
      "step": 19947
    },
    {
      "epoch": 1.0862483238379177,
      "grad_norm": 0.5585952771522082,
      "learning_rate": 9.090333199015868e-05,
      "loss": 12.2198,
      "step": 19948
    },
    {
      "epoch": 1.0863027778345007,
      "grad_norm": 0.5732565799501155,
      "learning_rate": 9.089455032843155e-05,
      "loss": 12.0317,
      "step": 19949
    },
    {
      "epoch": 1.0863572318310837,
      "grad_norm": 0.5765505756080281,
      "learning_rate": 9.088576873750992e-05,
      "loss": 12.2115,
      "step": 19950
    },
    {
      "epoch": 1.0864116858276667,
      "grad_norm": 0.5513923575328709,
      "learning_rate": 9.087698721746209e-05,
      "loss": 12.103,
      "step": 19951
    },
    {
      "epoch": 1.0864661398242497,
      "grad_norm": 0.5820435479206023,
      "learning_rate": 9.086820576835634e-05,
      "loss": 12.0404,
      "step": 19952
    },
    {
      "epoch": 1.0865205938208327,
      "grad_norm": 0.6022803029144533,
      "learning_rate": 9.085942439026092e-05,
      "loss": 12.0666,
      "step": 19953
    },
    {
      "epoch": 1.0865750478174157,
      "grad_norm": 0.5037362536968005,
      "learning_rate": 9.085064308324418e-05,
      "loss": 11.9326,
      "step": 19954
    },
    {
      "epoch": 1.0866295018139986,
      "grad_norm": 0.5629619369554842,
      "learning_rate": 9.084186184737437e-05,
      "loss": 12.075,
      "step": 19955
    },
    {
      "epoch": 1.0866839558105819,
      "grad_norm": 0.7293732390996722,
      "learning_rate": 9.083308068271977e-05,
      "loss": 12.0475,
      "step": 19956
    },
    {
      "epoch": 1.0867384098071649,
      "grad_norm": 0.6011848720393436,
      "learning_rate": 9.082429958934869e-05,
      "loss": 11.9534,
      "step": 19957
    },
    {
      "epoch": 1.0867928638037478,
      "grad_norm": 0.6314019833993705,
      "learning_rate": 9.08155185673294e-05,
      "loss": 12.1621,
      "step": 19958
    },
    {
      "epoch": 1.0868473178003308,
      "grad_norm": 0.5321012509526907,
      "learning_rate": 9.080673761673016e-05,
      "loss": 11.9822,
      "step": 19959
    },
    {
      "epoch": 1.0869017717969138,
      "grad_norm": 0.5358458161111209,
      "learning_rate": 9.079795673761927e-05,
      "loss": 12.0879,
      "step": 19960
    },
    {
      "epoch": 1.0869562257934968,
      "grad_norm": 0.5516116169410465,
      "learning_rate": 9.078917593006502e-05,
      "loss": 12.0031,
      "step": 19961
    },
    {
      "epoch": 1.0870106797900798,
      "grad_norm": 0.5659572430474552,
      "learning_rate": 9.078039519413567e-05,
      "loss": 12.1379,
      "step": 19962
    },
    {
      "epoch": 1.0870651337866628,
      "grad_norm": 0.5606247963131835,
      "learning_rate": 9.077161452989952e-05,
      "loss": 11.9215,
      "step": 19963
    },
    {
      "epoch": 1.0871195877832458,
      "grad_norm": 0.5748818529484704,
      "learning_rate": 9.076283393742484e-05,
      "loss": 11.9057,
      "step": 19964
    },
    {
      "epoch": 1.0871740417798288,
      "grad_norm": 0.6397330396077927,
      "learning_rate": 9.075405341677989e-05,
      "loss": 12.0713,
      "step": 19965
    },
    {
      "epoch": 1.087228495776412,
      "grad_norm": 0.5587576937613958,
      "learning_rate": 9.0745272968033e-05,
      "loss": 12.0238,
      "step": 19966
    },
    {
      "epoch": 1.087282949772995,
      "grad_norm": 0.5372686915159008,
      "learning_rate": 9.073649259125242e-05,
      "loss": 11.7746,
      "step": 19967
    },
    {
      "epoch": 1.087337403769578,
      "grad_norm": 0.5633905760899219,
      "learning_rate": 9.072771228650646e-05,
      "loss": 11.9728,
      "step": 19968
    },
    {
      "epoch": 1.087391857766161,
      "grad_norm": 0.5949007043908271,
      "learning_rate": 9.071893205386331e-05,
      "loss": 12.115,
      "step": 19969
    },
    {
      "epoch": 1.087446311762744,
      "grad_norm": 0.5807917281893856,
      "learning_rate": 9.071015189339131e-05,
      "loss": 11.9429,
      "step": 19970
    },
    {
      "epoch": 1.087500765759327,
      "grad_norm": 0.5756543238824169,
      "learning_rate": 9.070137180515875e-05,
      "loss": 12.053,
      "step": 19971
    },
    {
      "epoch": 1.08755521975591,
      "grad_norm": 0.5297283885852245,
      "learning_rate": 9.069259178923386e-05,
      "loss": 11.8516,
      "step": 19972
    },
    {
      "epoch": 1.087609673752493,
      "grad_norm": 0.5551320857290687,
      "learning_rate": 9.068381184568494e-05,
      "loss": 12.0597,
      "step": 19973
    },
    {
      "epoch": 1.087664127749076,
      "grad_norm": 0.5625803842858186,
      "learning_rate": 9.067503197458027e-05,
      "loss": 11.9823,
      "step": 19974
    },
    {
      "epoch": 1.087718581745659,
      "grad_norm": 0.5795953180190437,
      "learning_rate": 9.066625217598812e-05,
      "loss": 12.1733,
      "step": 19975
    },
    {
      "epoch": 1.087773035742242,
      "grad_norm": 0.5473935443239211,
      "learning_rate": 9.06574724499767e-05,
      "loss": 12.0024,
      "step": 19976
    },
    {
      "epoch": 1.087827489738825,
      "grad_norm": 0.5667668328190282,
      "learning_rate": 9.064869279661439e-05,
      "loss": 11.9631,
      "step": 19977
    },
    {
      "epoch": 1.087881943735408,
      "grad_norm": 0.5266704522455978,
      "learning_rate": 9.063991321596948e-05,
      "loss": 12.045,
      "step": 19978
    },
    {
      "epoch": 1.0879363977319911,
      "grad_norm": 0.5844567425175714,
      "learning_rate": 9.063113370811009e-05,
      "loss": 12.139,
      "step": 19979
    },
    {
      "epoch": 1.0879908517285741,
      "grad_norm": 0.5825710730257588,
      "learning_rate": 9.062235427310457e-05,
      "loss": 12.0948,
      "step": 19980
    },
    {
      "epoch": 1.0880453057251571,
      "grad_norm": 0.6299771286484094,
      "learning_rate": 9.061357491102122e-05,
      "loss": 12.0189,
      "step": 19981
    },
    {
      "epoch": 1.0880997597217401,
      "grad_norm": 0.5363310138015335,
      "learning_rate": 9.060479562192829e-05,
      "loss": 11.8844,
      "step": 19982
    },
    {
      "epoch": 1.088154213718323,
      "grad_norm": 0.5707402502112603,
      "learning_rate": 9.059601640589403e-05,
      "loss": 12.0814,
      "step": 19983
    },
    {
      "epoch": 1.088208667714906,
      "grad_norm": 0.5639069057481662,
      "learning_rate": 9.058723726298673e-05,
      "loss": 11.9652,
      "step": 19984
    },
    {
      "epoch": 1.088263121711489,
      "grad_norm": 0.738216354293828,
      "learning_rate": 9.057845819327466e-05,
      "loss": 12.0537,
      "step": 19985
    },
    {
      "epoch": 1.088317575708072,
      "grad_norm": 0.5441504601844046,
      "learning_rate": 9.056967919682608e-05,
      "loss": 12.0143,
      "step": 19986
    },
    {
      "epoch": 1.088372029704655,
      "grad_norm": 0.6504410720166462,
      "learning_rate": 9.056090027370923e-05,
      "loss": 12.1874,
      "step": 19987
    },
    {
      "epoch": 1.088426483701238,
      "grad_norm": 0.6195252656867826,
      "learning_rate": 9.055212142399245e-05,
      "loss": 12.0596,
      "step": 19988
    },
    {
      "epoch": 1.088480937697821,
      "grad_norm": 0.5673916365171481,
      "learning_rate": 9.054334264774394e-05,
      "loss": 12.1216,
      "step": 19989
    },
    {
      "epoch": 1.0885353916944043,
      "grad_norm": 0.5562905035399163,
      "learning_rate": 9.053456394503197e-05,
      "loss": 11.995,
      "step": 19990
    },
    {
      "epoch": 1.0885898456909873,
      "grad_norm": 0.6864049145436131,
      "learning_rate": 9.052578531592479e-05,
      "loss": 11.9271,
      "step": 19991
    },
    {
      "epoch": 1.0886442996875703,
      "grad_norm": 0.554609993264955,
      "learning_rate": 9.051700676049073e-05,
      "loss": 11.9662,
      "step": 19992
    },
    {
      "epoch": 1.0886987536841533,
      "grad_norm": 0.5390644716978882,
      "learning_rate": 9.050822827879801e-05,
      "loss": 11.9687,
      "step": 19993
    },
    {
      "epoch": 1.0887532076807362,
      "grad_norm": 0.6197942093873294,
      "learning_rate": 9.04994498709149e-05,
      "loss": 11.9846,
      "step": 19994
    },
    {
      "epoch": 1.0888076616773192,
      "grad_norm": 0.5532516314359079,
      "learning_rate": 9.049067153690965e-05,
      "loss": 12.0499,
      "step": 19995
    },
    {
      "epoch": 1.0888621156739022,
      "grad_norm": 0.5730859665342325,
      "learning_rate": 9.048189327685055e-05,
      "loss": 12.216,
      "step": 19996
    },
    {
      "epoch": 1.0889165696704852,
      "grad_norm": 0.5500219671907687,
      "learning_rate": 9.047311509080584e-05,
      "loss": 12.0319,
      "step": 19997
    },
    {
      "epoch": 1.0889710236670682,
      "grad_norm": 0.5921074346229978,
      "learning_rate": 9.04643369788438e-05,
      "loss": 12.1958,
      "step": 19998
    },
    {
      "epoch": 1.0890254776636512,
      "grad_norm": 0.6028458643295544,
      "learning_rate": 9.045555894103265e-05,
      "loss": 11.9421,
      "step": 19999
    },
    {
      "epoch": 1.0890799316602342,
      "grad_norm": 0.5073641549341452,
      "learning_rate": 9.044678097744068e-05,
      "loss": 12.0276,
      "step": 20000
    },
    {
      "epoch": 1.0891343856568172,
      "grad_norm": 0.6065992995903755,
      "learning_rate": 9.043800308813614e-05,
      "loss": 12.1435,
      "step": 20001
    },
    {
      "epoch": 1.0891888396534004,
      "grad_norm": 0.5643750133310085,
      "learning_rate": 9.042922527318728e-05,
      "loss": 12.0401,
      "step": 20002
    },
    {
      "epoch": 1.0892432936499834,
      "grad_norm": 0.6550555665516066,
      "learning_rate": 9.042044753266238e-05,
      "loss": 12.1827,
      "step": 20003
    },
    {
      "epoch": 1.0892977476465664,
      "grad_norm": 0.4899034343825559,
      "learning_rate": 9.04116698666297e-05,
      "loss": 12.0562,
      "step": 20004
    },
    {
      "epoch": 1.0893522016431494,
      "grad_norm": 0.6185593649271679,
      "learning_rate": 9.040289227515745e-05,
      "loss": 12.0924,
      "step": 20005
    },
    {
      "epoch": 1.0894066556397324,
      "grad_norm": 0.570663308596088,
      "learning_rate": 9.039411475831395e-05,
      "loss": 11.9767,
      "step": 20006
    },
    {
      "epoch": 1.0894611096363154,
      "grad_norm": 0.5268150422309706,
      "learning_rate": 9.038533731616741e-05,
      "loss": 12.0193,
      "step": 20007
    },
    {
      "epoch": 1.0895155636328984,
      "grad_norm": 0.5258947574793097,
      "learning_rate": 9.037655994878614e-05,
      "loss": 12.1018,
      "step": 20008
    },
    {
      "epoch": 1.0895700176294814,
      "grad_norm": 0.575274588880538,
      "learning_rate": 9.036778265623832e-05,
      "loss": 11.9754,
      "step": 20009
    },
    {
      "epoch": 1.0896244716260644,
      "grad_norm": 0.5802939253572615,
      "learning_rate": 9.035900543859224e-05,
      "loss": 12.0168,
      "step": 20010
    },
    {
      "epoch": 1.0896789256226473,
      "grad_norm": 0.5122724331256518,
      "learning_rate": 9.035022829591613e-05,
      "loss": 11.9511,
      "step": 20011
    },
    {
      "epoch": 1.0897333796192303,
      "grad_norm": 0.4769146188379691,
      "learning_rate": 9.03414512282783e-05,
      "loss": 11.9888,
      "step": 20012
    },
    {
      "epoch": 1.0897878336158136,
      "grad_norm": 0.5265505779491405,
      "learning_rate": 9.033267423574693e-05,
      "loss": 11.9005,
      "step": 20013
    },
    {
      "epoch": 1.0898422876123965,
      "grad_norm": 0.5333724765858119,
      "learning_rate": 9.032389731839031e-05,
      "loss": 12.0755,
      "step": 20014
    },
    {
      "epoch": 1.0898967416089795,
      "grad_norm": 0.5193352601069574,
      "learning_rate": 9.03151204762767e-05,
      "loss": 12.1701,
      "step": 20015
    },
    {
      "epoch": 1.0899511956055625,
      "grad_norm": 0.49464098071032003,
      "learning_rate": 9.030634370947433e-05,
      "loss": 12.0579,
      "step": 20016
    },
    {
      "epoch": 1.0900056496021455,
      "grad_norm": 0.568910824507174,
      "learning_rate": 9.029756701805147e-05,
      "loss": 12.0425,
      "step": 20017
    },
    {
      "epoch": 1.0900601035987285,
      "grad_norm": 0.5670081222423411,
      "learning_rate": 9.028879040207638e-05,
      "loss": 11.9655,
      "step": 20018
    },
    {
      "epoch": 1.0901145575953115,
      "grad_norm": 0.5209534940742387,
      "learning_rate": 9.028001386161724e-05,
      "loss": 12.0386,
      "step": 20019
    },
    {
      "epoch": 1.0901690115918945,
      "grad_norm": 0.5257222292295778,
      "learning_rate": 9.027123739674236e-05,
      "loss": 12.1165,
      "step": 20020
    },
    {
      "epoch": 1.0902234655884775,
      "grad_norm": 0.562692073697708,
      "learning_rate": 9.026246100751996e-05,
      "loss": 12.1326,
      "step": 20021
    },
    {
      "epoch": 1.0902779195850605,
      "grad_norm": 0.5356455064032006,
      "learning_rate": 9.02536846940183e-05,
      "loss": 12.1093,
      "step": 20022
    },
    {
      "epoch": 1.0903323735816435,
      "grad_norm": 0.6178913806743136,
      "learning_rate": 9.024490845630564e-05,
      "loss": 12.0731,
      "step": 20023
    },
    {
      "epoch": 1.0903868275782265,
      "grad_norm": 0.5086643420023287,
      "learning_rate": 9.023613229445018e-05,
      "loss": 12.0523,
      "step": 20024
    },
    {
      "epoch": 1.0904412815748097,
      "grad_norm": 0.5219656284534158,
      "learning_rate": 9.022735620852019e-05,
      "loss": 11.7372,
      "step": 20025
    },
    {
      "epoch": 1.0904957355713927,
      "grad_norm": 0.5162106532165133,
      "learning_rate": 9.021858019858393e-05,
      "loss": 11.9562,
      "step": 20026
    },
    {
      "epoch": 1.0905501895679757,
      "grad_norm": 0.5000648210155602,
      "learning_rate": 9.020980426470963e-05,
      "loss": 11.9445,
      "step": 20027
    },
    {
      "epoch": 1.0906046435645587,
      "grad_norm": 0.534566777693027,
      "learning_rate": 9.020102840696558e-05,
      "loss": 12.0463,
      "step": 20028
    },
    {
      "epoch": 1.0906590975611417,
      "grad_norm": 0.5624082349442754,
      "learning_rate": 9.01922526254199e-05,
      "loss": 11.9304,
      "step": 20029
    },
    {
      "epoch": 1.0907135515577246,
      "grad_norm": 0.551904556369177,
      "learning_rate": 9.018347692014095e-05,
      "loss": 11.9993,
      "step": 20030
    },
    {
      "epoch": 1.0907680055543076,
      "grad_norm": 0.5942305762390805,
      "learning_rate": 9.017470129119692e-05,
      "loss": 12.092,
      "step": 20031
    },
    {
      "epoch": 1.0908224595508906,
      "grad_norm": 0.5024243307789622,
      "learning_rate": 9.016592573865606e-05,
      "loss": 12.0821,
      "step": 20032
    },
    {
      "epoch": 1.0908769135474736,
      "grad_norm": 0.52087478799685,
      "learning_rate": 9.01571502625866e-05,
      "loss": 12.0875,
      "step": 20033
    },
    {
      "epoch": 1.0909313675440566,
      "grad_norm": 0.49630439482687516,
      "learning_rate": 9.014837486305682e-05,
      "loss": 12.0044,
      "step": 20034
    },
    {
      "epoch": 1.0909858215406396,
      "grad_norm": 0.5379210037700731,
      "learning_rate": 9.013959954013492e-05,
      "loss": 12.0714,
      "step": 20035
    },
    {
      "epoch": 1.0910402755372228,
      "grad_norm": 0.5401713498441595,
      "learning_rate": 9.01308242938891e-05,
      "loss": 11.8857,
      "step": 20036
    },
    {
      "epoch": 1.0910947295338058,
      "grad_norm": 0.5356267210334869,
      "learning_rate": 9.012204912438769e-05,
      "loss": 12.1584,
      "step": 20037
    },
    {
      "epoch": 1.0911491835303888,
      "grad_norm": 0.5685396376260402,
      "learning_rate": 9.011327403169891e-05,
      "loss": 12.0256,
      "step": 20038
    },
    {
      "epoch": 1.0912036375269718,
      "grad_norm": 0.6397245614974457,
      "learning_rate": 9.010449901589094e-05,
      "loss": 12.0091,
      "step": 20039
    },
    {
      "epoch": 1.0912580915235548,
      "grad_norm": 0.5338238370933304,
      "learning_rate": 9.009572407703201e-05,
      "loss": 11.9764,
      "step": 20040
    },
    {
      "epoch": 1.0913125455201378,
      "grad_norm": 0.5576473803063768,
      "learning_rate": 9.008694921519044e-05,
      "loss": 11.9742,
      "step": 20041
    },
    {
      "epoch": 1.0913669995167208,
      "grad_norm": 0.5860825346362053,
      "learning_rate": 9.00781744304344e-05,
      "loss": 12.0667,
      "step": 20042
    },
    {
      "epoch": 1.0914214535133038,
      "grad_norm": 0.5144441272948119,
      "learning_rate": 9.006939972283213e-05,
      "loss": 11.9995,
      "step": 20043
    },
    {
      "epoch": 1.0914759075098868,
      "grad_norm": 0.5611980194921362,
      "learning_rate": 9.006062509245188e-05,
      "loss": 12.1328,
      "step": 20044
    },
    {
      "epoch": 1.0915303615064698,
      "grad_norm": 0.5431351711512413,
      "learning_rate": 9.005185053936186e-05,
      "loss": 12.0958,
      "step": 20045
    },
    {
      "epoch": 1.0915848155030528,
      "grad_norm": 0.6682242249022382,
      "learning_rate": 9.004307606363033e-05,
      "loss": 12.0687,
      "step": 20046
    },
    {
      "epoch": 1.0916392694996357,
      "grad_norm": 0.5671285226395476,
      "learning_rate": 9.00343016653255e-05,
      "loss": 11.931,
      "step": 20047
    },
    {
      "epoch": 1.0916937234962187,
      "grad_norm": 0.5021756500744794,
      "learning_rate": 9.002552734451566e-05,
      "loss": 11.944,
      "step": 20048
    },
    {
      "epoch": 1.091748177492802,
      "grad_norm": 0.5825252285222624,
      "learning_rate": 9.001675310126897e-05,
      "loss": 12.0314,
      "step": 20049
    },
    {
      "epoch": 1.091802631489385,
      "grad_norm": 0.608147473703928,
      "learning_rate": 9.000797893565367e-05,
      "loss": 12.222,
      "step": 20050
    },
    {
      "epoch": 1.091857085485968,
      "grad_norm": 0.5639127355682383,
      "learning_rate": 8.999920484773798e-05,
      "loss": 11.8921,
      "step": 20051
    },
    {
      "epoch": 1.091911539482551,
      "grad_norm": 0.5647725739831364,
      "learning_rate": 8.999043083759017e-05,
      "loss": 11.9675,
      "step": 20052
    },
    {
      "epoch": 1.091965993479134,
      "grad_norm": 0.5480108093055222,
      "learning_rate": 8.998165690527846e-05,
      "loss": 12.0654,
      "step": 20053
    },
    {
      "epoch": 1.092020447475717,
      "grad_norm": 0.6067634857647759,
      "learning_rate": 8.997288305087104e-05,
      "loss": 12.0903,
      "step": 20054
    },
    {
      "epoch": 1.0920749014723,
      "grad_norm": 0.5585022925172852,
      "learning_rate": 8.996410927443619e-05,
      "loss": 11.9205,
      "step": 20055
    },
    {
      "epoch": 1.092129355468883,
      "grad_norm": 0.5333906347604619,
      "learning_rate": 8.99553355760421e-05,
      "loss": 12.0811,
      "step": 20056
    },
    {
      "epoch": 1.092183809465466,
      "grad_norm": 0.5919546618318178,
      "learning_rate": 8.994656195575699e-05,
      "loss": 12.0538,
      "step": 20057
    },
    {
      "epoch": 1.0922382634620489,
      "grad_norm": 0.5386646282360394,
      "learning_rate": 8.993778841364915e-05,
      "loss": 12.0325,
      "step": 20058
    },
    {
      "epoch": 1.0922927174586319,
      "grad_norm": 0.5627313682717267,
      "learning_rate": 8.992901494978671e-05,
      "loss": 12.0302,
      "step": 20059
    },
    {
      "epoch": 1.092347171455215,
      "grad_norm": 0.66080896523592,
      "learning_rate": 8.992024156423795e-05,
      "loss": 12.0468,
      "step": 20060
    },
    {
      "epoch": 1.092401625451798,
      "grad_norm": 0.5483386811294285,
      "learning_rate": 8.991146825707107e-05,
      "loss": 12.0382,
      "step": 20061
    },
    {
      "epoch": 1.092456079448381,
      "grad_norm": 0.5698024277752847,
      "learning_rate": 8.99026950283543e-05,
      "loss": 12.0269,
      "step": 20062
    },
    {
      "epoch": 1.092510533444964,
      "grad_norm": 0.5034213710347741,
      "learning_rate": 8.989392187815587e-05,
      "loss": 11.9569,
      "step": 20063
    },
    {
      "epoch": 1.092564987441547,
      "grad_norm": 0.5252457182569386,
      "learning_rate": 8.988514880654402e-05,
      "loss": 12.0722,
      "step": 20064
    },
    {
      "epoch": 1.09261944143813,
      "grad_norm": 0.5485424380270981,
      "learning_rate": 8.987637581358693e-05,
      "loss": 12.1475,
      "step": 20065
    },
    {
      "epoch": 1.092673895434713,
      "grad_norm": 0.5563476844267424,
      "learning_rate": 8.986760289935285e-05,
      "loss": 12.0159,
      "step": 20066
    },
    {
      "epoch": 1.092728349431296,
      "grad_norm": 0.5586996576538787,
      "learning_rate": 8.985883006390999e-05,
      "loss": 12.0525,
      "step": 20067
    },
    {
      "epoch": 1.092782803427879,
      "grad_norm": 0.5824339016281458,
      "learning_rate": 8.985005730732658e-05,
      "loss": 12.0977,
      "step": 20068
    },
    {
      "epoch": 1.092837257424462,
      "grad_norm": 0.5784505890381453,
      "learning_rate": 8.984128462967081e-05,
      "loss": 12.0047,
      "step": 20069
    },
    {
      "epoch": 1.092891711421045,
      "grad_norm": 0.52087938675949,
      "learning_rate": 8.983251203101092e-05,
      "loss": 12.0008,
      "step": 20070
    },
    {
      "epoch": 1.092946165417628,
      "grad_norm": 0.6487274880530316,
      "learning_rate": 8.982373951141511e-05,
      "loss": 12.0405,
      "step": 20071
    },
    {
      "epoch": 1.0930006194142112,
      "grad_norm": 0.5389941667346818,
      "learning_rate": 8.981496707095162e-05,
      "loss": 12.1138,
      "step": 20072
    },
    {
      "epoch": 1.0930550734107942,
      "grad_norm": 0.5928294869034778,
      "learning_rate": 8.980619470968865e-05,
      "loss": 11.8825,
      "step": 20073
    },
    {
      "epoch": 1.0931095274073772,
      "grad_norm": 0.5458634722915138,
      "learning_rate": 8.97974224276944e-05,
      "loss": 11.9042,
      "step": 20074
    },
    {
      "epoch": 1.0931639814039602,
      "grad_norm": 0.5636647294429769,
      "learning_rate": 8.978865022503712e-05,
      "loss": 12.1298,
      "step": 20075
    },
    {
      "epoch": 1.0932184354005432,
      "grad_norm": 0.5938311159587132,
      "learning_rate": 8.9779878101785e-05,
      "loss": 12.1066,
      "step": 20076
    },
    {
      "epoch": 1.0932728893971262,
      "grad_norm": 0.5960768360565969,
      "learning_rate": 8.977110605800628e-05,
      "loss": 12.0255,
      "step": 20077
    },
    {
      "epoch": 1.0933273433937092,
      "grad_norm": 0.576698533575002,
      "learning_rate": 8.976233409376916e-05,
      "loss": 11.9549,
      "step": 20078
    },
    {
      "epoch": 1.0933817973902922,
      "grad_norm": 0.6343789106509875,
      "learning_rate": 8.975356220914184e-05,
      "loss": 11.9698,
      "step": 20079
    },
    {
      "epoch": 1.0934362513868752,
      "grad_norm": 0.5420024279986068,
      "learning_rate": 8.974479040419254e-05,
      "loss": 11.9848,
      "step": 20080
    },
    {
      "epoch": 1.0934907053834582,
      "grad_norm": 0.580584398832918,
      "learning_rate": 8.973601867898946e-05,
      "loss": 12.1149,
      "step": 20081
    },
    {
      "epoch": 1.0935451593800412,
      "grad_norm": 0.6080394022495313,
      "learning_rate": 8.972724703360083e-05,
      "loss": 12.0821,
      "step": 20082
    },
    {
      "epoch": 1.0935996133766244,
      "grad_norm": 0.6053854430232528,
      "learning_rate": 8.971847546809482e-05,
      "loss": 11.8476,
      "step": 20083
    },
    {
      "epoch": 1.0936540673732074,
      "grad_norm": 0.5395166590259115,
      "learning_rate": 8.970970398253971e-05,
      "loss": 12.0407,
      "step": 20084
    },
    {
      "epoch": 1.0937085213697904,
      "grad_norm": 0.5644145067321656,
      "learning_rate": 8.970093257700362e-05,
      "loss": 12.0285,
      "step": 20085
    },
    {
      "epoch": 1.0937629753663733,
      "grad_norm": 0.5903288855279883,
      "learning_rate": 8.969216125155483e-05,
      "loss": 11.9965,
      "step": 20086
    },
    {
      "epoch": 1.0938174293629563,
      "grad_norm": 0.5131401448105591,
      "learning_rate": 8.968339000626154e-05,
      "loss": 12.0213,
      "step": 20087
    },
    {
      "epoch": 1.0938718833595393,
      "grad_norm": 0.5098649236324263,
      "learning_rate": 8.967461884119191e-05,
      "loss": 12.0144,
      "step": 20088
    },
    {
      "epoch": 1.0939263373561223,
      "grad_norm": 0.6312531703837619,
      "learning_rate": 8.966584775641423e-05,
      "loss": 12.0738,
      "step": 20089
    },
    {
      "epoch": 1.0939807913527053,
      "grad_norm": 0.5581019802511759,
      "learning_rate": 8.96570767519966e-05,
      "loss": 11.9121,
      "step": 20090
    },
    {
      "epoch": 1.0940352453492883,
      "grad_norm": 0.5632930161470753,
      "learning_rate": 8.96483058280073e-05,
      "loss": 11.974,
      "step": 20091
    },
    {
      "epoch": 1.0940896993458713,
      "grad_norm": 0.6088835188583688,
      "learning_rate": 8.963953498451449e-05,
      "loss": 12.0442,
      "step": 20092
    },
    {
      "epoch": 1.0941441533424543,
      "grad_norm": 0.5602940826984238,
      "learning_rate": 8.963076422158641e-05,
      "loss": 12.0847,
      "step": 20093
    },
    {
      "epoch": 1.0941986073390373,
      "grad_norm": 0.5516614152175064,
      "learning_rate": 8.962199353929123e-05,
      "loss": 12.0685,
      "step": 20094
    },
    {
      "epoch": 1.0942530613356205,
      "grad_norm": 0.5263358727348857,
      "learning_rate": 8.961322293769718e-05,
      "loss": 11.9879,
      "step": 20095
    },
    {
      "epoch": 1.0943075153322035,
      "grad_norm": 0.5990840897682311,
      "learning_rate": 8.960445241687242e-05,
      "loss": 12.176,
      "step": 20096
    },
    {
      "epoch": 1.0943619693287865,
      "grad_norm": 0.6522974451215476,
      "learning_rate": 8.95956819768852e-05,
      "loss": 12.2024,
      "step": 20097
    },
    {
      "epoch": 1.0944164233253695,
      "grad_norm": 0.501404330370195,
      "learning_rate": 8.958691161780371e-05,
      "loss": 12.0772,
      "step": 20098
    },
    {
      "epoch": 1.0944708773219525,
      "grad_norm": 0.5666126899727636,
      "learning_rate": 8.95781413396962e-05,
      "loss": 11.9306,
      "step": 20099
    },
    {
      "epoch": 1.0945253313185355,
      "grad_norm": 0.5762612189338195,
      "learning_rate": 8.956937114263072e-05,
      "loss": 12.0253,
      "step": 20100
    },
    {
      "epoch": 1.0945797853151185,
      "grad_norm": 0.5554150959529307,
      "learning_rate": 8.956060102667559e-05,
      "loss": 12.0569,
      "step": 20101
    },
    {
      "epoch": 1.0946342393117015,
      "grad_norm": 0.5361180229130249,
      "learning_rate": 8.955183099189897e-05,
      "loss": 12.0235,
      "step": 20102
    },
    {
      "epoch": 1.0946886933082844,
      "grad_norm": 0.5243410510388125,
      "learning_rate": 8.954306103836908e-05,
      "loss": 11.9835,
      "step": 20103
    },
    {
      "epoch": 1.0947431473048674,
      "grad_norm": 0.5157913590340827,
      "learning_rate": 8.95342911661541e-05,
      "loss": 12.0156,
      "step": 20104
    },
    {
      "epoch": 1.0947976013014504,
      "grad_norm": 0.5460772186703869,
      "learning_rate": 8.952552137532222e-05,
      "loss": 12.1232,
      "step": 20105
    },
    {
      "epoch": 1.0948520552980336,
      "grad_norm": 0.685412340208087,
      "learning_rate": 8.951675166594165e-05,
      "loss": 12.0335,
      "step": 20106
    },
    {
      "epoch": 1.0949065092946166,
      "grad_norm": 0.6091407511085071,
      "learning_rate": 8.950798203808054e-05,
      "loss": 11.9644,
      "step": 20107
    },
    {
      "epoch": 1.0949609632911996,
      "grad_norm": 0.6472951916502481,
      "learning_rate": 8.949921249180715e-05,
      "loss": 12.1011,
      "step": 20108
    },
    {
      "epoch": 1.0950154172877826,
      "grad_norm": 0.5905049314827779,
      "learning_rate": 8.94904430271897e-05,
      "loss": 12.1145,
      "step": 20109
    },
    {
      "epoch": 1.0950698712843656,
      "grad_norm": 0.5224146984482988,
      "learning_rate": 8.948167364429628e-05,
      "loss": 12.0614,
      "step": 20110
    },
    {
      "epoch": 1.0951243252809486,
      "grad_norm": 0.619051476455258,
      "learning_rate": 8.947290434319508e-05,
      "loss": 12.1789,
      "step": 20111
    },
    {
      "epoch": 1.0951787792775316,
      "grad_norm": 0.5208124207888019,
      "learning_rate": 8.94641351239544e-05,
      "loss": 12.024,
      "step": 20112
    },
    {
      "epoch": 1.0952332332741146,
      "grad_norm": 0.5526687975045929,
      "learning_rate": 8.945536598664235e-05,
      "loss": 12.1497,
      "step": 20113
    },
    {
      "epoch": 1.0952876872706976,
      "grad_norm": 0.5252404891630289,
      "learning_rate": 8.944659693132715e-05,
      "loss": 12.0619,
      "step": 20114
    },
    {
      "epoch": 1.0953421412672806,
      "grad_norm": 0.5319056381621993,
      "learning_rate": 8.943782795807698e-05,
      "loss": 12.0097,
      "step": 20115
    },
    {
      "epoch": 1.0953965952638636,
      "grad_norm": 0.49652797801671555,
      "learning_rate": 8.942905906696004e-05,
      "loss": 11.9446,
      "step": 20116
    },
    {
      "epoch": 1.0954510492604466,
      "grad_norm": 0.5454351264879337,
      "learning_rate": 8.94202902580445e-05,
      "loss": 11.8903,
      "step": 20117
    },
    {
      "epoch": 1.0955055032570296,
      "grad_norm": 0.5506251747120456,
      "learning_rate": 8.941152153139854e-05,
      "loss": 12.1033,
      "step": 20118
    },
    {
      "epoch": 1.0955599572536128,
      "grad_norm": 0.5990370360918348,
      "learning_rate": 8.940275288709041e-05,
      "loss": 12.1675,
      "step": 20119
    },
    {
      "epoch": 1.0956144112501958,
      "grad_norm": 0.5337433802798721,
      "learning_rate": 8.939398432518823e-05,
      "loss": 12.0019,
      "step": 20120
    },
    {
      "epoch": 1.0956688652467788,
      "grad_norm": 0.5445019440378159,
      "learning_rate": 8.93852158457602e-05,
      "loss": 12.0999,
      "step": 20121
    },
    {
      "epoch": 1.0957233192433617,
      "grad_norm": 0.6067101126084533,
      "learning_rate": 8.937644744887451e-05,
      "loss": 12.1753,
      "step": 20122
    },
    {
      "epoch": 1.0957777732399447,
      "grad_norm": 0.6552684164311637,
      "learning_rate": 8.936767913459932e-05,
      "loss": 12.1914,
      "step": 20123
    },
    {
      "epoch": 1.0958322272365277,
      "grad_norm": 0.5696952036299877,
      "learning_rate": 8.935891090300288e-05,
      "loss": 12.0084,
      "step": 20124
    },
    {
      "epoch": 1.0958866812331107,
      "grad_norm": 0.5644830707317485,
      "learning_rate": 8.935014275415332e-05,
      "loss": 12.0161,
      "step": 20125
    },
    {
      "epoch": 1.0959411352296937,
      "grad_norm": 0.5801519224720506,
      "learning_rate": 8.934137468811883e-05,
      "loss": 12.0711,
      "step": 20126
    },
    {
      "epoch": 1.0959955892262767,
      "grad_norm": 0.5415007310440003,
      "learning_rate": 8.93326067049676e-05,
      "loss": 12.0049,
      "step": 20127
    },
    {
      "epoch": 1.0960500432228597,
      "grad_norm": 0.6284333256617635,
      "learning_rate": 8.932383880476782e-05,
      "loss": 12.0675,
      "step": 20128
    },
    {
      "epoch": 1.0961044972194427,
      "grad_norm": 0.620463081209787,
      "learning_rate": 8.931507098758768e-05,
      "loss": 12.1533,
      "step": 20129
    },
    {
      "epoch": 1.096158951216026,
      "grad_norm": 0.5116319195529034,
      "learning_rate": 8.930630325349531e-05,
      "loss": 12.0879,
      "step": 20130
    },
    {
      "epoch": 1.096213405212609,
      "grad_norm": 0.5817236198966054,
      "learning_rate": 8.929753560255892e-05,
      "loss": 11.973,
      "step": 20131
    },
    {
      "epoch": 1.096267859209192,
      "grad_norm": 0.5274798141762556,
      "learning_rate": 8.928876803484669e-05,
      "loss": 11.9422,
      "step": 20132
    },
    {
      "epoch": 1.096322313205775,
      "grad_norm": 0.5738998062283146,
      "learning_rate": 8.92800005504268e-05,
      "loss": 12.0354,
      "step": 20133
    },
    {
      "epoch": 1.0963767672023579,
      "grad_norm": 0.5315282150985224,
      "learning_rate": 8.92712331493674e-05,
      "loss": 11.9889,
      "step": 20134
    },
    {
      "epoch": 1.0964312211989409,
      "grad_norm": 0.5160836765505249,
      "learning_rate": 8.926246583173672e-05,
      "loss": 11.7833,
      "step": 20135
    },
    {
      "epoch": 1.0964856751955239,
      "grad_norm": 0.539779799869873,
      "learning_rate": 8.925369859760288e-05,
      "loss": 12.0612,
      "step": 20136
    },
    {
      "epoch": 1.0965401291921069,
      "grad_norm": 0.5651266964591511,
      "learning_rate": 8.924493144703411e-05,
      "loss": 12.0528,
      "step": 20137
    },
    {
      "epoch": 1.0965945831886899,
      "grad_norm": 0.606502159416733,
      "learning_rate": 8.923616438009855e-05,
      "loss": 12.1592,
      "step": 20138
    },
    {
      "epoch": 1.0966490371852728,
      "grad_norm": 0.5998599114540937,
      "learning_rate": 8.92273973968644e-05,
      "loss": 11.994,
      "step": 20139
    },
    {
      "epoch": 1.0967034911818558,
      "grad_norm": 0.5697227085835541,
      "learning_rate": 8.921863049739979e-05,
      "loss": 12.0578,
      "step": 20140
    },
    {
      "epoch": 1.0967579451784388,
      "grad_norm": 0.5481736629362859,
      "learning_rate": 8.920986368177292e-05,
      "loss": 12.0411,
      "step": 20141
    },
    {
      "epoch": 1.096812399175022,
      "grad_norm": 0.5980455811264401,
      "learning_rate": 8.920109695005195e-05,
      "loss": 12.0798,
      "step": 20142
    },
    {
      "epoch": 1.096866853171605,
      "grad_norm": 0.5560172118990516,
      "learning_rate": 8.919233030230507e-05,
      "loss": 12.1112,
      "step": 20143
    },
    {
      "epoch": 1.096921307168188,
      "grad_norm": 0.5615406718987744,
      "learning_rate": 8.918356373860044e-05,
      "loss": 11.9977,
      "step": 20144
    },
    {
      "epoch": 1.096975761164771,
      "grad_norm": 0.5818790254722687,
      "learning_rate": 8.917479725900622e-05,
      "loss": 12.1381,
      "step": 20145
    },
    {
      "epoch": 1.097030215161354,
      "grad_norm": 0.6037929523246859,
      "learning_rate": 8.91660308635906e-05,
      "loss": 12.0707,
      "step": 20146
    },
    {
      "epoch": 1.097084669157937,
      "grad_norm": 0.5516240368100374,
      "learning_rate": 8.915726455242174e-05,
      "loss": 12.0374,
      "step": 20147
    },
    {
      "epoch": 1.09713912315452,
      "grad_norm": 0.5310517505703545,
      "learning_rate": 8.914849832556782e-05,
      "loss": 12.1701,
      "step": 20148
    },
    {
      "epoch": 1.097193577151103,
      "grad_norm": 0.6207564108102965,
      "learning_rate": 8.913973218309702e-05,
      "loss": 12.1254,
      "step": 20149
    },
    {
      "epoch": 1.097248031147686,
      "grad_norm": 0.5907823547374166,
      "learning_rate": 8.913096612507745e-05,
      "loss": 12.0863,
      "step": 20150
    },
    {
      "epoch": 1.097302485144269,
      "grad_norm": 0.5575367026254822,
      "learning_rate": 8.912220015157732e-05,
      "loss": 11.9744,
      "step": 20151
    },
    {
      "epoch": 1.097356939140852,
      "grad_norm": 0.6751885967052378,
      "learning_rate": 8.911343426266478e-05,
      "loss": 12.0798,
      "step": 20152
    },
    {
      "epoch": 1.0974113931374352,
      "grad_norm": 0.5988546304632669,
      "learning_rate": 8.9104668458408e-05,
      "loss": 12.0581,
      "step": 20153
    },
    {
      "epoch": 1.0974658471340182,
      "grad_norm": 0.5312345113132421,
      "learning_rate": 8.909590273887516e-05,
      "loss": 12.0057,
      "step": 20154
    },
    {
      "epoch": 1.0975203011306012,
      "grad_norm": 0.6099930599315821,
      "learning_rate": 8.908713710413438e-05,
      "loss": 12.2494,
      "step": 20155
    },
    {
      "epoch": 1.0975747551271842,
      "grad_norm": 0.614170068332695,
      "learning_rate": 8.907837155425385e-05,
      "loss": 12.0462,
      "step": 20156
    },
    {
      "epoch": 1.0976292091237672,
      "grad_norm": 0.5822291739120788,
      "learning_rate": 8.906960608930176e-05,
      "loss": 12.0543,
      "step": 20157
    },
    {
      "epoch": 1.0976836631203502,
      "grad_norm": 0.5188234845750623,
      "learning_rate": 8.906084070934623e-05,
      "loss": 12.0245,
      "step": 20158
    },
    {
      "epoch": 1.0977381171169331,
      "grad_norm": 0.5011474986009845,
      "learning_rate": 8.905207541445551e-05,
      "loss": 12.0592,
      "step": 20159
    },
    {
      "epoch": 1.0977925711135161,
      "grad_norm": 0.6494712131666551,
      "learning_rate": 8.904331020469759e-05,
      "loss": 12.1785,
      "step": 20160
    },
    {
      "epoch": 1.0978470251100991,
      "grad_norm": 0.5733330915908963,
      "learning_rate": 8.903454508014076e-05,
      "loss": 12.0438,
      "step": 20161
    },
    {
      "epoch": 1.0979014791066821,
      "grad_norm": 0.5601787541031239,
      "learning_rate": 8.902578004085315e-05,
      "loss": 12.0481,
      "step": 20162
    },
    {
      "epoch": 1.0979559331032651,
      "grad_norm": 0.6770310032193882,
      "learning_rate": 8.901701508690291e-05,
      "loss": 12.0976,
      "step": 20163
    },
    {
      "epoch": 1.098010387099848,
      "grad_norm": 0.6271802627011732,
      "learning_rate": 8.900825021835821e-05,
      "loss": 12.1657,
      "step": 20164
    },
    {
      "epoch": 1.0980648410964313,
      "grad_norm": 0.5596518808132004,
      "learning_rate": 8.89994854352872e-05,
      "loss": 11.9854,
      "step": 20165
    },
    {
      "epoch": 1.0981192950930143,
      "grad_norm": 0.5020805826317156,
      "learning_rate": 8.899072073775802e-05,
      "loss": 11.9333,
      "step": 20166
    },
    {
      "epoch": 1.0981737490895973,
      "grad_norm": 0.6042882510332298,
      "learning_rate": 8.898195612583886e-05,
      "loss": 12.1209,
      "step": 20167
    },
    {
      "epoch": 1.0982282030861803,
      "grad_norm": 0.652798989112454,
      "learning_rate": 8.897319159959783e-05,
      "loss": 12.0498,
      "step": 20168
    },
    {
      "epoch": 1.0982826570827633,
      "grad_norm": 0.5933847046017819,
      "learning_rate": 8.896442715910317e-05,
      "loss": 12.1471,
      "step": 20169
    },
    {
      "epoch": 1.0983371110793463,
      "grad_norm": 0.602944215692327,
      "learning_rate": 8.895566280442294e-05,
      "loss": 11.9518,
      "step": 20170
    },
    {
      "epoch": 1.0983915650759293,
      "grad_norm": 0.5666090980810462,
      "learning_rate": 8.894689853562532e-05,
      "loss": 12.0612,
      "step": 20171
    },
    {
      "epoch": 1.0984460190725123,
      "grad_norm": 0.5955716178058724,
      "learning_rate": 8.893813435277845e-05,
      "loss": 12.0099,
      "step": 20172
    },
    {
      "epoch": 1.0985004730690953,
      "grad_norm": 0.630128554425275,
      "learning_rate": 8.892937025595053e-05,
      "loss": 12.0716,
      "step": 20173
    },
    {
      "epoch": 1.0985549270656783,
      "grad_norm": 0.56439065043191,
      "learning_rate": 8.892060624520968e-05,
      "loss": 11.9565,
      "step": 20174
    },
    {
      "epoch": 1.0986093810622612,
      "grad_norm": 0.5866824579384988,
      "learning_rate": 8.891184232062405e-05,
      "loss": 11.9924,
      "step": 20175
    },
    {
      "epoch": 1.0986638350588445,
      "grad_norm": 0.7665723124085866,
      "learning_rate": 8.89030784822618e-05,
      "loss": 12.2208,
      "step": 20176
    },
    {
      "epoch": 1.0987182890554275,
      "grad_norm": 0.6276742135752813,
      "learning_rate": 8.889431473019108e-05,
      "loss": 12.0264,
      "step": 20177
    },
    {
      "epoch": 1.0987727430520104,
      "grad_norm": 0.579773130352321,
      "learning_rate": 8.888555106448e-05,
      "loss": 12.1391,
      "step": 20178
    },
    {
      "epoch": 1.0988271970485934,
      "grad_norm": 0.5597768164792385,
      "learning_rate": 8.88767874851968e-05,
      "loss": 12.1392,
      "step": 20179
    },
    {
      "epoch": 1.0988816510451764,
      "grad_norm": 0.5509950255852855,
      "learning_rate": 8.886802399240952e-05,
      "loss": 11.9973,
      "step": 20180
    },
    {
      "epoch": 1.0989361050417594,
      "grad_norm": 0.5743486501236731,
      "learning_rate": 8.885926058618636e-05,
      "loss": 12.0458,
      "step": 20181
    },
    {
      "epoch": 1.0989905590383424,
      "grad_norm": 0.5308475307375142,
      "learning_rate": 8.885049726659546e-05,
      "loss": 11.9559,
      "step": 20182
    },
    {
      "epoch": 1.0990450130349254,
      "grad_norm": 0.5065825279705609,
      "learning_rate": 8.884173403370494e-05,
      "loss": 11.9881,
      "step": 20183
    },
    {
      "epoch": 1.0990994670315084,
      "grad_norm": 0.5652712077381974,
      "learning_rate": 8.883297088758298e-05,
      "loss": 12.0564,
      "step": 20184
    },
    {
      "epoch": 1.0991539210280914,
      "grad_norm": 0.5415480303899827,
      "learning_rate": 8.882420782829772e-05,
      "loss": 12.0772,
      "step": 20185
    },
    {
      "epoch": 1.0992083750246744,
      "grad_norm": 0.5352570980250407,
      "learning_rate": 8.881544485591729e-05,
      "loss": 12.0056,
      "step": 20186
    },
    {
      "epoch": 1.0992628290212574,
      "grad_norm": 0.5173298551932756,
      "learning_rate": 8.880668197050984e-05,
      "loss": 12.0798,
      "step": 20187
    },
    {
      "epoch": 1.0993172830178404,
      "grad_norm": 0.48717873604380874,
      "learning_rate": 8.87979191721435e-05,
      "loss": 12.0184,
      "step": 20188
    },
    {
      "epoch": 1.0993717370144236,
      "grad_norm": 0.5597341952123152,
      "learning_rate": 8.878915646088646e-05,
      "loss": 11.9295,
      "step": 20189
    },
    {
      "epoch": 1.0994261910110066,
      "grad_norm": 0.5615709176867464,
      "learning_rate": 8.878039383680678e-05,
      "loss": 12.1507,
      "step": 20190
    },
    {
      "epoch": 1.0994806450075896,
      "grad_norm": 0.5740671509788972,
      "learning_rate": 8.877163129997265e-05,
      "loss": 12.022,
      "step": 20191
    },
    {
      "epoch": 1.0995350990041726,
      "grad_norm": 0.5352406945605978,
      "learning_rate": 8.876286885045218e-05,
      "loss": 12.1071,
      "step": 20192
    },
    {
      "epoch": 1.0995895530007556,
      "grad_norm": 0.5373223953618459,
      "learning_rate": 8.875410648831355e-05,
      "loss": 11.9739,
      "step": 20193
    },
    {
      "epoch": 1.0996440069973386,
      "grad_norm": 0.5296057941726166,
      "learning_rate": 8.874534421362484e-05,
      "loss": 11.9395,
      "step": 20194
    },
    {
      "epoch": 1.0996984609939215,
      "grad_norm": 0.5577670315621532,
      "learning_rate": 8.873658202645424e-05,
      "loss": 11.9845,
      "step": 20195
    },
    {
      "epoch": 1.0997529149905045,
      "grad_norm": 0.5335575364798968,
      "learning_rate": 8.872781992686987e-05,
      "loss": 12.0861,
      "step": 20196
    },
    {
      "epoch": 1.0998073689870875,
      "grad_norm": 0.5508209498312792,
      "learning_rate": 8.871905791493987e-05,
      "loss": 11.9597,
      "step": 20197
    },
    {
      "epoch": 1.0998618229836705,
      "grad_norm": 0.6250421100736349,
      "learning_rate": 8.871029599073235e-05,
      "loss": 12.1618,
      "step": 20198
    },
    {
      "epoch": 1.0999162769802535,
      "grad_norm": 0.591381183464453,
      "learning_rate": 8.87015341543155e-05,
      "loss": 12.1627,
      "step": 20199
    },
    {
      "epoch": 1.0999707309768367,
      "grad_norm": 0.5787438375914769,
      "learning_rate": 8.869277240575738e-05,
      "loss": 12.0613,
      "step": 20200
    },
    {
      "epoch": 1.1000251849734197,
      "grad_norm": 0.5910180724776823,
      "learning_rate": 8.868401074512616e-05,
      "loss": 11.9308,
      "step": 20201
    },
    {
      "epoch": 1.1000796389700027,
      "grad_norm": 0.5395654718912095,
      "learning_rate": 8.867524917248999e-05,
      "loss": 12.0461,
      "step": 20202
    },
    {
      "epoch": 1.1001340929665857,
      "grad_norm": 0.5436651477781418,
      "learning_rate": 8.866648768791697e-05,
      "loss": 12.1008,
      "step": 20203
    },
    {
      "epoch": 1.1001885469631687,
      "grad_norm": 0.6733763075191992,
      "learning_rate": 8.865772629147523e-05,
      "loss": 11.9532,
      "step": 20204
    },
    {
      "epoch": 1.1002430009597517,
      "grad_norm": 0.5931005392577273,
      "learning_rate": 8.86489649832329e-05,
      "loss": 12.0734,
      "step": 20205
    },
    {
      "epoch": 1.1002974549563347,
      "grad_norm": 0.5793400162860026,
      "learning_rate": 8.864020376325814e-05,
      "loss": 12.0362,
      "step": 20206
    },
    {
      "epoch": 1.1003519089529177,
      "grad_norm": 0.5345822138575493,
      "learning_rate": 8.863144263161906e-05,
      "loss": 12.0126,
      "step": 20207
    },
    {
      "epoch": 1.1004063629495007,
      "grad_norm": 0.5150319586228077,
      "learning_rate": 8.86226815883838e-05,
      "loss": 12.0538,
      "step": 20208
    },
    {
      "epoch": 1.1004608169460837,
      "grad_norm": 0.5139144427714485,
      "learning_rate": 8.86139206336205e-05,
      "loss": 11.9446,
      "step": 20209
    },
    {
      "epoch": 1.1005152709426667,
      "grad_norm": 0.5159596684044622,
      "learning_rate": 8.860515976739722e-05,
      "loss": 12.0148,
      "step": 20210
    },
    {
      "epoch": 1.1005697249392496,
      "grad_norm": 0.5954957669807992,
      "learning_rate": 8.859639898978213e-05,
      "loss": 11.9208,
      "step": 20211
    },
    {
      "epoch": 1.1006241789358329,
      "grad_norm": 0.47558291302665023,
      "learning_rate": 8.858763830084338e-05,
      "loss": 12.0234,
      "step": 20212
    },
    {
      "epoch": 1.1006786329324159,
      "grad_norm": 0.5723861581364824,
      "learning_rate": 8.857887770064905e-05,
      "loss": 12.1122,
      "step": 20213
    },
    {
      "epoch": 1.1007330869289988,
      "grad_norm": 0.5300311766221824,
      "learning_rate": 8.857011718926728e-05,
      "loss": 11.9661,
      "step": 20214
    },
    {
      "epoch": 1.1007875409255818,
      "grad_norm": 0.4804128022172825,
      "learning_rate": 8.85613567667662e-05,
      "loss": 11.9646,
      "step": 20215
    },
    {
      "epoch": 1.1008419949221648,
      "grad_norm": 0.6411895825078691,
      "learning_rate": 8.855259643321391e-05,
      "loss": 12.0086,
      "step": 20216
    },
    {
      "epoch": 1.1008964489187478,
      "grad_norm": 0.5045659544266304,
      "learning_rate": 8.854383618867857e-05,
      "loss": 11.9535,
      "step": 20217
    },
    {
      "epoch": 1.1009509029153308,
      "grad_norm": 0.5546269037658237,
      "learning_rate": 8.853507603322828e-05,
      "loss": 12.0156,
      "step": 20218
    },
    {
      "epoch": 1.1010053569119138,
      "grad_norm": 0.5472239519449394,
      "learning_rate": 8.852631596693115e-05,
      "loss": 12.0827,
      "step": 20219
    },
    {
      "epoch": 1.1010598109084968,
      "grad_norm": 0.6151265780247128,
      "learning_rate": 8.851755598985537e-05,
      "loss": 12.1356,
      "step": 20220
    },
    {
      "epoch": 1.1011142649050798,
      "grad_norm": 0.5877863943068412,
      "learning_rate": 8.850879610206894e-05,
      "loss": 12.0763,
      "step": 20221
    },
    {
      "epoch": 1.1011687189016628,
      "grad_norm": 0.5374816326611753,
      "learning_rate": 8.850003630364005e-05,
      "loss": 11.9959,
      "step": 20222
    },
    {
      "epoch": 1.101223172898246,
      "grad_norm": 0.6347521513899049,
      "learning_rate": 8.849127659463679e-05,
      "loss": 11.9279,
      "step": 20223
    },
    {
      "epoch": 1.101277626894829,
      "grad_norm": 0.5497226963963572,
      "learning_rate": 8.848251697512732e-05,
      "loss": 11.9248,
      "step": 20224
    },
    {
      "epoch": 1.101332080891412,
      "grad_norm": 0.5671760158929101,
      "learning_rate": 8.847375744517972e-05,
      "loss": 11.9878,
      "step": 20225
    },
    {
      "epoch": 1.101386534887995,
      "grad_norm": 0.505449844315798,
      "learning_rate": 8.846499800486211e-05,
      "loss": 12.0726,
      "step": 20226
    },
    {
      "epoch": 1.101440988884578,
      "grad_norm": 0.5733129507632411,
      "learning_rate": 8.845623865424262e-05,
      "loss": 12.1238,
      "step": 20227
    },
    {
      "epoch": 1.101495442881161,
      "grad_norm": 0.5792222536553667,
      "learning_rate": 8.844747939338933e-05,
      "loss": 12.0608,
      "step": 20228
    },
    {
      "epoch": 1.101549896877744,
      "grad_norm": 0.5357805512260676,
      "learning_rate": 8.843872022237039e-05,
      "loss": 12.0475,
      "step": 20229
    },
    {
      "epoch": 1.101604350874327,
      "grad_norm": 0.5568764403483965,
      "learning_rate": 8.842996114125395e-05,
      "loss": 11.9596,
      "step": 20230
    },
    {
      "epoch": 1.10165880487091,
      "grad_norm": 0.5652795661893997,
      "learning_rate": 8.842120215010803e-05,
      "loss": 12.1226,
      "step": 20231
    },
    {
      "epoch": 1.101713258867493,
      "grad_norm": 0.5696300592102768,
      "learning_rate": 8.841244324900076e-05,
      "loss": 12.1162,
      "step": 20232
    },
    {
      "epoch": 1.101767712864076,
      "grad_norm": 0.6205151314076434,
      "learning_rate": 8.84036844380003e-05,
      "loss": 12.1801,
      "step": 20233
    },
    {
      "epoch": 1.101822166860659,
      "grad_norm": 0.5466304713585389,
      "learning_rate": 8.839492571717473e-05,
      "loss": 11.9161,
      "step": 20234
    },
    {
      "epoch": 1.1018766208572421,
      "grad_norm": 0.5140270883111085,
      "learning_rate": 8.838616708659217e-05,
      "loss": 12.0871,
      "step": 20235
    },
    {
      "epoch": 1.1019310748538251,
      "grad_norm": 0.5832650218878539,
      "learning_rate": 8.83774085463207e-05,
      "loss": 12.0507,
      "step": 20236
    },
    {
      "epoch": 1.1019855288504081,
      "grad_norm": 0.6361237961683173,
      "learning_rate": 8.836865009642848e-05,
      "loss": 12.1832,
      "step": 20237
    },
    {
      "epoch": 1.1020399828469911,
      "grad_norm": 0.5433931408566341,
      "learning_rate": 8.835989173698358e-05,
      "loss": 12.112,
      "step": 20238
    },
    {
      "epoch": 1.102094436843574,
      "grad_norm": 0.531281605006899,
      "learning_rate": 8.835113346805408e-05,
      "loss": 12.1103,
      "step": 20239
    },
    {
      "epoch": 1.102148890840157,
      "grad_norm": 0.547987291454372,
      "learning_rate": 8.83423752897082e-05,
      "loss": 12.0016,
      "step": 20240
    },
    {
      "epoch": 1.10220334483674,
      "grad_norm": 0.6196430697296458,
      "learning_rate": 8.833361720201391e-05,
      "loss": 12.106,
      "step": 20241
    },
    {
      "epoch": 1.102257798833323,
      "grad_norm": 0.5075520452259732,
      "learning_rate": 8.832485920503937e-05,
      "loss": 12.064,
      "step": 20242
    },
    {
      "epoch": 1.102312252829906,
      "grad_norm": 0.573113397993567,
      "learning_rate": 8.831610129885266e-05,
      "loss": 12.1014,
      "step": 20243
    },
    {
      "epoch": 1.102366706826489,
      "grad_norm": 0.5467387171808178,
      "learning_rate": 8.830734348352195e-05,
      "loss": 12.0631,
      "step": 20244
    },
    {
      "epoch": 1.102421160823072,
      "grad_norm": 0.5386865259364804,
      "learning_rate": 8.829858575911527e-05,
      "loss": 12.0966,
      "step": 20245
    },
    {
      "epoch": 1.1024756148196553,
      "grad_norm": 0.5252737553681605,
      "learning_rate": 8.828982812570075e-05,
      "loss": 11.9106,
      "step": 20246
    },
    {
      "epoch": 1.1025300688162383,
      "grad_norm": 0.48756803933840664,
      "learning_rate": 8.82810705833465e-05,
      "loss": 11.9954,
      "step": 20247
    },
    {
      "epoch": 1.1025845228128213,
      "grad_norm": 0.5584157079026932,
      "learning_rate": 8.827231313212061e-05,
      "loss": 12.0073,
      "step": 20248
    },
    {
      "epoch": 1.1026389768094043,
      "grad_norm": 0.5613385083370157,
      "learning_rate": 8.826355577209118e-05,
      "loss": 12.0158,
      "step": 20249
    },
    {
      "epoch": 1.1026934308059873,
      "grad_norm": 0.5457529803685589,
      "learning_rate": 8.825479850332633e-05,
      "loss": 12.097,
      "step": 20250
    },
    {
      "epoch": 1.1027478848025702,
      "grad_norm": 0.567677049333743,
      "learning_rate": 8.824604132589412e-05,
      "loss": 12.0287,
      "step": 20251
    },
    {
      "epoch": 1.1028023387991532,
      "grad_norm": 0.4833128160398218,
      "learning_rate": 8.823728423986266e-05,
      "loss": 12.0636,
      "step": 20252
    },
    {
      "epoch": 1.1028567927957362,
      "grad_norm": 0.573142902524311,
      "learning_rate": 8.822852724530004e-05,
      "loss": 11.9064,
      "step": 20253
    },
    {
      "epoch": 1.1029112467923192,
      "grad_norm": 0.5058465021018069,
      "learning_rate": 8.821977034227435e-05,
      "loss": 12.0118,
      "step": 20254
    },
    {
      "epoch": 1.1029657007889022,
      "grad_norm": 0.5663852923558932,
      "learning_rate": 8.821101353085374e-05,
      "loss": 12.1247,
      "step": 20255
    },
    {
      "epoch": 1.1030201547854852,
      "grad_norm": 0.6070677912259818,
      "learning_rate": 8.820225681110624e-05,
      "loss": 12.1287,
      "step": 20256
    },
    {
      "epoch": 1.1030746087820682,
      "grad_norm": 0.541915999624938,
      "learning_rate": 8.819350018309999e-05,
      "loss": 12.1196,
      "step": 20257
    },
    {
      "epoch": 1.1031290627786514,
      "grad_norm": 0.5319353360682426,
      "learning_rate": 8.818474364690306e-05,
      "loss": 11.7641,
      "step": 20258
    },
    {
      "epoch": 1.1031835167752344,
      "grad_norm": 0.5757522620881451,
      "learning_rate": 8.817598720258353e-05,
      "loss": 11.9708,
      "step": 20259
    },
    {
      "epoch": 1.1032379707718174,
      "grad_norm": 0.6081238647231033,
      "learning_rate": 8.816723085020954e-05,
      "loss": 12.0519,
      "step": 20260
    },
    {
      "epoch": 1.1032924247684004,
      "grad_norm": 0.5427343785399987,
      "learning_rate": 8.815847458984911e-05,
      "loss": 12.0829,
      "step": 20261
    },
    {
      "epoch": 1.1033468787649834,
      "grad_norm": 0.5652734711236881,
      "learning_rate": 8.814971842157039e-05,
      "loss": 12.0157,
      "step": 20262
    },
    {
      "epoch": 1.1034013327615664,
      "grad_norm": 0.5534906612787915,
      "learning_rate": 8.814096234544143e-05,
      "loss": 12.0162,
      "step": 20263
    },
    {
      "epoch": 1.1034557867581494,
      "grad_norm": 0.5036252440792517,
      "learning_rate": 8.813220636153035e-05,
      "loss": 12.009,
      "step": 20264
    },
    {
      "epoch": 1.1035102407547324,
      "grad_norm": 0.5257087999044628,
      "learning_rate": 8.812345046990519e-05,
      "loss": 11.9534,
      "step": 20265
    },
    {
      "epoch": 1.1035646947513154,
      "grad_norm": 0.49793946160611324,
      "learning_rate": 8.81146946706341e-05,
      "loss": 12.03,
      "step": 20266
    },
    {
      "epoch": 1.1036191487478983,
      "grad_norm": 0.5546572372244591,
      "learning_rate": 8.810593896378513e-05,
      "loss": 11.9098,
      "step": 20267
    },
    {
      "epoch": 1.1036736027444813,
      "grad_norm": 0.5424649818390285,
      "learning_rate": 8.809718334942639e-05,
      "loss": 11.9215,
      "step": 20268
    },
    {
      "epoch": 1.1037280567410646,
      "grad_norm": 0.5866022695322437,
      "learning_rate": 8.808842782762592e-05,
      "loss": 12.0288,
      "step": 20269
    },
    {
      "epoch": 1.1037825107376475,
      "grad_norm": 0.5200616494258773,
      "learning_rate": 8.807967239845187e-05,
      "loss": 12.028,
      "step": 20270
    },
    {
      "epoch": 1.1038369647342305,
      "grad_norm": 0.471892641538079,
      "learning_rate": 8.807091706197228e-05,
      "loss": 11.9984,
      "step": 20271
    },
    {
      "epoch": 1.1038914187308135,
      "grad_norm": 0.5069317963591853,
      "learning_rate": 8.806216181825522e-05,
      "loss": 12.0616,
      "step": 20272
    },
    {
      "epoch": 1.1039458727273965,
      "grad_norm": 0.5181631910018965,
      "learning_rate": 8.805340666736878e-05,
      "loss": 12.1398,
      "step": 20273
    },
    {
      "epoch": 1.1040003267239795,
      "grad_norm": 0.5026001587635697,
      "learning_rate": 8.804465160938108e-05,
      "loss": 12.0038,
      "step": 20274
    },
    {
      "epoch": 1.1040547807205625,
      "grad_norm": 0.5208749547424139,
      "learning_rate": 8.803589664436017e-05,
      "loss": 11.9452,
      "step": 20275
    },
    {
      "epoch": 1.1041092347171455,
      "grad_norm": 0.5329006436594417,
      "learning_rate": 8.802714177237412e-05,
      "loss": 12.0868,
      "step": 20276
    },
    {
      "epoch": 1.1041636887137285,
      "grad_norm": 0.6238998929891365,
      "learning_rate": 8.801838699349101e-05,
      "loss": 12.08,
      "step": 20277
    },
    {
      "epoch": 1.1042181427103115,
      "grad_norm": 0.5713533790144977,
      "learning_rate": 8.800963230777896e-05,
      "loss": 12.0738,
      "step": 20278
    },
    {
      "epoch": 1.1042725967068945,
      "grad_norm": 0.5067501866499987,
      "learning_rate": 8.8000877715306e-05,
      "loss": 12.1064,
      "step": 20279
    },
    {
      "epoch": 1.1043270507034775,
      "grad_norm": 0.5696655009372665,
      "learning_rate": 8.799212321614029e-05,
      "loss": 12.0327,
      "step": 20280
    },
    {
      "epoch": 1.1043815047000605,
      "grad_norm": 0.5907525642848902,
      "learning_rate": 8.798336881034976e-05,
      "loss": 11.9661,
      "step": 20281
    },
    {
      "epoch": 1.1044359586966437,
      "grad_norm": 0.5029806148791618,
      "learning_rate": 8.797461449800262e-05,
      "loss": 12.0543,
      "step": 20282
    },
    {
      "epoch": 1.1044904126932267,
      "grad_norm": 0.5740037503429757,
      "learning_rate": 8.796586027916686e-05,
      "loss": 12.0909,
      "step": 20283
    },
    {
      "epoch": 1.1045448666898097,
      "grad_norm": 0.5359098765190763,
      "learning_rate": 8.795710615391061e-05,
      "loss": 12.0901,
      "step": 20284
    },
    {
      "epoch": 1.1045993206863927,
      "grad_norm": 0.5381826336912698,
      "learning_rate": 8.794835212230193e-05,
      "loss": 11.9385,
      "step": 20285
    },
    {
      "epoch": 1.1046537746829757,
      "grad_norm": 0.5164361531574544,
      "learning_rate": 8.793959818440887e-05,
      "loss": 11.9999,
      "step": 20286
    },
    {
      "epoch": 1.1047082286795586,
      "grad_norm": 0.4952044177832422,
      "learning_rate": 8.793084434029952e-05,
      "loss": 11.934,
      "step": 20287
    },
    {
      "epoch": 1.1047626826761416,
      "grad_norm": 0.5353442979609403,
      "learning_rate": 8.792209059004193e-05,
      "loss": 11.9958,
      "step": 20288
    },
    {
      "epoch": 1.1048171366727246,
      "grad_norm": 0.5198804593022757,
      "learning_rate": 8.79133369337042e-05,
      "loss": 11.8735,
      "step": 20289
    },
    {
      "epoch": 1.1048715906693076,
      "grad_norm": 0.5539164060775061,
      "learning_rate": 8.790458337135444e-05,
      "loss": 11.9442,
      "step": 20290
    },
    {
      "epoch": 1.1049260446658906,
      "grad_norm": 0.5149487649125211,
      "learning_rate": 8.789582990306062e-05,
      "loss": 12.0169,
      "step": 20291
    },
    {
      "epoch": 1.1049804986624736,
      "grad_norm": 0.5593324766953811,
      "learning_rate": 8.788707652889084e-05,
      "loss": 12.0518,
      "step": 20292
    },
    {
      "epoch": 1.1050349526590568,
      "grad_norm": 0.5223788666070568,
      "learning_rate": 8.78783232489132e-05,
      "loss": 12.0239,
      "step": 20293
    },
    {
      "epoch": 1.1050894066556398,
      "grad_norm": 0.6409900434279258,
      "learning_rate": 8.786957006319577e-05,
      "loss": 11.9321,
      "step": 20294
    },
    {
      "epoch": 1.1051438606522228,
      "grad_norm": 0.5991656646274143,
      "learning_rate": 8.786081697180659e-05,
      "loss": 12.1513,
      "step": 20295
    },
    {
      "epoch": 1.1051983146488058,
      "grad_norm": 0.5437244510373981,
      "learning_rate": 8.785206397481371e-05,
      "loss": 11.9969,
      "step": 20296
    },
    {
      "epoch": 1.1052527686453888,
      "grad_norm": 0.5410626905453222,
      "learning_rate": 8.784331107228525e-05,
      "loss": 12.117,
      "step": 20297
    },
    {
      "epoch": 1.1053072226419718,
      "grad_norm": 0.5280928701320662,
      "learning_rate": 8.783455826428921e-05,
      "loss": 12.0798,
      "step": 20298
    },
    {
      "epoch": 1.1053616766385548,
      "grad_norm": 0.5852564727198065,
      "learning_rate": 8.782580555089368e-05,
      "loss": 11.9185,
      "step": 20299
    },
    {
      "epoch": 1.1054161306351378,
      "grad_norm": 0.5640739042641488,
      "learning_rate": 8.78170529321668e-05,
      "loss": 12.045,
      "step": 20300
    },
    {
      "epoch": 1.1054705846317208,
      "grad_norm": 0.5711766805877603,
      "learning_rate": 8.780830040817651e-05,
      "loss": 12.0574,
      "step": 20301
    },
    {
      "epoch": 1.1055250386283038,
      "grad_norm": 0.5487375604738145,
      "learning_rate": 8.779954797899091e-05,
      "loss": 12.0689,
      "step": 20302
    },
    {
      "epoch": 1.1055794926248868,
      "grad_norm": 0.5463485657321694,
      "learning_rate": 8.779079564467807e-05,
      "loss": 12.0732,
      "step": 20303
    },
    {
      "epoch": 1.1056339466214697,
      "grad_norm": 0.5958010094516646,
      "learning_rate": 8.778204340530606e-05,
      "loss": 12.14,
      "step": 20304
    },
    {
      "epoch": 1.105688400618053,
      "grad_norm": 0.555501224415568,
      "learning_rate": 8.777329126094292e-05,
      "loss": 11.9945,
      "step": 20305
    },
    {
      "epoch": 1.105742854614636,
      "grad_norm": 0.5719709735434838,
      "learning_rate": 8.776453921165674e-05,
      "loss": 12.0651,
      "step": 20306
    },
    {
      "epoch": 1.105797308611219,
      "grad_norm": 0.5703447668093605,
      "learning_rate": 8.775578725751553e-05,
      "loss": 11.9732,
      "step": 20307
    },
    {
      "epoch": 1.105851762607802,
      "grad_norm": 0.584597127790104,
      "learning_rate": 8.77470353985874e-05,
      "loss": 12.0707,
      "step": 20308
    },
    {
      "epoch": 1.105906216604385,
      "grad_norm": 0.5099775442129576,
      "learning_rate": 8.773828363494036e-05,
      "loss": 11.9749,
      "step": 20309
    },
    {
      "epoch": 1.105960670600968,
      "grad_norm": 0.5249223983386117,
      "learning_rate": 8.77295319666425e-05,
      "loss": 11.933,
      "step": 20310
    },
    {
      "epoch": 1.106015124597551,
      "grad_norm": 0.5734136019444562,
      "learning_rate": 8.772078039376184e-05,
      "loss": 11.9393,
      "step": 20311
    },
    {
      "epoch": 1.106069578594134,
      "grad_norm": 0.6383360418108145,
      "learning_rate": 8.771202891636646e-05,
      "loss": 12.2153,
      "step": 20312
    },
    {
      "epoch": 1.106124032590717,
      "grad_norm": 0.564871378026721,
      "learning_rate": 8.770327753452438e-05,
      "loss": 12.0592,
      "step": 20313
    },
    {
      "epoch": 1.1061784865873,
      "grad_norm": 0.5003922451111757,
      "learning_rate": 8.769452624830367e-05,
      "loss": 11.9846,
      "step": 20314
    },
    {
      "epoch": 1.1062329405838829,
      "grad_norm": 0.6299781577491396,
      "learning_rate": 8.768577505777242e-05,
      "loss": 11.9441,
      "step": 20315
    },
    {
      "epoch": 1.106287394580466,
      "grad_norm": 0.6763607365341504,
      "learning_rate": 8.767702396299864e-05,
      "loss": 12.0107,
      "step": 20316
    },
    {
      "epoch": 1.106341848577049,
      "grad_norm": 0.5623560295924026,
      "learning_rate": 8.766827296405039e-05,
      "loss": 12.0481,
      "step": 20317
    },
    {
      "epoch": 1.106396302573632,
      "grad_norm": 0.5965173681242213,
      "learning_rate": 8.765952206099572e-05,
      "loss": 12.0934,
      "step": 20318
    },
    {
      "epoch": 1.106450756570215,
      "grad_norm": 0.5279141459995098,
      "learning_rate": 8.765077125390266e-05,
      "loss": 12.0243,
      "step": 20319
    },
    {
      "epoch": 1.106505210566798,
      "grad_norm": 0.5381397115637279,
      "learning_rate": 8.76420205428393e-05,
      "loss": 12.0341,
      "step": 20320
    },
    {
      "epoch": 1.106559664563381,
      "grad_norm": 0.5484374520949024,
      "learning_rate": 8.763326992787365e-05,
      "loss": 11.9527,
      "step": 20321
    },
    {
      "epoch": 1.106614118559964,
      "grad_norm": 0.5771416993957864,
      "learning_rate": 8.762451940907376e-05,
      "loss": 12.0113,
      "step": 20322
    },
    {
      "epoch": 1.106668572556547,
      "grad_norm": 0.5563674089393889,
      "learning_rate": 8.761576898650768e-05,
      "loss": 11.9658,
      "step": 20323
    },
    {
      "epoch": 1.10672302655313,
      "grad_norm": 0.5536476891827287,
      "learning_rate": 8.760701866024347e-05,
      "loss": 12.0788,
      "step": 20324
    },
    {
      "epoch": 1.106777480549713,
      "grad_norm": 0.5891056273868874,
      "learning_rate": 8.759826843034915e-05,
      "loss": 12.0831,
      "step": 20325
    },
    {
      "epoch": 1.106831934546296,
      "grad_norm": 0.599718056726197,
      "learning_rate": 8.758951829689275e-05,
      "loss": 12.0573,
      "step": 20326
    },
    {
      "epoch": 1.106886388542879,
      "grad_norm": 0.5371038737374279,
      "learning_rate": 8.758076825994237e-05,
      "loss": 12.0321,
      "step": 20327
    },
    {
      "epoch": 1.1069408425394622,
      "grad_norm": 0.6004552767530613,
      "learning_rate": 8.7572018319566e-05,
      "loss": 12.0596,
      "step": 20328
    },
    {
      "epoch": 1.1069952965360452,
      "grad_norm": 0.5865194360324422,
      "learning_rate": 8.756326847583171e-05,
      "loss": 12.0869,
      "step": 20329
    },
    {
      "epoch": 1.1070497505326282,
      "grad_norm": 0.5048031034073145,
      "learning_rate": 8.755451872880757e-05,
      "loss": 12.0092,
      "step": 20330
    },
    {
      "epoch": 1.1071042045292112,
      "grad_norm": 0.5322675844630659,
      "learning_rate": 8.754576907856154e-05,
      "loss": 12.0208,
      "step": 20331
    },
    {
      "epoch": 1.1071586585257942,
      "grad_norm": 0.5441646557030124,
      "learning_rate": 8.753701952516169e-05,
      "loss": 12.0182,
      "step": 20332
    },
    {
      "epoch": 1.1072131125223772,
      "grad_norm": 0.5608449926792899,
      "learning_rate": 8.752827006867607e-05,
      "loss": 11.9662,
      "step": 20333
    },
    {
      "epoch": 1.1072675665189602,
      "grad_norm": 0.5723882401594803,
      "learning_rate": 8.751952070917273e-05,
      "loss": 11.997,
      "step": 20334
    },
    {
      "epoch": 1.1073220205155432,
      "grad_norm": 0.5455646117400574,
      "learning_rate": 8.751077144671968e-05,
      "loss": 11.9964,
      "step": 20335
    },
    {
      "epoch": 1.1073764745121262,
      "grad_norm": 0.5869649085404693,
      "learning_rate": 8.750202228138497e-05,
      "loss": 12.0137,
      "step": 20336
    },
    {
      "epoch": 1.1074309285087092,
      "grad_norm": 0.4929235067216394,
      "learning_rate": 8.749327321323659e-05,
      "loss": 11.9886,
      "step": 20337
    },
    {
      "epoch": 1.1074853825052922,
      "grad_norm": 0.542015727952765,
      "learning_rate": 8.748452424234266e-05,
      "loss": 12.0229,
      "step": 20338
    },
    {
      "epoch": 1.1075398365018754,
      "grad_norm": 0.5877668144850737,
      "learning_rate": 8.747577536877117e-05,
      "loss": 11.9163,
      "step": 20339
    },
    {
      "epoch": 1.1075942904984584,
      "grad_norm": 0.5608190697937362,
      "learning_rate": 8.746702659259017e-05,
      "loss": 12.0286,
      "step": 20340
    },
    {
      "epoch": 1.1076487444950414,
      "grad_norm": 0.513750153259443,
      "learning_rate": 8.745827791386762e-05,
      "loss": 12.0426,
      "step": 20341
    },
    {
      "epoch": 1.1077031984916244,
      "grad_norm": 0.5838775241989291,
      "learning_rate": 8.744952933267163e-05,
      "loss": 11.9967,
      "step": 20342
    },
    {
      "epoch": 1.1077576524882073,
      "grad_norm": 0.5905911772599802,
      "learning_rate": 8.744078084907021e-05,
      "loss": 12.1397,
      "step": 20343
    },
    {
      "epoch": 1.1078121064847903,
      "grad_norm": 0.49872395232447336,
      "learning_rate": 8.743203246313136e-05,
      "loss": 11.8628,
      "step": 20344
    },
    {
      "epoch": 1.1078665604813733,
      "grad_norm": 0.5638960798286776,
      "learning_rate": 8.742328417492316e-05,
      "loss": 12.0169,
      "step": 20345
    },
    {
      "epoch": 1.1079210144779563,
      "grad_norm": 0.5880102201495002,
      "learning_rate": 8.74145359845136e-05,
      "loss": 12.0852,
      "step": 20346
    },
    {
      "epoch": 1.1079754684745393,
      "grad_norm": 0.5782451800864982,
      "learning_rate": 8.740578789197071e-05,
      "loss": 12.0772,
      "step": 20347
    },
    {
      "epoch": 1.1080299224711223,
      "grad_norm": 0.5032096969628987,
      "learning_rate": 8.739703989736252e-05,
      "loss": 11.8333,
      "step": 20348
    },
    {
      "epoch": 1.1080843764677053,
      "grad_norm": 0.5392127032883857,
      "learning_rate": 8.738829200075707e-05,
      "loss": 12.1177,
      "step": 20349
    },
    {
      "epoch": 1.1081388304642883,
      "grad_norm": 0.5569545293518601,
      "learning_rate": 8.737954420222243e-05,
      "loss": 11.9352,
      "step": 20350
    },
    {
      "epoch": 1.1081932844608713,
      "grad_norm": 0.5571004661020185,
      "learning_rate": 8.737079650182653e-05,
      "loss": 12.1027,
      "step": 20351
    },
    {
      "epoch": 1.1082477384574545,
      "grad_norm": 0.5202757639842748,
      "learning_rate": 8.73620488996374e-05,
      "loss": 11.9588,
      "step": 20352
    },
    {
      "epoch": 1.1083021924540375,
      "grad_norm": 0.5326218707584636,
      "learning_rate": 8.735330139572312e-05,
      "loss": 12.1588,
      "step": 20353
    },
    {
      "epoch": 1.1083566464506205,
      "grad_norm": 0.581581715781423,
      "learning_rate": 8.73445539901517e-05,
      "loss": 12.0124,
      "step": 20354
    },
    {
      "epoch": 1.1084111004472035,
      "grad_norm": 0.5643813781203146,
      "learning_rate": 8.733580668299113e-05,
      "loss": 12.0239,
      "step": 20355
    },
    {
      "epoch": 1.1084655544437865,
      "grad_norm": 0.4929896961613901,
      "learning_rate": 8.732705947430948e-05,
      "loss": 11.9148,
      "step": 20356
    },
    {
      "epoch": 1.1085200084403695,
      "grad_norm": 0.5223543752886799,
      "learning_rate": 8.731831236417472e-05,
      "loss": 11.9633,
      "step": 20357
    },
    {
      "epoch": 1.1085744624369525,
      "grad_norm": 0.5729833090573728,
      "learning_rate": 8.73095653526549e-05,
      "loss": 12.0722,
      "step": 20358
    },
    {
      "epoch": 1.1086289164335354,
      "grad_norm": 0.5251201813942938,
      "learning_rate": 8.7300818439818e-05,
      "loss": 12.0149,
      "step": 20359
    },
    {
      "epoch": 1.1086833704301184,
      "grad_norm": 0.558311215109325,
      "learning_rate": 8.729207162573214e-05,
      "loss": 11.9108,
      "step": 20360
    },
    {
      "epoch": 1.1087378244267014,
      "grad_norm": 0.5365779165147309,
      "learning_rate": 8.72833249104652e-05,
      "loss": 12.0024,
      "step": 20361
    },
    {
      "epoch": 1.1087922784232844,
      "grad_norm": 0.5151686326710411,
      "learning_rate": 8.727457829408527e-05,
      "loss": 11.9575,
      "step": 20362
    },
    {
      "epoch": 1.1088467324198676,
      "grad_norm": 0.576146484496607,
      "learning_rate": 8.726583177666034e-05,
      "loss": 12.146,
      "step": 20363
    },
    {
      "epoch": 1.1089011864164506,
      "grad_norm": 0.5604576831869068,
      "learning_rate": 8.725708535825845e-05,
      "loss": 12.0227,
      "step": 20364
    },
    {
      "epoch": 1.1089556404130336,
      "grad_norm": 0.5488182443925755,
      "learning_rate": 8.724833903894761e-05,
      "loss": 12.0595,
      "step": 20365
    },
    {
      "epoch": 1.1090100944096166,
      "grad_norm": 0.5702859955842696,
      "learning_rate": 8.72395928187958e-05,
      "loss": 12.0216,
      "step": 20366
    },
    {
      "epoch": 1.1090645484061996,
      "grad_norm": 0.630035859859341,
      "learning_rate": 8.723084669787107e-05,
      "loss": 12.0802,
      "step": 20367
    },
    {
      "epoch": 1.1091190024027826,
      "grad_norm": 0.5346963959972917,
      "learning_rate": 8.722210067624143e-05,
      "loss": 12.0565,
      "step": 20368
    },
    {
      "epoch": 1.1091734563993656,
      "grad_norm": 0.5638355168601141,
      "learning_rate": 8.721335475397486e-05,
      "loss": 12.0975,
      "step": 20369
    },
    {
      "epoch": 1.1092279103959486,
      "grad_norm": 0.544604177941654,
      "learning_rate": 8.72046089311394e-05,
      "loss": 11.9492,
      "step": 20370
    },
    {
      "epoch": 1.1092823643925316,
      "grad_norm": 0.5112490088018798,
      "learning_rate": 8.719586320780307e-05,
      "loss": 11.96,
      "step": 20371
    },
    {
      "epoch": 1.1093368183891146,
      "grad_norm": 0.5328271332370599,
      "learning_rate": 8.718711758403382e-05,
      "loss": 11.9347,
      "step": 20372
    },
    {
      "epoch": 1.1093912723856976,
      "grad_norm": 0.5674785663412966,
      "learning_rate": 8.717837205989969e-05,
      "loss": 11.9964,
      "step": 20373
    },
    {
      "epoch": 1.1094457263822806,
      "grad_norm": 0.5953098610929797,
      "learning_rate": 8.71696266354687e-05,
      "loss": 12.0257,
      "step": 20374
    },
    {
      "epoch": 1.1095001803788638,
      "grad_norm": 0.5529267917614739,
      "learning_rate": 8.716088131080882e-05,
      "loss": 11.9219,
      "step": 20375
    },
    {
      "epoch": 1.1095546343754468,
      "grad_norm": 0.612605982175686,
      "learning_rate": 8.715213608598811e-05,
      "loss": 12.0591,
      "step": 20376
    },
    {
      "epoch": 1.1096090883720298,
      "grad_norm": 0.6502316142012194,
      "learning_rate": 8.714339096107454e-05,
      "loss": 12.023,
      "step": 20377
    },
    {
      "epoch": 1.1096635423686128,
      "grad_norm": 0.5152773212699467,
      "learning_rate": 8.713464593613612e-05,
      "loss": 11.94,
      "step": 20378
    },
    {
      "epoch": 1.1097179963651957,
      "grad_norm": 0.5519647029298497,
      "learning_rate": 8.712590101124084e-05,
      "loss": 12.0978,
      "step": 20379
    },
    {
      "epoch": 1.1097724503617787,
      "grad_norm": 0.5142008477824004,
      "learning_rate": 8.711715618645671e-05,
      "loss": 12.1463,
      "step": 20380
    },
    {
      "epoch": 1.1098269043583617,
      "grad_norm": 0.5431164123937956,
      "learning_rate": 8.710841146185177e-05,
      "loss": 12.0085,
      "step": 20381
    },
    {
      "epoch": 1.1098813583549447,
      "grad_norm": 0.5465098198668166,
      "learning_rate": 8.709966683749396e-05,
      "loss": 11.8521,
      "step": 20382
    },
    {
      "epoch": 1.1099358123515277,
      "grad_norm": 0.5771607951376807,
      "learning_rate": 8.70909223134513e-05,
      "loss": 12.0074,
      "step": 20383
    },
    {
      "epoch": 1.1099902663481107,
      "grad_norm": 0.5732513893159404,
      "learning_rate": 8.70821778897918e-05,
      "loss": 12.0577,
      "step": 20384
    },
    {
      "epoch": 1.1100447203446937,
      "grad_norm": 0.5687566895392488,
      "learning_rate": 8.707343356658344e-05,
      "loss": 12.0488,
      "step": 20385
    },
    {
      "epoch": 1.110099174341277,
      "grad_norm": 0.5436380291903499,
      "learning_rate": 8.706468934389421e-05,
      "loss": 11.9659,
      "step": 20386
    },
    {
      "epoch": 1.11015362833786,
      "grad_norm": 0.5525146804435588,
      "learning_rate": 8.705594522179214e-05,
      "loss": 12.0789,
      "step": 20387
    },
    {
      "epoch": 1.110208082334443,
      "grad_norm": 0.5607722918179316,
      "learning_rate": 8.704720120034523e-05,
      "loss": 12.0246,
      "step": 20388
    },
    {
      "epoch": 1.110262536331026,
      "grad_norm": 0.5643884542944015,
      "learning_rate": 8.703845727962144e-05,
      "loss": 12.0696,
      "step": 20389
    },
    {
      "epoch": 1.1103169903276089,
      "grad_norm": 0.5559613977033397,
      "learning_rate": 8.702971345968879e-05,
      "loss": 12.0368,
      "step": 20390
    },
    {
      "epoch": 1.1103714443241919,
      "grad_norm": 0.5926001588421552,
      "learning_rate": 8.702096974061527e-05,
      "loss": 11.9943,
      "step": 20391
    },
    {
      "epoch": 1.1104258983207749,
      "grad_norm": 0.5782246297613625,
      "learning_rate": 8.701222612246887e-05,
      "loss": 12.0798,
      "step": 20392
    },
    {
      "epoch": 1.1104803523173579,
      "grad_norm": 0.5650841524659677,
      "learning_rate": 8.700348260531756e-05,
      "loss": 12.1178,
      "step": 20393
    },
    {
      "epoch": 1.1105348063139409,
      "grad_norm": 0.527666922247384,
      "learning_rate": 8.699473918922934e-05,
      "loss": 11.9583,
      "step": 20394
    },
    {
      "epoch": 1.1105892603105239,
      "grad_norm": 0.4831398814838757,
      "learning_rate": 8.698599587427223e-05,
      "loss": 12.0063,
      "step": 20395
    },
    {
      "epoch": 1.1106437143071068,
      "grad_norm": 0.6126566647506558,
      "learning_rate": 8.697725266051419e-05,
      "loss": 12.1569,
      "step": 20396
    },
    {
      "epoch": 1.1106981683036898,
      "grad_norm": 0.6469344452197978,
      "learning_rate": 8.696850954802319e-05,
      "loss": 12.0177,
      "step": 20397
    },
    {
      "epoch": 1.110752622300273,
      "grad_norm": 0.5665602869832791,
      "learning_rate": 8.695976653686726e-05,
      "loss": 12.167,
      "step": 20398
    },
    {
      "epoch": 1.110807076296856,
      "grad_norm": 0.5185074852028833,
      "learning_rate": 8.695102362711439e-05,
      "loss": 12.0196,
      "step": 20399
    },
    {
      "epoch": 1.110861530293439,
      "grad_norm": 0.5260885913422563,
      "learning_rate": 8.694228081883254e-05,
      "loss": 12.0168,
      "step": 20400
    },
    {
      "epoch": 1.110915984290022,
      "grad_norm": 0.6708326897002365,
      "learning_rate": 8.693353811208973e-05,
      "loss": 12.0019,
      "step": 20401
    },
    {
      "epoch": 1.110970438286605,
      "grad_norm": 0.5639695878205355,
      "learning_rate": 8.69247955069539e-05,
      "loss": 12.1039,
      "step": 20402
    },
    {
      "epoch": 1.111024892283188,
      "grad_norm": 0.5449293478224777,
      "learning_rate": 8.691605300349304e-05,
      "loss": 12.0094,
      "step": 20403
    },
    {
      "epoch": 1.111079346279771,
      "grad_norm": 0.5740721272416982,
      "learning_rate": 8.690731060177515e-05,
      "loss": 12.112,
      "step": 20404
    },
    {
      "epoch": 1.111133800276354,
      "grad_norm": 0.4930037984290305,
      "learning_rate": 8.68985683018682e-05,
      "loss": 12.0775,
      "step": 20405
    },
    {
      "epoch": 1.111188254272937,
      "grad_norm": 0.6027153909312933,
      "learning_rate": 8.68898261038402e-05,
      "loss": 11.9441,
      "step": 20406
    },
    {
      "epoch": 1.11124270826952,
      "grad_norm": 0.5260394731253033,
      "learning_rate": 8.68810840077591e-05,
      "loss": 11.9547,
      "step": 20407
    },
    {
      "epoch": 1.111297162266103,
      "grad_norm": 0.604684947929935,
      "learning_rate": 8.687234201369287e-05,
      "loss": 12.1559,
      "step": 20408
    },
    {
      "epoch": 1.1113516162626862,
      "grad_norm": 0.5853957597714421,
      "learning_rate": 8.686360012170954e-05,
      "loss": 11.8884,
      "step": 20409
    },
    {
      "epoch": 1.1114060702592692,
      "grad_norm": 0.5204564010716333,
      "learning_rate": 8.685485833187706e-05,
      "loss": 12.0701,
      "step": 20410
    },
    {
      "epoch": 1.1114605242558522,
      "grad_norm": 0.554571373235644,
      "learning_rate": 8.684611664426344e-05,
      "loss": 12.0565,
      "step": 20411
    },
    {
      "epoch": 1.1115149782524352,
      "grad_norm": 0.49278774796167696,
      "learning_rate": 8.683737505893655e-05,
      "loss": 11.9681,
      "step": 20412
    },
    {
      "epoch": 1.1115694322490182,
      "grad_norm": 0.5799618073088181,
      "learning_rate": 8.682863357596447e-05,
      "loss": 11.8597,
      "step": 20413
    },
    {
      "epoch": 1.1116238862456012,
      "grad_norm": 0.6174819718064491,
      "learning_rate": 8.681989219541516e-05,
      "loss": 12.0611,
      "step": 20414
    },
    {
      "epoch": 1.1116783402421841,
      "grad_norm": 0.5563880126339784,
      "learning_rate": 8.681115091735654e-05,
      "loss": 12.0307,
      "step": 20415
    },
    {
      "epoch": 1.1117327942387671,
      "grad_norm": 0.5428688058215628,
      "learning_rate": 8.680240974185665e-05,
      "loss": 11.9264,
      "step": 20416
    },
    {
      "epoch": 1.1117872482353501,
      "grad_norm": 0.6190313780390646,
      "learning_rate": 8.679366866898343e-05,
      "loss": 12.1029,
      "step": 20417
    },
    {
      "epoch": 1.1118417022319331,
      "grad_norm": 0.5624661924655168,
      "learning_rate": 8.678492769880486e-05,
      "loss": 12.0339,
      "step": 20418
    },
    {
      "epoch": 1.1118961562285161,
      "grad_norm": 0.5296270992929764,
      "learning_rate": 8.677618683138889e-05,
      "loss": 12.0566,
      "step": 20419
    },
    {
      "epoch": 1.1119506102250991,
      "grad_norm": 0.6563129787591534,
      "learning_rate": 8.676744606680352e-05,
      "loss": 12.0153,
      "step": 20420
    },
    {
      "epoch": 1.112005064221682,
      "grad_norm": 0.47750180191668606,
      "learning_rate": 8.675870540511675e-05,
      "loss": 12.0877,
      "step": 20421
    },
    {
      "epoch": 1.1120595182182653,
      "grad_norm": 0.5971189211479879,
      "learning_rate": 8.674996484639647e-05,
      "loss": 12.052,
      "step": 20422
    },
    {
      "epoch": 1.1121139722148483,
      "grad_norm": 0.5647061566072635,
      "learning_rate": 8.674122439071069e-05,
      "loss": 11.9557,
      "step": 20423
    },
    {
      "epoch": 1.1121684262114313,
      "grad_norm": 0.5718603926874921,
      "learning_rate": 8.673248403812735e-05,
      "loss": 12.0426,
      "step": 20424
    },
    {
      "epoch": 1.1122228802080143,
      "grad_norm": 0.5862622020485386,
      "learning_rate": 8.672374378871445e-05,
      "loss": 11.9336,
      "step": 20425
    },
    {
      "epoch": 1.1122773342045973,
      "grad_norm": 0.5319568509209703,
      "learning_rate": 8.671500364253995e-05,
      "loss": 12.0574,
      "step": 20426
    },
    {
      "epoch": 1.1123317882011803,
      "grad_norm": 0.5581882484166962,
      "learning_rate": 8.670626359967181e-05,
      "loss": 11.901,
      "step": 20427
    },
    {
      "epoch": 1.1123862421977633,
      "grad_norm": 0.5718910216535872,
      "learning_rate": 8.669752366017799e-05,
      "loss": 11.9544,
      "step": 20428
    },
    {
      "epoch": 1.1124406961943463,
      "grad_norm": 0.535119723211312,
      "learning_rate": 8.668878382412646e-05,
      "loss": 11.8905,
      "step": 20429
    },
    {
      "epoch": 1.1124951501909293,
      "grad_norm": 0.5585066425693939,
      "learning_rate": 8.668004409158519e-05,
      "loss": 12.0553,
      "step": 20430
    },
    {
      "epoch": 1.1125496041875123,
      "grad_norm": 0.5887672763553066,
      "learning_rate": 8.667130446262214e-05,
      "loss": 12.0295,
      "step": 20431
    },
    {
      "epoch": 1.1126040581840952,
      "grad_norm": 0.5512357232765116,
      "learning_rate": 8.666256493730525e-05,
      "loss": 12.0979,
      "step": 20432
    },
    {
      "epoch": 1.1126585121806785,
      "grad_norm": 0.5512854382094922,
      "learning_rate": 8.665382551570248e-05,
      "loss": 12.1492,
      "step": 20433
    },
    {
      "epoch": 1.1127129661772615,
      "grad_norm": 0.5461053454512196,
      "learning_rate": 8.664508619788181e-05,
      "loss": 12.1001,
      "step": 20434
    },
    {
      "epoch": 1.1127674201738444,
      "grad_norm": 0.6048445467025007,
      "learning_rate": 8.663634698391117e-05,
      "loss": 12.1017,
      "step": 20435
    },
    {
      "epoch": 1.1128218741704274,
      "grad_norm": 0.5082271795316599,
      "learning_rate": 8.662760787385854e-05,
      "loss": 12.0915,
      "step": 20436
    },
    {
      "epoch": 1.1128763281670104,
      "grad_norm": 0.5304286711573969,
      "learning_rate": 8.661886886779189e-05,
      "loss": 12.0,
      "step": 20437
    },
    {
      "epoch": 1.1129307821635934,
      "grad_norm": 0.641587832548868,
      "learning_rate": 8.661012996577915e-05,
      "loss": 12.0823,
      "step": 20438
    },
    {
      "epoch": 1.1129852361601764,
      "grad_norm": 0.5334425328832866,
      "learning_rate": 8.66013911678883e-05,
      "loss": 12.0591,
      "step": 20439
    },
    {
      "epoch": 1.1130396901567594,
      "grad_norm": 0.5508297466926332,
      "learning_rate": 8.659265247418727e-05,
      "loss": 11.9763,
      "step": 20440
    },
    {
      "epoch": 1.1130941441533424,
      "grad_norm": 0.6130747054086031,
      "learning_rate": 8.658391388474404e-05,
      "loss": 12.0046,
      "step": 20441
    },
    {
      "epoch": 1.1131485981499254,
      "grad_norm": 0.5395795817917738,
      "learning_rate": 8.657517539962654e-05,
      "loss": 12.0152,
      "step": 20442
    },
    {
      "epoch": 1.1132030521465084,
      "grad_norm": 0.6118794450591564,
      "learning_rate": 8.65664370189027e-05,
      "loss": 11.8971,
      "step": 20443
    },
    {
      "epoch": 1.1132575061430914,
      "grad_norm": 0.5236304783024495,
      "learning_rate": 8.655769874264052e-05,
      "loss": 11.9936,
      "step": 20444
    },
    {
      "epoch": 1.1133119601396746,
      "grad_norm": 0.6092612193110597,
      "learning_rate": 8.654896057090792e-05,
      "loss": 12.0868,
      "step": 20445
    },
    {
      "epoch": 1.1133664141362576,
      "grad_norm": 0.5321936221086799,
      "learning_rate": 8.654022250377283e-05,
      "loss": 12.0753,
      "step": 20446
    },
    {
      "epoch": 1.1134208681328406,
      "grad_norm": 0.6038692861905606,
      "learning_rate": 8.653148454130327e-05,
      "loss": 12.0654,
      "step": 20447
    },
    {
      "epoch": 1.1134753221294236,
      "grad_norm": 0.5325827501991652,
      "learning_rate": 8.652274668356713e-05,
      "loss": 12.0671,
      "step": 20448
    },
    {
      "epoch": 1.1135297761260066,
      "grad_norm": 0.5699565426641002,
      "learning_rate": 8.651400893063238e-05,
      "loss": 11.9932,
      "step": 20449
    },
    {
      "epoch": 1.1135842301225896,
      "grad_norm": 0.5456973476350043,
      "learning_rate": 8.650527128256695e-05,
      "loss": 12.0736,
      "step": 20450
    },
    {
      "epoch": 1.1136386841191726,
      "grad_norm": 0.6364466397889185,
      "learning_rate": 8.649653373943883e-05,
      "loss": 12.1013,
      "step": 20451
    },
    {
      "epoch": 1.1136931381157555,
      "grad_norm": 0.5958478372373719,
      "learning_rate": 8.648779630131589e-05,
      "loss": 11.9619,
      "step": 20452
    },
    {
      "epoch": 1.1137475921123385,
      "grad_norm": 0.5976355951406671,
      "learning_rate": 8.647905896826611e-05,
      "loss": 12.0252,
      "step": 20453
    },
    {
      "epoch": 1.1138020461089215,
      "grad_norm": 0.5255235963552749,
      "learning_rate": 8.647032174035744e-05,
      "loss": 12.146,
      "step": 20454
    },
    {
      "epoch": 1.1138565001055045,
      "grad_norm": 0.6238917781714524,
      "learning_rate": 8.646158461765782e-05,
      "loss": 12.003,
      "step": 20455
    },
    {
      "epoch": 1.1139109541020877,
      "grad_norm": 0.5990554357556224,
      "learning_rate": 8.645284760023519e-05,
      "loss": 11.9492,
      "step": 20456
    },
    {
      "epoch": 1.1139654080986707,
      "grad_norm": 0.5639341509059039,
      "learning_rate": 8.644411068815747e-05,
      "loss": 11.8785,
      "step": 20457
    },
    {
      "epoch": 1.1140198620952537,
      "grad_norm": 0.5225938502442502,
      "learning_rate": 8.643537388149263e-05,
      "loss": 12.0676,
      "step": 20458
    },
    {
      "epoch": 1.1140743160918367,
      "grad_norm": 0.6080400007278929,
      "learning_rate": 8.64266371803086e-05,
      "loss": 12.0214,
      "step": 20459
    },
    {
      "epoch": 1.1141287700884197,
      "grad_norm": 0.5962382006537994,
      "learning_rate": 8.641790058467332e-05,
      "loss": 12.0873,
      "step": 20460
    },
    {
      "epoch": 1.1141832240850027,
      "grad_norm": 0.5414322662812325,
      "learning_rate": 8.640916409465474e-05,
      "loss": 12.0977,
      "step": 20461
    },
    {
      "epoch": 1.1142376780815857,
      "grad_norm": 0.5442969491055653,
      "learning_rate": 8.640042771032076e-05,
      "loss": 11.9356,
      "step": 20462
    },
    {
      "epoch": 1.1142921320781687,
      "grad_norm": 0.5857955462241692,
      "learning_rate": 8.639169143173934e-05,
      "loss": 11.9875,
      "step": 20463
    },
    {
      "epoch": 1.1143465860747517,
      "grad_norm": 0.5722991266071655,
      "learning_rate": 8.63829552589784e-05,
      "loss": 11.9408,
      "step": 20464
    },
    {
      "epoch": 1.1144010400713347,
      "grad_norm": 0.5745187959531379,
      "learning_rate": 8.637421919210588e-05,
      "loss": 12.0425,
      "step": 20465
    },
    {
      "epoch": 1.1144554940679177,
      "grad_norm": 0.5737292188169556,
      "learning_rate": 8.636548323118974e-05,
      "loss": 12.011,
      "step": 20466
    },
    {
      "epoch": 1.1145099480645007,
      "grad_norm": 0.5953915731831864,
      "learning_rate": 8.635674737629786e-05,
      "loss": 11.9588,
      "step": 20467
    },
    {
      "epoch": 1.1145644020610839,
      "grad_norm": 0.6105710976926062,
      "learning_rate": 8.634801162749819e-05,
      "loss": 12.0818,
      "step": 20468
    },
    {
      "epoch": 1.1146188560576669,
      "grad_norm": 0.4860264760757827,
      "learning_rate": 8.63392759848587e-05,
      "loss": 11.9673,
      "step": 20469
    },
    {
      "epoch": 1.1146733100542499,
      "grad_norm": 0.4796813075837099,
      "learning_rate": 8.633054044844729e-05,
      "loss": 12.024,
      "step": 20470
    },
    {
      "epoch": 1.1147277640508328,
      "grad_norm": 0.5950842721789306,
      "learning_rate": 8.632180501833192e-05,
      "loss": 12.0347,
      "step": 20471
    },
    {
      "epoch": 1.1147822180474158,
      "grad_norm": 0.6542190907390353,
      "learning_rate": 8.631306969458047e-05,
      "loss": 12.1151,
      "step": 20472
    },
    {
      "epoch": 1.1148366720439988,
      "grad_norm": 0.5149019425783683,
      "learning_rate": 8.630433447726084e-05,
      "loss": 12.0373,
      "step": 20473
    },
    {
      "epoch": 1.1148911260405818,
      "grad_norm": 0.5401207092106911,
      "learning_rate": 8.629559936644103e-05,
      "loss": 12.0794,
      "step": 20474
    },
    {
      "epoch": 1.1149455800371648,
      "grad_norm": 0.5253189533374024,
      "learning_rate": 8.628686436218894e-05,
      "loss": 11.6059,
      "step": 20475
    },
    {
      "epoch": 1.1150000340337478,
      "grad_norm": 0.5998630256837638,
      "learning_rate": 8.627812946457249e-05,
      "loss": 12.0531,
      "step": 20476
    },
    {
      "epoch": 1.1150544880303308,
      "grad_norm": 0.5713336693894561,
      "learning_rate": 8.626939467365961e-05,
      "loss": 12.1183,
      "step": 20477
    },
    {
      "epoch": 1.1151089420269138,
      "grad_norm": 0.6539845428263784,
      "learning_rate": 8.626065998951821e-05,
      "loss": 12.0565,
      "step": 20478
    },
    {
      "epoch": 1.115163396023497,
      "grad_norm": 0.5795015131651489,
      "learning_rate": 8.625192541221623e-05,
      "loss": 12.1167,
      "step": 20479
    },
    {
      "epoch": 1.11521785002008,
      "grad_norm": 0.6687968959099096,
      "learning_rate": 8.624319094182157e-05,
      "loss": 11.9152,
      "step": 20480
    },
    {
      "epoch": 1.115272304016663,
      "grad_norm": 0.5662463661833551,
      "learning_rate": 8.623445657840222e-05,
      "loss": 12.0687,
      "step": 20481
    },
    {
      "epoch": 1.115326758013246,
      "grad_norm": 0.5569974547007831,
      "learning_rate": 8.622572232202599e-05,
      "loss": 12.0152,
      "step": 20482
    },
    {
      "epoch": 1.115381212009829,
      "grad_norm": 0.5356611435949286,
      "learning_rate": 8.621698817276087e-05,
      "loss": 11.987,
      "step": 20483
    },
    {
      "epoch": 1.115435666006412,
      "grad_norm": 0.5740871642686105,
      "learning_rate": 8.620825413067473e-05,
      "loss": 12.1727,
      "step": 20484
    },
    {
      "epoch": 1.115490120002995,
      "grad_norm": 0.5397304764024144,
      "learning_rate": 8.619952019583555e-05,
      "loss": 11.9573,
      "step": 20485
    },
    {
      "epoch": 1.115544573999578,
      "grad_norm": 0.5413113226753937,
      "learning_rate": 8.61907863683112e-05,
      "loss": 12.1108,
      "step": 20486
    },
    {
      "epoch": 1.115599027996161,
      "grad_norm": 0.5224978649881019,
      "learning_rate": 8.618205264816962e-05,
      "loss": 12.0286,
      "step": 20487
    },
    {
      "epoch": 1.115653481992744,
      "grad_norm": 0.49798079977913895,
      "learning_rate": 8.617331903547872e-05,
      "loss": 11.9539,
      "step": 20488
    },
    {
      "epoch": 1.115707935989327,
      "grad_norm": 0.5374865498581153,
      "learning_rate": 8.616458553030641e-05,
      "loss": 11.9546,
      "step": 20489
    },
    {
      "epoch": 1.11576238998591,
      "grad_norm": 0.5421550574793342,
      "learning_rate": 8.615585213272059e-05,
      "loss": 12.0558,
      "step": 20490
    },
    {
      "epoch": 1.115816843982493,
      "grad_norm": 0.5236490567634463,
      "learning_rate": 8.614711884278922e-05,
      "loss": 11.924,
      "step": 20491
    },
    {
      "epoch": 1.1158712979790761,
      "grad_norm": 0.5894706823009638,
      "learning_rate": 8.613838566058014e-05,
      "loss": 12.1306,
      "step": 20492
    },
    {
      "epoch": 1.1159257519756591,
      "grad_norm": 0.4902113230346238,
      "learning_rate": 8.612965258616133e-05,
      "loss": 11.9747,
      "step": 20493
    },
    {
      "epoch": 1.1159802059722421,
      "grad_norm": 0.5215407998834232,
      "learning_rate": 8.612091961960064e-05,
      "loss": 12.0714,
      "step": 20494
    },
    {
      "epoch": 1.1160346599688251,
      "grad_norm": 0.5243379429072624,
      "learning_rate": 8.611218676096599e-05,
      "loss": 12.0208,
      "step": 20495
    },
    {
      "epoch": 1.116089113965408,
      "grad_norm": 0.6471210179145048,
      "learning_rate": 8.610345401032532e-05,
      "loss": 12.1623,
      "step": 20496
    },
    {
      "epoch": 1.116143567961991,
      "grad_norm": 0.4908431841804648,
      "learning_rate": 8.609472136774654e-05,
      "loss": 11.8585,
      "step": 20497
    },
    {
      "epoch": 1.116198021958574,
      "grad_norm": 0.5336784621895109,
      "learning_rate": 8.608598883329752e-05,
      "loss": 11.7578,
      "step": 20498
    },
    {
      "epoch": 1.116252475955157,
      "grad_norm": 0.539822673026537,
      "learning_rate": 8.60772564070462e-05,
      "loss": 12.0355,
      "step": 20499
    },
    {
      "epoch": 1.11630692995174,
      "grad_norm": 0.5368551058329707,
      "learning_rate": 8.606852408906047e-05,
      "loss": 11.9642,
      "step": 20500
    },
    {
      "epoch": 1.116361383948323,
      "grad_norm": 0.5519632231326893,
      "learning_rate": 8.605979187940823e-05,
      "loss": 12.0173,
      "step": 20501
    },
    {
      "epoch": 1.1164158379449063,
      "grad_norm": 0.558058323641093,
      "learning_rate": 8.605105977815739e-05,
      "loss": 11.9624,
      "step": 20502
    },
    {
      "epoch": 1.1164702919414893,
      "grad_norm": 0.5412411084680292,
      "learning_rate": 8.604232778537584e-05,
      "loss": 12.0976,
      "step": 20503
    },
    {
      "epoch": 1.1165247459380723,
      "grad_norm": 0.6807011047904982,
      "learning_rate": 8.60335959011315e-05,
      "loss": 12.1143,
      "step": 20504
    },
    {
      "epoch": 1.1165791999346553,
      "grad_norm": 0.5167678272691687,
      "learning_rate": 8.602486412549225e-05,
      "loss": 11.9778,
      "step": 20505
    },
    {
      "epoch": 1.1166336539312383,
      "grad_norm": 0.5476282127282881,
      "learning_rate": 8.601613245852597e-05,
      "loss": 12.1037,
      "step": 20506
    },
    {
      "epoch": 1.1166881079278212,
      "grad_norm": 0.4953667512885385,
      "learning_rate": 8.600740090030062e-05,
      "loss": 11.962,
      "step": 20507
    },
    {
      "epoch": 1.1167425619244042,
      "grad_norm": 0.5890134000418888,
      "learning_rate": 8.599866945088406e-05,
      "loss": 12.0408,
      "step": 20508
    },
    {
      "epoch": 1.1167970159209872,
      "grad_norm": 0.5883335810102944,
      "learning_rate": 8.59899381103442e-05,
      "loss": 12.1385,
      "step": 20509
    },
    {
      "epoch": 1.1168514699175702,
      "grad_norm": 0.5502270776658903,
      "learning_rate": 8.598120687874893e-05,
      "loss": 11.9543,
      "step": 20510
    },
    {
      "epoch": 1.1169059239141532,
      "grad_norm": 0.5549793460829717,
      "learning_rate": 8.597247575616615e-05,
      "loss": 11.8679,
      "step": 20511
    },
    {
      "epoch": 1.1169603779107362,
      "grad_norm": 0.5659023150716123,
      "learning_rate": 8.596374474266378e-05,
      "loss": 12.1417,
      "step": 20512
    },
    {
      "epoch": 1.1170148319073192,
      "grad_norm": 0.5642468541772573,
      "learning_rate": 8.595501383830963e-05,
      "loss": 12.0823,
      "step": 20513
    },
    {
      "epoch": 1.1170692859039022,
      "grad_norm": 0.5123625037887571,
      "learning_rate": 8.594628304317168e-05,
      "loss": 12.0234,
      "step": 20514
    },
    {
      "epoch": 1.1171237399004854,
      "grad_norm": 0.47432507338245566,
      "learning_rate": 8.593755235731779e-05,
      "loss": 11.8732,
      "step": 20515
    },
    {
      "epoch": 1.1171781938970684,
      "grad_norm": 0.5304811960593719,
      "learning_rate": 8.592882178081584e-05,
      "loss": 11.974,
      "step": 20516
    },
    {
      "epoch": 1.1172326478936514,
      "grad_norm": 0.5517583033909015,
      "learning_rate": 8.59200913137337e-05,
      "loss": 12.0905,
      "step": 20517
    },
    {
      "epoch": 1.1172871018902344,
      "grad_norm": 0.46713763306512546,
      "learning_rate": 8.591136095613934e-05,
      "loss": 11.9744,
      "step": 20518
    },
    {
      "epoch": 1.1173415558868174,
      "grad_norm": 0.5486187992330056,
      "learning_rate": 8.590263070810058e-05,
      "loss": 12.0381,
      "step": 20519
    },
    {
      "epoch": 1.1173960098834004,
      "grad_norm": 0.5527723804788405,
      "learning_rate": 8.589390056968534e-05,
      "loss": 11.9795,
      "step": 20520
    },
    {
      "epoch": 1.1174504638799834,
      "grad_norm": 0.5018470446533754,
      "learning_rate": 8.588517054096147e-05,
      "loss": 12.0284,
      "step": 20521
    },
    {
      "epoch": 1.1175049178765664,
      "grad_norm": 0.5263256875552247,
      "learning_rate": 8.587644062199694e-05,
      "loss": 11.9745,
      "step": 20522
    },
    {
      "epoch": 1.1175593718731494,
      "grad_norm": 0.5830871320910996,
      "learning_rate": 8.586771081285952e-05,
      "loss": 12.161,
      "step": 20523
    },
    {
      "epoch": 1.1176138258697323,
      "grad_norm": 0.5812253637558997,
      "learning_rate": 8.585898111361716e-05,
      "loss": 11.9857,
      "step": 20524
    },
    {
      "epoch": 1.1176682798663153,
      "grad_norm": 0.6240263377907601,
      "learning_rate": 8.585025152433775e-05,
      "loss": 12.0903,
      "step": 20525
    },
    {
      "epoch": 1.1177227338628986,
      "grad_norm": 0.5347853323347552,
      "learning_rate": 8.584152204508916e-05,
      "loss": 12.0122,
      "step": 20526
    },
    {
      "epoch": 1.1177771878594815,
      "grad_norm": 0.5142875269130167,
      "learning_rate": 8.583279267593924e-05,
      "loss": 12.0679,
      "step": 20527
    },
    {
      "epoch": 1.1178316418560645,
      "grad_norm": 0.524234472729522,
      "learning_rate": 8.582406341695591e-05,
      "loss": 11.9171,
      "step": 20528
    },
    {
      "epoch": 1.1178860958526475,
      "grad_norm": 0.6411909173864685,
      "learning_rate": 8.581533426820703e-05,
      "loss": 11.9259,
      "step": 20529
    },
    {
      "epoch": 1.1179405498492305,
      "grad_norm": 0.5997938318227984,
      "learning_rate": 8.580660522976051e-05,
      "loss": 12.0067,
      "step": 20530
    },
    {
      "epoch": 1.1179950038458135,
      "grad_norm": 0.5543244796967985,
      "learning_rate": 8.57978763016842e-05,
      "loss": 11.9591,
      "step": 20531
    },
    {
      "epoch": 1.1180494578423965,
      "grad_norm": 0.511095836273182,
      "learning_rate": 8.578914748404603e-05,
      "loss": 11.9835,
      "step": 20532
    },
    {
      "epoch": 1.1181039118389795,
      "grad_norm": 0.527984273724554,
      "learning_rate": 8.578041877691376e-05,
      "loss": 12.0705,
      "step": 20533
    },
    {
      "epoch": 1.1181583658355625,
      "grad_norm": 0.5387854188171488,
      "learning_rate": 8.577169018035537e-05,
      "loss": 11.9265,
      "step": 20534
    },
    {
      "epoch": 1.1182128198321455,
      "grad_norm": 0.5687718892618461,
      "learning_rate": 8.576296169443872e-05,
      "loss": 11.8669,
      "step": 20535
    },
    {
      "epoch": 1.1182672738287285,
      "grad_norm": 0.5026518080973875,
      "learning_rate": 8.575423331923164e-05,
      "loss": 12.0234,
      "step": 20536
    },
    {
      "epoch": 1.1183217278253115,
      "grad_norm": 0.5401622858273115,
      "learning_rate": 8.574550505480204e-05,
      "loss": 12.0075,
      "step": 20537
    },
    {
      "epoch": 1.1183761818218947,
      "grad_norm": 0.5619456287560358,
      "learning_rate": 8.573677690121779e-05,
      "loss": 11.9651,
      "step": 20538
    },
    {
      "epoch": 1.1184306358184777,
      "grad_norm": 0.8599802351019422,
      "learning_rate": 8.572804885854676e-05,
      "loss": 12.0924,
      "step": 20539
    },
    {
      "epoch": 1.1184850898150607,
      "grad_norm": 0.5401716468800819,
      "learning_rate": 8.571932092685676e-05,
      "loss": 12.0439,
      "step": 20540
    },
    {
      "epoch": 1.1185395438116437,
      "grad_norm": 0.5613658674342403,
      "learning_rate": 8.571059310621577e-05,
      "loss": 11.9226,
      "step": 20541
    },
    {
      "epoch": 1.1185939978082267,
      "grad_norm": 0.5119920663394302,
      "learning_rate": 8.570186539669163e-05,
      "loss": 12.002,
      "step": 20542
    },
    {
      "epoch": 1.1186484518048097,
      "grad_norm": 0.5920825355337451,
      "learning_rate": 8.569313779835215e-05,
      "loss": 12.1506,
      "step": 20543
    },
    {
      "epoch": 1.1187029058013926,
      "grad_norm": 0.6027706362517788,
      "learning_rate": 8.568441031126519e-05,
      "loss": 11.9428,
      "step": 20544
    },
    {
      "epoch": 1.1187573597979756,
      "grad_norm": 0.5351833490287161,
      "learning_rate": 8.56756829354987e-05,
      "loss": 12.0721,
      "step": 20545
    },
    {
      "epoch": 1.1188118137945586,
      "grad_norm": 0.5572626660345444,
      "learning_rate": 8.56669556711205e-05,
      "loss": 11.9794,
      "step": 20546
    },
    {
      "epoch": 1.1188662677911416,
      "grad_norm": 0.5189188827745314,
      "learning_rate": 8.565822851819845e-05,
      "loss": 11.9398,
      "step": 20547
    },
    {
      "epoch": 1.1189207217877246,
      "grad_norm": 0.5874501284539313,
      "learning_rate": 8.564950147680043e-05,
      "loss": 12.1981,
      "step": 20548
    },
    {
      "epoch": 1.1189751757843078,
      "grad_norm": 0.5324668275406199,
      "learning_rate": 8.564077454699428e-05,
      "loss": 11.9317,
      "step": 20549
    },
    {
      "epoch": 1.1190296297808908,
      "grad_norm": 0.6009991727214898,
      "learning_rate": 8.563204772884787e-05,
      "loss": 12.2401,
      "step": 20550
    },
    {
      "epoch": 1.1190840837774738,
      "grad_norm": 0.5977241246014698,
      "learning_rate": 8.562332102242905e-05,
      "loss": 12.0362,
      "step": 20551
    },
    {
      "epoch": 1.1191385377740568,
      "grad_norm": 0.6377908932742733,
      "learning_rate": 8.561459442780578e-05,
      "loss": 12.2194,
      "step": 20552
    },
    {
      "epoch": 1.1191929917706398,
      "grad_norm": 0.6035991811431012,
      "learning_rate": 8.560586794504577e-05,
      "loss": 12.0193,
      "step": 20553
    },
    {
      "epoch": 1.1192474457672228,
      "grad_norm": 0.5325660866753731,
      "learning_rate": 8.559714157421695e-05,
      "loss": 12.063,
      "step": 20554
    },
    {
      "epoch": 1.1193018997638058,
      "grad_norm": 0.7824485461261439,
      "learning_rate": 8.558841531538715e-05,
      "loss": 12.02,
      "step": 20555
    },
    {
      "epoch": 1.1193563537603888,
      "grad_norm": 0.55243015110621,
      "learning_rate": 8.557968916862428e-05,
      "loss": 12.1466,
      "step": 20556
    },
    {
      "epoch": 1.1194108077569718,
      "grad_norm": 0.5050461586089697,
      "learning_rate": 8.557096313399615e-05,
      "loss": 11.994,
      "step": 20557
    },
    {
      "epoch": 1.1194652617535548,
      "grad_norm": 0.5906472124260127,
      "learning_rate": 8.556223721157064e-05,
      "loss": 12.1666,
      "step": 20558
    },
    {
      "epoch": 1.1195197157501378,
      "grad_norm": 0.5649664453950213,
      "learning_rate": 8.55535114014156e-05,
      "loss": 12.1033,
      "step": 20559
    },
    {
      "epoch": 1.1195741697467207,
      "grad_norm": 0.5679386653316959,
      "learning_rate": 8.554478570359887e-05,
      "loss": 12.0274,
      "step": 20560
    },
    {
      "epoch": 1.119628623743304,
      "grad_norm": 0.5260415529671344,
      "learning_rate": 8.553606011818832e-05,
      "loss": 12.1302,
      "step": 20561
    },
    {
      "epoch": 1.119683077739887,
      "grad_norm": 0.48550420063201516,
      "learning_rate": 8.55273346452518e-05,
      "loss": 11.9974,
      "step": 20562
    },
    {
      "epoch": 1.11973753173647,
      "grad_norm": 0.5596068466457854,
      "learning_rate": 8.551860928485715e-05,
      "loss": 11.95,
      "step": 20563
    },
    {
      "epoch": 1.119791985733053,
      "grad_norm": 0.5320204569567574,
      "learning_rate": 8.550988403707221e-05,
      "loss": 11.9891,
      "step": 20564
    },
    {
      "epoch": 1.119846439729636,
      "grad_norm": 0.5272038792017315,
      "learning_rate": 8.550115890196484e-05,
      "loss": 12.0433,
      "step": 20565
    },
    {
      "epoch": 1.119900893726219,
      "grad_norm": 0.5367106092079147,
      "learning_rate": 8.549243387960286e-05,
      "loss": 12.0138,
      "step": 20566
    },
    {
      "epoch": 1.119955347722802,
      "grad_norm": 0.5226903680437649,
      "learning_rate": 8.548370897005418e-05,
      "loss": 11.9826,
      "step": 20567
    },
    {
      "epoch": 1.120009801719385,
      "grad_norm": 0.6286936405364363,
      "learning_rate": 8.547498417338661e-05,
      "loss": 11.9526,
      "step": 20568
    },
    {
      "epoch": 1.120064255715968,
      "grad_norm": 0.5376852340755028,
      "learning_rate": 8.5466259489668e-05,
      "loss": 12.0677,
      "step": 20569
    },
    {
      "epoch": 1.120118709712551,
      "grad_norm": 0.5371857089677221,
      "learning_rate": 8.54575349189662e-05,
      "loss": 11.8697,
      "step": 20570
    },
    {
      "epoch": 1.120173163709134,
      "grad_norm": 0.589717476056914,
      "learning_rate": 8.544881046134905e-05,
      "loss": 11.8035,
      "step": 20571
    },
    {
      "epoch": 1.120227617705717,
      "grad_norm": 0.521908997324979,
      "learning_rate": 8.54400861168844e-05,
      "loss": 11.9903,
      "step": 20572
    },
    {
      "epoch": 1.1202820717023,
      "grad_norm": 0.5720192903917372,
      "learning_rate": 8.543136188564007e-05,
      "loss": 12.0403,
      "step": 20573
    },
    {
      "epoch": 1.120336525698883,
      "grad_norm": 0.6255945238675611,
      "learning_rate": 8.542263776768392e-05,
      "loss": 12.0523,
      "step": 20574
    },
    {
      "epoch": 1.120390979695466,
      "grad_norm": 0.569480533231531,
      "learning_rate": 8.541391376308376e-05,
      "loss": 12.0133,
      "step": 20575
    },
    {
      "epoch": 1.120445433692049,
      "grad_norm": 0.533932773695763,
      "learning_rate": 8.540518987190746e-05,
      "loss": 11.9363,
      "step": 20576
    },
    {
      "epoch": 1.120499887688632,
      "grad_norm": 0.5475647975327231,
      "learning_rate": 8.539646609422285e-05,
      "loss": 12.0669,
      "step": 20577
    },
    {
      "epoch": 1.120554341685215,
      "grad_norm": 0.5928832029061398,
      "learning_rate": 8.538774243009775e-05,
      "loss": 11.9212,
      "step": 20578
    },
    {
      "epoch": 1.120608795681798,
      "grad_norm": 0.5038069420499145,
      "learning_rate": 8.537901887960004e-05,
      "loss": 11.9198,
      "step": 20579
    },
    {
      "epoch": 1.120663249678381,
      "grad_norm": 0.5096724960488348,
      "learning_rate": 8.537029544279754e-05,
      "loss": 11.927,
      "step": 20580
    },
    {
      "epoch": 1.120717703674964,
      "grad_norm": 0.5598120008559261,
      "learning_rate": 8.536157211975806e-05,
      "loss": 12.149,
      "step": 20581
    },
    {
      "epoch": 1.120772157671547,
      "grad_norm": 0.564549478719887,
      "learning_rate": 8.535284891054947e-05,
      "loss": 12.011,
      "step": 20582
    },
    {
      "epoch": 1.12082661166813,
      "grad_norm": 0.5757046759921594,
      "learning_rate": 8.534412581523959e-05,
      "loss": 12.0404,
      "step": 20583
    },
    {
      "epoch": 1.120881065664713,
      "grad_norm": 0.5195550332959707,
      "learning_rate": 8.533540283389621e-05,
      "loss": 12.0446,
      "step": 20584
    },
    {
      "epoch": 1.1209355196612962,
      "grad_norm": 0.537669835620897,
      "learning_rate": 8.53266799665872e-05,
      "loss": 11.986,
      "step": 20585
    },
    {
      "epoch": 1.1209899736578792,
      "grad_norm": 0.5266346732329061,
      "learning_rate": 8.531795721338041e-05,
      "loss": 12.0639,
      "step": 20586
    },
    {
      "epoch": 1.1210444276544622,
      "grad_norm": 0.6063960080528453,
      "learning_rate": 8.530923457434364e-05,
      "loss": 12.0502,
      "step": 20587
    },
    {
      "epoch": 1.1210988816510452,
      "grad_norm": 0.5711788447519252,
      "learning_rate": 8.530051204954472e-05,
      "loss": 12.0723,
      "step": 20588
    },
    {
      "epoch": 1.1211533356476282,
      "grad_norm": 0.5400089161958422,
      "learning_rate": 8.529178963905147e-05,
      "loss": 12.1502,
      "step": 20589
    },
    {
      "epoch": 1.1212077896442112,
      "grad_norm": 0.5623434008351368,
      "learning_rate": 8.528306734293174e-05,
      "loss": 11.9905,
      "step": 20590
    },
    {
      "epoch": 1.1212622436407942,
      "grad_norm": 0.5020169980727249,
      "learning_rate": 8.527434516125335e-05,
      "loss": 12.04,
      "step": 20591
    },
    {
      "epoch": 1.1213166976373772,
      "grad_norm": 0.5083796763533629,
      "learning_rate": 8.526562309408417e-05,
      "loss": 12.041,
      "step": 20592
    },
    {
      "epoch": 1.1213711516339602,
      "grad_norm": 0.5386733757095622,
      "learning_rate": 8.525690114149191e-05,
      "loss": 12.0352,
      "step": 20593
    },
    {
      "epoch": 1.1214256056305432,
      "grad_norm": 0.5225178827581798,
      "learning_rate": 8.524817930354447e-05,
      "loss": 11.9339,
      "step": 20594
    },
    {
      "epoch": 1.1214800596271262,
      "grad_norm": 0.5441174805016625,
      "learning_rate": 8.523945758030966e-05,
      "loss": 11.9746,
      "step": 20595
    },
    {
      "epoch": 1.1215345136237094,
      "grad_norm": 0.5150008167747349,
      "learning_rate": 8.523073597185533e-05,
      "loss": 12.0508,
      "step": 20596
    },
    {
      "epoch": 1.1215889676202924,
      "grad_norm": 0.5307939347724793,
      "learning_rate": 8.522201447824925e-05,
      "loss": 11.5901,
      "step": 20597
    },
    {
      "epoch": 1.1216434216168754,
      "grad_norm": 0.5660549221776581,
      "learning_rate": 8.521329309955927e-05,
      "loss": 11.9772,
      "step": 20598
    },
    {
      "epoch": 1.1216978756134584,
      "grad_norm": 0.5287968192433967,
      "learning_rate": 8.520457183585321e-05,
      "loss": 12.0573,
      "step": 20599
    },
    {
      "epoch": 1.1217523296100413,
      "grad_norm": 0.6005800707822906,
      "learning_rate": 8.519585068719884e-05,
      "loss": 12.1486,
      "step": 20600
    },
    {
      "epoch": 1.1218067836066243,
      "grad_norm": 0.58718529211763,
      "learning_rate": 8.518712965366406e-05,
      "loss": 11.9406,
      "step": 20601
    },
    {
      "epoch": 1.1218612376032073,
      "grad_norm": 0.5926879722142123,
      "learning_rate": 8.517840873531669e-05,
      "loss": 12.0648,
      "step": 20602
    },
    {
      "epoch": 1.1219156915997903,
      "grad_norm": 0.555690531355676,
      "learning_rate": 8.516968793222443e-05,
      "loss": 12.0557,
      "step": 20603
    },
    {
      "epoch": 1.1219701455963733,
      "grad_norm": 0.5334045336469574,
      "learning_rate": 8.516096724445516e-05,
      "loss": 12.0405,
      "step": 20604
    },
    {
      "epoch": 1.1220245995929563,
      "grad_norm": 0.6051364508916515,
      "learning_rate": 8.515224667207671e-05,
      "loss": 11.9985,
      "step": 20605
    },
    {
      "epoch": 1.1220790535895393,
      "grad_norm": 0.5152068370069078,
      "learning_rate": 8.514352621515689e-05,
      "loss": 11.9404,
      "step": 20606
    },
    {
      "epoch": 1.1221335075861223,
      "grad_norm": 0.5700140228021777,
      "learning_rate": 8.51348058737635e-05,
      "loss": 12.0488,
      "step": 20607
    },
    {
      "epoch": 1.1221879615827055,
      "grad_norm": 0.5311689879860269,
      "learning_rate": 8.512608564796435e-05,
      "loss": 11.9787,
      "step": 20608
    },
    {
      "epoch": 1.1222424155792885,
      "grad_norm": 0.6004770577091992,
      "learning_rate": 8.511736553782725e-05,
      "loss": 11.8855,
      "step": 20609
    },
    {
      "epoch": 1.1222968695758715,
      "grad_norm": 0.572080055635242,
      "learning_rate": 8.510864554342e-05,
      "loss": 11.9507,
      "step": 20610
    },
    {
      "epoch": 1.1223513235724545,
      "grad_norm": 0.5223463898176213,
      "learning_rate": 8.509992566481042e-05,
      "loss": 11.9969,
      "step": 20611
    },
    {
      "epoch": 1.1224057775690375,
      "grad_norm": 0.5549006948857823,
      "learning_rate": 8.509120590206637e-05,
      "loss": 12.1299,
      "step": 20612
    },
    {
      "epoch": 1.1224602315656205,
      "grad_norm": 0.568669376047195,
      "learning_rate": 8.508248625525557e-05,
      "loss": 11.8601,
      "step": 20613
    },
    {
      "epoch": 1.1225146855622035,
      "grad_norm": 0.5619581181952749,
      "learning_rate": 8.507376672444585e-05,
      "loss": 12.1542,
      "step": 20614
    },
    {
      "epoch": 1.1225691395587865,
      "grad_norm": 0.5640475607492639,
      "learning_rate": 8.506504730970501e-05,
      "loss": 11.9929,
      "step": 20615
    },
    {
      "epoch": 1.1226235935553694,
      "grad_norm": 0.5495919073592045,
      "learning_rate": 8.505632801110087e-05,
      "loss": 11.9254,
      "step": 20616
    },
    {
      "epoch": 1.1226780475519524,
      "grad_norm": 0.5614813120214723,
      "learning_rate": 8.504760882870124e-05,
      "loss": 12.0447,
      "step": 20617
    },
    {
      "epoch": 1.1227325015485354,
      "grad_norm": 0.6016034729577902,
      "learning_rate": 8.503888976257392e-05,
      "loss": 12.0285,
      "step": 20618
    },
    {
      "epoch": 1.1227869555451186,
      "grad_norm": 0.5867644357431405,
      "learning_rate": 8.503017081278668e-05,
      "loss": 11.9198,
      "step": 20619
    },
    {
      "epoch": 1.1228414095417016,
      "grad_norm": 0.5127568851116143,
      "learning_rate": 8.502145197940736e-05,
      "loss": 11.9742,
      "step": 20620
    },
    {
      "epoch": 1.1228958635382846,
      "grad_norm": 0.5337297068088316,
      "learning_rate": 8.501273326250374e-05,
      "loss": 11.9856,
      "step": 20621
    },
    {
      "epoch": 1.1229503175348676,
      "grad_norm": 0.5461410367404895,
      "learning_rate": 8.500401466214364e-05,
      "loss": 12.0635,
      "step": 20622
    },
    {
      "epoch": 1.1230047715314506,
      "grad_norm": 0.5617180325957596,
      "learning_rate": 8.49952961783948e-05,
      "loss": 11.954,
      "step": 20623
    },
    {
      "epoch": 1.1230592255280336,
      "grad_norm": 0.510976442238758,
      "learning_rate": 8.498657781132509e-05,
      "loss": 11.8703,
      "step": 20624
    },
    {
      "epoch": 1.1231136795246166,
      "grad_norm": 0.5383942493352548,
      "learning_rate": 8.497785956100223e-05,
      "loss": 12.0471,
      "step": 20625
    },
    {
      "epoch": 1.1231681335211996,
      "grad_norm": 0.5160260825142511,
      "learning_rate": 8.496914142749407e-05,
      "loss": 11.9466,
      "step": 20626
    },
    {
      "epoch": 1.1232225875177826,
      "grad_norm": 0.5590712471310877,
      "learning_rate": 8.496042341086836e-05,
      "loss": 12.0509,
      "step": 20627
    },
    {
      "epoch": 1.1232770415143656,
      "grad_norm": 0.5524489743549429,
      "learning_rate": 8.495170551119296e-05,
      "loss": 11.9801,
      "step": 20628
    },
    {
      "epoch": 1.1233314955109486,
      "grad_norm": 0.5421061679085142,
      "learning_rate": 8.49429877285356e-05,
      "loss": 11.9793,
      "step": 20629
    },
    {
      "epoch": 1.1233859495075316,
      "grad_norm": 0.5565855504551699,
      "learning_rate": 8.49342700629641e-05,
      "loss": 12.1138,
      "step": 20630
    },
    {
      "epoch": 1.1234404035041148,
      "grad_norm": 0.5393647732434288,
      "learning_rate": 8.492555251454623e-05,
      "loss": 12.0277,
      "step": 20631
    },
    {
      "epoch": 1.1234948575006978,
      "grad_norm": 0.5653211907581519,
      "learning_rate": 8.491683508334983e-05,
      "loss": 12.0809,
      "step": 20632
    },
    {
      "epoch": 1.1235493114972808,
      "grad_norm": 0.5566333487947905,
      "learning_rate": 8.490811776944263e-05,
      "loss": 11.9899,
      "step": 20633
    },
    {
      "epoch": 1.1236037654938638,
      "grad_norm": 0.5489336043672067,
      "learning_rate": 8.489940057289243e-05,
      "loss": 12.0761,
      "step": 20634
    },
    {
      "epoch": 1.1236582194904468,
      "grad_norm": 0.5431071514195215,
      "learning_rate": 8.489068349376702e-05,
      "loss": 12.0434,
      "step": 20635
    },
    {
      "epoch": 1.1237126734870297,
      "grad_norm": 0.537853868515468,
      "learning_rate": 8.48819665321342e-05,
      "loss": 12.0849,
      "step": 20636
    },
    {
      "epoch": 1.1237671274836127,
      "grad_norm": 0.5511346778647371,
      "learning_rate": 8.487324968806173e-05,
      "loss": 11.9328,
      "step": 20637
    },
    {
      "epoch": 1.1238215814801957,
      "grad_norm": 0.5667308329210314,
      "learning_rate": 8.486453296161739e-05,
      "loss": 12.0361,
      "step": 20638
    },
    {
      "epoch": 1.1238760354767787,
      "grad_norm": 0.5471565466242325,
      "learning_rate": 8.485581635286901e-05,
      "loss": 12.0199,
      "step": 20639
    },
    {
      "epoch": 1.1239304894733617,
      "grad_norm": 0.6623247915833094,
      "learning_rate": 8.484709986188433e-05,
      "loss": 11.9894,
      "step": 20640
    },
    {
      "epoch": 1.1239849434699447,
      "grad_norm": 0.5511119629493311,
      "learning_rate": 8.483838348873116e-05,
      "loss": 12.0032,
      "step": 20641
    },
    {
      "epoch": 1.124039397466528,
      "grad_norm": 0.5424432658011794,
      "learning_rate": 8.482966723347726e-05,
      "loss": 12.0568,
      "step": 20642
    },
    {
      "epoch": 1.124093851463111,
      "grad_norm": 0.5191386336139402,
      "learning_rate": 8.48209510961904e-05,
      "loss": 11.7916,
      "step": 20643
    },
    {
      "epoch": 1.124148305459694,
      "grad_norm": 0.6078510380658935,
      "learning_rate": 8.481223507693838e-05,
      "loss": 12.0313,
      "step": 20644
    },
    {
      "epoch": 1.124202759456277,
      "grad_norm": 0.5488151658308369,
      "learning_rate": 8.480351917578896e-05,
      "loss": 12.1066,
      "step": 20645
    },
    {
      "epoch": 1.12425721345286,
      "grad_norm": 0.5545155623942513,
      "learning_rate": 8.479480339280992e-05,
      "loss": 12.0689,
      "step": 20646
    },
    {
      "epoch": 1.1243116674494429,
      "grad_norm": 0.5300148673499622,
      "learning_rate": 8.478608772806904e-05,
      "loss": 11.94,
      "step": 20647
    },
    {
      "epoch": 1.1243661214460259,
      "grad_norm": 0.6296302325770019,
      "learning_rate": 8.47773721816341e-05,
      "loss": 12.1124,
      "step": 20648
    },
    {
      "epoch": 1.1244205754426089,
      "grad_norm": 0.5481045776550506,
      "learning_rate": 8.476865675357284e-05,
      "loss": 11.9786,
      "step": 20649
    },
    {
      "epoch": 1.1244750294391919,
      "grad_norm": 0.5632361699445835,
      "learning_rate": 8.475994144395307e-05,
      "loss": 11.9742,
      "step": 20650
    },
    {
      "epoch": 1.1245294834357749,
      "grad_norm": 0.6013238189165129,
      "learning_rate": 8.475122625284257e-05,
      "loss": 12.0224,
      "step": 20651
    },
    {
      "epoch": 1.1245839374323578,
      "grad_norm": 0.6487355626787903,
      "learning_rate": 8.474251118030912e-05,
      "loss": 11.9843,
      "step": 20652
    },
    {
      "epoch": 1.1246383914289408,
      "grad_norm": 0.5609178175685393,
      "learning_rate": 8.47337962264204e-05,
      "loss": 11.9898,
      "step": 20653
    },
    {
      "epoch": 1.1246928454255238,
      "grad_norm": 0.5067012783158299,
      "learning_rate": 8.472508139124426e-05,
      "loss": 12.0245,
      "step": 20654
    },
    {
      "epoch": 1.124747299422107,
      "grad_norm": 0.5781508273221143,
      "learning_rate": 8.471636667484846e-05,
      "loss": 12.078,
      "step": 20655
    },
    {
      "epoch": 1.12480175341869,
      "grad_norm": 0.5277975985823132,
      "learning_rate": 8.470765207730075e-05,
      "loss": 12.0828,
      "step": 20656
    },
    {
      "epoch": 1.124856207415273,
      "grad_norm": 0.5802282833173706,
      "learning_rate": 8.469893759866892e-05,
      "loss": 11.9441,
      "step": 20657
    },
    {
      "epoch": 1.124910661411856,
      "grad_norm": 0.5611172893829233,
      "learning_rate": 8.46902232390207e-05,
      "loss": 12.1023,
      "step": 20658
    },
    {
      "epoch": 1.124965115408439,
      "grad_norm": 0.5192570995241735,
      "learning_rate": 8.468150899842387e-05,
      "loss": 12.0198,
      "step": 20659
    },
    {
      "epoch": 1.125019569405022,
      "grad_norm": 0.6822053457142676,
      "learning_rate": 8.467279487694617e-05,
      "loss": 12.0171,
      "step": 20660
    },
    {
      "epoch": 1.125074023401605,
      "grad_norm": 0.5516464625410687,
      "learning_rate": 8.46640808746554e-05,
      "loss": 12.0902,
      "step": 20661
    },
    {
      "epoch": 1.125128477398188,
      "grad_norm": 0.5654967257334075,
      "learning_rate": 8.465536699161934e-05,
      "loss": 12.095,
      "step": 20662
    },
    {
      "epoch": 1.125182931394771,
      "grad_norm": 0.5408078521278947,
      "learning_rate": 8.464665322790574e-05,
      "loss": 12.0886,
      "step": 20663
    },
    {
      "epoch": 1.125237385391354,
      "grad_norm": 0.595730130403624,
      "learning_rate": 8.463793958358228e-05,
      "loss": 12.0437,
      "step": 20664
    },
    {
      "epoch": 1.1252918393879372,
      "grad_norm": 0.6521803950094307,
      "learning_rate": 8.46292260587168e-05,
      "loss": 12.06,
      "step": 20665
    },
    {
      "epoch": 1.1253462933845202,
      "grad_norm": 0.4951956217474917,
      "learning_rate": 8.462051265337702e-05,
      "loss": 11.9553,
      "step": 20666
    },
    {
      "epoch": 1.1254007473811032,
      "grad_norm": 0.5791223500643684,
      "learning_rate": 8.461179936763074e-05,
      "loss": 11.9232,
      "step": 20667
    },
    {
      "epoch": 1.1254552013776862,
      "grad_norm": 0.5697549963250519,
      "learning_rate": 8.460308620154566e-05,
      "loss": 11.9406,
      "step": 20668
    },
    {
      "epoch": 1.1255096553742692,
      "grad_norm": 0.6110789914736281,
      "learning_rate": 8.459437315518959e-05,
      "loss": 12.1223,
      "step": 20669
    },
    {
      "epoch": 1.1255641093708522,
      "grad_norm": 0.5377100319564458,
      "learning_rate": 8.458566022863026e-05,
      "loss": 12.0876,
      "step": 20670
    },
    {
      "epoch": 1.1256185633674352,
      "grad_norm": 0.5596473554488341,
      "learning_rate": 8.457694742193538e-05,
      "loss": 12.0281,
      "step": 20671
    },
    {
      "epoch": 1.1256730173640181,
      "grad_norm": 0.6157363408101525,
      "learning_rate": 8.456823473517277e-05,
      "loss": 12.0079,
      "step": 20672
    },
    {
      "epoch": 1.1257274713606011,
      "grad_norm": 0.5958012714277305,
      "learning_rate": 8.45595221684102e-05,
      "loss": 12.0952,
      "step": 20673
    },
    {
      "epoch": 1.1257819253571841,
      "grad_norm": 0.5909040796004863,
      "learning_rate": 8.455080972171535e-05,
      "loss": 11.9936,
      "step": 20674
    },
    {
      "epoch": 1.1258363793537671,
      "grad_norm": 0.5259277364480356,
      "learning_rate": 8.454209739515594e-05,
      "loss": 11.9397,
      "step": 20675
    },
    {
      "epoch": 1.1258908333503501,
      "grad_norm": 0.6289245786484194,
      "learning_rate": 8.453338518879981e-05,
      "loss": 12.0402,
      "step": 20676
    },
    {
      "epoch": 1.125945287346933,
      "grad_norm": 0.5463792922285373,
      "learning_rate": 8.452467310271467e-05,
      "loss": 11.9932,
      "step": 20677
    },
    {
      "epoch": 1.1259997413435163,
      "grad_norm": 0.5498803039109769,
      "learning_rate": 8.451596113696827e-05,
      "loss": 12.1414,
      "step": 20678
    },
    {
      "epoch": 1.1260541953400993,
      "grad_norm": 0.5195443857787331,
      "learning_rate": 8.450724929162834e-05,
      "loss": 11.9772,
      "step": 20679
    },
    {
      "epoch": 1.1261086493366823,
      "grad_norm": 0.6088749495130134,
      "learning_rate": 8.449853756676265e-05,
      "loss": 12.1435,
      "step": 20680
    },
    {
      "epoch": 1.1261631033332653,
      "grad_norm": 0.4979402048421249,
      "learning_rate": 8.448982596243893e-05,
      "loss": 11.9464,
      "step": 20681
    },
    {
      "epoch": 1.1262175573298483,
      "grad_norm": 0.5471764648632118,
      "learning_rate": 8.448111447872493e-05,
      "loss": 11.9167,
      "step": 20682
    },
    {
      "epoch": 1.1262720113264313,
      "grad_norm": 0.5347362131973291,
      "learning_rate": 8.44724031156884e-05,
      "loss": 11.9314,
      "step": 20683
    },
    {
      "epoch": 1.1263264653230143,
      "grad_norm": 0.5300495252964539,
      "learning_rate": 8.446369187339704e-05,
      "loss": 11.9897,
      "step": 20684
    },
    {
      "epoch": 1.1263809193195973,
      "grad_norm": 0.4901735506020483,
      "learning_rate": 8.445498075191863e-05,
      "loss": 12.0088,
      "step": 20685
    },
    {
      "epoch": 1.1264353733161803,
      "grad_norm": 0.5439219481947628,
      "learning_rate": 8.44462697513209e-05,
      "loss": 12.0971,
      "step": 20686
    },
    {
      "epoch": 1.1264898273127633,
      "grad_norm": 0.5318345497439173,
      "learning_rate": 8.443755887167155e-05,
      "loss": 12.0603,
      "step": 20687
    },
    {
      "epoch": 1.1265442813093463,
      "grad_norm": 0.5827985748435938,
      "learning_rate": 8.442884811303837e-05,
      "loss": 11.9276,
      "step": 20688
    },
    {
      "epoch": 1.1265987353059295,
      "grad_norm": 0.5364340700027024,
      "learning_rate": 8.442013747548909e-05,
      "loss": 12.1521,
      "step": 20689
    },
    {
      "epoch": 1.1266531893025125,
      "grad_norm": 0.5170281003949356,
      "learning_rate": 8.441142695909143e-05,
      "loss": 11.9498,
      "step": 20690
    },
    {
      "epoch": 1.1267076432990955,
      "grad_norm": 0.5055257982311351,
      "learning_rate": 8.440271656391313e-05,
      "loss": 12.0405,
      "step": 20691
    },
    {
      "epoch": 1.1267620972956784,
      "grad_norm": 0.5428997872547985,
      "learning_rate": 8.439400629002192e-05,
      "loss": 12.1165,
      "step": 20692
    },
    {
      "epoch": 1.1268165512922614,
      "grad_norm": 0.5820501425587059,
      "learning_rate": 8.438529613748556e-05,
      "loss": 11.9797,
      "step": 20693
    },
    {
      "epoch": 1.1268710052888444,
      "grad_norm": 0.5406008973033072,
      "learning_rate": 8.437658610637172e-05,
      "loss": 11.9376,
      "step": 20694
    },
    {
      "epoch": 1.1269254592854274,
      "grad_norm": 0.5758219611985749,
      "learning_rate": 8.436787619674819e-05,
      "loss": 12.0318,
      "step": 20695
    },
    {
      "epoch": 1.1269799132820104,
      "grad_norm": 0.4994890911792227,
      "learning_rate": 8.435916640868266e-05,
      "loss": 11.9158,
      "step": 20696
    },
    {
      "epoch": 1.1270343672785934,
      "grad_norm": 0.575546458026418,
      "learning_rate": 8.435045674224287e-05,
      "loss": 11.7971,
      "step": 20697
    },
    {
      "epoch": 1.1270888212751764,
      "grad_norm": 0.5398853059585484,
      "learning_rate": 8.434174719749654e-05,
      "loss": 11.9874,
      "step": 20698
    },
    {
      "epoch": 1.1271432752717594,
      "grad_norm": 0.6294282540302789,
      "learning_rate": 8.433303777451143e-05,
      "loss": 12.1287,
      "step": 20699
    },
    {
      "epoch": 1.1271977292683424,
      "grad_norm": 0.7472791478523968,
      "learning_rate": 8.432432847335525e-05,
      "loss": 11.9709,
      "step": 20700
    },
    {
      "epoch": 1.1272521832649254,
      "grad_norm": 0.5856051668569312,
      "learning_rate": 8.431561929409571e-05,
      "loss": 11.9004,
      "step": 20701
    },
    {
      "epoch": 1.1273066372615086,
      "grad_norm": 0.5889737421383973,
      "learning_rate": 8.430691023680055e-05,
      "loss": 12.041,
      "step": 20702
    },
    {
      "epoch": 1.1273610912580916,
      "grad_norm": 0.5480485563764228,
      "learning_rate": 8.429820130153752e-05,
      "loss": 12.0473,
      "step": 20703
    },
    {
      "epoch": 1.1274155452546746,
      "grad_norm": 0.5528813501864324,
      "learning_rate": 8.428949248837428e-05,
      "loss": 12.1747,
      "step": 20704
    },
    {
      "epoch": 1.1274699992512576,
      "grad_norm": 0.5446607362507545,
      "learning_rate": 8.428078379737858e-05,
      "loss": 11.9402,
      "step": 20705
    },
    {
      "epoch": 1.1275244532478406,
      "grad_norm": 0.584286209918372,
      "learning_rate": 8.427207522861813e-05,
      "loss": 12.1177,
      "step": 20706
    },
    {
      "epoch": 1.1275789072444236,
      "grad_norm": 0.5559404672854191,
      "learning_rate": 8.426336678216066e-05,
      "loss": 12.0242,
      "step": 20707
    },
    {
      "epoch": 1.1276333612410065,
      "grad_norm": 0.6205962003625162,
      "learning_rate": 8.425465845807392e-05,
      "loss": 12.0569,
      "step": 20708
    },
    {
      "epoch": 1.1276878152375895,
      "grad_norm": 0.5098974570827257,
      "learning_rate": 8.424595025642555e-05,
      "loss": 12.053,
      "step": 20709
    },
    {
      "epoch": 1.1277422692341725,
      "grad_norm": 0.5212660381706562,
      "learning_rate": 8.423724217728334e-05,
      "loss": 12.102,
      "step": 20710
    },
    {
      "epoch": 1.1277967232307555,
      "grad_norm": 0.5943423464276083,
      "learning_rate": 8.422853422071497e-05,
      "loss": 12.042,
      "step": 20711
    },
    {
      "epoch": 1.1278511772273387,
      "grad_norm": 0.559493213992451,
      "learning_rate": 8.421982638678818e-05,
      "loss": 11.9594,
      "step": 20712
    },
    {
      "epoch": 1.1279056312239217,
      "grad_norm": 0.5556647353810512,
      "learning_rate": 8.421111867557068e-05,
      "loss": 12.1456,
      "step": 20713
    },
    {
      "epoch": 1.1279600852205047,
      "grad_norm": 0.6328214803425803,
      "learning_rate": 8.420241108713013e-05,
      "loss": 12.1421,
      "step": 20714
    },
    {
      "epoch": 1.1280145392170877,
      "grad_norm": 0.5926666590503891,
      "learning_rate": 8.419370362153431e-05,
      "loss": 12.0015,
      "step": 20715
    },
    {
      "epoch": 1.1280689932136707,
      "grad_norm": 0.5315547798326242,
      "learning_rate": 8.418499627885089e-05,
      "loss": 11.9904,
      "step": 20716
    },
    {
      "epoch": 1.1281234472102537,
      "grad_norm": 0.6096039674322927,
      "learning_rate": 8.417628905914758e-05,
      "loss": 12.0197,
      "step": 20717
    },
    {
      "epoch": 1.1281779012068367,
      "grad_norm": 0.5584602220091498,
      "learning_rate": 8.416758196249211e-05,
      "loss": 12.0328,
      "step": 20718
    },
    {
      "epoch": 1.1282323552034197,
      "grad_norm": 0.5960657696185826,
      "learning_rate": 8.415887498895219e-05,
      "loss": 11.9276,
      "step": 20719
    },
    {
      "epoch": 1.1282868092000027,
      "grad_norm": 0.6119079673118717,
      "learning_rate": 8.415016813859548e-05,
      "loss": 11.9412,
      "step": 20720
    },
    {
      "epoch": 1.1283412631965857,
      "grad_norm": 0.5498559512074462,
      "learning_rate": 8.414146141148972e-05,
      "loss": 11.9704,
      "step": 20721
    },
    {
      "epoch": 1.1283957171931687,
      "grad_norm": 0.542946266885336,
      "learning_rate": 8.413275480770266e-05,
      "loss": 12.0014,
      "step": 20722
    },
    {
      "epoch": 1.1284501711897517,
      "grad_norm": 0.5427884476921939,
      "learning_rate": 8.412404832730197e-05,
      "loss": 12.0812,
      "step": 20723
    },
    {
      "epoch": 1.1285046251863347,
      "grad_norm": 0.5462960786118958,
      "learning_rate": 8.411534197035529e-05,
      "loss": 11.9141,
      "step": 20724
    },
    {
      "epoch": 1.1285590791829179,
      "grad_norm": 0.565910848925007,
      "learning_rate": 8.41066357369304e-05,
      "loss": 11.9949,
      "step": 20725
    },
    {
      "epoch": 1.1286135331795009,
      "grad_norm": 0.5239689661926308,
      "learning_rate": 8.409792962709497e-05,
      "loss": 11.7701,
      "step": 20726
    },
    {
      "epoch": 1.1286679871760839,
      "grad_norm": 0.5462626931379434,
      "learning_rate": 8.40892236409167e-05,
      "loss": 12.043,
      "step": 20727
    },
    {
      "epoch": 1.1287224411726668,
      "grad_norm": 0.5312671667088449,
      "learning_rate": 8.40805177784633e-05,
      "loss": 12.0121,
      "step": 20728
    },
    {
      "epoch": 1.1287768951692498,
      "grad_norm": 0.6380397818351647,
      "learning_rate": 8.407181203980247e-05,
      "loss": 12.1194,
      "step": 20729
    },
    {
      "epoch": 1.1288313491658328,
      "grad_norm": 0.578529542017685,
      "learning_rate": 8.406310642500189e-05,
      "loss": 12.1208,
      "step": 20730
    },
    {
      "epoch": 1.1288858031624158,
      "grad_norm": 0.5497148691261249,
      "learning_rate": 8.405440093412927e-05,
      "loss": 11.9895,
      "step": 20731
    },
    {
      "epoch": 1.1289402571589988,
      "grad_norm": 0.5100078242558772,
      "learning_rate": 8.404569556725229e-05,
      "loss": 12.0079,
      "step": 20732
    },
    {
      "epoch": 1.1289947111555818,
      "grad_norm": 0.5535915308511297,
      "learning_rate": 8.403699032443871e-05,
      "loss": 12.0246,
      "step": 20733
    },
    {
      "epoch": 1.1290491651521648,
      "grad_norm": 0.6122926920072559,
      "learning_rate": 8.402828520575614e-05,
      "loss": 11.9979,
      "step": 20734
    },
    {
      "epoch": 1.129103619148748,
      "grad_norm": 0.5094572839458141,
      "learning_rate": 8.40195802112723e-05,
      "loss": 11.7763,
      "step": 20735
    },
    {
      "epoch": 1.129158073145331,
      "grad_norm": 0.5970999949646029,
      "learning_rate": 8.401087534105485e-05,
      "loss": 12.0442,
      "step": 20736
    },
    {
      "epoch": 1.129212527141914,
      "grad_norm": 0.5621315848332539,
      "learning_rate": 8.400217059517155e-05,
      "loss": 12.0884,
      "step": 20737
    },
    {
      "epoch": 1.129266981138497,
      "grad_norm": 0.6562930582146015,
      "learning_rate": 8.399346597369005e-05,
      "loss": 12.0669,
      "step": 20738
    },
    {
      "epoch": 1.12932143513508,
      "grad_norm": 0.574893916287573,
      "learning_rate": 8.398476147667803e-05,
      "loss": 12.0122,
      "step": 20739
    },
    {
      "epoch": 1.129375889131663,
      "grad_norm": 0.6457764055167019,
      "learning_rate": 8.39760571042032e-05,
      "loss": 11.9435,
      "step": 20740
    },
    {
      "epoch": 1.129430343128246,
      "grad_norm": 0.5280157277237386,
      "learning_rate": 8.396735285633324e-05,
      "loss": 12.0004,
      "step": 20741
    },
    {
      "epoch": 1.129484797124829,
      "grad_norm": 0.5123033419856629,
      "learning_rate": 8.395864873313584e-05,
      "loss": 11.9824,
      "step": 20742
    },
    {
      "epoch": 1.129539251121412,
      "grad_norm": 0.4986953868214987,
      "learning_rate": 8.394994473467869e-05,
      "loss": 11.8879,
      "step": 20743
    },
    {
      "epoch": 1.129593705117995,
      "grad_norm": 0.5713935244026771,
      "learning_rate": 8.394124086102943e-05,
      "loss": 11.8317,
      "step": 20744
    },
    {
      "epoch": 1.129648159114578,
      "grad_norm": 0.5064476313565307,
      "learning_rate": 8.393253711225579e-05,
      "loss": 11.8824,
      "step": 20745
    },
    {
      "epoch": 1.129702613111161,
      "grad_norm": 0.5383477103141034,
      "learning_rate": 8.392383348842543e-05,
      "loss": 12.062,
      "step": 20746
    },
    {
      "epoch": 1.129757067107744,
      "grad_norm": 0.6857506568266057,
      "learning_rate": 8.3915129989606e-05,
      "loss": 12.0129,
      "step": 20747
    },
    {
      "epoch": 1.1298115211043271,
      "grad_norm": 0.6066521891673776,
      "learning_rate": 8.390642661586528e-05,
      "loss": 12.1521,
      "step": 20748
    },
    {
      "epoch": 1.1298659751009101,
      "grad_norm": 0.604582705656562,
      "learning_rate": 8.389772336727084e-05,
      "loss": 12.0386,
      "step": 20749
    },
    {
      "epoch": 1.1299204290974931,
      "grad_norm": 0.545887271751831,
      "learning_rate": 8.388902024389042e-05,
      "loss": 12.0592,
      "step": 20750
    },
    {
      "epoch": 1.1299748830940761,
      "grad_norm": 0.6514439487333721,
      "learning_rate": 8.388031724579169e-05,
      "loss": 12.2167,
      "step": 20751
    },
    {
      "epoch": 1.1300293370906591,
      "grad_norm": 0.544262019270203,
      "learning_rate": 8.387161437304232e-05,
      "loss": 11.9321,
      "step": 20752
    },
    {
      "epoch": 1.130083791087242,
      "grad_norm": 0.5831675179655477,
      "learning_rate": 8.386291162570998e-05,
      "loss": 11.9923,
      "step": 20753
    },
    {
      "epoch": 1.130138245083825,
      "grad_norm": 0.5468630280688331,
      "learning_rate": 8.385420900386234e-05,
      "loss": 11.8361,
      "step": 20754
    },
    {
      "epoch": 1.130192699080408,
      "grad_norm": 0.5239437358619802,
      "learning_rate": 8.384550650756707e-05,
      "loss": 11.8683,
      "step": 20755
    },
    {
      "epoch": 1.130247153076991,
      "grad_norm": 0.5770978379625384,
      "learning_rate": 8.383680413689186e-05,
      "loss": 11.9685,
      "step": 20756
    },
    {
      "epoch": 1.130301607073574,
      "grad_norm": 0.5528615214271068,
      "learning_rate": 8.382810189190438e-05,
      "loss": 12.027,
      "step": 20757
    },
    {
      "epoch": 1.1303560610701573,
      "grad_norm": 0.5572352510964343,
      "learning_rate": 8.381939977267225e-05,
      "loss": 11.9764,
      "step": 20758
    },
    {
      "epoch": 1.1304105150667403,
      "grad_norm": 0.5438921781892165,
      "learning_rate": 8.38106977792632e-05,
      "loss": 11.9362,
      "step": 20759
    },
    {
      "epoch": 1.1304649690633233,
      "grad_norm": 0.6162761724338172,
      "learning_rate": 8.38019959117449e-05,
      "loss": 11.9992,
      "step": 20760
    },
    {
      "epoch": 1.1305194230599063,
      "grad_norm": 0.5644696651643153,
      "learning_rate": 8.379329417018497e-05,
      "loss": 11.9646,
      "step": 20761
    },
    {
      "epoch": 1.1305738770564893,
      "grad_norm": 0.661556556635499,
      "learning_rate": 8.378459255465112e-05,
      "loss": 12.0489,
      "step": 20762
    },
    {
      "epoch": 1.1306283310530723,
      "grad_norm": 0.5901163540090555,
      "learning_rate": 8.377589106521101e-05,
      "loss": 12.0038,
      "step": 20763
    },
    {
      "epoch": 1.1306827850496552,
      "grad_norm": 0.5887311343504869,
      "learning_rate": 8.376718970193229e-05,
      "loss": 12.1154,
      "step": 20764
    },
    {
      "epoch": 1.1307372390462382,
      "grad_norm": 0.6280187554043807,
      "learning_rate": 8.375848846488262e-05,
      "loss": 12.0463,
      "step": 20765
    },
    {
      "epoch": 1.1307916930428212,
      "grad_norm": 0.5887668368648779,
      "learning_rate": 8.374978735412965e-05,
      "loss": 12.0367,
      "step": 20766
    },
    {
      "epoch": 1.1308461470394042,
      "grad_norm": 0.5647871665134128,
      "learning_rate": 8.374108636974107e-05,
      "loss": 11.8751,
      "step": 20767
    },
    {
      "epoch": 1.1309006010359872,
      "grad_norm": 0.5375661928587627,
      "learning_rate": 8.373238551178453e-05,
      "loss": 12.0289,
      "step": 20768
    },
    {
      "epoch": 1.1309550550325702,
      "grad_norm": 0.5365046984815546,
      "learning_rate": 8.372368478032765e-05,
      "loss": 12.0516,
      "step": 20769
    },
    {
      "epoch": 1.1310095090291532,
      "grad_norm": 0.57074124799673,
      "learning_rate": 8.371498417543817e-05,
      "loss": 12.1869,
      "step": 20770
    },
    {
      "epoch": 1.1310639630257362,
      "grad_norm": 0.5485043911729385,
      "learning_rate": 8.37062836971837e-05,
      "loss": 12.1736,
      "step": 20771
    },
    {
      "epoch": 1.1311184170223194,
      "grad_norm": 0.5538806730895031,
      "learning_rate": 8.369758334563189e-05,
      "loss": 12.04,
      "step": 20772
    },
    {
      "epoch": 1.1311728710189024,
      "grad_norm": 0.5623222575053048,
      "learning_rate": 8.368888312085043e-05,
      "loss": 12.0119,
      "step": 20773
    },
    {
      "epoch": 1.1312273250154854,
      "grad_norm": 0.5860184149076423,
      "learning_rate": 8.368018302290694e-05,
      "loss": 12.0345,
      "step": 20774
    },
    {
      "epoch": 1.1312817790120684,
      "grad_norm": 0.5368386044729148,
      "learning_rate": 8.367148305186907e-05,
      "loss": 12.0709,
      "step": 20775
    },
    {
      "epoch": 1.1313362330086514,
      "grad_norm": 0.6503054542888462,
      "learning_rate": 8.366278320780449e-05,
      "loss": 12.1014,
      "step": 20776
    },
    {
      "epoch": 1.1313906870052344,
      "grad_norm": 0.5586135419144223,
      "learning_rate": 8.365408349078085e-05,
      "loss": 12.0123,
      "step": 20777
    },
    {
      "epoch": 1.1314451410018174,
      "grad_norm": 0.5508012796026294,
      "learning_rate": 8.36453839008658e-05,
      "loss": 11.9524,
      "step": 20778
    },
    {
      "epoch": 1.1314995949984004,
      "grad_norm": 0.5687890564258543,
      "learning_rate": 8.3636684438127e-05,
      "loss": 12.0393,
      "step": 20779
    },
    {
      "epoch": 1.1315540489949834,
      "grad_norm": 0.5505202177610155,
      "learning_rate": 8.362798510263205e-05,
      "loss": 11.9455,
      "step": 20780
    },
    {
      "epoch": 1.1316085029915663,
      "grad_norm": 0.5414750858393832,
      "learning_rate": 8.361928589444865e-05,
      "loss": 11.9601,
      "step": 20781
    },
    {
      "epoch": 1.1316629569881496,
      "grad_norm": 0.5258501960877228,
      "learning_rate": 8.361058681364442e-05,
      "loss": 11.9788,
      "step": 20782
    },
    {
      "epoch": 1.1317174109847326,
      "grad_norm": 0.5440550629269573,
      "learning_rate": 8.360188786028707e-05,
      "loss": 11.9825,
      "step": 20783
    },
    {
      "epoch": 1.1317718649813155,
      "grad_norm": 0.5351544035011582,
      "learning_rate": 8.359318903444416e-05,
      "loss": 12.0982,
      "step": 20784
    },
    {
      "epoch": 1.1318263189778985,
      "grad_norm": 0.5323293259840147,
      "learning_rate": 8.358449033618334e-05,
      "loss": 11.9676,
      "step": 20785
    },
    {
      "epoch": 1.1318807729744815,
      "grad_norm": 0.5529529749133446,
      "learning_rate": 8.357579176557228e-05,
      "loss": 11.8974,
      "step": 20786
    },
    {
      "epoch": 1.1319352269710645,
      "grad_norm": 0.590287551039442,
      "learning_rate": 8.356709332267863e-05,
      "loss": 11.9511,
      "step": 20787
    },
    {
      "epoch": 1.1319896809676475,
      "grad_norm": 0.5125296807104094,
      "learning_rate": 8.355839500757002e-05,
      "loss": 12.0024,
      "step": 20788
    },
    {
      "epoch": 1.1320441349642305,
      "grad_norm": 0.5804900466222254,
      "learning_rate": 8.354969682031407e-05,
      "loss": 12.1384,
      "step": 20789
    },
    {
      "epoch": 1.1320985889608135,
      "grad_norm": 0.5368063847193392,
      "learning_rate": 8.354099876097845e-05,
      "loss": 11.9719,
      "step": 20790
    },
    {
      "epoch": 1.1321530429573965,
      "grad_norm": 0.5779207309703169,
      "learning_rate": 8.353230082963078e-05,
      "loss": 11.9777,
      "step": 20791
    },
    {
      "epoch": 1.1322074969539795,
      "grad_norm": 0.5739429398866446,
      "learning_rate": 8.352360302633868e-05,
      "loss": 12.0905,
      "step": 20792
    },
    {
      "epoch": 1.1322619509505625,
      "grad_norm": 0.5300896274934617,
      "learning_rate": 8.351490535116987e-05,
      "loss": 12.0226,
      "step": 20793
    },
    {
      "epoch": 1.1323164049471455,
      "grad_norm": 0.571155766038941,
      "learning_rate": 8.35062078041919e-05,
      "loss": 12.0976,
      "step": 20794
    },
    {
      "epoch": 1.1323708589437287,
      "grad_norm": 0.564676725194143,
      "learning_rate": 8.349751038547239e-05,
      "loss": 11.9845,
      "step": 20795
    },
    {
      "epoch": 1.1324253129403117,
      "grad_norm": 0.5337945908674252,
      "learning_rate": 8.3488813095079e-05,
      "loss": 11.9669,
      "step": 20796
    },
    {
      "epoch": 1.1324797669368947,
      "grad_norm": 0.5344172893573053,
      "learning_rate": 8.348011593307939e-05,
      "loss": 11.9563,
      "step": 20797
    },
    {
      "epoch": 1.1325342209334777,
      "grad_norm": 0.5806123347477514,
      "learning_rate": 8.347141889954117e-05,
      "loss": 12.061,
      "step": 20798
    },
    {
      "epoch": 1.1325886749300607,
      "grad_norm": 0.5690743516430177,
      "learning_rate": 8.346272199453196e-05,
      "loss": 12.024,
      "step": 20799
    },
    {
      "epoch": 1.1326431289266436,
      "grad_norm": 0.5225427525492036,
      "learning_rate": 8.34540252181194e-05,
      "loss": 11.9932,
      "step": 20800
    },
    {
      "epoch": 1.1326975829232266,
      "grad_norm": 0.5668045251543957,
      "learning_rate": 8.344532857037113e-05,
      "loss": 11.9542,
      "step": 20801
    },
    {
      "epoch": 1.1327520369198096,
      "grad_norm": 0.591275603933892,
      "learning_rate": 8.343663205135474e-05,
      "loss": 12.0476,
      "step": 20802
    },
    {
      "epoch": 1.1328064909163926,
      "grad_norm": 0.5238304872807422,
      "learning_rate": 8.342793566113787e-05,
      "loss": 12.0662,
      "step": 20803
    },
    {
      "epoch": 1.1328609449129756,
      "grad_norm": 0.5531231281251932,
      "learning_rate": 8.341923939978821e-05,
      "loss": 11.9729,
      "step": 20804
    },
    {
      "epoch": 1.1329153989095588,
      "grad_norm": 0.5525939622220033,
      "learning_rate": 8.341054326737327e-05,
      "loss": 11.9551,
      "step": 20805
    },
    {
      "epoch": 1.1329698529061418,
      "grad_norm": 0.5020694928463673,
      "learning_rate": 8.340184726396076e-05,
      "loss": 11.9396,
      "step": 20806
    },
    {
      "epoch": 1.1330243069027248,
      "grad_norm": 0.4834336628662161,
      "learning_rate": 8.339315138961821e-05,
      "loss": 11.9399,
      "step": 20807
    },
    {
      "epoch": 1.1330787608993078,
      "grad_norm": 0.5434950420977097,
      "learning_rate": 8.338445564441335e-05,
      "loss": 12.0291,
      "step": 20808
    },
    {
      "epoch": 1.1331332148958908,
      "grad_norm": 0.5607777847547193,
      "learning_rate": 8.337576002841375e-05,
      "loss": 12.1495,
      "step": 20809
    },
    {
      "epoch": 1.1331876688924738,
      "grad_norm": 0.5272975127627831,
      "learning_rate": 8.336706454168701e-05,
      "loss": 12.1168,
      "step": 20810
    },
    {
      "epoch": 1.1332421228890568,
      "grad_norm": 0.5269370953237635,
      "learning_rate": 8.335836918430075e-05,
      "loss": 12.1105,
      "step": 20811
    },
    {
      "epoch": 1.1332965768856398,
      "grad_norm": 0.5280620247681213,
      "learning_rate": 8.334967395632264e-05,
      "loss": 12.0228,
      "step": 20812
    },
    {
      "epoch": 1.1333510308822228,
      "grad_norm": 0.5604435021674699,
      "learning_rate": 8.334097885782024e-05,
      "loss": 12.0042,
      "step": 20813
    },
    {
      "epoch": 1.1334054848788058,
      "grad_norm": 0.4949261064367269,
      "learning_rate": 8.333228388886121e-05,
      "loss": 11.9785,
      "step": 20814
    },
    {
      "epoch": 1.1334599388753888,
      "grad_norm": 0.5435580938905246,
      "learning_rate": 8.33235890495131e-05,
      "loss": 12.0545,
      "step": 20815
    },
    {
      "epoch": 1.1335143928719718,
      "grad_norm": 0.5391843622721241,
      "learning_rate": 8.331489433984357e-05,
      "loss": 12.1,
      "step": 20816
    },
    {
      "epoch": 1.1335688468685547,
      "grad_norm": 0.5577771887455933,
      "learning_rate": 8.330619975992021e-05,
      "loss": 12.0138,
      "step": 20817
    },
    {
      "epoch": 1.133623300865138,
      "grad_norm": 0.5249232630778278,
      "learning_rate": 8.329750530981064e-05,
      "loss": 11.9539,
      "step": 20818
    },
    {
      "epoch": 1.133677754861721,
      "grad_norm": 0.5104218897688741,
      "learning_rate": 8.328881098958246e-05,
      "loss": 12.0681,
      "step": 20819
    },
    {
      "epoch": 1.133732208858304,
      "grad_norm": 0.5857135309788942,
      "learning_rate": 8.32801167993033e-05,
      "loss": 11.8939,
      "step": 20820
    },
    {
      "epoch": 1.133786662854887,
      "grad_norm": 0.6147682545125815,
      "learning_rate": 8.327142273904078e-05,
      "loss": 12.0024,
      "step": 20821
    },
    {
      "epoch": 1.13384111685147,
      "grad_norm": 0.5800537403969274,
      "learning_rate": 8.326272880886245e-05,
      "loss": 12.0978,
      "step": 20822
    },
    {
      "epoch": 1.133895570848053,
      "grad_norm": 0.5212552774853718,
      "learning_rate": 8.325403500883597e-05,
      "loss": 11.96,
      "step": 20823
    },
    {
      "epoch": 1.133950024844636,
      "grad_norm": 0.5668164666074186,
      "learning_rate": 8.324534133902892e-05,
      "loss": 11.9927,
      "step": 20824
    },
    {
      "epoch": 1.134004478841219,
      "grad_norm": 0.5557225954644414,
      "learning_rate": 8.32366477995089e-05,
      "loss": 12.0414,
      "step": 20825
    },
    {
      "epoch": 1.134058932837802,
      "grad_norm": 0.536066984690846,
      "learning_rate": 8.322795439034352e-05,
      "loss": 12.0549,
      "step": 20826
    },
    {
      "epoch": 1.134113386834385,
      "grad_norm": 0.5094212987549769,
      "learning_rate": 8.321926111160038e-05,
      "loss": 11.9891,
      "step": 20827
    },
    {
      "epoch": 1.134167840830968,
      "grad_norm": 0.6266210587123217,
      "learning_rate": 8.321056796334707e-05,
      "loss": 12.0174,
      "step": 20828
    },
    {
      "epoch": 1.134222294827551,
      "grad_norm": 0.5920350735747615,
      "learning_rate": 8.320187494565121e-05,
      "loss": 12.1067,
      "step": 20829
    },
    {
      "epoch": 1.134276748824134,
      "grad_norm": 0.5157900799792909,
      "learning_rate": 8.319318205858036e-05,
      "loss": 11.9817,
      "step": 20830
    },
    {
      "epoch": 1.134331202820717,
      "grad_norm": 0.5484848096232319,
      "learning_rate": 8.318448930220216e-05,
      "loss": 11.996,
      "step": 20831
    },
    {
      "epoch": 1.1343856568173,
      "grad_norm": 0.4911315591287539,
      "learning_rate": 8.31757966765842e-05,
      "loss": 11.7924,
      "step": 20832
    },
    {
      "epoch": 1.134440110813883,
      "grad_norm": 0.48866603006549925,
      "learning_rate": 8.316710418179406e-05,
      "loss": 11.8883,
      "step": 20833
    },
    {
      "epoch": 1.134494564810466,
      "grad_norm": 0.5355135372081358,
      "learning_rate": 8.315841181789937e-05,
      "loss": 11.9627,
      "step": 20834
    },
    {
      "epoch": 1.134549018807049,
      "grad_norm": 0.5275092246816174,
      "learning_rate": 8.314971958496766e-05,
      "loss": 11.9631,
      "step": 20835
    },
    {
      "epoch": 1.134603472803632,
      "grad_norm": 0.5295825837600786,
      "learning_rate": 8.314102748306659e-05,
      "loss": 11.9989,
      "step": 20836
    },
    {
      "epoch": 1.134657926800215,
      "grad_norm": 0.5321704154501747,
      "learning_rate": 8.313233551226369e-05,
      "loss": 12.0102,
      "step": 20837
    },
    {
      "epoch": 1.134712380796798,
      "grad_norm": 0.5254003594455219,
      "learning_rate": 8.312364367262658e-05,
      "loss": 12.021,
      "step": 20838
    },
    {
      "epoch": 1.134766834793381,
      "grad_norm": 0.5216226383713083,
      "learning_rate": 8.311495196422284e-05,
      "loss": 11.9733,
      "step": 20839
    },
    {
      "epoch": 1.134821288789964,
      "grad_norm": 0.5459613177918216,
      "learning_rate": 8.310626038712007e-05,
      "loss": 12.0244,
      "step": 20840
    },
    {
      "epoch": 1.1348757427865472,
      "grad_norm": 0.5404943272558749,
      "learning_rate": 8.309756894138583e-05,
      "loss": 12.1336,
      "step": 20841
    },
    {
      "epoch": 1.1349301967831302,
      "grad_norm": 0.48839452700605696,
      "learning_rate": 8.308887762708776e-05,
      "loss": 12.013,
      "step": 20842
    },
    {
      "epoch": 1.1349846507797132,
      "grad_norm": 0.5520083087994961,
      "learning_rate": 8.30801864442934e-05,
      "loss": 12.0592,
      "step": 20843
    },
    {
      "epoch": 1.1350391047762962,
      "grad_norm": 0.5462084428929227,
      "learning_rate": 8.30714953930704e-05,
      "loss": 11.9144,
      "step": 20844
    },
    {
      "epoch": 1.1350935587728792,
      "grad_norm": 0.5471507247040064,
      "learning_rate": 8.306280447348622e-05,
      "loss": 11.8395,
      "step": 20845
    },
    {
      "epoch": 1.1351480127694622,
      "grad_norm": 0.5501453700212303,
      "learning_rate": 8.305411368560854e-05,
      "loss": 11.9186,
      "step": 20846
    },
    {
      "epoch": 1.1352024667660452,
      "grad_norm": 0.5454816436293293,
      "learning_rate": 8.304542302950491e-05,
      "loss": 11.9424,
      "step": 20847
    },
    {
      "epoch": 1.1352569207626282,
      "grad_norm": 0.4800147172461387,
      "learning_rate": 8.303673250524293e-05,
      "loss": 11.9729,
      "step": 20848
    },
    {
      "epoch": 1.1353113747592112,
      "grad_norm": 0.5395301056753729,
      "learning_rate": 8.302804211289015e-05,
      "loss": 11.9276,
      "step": 20849
    },
    {
      "epoch": 1.1353658287557942,
      "grad_norm": 0.545924304880344,
      "learning_rate": 8.301935185251415e-05,
      "loss": 11.9468,
      "step": 20850
    },
    {
      "epoch": 1.1354202827523772,
      "grad_norm": 0.5531911619321243,
      "learning_rate": 8.301066172418252e-05,
      "loss": 12.0504,
      "step": 20851
    },
    {
      "epoch": 1.1354747367489604,
      "grad_norm": 0.5295766318287469,
      "learning_rate": 8.300197172796283e-05,
      "loss": 12.0351,
      "step": 20852
    },
    {
      "epoch": 1.1355291907455434,
      "grad_norm": 0.5322049728341919,
      "learning_rate": 8.299328186392266e-05,
      "loss": 11.9966,
      "step": 20853
    },
    {
      "epoch": 1.1355836447421264,
      "grad_norm": 0.5081287053833177,
      "learning_rate": 8.298459213212964e-05,
      "loss": 11.9673,
      "step": 20854
    },
    {
      "epoch": 1.1356380987387094,
      "grad_norm": 0.6124065673217175,
      "learning_rate": 8.297590253265125e-05,
      "loss": 11.983,
      "step": 20855
    },
    {
      "epoch": 1.1356925527352923,
      "grad_norm": 0.53898433722122,
      "learning_rate": 8.296721306555505e-05,
      "loss": 11.9183,
      "step": 20856
    },
    {
      "epoch": 1.1357470067318753,
      "grad_norm": 0.5382650722310977,
      "learning_rate": 8.295852373090869e-05,
      "loss": 12.0799,
      "step": 20857
    },
    {
      "epoch": 1.1358014607284583,
      "grad_norm": 0.561854909778086,
      "learning_rate": 8.294983452877971e-05,
      "loss": 11.9602,
      "step": 20858
    },
    {
      "epoch": 1.1358559147250413,
      "grad_norm": 0.6108217648046498,
      "learning_rate": 8.294114545923567e-05,
      "loss": 11.9697,
      "step": 20859
    },
    {
      "epoch": 1.1359103687216243,
      "grad_norm": 0.5682286780819156,
      "learning_rate": 8.293245652234415e-05,
      "loss": 12.0881,
      "step": 20860
    },
    {
      "epoch": 1.1359648227182073,
      "grad_norm": 0.6408087879625746,
      "learning_rate": 8.29237677181727e-05,
      "loss": 12.0791,
      "step": 20861
    },
    {
      "epoch": 1.1360192767147903,
      "grad_norm": 0.5605205521100582,
      "learning_rate": 8.291507904678892e-05,
      "loss": 12.1151,
      "step": 20862
    },
    {
      "epoch": 1.1360737307113733,
      "grad_norm": 0.5605707280569578,
      "learning_rate": 8.29063905082603e-05,
      "loss": 12.0089,
      "step": 20863
    },
    {
      "epoch": 1.1361281847079563,
      "grad_norm": 0.5012292664038457,
      "learning_rate": 8.289770210265453e-05,
      "loss": 11.9938,
      "step": 20864
    },
    {
      "epoch": 1.1361826387045395,
      "grad_norm": 0.5684573322543377,
      "learning_rate": 8.288901383003907e-05,
      "loss": 12.0373,
      "step": 20865
    },
    {
      "epoch": 1.1362370927011225,
      "grad_norm": 0.6351742069265385,
      "learning_rate": 8.288032569048148e-05,
      "loss": 12.0249,
      "step": 20866
    },
    {
      "epoch": 1.1362915466977055,
      "grad_norm": 0.5222002852865439,
      "learning_rate": 8.287163768404934e-05,
      "loss": 12.0471,
      "step": 20867
    },
    {
      "epoch": 1.1363460006942885,
      "grad_norm": 0.5203810318401516,
      "learning_rate": 8.286294981081024e-05,
      "loss": 11.9294,
      "step": 20868
    },
    {
      "epoch": 1.1364004546908715,
      "grad_norm": 0.57972478770493,
      "learning_rate": 8.285426207083171e-05,
      "loss": 12.0424,
      "step": 20869
    },
    {
      "epoch": 1.1364549086874545,
      "grad_norm": 0.5544463610879068,
      "learning_rate": 8.284557446418133e-05,
      "loss": 12.0597,
      "step": 20870
    },
    {
      "epoch": 1.1365093626840375,
      "grad_norm": 0.5035943721139681,
      "learning_rate": 8.283688699092662e-05,
      "loss": 11.9878,
      "step": 20871
    },
    {
      "epoch": 1.1365638166806205,
      "grad_norm": 0.5344768186437175,
      "learning_rate": 8.282819965113516e-05,
      "loss": 12.0697,
      "step": 20872
    },
    {
      "epoch": 1.1366182706772034,
      "grad_norm": 0.5616838941972153,
      "learning_rate": 8.281951244487452e-05,
      "loss": 11.989,
      "step": 20873
    },
    {
      "epoch": 1.1366727246737864,
      "grad_norm": 0.5391340768095988,
      "learning_rate": 8.281082537221223e-05,
      "loss": 11.9472,
      "step": 20874
    },
    {
      "epoch": 1.1367271786703697,
      "grad_norm": 0.6130211524475987,
      "learning_rate": 8.280213843321583e-05,
      "loss": 12.1204,
      "step": 20875
    },
    {
      "epoch": 1.1367816326669526,
      "grad_norm": 0.522993663207442,
      "learning_rate": 8.279345162795288e-05,
      "loss": 12.0767,
      "step": 20876
    },
    {
      "epoch": 1.1368360866635356,
      "grad_norm": 0.5704096775661565,
      "learning_rate": 8.278476495649094e-05,
      "loss": 12.0927,
      "step": 20877
    },
    {
      "epoch": 1.1368905406601186,
      "grad_norm": 0.5182424211696456,
      "learning_rate": 8.277607841889754e-05,
      "loss": 11.9097,
      "step": 20878
    },
    {
      "epoch": 1.1369449946567016,
      "grad_norm": 0.5664499558712293,
      "learning_rate": 8.276739201524026e-05,
      "loss": 11.9551,
      "step": 20879
    },
    {
      "epoch": 1.1369994486532846,
      "grad_norm": 0.5124961895977534,
      "learning_rate": 8.275870574558661e-05,
      "loss": 11.9613,
      "step": 20880
    },
    {
      "epoch": 1.1370539026498676,
      "grad_norm": 0.49355180719984976,
      "learning_rate": 8.275001961000418e-05,
      "loss": 12.0667,
      "step": 20881
    },
    {
      "epoch": 1.1371083566464506,
      "grad_norm": 0.5412642353635919,
      "learning_rate": 8.27413336085605e-05,
      "loss": 12.0697,
      "step": 20882
    },
    {
      "epoch": 1.1371628106430336,
      "grad_norm": 0.5478756978199564,
      "learning_rate": 8.273264774132308e-05,
      "loss": 11.9211,
      "step": 20883
    },
    {
      "epoch": 1.1372172646396166,
      "grad_norm": 0.5521736725057819,
      "learning_rate": 8.272396200835952e-05,
      "loss": 12.0213,
      "step": 20884
    },
    {
      "epoch": 1.1372717186361996,
      "grad_norm": 0.5305067708369225,
      "learning_rate": 8.27152764097373e-05,
      "loss": 11.9613,
      "step": 20885
    },
    {
      "epoch": 1.1373261726327826,
      "grad_norm": 0.5280367029753679,
      "learning_rate": 8.270659094552399e-05,
      "loss": 12.0564,
      "step": 20886
    },
    {
      "epoch": 1.1373806266293656,
      "grad_norm": 0.523282340775394,
      "learning_rate": 8.269790561578713e-05,
      "loss": 11.9566,
      "step": 20887
    },
    {
      "epoch": 1.1374350806259488,
      "grad_norm": 0.5073888625187748,
      "learning_rate": 8.268922042059426e-05,
      "loss": 11.9844,
      "step": 20888
    },
    {
      "epoch": 1.1374895346225318,
      "grad_norm": 0.5630654298574181,
      "learning_rate": 8.268053536001291e-05,
      "loss": 11.9883,
      "step": 20889
    },
    {
      "epoch": 1.1375439886191148,
      "grad_norm": 0.49276678646697886,
      "learning_rate": 8.26718504341106e-05,
      "loss": 11.952,
      "step": 20890
    },
    {
      "epoch": 1.1375984426156978,
      "grad_norm": 0.6022868989355634,
      "learning_rate": 8.266316564295492e-05,
      "loss": 12.0448,
      "step": 20891
    },
    {
      "epoch": 1.1376528966122808,
      "grad_norm": 0.4980652688408364,
      "learning_rate": 8.265448098661337e-05,
      "loss": 12.0342,
      "step": 20892
    },
    {
      "epoch": 1.1377073506088637,
      "grad_norm": 0.4817094395889994,
      "learning_rate": 8.264579646515347e-05,
      "loss": 11.9591,
      "step": 20893
    },
    {
      "epoch": 1.1377618046054467,
      "grad_norm": 0.5647523062277882,
      "learning_rate": 8.26371120786428e-05,
      "loss": 12.1397,
      "step": 20894
    },
    {
      "epoch": 1.1378162586020297,
      "grad_norm": 0.5375555758113099,
      "learning_rate": 8.262842782714884e-05,
      "loss": 11.9929,
      "step": 20895
    },
    {
      "epoch": 1.1378707125986127,
      "grad_norm": 0.49082208075832834,
      "learning_rate": 8.261974371073913e-05,
      "loss": 11.9572,
      "step": 20896
    },
    {
      "epoch": 1.1379251665951957,
      "grad_norm": 0.5972417861121767,
      "learning_rate": 8.26110597294812e-05,
      "loss": 12.2294,
      "step": 20897
    },
    {
      "epoch": 1.137979620591779,
      "grad_norm": 0.5170383175147765,
      "learning_rate": 8.26023758834426e-05,
      "loss": 12.0637,
      "step": 20898
    },
    {
      "epoch": 1.138034074588362,
      "grad_norm": 0.5175739439103323,
      "learning_rate": 8.259369217269084e-05,
      "loss": 11.8461,
      "step": 20899
    },
    {
      "epoch": 1.138088528584945,
      "grad_norm": 0.6240727278872582,
      "learning_rate": 8.258500859729345e-05,
      "loss": 12.1108,
      "step": 20900
    },
    {
      "epoch": 1.138142982581528,
      "grad_norm": 0.5173425148012402,
      "learning_rate": 8.257632515731793e-05,
      "loss": 12.1129,
      "step": 20901
    },
    {
      "epoch": 1.138197436578111,
      "grad_norm": 0.5384766030862483,
      "learning_rate": 8.256764185283184e-05,
      "loss": 12.031,
      "step": 20902
    },
    {
      "epoch": 1.138251890574694,
      "grad_norm": 0.4990048614018597,
      "learning_rate": 8.25589586839027e-05,
      "loss": 11.9201,
      "step": 20903
    },
    {
      "epoch": 1.1383063445712769,
      "grad_norm": 0.5347219549756849,
      "learning_rate": 8.255027565059806e-05,
      "loss": 12.0611,
      "step": 20904
    },
    {
      "epoch": 1.1383607985678599,
      "grad_norm": 0.523376663435599,
      "learning_rate": 8.254159275298533e-05,
      "loss": 11.9876,
      "step": 20905
    },
    {
      "epoch": 1.1384152525644429,
      "grad_norm": 0.49233654599900334,
      "learning_rate": 8.253290999113215e-05,
      "loss": 12.0099,
      "step": 20906
    },
    {
      "epoch": 1.1384697065610259,
      "grad_norm": 0.5117814584178179,
      "learning_rate": 8.252422736510597e-05,
      "loss": 12.0075,
      "step": 20907
    },
    {
      "epoch": 1.1385241605576089,
      "grad_norm": 0.6113436907536076,
      "learning_rate": 8.251554487497436e-05,
      "loss": 12.0207,
      "step": 20908
    },
    {
      "epoch": 1.1385786145541918,
      "grad_norm": 0.5237525973981209,
      "learning_rate": 8.250686252080478e-05,
      "loss": 12.0911,
      "step": 20909
    },
    {
      "epoch": 1.1386330685507748,
      "grad_norm": 0.5453306120977369,
      "learning_rate": 8.249818030266476e-05,
      "loss": 11.9071,
      "step": 20910
    },
    {
      "epoch": 1.138687522547358,
      "grad_norm": 0.5957861253600286,
      "learning_rate": 8.248949822062182e-05,
      "loss": 12.0551,
      "step": 20911
    },
    {
      "epoch": 1.138741976543941,
      "grad_norm": 0.5245725217612074,
      "learning_rate": 8.248081627474349e-05,
      "loss": 12.0185,
      "step": 20912
    },
    {
      "epoch": 1.138796430540524,
      "grad_norm": 0.5930235919132725,
      "learning_rate": 8.247213446509728e-05,
      "loss": 11.9996,
      "step": 20913
    },
    {
      "epoch": 1.138850884537107,
      "grad_norm": 0.5450954577191609,
      "learning_rate": 8.246345279175073e-05,
      "loss": 11.9504,
      "step": 20914
    },
    {
      "epoch": 1.13890533853369,
      "grad_norm": 0.5530840617194773,
      "learning_rate": 8.245477125477125e-05,
      "loss": 12.01,
      "step": 20915
    },
    {
      "epoch": 1.138959792530273,
      "grad_norm": 0.5633213277208992,
      "learning_rate": 8.244608985422641e-05,
      "loss": 12.0457,
      "step": 20916
    },
    {
      "epoch": 1.139014246526856,
      "grad_norm": 0.6085785006143246,
      "learning_rate": 8.243740859018375e-05,
      "loss": 12.1239,
      "step": 20917
    },
    {
      "epoch": 1.139068700523439,
      "grad_norm": 0.5330646862283069,
      "learning_rate": 8.242872746271073e-05,
      "loss": 11.9485,
      "step": 20918
    },
    {
      "epoch": 1.139123154520022,
      "grad_norm": 0.5364267288916039,
      "learning_rate": 8.242004647187489e-05,
      "loss": 12.0183,
      "step": 20919
    },
    {
      "epoch": 1.139177608516605,
      "grad_norm": 0.5496380663520917,
      "learning_rate": 8.24113656177437e-05,
      "loss": 12.0552,
      "step": 20920
    },
    {
      "epoch": 1.139232062513188,
      "grad_norm": 0.5371979806699659,
      "learning_rate": 8.240268490038469e-05,
      "loss": 12.0438,
      "step": 20921
    },
    {
      "epoch": 1.1392865165097712,
      "grad_norm": 0.5498816405632554,
      "learning_rate": 8.239400431986535e-05,
      "loss": 11.9895,
      "step": 20922
    },
    {
      "epoch": 1.1393409705063542,
      "grad_norm": 0.5625453952844264,
      "learning_rate": 8.238532387625315e-05,
      "loss": 11.9807,
      "step": 20923
    },
    {
      "epoch": 1.1393954245029372,
      "grad_norm": 0.5611986634358086,
      "learning_rate": 8.23766435696157e-05,
      "loss": 12.0085,
      "step": 20924
    },
    {
      "epoch": 1.1394498784995202,
      "grad_norm": 0.5377483954179323,
      "learning_rate": 8.236796340002038e-05,
      "loss": 12.0947,
      "step": 20925
    },
    {
      "epoch": 1.1395043324961032,
      "grad_norm": 0.5262368616779615,
      "learning_rate": 8.235928336753475e-05,
      "loss": 11.9465,
      "step": 20926
    },
    {
      "epoch": 1.1395587864926862,
      "grad_norm": 0.5223377240065364,
      "learning_rate": 8.235060347222625e-05,
      "loss": 12.0398,
      "step": 20927
    },
    {
      "epoch": 1.1396132404892692,
      "grad_norm": 0.5642763894207888,
      "learning_rate": 8.234192371416245e-05,
      "loss": 11.8982,
      "step": 20928
    },
    {
      "epoch": 1.1396676944858521,
      "grad_norm": 0.5019674705192412,
      "learning_rate": 8.233324409341081e-05,
      "loss": 12.0971,
      "step": 20929
    },
    {
      "epoch": 1.1397221484824351,
      "grad_norm": 0.5285775151462369,
      "learning_rate": 8.232456461003882e-05,
      "loss": 11.9557,
      "step": 20930
    },
    {
      "epoch": 1.1397766024790181,
      "grad_norm": 0.5729624412586914,
      "learning_rate": 8.231588526411398e-05,
      "loss": 12.1298,
      "step": 20931
    },
    {
      "epoch": 1.1398310564756011,
      "grad_norm": 0.5606389022362214,
      "learning_rate": 8.230720605570379e-05,
      "loss": 11.9898,
      "step": 20932
    },
    {
      "epoch": 1.1398855104721841,
      "grad_norm": 0.5903555006233688,
      "learning_rate": 8.229852698487572e-05,
      "loss": 11.9642,
      "step": 20933
    },
    {
      "epoch": 1.139939964468767,
      "grad_norm": 0.6299819004174931,
      "learning_rate": 8.228984805169732e-05,
      "loss": 12.1306,
      "step": 20934
    },
    {
      "epoch": 1.1399944184653503,
      "grad_norm": 0.5577541227970585,
      "learning_rate": 8.228116925623599e-05,
      "loss": 12.029,
      "step": 20935
    },
    {
      "epoch": 1.1400488724619333,
      "grad_norm": 0.5175393691618557,
      "learning_rate": 8.227249059855926e-05,
      "loss": 12.1277,
      "step": 20936
    },
    {
      "epoch": 1.1401033264585163,
      "grad_norm": 0.5606078093504049,
      "learning_rate": 8.226381207873462e-05,
      "loss": 12.153,
      "step": 20937
    },
    {
      "epoch": 1.1401577804550993,
      "grad_norm": 0.5168465111345226,
      "learning_rate": 8.225513369682954e-05,
      "loss": 11.9966,
      "step": 20938
    },
    {
      "epoch": 1.1402122344516823,
      "grad_norm": 0.6631339416457368,
      "learning_rate": 8.224645545291151e-05,
      "loss": 12.0144,
      "step": 20939
    },
    {
      "epoch": 1.1402666884482653,
      "grad_norm": 0.6520044289618595,
      "learning_rate": 8.223777734704804e-05,
      "loss": 12.0726,
      "step": 20940
    },
    {
      "epoch": 1.1403211424448483,
      "grad_norm": 0.5234933027330751,
      "learning_rate": 8.222909937930658e-05,
      "loss": 11.9941,
      "step": 20941
    },
    {
      "epoch": 1.1403755964414313,
      "grad_norm": 0.5772608521354804,
      "learning_rate": 8.222042154975464e-05,
      "loss": 12.0001,
      "step": 20942
    },
    {
      "epoch": 1.1404300504380143,
      "grad_norm": 0.5533461011161972,
      "learning_rate": 8.221174385845967e-05,
      "loss": 11.9857,
      "step": 20943
    },
    {
      "epoch": 1.1404845044345973,
      "grad_norm": 0.5251306865625073,
      "learning_rate": 8.220306630548917e-05,
      "loss": 11.9284,
      "step": 20944
    },
    {
      "epoch": 1.1405389584311805,
      "grad_norm": 0.5049067693066229,
      "learning_rate": 8.219438889091062e-05,
      "loss": 11.9974,
      "step": 20945
    },
    {
      "epoch": 1.1405934124277635,
      "grad_norm": 0.562717881292853,
      "learning_rate": 8.218571161479148e-05,
      "loss": 12.0331,
      "step": 20946
    },
    {
      "epoch": 1.1406478664243465,
      "grad_norm": 0.5125144286918268,
      "learning_rate": 8.217703447719924e-05,
      "loss": 11.9266,
      "step": 20947
    },
    {
      "epoch": 1.1407023204209294,
      "grad_norm": 0.5769333365392472,
      "learning_rate": 8.216835747820135e-05,
      "loss": 12.0408,
      "step": 20948
    },
    {
      "epoch": 1.1407567744175124,
      "grad_norm": 0.5479203582995349,
      "learning_rate": 8.215968061786531e-05,
      "loss": 12.1124,
      "step": 20949
    },
    {
      "epoch": 1.1408112284140954,
      "grad_norm": 0.5596161916792809,
      "learning_rate": 8.215100389625857e-05,
      "loss": 12.0871,
      "step": 20950
    },
    {
      "epoch": 1.1408656824106784,
      "grad_norm": 0.5305381593496411,
      "learning_rate": 8.214232731344864e-05,
      "loss": 12.0144,
      "step": 20951
    },
    {
      "epoch": 1.1409201364072614,
      "grad_norm": 0.5416221276249467,
      "learning_rate": 8.213365086950296e-05,
      "loss": 12.0667,
      "step": 20952
    },
    {
      "epoch": 1.1409745904038444,
      "grad_norm": 0.5333267437537554,
      "learning_rate": 8.2124974564489e-05,
      "loss": 11.9662,
      "step": 20953
    },
    {
      "epoch": 1.1410290444004274,
      "grad_norm": 0.5099745896153601,
      "learning_rate": 8.211629839847426e-05,
      "loss": 11.9656,
      "step": 20954
    },
    {
      "epoch": 1.1410834983970104,
      "grad_norm": 0.5423766204134551,
      "learning_rate": 8.210762237152619e-05,
      "loss": 11.9546,
      "step": 20955
    },
    {
      "epoch": 1.1411379523935934,
      "grad_norm": 0.5532803593890093,
      "learning_rate": 8.209894648371222e-05,
      "loss": 12.0223,
      "step": 20956
    },
    {
      "epoch": 1.1411924063901764,
      "grad_norm": 0.5835951375321685,
      "learning_rate": 8.209027073509985e-05,
      "loss": 12.0279,
      "step": 20957
    },
    {
      "epoch": 1.1412468603867596,
      "grad_norm": 0.5905217931965142,
      "learning_rate": 8.208159512575654e-05,
      "loss": 12.1469,
      "step": 20958
    },
    {
      "epoch": 1.1413013143833426,
      "grad_norm": 0.5422334570641386,
      "learning_rate": 8.207291965574974e-05,
      "loss": 12.1288,
      "step": 20959
    },
    {
      "epoch": 1.1413557683799256,
      "grad_norm": 0.696890680767564,
      "learning_rate": 8.206424432514694e-05,
      "loss": 12.0775,
      "step": 20960
    },
    {
      "epoch": 1.1414102223765086,
      "grad_norm": 0.5585119847024153,
      "learning_rate": 8.205556913401555e-05,
      "loss": 11.99,
      "step": 20961
    },
    {
      "epoch": 1.1414646763730916,
      "grad_norm": 0.549365936991758,
      "learning_rate": 8.20468940824231e-05,
      "loss": 12.0516,
      "step": 20962
    },
    {
      "epoch": 1.1415191303696746,
      "grad_norm": 0.607436255760605,
      "learning_rate": 8.2038219170437e-05,
      "loss": 12.0653,
      "step": 20963
    },
    {
      "epoch": 1.1415735843662576,
      "grad_norm": 0.5655225055773133,
      "learning_rate": 8.202954439812472e-05,
      "loss": 12.1366,
      "step": 20964
    },
    {
      "epoch": 1.1416280383628405,
      "grad_norm": 0.5301801390339581,
      "learning_rate": 8.202086976555375e-05,
      "loss": 12.0433,
      "step": 20965
    },
    {
      "epoch": 1.1416824923594235,
      "grad_norm": 0.4744429712595387,
      "learning_rate": 8.201219527279147e-05,
      "loss": 11.8752,
      "step": 20966
    },
    {
      "epoch": 1.1417369463560065,
      "grad_norm": 0.507296115025727,
      "learning_rate": 8.200352091990539e-05,
      "loss": 11.9695,
      "step": 20967
    },
    {
      "epoch": 1.1417914003525897,
      "grad_norm": 0.5290589105091998,
      "learning_rate": 8.199484670696295e-05,
      "loss": 12.045,
      "step": 20968
    },
    {
      "epoch": 1.1418458543491727,
      "grad_norm": 0.7096101281605665,
      "learning_rate": 8.19861726340316e-05,
      "loss": 12.1148,
      "step": 20969
    },
    {
      "epoch": 1.1419003083457557,
      "grad_norm": 0.49424842042945244,
      "learning_rate": 8.197749870117879e-05,
      "loss": 12.0347,
      "step": 20970
    },
    {
      "epoch": 1.1419547623423387,
      "grad_norm": 0.5749252311480237,
      "learning_rate": 8.196882490847197e-05,
      "loss": 12.0663,
      "step": 20971
    },
    {
      "epoch": 1.1420092163389217,
      "grad_norm": 0.5670819687811587,
      "learning_rate": 8.196015125597858e-05,
      "loss": 12.0073,
      "step": 20972
    },
    {
      "epoch": 1.1420636703355047,
      "grad_norm": 0.555447687916874,
      "learning_rate": 8.195147774376609e-05,
      "loss": 12.0688,
      "step": 20973
    },
    {
      "epoch": 1.1421181243320877,
      "grad_norm": 0.5414378867694744,
      "learning_rate": 8.194280437190194e-05,
      "loss": 12.0326,
      "step": 20974
    },
    {
      "epoch": 1.1421725783286707,
      "grad_norm": 0.5773807343132444,
      "learning_rate": 8.19341311404536e-05,
      "loss": 12.0903,
      "step": 20975
    },
    {
      "epoch": 1.1422270323252537,
      "grad_norm": 0.5020329669878001,
      "learning_rate": 8.192545804948845e-05,
      "loss": 11.9684,
      "step": 20976
    },
    {
      "epoch": 1.1422814863218367,
      "grad_norm": 0.48234046727032104,
      "learning_rate": 8.191678509907396e-05,
      "loss": 11.9321,
      "step": 20977
    },
    {
      "epoch": 1.1423359403184197,
      "grad_norm": 0.567535834836397,
      "learning_rate": 8.190811228927761e-05,
      "loss": 12.0578,
      "step": 20978
    },
    {
      "epoch": 1.1423903943150027,
      "grad_norm": 0.5327119061845109,
      "learning_rate": 8.189943962016679e-05,
      "loss": 11.9767,
      "step": 20979
    },
    {
      "epoch": 1.1424448483115857,
      "grad_norm": 0.5632906184815956,
      "learning_rate": 8.189076709180898e-05,
      "loss": 12.0716,
      "step": 20980
    },
    {
      "epoch": 1.1424993023081689,
      "grad_norm": 0.593895236471705,
      "learning_rate": 8.188209470427159e-05,
      "loss": 12.1147,
      "step": 20981
    },
    {
      "epoch": 1.1425537563047519,
      "grad_norm": 0.520442931732901,
      "learning_rate": 8.187342245762209e-05,
      "loss": 11.8542,
      "step": 20982
    },
    {
      "epoch": 1.1426082103013349,
      "grad_norm": 0.5405669907945672,
      "learning_rate": 8.186475035192788e-05,
      "loss": 12.0936,
      "step": 20983
    },
    {
      "epoch": 1.1426626642979179,
      "grad_norm": 0.5492648899230916,
      "learning_rate": 8.185607838725639e-05,
      "loss": 11.9907,
      "step": 20984
    },
    {
      "epoch": 1.1427171182945008,
      "grad_norm": 0.5803230662974547,
      "learning_rate": 8.184740656367515e-05,
      "loss": 12.0568,
      "step": 20985
    },
    {
      "epoch": 1.1427715722910838,
      "grad_norm": 0.46560360122414524,
      "learning_rate": 8.183873488125147e-05,
      "loss": 11.9337,
      "step": 20986
    },
    {
      "epoch": 1.1428260262876668,
      "grad_norm": 0.49780376200500465,
      "learning_rate": 8.183006334005283e-05,
      "loss": 12.0872,
      "step": 20987
    },
    {
      "epoch": 1.1428804802842498,
      "grad_norm": 0.6016666680327781,
      "learning_rate": 8.182139194014665e-05,
      "loss": 12.0181,
      "step": 20988
    },
    {
      "epoch": 1.1429349342808328,
      "grad_norm": 0.6388919810710112,
      "learning_rate": 8.18127206816004e-05,
      "loss": 12.1541,
      "step": 20989
    },
    {
      "epoch": 1.1429893882774158,
      "grad_norm": 0.5375566325314137,
      "learning_rate": 8.180404956448147e-05,
      "loss": 12.0993,
      "step": 20990
    },
    {
      "epoch": 1.1430438422739988,
      "grad_norm": 0.5732752135944795,
      "learning_rate": 8.179537858885731e-05,
      "loss": 12.0784,
      "step": 20991
    },
    {
      "epoch": 1.143098296270582,
      "grad_norm": 0.5518683111626997,
      "learning_rate": 8.178670775479534e-05,
      "loss": 12.0159,
      "step": 20992
    },
    {
      "epoch": 1.143152750267165,
      "grad_norm": 0.5556360373222765,
      "learning_rate": 8.177803706236299e-05,
      "loss": 11.9894,
      "step": 20993
    },
    {
      "epoch": 1.143207204263748,
      "grad_norm": 0.5298063978603275,
      "learning_rate": 8.176936651162767e-05,
      "loss": 11.9655,
      "step": 20994
    },
    {
      "epoch": 1.143261658260331,
      "grad_norm": 0.5793338571832891,
      "learning_rate": 8.176069610265684e-05,
      "loss": 12.0498,
      "step": 20995
    },
    {
      "epoch": 1.143316112256914,
      "grad_norm": 0.6264256406282895,
      "learning_rate": 8.175202583551787e-05,
      "loss": 11.9412,
      "step": 20996
    },
    {
      "epoch": 1.143370566253497,
      "grad_norm": 0.5361244133679585,
      "learning_rate": 8.174335571027823e-05,
      "loss": 12.0836,
      "step": 20997
    },
    {
      "epoch": 1.14342502025008,
      "grad_norm": 0.5556546496980178,
      "learning_rate": 8.173468572700529e-05,
      "loss": 11.8692,
      "step": 20998
    },
    {
      "epoch": 1.143479474246663,
      "grad_norm": 0.5293104875373195,
      "learning_rate": 8.172601588576648e-05,
      "loss": 11.9413,
      "step": 20999
    },
    {
      "epoch": 1.143533928243246,
      "grad_norm": 0.5472832595633416,
      "learning_rate": 8.171734618662927e-05,
      "loss": 12.0546,
      "step": 21000
    },
    {
      "epoch": 1.143588382239829,
      "grad_norm": 0.5230902969254605,
      "learning_rate": 8.170867662966102e-05,
      "loss": 12.0118,
      "step": 21001
    },
    {
      "epoch": 1.143642836236412,
      "grad_norm": 0.5418456221500304,
      "learning_rate": 8.170000721492918e-05,
      "loss": 11.9179,
      "step": 21002
    },
    {
      "epoch": 1.143697290232995,
      "grad_norm": 0.5203268480355566,
      "learning_rate": 8.169133794250116e-05,
      "loss": 12.0247,
      "step": 21003
    },
    {
      "epoch": 1.143751744229578,
      "grad_norm": 0.5229551837980176,
      "learning_rate": 8.168266881244436e-05,
      "loss": 11.903,
      "step": 21004
    },
    {
      "epoch": 1.1438061982261611,
      "grad_norm": 0.5344584978853519,
      "learning_rate": 8.167399982482622e-05,
      "loss": 11.9609,
      "step": 21005
    },
    {
      "epoch": 1.1438606522227441,
      "grad_norm": 0.5228388886315154,
      "learning_rate": 8.166533097971412e-05,
      "loss": 12.05,
      "step": 21006
    },
    {
      "epoch": 1.1439151062193271,
      "grad_norm": 0.5711820104447795,
      "learning_rate": 8.165666227717546e-05,
      "loss": 11.9169,
      "step": 21007
    },
    {
      "epoch": 1.1439695602159101,
      "grad_norm": 0.5198358688515019,
      "learning_rate": 8.164799371727768e-05,
      "loss": 11.9617,
      "step": 21008
    },
    {
      "epoch": 1.1440240142124931,
      "grad_norm": 0.5458365819539345,
      "learning_rate": 8.163932530008817e-05,
      "loss": 12.09,
      "step": 21009
    },
    {
      "epoch": 1.144078468209076,
      "grad_norm": 0.5959144261997894,
      "learning_rate": 8.163065702567433e-05,
      "loss": 12.0294,
      "step": 21010
    },
    {
      "epoch": 1.144132922205659,
      "grad_norm": 0.5189776173020441,
      "learning_rate": 8.162198889410362e-05,
      "loss": 12.0019,
      "step": 21011
    },
    {
      "epoch": 1.144187376202242,
      "grad_norm": 0.5838636030572146,
      "learning_rate": 8.161332090544339e-05,
      "loss": 11.9223,
      "step": 21012
    },
    {
      "epoch": 1.144241830198825,
      "grad_norm": 0.553351221664354,
      "learning_rate": 8.160465305976107e-05,
      "loss": 12.0005,
      "step": 21013
    },
    {
      "epoch": 1.144296284195408,
      "grad_norm": 0.5587968080469131,
      "learning_rate": 8.159598535712405e-05,
      "loss": 11.8973,
      "step": 21014
    },
    {
      "epoch": 1.1443507381919913,
      "grad_norm": 0.6187991735080954,
      "learning_rate": 8.158731779759975e-05,
      "loss": 12.2249,
      "step": 21015
    },
    {
      "epoch": 1.1444051921885743,
      "grad_norm": 0.5533867559389424,
      "learning_rate": 8.157865038125552e-05,
      "loss": 12.02,
      "step": 21016
    },
    {
      "epoch": 1.1444596461851573,
      "grad_norm": 0.6128098583652383,
      "learning_rate": 8.156998310815882e-05,
      "loss": 12.2211,
      "step": 21017
    },
    {
      "epoch": 1.1445141001817403,
      "grad_norm": 0.5387386573489695,
      "learning_rate": 8.156131597837701e-05,
      "loss": 12.0796,
      "step": 21018
    },
    {
      "epoch": 1.1445685541783233,
      "grad_norm": 0.5328584040496055,
      "learning_rate": 8.15526489919775e-05,
      "loss": 11.837,
      "step": 21019
    },
    {
      "epoch": 1.1446230081749063,
      "grad_norm": 0.5710270942995136,
      "learning_rate": 8.154398214902769e-05,
      "loss": 12.1224,
      "step": 21020
    },
    {
      "epoch": 1.1446774621714892,
      "grad_norm": 0.5795974385270815,
      "learning_rate": 8.153531544959494e-05,
      "loss": 12.0274,
      "step": 21021
    },
    {
      "epoch": 1.1447319161680722,
      "grad_norm": 0.5319926295155886,
      "learning_rate": 8.15266488937467e-05,
      "loss": 11.9125,
      "step": 21022
    },
    {
      "epoch": 1.1447863701646552,
      "grad_norm": 0.5604516086391089,
      "learning_rate": 8.151798248155032e-05,
      "loss": 11.8868,
      "step": 21023
    },
    {
      "epoch": 1.1448408241612382,
      "grad_norm": 0.5489005774116194,
      "learning_rate": 8.150931621307323e-05,
      "loss": 11.9741,
      "step": 21024
    },
    {
      "epoch": 1.1448952781578212,
      "grad_norm": 0.5389533652488021,
      "learning_rate": 8.150065008838281e-05,
      "loss": 12.0303,
      "step": 21025
    },
    {
      "epoch": 1.1449497321544042,
      "grad_norm": 0.5405954217513335,
      "learning_rate": 8.149198410754641e-05,
      "loss": 11.9794,
      "step": 21026
    },
    {
      "epoch": 1.1450041861509872,
      "grad_norm": 0.5810190019682795,
      "learning_rate": 8.148331827063147e-05,
      "loss": 11.935,
      "step": 21027
    },
    {
      "epoch": 1.1450586401475704,
      "grad_norm": 0.7319128147122567,
      "learning_rate": 8.147465257770532e-05,
      "loss": 12.0482,
      "step": 21028
    },
    {
      "epoch": 1.1451130941441534,
      "grad_norm": 0.5822036562162385,
      "learning_rate": 8.14659870288354e-05,
      "loss": 12.0907,
      "step": 21029
    },
    {
      "epoch": 1.1451675481407364,
      "grad_norm": 0.5532352398438115,
      "learning_rate": 8.145732162408907e-05,
      "loss": 11.9685,
      "step": 21030
    },
    {
      "epoch": 1.1452220021373194,
      "grad_norm": 0.5712018466987027,
      "learning_rate": 8.144865636353371e-05,
      "loss": 12.1183,
      "step": 21031
    },
    {
      "epoch": 1.1452764561339024,
      "grad_norm": 0.54110084551623,
      "learning_rate": 8.14399912472367e-05,
      "loss": 12.0274,
      "step": 21032
    },
    {
      "epoch": 1.1453309101304854,
      "grad_norm": 0.5363305858709243,
      "learning_rate": 8.143132627526545e-05,
      "loss": 12.0609,
      "step": 21033
    },
    {
      "epoch": 1.1453853641270684,
      "grad_norm": 0.5843397847758113,
      "learning_rate": 8.142266144768729e-05,
      "loss": 12.0073,
      "step": 21034
    },
    {
      "epoch": 1.1454398181236514,
      "grad_norm": 0.5350188464651414,
      "learning_rate": 8.141399676456972e-05,
      "loss": 12.0521,
      "step": 21035
    },
    {
      "epoch": 1.1454942721202344,
      "grad_norm": 0.5420875012653492,
      "learning_rate": 8.140533222597995e-05,
      "loss": 12.0326,
      "step": 21036
    },
    {
      "epoch": 1.1455487261168174,
      "grad_norm": 0.5653638168090159,
      "learning_rate": 8.139666783198542e-05,
      "loss": 11.9954,
      "step": 21037
    },
    {
      "epoch": 1.1456031801134006,
      "grad_norm": 0.5723196320718857,
      "learning_rate": 8.138800358265354e-05,
      "loss": 12.0604,
      "step": 21038
    },
    {
      "epoch": 1.1456576341099836,
      "grad_norm": 0.5773998459245281,
      "learning_rate": 8.137933947805169e-05,
      "loss": 12.0177,
      "step": 21039
    },
    {
      "epoch": 1.1457120881065666,
      "grad_norm": 0.51146925626736,
      "learning_rate": 8.13706755182472e-05,
      "loss": 12.0252,
      "step": 21040
    },
    {
      "epoch": 1.1457665421031495,
      "grad_norm": 0.6265872521892802,
      "learning_rate": 8.136201170330746e-05,
      "loss": 12.0123,
      "step": 21041
    },
    {
      "epoch": 1.1458209960997325,
      "grad_norm": 0.4907310093920785,
      "learning_rate": 8.135334803329983e-05,
      "loss": 11.9729,
      "step": 21042
    },
    {
      "epoch": 1.1458754500963155,
      "grad_norm": 0.5463913693585858,
      "learning_rate": 8.134468450829172e-05,
      "loss": 11.9928,
      "step": 21043
    },
    {
      "epoch": 1.1459299040928985,
      "grad_norm": 0.5455569596044371,
      "learning_rate": 8.133602112835043e-05,
      "loss": 11.9847,
      "step": 21044
    },
    {
      "epoch": 1.1459843580894815,
      "grad_norm": 0.5316381424131947,
      "learning_rate": 8.132735789354346e-05,
      "loss": 11.9869,
      "step": 21045
    },
    {
      "epoch": 1.1460388120860645,
      "grad_norm": 0.5544940775590294,
      "learning_rate": 8.131869480393803e-05,
      "loss": 12.0745,
      "step": 21046
    },
    {
      "epoch": 1.1460932660826475,
      "grad_norm": 0.6004156000299287,
      "learning_rate": 8.131003185960154e-05,
      "loss": 12.0626,
      "step": 21047
    },
    {
      "epoch": 1.1461477200792305,
      "grad_norm": 0.5218071904391417,
      "learning_rate": 8.130136906060137e-05,
      "loss": 12.0363,
      "step": 21048
    },
    {
      "epoch": 1.1462021740758135,
      "grad_norm": 0.5389146853602708,
      "learning_rate": 8.129270640700492e-05,
      "loss": 12.006,
      "step": 21049
    },
    {
      "epoch": 1.1462566280723965,
      "grad_norm": 0.6203417650257257,
      "learning_rate": 8.128404389887953e-05,
      "loss": 12.0354,
      "step": 21050
    },
    {
      "epoch": 1.1463110820689797,
      "grad_norm": 0.5902856739990013,
      "learning_rate": 8.127538153629253e-05,
      "loss": 12.0125,
      "step": 21051
    },
    {
      "epoch": 1.1463655360655627,
      "grad_norm": 0.5306534686815397,
      "learning_rate": 8.126671931931131e-05,
      "loss": 12.1156,
      "step": 21052
    },
    {
      "epoch": 1.1464199900621457,
      "grad_norm": 0.5373704830394005,
      "learning_rate": 8.125805724800323e-05,
      "loss": 11.9522,
      "step": 21053
    },
    {
      "epoch": 1.1464744440587287,
      "grad_norm": 0.5377695949027106,
      "learning_rate": 8.124939532243564e-05,
      "loss": 11.9527,
      "step": 21054
    },
    {
      "epoch": 1.1465288980553117,
      "grad_norm": 0.5911281005314784,
      "learning_rate": 8.12407335426759e-05,
      "loss": 12.1034,
      "step": 21055
    },
    {
      "epoch": 1.1465833520518947,
      "grad_norm": 0.4977620166418643,
      "learning_rate": 8.123207190879136e-05,
      "loss": 11.9416,
      "step": 21056
    },
    {
      "epoch": 1.1466378060484776,
      "grad_norm": 0.5116839969984828,
      "learning_rate": 8.122341042084938e-05,
      "loss": 12.0125,
      "step": 21057
    },
    {
      "epoch": 1.1466922600450606,
      "grad_norm": 0.537829407364425,
      "learning_rate": 8.12147490789173e-05,
      "loss": 12.0088,
      "step": 21058
    },
    {
      "epoch": 1.1467467140416436,
      "grad_norm": 0.5566024102714832,
      "learning_rate": 8.120608788306245e-05,
      "loss": 11.992,
      "step": 21059
    },
    {
      "epoch": 1.1468011680382266,
      "grad_norm": 0.5172239653070049,
      "learning_rate": 8.119742683335225e-05,
      "loss": 12.038,
      "step": 21060
    },
    {
      "epoch": 1.1468556220348098,
      "grad_norm": 0.5081733170513308,
      "learning_rate": 8.1188765929854e-05,
      "loss": 11.9034,
      "step": 21061
    },
    {
      "epoch": 1.1469100760313928,
      "grad_norm": 0.6079477794406521,
      "learning_rate": 8.118010517263506e-05,
      "loss": 11.9627,
      "step": 21062
    },
    {
      "epoch": 1.1469645300279758,
      "grad_norm": 0.5826173534014737,
      "learning_rate": 8.11714445617628e-05,
      "loss": 12.0594,
      "step": 21063
    },
    {
      "epoch": 1.1470189840245588,
      "grad_norm": 0.5632061995460318,
      "learning_rate": 8.116278409730452e-05,
      "loss": 11.9167,
      "step": 21064
    },
    {
      "epoch": 1.1470734380211418,
      "grad_norm": 0.5485729104136255,
      "learning_rate": 8.115412377932762e-05,
      "loss": 12.1227,
      "step": 21065
    },
    {
      "epoch": 1.1471278920177248,
      "grad_norm": 0.5528861036138745,
      "learning_rate": 8.11454636078994e-05,
      "loss": 12.0376,
      "step": 21066
    },
    {
      "epoch": 1.1471823460143078,
      "grad_norm": 0.5574868688652987,
      "learning_rate": 8.11368035830872e-05,
      "loss": 11.8677,
      "step": 21067
    },
    {
      "epoch": 1.1472368000108908,
      "grad_norm": 0.5818010753479786,
      "learning_rate": 8.112814370495839e-05,
      "loss": 12.0715,
      "step": 21068
    },
    {
      "epoch": 1.1472912540074738,
      "grad_norm": 0.6798929789492493,
      "learning_rate": 8.11194839735803e-05,
      "loss": 12.1723,
      "step": 21069
    },
    {
      "epoch": 1.1473457080040568,
      "grad_norm": 0.5422378399543559,
      "learning_rate": 8.111082438902025e-05,
      "loss": 12.065,
      "step": 21070
    },
    {
      "epoch": 1.1474001620006398,
      "grad_norm": 0.5122511501381658,
      "learning_rate": 8.110216495134562e-05,
      "loss": 11.9232,
      "step": 21071
    },
    {
      "epoch": 1.1474546159972228,
      "grad_norm": 0.5453350705386038,
      "learning_rate": 8.10935056606237e-05,
      "loss": 12.0262,
      "step": 21072
    },
    {
      "epoch": 1.1475090699938058,
      "grad_norm": 0.6438196027922032,
      "learning_rate": 8.108484651692188e-05,
      "loss": 12.0422,
      "step": 21073
    },
    {
      "epoch": 1.1475635239903887,
      "grad_norm": 0.6528455938916412,
      "learning_rate": 8.107618752030745e-05,
      "loss": 12.0682,
      "step": 21074
    },
    {
      "epoch": 1.147617977986972,
      "grad_norm": 0.5451997878687775,
      "learning_rate": 8.10675286708478e-05,
      "loss": 12.0776,
      "step": 21075
    },
    {
      "epoch": 1.147672431983555,
      "grad_norm": 0.520908444699066,
      "learning_rate": 8.105886996861017e-05,
      "loss": 11.988,
      "step": 21076
    },
    {
      "epoch": 1.147726885980138,
      "grad_norm": 0.5852947218268076,
      "learning_rate": 8.105021141366196e-05,
      "loss": 12.0616,
      "step": 21077
    },
    {
      "epoch": 1.147781339976721,
      "grad_norm": 0.5432544667041285,
      "learning_rate": 8.104155300607049e-05,
      "loss": 11.9157,
      "step": 21078
    },
    {
      "epoch": 1.147835793973304,
      "grad_norm": 0.4872885499050939,
      "learning_rate": 8.103289474590308e-05,
      "loss": 11.9543,
      "step": 21079
    },
    {
      "epoch": 1.147890247969887,
      "grad_norm": 0.5648118091473893,
      "learning_rate": 8.102423663322704e-05,
      "loss": 11.9183,
      "step": 21080
    },
    {
      "epoch": 1.14794470196647,
      "grad_norm": 0.5579987798738413,
      "learning_rate": 8.101557866810972e-05,
      "loss": 12.1045,
      "step": 21081
    },
    {
      "epoch": 1.147999155963053,
      "grad_norm": 0.5483380124450635,
      "learning_rate": 8.100692085061847e-05,
      "loss": 12.0214,
      "step": 21082
    },
    {
      "epoch": 1.148053609959636,
      "grad_norm": 0.532668117450102,
      "learning_rate": 8.099826318082057e-05,
      "loss": 12.1756,
      "step": 21083
    },
    {
      "epoch": 1.148108063956219,
      "grad_norm": 0.5636987922499963,
      "learning_rate": 8.098960565878337e-05,
      "loss": 11.9614,
      "step": 21084
    },
    {
      "epoch": 1.148162517952802,
      "grad_norm": 0.5606659244042904,
      "learning_rate": 8.098094828457424e-05,
      "loss": 12.1564,
      "step": 21085
    },
    {
      "epoch": 1.148216971949385,
      "grad_norm": 0.5411349597727095,
      "learning_rate": 8.097229105826036e-05,
      "loss": 11.9888,
      "step": 21086
    },
    {
      "epoch": 1.148271425945968,
      "grad_norm": 0.6505933994179495,
      "learning_rate": 8.096363397990917e-05,
      "loss": 11.8961,
      "step": 21087
    },
    {
      "epoch": 1.148325879942551,
      "grad_norm": 0.495973865804788,
      "learning_rate": 8.095497704958795e-05,
      "loss": 11.969,
      "step": 21088
    },
    {
      "epoch": 1.148380333939134,
      "grad_norm": 0.48686591604475277,
      "learning_rate": 8.094632026736403e-05,
      "loss": 11.9074,
      "step": 21089
    },
    {
      "epoch": 1.148434787935717,
      "grad_norm": 0.5323346380652924,
      "learning_rate": 8.093766363330471e-05,
      "loss": 11.9727,
      "step": 21090
    },
    {
      "epoch": 1.1484892419323,
      "grad_norm": 0.5178261071161707,
      "learning_rate": 8.092900714747731e-05,
      "loss": 11.9483,
      "step": 21091
    },
    {
      "epoch": 1.148543695928883,
      "grad_norm": 0.543103451995716,
      "learning_rate": 8.092035080994917e-05,
      "loss": 11.9492,
      "step": 21092
    },
    {
      "epoch": 1.148598149925466,
      "grad_norm": 0.5291390005538827,
      "learning_rate": 8.091169462078754e-05,
      "loss": 12.0143,
      "step": 21093
    },
    {
      "epoch": 1.148652603922049,
      "grad_norm": 0.560070717878981,
      "learning_rate": 8.09030385800598e-05,
      "loss": 12.1458,
      "step": 21094
    },
    {
      "epoch": 1.148707057918632,
      "grad_norm": 0.5588331491242884,
      "learning_rate": 8.089438268783323e-05,
      "loss": 12.0294,
      "step": 21095
    },
    {
      "epoch": 1.148761511915215,
      "grad_norm": 0.5007083159102311,
      "learning_rate": 8.08857269441752e-05,
      "loss": 11.9746,
      "step": 21096
    },
    {
      "epoch": 1.148815965911798,
      "grad_norm": 0.5677994203029244,
      "learning_rate": 8.087707134915288e-05,
      "loss": 11.9238,
      "step": 21097
    },
    {
      "epoch": 1.1488704199083812,
      "grad_norm": 0.5813209840921494,
      "learning_rate": 8.08684159028337e-05,
      "loss": 12.088,
      "step": 21098
    },
    {
      "epoch": 1.1489248739049642,
      "grad_norm": 0.5525302957048495,
      "learning_rate": 8.085976060528491e-05,
      "loss": 12.0606,
      "step": 21099
    },
    {
      "epoch": 1.1489793279015472,
      "grad_norm": 0.5496389271612488,
      "learning_rate": 8.085110545657385e-05,
      "loss": 12.0706,
      "step": 21100
    },
    {
      "epoch": 1.1490337818981302,
      "grad_norm": 0.5482635604991649,
      "learning_rate": 8.084245045676779e-05,
      "loss": 12.0023,
      "step": 21101
    },
    {
      "epoch": 1.1490882358947132,
      "grad_norm": 0.5469133095471433,
      "learning_rate": 8.083379560593406e-05,
      "loss": 12.0108,
      "step": 21102
    },
    {
      "epoch": 1.1491426898912962,
      "grad_norm": 0.5482534608450617,
      "learning_rate": 8.082514090413994e-05,
      "loss": 12.0259,
      "step": 21103
    },
    {
      "epoch": 1.1491971438878792,
      "grad_norm": 0.5767821949269134,
      "learning_rate": 8.081648635145272e-05,
      "loss": 12.1142,
      "step": 21104
    },
    {
      "epoch": 1.1492515978844622,
      "grad_norm": 0.5878543688318114,
      "learning_rate": 8.080783194793975e-05,
      "loss": 11.9575,
      "step": 21105
    },
    {
      "epoch": 1.1493060518810452,
      "grad_norm": 0.537615416805045,
      "learning_rate": 8.079917769366833e-05,
      "loss": 11.9365,
      "step": 21106
    },
    {
      "epoch": 1.1493605058776282,
      "grad_norm": 0.5902894231903612,
      "learning_rate": 8.079052358870568e-05,
      "loss": 12.0805,
      "step": 21107
    },
    {
      "epoch": 1.1494149598742114,
      "grad_norm": 0.5797267215268892,
      "learning_rate": 8.078186963311912e-05,
      "loss": 12.0357,
      "step": 21108
    },
    {
      "epoch": 1.1494694138707944,
      "grad_norm": 0.5578950816383108,
      "learning_rate": 8.0773215826976e-05,
      "loss": 12.0792,
      "step": 21109
    },
    {
      "epoch": 1.1495238678673774,
      "grad_norm": 0.531064236981189,
      "learning_rate": 8.076456217034356e-05,
      "loss": 12.0386,
      "step": 21110
    },
    {
      "epoch": 1.1495783218639604,
      "grad_norm": 0.508623729429057,
      "learning_rate": 8.075590866328911e-05,
      "loss": 12.0248,
      "step": 21111
    },
    {
      "epoch": 1.1496327758605434,
      "grad_norm": 0.5843775193424045,
      "learning_rate": 8.074725530587996e-05,
      "loss": 12.0742,
      "step": 21112
    },
    {
      "epoch": 1.1496872298571263,
      "grad_norm": 0.4926905519642264,
      "learning_rate": 8.073860209818336e-05,
      "loss": 11.9017,
      "step": 21113
    },
    {
      "epoch": 1.1497416838537093,
      "grad_norm": 0.5328678072904144,
      "learning_rate": 8.072994904026663e-05,
      "loss": 11.9381,
      "step": 21114
    },
    {
      "epoch": 1.1497961378502923,
      "grad_norm": 0.526885307689961,
      "learning_rate": 8.072129613219703e-05,
      "loss": 11.9485,
      "step": 21115
    },
    {
      "epoch": 1.1498505918468753,
      "grad_norm": 0.5497004056654098,
      "learning_rate": 8.071264337404192e-05,
      "loss": 12.0153,
      "step": 21116
    },
    {
      "epoch": 1.1499050458434583,
      "grad_norm": 0.5863572104234963,
      "learning_rate": 8.070399076586849e-05,
      "loss": 11.9572,
      "step": 21117
    },
    {
      "epoch": 1.1499594998400413,
      "grad_norm": 0.5615088944244939,
      "learning_rate": 8.069533830774407e-05,
      "loss": 12.0666,
      "step": 21118
    },
    {
      "epoch": 1.1500139538366243,
      "grad_norm": 0.7323211513084971,
      "learning_rate": 8.06866859997359e-05,
      "loss": 12.0178,
      "step": 21119
    },
    {
      "epoch": 1.1500684078332073,
      "grad_norm": 0.5678232451995301,
      "learning_rate": 8.067803384191133e-05,
      "loss": 11.9026,
      "step": 21120
    },
    {
      "epoch": 1.1501228618297905,
      "grad_norm": 0.5446422610668087,
      "learning_rate": 8.066938183433762e-05,
      "loss": 11.9061,
      "step": 21121
    },
    {
      "epoch": 1.1501773158263735,
      "grad_norm": 0.537432196594231,
      "learning_rate": 8.066072997708203e-05,
      "loss": 11.9063,
      "step": 21122
    },
    {
      "epoch": 1.1502317698229565,
      "grad_norm": 0.557170520357956,
      "learning_rate": 8.065207827021184e-05,
      "loss": 11.9433,
      "step": 21123
    },
    {
      "epoch": 1.1502862238195395,
      "grad_norm": 0.7040948029689192,
      "learning_rate": 8.064342671379435e-05,
      "loss": 12.1554,
      "step": 21124
    },
    {
      "epoch": 1.1503406778161225,
      "grad_norm": 0.6032339334585413,
      "learning_rate": 8.063477530789681e-05,
      "loss": 11.9106,
      "step": 21125
    },
    {
      "epoch": 1.1503951318127055,
      "grad_norm": 0.5399728029226735,
      "learning_rate": 8.062612405258652e-05,
      "loss": 11.9057,
      "step": 21126
    },
    {
      "epoch": 1.1504495858092885,
      "grad_norm": 0.4955261796241731,
      "learning_rate": 8.061747294793071e-05,
      "loss": 12.0238,
      "step": 21127
    },
    {
      "epoch": 1.1505040398058715,
      "grad_norm": 0.5855436553356044,
      "learning_rate": 8.06088219939967e-05,
      "loss": 12.0454,
      "step": 21128
    },
    {
      "epoch": 1.1505584938024545,
      "grad_norm": 0.605434000788639,
      "learning_rate": 8.060017119085173e-05,
      "loss": 12.1447,
      "step": 21129
    },
    {
      "epoch": 1.1506129477990374,
      "grad_norm": 0.5567938188227343,
      "learning_rate": 8.059152053856307e-05,
      "loss": 11.985,
      "step": 21130
    },
    {
      "epoch": 1.1506674017956207,
      "grad_norm": 0.5553514637675814,
      "learning_rate": 8.058287003719802e-05,
      "loss": 12.0672,
      "step": 21131
    },
    {
      "epoch": 1.1507218557922037,
      "grad_norm": 0.5184625714277747,
      "learning_rate": 8.057421968682383e-05,
      "loss": 12.0448,
      "step": 21132
    },
    {
      "epoch": 1.1507763097887866,
      "grad_norm": 0.5204256390067515,
      "learning_rate": 8.056556948750777e-05,
      "loss": 11.9548,
      "step": 21133
    },
    {
      "epoch": 1.1508307637853696,
      "grad_norm": 0.5953661138703211,
      "learning_rate": 8.055691943931707e-05,
      "loss": 12.0341,
      "step": 21134
    },
    {
      "epoch": 1.1508852177819526,
      "grad_norm": 0.8378579415423998,
      "learning_rate": 8.054826954231906e-05,
      "loss": 12.039,
      "step": 21135
    },
    {
      "epoch": 1.1509396717785356,
      "grad_norm": 0.5715107785392818,
      "learning_rate": 8.053961979658098e-05,
      "loss": 12.1688,
      "step": 21136
    },
    {
      "epoch": 1.1509941257751186,
      "grad_norm": 0.563690352731225,
      "learning_rate": 8.053097020217006e-05,
      "loss": 12.0174,
      "step": 21137
    },
    {
      "epoch": 1.1510485797717016,
      "grad_norm": 0.5488959381633189,
      "learning_rate": 8.052232075915359e-05,
      "loss": 12.0471,
      "step": 21138
    },
    {
      "epoch": 1.1511030337682846,
      "grad_norm": 0.5663448934394109,
      "learning_rate": 8.05136714675988e-05,
      "loss": 12.1247,
      "step": 21139
    },
    {
      "epoch": 1.1511574877648676,
      "grad_norm": 0.5685910922950765,
      "learning_rate": 8.050502232757297e-05,
      "loss": 12.0227,
      "step": 21140
    },
    {
      "epoch": 1.1512119417614506,
      "grad_norm": 0.5900570812695328,
      "learning_rate": 8.049637333914336e-05,
      "loss": 11.9444,
      "step": 21141
    },
    {
      "epoch": 1.1512663957580336,
      "grad_norm": 0.5252079331707762,
      "learning_rate": 8.04877245023772e-05,
      "loss": 12.0593,
      "step": 21142
    },
    {
      "epoch": 1.1513208497546166,
      "grad_norm": 0.5511587861500431,
      "learning_rate": 8.047907581734178e-05,
      "loss": 12.0923,
      "step": 21143
    },
    {
      "epoch": 1.1513753037511998,
      "grad_norm": 0.47982691738561195,
      "learning_rate": 8.047042728410436e-05,
      "loss": 11.9816,
      "step": 21144
    },
    {
      "epoch": 1.1514297577477828,
      "grad_norm": 0.5415394760446558,
      "learning_rate": 8.046177890273216e-05,
      "loss": 12.122,
      "step": 21145
    },
    {
      "epoch": 1.1514842117443658,
      "grad_norm": 0.5198899441470982,
      "learning_rate": 8.045313067329248e-05,
      "loss": 11.7322,
      "step": 21146
    },
    {
      "epoch": 1.1515386657409488,
      "grad_norm": 0.5429851464257295,
      "learning_rate": 8.044448259585249e-05,
      "loss": 12.0108,
      "step": 21147
    },
    {
      "epoch": 1.1515931197375318,
      "grad_norm": 0.5622638318610981,
      "learning_rate": 8.043583467047949e-05,
      "loss": 12.1389,
      "step": 21148
    },
    {
      "epoch": 1.1516475737341147,
      "grad_norm": 0.530104001255593,
      "learning_rate": 8.042718689724072e-05,
      "loss": 11.9547,
      "step": 21149
    },
    {
      "epoch": 1.1517020277306977,
      "grad_norm": 0.5883781883269091,
      "learning_rate": 8.041853927620345e-05,
      "loss": 12.0862,
      "step": 21150
    },
    {
      "epoch": 1.1517564817272807,
      "grad_norm": 0.5161862523816715,
      "learning_rate": 8.040989180743487e-05,
      "loss": 11.8887,
      "step": 21151
    },
    {
      "epoch": 1.1518109357238637,
      "grad_norm": 0.5728392481592719,
      "learning_rate": 8.040124449100226e-05,
      "loss": 12.0205,
      "step": 21152
    },
    {
      "epoch": 1.1518653897204467,
      "grad_norm": 0.5858138955307554,
      "learning_rate": 8.039259732697286e-05,
      "loss": 12.0949,
      "step": 21153
    },
    {
      "epoch": 1.1519198437170297,
      "grad_norm": 0.5524384241816089,
      "learning_rate": 8.038395031541392e-05,
      "loss": 12.0203,
      "step": 21154
    },
    {
      "epoch": 1.151974297713613,
      "grad_norm": 0.535014389337564,
      "learning_rate": 8.037530345639267e-05,
      "loss": 11.9085,
      "step": 21155
    },
    {
      "epoch": 1.152028751710196,
      "grad_norm": 0.6554376807816443,
      "learning_rate": 8.036665674997639e-05,
      "loss": 11.9767,
      "step": 21156
    },
    {
      "epoch": 1.152083205706779,
      "grad_norm": 0.5947689449903816,
      "learning_rate": 8.035801019623224e-05,
      "loss": 12.0815,
      "step": 21157
    },
    {
      "epoch": 1.152137659703362,
      "grad_norm": 0.548992716856938,
      "learning_rate": 8.034936379522749e-05,
      "loss": 11.892,
      "step": 21158
    },
    {
      "epoch": 1.152192113699945,
      "grad_norm": 0.5515032063781492,
      "learning_rate": 8.034071754702938e-05,
      "loss": 11.9991,
      "step": 21159
    },
    {
      "epoch": 1.152246567696528,
      "grad_norm": 0.5111353487376212,
      "learning_rate": 8.033207145170516e-05,
      "loss": 11.8936,
      "step": 21160
    },
    {
      "epoch": 1.1523010216931109,
      "grad_norm": 0.6442684090397887,
      "learning_rate": 8.032342550932206e-05,
      "loss": 12.0178,
      "step": 21161
    },
    {
      "epoch": 1.1523554756896939,
      "grad_norm": 0.5372293081498746,
      "learning_rate": 8.03147797199473e-05,
      "loss": 11.7815,
      "step": 21162
    },
    {
      "epoch": 1.1524099296862769,
      "grad_norm": 0.531260849518373,
      "learning_rate": 8.030613408364812e-05,
      "loss": 12.1365,
      "step": 21163
    },
    {
      "epoch": 1.1524643836828599,
      "grad_norm": 0.5124011686686147,
      "learning_rate": 8.029748860049168e-05,
      "loss": 12.0414,
      "step": 21164
    },
    {
      "epoch": 1.1525188376794429,
      "grad_norm": 0.5123375927676078,
      "learning_rate": 8.028884327054534e-05,
      "loss": 11.8998,
      "step": 21165
    },
    {
      "epoch": 1.1525732916760258,
      "grad_norm": 0.5245392110966306,
      "learning_rate": 8.028019809387629e-05,
      "loss": 11.9568,
      "step": 21166
    },
    {
      "epoch": 1.1526277456726088,
      "grad_norm": 0.5271637670116032,
      "learning_rate": 8.027155307055167e-05,
      "loss": 11.9273,
      "step": 21167
    },
    {
      "epoch": 1.152682199669192,
      "grad_norm": 0.4852455642741703,
      "learning_rate": 8.026290820063876e-05,
      "loss": 11.9551,
      "step": 21168
    },
    {
      "epoch": 1.152736653665775,
      "grad_norm": 0.5851320866655585,
      "learning_rate": 8.02542634842048e-05,
      "loss": 11.9567,
      "step": 21169
    },
    {
      "epoch": 1.152791107662358,
      "grad_norm": 0.5351084480055083,
      "learning_rate": 8.024561892131699e-05,
      "loss": 11.8094,
      "step": 21170
    },
    {
      "epoch": 1.152845561658941,
      "grad_norm": 0.5882382168396529,
      "learning_rate": 8.023697451204258e-05,
      "loss": 12.1339,
      "step": 21171
    },
    {
      "epoch": 1.152900015655524,
      "grad_norm": 0.6477972692190181,
      "learning_rate": 8.022833025644875e-05,
      "loss": 12.2011,
      "step": 21172
    },
    {
      "epoch": 1.152954469652107,
      "grad_norm": 0.5532746213673208,
      "learning_rate": 8.021968615460275e-05,
      "loss": 12.0372,
      "step": 21173
    },
    {
      "epoch": 1.15300892364869,
      "grad_norm": 0.5286103616233467,
      "learning_rate": 8.021104220657178e-05,
      "loss": 11.9014,
      "step": 21174
    },
    {
      "epoch": 1.153063377645273,
      "grad_norm": 0.57629068999438,
      "learning_rate": 8.020239841242305e-05,
      "loss": 12.0307,
      "step": 21175
    },
    {
      "epoch": 1.153117831641856,
      "grad_norm": 0.5705025655303789,
      "learning_rate": 8.019375477222386e-05,
      "loss": 12.0665,
      "step": 21176
    },
    {
      "epoch": 1.153172285638439,
      "grad_norm": 0.5119161902042083,
      "learning_rate": 8.01851112860413e-05,
      "loss": 12.0077,
      "step": 21177
    },
    {
      "epoch": 1.1532267396350222,
      "grad_norm": 0.5084787593407395,
      "learning_rate": 8.017646795394264e-05,
      "loss": 11.99,
      "step": 21178
    },
    {
      "epoch": 1.1532811936316052,
      "grad_norm": 0.5625113833150053,
      "learning_rate": 8.016782477599507e-05,
      "loss": 11.9331,
      "step": 21179
    },
    {
      "epoch": 1.1533356476281882,
      "grad_norm": 0.5683958156635728,
      "learning_rate": 8.015918175226584e-05,
      "loss": 12.0022,
      "step": 21180
    },
    {
      "epoch": 1.1533901016247712,
      "grad_norm": 0.552218824644041,
      "learning_rate": 8.015053888282215e-05,
      "loss": 11.9492,
      "step": 21181
    },
    {
      "epoch": 1.1534445556213542,
      "grad_norm": 0.538202368181314,
      "learning_rate": 8.014189616773117e-05,
      "loss": 11.9867,
      "step": 21182
    },
    {
      "epoch": 1.1534990096179372,
      "grad_norm": 0.5772816653059827,
      "learning_rate": 8.013325360706017e-05,
      "loss": 11.9636,
      "step": 21183
    },
    {
      "epoch": 1.1535534636145202,
      "grad_norm": 0.5305500868198957,
      "learning_rate": 8.012461120087631e-05,
      "loss": 11.9883,
      "step": 21184
    },
    {
      "epoch": 1.1536079176111032,
      "grad_norm": 0.6269183943377432,
      "learning_rate": 8.01159689492468e-05,
      "loss": 12.0116,
      "step": 21185
    },
    {
      "epoch": 1.1536623716076861,
      "grad_norm": 0.5043439783473882,
      "learning_rate": 8.010732685223888e-05,
      "loss": 11.9738,
      "step": 21186
    },
    {
      "epoch": 1.1537168256042691,
      "grad_norm": 0.5478230975249527,
      "learning_rate": 8.009868490991969e-05,
      "loss": 12.1364,
      "step": 21187
    },
    {
      "epoch": 1.1537712796008521,
      "grad_norm": 0.5167076642854395,
      "learning_rate": 8.009004312235648e-05,
      "loss": 11.9932,
      "step": 21188
    },
    {
      "epoch": 1.1538257335974351,
      "grad_norm": 0.591628232371505,
      "learning_rate": 8.008140148961641e-05,
      "loss": 12.0235,
      "step": 21189
    },
    {
      "epoch": 1.1538801875940181,
      "grad_norm": 0.5232239089879509,
      "learning_rate": 8.007276001176672e-05,
      "loss": 12.0338,
      "step": 21190
    },
    {
      "epoch": 1.1539346415906013,
      "grad_norm": 0.5218900289117727,
      "learning_rate": 8.006411868887456e-05,
      "loss": 12.0667,
      "step": 21191
    },
    {
      "epoch": 1.1539890955871843,
      "grad_norm": 0.5002955849688884,
      "learning_rate": 8.005547752100718e-05,
      "loss": 12.0313,
      "step": 21192
    },
    {
      "epoch": 1.1540435495837673,
      "grad_norm": 0.6418628227847089,
      "learning_rate": 8.004683650823175e-05,
      "loss": 12.0277,
      "step": 21193
    },
    {
      "epoch": 1.1540980035803503,
      "grad_norm": 0.5421520459976056,
      "learning_rate": 8.003819565061548e-05,
      "loss": 12.0242,
      "step": 21194
    },
    {
      "epoch": 1.1541524575769333,
      "grad_norm": 0.5854748086041517,
      "learning_rate": 8.002955494822553e-05,
      "loss": 11.9826,
      "step": 21195
    },
    {
      "epoch": 1.1542069115735163,
      "grad_norm": 0.5257715022701855,
      "learning_rate": 8.002091440112914e-05,
      "loss": 11.9302,
      "step": 21196
    },
    {
      "epoch": 1.1542613655700993,
      "grad_norm": 0.546327156552604,
      "learning_rate": 8.001227400939345e-05,
      "loss": 11.973,
      "step": 21197
    },
    {
      "epoch": 1.1543158195666823,
      "grad_norm": 0.594397566181094,
      "learning_rate": 8.000363377308566e-05,
      "loss": 12.0479,
      "step": 21198
    },
    {
      "epoch": 1.1543702735632653,
      "grad_norm": 0.5339239788357752,
      "learning_rate": 7.999499369227298e-05,
      "loss": 11.9646,
      "step": 21199
    },
    {
      "epoch": 1.1544247275598483,
      "grad_norm": 0.5500296085494902,
      "learning_rate": 7.998635376702257e-05,
      "loss": 11.9813,
      "step": 21200
    },
    {
      "epoch": 1.1544791815564315,
      "grad_norm": 0.5488862208409117,
      "learning_rate": 7.997771399740163e-05,
      "loss": 12.0284,
      "step": 21201
    },
    {
      "epoch": 1.1545336355530145,
      "grad_norm": 0.5354108211709937,
      "learning_rate": 7.996907438347734e-05,
      "loss": 11.8456,
      "step": 21202
    },
    {
      "epoch": 1.1545880895495975,
      "grad_norm": 0.5203275691147052,
      "learning_rate": 7.99604349253169e-05,
      "loss": 12.0175,
      "step": 21203
    },
    {
      "epoch": 1.1546425435461805,
      "grad_norm": 0.5707911160642656,
      "learning_rate": 7.995179562298746e-05,
      "loss": 11.8751,
      "step": 21204
    },
    {
      "epoch": 1.1546969975427634,
      "grad_norm": 0.5927001165041338,
      "learning_rate": 7.994315647655624e-05,
      "loss": 12.0275,
      "step": 21205
    },
    {
      "epoch": 1.1547514515393464,
      "grad_norm": 0.5277214276087671,
      "learning_rate": 7.993451748609042e-05,
      "loss": 11.9848,
      "step": 21206
    },
    {
      "epoch": 1.1548059055359294,
      "grad_norm": 0.5769471969029595,
      "learning_rate": 7.992587865165713e-05,
      "loss": 11.9488,
      "step": 21207
    },
    {
      "epoch": 1.1548603595325124,
      "grad_norm": 0.5496266310399047,
      "learning_rate": 7.991723997332358e-05,
      "loss": 12.0007,
      "step": 21208
    },
    {
      "epoch": 1.1549148135290954,
      "grad_norm": 0.6594062965515542,
      "learning_rate": 7.990860145115694e-05,
      "loss": 12.0828,
      "step": 21209
    },
    {
      "epoch": 1.1549692675256784,
      "grad_norm": 0.5677854047819902,
      "learning_rate": 7.989996308522437e-05,
      "loss": 12.0297,
      "step": 21210
    },
    {
      "epoch": 1.1550237215222614,
      "grad_norm": 0.551883355920494,
      "learning_rate": 7.989132487559307e-05,
      "loss": 12.0298,
      "step": 21211
    },
    {
      "epoch": 1.1550781755188444,
      "grad_norm": 0.5323634440159634,
      "learning_rate": 7.98826868223302e-05,
      "loss": 11.9089,
      "step": 21212
    },
    {
      "epoch": 1.1551326295154274,
      "grad_norm": 0.5037670116250631,
      "learning_rate": 7.987404892550289e-05,
      "loss": 11.9592,
      "step": 21213
    },
    {
      "epoch": 1.1551870835120106,
      "grad_norm": 0.5199204506116677,
      "learning_rate": 7.98654111851784e-05,
      "loss": 11.9533,
      "step": 21214
    },
    {
      "epoch": 1.1552415375085936,
      "grad_norm": 0.6107623890788886,
      "learning_rate": 7.985677360142384e-05,
      "loss": 12.0708,
      "step": 21215
    },
    {
      "epoch": 1.1552959915051766,
      "grad_norm": 0.5511599905622395,
      "learning_rate": 7.984813617430644e-05,
      "loss": 12.1261,
      "step": 21216
    },
    {
      "epoch": 1.1553504455017596,
      "grad_norm": 0.5115863465619226,
      "learning_rate": 7.983949890389322e-05,
      "loss": 11.9577,
      "step": 21217
    },
    {
      "epoch": 1.1554048994983426,
      "grad_norm": 0.5637422061740728,
      "learning_rate": 7.983086179025148e-05,
      "loss": 12.0071,
      "step": 21218
    },
    {
      "epoch": 1.1554593534949256,
      "grad_norm": 0.4981769207547077,
      "learning_rate": 7.982222483344834e-05,
      "loss": 11.9628,
      "step": 21219
    },
    {
      "epoch": 1.1555138074915086,
      "grad_norm": 0.5414525482918751,
      "learning_rate": 7.981358803355095e-05,
      "loss": 11.975,
      "step": 21220
    },
    {
      "epoch": 1.1555682614880916,
      "grad_norm": 0.5530668572217571,
      "learning_rate": 7.980495139062649e-05,
      "loss": 12.0245,
      "step": 21221
    },
    {
      "epoch": 1.1556227154846745,
      "grad_norm": 0.5786923404746341,
      "learning_rate": 7.979631490474213e-05,
      "loss": 11.9814,
      "step": 21222
    },
    {
      "epoch": 1.1556771694812575,
      "grad_norm": 0.6240826110107114,
      "learning_rate": 7.978767857596499e-05,
      "loss": 12.1326,
      "step": 21223
    },
    {
      "epoch": 1.1557316234778405,
      "grad_norm": 0.49542183088818065,
      "learning_rate": 7.977904240436224e-05,
      "loss": 11.7954,
      "step": 21224
    },
    {
      "epoch": 1.1557860774744237,
      "grad_norm": 0.556239010444873,
      "learning_rate": 7.977040639000107e-05,
      "loss": 12.0706,
      "step": 21225
    },
    {
      "epoch": 1.1558405314710067,
      "grad_norm": 0.5166307537336219,
      "learning_rate": 7.976177053294867e-05,
      "loss": 11.9495,
      "step": 21226
    },
    {
      "epoch": 1.1558949854675897,
      "grad_norm": 0.5862827674070573,
      "learning_rate": 7.975313483327206e-05,
      "loss": 12.0701,
      "step": 21227
    },
    {
      "epoch": 1.1559494394641727,
      "grad_norm": 0.5799460978782308,
      "learning_rate": 7.974449929103847e-05,
      "loss": 11.9343,
      "step": 21228
    },
    {
      "epoch": 1.1560038934607557,
      "grad_norm": 0.5799848983304122,
      "learning_rate": 7.973586390631508e-05,
      "loss": 11.9795,
      "step": 21229
    },
    {
      "epoch": 1.1560583474573387,
      "grad_norm": 0.5479810690267198,
      "learning_rate": 7.9727228679169e-05,
      "loss": 12.0038,
      "step": 21230
    },
    {
      "epoch": 1.1561128014539217,
      "grad_norm": 0.5268258755091737,
      "learning_rate": 7.971859360966739e-05,
      "loss": 11.9109,
      "step": 21231
    },
    {
      "epoch": 1.1561672554505047,
      "grad_norm": 0.5399960113306216,
      "learning_rate": 7.970995869787738e-05,
      "loss": 12.1343,
      "step": 21232
    },
    {
      "epoch": 1.1562217094470877,
      "grad_norm": 0.5416405261216394,
      "learning_rate": 7.970132394386616e-05,
      "loss": 11.9387,
      "step": 21233
    },
    {
      "epoch": 1.1562761634436707,
      "grad_norm": 0.4994698027199779,
      "learning_rate": 7.969268934770084e-05,
      "loss": 11.9743,
      "step": 21234
    },
    {
      "epoch": 1.1563306174402537,
      "grad_norm": 0.5605141229754358,
      "learning_rate": 7.968405490944855e-05,
      "loss": 11.9807,
      "step": 21235
    },
    {
      "epoch": 1.1563850714368367,
      "grad_norm": 0.5164187375636196,
      "learning_rate": 7.967542062917648e-05,
      "loss": 11.9675,
      "step": 21236
    },
    {
      "epoch": 1.1564395254334197,
      "grad_norm": 0.6047170965949826,
      "learning_rate": 7.966678650695179e-05,
      "loss": 12.0729,
      "step": 21237
    },
    {
      "epoch": 1.1564939794300029,
      "grad_norm": 0.5670926295472087,
      "learning_rate": 7.965815254284152e-05,
      "loss": 12.0425,
      "step": 21238
    },
    {
      "epoch": 1.1565484334265859,
      "grad_norm": 0.5591787087438296,
      "learning_rate": 7.964951873691289e-05,
      "loss": 12.0772,
      "step": 21239
    },
    {
      "epoch": 1.1566028874231689,
      "grad_norm": 0.5418566161904471,
      "learning_rate": 7.964088508923297e-05,
      "loss": 12.067,
      "step": 21240
    },
    {
      "epoch": 1.1566573414197518,
      "grad_norm": 0.60068859839871,
      "learning_rate": 7.963225159986899e-05,
      "loss": 12.1196,
      "step": 21241
    },
    {
      "epoch": 1.1567117954163348,
      "grad_norm": 0.6292034462025742,
      "learning_rate": 7.962361826888802e-05,
      "loss": 12.1183,
      "step": 21242
    },
    {
      "epoch": 1.1567662494129178,
      "grad_norm": 0.5440244217598444,
      "learning_rate": 7.961498509635722e-05,
      "loss": 11.9312,
      "step": 21243
    },
    {
      "epoch": 1.1568207034095008,
      "grad_norm": 0.6557745520530629,
      "learning_rate": 7.96063520823437e-05,
      "loss": 12.0664,
      "step": 21244
    },
    {
      "epoch": 1.1568751574060838,
      "grad_norm": 0.51055205754822,
      "learning_rate": 7.959771922691463e-05,
      "loss": 12.0452,
      "step": 21245
    },
    {
      "epoch": 1.1569296114026668,
      "grad_norm": 0.5914900864544089,
      "learning_rate": 7.95890865301371e-05,
      "loss": 11.9728,
      "step": 21246
    },
    {
      "epoch": 1.1569840653992498,
      "grad_norm": 0.5354573851902568,
      "learning_rate": 7.958045399207827e-05,
      "loss": 11.9417,
      "step": 21247
    },
    {
      "epoch": 1.157038519395833,
      "grad_norm": 0.5204337816259289,
      "learning_rate": 7.957182161280526e-05,
      "loss": 12.0514,
      "step": 21248
    },
    {
      "epoch": 1.157092973392416,
      "grad_norm": 0.5252946339072717,
      "learning_rate": 7.956318939238517e-05,
      "loss": 11.989,
      "step": 21249
    },
    {
      "epoch": 1.157147427388999,
      "grad_norm": 0.5091175709706328,
      "learning_rate": 7.955455733088516e-05,
      "loss": 12.0704,
      "step": 21250
    },
    {
      "epoch": 1.157201881385582,
      "grad_norm": 0.5174848837304137,
      "learning_rate": 7.954592542837229e-05,
      "loss": 12.0009,
      "step": 21251
    },
    {
      "epoch": 1.157256335382165,
      "grad_norm": 0.5023318740844385,
      "learning_rate": 7.953729368491378e-05,
      "loss": 11.8673,
      "step": 21252
    },
    {
      "epoch": 1.157310789378748,
      "grad_norm": 0.6140836736504344,
      "learning_rate": 7.95286621005767e-05,
      "loss": 11.9944,
      "step": 21253
    },
    {
      "epoch": 1.157365243375331,
      "grad_norm": 0.5182919984248456,
      "learning_rate": 7.952003067542818e-05,
      "loss": 12.1002,
      "step": 21254
    },
    {
      "epoch": 1.157419697371914,
      "grad_norm": 0.5287698514499525,
      "learning_rate": 7.951139940953533e-05,
      "loss": 12.0127,
      "step": 21255
    },
    {
      "epoch": 1.157474151368497,
      "grad_norm": 0.5765625554102265,
      "learning_rate": 7.950276830296527e-05,
      "loss": 12.0541,
      "step": 21256
    },
    {
      "epoch": 1.15752860536508,
      "grad_norm": 0.5937487107232046,
      "learning_rate": 7.949413735578517e-05,
      "loss": 12.071,
      "step": 21257
    },
    {
      "epoch": 1.157583059361663,
      "grad_norm": 0.5589663206974158,
      "learning_rate": 7.948550656806205e-05,
      "loss": 11.8881,
      "step": 21258
    },
    {
      "epoch": 1.157637513358246,
      "grad_norm": 0.666793727274845,
      "learning_rate": 7.947687593986308e-05,
      "loss": 12.0645,
      "step": 21259
    },
    {
      "epoch": 1.157691967354829,
      "grad_norm": 0.6111311427044318,
      "learning_rate": 7.946824547125536e-05,
      "loss": 11.9878,
      "step": 21260
    },
    {
      "epoch": 1.1577464213514121,
      "grad_norm": 0.5227747121282392,
      "learning_rate": 7.945961516230601e-05,
      "loss": 11.9656,
      "step": 21261
    },
    {
      "epoch": 1.1578008753479951,
      "grad_norm": 0.5919891131184604,
      "learning_rate": 7.94509850130821e-05,
      "loss": 11.9137,
      "step": 21262
    },
    {
      "epoch": 1.1578553293445781,
      "grad_norm": 0.5526761338976475,
      "learning_rate": 7.944235502365083e-05,
      "loss": 11.9691,
      "step": 21263
    },
    {
      "epoch": 1.1579097833411611,
      "grad_norm": 0.5191002217963598,
      "learning_rate": 7.943372519407924e-05,
      "loss": 11.9662,
      "step": 21264
    },
    {
      "epoch": 1.1579642373377441,
      "grad_norm": 0.5784555927342834,
      "learning_rate": 7.942509552443445e-05,
      "loss": 12.0409,
      "step": 21265
    },
    {
      "epoch": 1.158018691334327,
      "grad_norm": 0.5888502247943441,
      "learning_rate": 7.941646601478357e-05,
      "loss": 12.0134,
      "step": 21266
    },
    {
      "epoch": 1.15807314533091,
      "grad_norm": 0.5631901823699048,
      "learning_rate": 7.940783666519372e-05,
      "loss": 11.9675,
      "step": 21267
    },
    {
      "epoch": 1.158127599327493,
      "grad_norm": 0.5360246408255919,
      "learning_rate": 7.939920747573195e-05,
      "loss": 12.0723,
      "step": 21268
    },
    {
      "epoch": 1.158182053324076,
      "grad_norm": 0.5450568280593008,
      "learning_rate": 7.939057844646542e-05,
      "loss": 11.9873,
      "step": 21269
    },
    {
      "epoch": 1.158236507320659,
      "grad_norm": 0.7460199505110907,
      "learning_rate": 7.93819495774612e-05,
      "loss": 12.0658,
      "step": 21270
    },
    {
      "epoch": 1.1582909613172423,
      "grad_norm": 0.6202295260594721,
      "learning_rate": 7.937332086878639e-05,
      "loss": 12.0706,
      "step": 21271
    },
    {
      "epoch": 1.1583454153138253,
      "grad_norm": 0.5186259869149088,
      "learning_rate": 7.93646923205081e-05,
      "loss": 12.0008,
      "step": 21272
    },
    {
      "epoch": 1.1583998693104083,
      "grad_norm": 0.5389499880625561,
      "learning_rate": 7.935606393269341e-05,
      "loss": 12.0292,
      "step": 21273
    },
    {
      "epoch": 1.1584543233069913,
      "grad_norm": 0.6026633532668297,
      "learning_rate": 7.934743570540944e-05,
      "loss": 11.9039,
      "step": 21274
    },
    {
      "epoch": 1.1585087773035743,
      "grad_norm": 0.5740250909525888,
      "learning_rate": 7.933880763872328e-05,
      "loss": 12.0317,
      "step": 21275
    },
    {
      "epoch": 1.1585632313001573,
      "grad_norm": 0.5241001314895766,
      "learning_rate": 7.933017973270202e-05,
      "loss": 12.0434,
      "step": 21276
    },
    {
      "epoch": 1.1586176852967403,
      "grad_norm": 0.5617167280592977,
      "learning_rate": 7.932155198741276e-05,
      "loss": 12.1402,
      "step": 21277
    },
    {
      "epoch": 1.1586721392933232,
      "grad_norm": 0.5273417370388322,
      "learning_rate": 7.931292440292258e-05,
      "loss": 11.9136,
      "step": 21278
    },
    {
      "epoch": 1.1587265932899062,
      "grad_norm": 0.5187379134264419,
      "learning_rate": 7.930429697929855e-05,
      "loss": 12.0329,
      "step": 21279
    },
    {
      "epoch": 1.1587810472864892,
      "grad_norm": 0.5208629925119909,
      "learning_rate": 7.929566971660777e-05,
      "loss": 11.9753,
      "step": 21280
    },
    {
      "epoch": 1.1588355012830722,
      "grad_norm": 0.5019973934481978,
      "learning_rate": 7.928704261491735e-05,
      "loss": 11.9225,
      "step": 21281
    },
    {
      "epoch": 1.1588899552796552,
      "grad_norm": 0.5206683386613279,
      "learning_rate": 7.927841567429435e-05,
      "loss": 12.0117,
      "step": 21282
    },
    {
      "epoch": 1.1589444092762382,
      "grad_norm": 0.5106389297711984,
      "learning_rate": 7.926978889480587e-05,
      "loss": 12.0175,
      "step": 21283
    },
    {
      "epoch": 1.1589988632728214,
      "grad_norm": 0.5863089332292762,
      "learning_rate": 7.926116227651896e-05,
      "loss": 12.0695,
      "step": 21284
    },
    {
      "epoch": 1.1590533172694044,
      "grad_norm": 0.5314155203167451,
      "learning_rate": 7.925253581950077e-05,
      "loss": 11.9583,
      "step": 21285
    },
    {
      "epoch": 1.1591077712659874,
      "grad_norm": 0.5579858506104858,
      "learning_rate": 7.924390952381832e-05,
      "loss": 12.0124,
      "step": 21286
    },
    {
      "epoch": 1.1591622252625704,
      "grad_norm": 0.5483341561174498,
      "learning_rate": 7.923528338953876e-05,
      "loss": 12.0567,
      "step": 21287
    },
    {
      "epoch": 1.1592166792591534,
      "grad_norm": 0.5681173540784744,
      "learning_rate": 7.922665741672905e-05,
      "loss": 12.078,
      "step": 21288
    },
    {
      "epoch": 1.1592711332557364,
      "grad_norm": 0.5228607695356412,
      "learning_rate": 7.921803160545637e-05,
      "loss": 12.0183,
      "step": 21289
    },
    {
      "epoch": 1.1593255872523194,
      "grad_norm": 0.5857548834210604,
      "learning_rate": 7.920940595578775e-05,
      "loss": 12.0019,
      "step": 21290
    },
    {
      "epoch": 1.1593800412489024,
      "grad_norm": 0.523380359704805,
      "learning_rate": 7.920078046779028e-05,
      "loss": 12.037,
      "step": 21291
    },
    {
      "epoch": 1.1594344952454854,
      "grad_norm": 0.5778553722466274,
      "learning_rate": 7.919215514153103e-05,
      "loss": 12.0122,
      "step": 21292
    },
    {
      "epoch": 1.1594889492420684,
      "grad_norm": 0.5528023586460237,
      "learning_rate": 7.918352997707708e-05,
      "loss": 12.0816,
      "step": 21293
    },
    {
      "epoch": 1.1595434032386516,
      "grad_norm": 0.5587668613138254,
      "learning_rate": 7.917490497449547e-05,
      "loss": 11.8789,
      "step": 21294
    },
    {
      "epoch": 1.1595978572352346,
      "grad_norm": 0.555394070804435,
      "learning_rate": 7.916628013385331e-05,
      "loss": 12.0878,
      "step": 21295
    },
    {
      "epoch": 1.1596523112318176,
      "grad_norm": 0.5488890422522659,
      "learning_rate": 7.915765545521761e-05,
      "loss": 12.0984,
      "step": 21296
    },
    {
      "epoch": 1.1597067652284005,
      "grad_norm": 0.546272098096521,
      "learning_rate": 7.914903093865555e-05,
      "loss": 11.9192,
      "step": 21297
    },
    {
      "epoch": 1.1597612192249835,
      "grad_norm": 0.6543254243942339,
      "learning_rate": 7.91404065842341e-05,
      "loss": 12.0496,
      "step": 21298
    },
    {
      "epoch": 1.1598156732215665,
      "grad_norm": 0.6017006398399886,
      "learning_rate": 7.913178239202032e-05,
      "loss": 12.1064,
      "step": 21299
    },
    {
      "epoch": 1.1598701272181495,
      "grad_norm": 0.5230995730140959,
      "learning_rate": 7.912315836208128e-05,
      "loss": 12.0087,
      "step": 21300
    },
    {
      "epoch": 1.1599245812147325,
      "grad_norm": 0.5614196050577605,
      "learning_rate": 7.911453449448409e-05,
      "loss": 12.0527,
      "step": 21301
    },
    {
      "epoch": 1.1599790352113155,
      "grad_norm": 0.4539513549397306,
      "learning_rate": 7.910591078929578e-05,
      "loss": 12.0398,
      "step": 21302
    },
    {
      "epoch": 1.1600334892078985,
      "grad_norm": 0.5356462699840768,
      "learning_rate": 7.909728724658342e-05,
      "loss": 12.0782,
      "step": 21303
    },
    {
      "epoch": 1.1600879432044815,
      "grad_norm": 0.5066520142058429,
      "learning_rate": 7.908866386641404e-05,
      "loss": 11.807,
      "step": 21304
    },
    {
      "epoch": 1.1601423972010645,
      "grad_norm": 0.5668869110468674,
      "learning_rate": 7.908004064885475e-05,
      "loss": 12.0233,
      "step": 21305
    },
    {
      "epoch": 1.1601968511976475,
      "grad_norm": 0.5285924976149727,
      "learning_rate": 7.907141759397255e-05,
      "loss": 12.1045,
      "step": 21306
    },
    {
      "epoch": 1.1602513051942305,
      "grad_norm": 0.546543507842577,
      "learning_rate": 7.906279470183453e-05,
      "loss": 12.0654,
      "step": 21307
    },
    {
      "epoch": 1.1603057591908137,
      "grad_norm": 0.5385015810317186,
      "learning_rate": 7.905417197250772e-05,
      "loss": 12.1201,
      "step": 21308
    },
    {
      "epoch": 1.1603602131873967,
      "grad_norm": 0.5577951474519199,
      "learning_rate": 7.904554940605918e-05,
      "loss": 11.9945,
      "step": 21309
    },
    {
      "epoch": 1.1604146671839797,
      "grad_norm": 0.5469466081823766,
      "learning_rate": 7.903692700255596e-05,
      "loss": 12.0259,
      "step": 21310
    },
    {
      "epoch": 1.1604691211805627,
      "grad_norm": 0.5346637133607479,
      "learning_rate": 7.902830476206509e-05,
      "loss": 11.9273,
      "step": 21311
    },
    {
      "epoch": 1.1605235751771457,
      "grad_norm": 0.5918326644363558,
      "learning_rate": 7.901968268465366e-05,
      "loss": 12.0651,
      "step": 21312
    },
    {
      "epoch": 1.1605780291737287,
      "grad_norm": 0.5384818229036822,
      "learning_rate": 7.90110607703887e-05,
      "loss": 11.9657,
      "step": 21313
    },
    {
      "epoch": 1.1606324831703116,
      "grad_norm": 0.6664454169675025,
      "learning_rate": 7.900243901933726e-05,
      "loss": 12.014,
      "step": 21314
    },
    {
      "epoch": 1.1606869371668946,
      "grad_norm": 0.5937989905265902,
      "learning_rate": 7.899381743156636e-05,
      "loss": 12.1148,
      "step": 21315
    },
    {
      "epoch": 1.1607413911634776,
      "grad_norm": 0.5574320242817628,
      "learning_rate": 7.898519600714304e-05,
      "loss": 11.9253,
      "step": 21316
    },
    {
      "epoch": 1.1607958451600606,
      "grad_norm": 0.669922264965814,
      "learning_rate": 7.897657474613442e-05,
      "loss": 11.993,
      "step": 21317
    },
    {
      "epoch": 1.1608502991566438,
      "grad_norm": 0.5635964816746843,
      "learning_rate": 7.896795364860743e-05,
      "loss": 11.9512,
      "step": 21318
    },
    {
      "epoch": 1.1609047531532268,
      "grad_norm": 0.5431889182245252,
      "learning_rate": 7.895933271462919e-05,
      "loss": 12.103,
      "step": 21319
    },
    {
      "epoch": 1.1609592071498098,
      "grad_norm": 0.5465032975915614,
      "learning_rate": 7.895071194426669e-05,
      "loss": 11.995,
      "step": 21320
    },
    {
      "epoch": 1.1610136611463928,
      "grad_norm": 0.6067239504686526,
      "learning_rate": 7.894209133758698e-05,
      "loss": 12.0264,
      "step": 21321
    },
    {
      "epoch": 1.1610681151429758,
      "grad_norm": 0.5230913732754815,
      "learning_rate": 7.893347089465707e-05,
      "loss": 11.9428,
      "step": 21322
    },
    {
      "epoch": 1.1611225691395588,
      "grad_norm": 0.554245263055647,
      "learning_rate": 7.892485061554407e-05,
      "loss": 11.9681,
      "step": 21323
    },
    {
      "epoch": 1.1611770231361418,
      "grad_norm": 0.5982935869902283,
      "learning_rate": 7.891623050031495e-05,
      "loss": 11.9538,
      "step": 21324
    },
    {
      "epoch": 1.1612314771327248,
      "grad_norm": 0.6120429744966326,
      "learning_rate": 7.890761054903675e-05,
      "loss": 11.9878,
      "step": 21325
    },
    {
      "epoch": 1.1612859311293078,
      "grad_norm": 0.5260012624394336,
      "learning_rate": 7.889899076177651e-05,
      "loss": 12.005,
      "step": 21326
    },
    {
      "epoch": 1.1613403851258908,
      "grad_norm": 0.5592786523396323,
      "learning_rate": 7.88903711386013e-05,
      "loss": 11.9804,
      "step": 21327
    },
    {
      "epoch": 1.1613948391224738,
      "grad_norm": 0.641811322354292,
      "learning_rate": 7.888175167957807e-05,
      "loss": 12.0675,
      "step": 21328
    },
    {
      "epoch": 1.1614492931190568,
      "grad_norm": 0.5965510411636518,
      "learning_rate": 7.887313238477387e-05,
      "loss": 11.9095,
      "step": 21329
    },
    {
      "epoch": 1.1615037471156398,
      "grad_norm": 0.5483269925942232,
      "learning_rate": 7.886451325425574e-05,
      "loss": 12.0477,
      "step": 21330
    },
    {
      "epoch": 1.161558201112223,
      "grad_norm": 0.5454992293707681,
      "learning_rate": 7.88558942880907e-05,
      "loss": 11.9792,
      "step": 21331
    },
    {
      "epoch": 1.161612655108806,
      "grad_norm": 0.5596179934384802,
      "learning_rate": 7.884727548634578e-05,
      "loss": 12.11,
      "step": 21332
    },
    {
      "epoch": 1.161667109105389,
      "grad_norm": 0.6544838598092451,
      "learning_rate": 7.883865684908797e-05,
      "loss": 11.9894,
      "step": 21333
    },
    {
      "epoch": 1.161721563101972,
      "grad_norm": 0.5976716142108864,
      "learning_rate": 7.883003837638433e-05,
      "loss": 12.1143,
      "step": 21334
    },
    {
      "epoch": 1.161776017098555,
      "grad_norm": 0.6820796469692685,
      "learning_rate": 7.882142006830186e-05,
      "loss": 12.1575,
      "step": 21335
    },
    {
      "epoch": 1.161830471095138,
      "grad_norm": 0.5891571395628682,
      "learning_rate": 7.881280192490759e-05,
      "loss": 11.9877,
      "step": 21336
    },
    {
      "epoch": 1.161884925091721,
      "grad_norm": 0.6234461921548158,
      "learning_rate": 7.880418394626852e-05,
      "loss": 12.0753,
      "step": 21337
    },
    {
      "epoch": 1.161939379088304,
      "grad_norm": 0.5503187924707625,
      "learning_rate": 7.879556613245168e-05,
      "loss": 11.9641,
      "step": 21338
    },
    {
      "epoch": 1.161993833084887,
      "grad_norm": 0.5876566938983689,
      "learning_rate": 7.878694848352406e-05,
      "loss": 12.0456,
      "step": 21339
    },
    {
      "epoch": 1.16204828708147,
      "grad_norm": 0.5923178089296206,
      "learning_rate": 7.877833099955269e-05,
      "loss": 12.075,
      "step": 21340
    },
    {
      "epoch": 1.1621027410780531,
      "grad_norm": 0.5391073852811503,
      "learning_rate": 7.876971368060457e-05,
      "loss": 11.9976,
      "step": 21341
    },
    {
      "epoch": 1.162157195074636,
      "grad_norm": 0.5005271848541565,
      "learning_rate": 7.876109652674672e-05,
      "loss": 11.874,
      "step": 21342
    },
    {
      "epoch": 1.162211649071219,
      "grad_norm": 0.6154958135301781,
      "learning_rate": 7.875247953804615e-05,
      "loss": 12.0391,
      "step": 21343
    },
    {
      "epoch": 1.162266103067802,
      "grad_norm": 0.5388557819299863,
      "learning_rate": 7.874386271456986e-05,
      "loss": 11.933,
      "step": 21344
    },
    {
      "epoch": 1.162320557064385,
      "grad_norm": 0.576313255775912,
      "learning_rate": 7.873524605638483e-05,
      "loss": 12.0524,
      "step": 21345
    },
    {
      "epoch": 1.162375011060968,
      "grad_norm": 0.5233283920183274,
      "learning_rate": 7.872662956355812e-05,
      "loss": 12.0269,
      "step": 21346
    },
    {
      "epoch": 1.162429465057551,
      "grad_norm": 0.5535799199008398,
      "learning_rate": 7.871801323615675e-05,
      "loss": 12.0384,
      "step": 21347
    },
    {
      "epoch": 1.162483919054134,
      "grad_norm": 0.5528278765162272,
      "learning_rate": 7.870939707424762e-05,
      "loss": 12.034,
      "step": 21348
    },
    {
      "epoch": 1.162538373050717,
      "grad_norm": 0.5317526454774004,
      "learning_rate": 7.870078107789778e-05,
      "loss": 12.0472,
      "step": 21349
    },
    {
      "epoch": 1.1625928270473,
      "grad_norm": 0.5714501615963714,
      "learning_rate": 7.869216524717426e-05,
      "loss": 11.973,
      "step": 21350
    },
    {
      "epoch": 1.162647281043883,
      "grad_norm": 0.5215241127983764,
      "learning_rate": 7.868354958214404e-05,
      "loss": 11.9012,
      "step": 21351
    },
    {
      "epoch": 1.162701735040466,
      "grad_norm": 0.4966222070787855,
      "learning_rate": 7.867493408287412e-05,
      "loss": 11.972,
      "step": 21352
    },
    {
      "epoch": 1.162756189037049,
      "grad_norm": 0.580673890672549,
      "learning_rate": 7.866631874943148e-05,
      "loss": 11.9781,
      "step": 21353
    },
    {
      "epoch": 1.1628106430336322,
      "grad_norm": 0.6109334189838569,
      "learning_rate": 7.865770358188312e-05,
      "loss": 12.0469,
      "step": 21354
    },
    {
      "epoch": 1.1628650970302152,
      "grad_norm": 0.6331103900132437,
      "learning_rate": 7.864908858029604e-05,
      "loss": 11.9192,
      "step": 21355
    },
    {
      "epoch": 1.1629195510267982,
      "grad_norm": 0.5767142908888633,
      "learning_rate": 7.86404737447372e-05,
      "loss": 12.213,
      "step": 21356
    },
    {
      "epoch": 1.1629740050233812,
      "grad_norm": 0.6159866567963135,
      "learning_rate": 7.863185907527369e-05,
      "loss": 12.0352,
      "step": 21357
    },
    {
      "epoch": 1.1630284590199642,
      "grad_norm": 0.5060434680232894,
      "learning_rate": 7.862324457197237e-05,
      "loss": 11.8988,
      "step": 21358
    },
    {
      "epoch": 1.1630829130165472,
      "grad_norm": 0.660432155580146,
      "learning_rate": 7.86146302349003e-05,
      "loss": 12.1318,
      "step": 21359
    },
    {
      "epoch": 1.1631373670131302,
      "grad_norm": 0.6432927532773697,
      "learning_rate": 7.860601606412444e-05,
      "loss": 11.8871,
      "step": 21360
    },
    {
      "epoch": 1.1631918210097132,
      "grad_norm": 0.5569170269303703,
      "learning_rate": 7.859740205971178e-05,
      "loss": 12.0168,
      "step": 21361
    },
    {
      "epoch": 1.1632462750062962,
      "grad_norm": 0.5579488273407743,
      "learning_rate": 7.858878822172933e-05,
      "loss": 12.0225,
      "step": 21362
    },
    {
      "epoch": 1.1633007290028792,
      "grad_norm": 0.5476491036133607,
      "learning_rate": 7.858017455024405e-05,
      "loss": 11.8573,
      "step": 21363
    },
    {
      "epoch": 1.1633551829994624,
      "grad_norm": 0.5435103348402449,
      "learning_rate": 7.857156104532293e-05,
      "loss": 11.998,
      "step": 21364
    },
    {
      "epoch": 1.1634096369960454,
      "grad_norm": 0.5526049723870691,
      "learning_rate": 7.856294770703292e-05,
      "loss": 11.9178,
      "step": 21365
    },
    {
      "epoch": 1.1634640909926284,
      "grad_norm": 0.6471772261019824,
      "learning_rate": 7.855433453544105e-05,
      "loss": 12.1013,
      "step": 21366
    },
    {
      "epoch": 1.1635185449892114,
      "grad_norm": 0.5763607212315821,
      "learning_rate": 7.854572153061428e-05,
      "loss": 11.945,
      "step": 21367
    },
    {
      "epoch": 1.1635729989857944,
      "grad_norm": 0.5876161848674404,
      "learning_rate": 7.853710869261957e-05,
      "loss": 11.9471,
      "step": 21368
    },
    {
      "epoch": 1.1636274529823774,
      "grad_norm": 0.5490189329583168,
      "learning_rate": 7.85284960215239e-05,
      "loss": 11.9542,
      "step": 21369
    },
    {
      "epoch": 1.1636819069789603,
      "grad_norm": 0.5655948032052398,
      "learning_rate": 7.851988351739423e-05,
      "loss": 12.0385,
      "step": 21370
    },
    {
      "epoch": 1.1637363609755433,
      "grad_norm": 0.6149743281232586,
      "learning_rate": 7.851127118029753e-05,
      "loss": 11.9412,
      "step": 21371
    },
    {
      "epoch": 1.1637908149721263,
      "grad_norm": 0.5610191324150361,
      "learning_rate": 7.850265901030081e-05,
      "loss": 11.9229,
      "step": 21372
    },
    {
      "epoch": 1.1638452689687093,
      "grad_norm": 0.5474582047453633,
      "learning_rate": 7.849404700747103e-05,
      "loss": 11.9991,
      "step": 21373
    },
    {
      "epoch": 1.1638997229652923,
      "grad_norm": 0.5261407730884615,
      "learning_rate": 7.848543517187514e-05,
      "loss": 11.9588,
      "step": 21374
    },
    {
      "epoch": 1.1639541769618753,
      "grad_norm": 0.6201605935308045,
      "learning_rate": 7.847682350358012e-05,
      "loss": 12.0262,
      "step": 21375
    },
    {
      "epoch": 1.1640086309584583,
      "grad_norm": 0.5875726654817186,
      "learning_rate": 7.846821200265292e-05,
      "loss": 12.1394,
      "step": 21376
    },
    {
      "epoch": 1.1640630849550415,
      "grad_norm": 0.545933145885558,
      "learning_rate": 7.845960066916052e-05,
      "loss": 11.8759,
      "step": 21377
    },
    {
      "epoch": 1.1641175389516245,
      "grad_norm": 0.5308942247364341,
      "learning_rate": 7.845098950316991e-05,
      "loss": 12.0471,
      "step": 21378
    },
    {
      "epoch": 1.1641719929482075,
      "grad_norm": 0.5420715926627245,
      "learning_rate": 7.844237850474798e-05,
      "loss": 11.8491,
      "step": 21379
    },
    {
      "epoch": 1.1642264469447905,
      "grad_norm": 0.5368358898850524,
      "learning_rate": 7.843376767396174e-05,
      "loss": 12.0462,
      "step": 21380
    },
    {
      "epoch": 1.1642809009413735,
      "grad_norm": 0.5109476748719622,
      "learning_rate": 7.842515701087813e-05,
      "loss": 12.0705,
      "step": 21381
    },
    {
      "epoch": 1.1643353549379565,
      "grad_norm": 0.528363230321364,
      "learning_rate": 7.841654651556409e-05,
      "loss": 11.8376,
      "step": 21382
    },
    {
      "epoch": 1.1643898089345395,
      "grad_norm": 0.5139255861100533,
      "learning_rate": 7.840793618808664e-05,
      "loss": 11.8097,
      "step": 21383
    },
    {
      "epoch": 1.1644442629311225,
      "grad_norm": 0.5236125064547095,
      "learning_rate": 7.839932602851269e-05,
      "loss": 12.0464,
      "step": 21384
    },
    {
      "epoch": 1.1644987169277055,
      "grad_norm": 0.5277906866132869,
      "learning_rate": 7.839071603690922e-05,
      "loss": 12.0775,
      "step": 21385
    },
    {
      "epoch": 1.1645531709242884,
      "grad_norm": 0.5543228153366013,
      "learning_rate": 7.838210621334316e-05,
      "loss": 12.0554,
      "step": 21386
    },
    {
      "epoch": 1.1646076249208714,
      "grad_norm": 0.5684060371693637,
      "learning_rate": 7.837349655788146e-05,
      "loss": 11.8817,
      "step": 21387
    },
    {
      "epoch": 1.1646620789174547,
      "grad_norm": 0.5108695795591632,
      "learning_rate": 7.836488707059109e-05,
      "loss": 11.7814,
      "step": 21388
    },
    {
      "epoch": 1.1647165329140376,
      "grad_norm": 0.5589237735708477,
      "learning_rate": 7.835627775153899e-05,
      "loss": 11.9772,
      "step": 21389
    },
    {
      "epoch": 1.1647709869106206,
      "grad_norm": 0.5192688732247119,
      "learning_rate": 7.834766860079208e-05,
      "loss": 11.8846,
      "step": 21390
    },
    {
      "epoch": 1.1648254409072036,
      "grad_norm": 0.6108269002553003,
      "learning_rate": 7.833905961841734e-05,
      "loss": 11.9929,
      "step": 21391
    },
    {
      "epoch": 1.1648798949037866,
      "grad_norm": 0.5471500568396419,
      "learning_rate": 7.833045080448172e-05,
      "loss": 12.1352,
      "step": 21392
    },
    {
      "epoch": 1.1649343489003696,
      "grad_norm": 0.6343554048826865,
      "learning_rate": 7.832184215905211e-05,
      "loss": 11.9481,
      "step": 21393
    },
    {
      "epoch": 1.1649888028969526,
      "grad_norm": 0.5065980631943604,
      "learning_rate": 7.831323368219551e-05,
      "loss": 11.9507,
      "step": 21394
    },
    {
      "epoch": 1.1650432568935356,
      "grad_norm": 0.5531899671964808,
      "learning_rate": 7.830462537397884e-05,
      "loss": 11.8348,
      "step": 21395
    },
    {
      "epoch": 1.1650977108901186,
      "grad_norm": 0.564211332494212,
      "learning_rate": 7.829601723446904e-05,
      "loss": 11.9069,
      "step": 21396
    },
    {
      "epoch": 1.1651521648867016,
      "grad_norm": 0.508681001390616,
      "learning_rate": 7.828740926373307e-05,
      "loss": 12.025,
      "step": 21397
    },
    {
      "epoch": 1.1652066188832846,
      "grad_norm": 0.5496652300910502,
      "learning_rate": 7.827880146183785e-05,
      "loss": 11.9943,
      "step": 21398
    },
    {
      "epoch": 1.1652610728798676,
      "grad_norm": 0.5495858512184852,
      "learning_rate": 7.82701938288503e-05,
      "loss": 12.0881,
      "step": 21399
    },
    {
      "epoch": 1.1653155268764506,
      "grad_norm": 0.558278490805566,
      "learning_rate": 7.826158636483736e-05,
      "loss": 11.9001,
      "step": 21400
    },
    {
      "epoch": 1.1653699808730338,
      "grad_norm": 0.5387885611485658,
      "learning_rate": 7.825297906986597e-05,
      "loss": 12.0711,
      "step": 21401
    },
    {
      "epoch": 1.1654244348696168,
      "grad_norm": 0.4769160878218986,
      "learning_rate": 7.824437194400307e-05,
      "loss": 11.9543,
      "step": 21402
    },
    {
      "epoch": 1.1654788888661998,
      "grad_norm": 0.5122308867949037,
      "learning_rate": 7.823576498731557e-05,
      "loss": 12.0465,
      "step": 21403
    },
    {
      "epoch": 1.1655333428627828,
      "grad_norm": 0.5677880671335251,
      "learning_rate": 7.822715819987042e-05,
      "loss": 11.9772,
      "step": 21404
    },
    {
      "epoch": 1.1655877968593658,
      "grad_norm": 0.49972644574935027,
      "learning_rate": 7.821855158173453e-05,
      "loss": 12.0451,
      "step": 21405
    },
    {
      "epoch": 1.1656422508559487,
      "grad_norm": 0.5070277872774278,
      "learning_rate": 7.820994513297484e-05,
      "loss": 12.0071,
      "step": 21406
    },
    {
      "epoch": 1.1656967048525317,
      "grad_norm": 0.5061112900393233,
      "learning_rate": 7.820133885365827e-05,
      "loss": 11.9472,
      "step": 21407
    },
    {
      "epoch": 1.1657511588491147,
      "grad_norm": 0.546498177730851,
      "learning_rate": 7.819273274385179e-05,
      "loss": 12.0763,
      "step": 21408
    },
    {
      "epoch": 1.1658056128456977,
      "grad_norm": 0.5124945710238951,
      "learning_rate": 7.818412680362222e-05,
      "loss": 12.0112,
      "step": 21409
    },
    {
      "epoch": 1.1658600668422807,
      "grad_norm": 0.5119162206929523,
      "learning_rate": 7.817552103303654e-05,
      "loss": 11.9779,
      "step": 21410
    },
    {
      "epoch": 1.165914520838864,
      "grad_norm": 0.5696860492088878,
      "learning_rate": 7.81669154321617e-05,
      "loss": 11.923,
      "step": 21411
    },
    {
      "epoch": 1.165968974835447,
      "grad_norm": 0.5315858939238846,
      "learning_rate": 7.815831000106457e-05,
      "loss": 12.0212,
      "step": 21412
    },
    {
      "epoch": 1.16602342883203,
      "grad_norm": 0.5426692120256107,
      "learning_rate": 7.814970473981208e-05,
      "loss": 11.9583,
      "step": 21413
    },
    {
      "epoch": 1.166077882828613,
      "grad_norm": 0.5719014979707006,
      "learning_rate": 7.814109964847115e-05,
      "loss": 11.8939,
      "step": 21414
    },
    {
      "epoch": 1.166132336825196,
      "grad_norm": 0.5993287668117351,
      "learning_rate": 7.81324947271087e-05,
      "loss": 11.9861,
      "step": 21415
    },
    {
      "epoch": 1.166186790821779,
      "grad_norm": 0.5455525230575496,
      "learning_rate": 7.812388997579161e-05,
      "loss": 11.9907,
      "step": 21416
    },
    {
      "epoch": 1.1662412448183619,
      "grad_norm": 0.5466803659336565,
      "learning_rate": 7.811528539458686e-05,
      "loss": 11.9992,
      "step": 21417
    },
    {
      "epoch": 1.1662956988149449,
      "grad_norm": 0.5648320572273463,
      "learning_rate": 7.810668098356134e-05,
      "loss": 11.9194,
      "step": 21418
    },
    {
      "epoch": 1.1663501528115279,
      "grad_norm": 0.5427405229639615,
      "learning_rate": 7.809807674278191e-05,
      "loss": 11.7762,
      "step": 21419
    },
    {
      "epoch": 1.1664046068081109,
      "grad_norm": 0.5612093245163017,
      "learning_rate": 7.808947267231549e-05,
      "loss": 12.1771,
      "step": 21420
    },
    {
      "epoch": 1.1664590608046939,
      "grad_norm": 0.5351677144029716,
      "learning_rate": 7.808086877222903e-05,
      "loss": 11.9764,
      "step": 21421
    },
    {
      "epoch": 1.1665135148012769,
      "grad_norm": 0.6642511035229242,
      "learning_rate": 7.80722650425894e-05,
      "loss": 12.0197,
      "step": 21422
    },
    {
      "epoch": 1.1665679687978598,
      "grad_norm": 0.5569335625516744,
      "learning_rate": 7.806366148346352e-05,
      "loss": 12.0209,
      "step": 21423
    },
    {
      "epoch": 1.166622422794443,
      "grad_norm": 0.5508548278692025,
      "learning_rate": 7.805505809491828e-05,
      "loss": 11.9193,
      "step": 21424
    },
    {
      "epoch": 1.166676876791026,
      "grad_norm": 0.5410126673008115,
      "learning_rate": 7.804645487702058e-05,
      "loss": 12.0918,
      "step": 21425
    },
    {
      "epoch": 1.166731330787609,
      "grad_norm": 0.5194402193255873,
      "learning_rate": 7.803785182983735e-05,
      "loss": 12.0509,
      "step": 21426
    },
    {
      "epoch": 1.166785784784192,
      "grad_norm": 0.5470937659480506,
      "learning_rate": 7.802924895343543e-05,
      "loss": 12.0272,
      "step": 21427
    },
    {
      "epoch": 1.166840238780775,
      "grad_norm": 0.5389157927800435,
      "learning_rate": 7.802064624788183e-05,
      "loss": 12.1264,
      "step": 21428
    },
    {
      "epoch": 1.166894692777358,
      "grad_norm": 0.5848160886153482,
      "learning_rate": 7.801204371324332e-05,
      "loss": 12.0118,
      "step": 21429
    },
    {
      "epoch": 1.166949146773941,
      "grad_norm": 0.489528793970145,
      "learning_rate": 7.800344134958685e-05,
      "loss": 11.9481,
      "step": 21430
    },
    {
      "epoch": 1.167003600770524,
      "grad_norm": 0.51385469906854,
      "learning_rate": 7.799483915697928e-05,
      "loss": 11.8606,
      "step": 21431
    },
    {
      "epoch": 1.167058054767107,
      "grad_norm": 0.6091393168735681,
      "learning_rate": 7.798623713548757e-05,
      "loss": 11.9637,
      "step": 21432
    },
    {
      "epoch": 1.16711250876369,
      "grad_norm": 0.5699559473013444,
      "learning_rate": 7.797763528517855e-05,
      "loss": 12.0566,
      "step": 21433
    },
    {
      "epoch": 1.1671669627602732,
      "grad_norm": 0.4878021996624549,
      "learning_rate": 7.796903360611915e-05,
      "loss": 11.9056,
      "step": 21434
    },
    {
      "epoch": 1.1672214167568562,
      "grad_norm": 0.5638929882084674,
      "learning_rate": 7.796043209837622e-05,
      "loss": 11.765,
      "step": 21435
    },
    {
      "epoch": 1.1672758707534392,
      "grad_norm": 0.5080449196689862,
      "learning_rate": 7.795183076201669e-05,
      "loss": 11.9961,
      "step": 21436
    },
    {
      "epoch": 1.1673303247500222,
      "grad_norm": 0.5320629353679006,
      "learning_rate": 7.794322959710741e-05,
      "loss": 11.9381,
      "step": 21437
    },
    {
      "epoch": 1.1673847787466052,
      "grad_norm": 0.5284954434762772,
      "learning_rate": 7.79346286037153e-05,
      "loss": 11.8977,
      "step": 21438
    },
    {
      "epoch": 1.1674392327431882,
      "grad_norm": 0.5356057188016835,
      "learning_rate": 7.792602778190717e-05,
      "loss": 11.9503,
      "step": 21439
    },
    {
      "epoch": 1.1674936867397712,
      "grad_norm": 0.5816844473371946,
      "learning_rate": 7.791742713174998e-05,
      "loss": 11.9861,
      "step": 21440
    },
    {
      "epoch": 1.1675481407363542,
      "grad_norm": 0.5891457658875092,
      "learning_rate": 7.790882665331057e-05,
      "loss": 12.0884,
      "step": 21441
    },
    {
      "epoch": 1.1676025947329371,
      "grad_norm": 0.5551175021627227,
      "learning_rate": 7.790022634665584e-05,
      "loss": 11.8205,
      "step": 21442
    },
    {
      "epoch": 1.1676570487295201,
      "grad_norm": 0.5438903206830921,
      "learning_rate": 7.789162621185263e-05,
      "loss": 12.1324,
      "step": 21443
    },
    {
      "epoch": 1.1677115027261031,
      "grad_norm": 0.5401067958423126,
      "learning_rate": 7.788302624896784e-05,
      "loss": 11.8797,
      "step": 21444
    },
    {
      "epoch": 1.1677659567226861,
      "grad_norm": 0.5592458681040414,
      "learning_rate": 7.787442645806837e-05,
      "loss": 11.9851,
      "step": 21445
    },
    {
      "epoch": 1.1678204107192691,
      "grad_norm": 0.575958398770358,
      "learning_rate": 7.786582683922107e-05,
      "loss": 11.9914,
      "step": 21446
    },
    {
      "epoch": 1.1678748647158523,
      "grad_norm": 0.5036211915335588,
      "learning_rate": 7.78572273924928e-05,
      "loss": 11.9179,
      "step": 21447
    },
    {
      "epoch": 1.1679293187124353,
      "grad_norm": 0.5415017112148952,
      "learning_rate": 7.784862811795048e-05,
      "loss": 12.0637,
      "step": 21448
    },
    {
      "epoch": 1.1679837727090183,
      "grad_norm": 0.6061407429736834,
      "learning_rate": 7.784002901566091e-05,
      "loss": 12.0024,
      "step": 21449
    },
    {
      "epoch": 1.1680382267056013,
      "grad_norm": 0.6181644418935153,
      "learning_rate": 7.783143008569099e-05,
      "loss": 12.0362,
      "step": 21450
    },
    {
      "epoch": 1.1680926807021843,
      "grad_norm": 0.5514534086004671,
      "learning_rate": 7.782283132810759e-05,
      "loss": 12.0524,
      "step": 21451
    },
    {
      "epoch": 1.1681471346987673,
      "grad_norm": 0.5047071015530673,
      "learning_rate": 7.781423274297757e-05,
      "loss": 11.9453,
      "step": 21452
    },
    {
      "epoch": 1.1682015886953503,
      "grad_norm": 0.52471953065119,
      "learning_rate": 7.78056343303678e-05,
      "loss": 11.9401,
      "step": 21453
    },
    {
      "epoch": 1.1682560426919333,
      "grad_norm": 0.6706785428933535,
      "learning_rate": 7.779703609034514e-05,
      "loss": 11.9634,
      "step": 21454
    },
    {
      "epoch": 1.1683104966885163,
      "grad_norm": 0.5778686607698854,
      "learning_rate": 7.778843802297645e-05,
      "loss": 12.1274,
      "step": 21455
    },
    {
      "epoch": 1.1683649506850993,
      "grad_norm": 0.5590734244778683,
      "learning_rate": 7.777984012832859e-05,
      "loss": 11.9344,
      "step": 21456
    },
    {
      "epoch": 1.1684194046816823,
      "grad_norm": 0.5578065982592294,
      "learning_rate": 7.777124240646842e-05,
      "loss": 12.0206,
      "step": 21457
    },
    {
      "epoch": 1.1684738586782655,
      "grad_norm": 0.5428385181545224,
      "learning_rate": 7.77626448574628e-05,
      "loss": 12.0342,
      "step": 21458
    },
    {
      "epoch": 1.1685283126748485,
      "grad_norm": 0.6147564819772413,
      "learning_rate": 7.77540474813786e-05,
      "loss": 12.1248,
      "step": 21459
    },
    {
      "epoch": 1.1685827666714315,
      "grad_norm": 0.5909814840434413,
      "learning_rate": 7.774545027828265e-05,
      "loss": 11.9908,
      "step": 21460
    },
    {
      "epoch": 1.1686372206680145,
      "grad_norm": 0.5652429534110385,
      "learning_rate": 7.773685324824178e-05,
      "loss": 11.8524,
      "step": 21461
    },
    {
      "epoch": 1.1686916746645974,
      "grad_norm": 0.6087531963234392,
      "learning_rate": 7.77282563913229e-05,
      "loss": 11.9695,
      "step": 21462
    },
    {
      "epoch": 1.1687461286611804,
      "grad_norm": 0.5512132667893598,
      "learning_rate": 7.771965970759281e-05,
      "loss": 12.0036,
      "step": 21463
    },
    {
      "epoch": 1.1688005826577634,
      "grad_norm": 0.5260982059537934,
      "learning_rate": 7.771106319711841e-05,
      "loss": 11.9306,
      "step": 21464
    },
    {
      "epoch": 1.1688550366543464,
      "grad_norm": 0.579546662874138,
      "learning_rate": 7.770246685996648e-05,
      "loss": 11.9179,
      "step": 21465
    },
    {
      "epoch": 1.1689094906509294,
      "grad_norm": 0.5552546566089984,
      "learning_rate": 7.769387069620394e-05,
      "loss": 12.0308,
      "step": 21466
    },
    {
      "epoch": 1.1689639446475124,
      "grad_norm": 0.5320251798810329,
      "learning_rate": 7.76852747058976e-05,
      "loss": 11.9921,
      "step": 21467
    },
    {
      "epoch": 1.1690183986440954,
      "grad_norm": 0.5515518523649794,
      "learning_rate": 7.767667888911434e-05,
      "loss": 12.0567,
      "step": 21468
    },
    {
      "epoch": 1.1690728526406784,
      "grad_norm": 0.5240191542993571,
      "learning_rate": 7.766808324592091e-05,
      "loss": 11.94,
      "step": 21469
    },
    {
      "epoch": 1.1691273066372614,
      "grad_norm": 0.5596577161314998,
      "learning_rate": 7.765948777638423e-05,
      "loss": 11.9289,
      "step": 21470
    },
    {
      "epoch": 1.1691817606338446,
      "grad_norm": 0.5554941562208867,
      "learning_rate": 7.765089248057114e-05,
      "loss": 11.9961,
      "step": 21471
    },
    {
      "epoch": 1.1692362146304276,
      "grad_norm": 0.5683779647278253,
      "learning_rate": 7.764229735854843e-05,
      "loss": 12.1373,
      "step": 21472
    },
    {
      "epoch": 1.1692906686270106,
      "grad_norm": 0.5305858502239906,
      "learning_rate": 7.763370241038297e-05,
      "loss": 12.0558,
      "step": 21473
    },
    {
      "epoch": 1.1693451226235936,
      "grad_norm": 0.5020052882068247,
      "learning_rate": 7.762510763614161e-05,
      "loss": 11.8852,
      "step": 21474
    },
    {
      "epoch": 1.1693995766201766,
      "grad_norm": 0.5547686098086152,
      "learning_rate": 7.761651303589114e-05,
      "loss": 11.9912,
      "step": 21475
    },
    {
      "epoch": 1.1694540306167596,
      "grad_norm": 0.5308770890161167,
      "learning_rate": 7.76079186096984e-05,
      "loss": 11.9896,
      "step": 21476
    },
    {
      "epoch": 1.1695084846133426,
      "grad_norm": 0.5203878529266085,
      "learning_rate": 7.759932435763027e-05,
      "loss": 11.8653,
      "step": 21477
    },
    {
      "epoch": 1.1695629386099256,
      "grad_norm": 0.5600808721754098,
      "learning_rate": 7.75907302797536e-05,
      "loss": 12.0303,
      "step": 21478
    },
    {
      "epoch": 1.1696173926065085,
      "grad_norm": 0.5237632925496599,
      "learning_rate": 7.758213637613512e-05,
      "loss": 11.7995,
      "step": 21479
    },
    {
      "epoch": 1.1696718466030915,
      "grad_norm": 0.5309656644064694,
      "learning_rate": 7.757354264684168e-05,
      "loss": 12.0239,
      "step": 21480
    },
    {
      "epoch": 1.1697263005996748,
      "grad_norm": 0.6012737482998177,
      "learning_rate": 7.756494909194016e-05,
      "loss": 12.0147,
      "step": 21481
    },
    {
      "epoch": 1.1697807545962577,
      "grad_norm": 0.5278493531239027,
      "learning_rate": 7.755635571149735e-05,
      "loss": 11.9673,
      "step": 21482
    },
    {
      "epoch": 1.1698352085928407,
      "grad_norm": 0.6371868709985388,
      "learning_rate": 7.75477625055801e-05,
      "loss": 11.9891,
      "step": 21483
    },
    {
      "epoch": 1.1698896625894237,
      "grad_norm": 0.5384168711632088,
      "learning_rate": 7.75391694742552e-05,
      "loss": 11.9979,
      "step": 21484
    },
    {
      "epoch": 1.1699441165860067,
      "grad_norm": 0.7231925061144907,
      "learning_rate": 7.753057661758949e-05,
      "loss": 11.9619,
      "step": 21485
    },
    {
      "epoch": 1.1699985705825897,
      "grad_norm": 0.6527073810977893,
      "learning_rate": 7.752198393564977e-05,
      "loss": 11.9728,
      "step": 21486
    },
    {
      "epoch": 1.1700530245791727,
      "grad_norm": 0.5188787105760849,
      "learning_rate": 7.751339142850288e-05,
      "loss": 12.0277,
      "step": 21487
    },
    {
      "epoch": 1.1701074785757557,
      "grad_norm": 0.4975516801920765,
      "learning_rate": 7.750479909621568e-05,
      "loss": 12.0173,
      "step": 21488
    },
    {
      "epoch": 1.1701619325723387,
      "grad_norm": 0.5568290775559681,
      "learning_rate": 7.749620693885489e-05,
      "loss": 11.9154,
      "step": 21489
    },
    {
      "epoch": 1.1702163865689217,
      "grad_norm": 0.592378744097072,
      "learning_rate": 7.748761495648736e-05,
      "loss": 11.9838,
      "step": 21490
    },
    {
      "epoch": 1.1702708405655047,
      "grad_norm": 0.5392249836439629,
      "learning_rate": 7.74790231491799e-05,
      "loss": 11.8422,
      "step": 21491
    },
    {
      "epoch": 1.1703252945620877,
      "grad_norm": 0.5828539964433967,
      "learning_rate": 7.747043151699935e-05,
      "loss": 12.0813,
      "step": 21492
    },
    {
      "epoch": 1.1703797485586707,
      "grad_norm": 0.5842192642892001,
      "learning_rate": 7.74618400600125e-05,
      "loss": 12.0662,
      "step": 21493
    },
    {
      "epoch": 1.1704342025552539,
      "grad_norm": 0.49210379808848115,
      "learning_rate": 7.745324877828617e-05,
      "loss": 11.9989,
      "step": 21494
    },
    {
      "epoch": 1.1704886565518369,
      "grad_norm": 0.5979440408864463,
      "learning_rate": 7.744465767188715e-05,
      "loss": 11.8663,
      "step": 21495
    },
    {
      "epoch": 1.1705431105484199,
      "grad_norm": 0.5539942505207179,
      "learning_rate": 7.743606674088227e-05,
      "loss": 12.0314,
      "step": 21496
    },
    {
      "epoch": 1.1705975645450029,
      "grad_norm": 0.548218170252454,
      "learning_rate": 7.74274759853383e-05,
      "loss": 12.1344,
      "step": 21497
    },
    {
      "epoch": 1.1706520185415858,
      "grad_norm": 0.5625011843254378,
      "learning_rate": 7.74188854053221e-05,
      "loss": 12.1762,
      "step": 21498
    },
    {
      "epoch": 1.1707064725381688,
      "grad_norm": 0.5460425827148898,
      "learning_rate": 7.74102950009004e-05,
      "loss": 12.0292,
      "step": 21499
    },
    {
      "epoch": 1.1707609265347518,
      "grad_norm": 0.5408142850430533,
      "learning_rate": 7.740170477214003e-05,
      "loss": 12.0924,
      "step": 21500
    },
    {
      "epoch": 1.1708153805313348,
      "grad_norm": 0.5781869069480234,
      "learning_rate": 7.739311471910781e-05,
      "loss": 12.073,
      "step": 21501
    },
    {
      "epoch": 1.1708698345279178,
      "grad_norm": 0.5392614362905274,
      "learning_rate": 7.738452484187052e-05,
      "loss": 11.9822,
      "step": 21502
    },
    {
      "epoch": 1.1709242885245008,
      "grad_norm": 0.5686113171382908,
      "learning_rate": 7.737593514049492e-05,
      "loss": 11.9301,
      "step": 21503
    },
    {
      "epoch": 1.170978742521084,
      "grad_norm": 0.6093378513520353,
      "learning_rate": 7.736734561504787e-05,
      "loss": 11.9898,
      "step": 21504
    },
    {
      "epoch": 1.171033196517667,
      "grad_norm": 0.5521503110204157,
      "learning_rate": 7.735875626559614e-05,
      "loss": 12.0431,
      "step": 21505
    },
    {
      "epoch": 1.17108765051425,
      "grad_norm": 0.5503166172575223,
      "learning_rate": 7.735016709220652e-05,
      "loss": 12.0454,
      "step": 21506
    },
    {
      "epoch": 1.171142104510833,
      "grad_norm": 0.5399224442906748,
      "learning_rate": 7.73415780949458e-05,
      "loss": 11.8754,
      "step": 21507
    },
    {
      "epoch": 1.171196558507416,
      "grad_norm": 0.6172229676242614,
      "learning_rate": 7.733298927388077e-05,
      "loss": 12.1151,
      "step": 21508
    },
    {
      "epoch": 1.171251012503999,
      "grad_norm": 0.5221733846267943,
      "learning_rate": 7.732440062907822e-05,
      "loss": 12.1128,
      "step": 21509
    },
    {
      "epoch": 1.171305466500582,
      "grad_norm": 0.5535158775303224,
      "learning_rate": 7.731581216060492e-05,
      "loss": 11.9696,
      "step": 21510
    },
    {
      "epoch": 1.171359920497165,
      "grad_norm": 0.5784357095092382,
      "learning_rate": 7.730722386852768e-05,
      "loss": 12.0313,
      "step": 21511
    },
    {
      "epoch": 1.171414374493748,
      "grad_norm": 0.5411237223474381,
      "learning_rate": 7.729863575291326e-05,
      "loss": 11.9306,
      "step": 21512
    },
    {
      "epoch": 1.171468828490331,
      "grad_norm": 0.5534941810446697,
      "learning_rate": 7.729004781382847e-05,
      "loss": 11.8688,
      "step": 21513
    },
    {
      "epoch": 1.171523282486914,
      "grad_norm": 0.5361318082877433,
      "learning_rate": 7.728146005134005e-05,
      "loss": 12.0888,
      "step": 21514
    },
    {
      "epoch": 1.171577736483497,
      "grad_norm": 0.5330867725931275,
      "learning_rate": 7.727287246551482e-05,
      "loss": 11.9465,
      "step": 21515
    },
    {
      "epoch": 1.17163219048008,
      "grad_norm": 0.6008539322265298,
      "learning_rate": 7.726428505641955e-05,
      "loss": 12.0014,
      "step": 21516
    },
    {
      "epoch": 1.1716866444766632,
      "grad_norm": 0.5886242340830706,
      "learning_rate": 7.725569782412102e-05,
      "loss": 11.9946,
      "step": 21517
    },
    {
      "epoch": 1.1717410984732461,
      "grad_norm": 0.5120914005227445,
      "learning_rate": 7.7247110768686e-05,
      "loss": 11.9638,
      "step": 21518
    },
    {
      "epoch": 1.1717955524698291,
      "grad_norm": 0.5123307090634064,
      "learning_rate": 7.723852389018126e-05,
      "loss": 11.9975,
      "step": 21519
    },
    {
      "epoch": 1.1718500064664121,
      "grad_norm": 0.5193406394637023,
      "learning_rate": 7.722993718867357e-05,
      "loss": 11.958,
      "step": 21520
    },
    {
      "epoch": 1.1719044604629951,
      "grad_norm": 0.5594636417028016,
      "learning_rate": 7.72213506642297e-05,
      "loss": 11.9261,
      "step": 21521
    },
    {
      "epoch": 1.1719589144595781,
      "grad_norm": 0.5427898649075381,
      "learning_rate": 7.721276431691643e-05,
      "loss": 12.0003,
      "step": 21522
    },
    {
      "epoch": 1.172013368456161,
      "grad_norm": 0.5094372058973505,
      "learning_rate": 7.720417814680052e-05,
      "loss": 12.0733,
      "step": 21523
    },
    {
      "epoch": 1.172067822452744,
      "grad_norm": 0.5052121062928325,
      "learning_rate": 7.719559215394875e-05,
      "loss": 12.0188,
      "step": 21524
    },
    {
      "epoch": 1.172122276449327,
      "grad_norm": 0.5406966806318048,
      "learning_rate": 7.718700633842787e-05,
      "loss": 11.9344,
      "step": 21525
    },
    {
      "epoch": 1.17217673044591,
      "grad_norm": 0.5726195194524218,
      "learning_rate": 7.717842070030467e-05,
      "loss": 11.9564,
      "step": 21526
    },
    {
      "epoch": 1.172231184442493,
      "grad_norm": 0.49501450545394016,
      "learning_rate": 7.71698352396459e-05,
      "loss": 11.8853,
      "step": 21527
    },
    {
      "epoch": 1.1722856384390763,
      "grad_norm": 0.5695613901911343,
      "learning_rate": 7.71612499565183e-05,
      "loss": 12.0181,
      "step": 21528
    },
    {
      "epoch": 1.1723400924356593,
      "grad_norm": 0.537949886667884,
      "learning_rate": 7.715266485098868e-05,
      "loss": 11.9117,
      "step": 21529
    },
    {
      "epoch": 1.1723945464322423,
      "grad_norm": 0.5456325926834007,
      "learning_rate": 7.714407992312376e-05,
      "loss": 11.9595,
      "step": 21530
    },
    {
      "epoch": 1.1724490004288253,
      "grad_norm": 0.600319765366184,
      "learning_rate": 7.71354951729903e-05,
      "loss": 12.0551,
      "step": 21531
    },
    {
      "epoch": 1.1725034544254083,
      "grad_norm": 0.6105793578343066,
      "learning_rate": 7.712691060065507e-05,
      "loss": 11.9822,
      "step": 21532
    },
    {
      "epoch": 1.1725579084219913,
      "grad_norm": 0.505282344779516,
      "learning_rate": 7.711832620618482e-05,
      "loss": 11.9356,
      "step": 21533
    },
    {
      "epoch": 1.1726123624185743,
      "grad_norm": 0.5079079065072811,
      "learning_rate": 7.710974198964629e-05,
      "loss": 11.9677,
      "step": 21534
    },
    {
      "epoch": 1.1726668164151572,
      "grad_norm": 0.5928270465117728,
      "learning_rate": 7.710115795110625e-05,
      "loss": 11.9835,
      "step": 21535
    },
    {
      "epoch": 1.1727212704117402,
      "grad_norm": 0.5729532760680833,
      "learning_rate": 7.709257409063142e-05,
      "loss": 11.918,
      "step": 21536
    },
    {
      "epoch": 1.1727757244083232,
      "grad_norm": 0.6133392232219819,
      "learning_rate": 7.70839904082886e-05,
      "loss": 12.0699,
      "step": 21537
    },
    {
      "epoch": 1.1728301784049062,
      "grad_norm": 0.5242315398613062,
      "learning_rate": 7.707540690414452e-05,
      "loss": 12.1721,
      "step": 21538
    },
    {
      "epoch": 1.1728846324014892,
      "grad_norm": 0.48245773539151876,
      "learning_rate": 7.706682357826595e-05,
      "loss": 11.9284,
      "step": 21539
    },
    {
      "epoch": 1.1729390863980722,
      "grad_norm": 0.5321526831624755,
      "learning_rate": 7.705824043071957e-05,
      "loss": 12.0253,
      "step": 21540
    },
    {
      "epoch": 1.1729935403946554,
      "grad_norm": 0.5277659031414554,
      "learning_rate": 7.704965746157215e-05,
      "loss": 11.9888,
      "step": 21541
    },
    {
      "epoch": 1.1730479943912384,
      "grad_norm": 0.5525367153991201,
      "learning_rate": 7.704107467089045e-05,
      "loss": 11.9003,
      "step": 21542
    },
    {
      "epoch": 1.1731024483878214,
      "grad_norm": 0.5292540902969628,
      "learning_rate": 7.703249205874121e-05,
      "loss": 11.7963,
      "step": 21543
    },
    {
      "epoch": 1.1731569023844044,
      "grad_norm": 0.4988636316086766,
      "learning_rate": 7.702390962519117e-05,
      "loss": 11.9743,
      "step": 21544
    },
    {
      "epoch": 1.1732113563809874,
      "grad_norm": 0.5287904304911054,
      "learning_rate": 7.701532737030706e-05,
      "loss": 11.9094,
      "step": 21545
    },
    {
      "epoch": 1.1732658103775704,
      "grad_norm": 0.5412664155879475,
      "learning_rate": 7.70067452941556e-05,
      "loss": 11.9634,
      "step": 21546
    },
    {
      "epoch": 1.1733202643741534,
      "grad_norm": 0.5867668093510661,
      "learning_rate": 7.699816339680357e-05,
      "loss": 12.0047,
      "step": 21547
    },
    {
      "epoch": 1.1733747183707364,
      "grad_norm": 0.5837210487251758,
      "learning_rate": 7.698958167831763e-05,
      "loss": 12.0087,
      "step": 21548
    },
    {
      "epoch": 1.1734291723673194,
      "grad_norm": 0.5097681136664706,
      "learning_rate": 7.698100013876465e-05,
      "loss": 12.0246,
      "step": 21549
    },
    {
      "epoch": 1.1734836263639024,
      "grad_norm": 0.5253903708182228,
      "learning_rate": 7.697241877821121e-05,
      "loss": 11.9348,
      "step": 21550
    },
    {
      "epoch": 1.1735380803604856,
      "grad_norm": 0.6293028844089644,
      "learning_rate": 7.696383759672412e-05,
      "loss": 11.9706,
      "step": 21551
    },
    {
      "epoch": 1.1735925343570686,
      "grad_norm": 0.5576519320191654,
      "learning_rate": 7.695525659437006e-05,
      "loss": 12.0808,
      "step": 21552
    },
    {
      "epoch": 1.1736469883536516,
      "grad_norm": 0.5181540665008444,
      "learning_rate": 7.694667577121582e-05,
      "loss": 12.0329,
      "step": 21553
    },
    {
      "epoch": 1.1737014423502345,
      "grad_norm": 0.5389763408112704,
      "learning_rate": 7.69380951273281e-05,
      "loss": 12.0049,
      "step": 21554
    },
    {
      "epoch": 1.1737558963468175,
      "grad_norm": 0.5413564993798322,
      "learning_rate": 7.69295146627736e-05,
      "loss": 12.0269,
      "step": 21555
    },
    {
      "epoch": 1.1738103503434005,
      "grad_norm": 0.7198102054487848,
      "learning_rate": 7.692093437761908e-05,
      "loss": 12.1316,
      "step": 21556
    },
    {
      "epoch": 1.1738648043399835,
      "grad_norm": 0.5786289001723626,
      "learning_rate": 7.691235427193123e-05,
      "loss": 12.0126,
      "step": 21557
    },
    {
      "epoch": 1.1739192583365665,
      "grad_norm": 0.513170917273,
      "learning_rate": 7.690377434577681e-05,
      "loss": 11.8932,
      "step": 21558
    },
    {
      "epoch": 1.1739737123331495,
      "grad_norm": 0.5374406423491991,
      "learning_rate": 7.68951945992225e-05,
      "loss": 12.0512,
      "step": 21559
    },
    {
      "epoch": 1.1740281663297325,
      "grad_norm": 0.5724407425139194,
      "learning_rate": 7.688661503233503e-05,
      "loss": 12.0738,
      "step": 21560
    },
    {
      "epoch": 1.1740826203263155,
      "grad_norm": 0.5209037313444472,
      "learning_rate": 7.687803564518112e-05,
      "loss": 11.9837,
      "step": 21561
    },
    {
      "epoch": 1.1741370743228985,
      "grad_norm": 0.5103557591440533,
      "learning_rate": 7.686945643782747e-05,
      "loss": 11.9613,
      "step": 21562
    },
    {
      "epoch": 1.1741915283194815,
      "grad_norm": 0.5317941678247606,
      "learning_rate": 7.68608774103408e-05,
      "loss": 11.9773,
      "step": 21563
    },
    {
      "epoch": 1.1742459823160647,
      "grad_norm": 0.5251832866116968,
      "learning_rate": 7.685229856278784e-05,
      "loss": 11.9801,
      "step": 21564
    },
    {
      "epoch": 1.1743004363126477,
      "grad_norm": 0.5508580766149924,
      "learning_rate": 7.684371989523528e-05,
      "loss": 12.0116,
      "step": 21565
    },
    {
      "epoch": 1.1743548903092307,
      "grad_norm": 0.541710490167256,
      "learning_rate": 7.683514140774985e-05,
      "loss": 12.0006,
      "step": 21566
    },
    {
      "epoch": 1.1744093443058137,
      "grad_norm": 0.5776584803130658,
      "learning_rate": 7.682656310039826e-05,
      "loss": 11.9537,
      "step": 21567
    },
    {
      "epoch": 1.1744637983023967,
      "grad_norm": 0.5629310797797762,
      "learning_rate": 7.681798497324716e-05,
      "loss": 11.9284,
      "step": 21568
    },
    {
      "epoch": 1.1745182522989797,
      "grad_norm": 0.5587841189061926,
      "learning_rate": 7.680940702636335e-05,
      "loss": 11.9966,
      "step": 21569
    },
    {
      "epoch": 1.1745727062955627,
      "grad_norm": 0.6449658253645129,
      "learning_rate": 7.680082925981346e-05,
      "loss": 12.0528,
      "step": 21570
    },
    {
      "epoch": 1.1746271602921456,
      "grad_norm": 0.5749180238364583,
      "learning_rate": 7.67922516736642e-05,
      "loss": 12.0674,
      "step": 21571
    },
    {
      "epoch": 1.1746816142887286,
      "grad_norm": 0.6287072461563051,
      "learning_rate": 7.678367426798228e-05,
      "loss": 12.0589,
      "step": 21572
    },
    {
      "epoch": 1.1747360682853116,
      "grad_norm": 0.6296938676638881,
      "learning_rate": 7.67750970428344e-05,
      "loss": 12.1065,
      "step": 21573
    },
    {
      "epoch": 1.1747905222818948,
      "grad_norm": 0.5412884507051381,
      "learning_rate": 7.676651999828726e-05,
      "loss": 11.9356,
      "step": 21574
    },
    {
      "epoch": 1.1748449762784778,
      "grad_norm": 0.5511691378546619,
      "learning_rate": 7.675794313440756e-05,
      "loss": 11.9952,
      "step": 21575
    },
    {
      "epoch": 1.1748994302750608,
      "grad_norm": 0.6065834705687223,
      "learning_rate": 7.6749366451262e-05,
      "loss": 11.9431,
      "step": 21576
    },
    {
      "epoch": 1.1749538842716438,
      "grad_norm": 0.5623202627458568,
      "learning_rate": 7.674078994891727e-05,
      "loss": 12.0089,
      "step": 21577
    },
    {
      "epoch": 1.1750083382682268,
      "grad_norm": 0.5780561703387903,
      "learning_rate": 7.673221362744005e-05,
      "loss": 11.9446,
      "step": 21578
    },
    {
      "epoch": 1.1750627922648098,
      "grad_norm": 0.502136544401422,
      "learning_rate": 7.672363748689706e-05,
      "loss": 11.9293,
      "step": 21579
    },
    {
      "epoch": 1.1751172462613928,
      "grad_norm": 0.494545000401578,
      "learning_rate": 7.671506152735495e-05,
      "loss": 11.9548,
      "step": 21580
    },
    {
      "epoch": 1.1751717002579758,
      "grad_norm": 0.5151063631831978,
      "learning_rate": 7.670648574888042e-05,
      "loss": 11.9267,
      "step": 21581
    },
    {
      "epoch": 1.1752261542545588,
      "grad_norm": 0.6079219359619696,
      "learning_rate": 7.669791015154017e-05,
      "loss": 11.9529,
      "step": 21582
    },
    {
      "epoch": 1.1752806082511418,
      "grad_norm": 0.5665895568151611,
      "learning_rate": 7.668933473540087e-05,
      "loss": 11.9499,
      "step": 21583
    },
    {
      "epoch": 1.1753350622477248,
      "grad_norm": 0.5909458742546837,
      "learning_rate": 7.668075950052922e-05,
      "loss": 12.0102,
      "step": 21584
    },
    {
      "epoch": 1.1753895162443078,
      "grad_norm": 0.5969256567130299,
      "learning_rate": 7.667218444699187e-05,
      "loss": 11.8862,
      "step": 21585
    },
    {
      "epoch": 1.1754439702408908,
      "grad_norm": 0.5237406253917345,
      "learning_rate": 7.666360957485554e-05,
      "loss": 11.9454,
      "step": 21586
    },
    {
      "epoch": 1.175498424237474,
      "grad_norm": 0.5921875596250283,
      "learning_rate": 7.66550348841869e-05,
      "loss": 12.0029,
      "step": 21587
    },
    {
      "epoch": 1.175552878234057,
      "grad_norm": 0.5465794547164644,
      "learning_rate": 7.664646037505263e-05,
      "loss": 11.8206,
      "step": 21588
    },
    {
      "epoch": 1.17560733223064,
      "grad_norm": 0.5875587815375578,
      "learning_rate": 7.663788604751943e-05,
      "loss": 12.0814,
      "step": 21589
    },
    {
      "epoch": 1.175661786227223,
      "grad_norm": 0.5642222664525203,
      "learning_rate": 7.66293119016539e-05,
      "loss": 12.0113,
      "step": 21590
    },
    {
      "epoch": 1.175716240223806,
      "grad_norm": 0.5204404510690026,
      "learning_rate": 7.662073793752278e-05,
      "loss": 11.9023,
      "step": 21591
    },
    {
      "epoch": 1.175770694220389,
      "grad_norm": 0.5556397579875381,
      "learning_rate": 7.661216415519273e-05,
      "loss": 11.9804,
      "step": 21592
    },
    {
      "epoch": 1.175825148216972,
      "grad_norm": 0.5640496495315787,
      "learning_rate": 7.660359055473039e-05,
      "loss": 12.0748,
      "step": 21593
    },
    {
      "epoch": 1.175879602213555,
      "grad_norm": 0.49745883549555725,
      "learning_rate": 7.659501713620246e-05,
      "loss": 11.9385,
      "step": 21594
    },
    {
      "epoch": 1.175934056210138,
      "grad_norm": 0.5404957254428031,
      "learning_rate": 7.65864438996756e-05,
      "loss": 11.916,
      "step": 21595
    },
    {
      "epoch": 1.175988510206721,
      "grad_norm": 0.5242902139557931,
      "learning_rate": 7.657787084521649e-05,
      "loss": 12.1823,
      "step": 21596
    },
    {
      "epoch": 1.1760429642033041,
      "grad_norm": 0.5657329966808164,
      "learning_rate": 7.656929797289177e-05,
      "loss": 12.0118,
      "step": 21597
    },
    {
      "epoch": 1.1760974181998871,
      "grad_norm": 0.5700382513723844,
      "learning_rate": 7.656072528276811e-05,
      "loss": 12.0433,
      "step": 21598
    },
    {
      "epoch": 1.17615187219647,
      "grad_norm": 0.5633965987301115,
      "learning_rate": 7.655215277491225e-05,
      "loss": 12.0989,
      "step": 21599
    },
    {
      "epoch": 1.176206326193053,
      "grad_norm": 0.5158507634434333,
      "learning_rate": 7.654358044939073e-05,
      "loss": 12.009,
      "step": 21600
    },
    {
      "epoch": 1.176260780189636,
      "grad_norm": 0.5670445956130047,
      "learning_rate": 7.653500830627023e-05,
      "loss": 12.0958,
      "step": 21601
    },
    {
      "epoch": 1.176315234186219,
      "grad_norm": 0.5698455091294612,
      "learning_rate": 7.652643634561748e-05,
      "loss": 12.001,
      "step": 21602
    },
    {
      "epoch": 1.176369688182802,
      "grad_norm": 0.5565254874646226,
      "learning_rate": 7.651786456749908e-05,
      "loss": 12.017,
      "step": 21603
    },
    {
      "epoch": 1.176424142179385,
      "grad_norm": 0.5624231475396476,
      "learning_rate": 7.650929297198171e-05,
      "loss": 11.9315,
      "step": 21604
    },
    {
      "epoch": 1.176478596175968,
      "grad_norm": 0.6829559344536329,
      "learning_rate": 7.650072155913203e-05,
      "loss": 11.9368,
      "step": 21605
    },
    {
      "epoch": 1.176533050172551,
      "grad_norm": 0.5266904437228023,
      "learning_rate": 7.649215032901666e-05,
      "loss": 12.1225,
      "step": 21606
    },
    {
      "epoch": 1.176587504169134,
      "grad_norm": 0.5355267032043709,
      "learning_rate": 7.648357928170228e-05,
      "loss": 11.8722,
      "step": 21607
    },
    {
      "epoch": 1.176641958165717,
      "grad_norm": 0.5974199310957993,
      "learning_rate": 7.647500841725553e-05,
      "loss": 11.9625,
      "step": 21608
    },
    {
      "epoch": 1.1766964121623,
      "grad_norm": 0.5469317279833944,
      "learning_rate": 7.646643773574309e-05,
      "loss": 11.9221,
      "step": 21609
    },
    {
      "epoch": 1.176750866158883,
      "grad_norm": 0.5394380790937495,
      "learning_rate": 7.645786723723156e-05,
      "loss": 11.8321,
      "step": 21610
    },
    {
      "epoch": 1.1768053201554662,
      "grad_norm": 0.5701565621650369,
      "learning_rate": 7.644929692178758e-05,
      "loss": 12.1174,
      "step": 21611
    },
    {
      "epoch": 1.1768597741520492,
      "grad_norm": 0.5291252400811727,
      "learning_rate": 7.644072678947781e-05,
      "loss": 12.0025,
      "step": 21612
    },
    {
      "epoch": 1.1769142281486322,
      "grad_norm": 0.5223918746204956,
      "learning_rate": 7.643215684036891e-05,
      "loss": 12.0317,
      "step": 21613
    },
    {
      "epoch": 1.1769686821452152,
      "grad_norm": 0.6607752055855449,
      "learning_rate": 7.642358707452752e-05,
      "loss": 12.2902,
      "step": 21614
    },
    {
      "epoch": 1.1770231361417982,
      "grad_norm": 0.5726191469782652,
      "learning_rate": 7.641501749202028e-05,
      "loss": 12.0658,
      "step": 21615
    },
    {
      "epoch": 1.1770775901383812,
      "grad_norm": 0.6062855462084796,
      "learning_rate": 7.640644809291381e-05,
      "loss": 12.0044,
      "step": 21616
    },
    {
      "epoch": 1.1771320441349642,
      "grad_norm": 0.5062403174202712,
      "learning_rate": 7.639787887727474e-05,
      "loss": 11.9501,
      "step": 21617
    },
    {
      "epoch": 1.1771864981315472,
      "grad_norm": 0.5440039427500988,
      "learning_rate": 7.638930984516975e-05,
      "loss": 12.0606,
      "step": 21618
    },
    {
      "epoch": 1.1772409521281302,
      "grad_norm": 0.6159668397552996,
      "learning_rate": 7.638074099666545e-05,
      "loss": 12.1079,
      "step": 21619
    },
    {
      "epoch": 1.1772954061247132,
      "grad_norm": 0.5651141397183372,
      "learning_rate": 7.637217233182845e-05,
      "loss": 12.1082,
      "step": 21620
    },
    {
      "epoch": 1.1773498601212964,
      "grad_norm": 0.5454780345859144,
      "learning_rate": 7.63636038507254e-05,
      "loss": 11.9098,
      "step": 21621
    },
    {
      "epoch": 1.1774043141178794,
      "grad_norm": 0.5504587016802348,
      "learning_rate": 7.635503555342294e-05,
      "loss": 11.9108,
      "step": 21622
    },
    {
      "epoch": 1.1774587681144624,
      "grad_norm": 0.5042810541540745,
      "learning_rate": 7.634646743998765e-05,
      "loss": 12.0666,
      "step": 21623
    },
    {
      "epoch": 1.1775132221110454,
      "grad_norm": 0.5973572988736263,
      "learning_rate": 7.633789951048622e-05,
      "loss": 12.0896,
      "step": 21624
    },
    {
      "epoch": 1.1775676761076284,
      "grad_norm": 0.5247466352235615,
      "learning_rate": 7.632933176498527e-05,
      "loss": 11.9067,
      "step": 21625
    },
    {
      "epoch": 1.1776221301042114,
      "grad_norm": 0.6310328034642694,
      "learning_rate": 7.632076420355139e-05,
      "loss": 12.0383,
      "step": 21626
    },
    {
      "epoch": 1.1776765841007943,
      "grad_norm": 0.515311925378349,
      "learning_rate": 7.631219682625123e-05,
      "loss": 12.053,
      "step": 21627
    },
    {
      "epoch": 1.1777310380973773,
      "grad_norm": 0.6070465029960924,
      "learning_rate": 7.630362963315138e-05,
      "loss": 12.006,
      "step": 21628
    },
    {
      "epoch": 1.1777854920939603,
      "grad_norm": 0.5652235188360191,
      "learning_rate": 7.629506262431852e-05,
      "loss": 11.9344,
      "step": 21629
    },
    {
      "epoch": 1.1778399460905433,
      "grad_norm": 0.5773397818903766,
      "learning_rate": 7.62864957998192e-05,
      "loss": 12.0128,
      "step": 21630
    },
    {
      "epoch": 1.1778944000871263,
      "grad_norm": 0.6079772690009896,
      "learning_rate": 7.627792915972006e-05,
      "loss": 12.0603,
      "step": 21631
    },
    {
      "epoch": 1.1779488540837093,
      "grad_norm": 0.5497045741866009,
      "learning_rate": 7.626936270408774e-05,
      "loss": 11.927,
      "step": 21632
    },
    {
      "epoch": 1.1780033080802923,
      "grad_norm": 0.5537199405150324,
      "learning_rate": 7.626079643298882e-05,
      "loss": 12.0391,
      "step": 21633
    },
    {
      "epoch": 1.1780577620768755,
      "grad_norm": 0.5762763130172146,
      "learning_rate": 7.62522303464899e-05,
      "loss": 12.0694,
      "step": 21634
    },
    {
      "epoch": 1.1781122160734585,
      "grad_norm": 0.5431762987688716,
      "learning_rate": 7.624366444465764e-05,
      "loss": 11.958,
      "step": 21635
    },
    {
      "epoch": 1.1781666700700415,
      "grad_norm": 0.5582454172540878,
      "learning_rate": 7.623509872755866e-05,
      "loss": 12.1113,
      "step": 21636
    },
    {
      "epoch": 1.1782211240666245,
      "grad_norm": 0.5281912605122874,
      "learning_rate": 7.622653319525951e-05,
      "loss": 12.1244,
      "step": 21637
    },
    {
      "epoch": 1.1782755780632075,
      "grad_norm": 0.5317912076200442,
      "learning_rate": 7.621796784782683e-05,
      "loss": 11.9328,
      "step": 21638
    },
    {
      "epoch": 1.1783300320597905,
      "grad_norm": 0.5226098405298144,
      "learning_rate": 7.620940268532724e-05,
      "loss": 11.9744,
      "step": 21639
    },
    {
      "epoch": 1.1783844860563735,
      "grad_norm": 0.5211359218380267,
      "learning_rate": 7.620083770782731e-05,
      "loss": 11.7664,
      "step": 21640
    },
    {
      "epoch": 1.1784389400529565,
      "grad_norm": 0.5195296694323369,
      "learning_rate": 7.619227291539364e-05,
      "loss": 11.9052,
      "step": 21641
    },
    {
      "epoch": 1.1784933940495395,
      "grad_norm": 0.5125149024680569,
      "learning_rate": 7.618370830809287e-05,
      "loss": 12.0138,
      "step": 21642
    },
    {
      "epoch": 1.1785478480461224,
      "grad_norm": 0.5779339418846191,
      "learning_rate": 7.617514388599158e-05,
      "loss": 12.0039,
      "step": 21643
    },
    {
      "epoch": 1.1786023020427057,
      "grad_norm": 0.6368765557088946,
      "learning_rate": 7.616657964915634e-05,
      "loss": 12.0219,
      "step": 21644
    },
    {
      "epoch": 1.1786567560392887,
      "grad_norm": 0.6600037004525121,
      "learning_rate": 7.615801559765378e-05,
      "loss": 11.9651,
      "step": 21645
    },
    {
      "epoch": 1.1787112100358716,
      "grad_norm": 0.5504352862620744,
      "learning_rate": 7.614945173155049e-05,
      "loss": 12.0223,
      "step": 21646
    },
    {
      "epoch": 1.1787656640324546,
      "grad_norm": 0.5182725241016743,
      "learning_rate": 7.614088805091308e-05,
      "loss": 12.006,
      "step": 21647
    },
    {
      "epoch": 1.1788201180290376,
      "grad_norm": 0.5330077841931283,
      "learning_rate": 7.613232455580811e-05,
      "loss": 11.8409,
      "step": 21648
    },
    {
      "epoch": 1.1788745720256206,
      "grad_norm": 0.5525951312144873,
      "learning_rate": 7.612376124630224e-05,
      "loss": 12.0381,
      "step": 21649
    },
    {
      "epoch": 1.1789290260222036,
      "grad_norm": 0.5190346790686747,
      "learning_rate": 7.611519812246194e-05,
      "loss": 11.9219,
      "step": 21650
    },
    {
      "epoch": 1.1789834800187866,
      "grad_norm": 0.5979838501324751,
      "learning_rate": 7.610663518435388e-05,
      "loss": 12.0164,
      "step": 21651
    },
    {
      "epoch": 1.1790379340153696,
      "grad_norm": 0.5470158345829987,
      "learning_rate": 7.609807243204464e-05,
      "loss": 12.0324,
      "step": 21652
    },
    {
      "epoch": 1.1790923880119526,
      "grad_norm": 0.5308636105202947,
      "learning_rate": 7.60895098656008e-05,
      "loss": 12.0483,
      "step": 21653
    },
    {
      "epoch": 1.1791468420085356,
      "grad_norm": 0.500336870950848,
      "learning_rate": 7.608094748508893e-05,
      "loss": 11.7895,
      "step": 21654
    },
    {
      "epoch": 1.1792012960051186,
      "grad_norm": 0.5666076552703959,
      "learning_rate": 7.607238529057563e-05,
      "loss": 11.9856,
      "step": 21655
    },
    {
      "epoch": 1.1792557500017016,
      "grad_norm": 0.5528134284284965,
      "learning_rate": 7.606382328212748e-05,
      "loss": 12.012,
      "step": 21656
    },
    {
      "epoch": 1.1793102039982848,
      "grad_norm": 0.5439674157238209,
      "learning_rate": 7.605526145981103e-05,
      "loss": 11.9978,
      "step": 21657
    },
    {
      "epoch": 1.1793646579948678,
      "grad_norm": 0.5572339329752212,
      "learning_rate": 7.604669982369289e-05,
      "loss": 11.9457,
      "step": 21658
    },
    {
      "epoch": 1.1794191119914508,
      "grad_norm": 0.5019183247288584,
      "learning_rate": 7.603813837383968e-05,
      "loss": 12.0303,
      "step": 21659
    },
    {
      "epoch": 1.1794735659880338,
      "grad_norm": 0.5234298019560105,
      "learning_rate": 7.602957711031788e-05,
      "loss": 12.0421,
      "step": 21660
    },
    {
      "epoch": 1.1795280199846168,
      "grad_norm": 0.5988081752598906,
      "learning_rate": 7.60210160331941e-05,
      "loss": 12.0151,
      "step": 21661
    },
    {
      "epoch": 1.1795824739811998,
      "grad_norm": 0.5503929227153116,
      "learning_rate": 7.601245514253494e-05,
      "loss": 12.022,
      "step": 21662
    },
    {
      "epoch": 1.1796369279777827,
      "grad_norm": 0.5875432886014211,
      "learning_rate": 7.600389443840694e-05,
      "loss": 12.1746,
      "step": 21663
    },
    {
      "epoch": 1.1796913819743657,
      "grad_norm": 0.5548592242777497,
      "learning_rate": 7.599533392087667e-05,
      "loss": 12.0658,
      "step": 21664
    },
    {
      "epoch": 1.1797458359709487,
      "grad_norm": 0.5441249909569249,
      "learning_rate": 7.598677359001074e-05,
      "loss": 12.0069,
      "step": 21665
    },
    {
      "epoch": 1.1798002899675317,
      "grad_norm": 0.5411189061569485,
      "learning_rate": 7.597821344587566e-05,
      "loss": 11.9706,
      "step": 21666
    },
    {
      "epoch": 1.179854743964115,
      "grad_norm": 0.5401778057740976,
      "learning_rate": 7.596965348853804e-05,
      "loss": 11.9824,
      "step": 21667
    },
    {
      "epoch": 1.179909197960698,
      "grad_norm": 0.5358619510051359,
      "learning_rate": 7.59610937180644e-05,
      "loss": 11.9398,
      "step": 21668
    },
    {
      "epoch": 1.179963651957281,
      "grad_norm": 0.5755059795425265,
      "learning_rate": 7.595253413452133e-05,
      "loss": 11.9483,
      "step": 21669
    },
    {
      "epoch": 1.180018105953864,
      "grad_norm": 0.6660613258187639,
      "learning_rate": 7.594397473797545e-05,
      "loss": 11.9965,
      "step": 21670
    },
    {
      "epoch": 1.180072559950447,
      "grad_norm": 0.5245059523204031,
      "learning_rate": 7.593541552849319e-05,
      "loss": 11.9452,
      "step": 21671
    },
    {
      "epoch": 1.18012701394703,
      "grad_norm": 0.5805989717286782,
      "learning_rate": 7.592685650614118e-05,
      "loss": 12.0659,
      "step": 21672
    },
    {
      "epoch": 1.180181467943613,
      "grad_norm": 0.5281261993262514,
      "learning_rate": 7.591829767098598e-05,
      "loss": 11.9192,
      "step": 21673
    },
    {
      "epoch": 1.1802359219401959,
      "grad_norm": 0.5152599095149386,
      "learning_rate": 7.590973902309413e-05,
      "loss": 12.0361,
      "step": 21674
    },
    {
      "epoch": 1.1802903759367789,
      "grad_norm": 0.5273053352868992,
      "learning_rate": 7.59011805625322e-05,
      "loss": 11.8206,
      "step": 21675
    },
    {
      "epoch": 1.1803448299333619,
      "grad_norm": 0.6876622776737558,
      "learning_rate": 7.589262228936674e-05,
      "loss": 11.9064,
      "step": 21676
    },
    {
      "epoch": 1.1803992839299449,
      "grad_norm": 0.517094720883418,
      "learning_rate": 7.588406420366427e-05,
      "loss": 11.9869,
      "step": 21677
    },
    {
      "epoch": 1.1804537379265279,
      "grad_norm": 0.6099536909062926,
      "learning_rate": 7.587550630549136e-05,
      "loss": 11.8845,
      "step": 21678
    },
    {
      "epoch": 1.1805081919231109,
      "grad_norm": 0.6056931162179835,
      "learning_rate": 7.586694859491455e-05,
      "loss": 12.1122,
      "step": 21679
    },
    {
      "epoch": 1.180562645919694,
      "grad_norm": 0.5603632374089449,
      "learning_rate": 7.585839107200046e-05,
      "loss": 11.96,
      "step": 21680
    },
    {
      "epoch": 1.180617099916277,
      "grad_norm": 0.5319769323960366,
      "learning_rate": 7.584983373681552e-05,
      "loss": 11.9465,
      "step": 21681
    },
    {
      "epoch": 1.18067155391286,
      "grad_norm": 0.5019727975318679,
      "learning_rate": 7.584127658942632e-05,
      "loss": 11.7859,
      "step": 21682
    },
    {
      "epoch": 1.180726007909443,
      "grad_norm": 0.5351146581359351,
      "learning_rate": 7.58327196298994e-05,
      "loss": 12.0504,
      "step": 21683
    },
    {
      "epoch": 1.180780461906026,
      "grad_norm": 0.5645373231965284,
      "learning_rate": 7.582416285830132e-05,
      "loss": 12.011,
      "step": 21684
    },
    {
      "epoch": 1.180834915902609,
      "grad_norm": 0.5979440946985238,
      "learning_rate": 7.58156062746986e-05,
      "loss": 12.0611,
      "step": 21685
    },
    {
      "epoch": 1.180889369899192,
      "grad_norm": 0.5055138933136387,
      "learning_rate": 7.580704987915777e-05,
      "loss": 12.0173,
      "step": 21686
    },
    {
      "epoch": 1.180943823895775,
      "grad_norm": 0.4761881341162523,
      "learning_rate": 7.579849367174539e-05,
      "loss": 11.9772,
      "step": 21687
    },
    {
      "epoch": 1.180998277892358,
      "grad_norm": 0.6189338825141933,
      "learning_rate": 7.578993765252798e-05,
      "loss": 11.8785,
      "step": 21688
    },
    {
      "epoch": 1.181052731888941,
      "grad_norm": 0.656185572498181,
      "learning_rate": 7.578138182157208e-05,
      "loss": 12.0732,
      "step": 21689
    },
    {
      "epoch": 1.181107185885524,
      "grad_norm": 0.5553444220352391,
      "learning_rate": 7.577282617894423e-05,
      "loss": 11.8689,
      "step": 21690
    },
    {
      "epoch": 1.1811616398821072,
      "grad_norm": 0.6184382676376969,
      "learning_rate": 7.576427072471093e-05,
      "loss": 12.1293,
      "step": 21691
    },
    {
      "epoch": 1.1812160938786902,
      "grad_norm": 0.5651102836980537,
      "learning_rate": 7.575571545893873e-05,
      "loss": 12.1049,
      "step": 21692
    },
    {
      "epoch": 1.1812705478752732,
      "grad_norm": 0.5838191518166306,
      "learning_rate": 7.574716038169414e-05,
      "loss": 12.0283,
      "step": 21693
    },
    {
      "epoch": 1.1813250018718562,
      "grad_norm": 0.5161780662387591,
      "learning_rate": 7.573860549304368e-05,
      "loss": 11.9599,
      "step": 21694
    },
    {
      "epoch": 1.1813794558684392,
      "grad_norm": 0.5102349093103432,
      "learning_rate": 7.573005079305392e-05,
      "loss": 11.8245,
      "step": 21695
    },
    {
      "epoch": 1.1814339098650222,
      "grad_norm": 0.6195600594406752,
      "learning_rate": 7.572149628179135e-05,
      "loss": 12.0737,
      "step": 21696
    },
    {
      "epoch": 1.1814883638616052,
      "grad_norm": 0.639473472962265,
      "learning_rate": 7.57129419593225e-05,
      "loss": 12.0344,
      "step": 21697
    },
    {
      "epoch": 1.1815428178581882,
      "grad_norm": 0.495696446874667,
      "learning_rate": 7.570438782571388e-05,
      "loss": 11.9619,
      "step": 21698
    },
    {
      "epoch": 1.1815972718547711,
      "grad_norm": 0.5520409378708824,
      "learning_rate": 7.569583388103201e-05,
      "loss": 11.9519,
      "step": 21699
    },
    {
      "epoch": 1.1816517258513541,
      "grad_norm": 0.5965402705952677,
      "learning_rate": 7.568728012534345e-05,
      "loss": 12.1446,
      "step": 21700
    },
    {
      "epoch": 1.1817061798479371,
      "grad_norm": 0.5520543923776201,
      "learning_rate": 7.567872655871464e-05,
      "loss": 11.9793,
      "step": 21701
    },
    {
      "epoch": 1.1817606338445201,
      "grad_norm": 0.512609167119209,
      "learning_rate": 7.567017318121214e-05,
      "loss": 11.9533,
      "step": 21702
    },
    {
      "epoch": 1.1818150878411031,
      "grad_norm": 0.5471941486705536,
      "learning_rate": 7.566161999290246e-05,
      "loss": 11.9136,
      "step": 21703
    },
    {
      "epoch": 1.1818695418376863,
      "grad_norm": 0.7079993458364057,
      "learning_rate": 7.565306699385208e-05,
      "loss": 11.9342,
      "step": 21704
    },
    {
      "epoch": 1.1819239958342693,
      "grad_norm": 0.560453505923944,
      "learning_rate": 7.564451418412756e-05,
      "loss": 11.854,
      "step": 21705
    },
    {
      "epoch": 1.1819784498308523,
      "grad_norm": 0.566423257294691,
      "learning_rate": 7.563596156379536e-05,
      "loss": 12.0045,
      "step": 21706
    },
    {
      "epoch": 1.1820329038274353,
      "grad_norm": 0.5686640800327653,
      "learning_rate": 7.562740913292201e-05,
      "loss": 11.8475,
      "step": 21707
    },
    {
      "epoch": 1.1820873578240183,
      "grad_norm": 0.5770219785563593,
      "learning_rate": 7.561885689157402e-05,
      "loss": 11.9858,
      "step": 21708
    },
    {
      "epoch": 1.1821418118206013,
      "grad_norm": 0.530537262039549,
      "learning_rate": 7.56103048398179e-05,
      "loss": 11.9835,
      "step": 21709
    },
    {
      "epoch": 1.1821962658171843,
      "grad_norm": 0.5525937311384657,
      "learning_rate": 7.560175297772016e-05,
      "loss": 11.9754,
      "step": 21710
    },
    {
      "epoch": 1.1822507198137673,
      "grad_norm": 0.5695581815367907,
      "learning_rate": 7.559320130534724e-05,
      "loss": 12.0579,
      "step": 21711
    },
    {
      "epoch": 1.1823051738103503,
      "grad_norm": 0.5483632717658697,
      "learning_rate": 7.558464982276569e-05,
      "loss": 11.8674,
      "step": 21712
    },
    {
      "epoch": 1.1823596278069333,
      "grad_norm": 0.6059415138577536,
      "learning_rate": 7.5576098530042e-05,
      "loss": 11.7938,
      "step": 21713
    },
    {
      "epoch": 1.1824140818035165,
      "grad_norm": 0.5610916969377091,
      "learning_rate": 7.556754742724267e-05,
      "loss": 11.7845,
      "step": 21714
    },
    {
      "epoch": 1.1824685358000995,
      "grad_norm": 0.5227998543254345,
      "learning_rate": 7.555899651443417e-05,
      "loss": 12.0602,
      "step": 21715
    },
    {
      "epoch": 1.1825229897966825,
      "grad_norm": 0.5181757698245862,
      "learning_rate": 7.555044579168303e-05,
      "loss": 11.9774,
      "step": 21716
    },
    {
      "epoch": 1.1825774437932655,
      "grad_norm": 0.5710367409027894,
      "learning_rate": 7.554189525905569e-05,
      "loss": 12.0563,
      "step": 21717
    },
    {
      "epoch": 1.1826318977898485,
      "grad_norm": 0.46707603485019517,
      "learning_rate": 7.553334491661871e-05,
      "loss": 11.9401,
      "step": 21718
    },
    {
      "epoch": 1.1826863517864314,
      "grad_norm": 0.557364000105476,
      "learning_rate": 7.552479476443854e-05,
      "loss": 12.0401,
      "step": 21719
    },
    {
      "epoch": 1.1827408057830144,
      "grad_norm": 0.5968173852566319,
      "learning_rate": 7.55162448025817e-05,
      "loss": 11.9308,
      "step": 21720
    },
    {
      "epoch": 1.1827952597795974,
      "grad_norm": 0.6053471368379904,
      "learning_rate": 7.550769503111459e-05,
      "loss": 12.148,
      "step": 21721
    },
    {
      "epoch": 1.1828497137761804,
      "grad_norm": 0.5733610961834373,
      "learning_rate": 7.549914545010377e-05,
      "loss": 11.9982,
      "step": 21722
    },
    {
      "epoch": 1.1829041677727634,
      "grad_norm": 0.521571558531158,
      "learning_rate": 7.54905960596157e-05,
      "loss": 12.0328,
      "step": 21723
    },
    {
      "epoch": 1.1829586217693464,
      "grad_norm": 0.563622430367173,
      "learning_rate": 7.548204685971688e-05,
      "loss": 12.0217,
      "step": 21724
    },
    {
      "epoch": 1.1830130757659294,
      "grad_norm": 0.5433801146102819,
      "learning_rate": 7.547349785047376e-05,
      "loss": 11.926,
      "step": 21725
    },
    {
      "epoch": 1.1830675297625124,
      "grad_norm": 0.5531502300311864,
      "learning_rate": 7.546494903195284e-05,
      "loss": 11.7903,
      "step": 21726
    },
    {
      "epoch": 1.1831219837590956,
      "grad_norm": 0.5920247200071598,
      "learning_rate": 7.54564004042206e-05,
      "loss": 12.1427,
      "step": 21727
    },
    {
      "epoch": 1.1831764377556786,
      "grad_norm": 0.5443411208538833,
      "learning_rate": 7.544785196734347e-05,
      "loss": 12.0634,
      "step": 21728
    },
    {
      "epoch": 1.1832308917522616,
      "grad_norm": 0.6103023491678193,
      "learning_rate": 7.543930372138799e-05,
      "loss": 12.0133,
      "step": 21729
    },
    {
      "epoch": 1.1832853457488446,
      "grad_norm": 0.5284839116957317,
      "learning_rate": 7.543075566642063e-05,
      "loss": 11.9542,
      "step": 21730
    },
    {
      "epoch": 1.1833397997454276,
      "grad_norm": 0.5370732737599623,
      "learning_rate": 7.542220780250781e-05,
      "loss": 11.9737,
      "step": 21731
    },
    {
      "epoch": 1.1833942537420106,
      "grad_norm": 0.5081119754958391,
      "learning_rate": 7.5413660129716e-05,
      "loss": 11.9504,
      "step": 21732
    },
    {
      "epoch": 1.1834487077385936,
      "grad_norm": 0.5198359969619085,
      "learning_rate": 7.540511264811172e-05,
      "loss": 11.9739,
      "step": 21733
    },
    {
      "epoch": 1.1835031617351766,
      "grad_norm": 0.5294198800766511,
      "learning_rate": 7.53965653577614e-05,
      "loss": 12.0196,
      "step": 21734
    },
    {
      "epoch": 1.1835576157317595,
      "grad_norm": 0.6708523712972554,
      "learning_rate": 7.538801825873151e-05,
      "loss": 12.0584,
      "step": 21735
    },
    {
      "epoch": 1.1836120697283425,
      "grad_norm": 0.5905873501532606,
      "learning_rate": 7.537947135108852e-05,
      "loss": 12.0055,
      "step": 21736
    },
    {
      "epoch": 1.1836665237249258,
      "grad_norm": 0.5280514428250044,
      "learning_rate": 7.537092463489888e-05,
      "loss": 11.9186,
      "step": 21737
    },
    {
      "epoch": 1.1837209777215087,
      "grad_norm": 0.511630308174885,
      "learning_rate": 7.536237811022908e-05,
      "loss": 11.8857,
      "step": 21738
    },
    {
      "epoch": 1.1837754317180917,
      "grad_norm": 0.5773849467589317,
      "learning_rate": 7.535383177714553e-05,
      "loss": 12.0339,
      "step": 21739
    },
    {
      "epoch": 1.1838298857146747,
      "grad_norm": 0.5902880598466377,
      "learning_rate": 7.534528563571478e-05,
      "loss": 12.0105,
      "step": 21740
    },
    {
      "epoch": 1.1838843397112577,
      "grad_norm": 0.5576658102997276,
      "learning_rate": 7.533673968600317e-05,
      "loss": 11.8993,
      "step": 21741
    },
    {
      "epoch": 1.1839387937078407,
      "grad_norm": 0.5011873542817606,
      "learning_rate": 7.532819392807723e-05,
      "loss": 11.9773,
      "step": 21742
    },
    {
      "epoch": 1.1839932477044237,
      "grad_norm": 0.5056615027385634,
      "learning_rate": 7.531964836200336e-05,
      "loss": 11.9591,
      "step": 21743
    },
    {
      "epoch": 1.1840477017010067,
      "grad_norm": 0.5902346509462059,
      "learning_rate": 7.531110298784807e-05,
      "loss": 12.0789,
      "step": 21744
    },
    {
      "epoch": 1.1841021556975897,
      "grad_norm": 0.5357587708500191,
      "learning_rate": 7.530255780567777e-05,
      "loss": 11.9819,
      "step": 21745
    },
    {
      "epoch": 1.1841566096941727,
      "grad_norm": 0.5714466400676492,
      "learning_rate": 7.529401281555892e-05,
      "loss": 11.9447,
      "step": 21746
    },
    {
      "epoch": 1.1842110636907557,
      "grad_norm": 0.6919689987784267,
      "learning_rate": 7.528546801755799e-05,
      "loss": 12.066,
      "step": 21747
    },
    {
      "epoch": 1.1842655176873387,
      "grad_norm": 0.569462809235025,
      "learning_rate": 7.52769234117414e-05,
      "loss": 12.0948,
      "step": 21748
    },
    {
      "epoch": 1.1843199716839217,
      "grad_norm": 0.5582325725628782,
      "learning_rate": 7.526837899817559e-05,
      "loss": 11.9375,
      "step": 21749
    },
    {
      "epoch": 1.1843744256805049,
      "grad_norm": 0.5830476014734609,
      "learning_rate": 7.525983477692703e-05,
      "loss": 12.0411,
      "step": 21750
    },
    {
      "epoch": 1.1844288796770879,
      "grad_norm": 0.6654803020952056,
      "learning_rate": 7.525129074806213e-05,
      "loss": 12.0759,
      "step": 21751
    },
    {
      "epoch": 1.1844833336736709,
      "grad_norm": 0.5128129143192934,
      "learning_rate": 7.524274691164734e-05,
      "loss": 11.9515,
      "step": 21752
    },
    {
      "epoch": 1.1845377876702539,
      "grad_norm": 0.6029074652028825,
      "learning_rate": 7.523420326774911e-05,
      "loss": 11.9985,
      "step": 21753
    },
    {
      "epoch": 1.1845922416668369,
      "grad_norm": 0.5067885544790296,
      "learning_rate": 7.522565981643387e-05,
      "loss": 11.8589,
      "step": 21754
    },
    {
      "epoch": 1.1846466956634198,
      "grad_norm": 0.5158502405510174,
      "learning_rate": 7.521711655776802e-05,
      "loss": 11.9687,
      "step": 21755
    },
    {
      "epoch": 1.1847011496600028,
      "grad_norm": 0.5380068678218914,
      "learning_rate": 7.520857349181806e-05,
      "loss": 12.0553,
      "step": 21756
    },
    {
      "epoch": 1.1847556036565858,
      "grad_norm": 0.5258209740523851,
      "learning_rate": 7.520003061865038e-05,
      "loss": 11.9678,
      "step": 21757
    },
    {
      "epoch": 1.1848100576531688,
      "grad_norm": 0.5674122484543469,
      "learning_rate": 7.519148793833143e-05,
      "loss": 12.027,
      "step": 21758
    },
    {
      "epoch": 1.1848645116497518,
      "grad_norm": 0.5415370607754529,
      "learning_rate": 7.518294545092763e-05,
      "loss": 11.8541,
      "step": 21759
    },
    {
      "epoch": 1.1849189656463348,
      "grad_norm": 0.5215861042928728,
      "learning_rate": 7.517440315650542e-05,
      "loss": 11.8873,
      "step": 21760
    },
    {
      "epoch": 1.184973419642918,
      "grad_norm": 0.7521342107961826,
      "learning_rate": 7.51658610551312e-05,
      "loss": 12.246,
      "step": 21761
    },
    {
      "epoch": 1.185027873639501,
      "grad_norm": 0.5522051040985905,
      "learning_rate": 7.51573191468714e-05,
      "loss": 11.9465,
      "step": 21762
    },
    {
      "epoch": 1.185082327636084,
      "grad_norm": 0.5986971920610508,
      "learning_rate": 7.514877743179248e-05,
      "loss": 12.0501,
      "step": 21763
    },
    {
      "epoch": 1.185136781632667,
      "grad_norm": 0.6577498856361116,
      "learning_rate": 7.514023590996081e-05,
      "loss": 12.0744,
      "step": 21764
    },
    {
      "epoch": 1.18519123562925,
      "grad_norm": 0.5709753816871418,
      "learning_rate": 7.513169458144284e-05,
      "loss": 11.8767,
      "step": 21765
    },
    {
      "epoch": 1.185245689625833,
      "grad_norm": 0.5424432614333545,
      "learning_rate": 7.512315344630496e-05,
      "loss": 11.942,
      "step": 21766
    },
    {
      "epoch": 1.185300143622416,
      "grad_norm": 0.5524916153826994,
      "learning_rate": 7.511461250461364e-05,
      "loss": 12.0522,
      "step": 21767
    },
    {
      "epoch": 1.185354597618999,
      "grad_norm": 0.5386178542301641,
      "learning_rate": 7.510607175643525e-05,
      "loss": 12.0999,
      "step": 21768
    },
    {
      "epoch": 1.185409051615582,
      "grad_norm": 0.5302354868198255,
      "learning_rate": 7.509753120183624e-05,
      "loss": 12.0432,
      "step": 21769
    },
    {
      "epoch": 1.185463505612165,
      "grad_norm": 0.5581106828952281,
      "learning_rate": 7.508899084088301e-05,
      "loss": 12.0233,
      "step": 21770
    },
    {
      "epoch": 1.185517959608748,
      "grad_norm": 0.5425977101660957,
      "learning_rate": 7.508045067364194e-05,
      "loss": 12.047,
      "step": 21771
    },
    {
      "epoch": 1.185572413605331,
      "grad_norm": 0.593887575565652,
      "learning_rate": 7.507191070017948e-05,
      "loss": 11.974,
      "step": 21772
    },
    {
      "epoch": 1.185626867601914,
      "grad_norm": 0.5225432690410278,
      "learning_rate": 7.506337092056202e-05,
      "loss": 12.0737,
      "step": 21773
    },
    {
      "epoch": 1.1856813215984972,
      "grad_norm": 0.6143012266055953,
      "learning_rate": 7.505483133485595e-05,
      "loss": 11.9907,
      "step": 21774
    },
    {
      "epoch": 1.1857357755950801,
      "grad_norm": 0.5629345433031676,
      "learning_rate": 7.504629194312773e-05,
      "loss": 11.9683,
      "step": 21775
    },
    {
      "epoch": 1.1857902295916631,
      "grad_norm": 0.581947804657824,
      "learning_rate": 7.50377527454437e-05,
      "loss": 11.9701,
      "step": 21776
    },
    {
      "epoch": 1.1858446835882461,
      "grad_norm": 0.5147110297629652,
      "learning_rate": 7.502921374187029e-05,
      "loss": 11.9844,
      "step": 21777
    },
    {
      "epoch": 1.1858991375848291,
      "grad_norm": 0.5407679106530497,
      "learning_rate": 7.50206749324739e-05,
      "loss": 11.8965,
      "step": 21778
    },
    {
      "epoch": 1.1859535915814121,
      "grad_norm": 0.5643712912136952,
      "learning_rate": 7.501213631732095e-05,
      "loss": 12.0641,
      "step": 21779
    },
    {
      "epoch": 1.186008045577995,
      "grad_norm": 0.6000764496420661,
      "learning_rate": 7.500359789647785e-05,
      "loss": 12.1549,
      "step": 21780
    },
    {
      "epoch": 1.186062499574578,
      "grad_norm": 0.5520801822640712,
      "learning_rate": 7.49950596700109e-05,
      "loss": 11.9133,
      "step": 21781
    },
    {
      "epoch": 1.186116953571161,
      "grad_norm": 0.6109658539324219,
      "learning_rate": 7.498652163798658e-05,
      "loss": 11.9836,
      "step": 21782
    },
    {
      "epoch": 1.186171407567744,
      "grad_norm": 0.5252949920892036,
      "learning_rate": 7.497798380047127e-05,
      "loss": 12.1093,
      "step": 21783
    },
    {
      "epoch": 1.1862258615643273,
      "grad_norm": 0.6179409760128146,
      "learning_rate": 7.496944615753136e-05,
      "loss": 12.0782,
      "step": 21784
    },
    {
      "epoch": 1.1862803155609103,
      "grad_norm": 0.5840217304614371,
      "learning_rate": 7.496090870923323e-05,
      "loss": 11.9846,
      "step": 21785
    },
    {
      "epoch": 1.1863347695574933,
      "grad_norm": 0.594856080206936,
      "learning_rate": 7.495237145564327e-05,
      "loss": 12.0495,
      "step": 21786
    },
    {
      "epoch": 1.1863892235540763,
      "grad_norm": 0.5039334530956467,
      "learning_rate": 7.494383439682787e-05,
      "loss": 11.902,
      "step": 21787
    },
    {
      "epoch": 1.1864436775506593,
      "grad_norm": 0.48203965829591866,
      "learning_rate": 7.493529753285339e-05,
      "loss": 11.7758,
      "step": 21788
    },
    {
      "epoch": 1.1864981315472423,
      "grad_norm": 0.5000380138956201,
      "learning_rate": 7.492676086378626e-05,
      "loss": 11.9807,
      "step": 21789
    },
    {
      "epoch": 1.1865525855438253,
      "grad_norm": 0.5560242956456776,
      "learning_rate": 7.491822438969289e-05,
      "loss": 11.9622,
      "step": 21790
    },
    {
      "epoch": 1.1866070395404082,
      "grad_norm": 0.561093459286848,
      "learning_rate": 7.490968811063956e-05,
      "loss": 11.9649,
      "step": 21791
    },
    {
      "epoch": 1.1866614935369912,
      "grad_norm": 0.569981991409385,
      "learning_rate": 7.49011520266927e-05,
      "loss": 12.2138,
      "step": 21792
    },
    {
      "epoch": 1.1867159475335742,
      "grad_norm": 0.5575985421141544,
      "learning_rate": 7.489261613791871e-05,
      "loss": 11.9822,
      "step": 21793
    },
    {
      "epoch": 1.1867704015301572,
      "grad_norm": 0.4961472956487318,
      "learning_rate": 7.488408044438393e-05,
      "loss": 12.0297,
      "step": 21794
    },
    {
      "epoch": 1.1868248555267402,
      "grad_norm": 0.5311876448727464,
      "learning_rate": 7.487554494615476e-05,
      "loss": 12.059,
      "step": 21795
    },
    {
      "epoch": 1.1868793095233232,
      "grad_norm": 0.5626886924773438,
      "learning_rate": 7.486700964329758e-05,
      "loss": 11.9969,
      "step": 21796
    },
    {
      "epoch": 1.1869337635199064,
      "grad_norm": 0.5843387177835994,
      "learning_rate": 7.485847453587873e-05,
      "loss": 12.0714,
      "step": 21797
    },
    {
      "epoch": 1.1869882175164894,
      "grad_norm": 0.5189952502279995,
      "learning_rate": 7.48499396239646e-05,
      "loss": 12.1539,
      "step": 21798
    },
    {
      "epoch": 1.1870426715130724,
      "grad_norm": 0.5209743899056666,
      "learning_rate": 7.484140490762158e-05,
      "loss": 11.8418,
      "step": 21799
    },
    {
      "epoch": 1.1870971255096554,
      "grad_norm": 0.4994690385578024,
      "learning_rate": 7.483287038691601e-05,
      "loss": 11.8865,
      "step": 21800
    },
    {
      "epoch": 1.1871515795062384,
      "grad_norm": 0.5424976862399479,
      "learning_rate": 7.482433606191426e-05,
      "loss": 12.0755,
      "step": 21801
    },
    {
      "epoch": 1.1872060335028214,
      "grad_norm": 0.5428216072561727,
      "learning_rate": 7.481580193268267e-05,
      "loss": 11.8406,
      "step": 21802
    },
    {
      "epoch": 1.1872604874994044,
      "grad_norm": 0.611819389257739,
      "learning_rate": 7.480726799928764e-05,
      "loss": 11.9912,
      "step": 21803
    },
    {
      "epoch": 1.1873149414959874,
      "grad_norm": 0.5115133464397964,
      "learning_rate": 7.47987342617955e-05,
      "loss": 12.0168,
      "step": 21804
    },
    {
      "epoch": 1.1873693954925704,
      "grad_norm": 0.6358308408573634,
      "learning_rate": 7.479020072027266e-05,
      "loss": 12.166,
      "step": 21805
    },
    {
      "epoch": 1.1874238494891534,
      "grad_norm": 0.5419860484051422,
      "learning_rate": 7.478166737478543e-05,
      "loss": 12.1022,
      "step": 21806
    },
    {
      "epoch": 1.1874783034857366,
      "grad_norm": 0.50538323000821,
      "learning_rate": 7.477313422540017e-05,
      "loss": 11.9044,
      "step": 21807
    },
    {
      "epoch": 1.1875327574823196,
      "grad_norm": 0.5468604627982997,
      "learning_rate": 7.476460127218328e-05,
      "loss": 12.081,
      "step": 21808
    },
    {
      "epoch": 1.1875872114789026,
      "grad_norm": 0.5568421073854486,
      "learning_rate": 7.475606851520107e-05,
      "loss": 12.0435,
      "step": 21809
    },
    {
      "epoch": 1.1876416654754856,
      "grad_norm": 0.6510994234584724,
      "learning_rate": 7.474753595451992e-05,
      "loss": 12.0943,
      "step": 21810
    },
    {
      "epoch": 1.1876961194720685,
      "grad_norm": 0.5522041966154742,
      "learning_rate": 7.473900359020615e-05,
      "loss": 12.0387,
      "step": 21811
    },
    {
      "epoch": 1.1877505734686515,
      "grad_norm": 0.6119962614850255,
      "learning_rate": 7.473047142232611e-05,
      "loss": 12.0829,
      "step": 21812
    },
    {
      "epoch": 1.1878050274652345,
      "grad_norm": 0.5305426444618092,
      "learning_rate": 7.472193945094619e-05,
      "loss": 12.0808,
      "step": 21813
    },
    {
      "epoch": 1.1878594814618175,
      "grad_norm": 0.6337365180981813,
      "learning_rate": 7.47134076761327e-05,
      "loss": 12.129,
      "step": 21814
    },
    {
      "epoch": 1.1879139354584005,
      "grad_norm": 0.5250699532031685,
      "learning_rate": 7.470487609795197e-05,
      "loss": 11.983,
      "step": 21815
    },
    {
      "epoch": 1.1879683894549835,
      "grad_norm": 0.5471512499498226,
      "learning_rate": 7.469634471647037e-05,
      "loss": 12.1291,
      "step": 21816
    },
    {
      "epoch": 1.1880228434515665,
      "grad_norm": 0.5467733525912786,
      "learning_rate": 7.468781353175425e-05,
      "loss": 11.9967,
      "step": 21817
    },
    {
      "epoch": 1.1880772974481495,
      "grad_norm": 0.50643789381646,
      "learning_rate": 7.467928254386993e-05,
      "loss": 11.8694,
      "step": 21818
    },
    {
      "epoch": 1.1881317514447325,
      "grad_norm": 0.6046764353226523,
      "learning_rate": 7.467075175288377e-05,
      "loss": 12.0005,
      "step": 21819
    },
    {
      "epoch": 1.1881862054413157,
      "grad_norm": 0.5516677753495105,
      "learning_rate": 7.466222115886208e-05,
      "loss": 11.8673,
      "step": 21820
    },
    {
      "epoch": 1.1882406594378987,
      "grad_norm": 0.5172722454661293,
      "learning_rate": 7.465369076187124e-05,
      "loss": 11.9521,
      "step": 21821
    },
    {
      "epoch": 1.1882951134344817,
      "grad_norm": 0.49449399673176725,
      "learning_rate": 7.46451605619775e-05,
      "loss": 11.9785,
      "step": 21822
    },
    {
      "epoch": 1.1883495674310647,
      "grad_norm": 0.6243888753730924,
      "learning_rate": 7.463663055924728e-05,
      "loss": 11.9585,
      "step": 21823
    },
    {
      "epoch": 1.1884040214276477,
      "grad_norm": 0.5698569554828076,
      "learning_rate": 7.462810075374685e-05,
      "loss": 11.8915,
      "step": 21824
    },
    {
      "epoch": 1.1884584754242307,
      "grad_norm": 0.5130897372640837,
      "learning_rate": 7.461957114554256e-05,
      "loss": 12.0606,
      "step": 21825
    },
    {
      "epoch": 1.1885129294208137,
      "grad_norm": 0.585214140568694,
      "learning_rate": 7.461104173470075e-05,
      "loss": 12.0466,
      "step": 21826
    },
    {
      "epoch": 1.1885673834173967,
      "grad_norm": 0.8057724923497462,
      "learning_rate": 7.460251252128774e-05,
      "loss": 12.0085,
      "step": 21827
    },
    {
      "epoch": 1.1886218374139796,
      "grad_norm": 0.5206311767301525,
      "learning_rate": 7.459398350536985e-05,
      "loss": 12.0089,
      "step": 21828
    },
    {
      "epoch": 1.1886762914105626,
      "grad_norm": 0.5250316044303713,
      "learning_rate": 7.458545468701341e-05,
      "loss": 12.0066,
      "step": 21829
    },
    {
      "epoch": 1.1887307454071459,
      "grad_norm": 0.5309614924213232,
      "learning_rate": 7.457692606628473e-05,
      "loss": 11.9824,
      "step": 21830
    },
    {
      "epoch": 1.1887851994037288,
      "grad_norm": 0.6201426355334324,
      "learning_rate": 7.456839764325016e-05,
      "loss": 11.9513,
      "step": 21831
    },
    {
      "epoch": 1.1888396534003118,
      "grad_norm": 0.6264038379079032,
      "learning_rate": 7.455986941797597e-05,
      "loss": 12.1016,
      "step": 21832
    },
    {
      "epoch": 1.1888941073968948,
      "grad_norm": 0.5442003806645258,
      "learning_rate": 7.45513413905285e-05,
      "loss": 12.0651,
      "step": 21833
    },
    {
      "epoch": 1.1889485613934778,
      "grad_norm": 0.5446554848843562,
      "learning_rate": 7.454281356097407e-05,
      "loss": 11.9691,
      "step": 21834
    },
    {
      "epoch": 1.1890030153900608,
      "grad_norm": 0.6018851851438473,
      "learning_rate": 7.453428592937901e-05,
      "loss": 12.0177,
      "step": 21835
    },
    {
      "epoch": 1.1890574693866438,
      "grad_norm": 0.5436708939061847,
      "learning_rate": 7.45257584958096e-05,
      "loss": 11.8384,
      "step": 21836
    },
    {
      "epoch": 1.1891119233832268,
      "grad_norm": 0.5096974488486813,
      "learning_rate": 7.451723126033214e-05,
      "loss": 11.8501,
      "step": 21837
    },
    {
      "epoch": 1.1891663773798098,
      "grad_norm": 0.6681927843726951,
      "learning_rate": 7.450870422301298e-05,
      "loss": 12.0757,
      "step": 21838
    },
    {
      "epoch": 1.1892208313763928,
      "grad_norm": 0.5688654042324359,
      "learning_rate": 7.450017738391841e-05,
      "loss": 11.7817,
      "step": 21839
    },
    {
      "epoch": 1.1892752853729758,
      "grad_norm": 0.5095351042546655,
      "learning_rate": 7.449165074311475e-05,
      "loss": 11.9059,
      "step": 21840
    },
    {
      "epoch": 1.1893297393695588,
      "grad_norm": 0.5143100203772302,
      "learning_rate": 7.448312430066831e-05,
      "loss": 12.0267,
      "step": 21841
    },
    {
      "epoch": 1.1893841933661418,
      "grad_norm": 0.5155825069420525,
      "learning_rate": 7.447459805664534e-05,
      "loss": 11.8781,
      "step": 21842
    },
    {
      "epoch": 1.1894386473627248,
      "grad_norm": 0.5223028192969036,
      "learning_rate": 7.446607201111219e-05,
      "loss": 12.0149,
      "step": 21843
    },
    {
      "epoch": 1.189493101359308,
      "grad_norm": 0.5689111220899553,
      "learning_rate": 7.445754616413514e-05,
      "loss": 11.7965,
      "step": 21844
    },
    {
      "epoch": 1.189547555355891,
      "grad_norm": 0.5745430197939587,
      "learning_rate": 7.444902051578049e-05,
      "loss": 12.1652,
      "step": 21845
    },
    {
      "epoch": 1.189602009352474,
      "grad_norm": 0.5448431461543952,
      "learning_rate": 7.444049506611454e-05,
      "loss": 11.8951,
      "step": 21846
    },
    {
      "epoch": 1.189656463349057,
      "grad_norm": 0.5872918584595909,
      "learning_rate": 7.44319698152036e-05,
      "loss": 12.0168,
      "step": 21847
    },
    {
      "epoch": 1.18971091734564,
      "grad_norm": 0.5226448915531637,
      "learning_rate": 7.442344476311393e-05,
      "loss": 12.0349,
      "step": 21848
    },
    {
      "epoch": 1.189765371342223,
      "grad_norm": 0.5432356020572908,
      "learning_rate": 7.441491990991185e-05,
      "loss": 11.9819,
      "step": 21849
    },
    {
      "epoch": 1.189819825338806,
      "grad_norm": 0.5628765091838128,
      "learning_rate": 7.440639525566365e-05,
      "loss": 12.068,
      "step": 21850
    },
    {
      "epoch": 1.189874279335389,
      "grad_norm": 0.531129064044447,
      "learning_rate": 7.439787080043565e-05,
      "loss": 11.977,
      "step": 21851
    },
    {
      "epoch": 1.189928733331972,
      "grad_norm": 0.5306473030453958,
      "learning_rate": 7.438934654429407e-05,
      "loss": 11.9767,
      "step": 21852
    },
    {
      "epoch": 1.189983187328555,
      "grad_norm": 0.6468496902013098,
      "learning_rate": 7.438082248730521e-05,
      "loss": 12.0752,
      "step": 21853
    },
    {
      "epoch": 1.1900376413251381,
      "grad_norm": 0.662395989981953,
      "learning_rate": 7.437229862953536e-05,
      "loss": 12.1451,
      "step": 21854
    },
    {
      "epoch": 1.190092095321721,
      "grad_norm": 0.5634303787723975,
      "learning_rate": 7.436377497105085e-05,
      "loss": 11.8939,
      "step": 21855
    },
    {
      "epoch": 1.190146549318304,
      "grad_norm": 0.5284039031704981,
      "learning_rate": 7.43552515119179e-05,
      "loss": 12.0463,
      "step": 21856
    },
    {
      "epoch": 1.190201003314887,
      "grad_norm": 0.5481202976593118,
      "learning_rate": 7.434672825220283e-05,
      "loss": 12.0324,
      "step": 21857
    },
    {
      "epoch": 1.19025545731147,
      "grad_norm": 0.5671166962659867,
      "learning_rate": 7.433820519197189e-05,
      "loss": 12.0328,
      "step": 21858
    },
    {
      "epoch": 1.190309911308053,
      "grad_norm": 0.552763651588234,
      "learning_rate": 7.432968233129139e-05,
      "loss": 12.0551,
      "step": 21859
    },
    {
      "epoch": 1.190364365304636,
      "grad_norm": 0.5391150052421613,
      "learning_rate": 7.432115967022754e-05,
      "loss": 11.999,
      "step": 21860
    },
    {
      "epoch": 1.190418819301219,
      "grad_norm": 0.5515344211903856,
      "learning_rate": 7.431263720884674e-05,
      "loss": 11.9639,
      "step": 21861
    },
    {
      "epoch": 1.190473273297802,
      "grad_norm": 0.4927736099365677,
      "learning_rate": 7.430411494721512e-05,
      "loss": 12.0217,
      "step": 21862
    },
    {
      "epoch": 1.190527727294385,
      "grad_norm": 0.6616589330031137,
      "learning_rate": 7.429559288539903e-05,
      "loss": 12.0721,
      "step": 21863
    },
    {
      "epoch": 1.190582181290968,
      "grad_norm": 0.5585836406806164,
      "learning_rate": 7.428707102346469e-05,
      "loss": 11.963,
      "step": 21864
    },
    {
      "epoch": 1.190636635287551,
      "grad_norm": 0.5778445956640136,
      "learning_rate": 7.427854936147841e-05,
      "loss": 12.1283,
      "step": 21865
    },
    {
      "epoch": 1.190691089284134,
      "grad_norm": 0.5773331365814527,
      "learning_rate": 7.427002789950645e-05,
      "loss": 12.102,
      "step": 21866
    },
    {
      "epoch": 1.1907455432807172,
      "grad_norm": 0.6090155689074539,
      "learning_rate": 7.426150663761508e-05,
      "loss": 12.058,
      "step": 21867
    },
    {
      "epoch": 1.1907999972773002,
      "grad_norm": 0.5268971066809944,
      "learning_rate": 7.425298557587054e-05,
      "loss": 11.9267,
      "step": 21868
    },
    {
      "epoch": 1.1908544512738832,
      "grad_norm": 0.5352024391379133,
      "learning_rate": 7.42444647143391e-05,
      "loss": 11.8219,
      "step": 21869
    },
    {
      "epoch": 1.1909089052704662,
      "grad_norm": 0.5267535550203765,
      "learning_rate": 7.423594405308703e-05,
      "loss": 11.9724,
      "step": 21870
    },
    {
      "epoch": 1.1909633592670492,
      "grad_norm": 0.5316513047581957,
      "learning_rate": 7.42274235921806e-05,
      "loss": 12.0106,
      "step": 21871
    },
    {
      "epoch": 1.1910178132636322,
      "grad_norm": 0.5899821522090289,
      "learning_rate": 7.421890333168602e-05,
      "loss": 12.1346,
      "step": 21872
    },
    {
      "epoch": 1.1910722672602152,
      "grad_norm": 0.5992321160668765,
      "learning_rate": 7.421038327166958e-05,
      "loss": 12.0992,
      "step": 21873
    },
    {
      "epoch": 1.1911267212567982,
      "grad_norm": 0.5220059104085887,
      "learning_rate": 7.420186341219751e-05,
      "loss": 12.0391,
      "step": 21874
    },
    {
      "epoch": 1.1911811752533812,
      "grad_norm": 0.5483952568272,
      "learning_rate": 7.419334375333606e-05,
      "loss": 12.0126,
      "step": 21875
    },
    {
      "epoch": 1.1912356292499642,
      "grad_norm": 0.5386623185595957,
      "learning_rate": 7.418482429515152e-05,
      "loss": 12.078,
      "step": 21876
    },
    {
      "epoch": 1.1912900832465474,
      "grad_norm": 0.5439532173277934,
      "learning_rate": 7.417630503771013e-05,
      "loss": 12.0004,
      "step": 21877
    },
    {
      "epoch": 1.1913445372431304,
      "grad_norm": 0.5736166552730142,
      "learning_rate": 7.41677859810781e-05,
      "loss": 12.0908,
      "step": 21878
    },
    {
      "epoch": 1.1913989912397134,
      "grad_norm": 0.5648809372426857,
      "learning_rate": 7.41592671253217e-05,
      "loss": 11.9778,
      "step": 21879
    },
    {
      "epoch": 1.1914534452362964,
      "grad_norm": 0.48510183035173565,
      "learning_rate": 7.41507484705072e-05,
      "loss": 11.9495,
      "step": 21880
    },
    {
      "epoch": 1.1915078992328794,
      "grad_norm": 0.5211179210723154,
      "learning_rate": 7.41422300167008e-05,
      "loss": 11.9781,
      "step": 21881
    },
    {
      "epoch": 1.1915623532294624,
      "grad_norm": 0.5299213249093413,
      "learning_rate": 7.413371176396876e-05,
      "loss": 12.0785,
      "step": 21882
    },
    {
      "epoch": 1.1916168072260453,
      "grad_norm": 0.49308108682965807,
      "learning_rate": 7.41251937123773e-05,
      "loss": 11.9841,
      "step": 21883
    },
    {
      "epoch": 1.1916712612226283,
      "grad_norm": 0.5342328033936905,
      "learning_rate": 7.41166758619927e-05,
      "loss": 11.8857,
      "step": 21884
    },
    {
      "epoch": 1.1917257152192113,
      "grad_norm": 0.5273168195257198,
      "learning_rate": 7.410815821288113e-05,
      "loss": 12.0555,
      "step": 21885
    },
    {
      "epoch": 1.1917801692157943,
      "grad_norm": 0.5528610747944571,
      "learning_rate": 7.409964076510886e-05,
      "loss": 11.9563,
      "step": 21886
    },
    {
      "epoch": 1.1918346232123773,
      "grad_norm": 0.5396212993556745,
      "learning_rate": 7.409112351874214e-05,
      "loss": 11.9265,
      "step": 21887
    },
    {
      "epoch": 1.1918890772089603,
      "grad_norm": 0.5604629835030344,
      "learning_rate": 7.40826064738472e-05,
      "loss": 11.9565,
      "step": 21888
    },
    {
      "epoch": 1.1919435312055433,
      "grad_norm": 0.5256247292380115,
      "learning_rate": 7.407408963049027e-05,
      "loss": 11.9664,
      "step": 21889
    },
    {
      "epoch": 1.1919979852021265,
      "grad_norm": 0.5554092422140184,
      "learning_rate": 7.406557298873754e-05,
      "loss": 11.9756,
      "step": 21890
    },
    {
      "epoch": 1.1920524391987095,
      "grad_norm": 0.5385375597876597,
      "learning_rate": 7.40570565486553e-05,
      "loss": 11.9556,
      "step": 21891
    },
    {
      "epoch": 1.1921068931952925,
      "grad_norm": 0.5466236276669121,
      "learning_rate": 7.404854031030971e-05,
      "loss": 11.9882,
      "step": 21892
    },
    {
      "epoch": 1.1921613471918755,
      "grad_norm": 0.610434010639925,
      "learning_rate": 7.404002427376703e-05,
      "loss": 11.9038,
      "step": 21893
    },
    {
      "epoch": 1.1922158011884585,
      "grad_norm": 0.5481300770223017,
      "learning_rate": 7.403150843909348e-05,
      "loss": 11.7977,
      "step": 21894
    },
    {
      "epoch": 1.1922702551850415,
      "grad_norm": 0.5432539067052593,
      "learning_rate": 7.402299280635526e-05,
      "loss": 11.8553,
      "step": 21895
    },
    {
      "epoch": 1.1923247091816245,
      "grad_norm": 0.5326296538994771,
      "learning_rate": 7.401447737561862e-05,
      "loss": 12.0045,
      "step": 21896
    },
    {
      "epoch": 1.1923791631782075,
      "grad_norm": 0.48761894352387997,
      "learning_rate": 7.400596214694973e-05,
      "loss": 11.9545,
      "step": 21897
    },
    {
      "epoch": 1.1924336171747905,
      "grad_norm": 0.5777590971349343,
      "learning_rate": 7.399744712041485e-05,
      "loss": 12.0196,
      "step": 21898
    },
    {
      "epoch": 1.1924880711713735,
      "grad_norm": 0.5979944522655282,
      "learning_rate": 7.398893229608019e-05,
      "loss": 11.9432,
      "step": 21899
    },
    {
      "epoch": 1.1925425251679567,
      "grad_norm": 0.5351970404514349,
      "learning_rate": 7.398041767401196e-05,
      "loss": 11.9262,
      "step": 21900
    },
    {
      "epoch": 1.1925969791645397,
      "grad_norm": 0.5851772163331859,
      "learning_rate": 7.397190325427641e-05,
      "loss": 12.0415,
      "step": 21901
    },
    {
      "epoch": 1.1926514331611227,
      "grad_norm": 0.5520233932528713,
      "learning_rate": 7.396338903693964e-05,
      "loss": 11.9481,
      "step": 21902
    },
    {
      "epoch": 1.1927058871577056,
      "grad_norm": 0.5674022158261073,
      "learning_rate": 7.395487502206794e-05,
      "loss": 11.8746,
      "step": 21903
    },
    {
      "epoch": 1.1927603411542886,
      "grad_norm": 0.5574839517552891,
      "learning_rate": 7.394636120972749e-05,
      "loss": 11.9236,
      "step": 21904
    },
    {
      "epoch": 1.1928147951508716,
      "grad_norm": 0.5526424355538554,
      "learning_rate": 7.393784759998452e-05,
      "loss": 11.8955,
      "step": 21905
    },
    {
      "epoch": 1.1928692491474546,
      "grad_norm": 0.5830260492488549,
      "learning_rate": 7.392933419290522e-05,
      "loss": 11.9274,
      "step": 21906
    },
    {
      "epoch": 1.1929237031440376,
      "grad_norm": 0.5597230753997932,
      "learning_rate": 7.392082098855577e-05,
      "loss": 12.0131,
      "step": 21907
    },
    {
      "epoch": 1.1929781571406206,
      "grad_norm": 0.6554421652823247,
      "learning_rate": 7.391230798700242e-05,
      "loss": 12.031,
      "step": 21908
    },
    {
      "epoch": 1.1930326111372036,
      "grad_norm": 0.5903428092902078,
      "learning_rate": 7.390379518831129e-05,
      "loss": 12.0472,
      "step": 21909
    },
    {
      "epoch": 1.1930870651337866,
      "grad_norm": 0.5491708507929751,
      "learning_rate": 7.389528259254866e-05,
      "loss": 11.9285,
      "step": 21910
    },
    {
      "epoch": 1.1931415191303696,
      "grad_norm": 0.5961736138488625,
      "learning_rate": 7.388677019978072e-05,
      "loss": 11.9876,
      "step": 21911
    },
    {
      "epoch": 1.1931959731269526,
      "grad_norm": 0.5362989632073789,
      "learning_rate": 7.387825801007359e-05,
      "loss": 11.9254,
      "step": 21912
    },
    {
      "epoch": 1.1932504271235358,
      "grad_norm": 0.59063688744146,
      "learning_rate": 7.38697460234935e-05,
      "loss": 12.0482,
      "step": 21913
    },
    {
      "epoch": 1.1933048811201188,
      "grad_norm": 0.538421744485904,
      "learning_rate": 7.386123424010667e-05,
      "loss": 11.9666,
      "step": 21914
    },
    {
      "epoch": 1.1933593351167018,
      "grad_norm": 0.5636419177521339,
      "learning_rate": 7.385272265997924e-05,
      "loss": 11.8589,
      "step": 21915
    },
    {
      "epoch": 1.1934137891132848,
      "grad_norm": 0.5653417044781114,
      "learning_rate": 7.384421128317746e-05,
      "loss": 11.9675,
      "step": 21916
    },
    {
      "epoch": 1.1934682431098678,
      "grad_norm": 0.5327041947186542,
      "learning_rate": 7.383570010976746e-05,
      "loss": 11.9112,
      "step": 21917
    },
    {
      "epoch": 1.1935226971064508,
      "grad_norm": 0.5864609024597708,
      "learning_rate": 7.382718913981543e-05,
      "loss": 11.9947,
      "step": 21918
    },
    {
      "epoch": 1.1935771511030338,
      "grad_norm": 0.6286781208678304,
      "learning_rate": 7.381867837338758e-05,
      "loss": 11.9866,
      "step": 21919
    },
    {
      "epoch": 1.1936316050996167,
      "grad_norm": 0.5571969791110689,
      "learning_rate": 7.381016781055007e-05,
      "loss": 11.8934,
      "step": 21920
    },
    {
      "epoch": 1.1936860590961997,
      "grad_norm": 0.5391164813349384,
      "learning_rate": 7.380165745136914e-05,
      "loss": 11.9263,
      "step": 21921
    },
    {
      "epoch": 1.1937405130927827,
      "grad_norm": 0.5685581972310029,
      "learning_rate": 7.379314729591086e-05,
      "loss": 11.9015,
      "step": 21922
    },
    {
      "epoch": 1.1937949670893657,
      "grad_norm": 0.5347922702222191,
      "learning_rate": 7.378463734424148e-05,
      "loss": 11.9034,
      "step": 21923
    },
    {
      "epoch": 1.193849421085949,
      "grad_norm": 0.5334756761875965,
      "learning_rate": 7.377612759642714e-05,
      "loss": 11.8916,
      "step": 21924
    },
    {
      "epoch": 1.193903875082532,
      "grad_norm": 0.5378746969796901,
      "learning_rate": 7.376761805253402e-05,
      "loss": 11.9487,
      "step": 21925
    },
    {
      "epoch": 1.193958329079115,
      "grad_norm": 0.5632364163634869,
      "learning_rate": 7.375910871262832e-05,
      "loss": 11.9742,
      "step": 21926
    },
    {
      "epoch": 1.194012783075698,
      "grad_norm": 0.5509891768269785,
      "learning_rate": 7.37505995767762e-05,
      "loss": 11.9957,
      "step": 21927
    },
    {
      "epoch": 1.194067237072281,
      "grad_norm": 0.5028240155737436,
      "learning_rate": 7.37420906450438e-05,
      "loss": 11.9756,
      "step": 21928
    },
    {
      "epoch": 1.194121691068864,
      "grad_norm": 0.5700726798776007,
      "learning_rate": 7.373358191749732e-05,
      "loss": 11.9226,
      "step": 21929
    },
    {
      "epoch": 1.194176145065447,
      "grad_norm": 0.640326226880041,
      "learning_rate": 7.372507339420291e-05,
      "loss": 12.0049,
      "step": 21930
    },
    {
      "epoch": 1.1942305990620299,
      "grad_norm": 0.5734539135195217,
      "learning_rate": 7.371656507522676e-05,
      "loss": 12.0835,
      "step": 21931
    },
    {
      "epoch": 1.1942850530586129,
      "grad_norm": 0.5266682529185589,
      "learning_rate": 7.370805696063499e-05,
      "loss": 11.9226,
      "step": 21932
    },
    {
      "epoch": 1.1943395070551959,
      "grad_norm": 0.5321589207592013,
      "learning_rate": 7.369954905049376e-05,
      "loss": 11.8316,
      "step": 21933
    },
    {
      "epoch": 1.1943939610517789,
      "grad_norm": 0.6018036908920806,
      "learning_rate": 7.369104134486926e-05,
      "loss": 11.9235,
      "step": 21934
    },
    {
      "epoch": 1.1944484150483619,
      "grad_norm": 0.5212117869171325,
      "learning_rate": 7.368253384382761e-05,
      "loss": 12.0144,
      "step": 21935
    },
    {
      "epoch": 1.1945028690449448,
      "grad_norm": 0.55760427495541,
      "learning_rate": 7.367402654743503e-05,
      "loss": 11.8912,
      "step": 21936
    },
    {
      "epoch": 1.194557323041528,
      "grad_norm": 0.5723866896650784,
      "learning_rate": 7.36655194557576e-05,
      "loss": 12.0242,
      "step": 21937
    },
    {
      "epoch": 1.194611777038111,
      "grad_norm": 0.6439026464612314,
      "learning_rate": 7.365701256886152e-05,
      "loss": 12.0879,
      "step": 21938
    },
    {
      "epoch": 1.194666231034694,
      "grad_norm": 0.6360830880730395,
      "learning_rate": 7.364850588681293e-05,
      "loss": 11.77,
      "step": 21939
    },
    {
      "epoch": 1.194720685031277,
      "grad_norm": 0.5790733971628051,
      "learning_rate": 7.363999940967799e-05,
      "loss": 12.0057,
      "step": 21940
    },
    {
      "epoch": 1.19477513902786,
      "grad_norm": 0.5885949248156588,
      "learning_rate": 7.363149313752284e-05,
      "loss": 11.979,
      "step": 21941
    },
    {
      "epoch": 1.194829593024443,
      "grad_norm": 0.5387787434352426,
      "learning_rate": 7.362298707041361e-05,
      "loss": 11.9989,
      "step": 21942
    },
    {
      "epoch": 1.194884047021026,
      "grad_norm": 0.5671488692927811,
      "learning_rate": 7.361448120841645e-05,
      "loss": 11.9839,
      "step": 21943
    },
    {
      "epoch": 1.194938501017609,
      "grad_norm": 0.5458994071998429,
      "learning_rate": 7.360597555159752e-05,
      "loss": 12.0155,
      "step": 21944
    },
    {
      "epoch": 1.194992955014192,
      "grad_norm": 0.5285626015856262,
      "learning_rate": 7.359747010002294e-05,
      "loss": 11.8917,
      "step": 21945
    },
    {
      "epoch": 1.195047409010775,
      "grad_norm": 0.5591984973781156,
      "learning_rate": 7.358896485375883e-05,
      "loss": 11.9563,
      "step": 21946
    },
    {
      "epoch": 1.1951018630073582,
      "grad_norm": 0.5750925795714951,
      "learning_rate": 7.358045981287141e-05,
      "loss": 12.0029,
      "step": 21947
    },
    {
      "epoch": 1.1951563170039412,
      "grad_norm": 0.5418965446382998,
      "learning_rate": 7.357195497742673e-05,
      "loss": 11.9467,
      "step": 21948
    },
    {
      "epoch": 1.1952107710005242,
      "grad_norm": 0.539027786051126,
      "learning_rate": 7.356345034749098e-05,
      "loss": 11.9311,
      "step": 21949
    },
    {
      "epoch": 1.1952652249971072,
      "grad_norm": 0.5456273462070516,
      "learning_rate": 7.355494592313026e-05,
      "loss": 12.0312,
      "step": 21950
    },
    {
      "epoch": 1.1953196789936902,
      "grad_norm": 0.5476710821233999,
      "learning_rate": 7.354644170441075e-05,
      "loss": 11.9479,
      "step": 21951
    },
    {
      "epoch": 1.1953741329902732,
      "grad_norm": 0.5823357085415194,
      "learning_rate": 7.353793769139851e-05,
      "loss": 12.0686,
      "step": 21952
    },
    {
      "epoch": 1.1954285869868562,
      "grad_norm": 0.5753955765326135,
      "learning_rate": 7.352943388415973e-05,
      "loss": 12.0287,
      "step": 21953
    },
    {
      "epoch": 1.1954830409834392,
      "grad_norm": 0.5315100817541516,
      "learning_rate": 7.35209302827605e-05,
      "loss": 11.9147,
      "step": 21954
    },
    {
      "epoch": 1.1955374949800222,
      "grad_norm": 0.541929125728084,
      "learning_rate": 7.351242688726693e-05,
      "loss": 11.989,
      "step": 21955
    },
    {
      "epoch": 1.1955919489766051,
      "grad_norm": 0.58048912274076,
      "learning_rate": 7.350392369774521e-05,
      "loss": 12.0564,
      "step": 21956
    },
    {
      "epoch": 1.1956464029731881,
      "grad_norm": 0.5415460020106003,
      "learning_rate": 7.34954207142614e-05,
      "loss": 11.9536,
      "step": 21957
    },
    {
      "epoch": 1.1957008569697711,
      "grad_norm": 0.5395732840520236,
      "learning_rate": 7.348691793688162e-05,
      "loss": 12.0654,
      "step": 21958
    },
    {
      "epoch": 1.1957553109663541,
      "grad_norm": 0.5234919368691676,
      "learning_rate": 7.347841536567205e-05,
      "loss": 11.7763,
      "step": 21959
    },
    {
      "epoch": 1.1958097649629373,
      "grad_norm": 0.5268100722474017,
      "learning_rate": 7.346991300069876e-05,
      "loss": 11.9582,
      "step": 21960
    },
    {
      "epoch": 1.1958642189595203,
      "grad_norm": 0.6006650936852125,
      "learning_rate": 7.346141084202787e-05,
      "loss": 12.0406,
      "step": 21961
    },
    {
      "epoch": 1.1959186729561033,
      "grad_norm": 0.5727018594013691,
      "learning_rate": 7.345290888972554e-05,
      "loss": 12.0073,
      "step": 21962
    },
    {
      "epoch": 1.1959731269526863,
      "grad_norm": 0.5304001919462159,
      "learning_rate": 7.344440714385781e-05,
      "loss": 12.0074,
      "step": 21963
    },
    {
      "epoch": 1.1960275809492693,
      "grad_norm": 0.5307128137436281,
      "learning_rate": 7.34359056044908e-05,
      "loss": 12.0686,
      "step": 21964
    },
    {
      "epoch": 1.1960820349458523,
      "grad_norm": 0.6126776099286598,
      "learning_rate": 7.342740427169068e-05,
      "loss": 12.0333,
      "step": 21965
    },
    {
      "epoch": 1.1961364889424353,
      "grad_norm": 0.5468384048153477,
      "learning_rate": 7.34189031455235e-05,
      "loss": 11.9079,
      "step": 21966
    },
    {
      "epoch": 1.1961909429390183,
      "grad_norm": 0.5427062902505094,
      "learning_rate": 7.341040222605539e-05,
      "loss": 11.9996,
      "step": 21967
    },
    {
      "epoch": 1.1962453969356013,
      "grad_norm": 0.5923454382568167,
      "learning_rate": 7.340190151335245e-05,
      "loss": 12.1062,
      "step": 21968
    },
    {
      "epoch": 1.1962998509321843,
      "grad_norm": 0.5215597390807284,
      "learning_rate": 7.339340100748078e-05,
      "loss": 12.0299,
      "step": 21969
    },
    {
      "epoch": 1.1963543049287675,
      "grad_norm": 0.5252418576237954,
      "learning_rate": 7.338490070850649e-05,
      "loss": 11.9611,
      "step": 21970
    },
    {
      "epoch": 1.1964087589253505,
      "grad_norm": 0.5642377871979222,
      "learning_rate": 7.337640061649566e-05,
      "loss": 12.0417,
      "step": 21971
    },
    {
      "epoch": 1.1964632129219335,
      "grad_norm": 0.5532303926034008,
      "learning_rate": 7.336790073151447e-05,
      "loss": 11.9488,
      "step": 21972
    },
    {
      "epoch": 1.1965176669185165,
      "grad_norm": 0.5866435522997976,
      "learning_rate": 7.335940105362888e-05,
      "loss": 12.0359,
      "step": 21973
    },
    {
      "epoch": 1.1965721209150995,
      "grad_norm": 0.5965417508798888,
      "learning_rate": 7.33509015829051e-05,
      "loss": 12.0077,
      "step": 21974
    },
    {
      "epoch": 1.1966265749116825,
      "grad_norm": 0.5517370879178353,
      "learning_rate": 7.334240231940914e-05,
      "loss": 11.9778,
      "step": 21975
    },
    {
      "epoch": 1.1966810289082654,
      "grad_norm": 0.5736355459938184,
      "learning_rate": 7.333390326320715e-05,
      "loss": 11.9716,
      "step": 21976
    },
    {
      "epoch": 1.1967354829048484,
      "grad_norm": 0.6115971430366769,
      "learning_rate": 7.332540441436519e-05,
      "loss": 12.0461,
      "step": 21977
    },
    {
      "epoch": 1.1967899369014314,
      "grad_norm": 0.5074682437089225,
      "learning_rate": 7.331690577294936e-05,
      "loss": 11.7933,
      "step": 21978
    },
    {
      "epoch": 1.1968443908980144,
      "grad_norm": 0.5400402430781936,
      "learning_rate": 7.330840733902575e-05,
      "loss": 11.9894,
      "step": 21979
    },
    {
      "epoch": 1.1968988448945974,
      "grad_norm": 0.5863845049108035,
      "learning_rate": 7.329990911266043e-05,
      "loss": 11.9397,
      "step": 21980
    },
    {
      "epoch": 1.1969532988911804,
      "grad_norm": 0.5672006204768686,
      "learning_rate": 7.32914110939195e-05,
      "loss": 12.021,
      "step": 21981
    },
    {
      "epoch": 1.1970077528877634,
      "grad_norm": 0.5649091205380448,
      "learning_rate": 7.32829132828691e-05,
      "loss": 11.9162,
      "step": 21982
    },
    {
      "epoch": 1.1970622068843466,
      "grad_norm": 0.5193042287535758,
      "learning_rate": 7.327441567957518e-05,
      "loss": 11.8243,
      "step": 21983
    },
    {
      "epoch": 1.1971166608809296,
      "grad_norm": 0.6211147748830858,
      "learning_rate": 7.326591828410388e-05,
      "loss": 12.1122,
      "step": 21984
    },
    {
      "epoch": 1.1971711148775126,
      "grad_norm": 0.5599456933873733,
      "learning_rate": 7.32574210965213e-05,
      "loss": 12.0533,
      "step": 21985
    },
    {
      "epoch": 1.1972255688740956,
      "grad_norm": 0.584362125676465,
      "learning_rate": 7.324892411689348e-05,
      "loss": 11.9424,
      "step": 21986
    },
    {
      "epoch": 1.1972800228706786,
      "grad_norm": 0.6199258810366467,
      "learning_rate": 7.324042734528653e-05,
      "loss": 11.9043,
      "step": 21987
    },
    {
      "epoch": 1.1973344768672616,
      "grad_norm": 0.566877086814113,
      "learning_rate": 7.32319307817665e-05,
      "loss": 11.9762,
      "step": 21988
    },
    {
      "epoch": 1.1973889308638446,
      "grad_norm": 0.5703254437265531,
      "learning_rate": 7.322343442639948e-05,
      "loss": 12.0712,
      "step": 21989
    },
    {
      "epoch": 1.1974433848604276,
      "grad_norm": 0.6513271552859072,
      "learning_rate": 7.32149382792515e-05,
      "loss": 11.9653,
      "step": 21990
    },
    {
      "epoch": 1.1974978388570106,
      "grad_norm": 0.5497910689981529,
      "learning_rate": 7.320644234038865e-05,
      "loss": 12.0185,
      "step": 21991
    },
    {
      "epoch": 1.1975522928535935,
      "grad_norm": 0.543022473176367,
      "learning_rate": 7.319794660987704e-05,
      "loss": 12.0448,
      "step": 21992
    },
    {
      "epoch": 1.1976067468501765,
      "grad_norm": 0.5516768398291159,
      "learning_rate": 7.318945108778267e-05,
      "loss": 12.062,
      "step": 21993
    },
    {
      "epoch": 1.1976612008467598,
      "grad_norm": 0.5261862423612065,
      "learning_rate": 7.318095577417161e-05,
      "loss": 11.9819,
      "step": 21994
    },
    {
      "epoch": 1.1977156548433427,
      "grad_norm": 0.5705506301206956,
      "learning_rate": 7.317246066910992e-05,
      "loss": 11.9227,
      "step": 21995
    },
    {
      "epoch": 1.1977701088399257,
      "grad_norm": 0.5288704172122437,
      "learning_rate": 7.31639657726637e-05,
      "loss": 12.0418,
      "step": 21996
    },
    {
      "epoch": 1.1978245628365087,
      "grad_norm": 0.5464414712091592,
      "learning_rate": 7.315547108489897e-05,
      "loss": 11.859,
      "step": 21997
    },
    {
      "epoch": 1.1978790168330917,
      "grad_norm": 0.5154675847793856,
      "learning_rate": 7.314697660588181e-05,
      "loss": 11.9805,
      "step": 21998
    },
    {
      "epoch": 1.1979334708296747,
      "grad_norm": 0.54948897474018,
      "learning_rate": 7.313848233567826e-05,
      "loss": 11.9353,
      "step": 21999
    },
    {
      "epoch": 1.1979879248262577,
      "grad_norm": 0.5794043846753852,
      "learning_rate": 7.312998827435438e-05,
      "loss": 11.8999,
      "step": 22000
    },
    {
      "epoch": 1.1980423788228407,
      "grad_norm": 0.5547427345883039,
      "learning_rate": 7.312149442197623e-05,
      "loss": 11.9735,
      "step": 22001
    },
    {
      "epoch": 1.1980968328194237,
      "grad_norm": 0.5325978083539448,
      "learning_rate": 7.311300077860986e-05,
      "loss": 12.0776,
      "step": 22002
    },
    {
      "epoch": 1.1981512868160067,
      "grad_norm": 0.6224752343369995,
      "learning_rate": 7.310450734432125e-05,
      "loss": 12.2011,
      "step": 22003
    },
    {
      "epoch": 1.1982057408125897,
      "grad_norm": 0.5198020646625031,
      "learning_rate": 7.309601411917655e-05,
      "loss": 11.8813,
      "step": 22004
    },
    {
      "epoch": 1.1982601948091727,
      "grad_norm": 0.5493944469541123,
      "learning_rate": 7.308752110324173e-05,
      "loss": 11.9463,
      "step": 22005
    },
    {
      "epoch": 1.1983146488057557,
      "grad_norm": 0.6173523081537328,
      "learning_rate": 7.307902829658286e-05,
      "loss": 12.0457,
      "step": 22006
    },
    {
      "epoch": 1.1983691028023389,
      "grad_norm": 0.567660112391515,
      "learning_rate": 7.307053569926597e-05,
      "loss": 12.0803,
      "step": 22007
    },
    {
      "epoch": 1.1984235567989219,
      "grad_norm": 0.5880222453600811,
      "learning_rate": 7.30620433113571e-05,
      "loss": 12.0014,
      "step": 22008
    },
    {
      "epoch": 1.1984780107955049,
      "grad_norm": 0.5611166320327582,
      "learning_rate": 7.305355113292233e-05,
      "loss": 11.9751,
      "step": 22009
    },
    {
      "epoch": 1.1985324647920879,
      "grad_norm": 0.5648240770131326,
      "learning_rate": 7.304505916402766e-05,
      "loss": 12.0712,
      "step": 22010
    },
    {
      "epoch": 1.1985869187886709,
      "grad_norm": 0.5093535437447665,
      "learning_rate": 7.303656740473914e-05,
      "loss": 11.9333,
      "step": 22011
    },
    {
      "epoch": 1.1986413727852538,
      "grad_norm": 0.5327392268729846,
      "learning_rate": 7.302807585512281e-05,
      "loss": 11.957,
      "step": 22012
    },
    {
      "epoch": 1.1986958267818368,
      "grad_norm": 0.559389433152678,
      "learning_rate": 7.301958451524464e-05,
      "loss": 12.0161,
      "step": 22013
    },
    {
      "epoch": 1.1987502807784198,
      "grad_norm": 0.5470471857605839,
      "learning_rate": 7.301109338517074e-05,
      "loss": 11.9553,
      "step": 22014
    },
    {
      "epoch": 1.1988047347750028,
      "grad_norm": 0.47837328899612946,
      "learning_rate": 7.300260246496708e-05,
      "loss": 11.8601,
      "step": 22015
    },
    {
      "epoch": 1.1988591887715858,
      "grad_norm": 0.5899237141298945,
      "learning_rate": 7.299411175469972e-05,
      "loss": 11.977,
      "step": 22016
    },
    {
      "epoch": 1.198913642768169,
      "grad_norm": 0.5375233053095325,
      "learning_rate": 7.298562125443466e-05,
      "loss": 11.9505,
      "step": 22017
    },
    {
      "epoch": 1.198968096764752,
      "grad_norm": 0.48675226641449054,
      "learning_rate": 7.297713096423794e-05,
      "loss": 12.0649,
      "step": 22018
    },
    {
      "epoch": 1.199022550761335,
      "grad_norm": 0.5555814473334991,
      "learning_rate": 7.296864088417559e-05,
      "loss": 12.0215,
      "step": 22019
    },
    {
      "epoch": 1.199077004757918,
      "grad_norm": 0.5859006106638064,
      "learning_rate": 7.296015101431362e-05,
      "loss": 12.0142,
      "step": 22020
    },
    {
      "epoch": 1.199131458754501,
      "grad_norm": 0.5139311204066148,
      "learning_rate": 7.295166135471807e-05,
      "loss": 11.8954,
      "step": 22021
    },
    {
      "epoch": 1.199185912751084,
      "grad_norm": 0.55179853803515,
      "learning_rate": 7.294317190545494e-05,
      "loss": 12.0594,
      "step": 22022
    },
    {
      "epoch": 1.199240366747667,
      "grad_norm": 0.5353701993641484,
      "learning_rate": 7.293468266659023e-05,
      "loss": 11.9732,
      "step": 22023
    },
    {
      "epoch": 1.19929482074425,
      "grad_norm": 0.5464661093525545,
      "learning_rate": 7.292619363818995e-05,
      "loss": 11.9928,
      "step": 22024
    },
    {
      "epoch": 1.199349274740833,
      "grad_norm": 0.6882576058037276,
      "learning_rate": 7.291770482032014e-05,
      "loss": 12.1068,
      "step": 22025
    },
    {
      "epoch": 1.199403728737416,
      "grad_norm": 0.5362359921119237,
      "learning_rate": 7.29092162130468e-05,
      "loss": 11.9675,
      "step": 22026
    },
    {
      "epoch": 1.199458182733999,
      "grad_norm": 0.560517441415461,
      "learning_rate": 7.290072781643595e-05,
      "loss": 11.9185,
      "step": 22027
    },
    {
      "epoch": 1.199512636730582,
      "grad_norm": 0.57787623729744,
      "learning_rate": 7.289223963055357e-05,
      "loss": 12.0093,
      "step": 22028
    },
    {
      "epoch": 1.199567090727165,
      "grad_norm": 0.5177218457333005,
      "learning_rate": 7.288375165546567e-05,
      "loss": 11.8983,
      "step": 22029
    },
    {
      "epoch": 1.1996215447237482,
      "grad_norm": 0.637719324734241,
      "learning_rate": 7.287526389123827e-05,
      "loss": 12.0996,
      "step": 22030
    },
    {
      "epoch": 1.1996759987203311,
      "grad_norm": 0.5255124147758452,
      "learning_rate": 7.286677633793737e-05,
      "loss": 11.9215,
      "step": 22031
    },
    {
      "epoch": 1.1997304527169141,
      "grad_norm": 0.48575137124148565,
      "learning_rate": 7.285828899562902e-05,
      "loss": 11.8506,
      "step": 22032
    },
    {
      "epoch": 1.1997849067134971,
      "grad_norm": 0.6130569886618931,
      "learning_rate": 7.28498018643791e-05,
      "loss": 11.9346,
      "step": 22033
    },
    {
      "epoch": 1.1998393607100801,
      "grad_norm": 0.5802945288033231,
      "learning_rate": 7.284131494425369e-05,
      "loss": 12.0002,
      "step": 22034
    },
    {
      "epoch": 1.1998938147066631,
      "grad_norm": 0.5819325035949174,
      "learning_rate": 7.283282823531877e-05,
      "loss": 12.0333,
      "step": 22035
    },
    {
      "epoch": 1.1999482687032461,
      "grad_norm": 0.5236939960500671,
      "learning_rate": 7.282434173764035e-05,
      "loss": 11.9424,
      "step": 22036
    },
    {
      "epoch": 1.200002722699829,
      "grad_norm": 0.5966003358843548,
      "learning_rate": 7.281585545128438e-05,
      "loss": 11.9131,
      "step": 22037
    },
    {
      "epoch": 1.200057176696412,
      "grad_norm": 0.5546101784497532,
      "learning_rate": 7.28073693763169e-05,
      "loss": 12.0133,
      "step": 22038
    },
    {
      "epoch": 1.200111630692995,
      "grad_norm": 0.5156972528773146,
      "learning_rate": 7.279888351280386e-05,
      "loss": 12.0682,
      "step": 22039
    },
    {
      "epoch": 1.2001660846895783,
      "grad_norm": 0.5116502255674945,
      "learning_rate": 7.279039786081124e-05,
      "loss": 12.0056,
      "step": 22040
    },
    {
      "epoch": 1.2002205386861613,
      "grad_norm": 0.55062344944069,
      "learning_rate": 7.278191242040508e-05,
      "loss": 12.006,
      "step": 22041
    },
    {
      "epoch": 1.2002749926827443,
      "grad_norm": 0.5618602417553507,
      "learning_rate": 7.277342719165137e-05,
      "loss": 12.0006,
      "step": 22042
    },
    {
      "epoch": 1.2003294466793273,
      "grad_norm": 0.5445278229766826,
      "learning_rate": 7.276494217461602e-05,
      "loss": 12.0291,
      "step": 22043
    },
    {
      "epoch": 1.2003839006759103,
      "grad_norm": 0.600400346533125,
      "learning_rate": 7.275645736936503e-05,
      "loss": 12.0862,
      "step": 22044
    },
    {
      "epoch": 1.2004383546724933,
      "grad_norm": 0.568299995987044,
      "learning_rate": 7.27479727759644e-05,
      "loss": 12.0462,
      "step": 22045
    },
    {
      "epoch": 1.2004928086690763,
      "grad_norm": 0.5273174219750538,
      "learning_rate": 7.273948839448011e-05,
      "loss": 12.0002,
      "step": 22046
    },
    {
      "epoch": 1.2005472626656593,
      "grad_norm": 0.568077861293959,
      "learning_rate": 7.273100422497813e-05,
      "loss": 11.8807,
      "step": 22047
    },
    {
      "epoch": 1.2006017166622422,
      "grad_norm": 0.5055822204763117,
      "learning_rate": 7.272252026752444e-05,
      "loss": 11.9585,
      "step": 22048
    },
    {
      "epoch": 1.2006561706588252,
      "grad_norm": 0.5276861482917721,
      "learning_rate": 7.271403652218501e-05,
      "loss": 11.9588,
      "step": 22049
    },
    {
      "epoch": 1.2007106246554082,
      "grad_norm": 0.4972628498060447,
      "learning_rate": 7.27055529890258e-05,
      "loss": 11.8181,
      "step": 22050
    },
    {
      "epoch": 1.2007650786519912,
      "grad_norm": 0.6470355406641413,
      "learning_rate": 7.269706966811278e-05,
      "loss": 11.9526,
      "step": 22051
    },
    {
      "epoch": 1.2008195326485742,
      "grad_norm": 0.556983911002417,
      "learning_rate": 7.268858655951196e-05,
      "loss": 11.9277,
      "step": 22052
    },
    {
      "epoch": 1.2008739866451574,
      "grad_norm": 0.5454720828432393,
      "learning_rate": 7.268010366328926e-05,
      "loss": 11.9055,
      "step": 22053
    },
    {
      "epoch": 1.2009284406417404,
      "grad_norm": 0.5822493836480481,
      "learning_rate": 7.267162097951063e-05,
      "loss": 12.0546,
      "step": 22054
    },
    {
      "epoch": 1.2009828946383234,
      "grad_norm": 0.5324998231694357,
      "learning_rate": 7.266313850824209e-05,
      "loss": 11.9716,
      "step": 22055
    },
    {
      "epoch": 1.2010373486349064,
      "grad_norm": 0.5819054785121974,
      "learning_rate": 7.26546562495495e-05,
      "loss": 12.0384,
      "step": 22056
    },
    {
      "epoch": 1.2010918026314894,
      "grad_norm": 0.5133980500777363,
      "learning_rate": 7.264617420349895e-05,
      "loss": 11.9353,
      "step": 22057
    },
    {
      "epoch": 1.2011462566280724,
      "grad_norm": 0.5658558153461438,
      "learning_rate": 7.263769237015631e-05,
      "loss": 11.9319,
      "step": 22058
    },
    {
      "epoch": 1.2012007106246554,
      "grad_norm": 0.47454789496767735,
      "learning_rate": 7.26292107495876e-05,
      "loss": 11.975,
      "step": 22059
    },
    {
      "epoch": 1.2012551646212384,
      "grad_norm": 0.5440066283558274,
      "learning_rate": 7.262072934185871e-05,
      "loss": 11.9386,
      "step": 22060
    },
    {
      "epoch": 1.2013096186178214,
      "grad_norm": 0.5391034435579533,
      "learning_rate": 7.261224814703562e-05,
      "loss": 12.0825,
      "step": 22061
    },
    {
      "epoch": 1.2013640726144044,
      "grad_norm": 0.584625249614327,
      "learning_rate": 7.260376716518431e-05,
      "loss": 11.8243,
      "step": 22062
    },
    {
      "epoch": 1.2014185266109874,
      "grad_norm": 0.5094079871757436,
      "learning_rate": 7.259528639637068e-05,
      "loss": 12.0372,
      "step": 22063
    },
    {
      "epoch": 1.2014729806075706,
      "grad_norm": 0.7009737554586154,
      "learning_rate": 7.258680584066069e-05,
      "loss": 12.017,
      "step": 22064
    },
    {
      "epoch": 1.2015274346041536,
      "grad_norm": 0.5222235715445752,
      "learning_rate": 7.25783254981203e-05,
      "loss": 12.0611,
      "step": 22065
    },
    {
      "epoch": 1.2015818886007366,
      "grad_norm": 0.6015498109875128,
      "learning_rate": 7.256984536881545e-05,
      "loss": 11.9995,
      "step": 22066
    },
    {
      "epoch": 1.2016363425973196,
      "grad_norm": 0.5289823743578922,
      "learning_rate": 7.256136545281207e-05,
      "loss": 11.8069,
      "step": 22067
    },
    {
      "epoch": 1.2016907965939025,
      "grad_norm": 0.5015003605636306,
      "learning_rate": 7.255288575017612e-05,
      "loss": 11.9141,
      "step": 22068
    },
    {
      "epoch": 1.2017452505904855,
      "grad_norm": 0.5378912269149257,
      "learning_rate": 7.254440626097354e-05,
      "loss": 11.9026,
      "step": 22069
    },
    {
      "epoch": 1.2017997045870685,
      "grad_norm": 0.6546457360339119,
      "learning_rate": 7.253592698527025e-05,
      "loss": 11.9585,
      "step": 22070
    },
    {
      "epoch": 1.2018541585836515,
      "grad_norm": 0.5534231465095167,
      "learning_rate": 7.252744792313223e-05,
      "loss": 11.9595,
      "step": 22071
    },
    {
      "epoch": 1.2019086125802345,
      "grad_norm": 0.5249925229316503,
      "learning_rate": 7.251896907462537e-05,
      "loss": 11.9611,
      "step": 22072
    },
    {
      "epoch": 1.2019630665768175,
      "grad_norm": 0.5746386431126431,
      "learning_rate": 7.25104904398156e-05,
      "loss": 12.0867,
      "step": 22073
    },
    {
      "epoch": 1.2020175205734005,
      "grad_norm": 0.5710153979610945,
      "learning_rate": 7.250201201876888e-05,
      "loss": 11.9604,
      "step": 22074
    },
    {
      "epoch": 1.2020719745699835,
      "grad_norm": 0.5717775825974136,
      "learning_rate": 7.249353381155111e-05,
      "loss": 12.0821,
      "step": 22075
    },
    {
      "epoch": 1.2021264285665665,
      "grad_norm": 0.5998918160103405,
      "learning_rate": 7.248505581822825e-05,
      "loss": 12.0195,
      "step": 22076
    },
    {
      "epoch": 1.2021808825631497,
      "grad_norm": 0.5254032594783522,
      "learning_rate": 7.247657803886619e-05,
      "loss": 11.8963,
      "step": 22077
    },
    {
      "epoch": 1.2022353365597327,
      "grad_norm": 0.5537859402263627,
      "learning_rate": 7.246810047353087e-05,
      "loss": 11.9807,
      "step": 22078
    },
    {
      "epoch": 1.2022897905563157,
      "grad_norm": 0.5903931312597357,
      "learning_rate": 7.245962312228823e-05,
      "loss": 11.9017,
      "step": 22079
    },
    {
      "epoch": 1.2023442445528987,
      "grad_norm": 0.5417485964314547,
      "learning_rate": 7.245114598520419e-05,
      "loss": 11.9962,
      "step": 22080
    },
    {
      "epoch": 1.2023986985494817,
      "grad_norm": 0.5291549933040715,
      "learning_rate": 7.244266906234465e-05,
      "loss": 11.8927,
      "step": 22081
    },
    {
      "epoch": 1.2024531525460647,
      "grad_norm": 0.609881952281548,
      "learning_rate": 7.243419235377556e-05,
      "loss": 11.9386,
      "step": 22082
    },
    {
      "epoch": 1.2025076065426477,
      "grad_norm": 0.5623469278131573,
      "learning_rate": 7.242571585956279e-05,
      "loss": 11.9458,
      "step": 22083
    },
    {
      "epoch": 1.2025620605392306,
      "grad_norm": 0.547022649623105,
      "learning_rate": 7.241723957977229e-05,
      "loss": 11.9944,
      "step": 22084
    },
    {
      "epoch": 1.2026165145358136,
      "grad_norm": 0.5352473947782485,
      "learning_rate": 7.240876351446995e-05,
      "loss": 11.8539,
      "step": 22085
    },
    {
      "epoch": 1.2026709685323966,
      "grad_norm": 0.6625965168116763,
      "learning_rate": 7.240028766372168e-05,
      "loss": 12.0785,
      "step": 22086
    },
    {
      "epoch": 1.2027254225289798,
      "grad_norm": 0.718071770523711,
      "learning_rate": 7.239181202759342e-05,
      "loss": 12.1267,
      "step": 22087
    },
    {
      "epoch": 1.2027798765255628,
      "grad_norm": 0.5455341048641585,
      "learning_rate": 7.238333660615105e-05,
      "loss": 11.9934,
      "step": 22088
    },
    {
      "epoch": 1.2028343305221458,
      "grad_norm": 0.5202972777844755,
      "learning_rate": 7.237486139946046e-05,
      "loss": 11.9052,
      "step": 22089
    },
    {
      "epoch": 1.2028887845187288,
      "grad_norm": 0.595001783655921,
      "learning_rate": 7.236638640758761e-05,
      "loss": 12.0565,
      "step": 22090
    },
    {
      "epoch": 1.2029432385153118,
      "grad_norm": 0.5074587400834651,
      "learning_rate": 7.235791163059839e-05,
      "loss": 12.0901,
      "step": 22091
    },
    {
      "epoch": 1.2029976925118948,
      "grad_norm": 0.7264258267848539,
      "learning_rate": 7.23494370685587e-05,
      "loss": 11.8703,
      "step": 22092
    },
    {
      "epoch": 1.2030521465084778,
      "grad_norm": 0.5209063478833043,
      "learning_rate": 7.234096272153438e-05,
      "loss": 11.948,
      "step": 22093
    },
    {
      "epoch": 1.2031066005050608,
      "grad_norm": 0.5365000809312424,
      "learning_rate": 7.233248858959139e-05,
      "loss": 11.9564,
      "step": 22094
    },
    {
      "epoch": 1.2031610545016438,
      "grad_norm": 0.5794642254238294,
      "learning_rate": 7.232401467279559e-05,
      "loss": 11.9069,
      "step": 22095
    },
    {
      "epoch": 1.2032155084982268,
      "grad_norm": 0.5413611195574631,
      "learning_rate": 7.231554097121291e-05,
      "loss": 12.0539,
      "step": 22096
    },
    {
      "epoch": 1.2032699624948098,
      "grad_norm": 0.5224053850158283,
      "learning_rate": 7.230706748490923e-05,
      "loss": 11.934,
      "step": 22097
    },
    {
      "epoch": 1.2033244164913928,
      "grad_norm": 0.509226677967835,
      "learning_rate": 7.229859421395042e-05,
      "loss": 11.9446,
      "step": 22098
    },
    {
      "epoch": 1.2033788704879758,
      "grad_norm": 0.5373899283698169,
      "learning_rate": 7.22901211584024e-05,
      "loss": 12.1334,
      "step": 22099
    },
    {
      "epoch": 1.203433324484559,
      "grad_norm": 0.5349429933355131,
      "learning_rate": 7.228164831833102e-05,
      "loss": 11.9687,
      "step": 22100
    },
    {
      "epoch": 1.203487778481142,
      "grad_norm": 0.5473400868933365,
      "learning_rate": 7.22731756938022e-05,
      "loss": 12.0422,
      "step": 22101
    },
    {
      "epoch": 1.203542232477725,
      "grad_norm": 0.5280914671988042,
      "learning_rate": 7.226470328488184e-05,
      "loss": 11.9748,
      "step": 22102
    },
    {
      "epoch": 1.203596686474308,
      "grad_norm": 0.5490625735743139,
      "learning_rate": 7.225623109163584e-05,
      "loss": 11.9687,
      "step": 22103
    },
    {
      "epoch": 1.203651140470891,
      "grad_norm": 0.5523105319659944,
      "learning_rate": 7.224775911412996e-05,
      "loss": 11.7874,
      "step": 22104
    },
    {
      "epoch": 1.203705594467474,
      "grad_norm": 0.5269022681343605,
      "learning_rate": 7.223928735243019e-05,
      "loss": 11.9339,
      "step": 22105
    },
    {
      "epoch": 1.203760048464057,
      "grad_norm": 0.5386341462858246,
      "learning_rate": 7.223081580660236e-05,
      "loss": 12.1385,
      "step": 22106
    },
    {
      "epoch": 1.20381450246064,
      "grad_norm": 0.5050447802853469,
      "learning_rate": 7.222234447671239e-05,
      "loss": 11.9921,
      "step": 22107
    },
    {
      "epoch": 1.203868956457223,
      "grad_norm": 0.5736470925018637,
      "learning_rate": 7.22138733628261e-05,
      "loss": 12.0373,
      "step": 22108
    },
    {
      "epoch": 1.203923410453806,
      "grad_norm": 0.5264139579520998,
      "learning_rate": 7.22054024650094e-05,
      "loss": 11.9781,
      "step": 22109
    },
    {
      "epoch": 1.2039778644503891,
      "grad_norm": 0.5299022741525633,
      "learning_rate": 7.219693178332816e-05,
      "loss": 11.8533,
      "step": 22110
    },
    {
      "epoch": 1.2040323184469721,
      "grad_norm": 0.5799698616022124,
      "learning_rate": 7.218846131784824e-05,
      "loss": 12.0601,
      "step": 22111
    },
    {
      "epoch": 1.204086772443555,
      "grad_norm": 0.5806676006289738,
      "learning_rate": 7.217999106863549e-05,
      "loss": 11.9913,
      "step": 22112
    },
    {
      "epoch": 1.204141226440138,
      "grad_norm": 0.5501912701320303,
      "learning_rate": 7.217152103575584e-05,
      "loss": 11.9818,
      "step": 22113
    },
    {
      "epoch": 1.204195680436721,
      "grad_norm": 0.5655415207289628,
      "learning_rate": 7.216305121927508e-05,
      "loss": 11.9584,
      "step": 22114
    },
    {
      "epoch": 1.204250134433304,
      "grad_norm": 0.592532109262495,
      "learning_rate": 7.21545816192591e-05,
      "loss": 11.9461,
      "step": 22115
    },
    {
      "epoch": 1.204304588429887,
      "grad_norm": 0.48483512667346723,
      "learning_rate": 7.214611223577375e-05,
      "loss": 12.1258,
      "step": 22116
    },
    {
      "epoch": 1.20435904242647,
      "grad_norm": 0.6021119418504816,
      "learning_rate": 7.213764306888492e-05,
      "loss": 11.9288,
      "step": 22117
    },
    {
      "epoch": 1.204413496423053,
      "grad_norm": 0.5611011265865317,
      "learning_rate": 7.212917411865844e-05,
      "loss": 12.0514,
      "step": 22118
    },
    {
      "epoch": 1.204467950419636,
      "grad_norm": 0.5305076937015334,
      "learning_rate": 7.212070538516017e-05,
      "loss": 11.9955,
      "step": 22119
    },
    {
      "epoch": 1.204522404416219,
      "grad_norm": 0.5534774058940248,
      "learning_rate": 7.2112236868456e-05,
      "loss": 12.0304,
      "step": 22120
    },
    {
      "epoch": 1.204576858412802,
      "grad_norm": 0.5845141801582963,
      "learning_rate": 7.210376856861175e-05,
      "loss": 12.0212,
      "step": 22121
    },
    {
      "epoch": 1.204631312409385,
      "grad_norm": 0.5383471489402331,
      "learning_rate": 7.209530048569325e-05,
      "loss": 11.99,
      "step": 22122
    },
    {
      "epoch": 1.2046857664059683,
      "grad_norm": 0.5803132674896115,
      "learning_rate": 7.208683261976641e-05,
      "loss": 12.0276,
      "step": 22123
    },
    {
      "epoch": 1.2047402204025512,
      "grad_norm": 0.5633793886050223,
      "learning_rate": 7.207836497089701e-05,
      "loss": 12.0332,
      "step": 22124
    },
    {
      "epoch": 1.2047946743991342,
      "grad_norm": 0.5597780825700696,
      "learning_rate": 7.206989753915092e-05,
      "loss": 12.1111,
      "step": 22125
    },
    {
      "epoch": 1.2048491283957172,
      "grad_norm": 0.5462479356952539,
      "learning_rate": 7.2061430324594e-05,
      "loss": 12.0149,
      "step": 22126
    },
    {
      "epoch": 1.2049035823923002,
      "grad_norm": 0.5657041967252572,
      "learning_rate": 7.205296332729206e-05,
      "loss": 11.9446,
      "step": 22127
    },
    {
      "epoch": 1.2049580363888832,
      "grad_norm": 0.560924352421728,
      "learning_rate": 7.2044496547311e-05,
      "loss": 12.0223,
      "step": 22128
    },
    {
      "epoch": 1.2050124903854662,
      "grad_norm": 0.5336889827650598,
      "learning_rate": 7.203602998471661e-05,
      "loss": 11.9153,
      "step": 22129
    },
    {
      "epoch": 1.2050669443820492,
      "grad_norm": 0.5347407201619592,
      "learning_rate": 7.202756363957473e-05,
      "loss": 11.9628,
      "step": 22130
    },
    {
      "epoch": 1.2051213983786322,
      "grad_norm": 0.5443814156617037,
      "learning_rate": 7.201909751195122e-05,
      "loss": 11.9808,
      "step": 22131
    },
    {
      "epoch": 1.2051758523752152,
      "grad_norm": 0.5346787885470732,
      "learning_rate": 7.201063160191191e-05,
      "loss": 11.9713,
      "step": 22132
    },
    {
      "epoch": 1.2052303063717984,
      "grad_norm": 0.5654284677731318,
      "learning_rate": 7.200216590952262e-05,
      "loss": 11.8919,
      "step": 22133
    },
    {
      "epoch": 1.2052847603683814,
      "grad_norm": 0.5169820110637174,
      "learning_rate": 7.199370043484917e-05,
      "loss": 11.9123,
      "step": 22134
    },
    {
      "epoch": 1.2053392143649644,
      "grad_norm": 0.5622142007958477,
      "learning_rate": 7.198523517795741e-05,
      "loss": 11.9989,
      "step": 22135
    },
    {
      "epoch": 1.2053936683615474,
      "grad_norm": 0.5471605534669074,
      "learning_rate": 7.197677013891315e-05,
      "loss": 11.9656,
      "step": 22136
    },
    {
      "epoch": 1.2054481223581304,
      "grad_norm": 0.555817997447976,
      "learning_rate": 7.196830531778222e-05,
      "loss": 12.0999,
      "step": 22137
    },
    {
      "epoch": 1.2055025763547134,
      "grad_norm": 0.5182694496488273,
      "learning_rate": 7.195984071463045e-05,
      "loss": 11.8154,
      "step": 22138
    },
    {
      "epoch": 1.2055570303512964,
      "grad_norm": 0.6090367274971842,
      "learning_rate": 7.195137632952367e-05,
      "loss": 12.0171,
      "step": 22139
    },
    {
      "epoch": 1.2056114843478793,
      "grad_norm": 0.529902810752365,
      "learning_rate": 7.194291216252769e-05,
      "loss": 11.9217,
      "step": 22140
    },
    {
      "epoch": 1.2056659383444623,
      "grad_norm": 0.6266915580456607,
      "learning_rate": 7.193444821370833e-05,
      "loss": 12.1702,
      "step": 22141
    },
    {
      "epoch": 1.2057203923410453,
      "grad_norm": 0.5324727183403394,
      "learning_rate": 7.192598448313141e-05,
      "loss": 12.0653,
      "step": 22142
    },
    {
      "epoch": 1.2057748463376283,
      "grad_norm": 0.5717854460507498,
      "learning_rate": 7.191752097086275e-05,
      "loss": 11.9419,
      "step": 22143
    },
    {
      "epoch": 1.2058293003342113,
      "grad_norm": 0.5500219032678746,
      "learning_rate": 7.190905767696816e-05,
      "loss": 11.9897,
      "step": 22144
    },
    {
      "epoch": 1.2058837543307943,
      "grad_norm": 0.6058155988404833,
      "learning_rate": 7.190059460151342e-05,
      "loss": 11.987,
      "step": 22145
    },
    {
      "epoch": 1.2059382083273773,
      "grad_norm": 0.5239431508726817,
      "learning_rate": 7.189213174456439e-05,
      "loss": 12.0315,
      "step": 22146
    },
    {
      "epoch": 1.2059926623239605,
      "grad_norm": 0.5485371883024152,
      "learning_rate": 7.188366910618684e-05,
      "loss": 11.9058,
      "step": 22147
    },
    {
      "epoch": 1.2060471163205435,
      "grad_norm": 0.5739704399628821,
      "learning_rate": 7.187520668644662e-05,
      "loss": 11.917,
      "step": 22148
    },
    {
      "epoch": 1.2061015703171265,
      "grad_norm": 0.5965933753043204,
      "learning_rate": 7.186674448540947e-05,
      "loss": 12.0307,
      "step": 22149
    },
    {
      "epoch": 1.2061560243137095,
      "grad_norm": 0.5515978531681733,
      "learning_rate": 7.185828250314126e-05,
      "loss": 12.0179,
      "step": 22150
    },
    {
      "epoch": 1.2062104783102925,
      "grad_norm": 0.5232068825150631,
      "learning_rate": 7.184982073970776e-05,
      "loss": 12.0243,
      "step": 22151
    },
    {
      "epoch": 1.2062649323068755,
      "grad_norm": 0.558244250566568,
      "learning_rate": 7.184135919517479e-05,
      "loss": 11.8912,
      "step": 22152
    },
    {
      "epoch": 1.2063193863034585,
      "grad_norm": 0.5243481543708333,
      "learning_rate": 7.183289786960813e-05,
      "loss": 12.0872,
      "step": 22153
    },
    {
      "epoch": 1.2063738403000415,
      "grad_norm": 0.5432256391753869,
      "learning_rate": 7.182443676307357e-05,
      "loss": 12.0196,
      "step": 22154
    },
    {
      "epoch": 1.2064282942966245,
      "grad_norm": 0.5113051679382367,
      "learning_rate": 7.181597587563691e-05,
      "loss": 11.969,
      "step": 22155
    },
    {
      "epoch": 1.2064827482932075,
      "grad_norm": 0.5187867259613202,
      "learning_rate": 7.180751520736395e-05,
      "loss": 11.9699,
      "step": 22156
    },
    {
      "epoch": 1.2065372022897907,
      "grad_norm": 0.536965619191231,
      "learning_rate": 7.17990547583205e-05,
      "loss": 11.9784,
      "step": 22157
    },
    {
      "epoch": 1.2065916562863737,
      "grad_norm": 0.6097347835717006,
      "learning_rate": 7.17905945285723e-05,
      "loss": 12.1263,
      "step": 22158
    },
    {
      "epoch": 1.2066461102829567,
      "grad_norm": 0.5507974083113052,
      "learning_rate": 7.178213451818519e-05,
      "loss": 11.9288,
      "step": 22159
    },
    {
      "epoch": 1.2067005642795396,
      "grad_norm": 0.5705949664999904,
      "learning_rate": 7.177367472722492e-05,
      "loss": 11.9414,
      "step": 22160
    },
    {
      "epoch": 1.2067550182761226,
      "grad_norm": 0.8062485571839195,
      "learning_rate": 7.176521515575725e-05,
      "loss": 11.9337,
      "step": 22161
    },
    {
      "epoch": 1.2068094722727056,
      "grad_norm": 0.5864169340682323,
      "learning_rate": 7.175675580384806e-05,
      "loss": 11.9364,
      "step": 22162
    },
    {
      "epoch": 1.2068639262692886,
      "grad_norm": 0.5337505307724522,
      "learning_rate": 7.17482966715631e-05,
      "loss": 11.997,
      "step": 22163
    },
    {
      "epoch": 1.2069183802658716,
      "grad_norm": 0.5436177172422525,
      "learning_rate": 7.173983775896807e-05,
      "loss": 12.0018,
      "step": 22164
    },
    {
      "epoch": 1.2069728342624546,
      "grad_norm": 0.5648042686347058,
      "learning_rate": 7.17313790661288e-05,
      "loss": 12.0179,
      "step": 22165
    },
    {
      "epoch": 1.2070272882590376,
      "grad_norm": 0.5700943268038059,
      "learning_rate": 7.172292059311108e-05,
      "loss": 12.0263,
      "step": 22166
    },
    {
      "epoch": 1.2070817422556206,
      "grad_norm": 0.5838511277616565,
      "learning_rate": 7.171446233998067e-05,
      "loss": 11.9884,
      "step": 22167
    },
    {
      "epoch": 1.2071361962522036,
      "grad_norm": 0.5887753007205588,
      "learning_rate": 7.170600430680335e-05,
      "loss": 12.0172,
      "step": 22168
    },
    {
      "epoch": 1.2071906502487866,
      "grad_norm": 0.5278608586343595,
      "learning_rate": 7.169754649364487e-05,
      "loss": 11.9377,
      "step": 22169
    },
    {
      "epoch": 1.2072451042453698,
      "grad_norm": 0.5147691660591388,
      "learning_rate": 7.168908890057102e-05,
      "loss": 11.9707,
      "step": 22170
    },
    {
      "epoch": 1.2072995582419528,
      "grad_norm": 0.5332753264935212,
      "learning_rate": 7.168063152764756e-05,
      "loss": 11.8652,
      "step": 22171
    },
    {
      "epoch": 1.2073540122385358,
      "grad_norm": 0.5971722709092365,
      "learning_rate": 7.167217437494024e-05,
      "loss": 11.9451,
      "step": 22172
    },
    {
      "epoch": 1.2074084662351188,
      "grad_norm": 0.5479329384980722,
      "learning_rate": 7.166371744251492e-05,
      "loss": 11.9568,
      "step": 22173
    },
    {
      "epoch": 1.2074629202317018,
      "grad_norm": 0.5476576886552974,
      "learning_rate": 7.165526073043723e-05,
      "loss": 11.9433,
      "step": 22174
    },
    {
      "epoch": 1.2075173742282848,
      "grad_norm": 0.5637277927302408,
      "learning_rate": 7.164680423877299e-05,
      "loss": 12.0464,
      "step": 22175
    },
    {
      "epoch": 1.2075718282248677,
      "grad_norm": 0.6468205685455182,
      "learning_rate": 7.163834796758794e-05,
      "loss": 12.0958,
      "step": 22176
    },
    {
      "epoch": 1.2076262822214507,
      "grad_norm": 0.5717610794146486,
      "learning_rate": 7.162989191694787e-05,
      "loss": 12.0327,
      "step": 22177
    },
    {
      "epoch": 1.2076807362180337,
      "grad_norm": 0.5335832001968199,
      "learning_rate": 7.162143608691851e-05,
      "loss": 12.0202,
      "step": 22178
    },
    {
      "epoch": 1.2077351902146167,
      "grad_norm": 0.6300336057485512,
      "learning_rate": 7.161298047756561e-05,
      "loss": 11.9623,
      "step": 22179
    },
    {
      "epoch": 1.2077896442112,
      "grad_norm": 0.7084449429386043,
      "learning_rate": 7.160452508895497e-05,
      "loss": 11.8882,
      "step": 22180
    },
    {
      "epoch": 1.207844098207783,
      "grad_norm": 0.6087104624616091,
      "learning_rate": 7.15960699211523e-05,
      "loss": 12.0099,
      "step": 22181
    },
    {
      "epoch": 1.207898552204366,
      "grad_norm": 0.550174407470673,
      "learning_rate": 7.158761497422331e-05,
      "loss": 11.918,
      "step": 22182
    },
    {
      "epoch": 1.207953006200949,
      "grad_norm": 0.5466120104557948,
      "learning_rate": 7.157916024823386e-05,
      "loss": 11.9994,
      "step": 22183
    },
    {
      "epoch": 1.208007460197532,
      "grad_norm": 0.541649744902221,
      "learning_rate": 7.157070574324957e-05,
      "loss": 12.0657,
      "step": 22184
    },
    {
      "epoch": 1.208061914194115,
      "grad_norm": 0.6087705722689702,
      "learning_rate": 7.156225145933625e-05,
      "loss": 11.9982,
      "step": 22185
    },
    {
      "epoch": 1.208116368190698,
      "grad_norm": 0.5725121504350018,
      "learning_rate": 7.155379739655965e-05,
      "loss": 12.0643,
      "step": 22186
    },
    {
      "epoch": 1.208170822187281,
      "grad_norm": 0.6066352387285385,
      "learning_rate": 7.154534355498545e-05,
      "loss": 12.0684,
      "step": 22187
    },
    {
      "epoch": 1.2082252761838639,
      "grad_norm": 0.57731847318181,
      "learning_rate": 7.153688993467946e-05,
      "loss": 11.9755,
      "step": 22188
    },
    {
      "epoch": 1.2082797301804469,
      "grad_norm": 0.6293896024435822,
      "learning_rate": 7.15284365357074e-05,
      "loss": 12.1648,
      "step": 22189
    },
    {
      "epoch": 1.2083341841770299,
      "grad_norm": 0.4920126587746231,
      "learning_rate": 7.151998335813497e-05,
      "loss": 12.0314,
      "step": 22190
    },
    {
      "epoch": 1.2083886381736129,
      "grad_norm": 0.5472210948432139,
      "learning_rate": 7.151153040202794e-05,
      "loss": 12.0244,
      "step": 22191
    },
    {
      "epoch": 1.2084430921701959,
      "grad_norm": 0.5024160976264329,
      "learning_rate": 7.150307766745202e-05,
      "loss": 11.8616,
      "step": 22192
    },
    {
      "epoch": 1.208497546166779,
      "grad_norm": 0.5253886343420642,
      "learning_rate": 7.149462515447296e-05,
      "loss": 11.9704,
      "step": 22193
    },
    {
      "epoch": 1.208552000163362,
      "grad_norm": 0.607827313687678,
      "learning_rate": 7.148617286315646e-05,
      "loss": 11.9671,
      "step": 22194
    },
    {
      "epoch": 1.208606454159945,
      "grad_norm": 0.5251209836074804,
      "learning_rate": 7.147772079356827e-05,
      "loss": 11.8412,
      "step": 22195
    },
    {
      "epoch": 1.208660908156528,
      "grad_norm": 0.67730418799882,
      "learning_rate": 7.14692689457741e-05,
      "loss": 12.0243,
      "step": 22196
    },
    {
      "epoch": 1.208715362153111,
      "grad_norm": 0.5731436442116857,
      "learning_rate": 7.14608173198397e-05,
      "loss": 11.9742,
      "step": 22197
    },
    {
      "epoch": 1.208769816149694,
      "grad_norm": 0.645929462247328,
      "learning_rate": 7.145236591583072e-05,
      "loss": 12.0492,
      "step": 22198
    },
    {
      "epoch": 1.208824270146277,
      "grad_norm": 0.6119640998397153,
      "learning_rate": 7.144391473381296e-05,
      "loss": 12.057,
      "step": 22199
    },
    {
      "epoch": 1.20887872414286,
      "grad_norm": 0.5738906898356876,
      "learning_rate": 7.143546377385211e-05,
      "loss": 12.0368,
      "step": 22200
    },
    {
      "epoch": 1.208933178139443,
      "grad_norm": 0.6179311492166093,
      "learning_rate": 7.142701303601388e-05,
      "loss": 11.8716,
      "step": 22201
    },
    {
      "epoch": 1.208987632136026,
      "grad_norm": 0.5294809640433394,
      "learning_rate": 7.1418562520364e-05,
      "loss": 11.9695,
      "step": 22202
    },
    {
      "epoch": 1.2090420861326092,
      "grad_norm": 0.5285947080877262,
      "learning_rate": 7.141011222696818e-05,
      "loss": 11.887,
      "step": 22203
    },
    {
      "epoch": 1.2090965401291922,
      "grad_norm": 0.5063114473792395,
      "learning_rate": 7.14016621558921e-05,
      "loss": 11.9013,
      "step": 22204
    },
    {
      "epoch": 1.2091509941257752,
      "grad_norm": 0.578170449929612,
      "learning_rate": 7.139321230720151e-05,
      "loss": 11.9329,
      "step": 22205
    },
    {
      "epoch": 1.2092054481223582,
      "grad_norm": 0.5967577483165094,
      "learning_rate": 7.138476268096208e-05,
      "loss": 12.0208,
      "step": 22206
    },
    {
      "epoch": 1.2092599021189412,
      "grad_norm": 0.5141582545857023,
      "learning_rate": 7.137631327723952e-05,
      "loss": 11.9588,
      "step": 22207
    },
    {
      "epoch": 1.2093143561155242,
      "grad_norm": 0.5621523592159323,
      "learning_rate": 7.136786409609957e-05,
      "loss": 11.9533,
      "step": 22208
    },
    {
      "epoch": 1.2093688101121072,
      "grad_norm": 0.5581547531966375,
      "learning_rate": 7.135941513760792e-05,
      "loss": 11.8932,
      "step": 22209
    },
    {
      "epoch": 1.2094232641086902,
      "grad_norm": 0.5387537878154146,
      "learning_rate": 7.135096640183022e-05,
      "loss": 12.1501,
      "step": 22210
    },
    {
      "epoch": 1.2094777181052732,
      "grad_norm": 0.6273124702833844,
      "learning_rate": 7.134251788883224e-05,
      "loss": 11.9529,
      "step": 22211
    },
    {
      "epoch": 1.2095321721018562,
      "grad_norm": 0.6198131612178198,
      "learning_rate": 7.133406959867965e-05,
      "loss": 12.0887,
      "step": 22212
    },
    {
      "epoch": 1.2095866260984391,
      "grad_norm": 0.5908403742200561,
      "learning_rate": 7.132562153143819e-05,
      "loss": 11.9164,
      "step": 22213
    },
    {
      "epoch": 1.2096410800950221,
      "grad_norm": 0.508816991417017,
      "learning_rate": 7.131717368717342e-05,
      "loss": 11.9498,
      "step": 22214
    },
    {
      "epoch": 1.2096955340916051,
      "grad_norm": 0.5326215958832874,
      "learning_rate": 7.130872606595114e-05,
      "loss": 11.9998,
      "step": 22215
    },
    {
      "epoch": 1.2097499880881883,
      "grad_norm": 0.5395288580198045,
      "learning_rate": 7.130027866783703e-05,
      "loss": 12.0351,
      "step": 22216
    },
    {
      "epoch": 1.2098044420847713,
      "grad_norm": 0.5535370494809941,
      "learning_rate": 7.129183149289677e-05,
      "loss": 12.0239,
      "step": 22217
    },
    {
      "epoch": 1.2098588960813543,
      "grad_norm": 0.5553002147851895,
      "learning_rate": 7.128338454119603e-05,
      "loss": 12.0336,
      "step": 22218
    },
    {
      "epoch": 1.2099133500779373,
      "grad_norm": 0.5585794211770257,
      "learning_rate": 7.127493781280052e-05,
      "loss": 11.8861,
      "step": 22219
    },
    {
      "epoch": 1.2099678040745203,
      "grad_norm": 0.5464541689008439,
      "learning_rate": 7.12664913077759e-05,
      "loss": 11.9994,
      "step": 22220
    },
    {
      "epoch": 1.2100222580711033,
      "grad_norm": 0.5440864382681991,
      "learning_rate": 7.125804502618784e-05,
      "loss": 12.0383,
      "step": 22221
    },
    {
      "epoch": 1.2100767120676863,
      "grad_norm": 0.6148614300199156,
      "learning_rate": 7.124959896810207e-05,
      "loss": 12.1397,
      "step": 22222
    },
    {
      "epoch": 1.2101311660642693,
      "grad_norm": 0.5157349583589322,
      "learning_rate": 7.124115313358428e-05,
      "loss": 12.0791,
      "step": 22223
    },
    {
      "epoch": 1.2101856200608523,
      "grad_norm": 0.5429489920842572,
      "learning_rate": 7.123270752270005e-05,
      "loss": 12.0118,
      "step": 22224
    },
    {
      "epoch": 1.2102400740574353,
      "grad_norm": 0.5254134460917859,
      "learning_rate": 7.122426213551513e-05,
      "loss": 11.9898,
      "step": 22225
    },
    {
      "epoch": 1.2102945280540183,
      "grad_norm": 0.5446806538940857,
      "learning_rate": 7.121581697209516e-05,
      "loss": 11.9785,
      "step": 22226
    },
    {
      "epoch": 1.2103489820506015,
      "grad_norm": 0.5197203705656492,
      "learning_rate": 7.120737203250582e-05,
      "loss": 11.9456,
      "step": 22227
    },
    {
      "epoch": 1.2104034360471845,
      "grad_norm": 0.5630397593922395,
      "learning_rate": 7.11989273168128e-05,
      "loss": 11.984,
      "step": 22228
    },
    {
      "epoch": 1.2104578900437675,
      "grad_norm": 0.6025670215102922,
      "learning_rate": 7.119048282508176e-05,
      "loss": 12.1266,
      "step": 22229
    },
    {
      "epoch": 1.2105123440403505,
      "grad_norm": 0.5154841225457294,
      "learning_rate": 7.118203855737833e-05,
      "loss": 11.9637,
      "step": 22230
    },
    {
      "epoch": 1.2105667980369335,
      "grad_norm": 0.5638156237527165,
      "learning_rate": 7.117359451376822e-05,
      "loss": 11.9354,
      "step": 22231
    },
    {
      "epoch": 1.2106212520335164,
      "grad_norm": 0.5660231263125333,
      "learning_rate": 7.116515069431704e-05,
      "loss": 11.9912,
      "step": 22232
    },
    {
      "epoch": 1.2106757060300994,
      "grad_norm": 0.6107715028659775,
      "learning_rate": 7.115670709909056e-05,
      "loss": 11.9319,
      "step": 22233
    },
    {
      "epoch": 1.2107301600266824,
      "grad_norm": 0.6017035343736808,
      "learning_rate": 7.114826372815432e-05,
      "loss": 11.9558,
      "step": 22234
    },
    {
      "epoch": 1.2107846140232654,
      "grad_norm": 0.5229652175836073,
      "learning_rate": 7.113982058157402e-05,
      "loss": 11.8972,
      "step": 22235
    },
    {
      "epoch": 1.2108390680198484,
      "grad_norm": 0.5701160898836272,
      "learning_rate": 7.113137765941528e-05,
      "loss": 11.9585,
      "step": 22236
    },
    {
      "epoch": 1.2108935220164314,
      "grad_norm": 0.553893458828715,
      "learning_rate": 7.112293496174381e-05,
      "loss": 11.9503,
      "step": 22237
    },
    {
      "epoch": 1.2109479760130144,
      "grad_norm": 0.4892356897325319,
      "learning_rate": 7.111449248862525e-05,
      "loss": 11.9463,
      "step": 22238
    },
    {
      "epoch": 1.2110024300095974,
      "grad_norm": 0.5822877339074625,
      "learning_rate": 7.110605024012524e-05,
      "loss": 11.9796,
      "step": 22239
    },
    {
      "epoch": 1.2110568840061806,
      "grad_norm": 0.5853273380034315,
      "learning_rate": 7.109760821630943e-05,
      "loss": 12.045,
      "step": 22240
    },
    {
      "epoch": 1.2111113380027636,
      "grad_norm": 0.5656663684305093,
      "learning_rate": 7.108916641724345e-05,
      "loss": 12.0361,
      "step": 22241
    },
    {
      "epoch": 1.2111657919993466,
      "grad_norm": 0.5540103396672027,
      "learning_rate": 7.108072484299299e-05,
      "loss": 11.9424,
      "step": 22242
    },
    {
      "epoch": 1.2112202459959296,
      "grad_norm": 0.5130192052115216,
      "learning_rate": 7.107228349362368e-05,
      "loss": 11.9045,
      "step": 22243
    },
    {
      "epoch": 1.2112746999925126,
      "grad_norm": 0.563121879580242,
      "learning_rate": 7.106384236920109e-05,
      "loss": 12.0127,
      "step": 22244
    },
    {
      "epoch": 1.2113291539890956,
      "grad_norm": 0.5140509319795811,
      "learning_rate": 7.105540146979095e-05,
      "loss": 11.9238,
      "step": 22245
    },
    {
      "epoch": 1.2113836079856786,
      "grad_norm": 0.5607506092057514,
      "learning_rate": 7.104696079545886e-05,
      "loss": 11.9958,
      "step": 22246
    },
    {
      "epoch": 1.2114380619822616,
      "grad_norm": 0.5316034142149118,
      "learning_rate": 7.10385203462704e-05,
      "loss": 11.6815,
      "step": 22247
    },
    {
      "epoch": 1.2114925159788446,
      "grad_norm": 0.5238273300317361,
      "learning_rate": 7.10300801222913e-05,
      "loss": 12.1356,
      "step": 22248
    },
    {
      "epoch": 1.2115469699754275,
      "grad_norm": 0.5737696562662198,
      "learning_rate": 7.102164012358719e-05,
      "loss": 11.9107,
      "step": 22249
    },
    {
      "epoch": 1.2116014239720108,
      "grad_norm": 0.4843855987877133,
      "learning_rate": 7.101320035022363e-05,
      "loss": 11.9216,
      "step": 22250
    },
    {
      "epoch": 1.2116558779685938,
      "grad_norm": 0.5321854172161506,
      "learning_rate": 7.10047608022663e-05,
      "loss": 12.0449,
      "step": 22251
    },
    {
      "epoch": 1.2117103319651767,
      "grad_norm": 0.542024674210292,
      "learning_rate": 7.099632147978081e-05,
      "loss": 11.9542,
      "step": 22252
    },
    {
      "epoch": 1.2117647859617597,
      "grad_norm": 0.5624187728852066,
      "learning_rate": 7.098788238283278e-05,
      "loss": 11.9009,
      "step": 22253
    },
    {
      "epoch": 1.2118192399583427,
      "grad_norm": 0.545018263851282,
      "learning_rate": 7.097944351148787e-05,
      "loss": 11.9881,
      "step": 22254
    },
    {
      "epoch": 1.2118736939549257,
      "grad_norm": 0.547322436937603,
      "learning_rate": 7.097100486581165e-05,
      "loss": 12.1616,
      "step": 22255
    },
    {
      "epoch": 1.2119281479515087,
      "grad_norm": 0.5854550211063693,
      "learning_rate": 7.096256644586976e-05,
      "loss": 12.0093,
      "step": 22256
    },
    {
      "epoch": 1.2119826019480917,
      "grad_norm": 0.5294357935560317,
      "learning_rate": 7.095412825172784e-05,
      "loss": 11.9794,
      "step": 22257
    },
    {
      "epoch": 1.2120370559446747,
      "grad_norm": 0.6303869108387259,
      "learning_rate": 7.094569028345146e-05,
      "loss": 12.0509,
      "step": 22258
    },
    {
      "epoch": 1.2120915099412577,
      "grad_norm": 0.5067586615831682,
      "learning_rate": 7.093725254110627e-05,
      "loss": 11.9189,
      "step": 22259
    },
    {
      "epoch": 1.2121459639378407,
      "grad_norm": 0.5578965016345673,
      "learning_rate": 7.09288150247579e-05,
      "loss": 11.9744,
      "step": 22260
    },
    {
      "epoch": 1.2122004179344237,
      "grad_norm": 0.7391896317516873,
      "learning_rate": 7.092037773447193e-05,
      "loss": 11.9811,
      "step": 22261
    },
    {
      "epoch": 1.2122548719310067,
      "grad_norm": 0.5437859337146264,
      "learning_rate": 7.091194067031398e-05,
      "loss": 11.9437,
      "step": 22262
    },
    {
      "epoch": 1.2123093259275899,
      "grad_norm": 0.5742713919066256,
      "learning_rate": 7.090350383234966e-05,
      "loss": 11.929,
      "step": 22263
    },
    {
      "epoch": 1.2123637799241729,
      "grad_norm": 0.5562476907318756,
      "learning_rate": 7.08950672206446e-05,
      "loss": 11.9208,
      "step": 22264
    },
    {
      "epoch": 1.2124182339207559,
      "grad_norm": 0.5611820438255171,
      "learning_rate": 7.088663083526434e-05,
      "loss": 11.9286,
      "step": 22265
    },
    {
      "epoch": 1.2124726879173389,
      "grad_norm": 0.5372397433175323,
      "learning_rate": 7.087819467627454e-05,
      "loss": 11.978,
      "step": 22266
    },
    {
      "epoch": 1.2125271419139219,
      "grad_norm": 0.5401658270660569,
      "learning_rate": 7.086975874374077e-05,
      "loss": 12.0468,
      "step": 22267
    },
    {
      "epoch": 1.2125815959105049,
      "grad_norm": 0.5468530284162061,
      "learning_rate": 7.086132303772864e-05,
      "loss": 11.9912,
      "step": 22268
    },
    {
      "epoch": 1.2126360499070878,
      "grad_norm": 0.5289231617253853,
      "learning_rate": 7.085288755830375e-05,
      "loss": 11.9488,
      "step": 22269
    },
    {
      "epoch": 1.2126905039036708,
      "grad_norm": 0.4974327342037205,
      "learning_rate": 7.084445230553167e-05,
      "loss": 12.022,
      "step": 22270
    },
    {
      "epoch": 1.2127449579002538,
      "grad_norm": 0.5586721468309652,
      "learning_rate": 7.083601727947806e-05,
      "loss": 12.0494,
      "step": 22271
    },
    {
      "epoch": 1.2127994118968368,
      "grad_norm": 0.5432637061339359,
      "learning_rate": 7.082758248020844e-05,
      "loss": 12.0238,
      "step": 22272
    },
    {
      "epoch": 1.21285386589342,
      "grad_norm": 0.5474105843631708,
      "learning_rate": 7.081914790778845e-05,
      "loss": 11.9483,
      "step": 22273
    },
    {
      "epoch": 1.212908319890003,
      "grad_norm": 0.5318146688778155,
      "learning_rate": 7.081071356228368e-05,
      "loss": 11.7935,
      "step": 22274
    },
    {
      "epoch": 1.212962773886586,
      "grad_norm": 0.5240733562038657,
      "learning_rate": 7.080227944375968e-05,
      "loss": 11.9969,
      "step": 22275
    },
    {
      "epoch": 1.213017227883169,
      "grad_norm": 0.5855271049242533,
      "learning_rate": 7.079384555228204e-05,
      "loss": 12.1057,
      "step": 22276
    },
    {
      "epoch": 1.213071681879752,
      "grad_norm": 0.5257915104221932,
      "learning_rate": 7.078541188791636e-05,
      "loss": 11.9402,
      "step": 22277
    },
    {
      "epoch": 1.213126135876335,
      "grad_norm": 0.5357458275395512,
      "learning_rate": 7.077697845072821e-05,
      "loss": 11.8799,
      "step": 22278
    },
    {
      "epoch": 1.213180589872918,
      "grad_norm": 0.5892448472689967,
      "learning_rate": 7.076854524078318e-05,
      "loss": 12.0626,
      "step": 22279
    },
    {
      "epoch": 1.213235043869501,
      "grad_norm": 0.5561518650997543,
      "learning_rate": 7.076011225814685e-05,
      "loss": 11.994,
      "step": 22280
    },
    {
      "epoch": 1.213289497866084,
      "grad_norm": 0.5671959921232491,
      "learning_rate": 7.075167950288477e-05,
      "loss": 11.9708,
      "step": 22281
    },
    {
      "epoch": 1.213343951862667,
      "grad_norm": 0.5241208392779223,
      "learning_rate": 7.074324697506255e-05,
      "loss": 11.8656,
      "step": 22282
    },
    {
      "epoch": 1.21339840585925,
      "grad_norm": 0.6521234757186432,
      "learning_rate": 7.073481467474575e-05,
      "loss": 12.013,
      "step": 22283
    },
    {
      "epoch": 1.213452859855833,
      "grad_norm": 0.48788085413657317,
      "learning_rate": 7.072638260199997e-05,
      "loss": 12.0519,
      "step": 22284
    },
    {
      "epoch": 1.213507313852416,
      "grad_norm": 0.563264647372284,
      "learning_rate": 7.07179507568907e-05,
      "loss": 11.8737,
      "step": 22285
    },
    {
      "epoch": 1.2135617678489992,
      "grad_norm": 0.594163666289845,
      "learning_rate": 7.070951913948359e-05,
      "loss": 12.0171,
      "step": 22286
    },
    {
      "epoch": 1.2136162218455822,
      "grad_norm": 0.5633618600797079,
      "learning_rate": 7.070108774984415e-05,
      "loss": 12.028,
      "step": 22287
    },
    {
      "epoch": 1.2136706758421651,
      "grad_norm": 0.5191886984552105,
      "learning_rate": 7.069265658803796e-05,
      "loss": 11.9681,
      "step": 22288
    },
    {
      "epoch": 1.2137251298387481,
      "grad_norm": 0.537810626953105,
      "learning_rate": 7.06842256541306e-05,
      "loss": 12.0174,
      "step": 22289
    },
    {
      "epoch": 1.2137795838353311,
      "grad_norm": 0.5692931708589785,
      "learning_rate": 7.067579494818761e-05,
      "loss": 11.9745,
      "step": 22290
    },
    {
      "epoch": 1.2138340378319141,
      "grad_norm": 0.6202625084016253,
      "learning_rate": 7.066736447027455e-05,
      "loss": 11.9476,
      "step": 22291
    },
    {
      "epoch": 1.2138884918284971,
      "grad_norm": 0.5324484456463837,
      "learning_rate": 7.065893422045698e-05,
      "loss": 12.1948,
      "step": 22292
    },
    {
      "epoch": 1.21394294582508,
      "grad_norm": 0.5021182461547399,
      "learning_rate": 7.065050419880046e-05,
      "loss": 11.9584,
      "step": 22293
    },
    {
      "epoch": 1.213997399821663,
      "grad_norm": 0.5939087172169644,
      "learning_rate": 7.064207440537061e-05,
      "loss": 12.0118,
      "step": 22294
    },
    {
      "epoch": 1.214051853818246,
      "grad_norm": 0.5634578786794503,
      "learning_rate": 7.063364484023285e-05,
      "loss": 11.9904,
      "step": 22295
    },
    {
      "epoch": 1.214106307814829,
      "grad_norm": 0.5484709194305013,
      "learning_rate": 7.062521550345277e-05,
      "loss": 11.8776,
      "step": 22296
    },
    {
      "epoch": 1.2141607618114123,
      "grad_norm": 0.5421099908982828,
      "learning_rate": 7.061678639509596e-05,
      "loss": 12.0019,
      "step": 22297
    },
    {
      "epoch": 1.2142152158079953,
      "grad_norm": 0.4991751723091259,
      "learning_rate": 7.060835751522797e-05,
      "loss": 11.9418,
      "step": 22298
    },
    {
      "epoch": 1.2142696698045783,
      "grad_norm": 0.577553957827333,
      "learning_rate": 7.059992886391428e-05,
      "loss": 12.0586,
      "step": 22299
    },
    {
      "epoch": 1.2143241238011613,
      "grad_norm": 0.5255331978796294,
      "learning_rate": 7.059150044122052e-05,
      "loss": 11.9671,
      "step": 22300
    },
    {
      "epoch": 1.2143785777977443,
      "grad_norm": 0.5196347588231244,
      "learning_rate": 7.058307224721216e-05,
      "loss": 11.8502,
      "step": 22301
    },
    {
      "epoch": 1.2144330317943273,
      "grad_norm": 0.5355706344964161,
      "learning_rate": 7.057464428195476e-05,
      "loss": 11.8549,
      "step": 22302
    },
    {
      "epoch": 1.2144874857909103,
      "grad_norm": 0.4955177547160181,
      "learning_rate": 7.056621654551385e-05,
      "loss": 11.8426,
      "step": 22303
    },
    {
      "epoch": 1.2145419397874933,
      "grad_norm": 0.523880147256863,
      "learning_rate": 7.055778903795502e-05,
      "loss": 11.9664,
      "step": 22304
    },
    {
      "epoch": 1.2145963937840762,
      "grad_norm": 0.5350504152097167,
      "learning_rate": 7.054936175934375e-05,
      "loss": 11.8836,
      "step": 22305
    },
    {
      "epoch": 1.2146508477806592,
      "grad_norm": 0.5548101050832339,
      "learning_rate": 7.054093470974557e-05,
      "loss": 11.903,
      "step": 22306
    },
    {
      "epoch": 1.2147053017772422,
      "grad_norm": 0.6393004098557902,
      "learning_rate": 7.053250788922599e-05,
      "loss": 12.1723,
      "step": 22307
    },
    {
      "epoch": 1.2147597557738252,
      "grad_norm": 0.5315787203549713,
      "learning_rate": 7.05240812978506e-05,
      "loss": 12.0701,
      "step": 22308
    },
    {
      "epoch": 1.2148142097704082,
      "grad_norm": 0.5644005983604934,
      "learning_rate": 7.051565493568488e-05,
      "loss": 11.8484,
      "step": 22309
    },
    {
      "epoch": 1.2148686637669914,
      "grad_norm": 0.5338038664794075,
      "learning_rate": 7.050722880279439e-05,
      "loss": 11.9557,
      "step": 22310
    },
    {
      "epoch": 1.2149231177635744,
      "grad_norm": 0.5581254134054306,
      "learning_rate": 7.049880289924464e-05,
      "loss": 11.9729,
      "step": 22311
    },
    {
      "epoch": 1.2149775717601574,
      "grad_norm": 0.5450984713680336,
      "learning_rate": 7.049037722510113e-05,
      "loss": 11.8956,
      "step": 22312
    },
    {
      "epoch": 1.2150320257567404,
      "grad_norm": 0.6100547288983047,
      "learning_rate": 7.04819517804294e-05,
      "loss": 12.0668,
      "step": 22313
    },
    {
      "epoch": 1.2150864797533234,
      "grad_norm": 0.5259666863737034,
      "learning_rate": 7.047352656529498e-05,
      "loss": 12.1377,
      "step": 22314
    },
    {
      "epoch": 1.2151409337499064,
      "grad_norm": 0.5224156702911438,
      "learning_rate": 7.046510157976336e-05,
      "loss": 11.9721,
      "step": 22315
    },
    {
      "epoch": 1.2151953877464894,
      "grad_norm": 0.5203314355568164,
      "learning_rate": 7.045667682390004e-05,
      "loss": 11.9608,
      "step": 22316
    },
    {
      "epoch": 1.2152498417430724,
      "grad_norm": 0.5803029169990991,
      "learning_rate": 7.04482522977706e-05,
      "loss": 11.9736,
      "step": 22317
    },
    {
      "epoch": 1.2153042957396554,
      "grad_norm": 0.55320366865983,
      "learning_rate": 7.043982800144046e-05,
      "loss": 11.9466,
      "step": 22318
    },
    {
      "epoch": 1.2153587497362384,
      "grad_norm": 0.5532418365233676,
      "learning_rate": 7.043140393497518e-05,
      "loss": 12.005,
      "step": 22319
    },
    {
      "epoch": 1.2154132037328216,
      "grad_norm": 0.5154581249122987,
      "learning_rate": 7.042298009844027e-05,
      "loss": 11.7698,
      "step": 22320
    },
    {
      "epoch": 1.2154676577294046,
      "grad_norm": 0.5570806186394629,
      "learning_rate": 7.041455649190123e-05,
      "loss": 12.0179,
      "step": 22321
    },
    {
      "epoch": 1.2155221117259876,
      "grad_norm": 0.6075919078474695,
      "learning_rate": 7.040613311542357e-05,
      "loss": 12.0396,
      "step": 22322
    },
    {
      "epoch": 1.2155765657225706,
      "grad_norm": 0.5666390942113456,
      "learning_rate": 7.039770996907277e-05,
      "loss": 11.9011,
      "step": 22323
    },
    {
      "epoch": 1.2156310197191535,
      "grad_norm": 0.5092742503468421,
      "learning_rate": 7.038928705291436e-05,
      "loss": 11.926,
      "step": 22324
    },
    {
      "epoch": 1.2156854737157365,
      "grad_norm": 0.5501225754048922,
      "learning_rate": 7.038086436701381e-05,
      "loss": 11.8644,
      "step": 22325
    },
    {
      "epoch": 1.2157399277123195,
      "grad_norm": 0.5719510745672549,
      "learning_rate": 7.037244191143661e-05,
      "loss": 12.0941,
      "step": 22326
    },
    {
      "epoch": 1.2157943817089025,
      "grad_norm": 0.5577233311053958,
      "learning_rate": 7.03640196862483e-05,
      "loss": 11.9658,
      "step": 22327
    },
    {
      "epoch": 1.2158488357054855,
      "grad_norm": 0.554705498761114,
      "learning_rate": 7.035559769151432e-05,
      "loss": 11.8951,
      "step": 22328
    },
    {
      "epoch": 1.2159032897020685,
      "grad_norm": 0.6049096053195613,
      "learning_rate": 7.034717592730018e-05,
      "loss": 12.1224,
      "step": 22329
    },
    {
      "epoch": 1.2159577436986515,
      "grad_norm": 0.5216970000678234,
      "learning_rate": 7.033875439367137e-05,
      "loss": 11.9001,
      "step": 22330
    },
    {
      "epoch": 1.2160121976952345,
      "grad_norm": 0.5569509677271596,
      "learning_rate": 7.033033309069339e-05,
      "loss": 12.0039,
      "step": 22331
    },
    {
      "epoch": 1.2160666516918175,
      "grad_norm": 0.5508933269962223,
      "learning_rate": 7.03219120184317e-05,
      "loss": 11.99,
      "step": 22332
    },
    {
      "epoch": 1.2161211056884007,
      "grad_norm": 0.5963243297822097,
      "learning_rate": 7.031349117695183e-05,
      "loss": 11.9617,
      "step": 22333
    },
    {
      "epoch": 1.2161755596849837,
      "grad_norm": 0.5379930064043972,
      "learning_rate": 7.030507056631923e-05,
      "loss": 11.9727,
      "step": 22334
    },
    {
      "epoch": 1.2162300136815667,
      "grad_norm": 0.5386747676124668,
      "learning_rate": 7.029665018659936e-05,
      "loss": 11.9948,
      "step": 22335
    },
    {
      "epoch": 1.2162844676781497,
      "grad_norm": 0.5723447515807047,
      "learning_rate": 7.028823003785774e-05,
      "loss": 11.9454,
      "step": 22336
    },
    {
      "epoch": 1.2163389216747327,
      "grad_norm": 0.583893023561163,
      "learning_rate": 7.027981012015981e-05,
      "loss": 11.9514,
      "step": 22337
    },
    {
      "epoch": 1.2163933756713157,
      "grad_norm": 0.5887967027117306,
      "learning_rate": 7.027139043357106e-05,
      "loss": 12.0417,
      "step": 22338
    },
    {
      "epoch": 1.2164478296678987,
      "grad_norm": 0.5689338247182932,
      "learning_rate": 7.026297097815697e-05,
      "loss": 12.0591,
      "step": 22339
    },
    {
      "epoch": 1.2165022836644817,
      "grad_norm": 0.5141804373708683,
      "learning_rate": 7.025455175398299e-05,
      "loss": 12.0649,
      "step": 22340
    },
    {
      "epoch": 1.2165567376610646,
      "grad_norm": 0.7292763687760996,
      "learning_rate": 7.02461327611146e-05,
      "loss": 12.0871,
      "step": 22341
    },
    {
      "epoch": 1.2166111916576476,
      "grad_norm": 0.5518115603447546,
      "learning_rate": 7.02377139996173e-05,
      "loss": 12.036,
      "step": 22342
    },
    {
      "epoch": 1.2166656456542309,
      "grad_norm": 0.5506875059426861,
      "learning_rate": 7.02292954695565e-05,
      "loss": 11.9148,
      "step": 22343
    },
    {
      "epoch": 1.2167200996508138,
      "grad_norm": 0.7162991465665424,
      "learning_rate": 7.022087717099775e-05,
      "loss": 11.9591,
      "step": 22344
    },
    {
      "epoch": 1.2167745536473968,
      "grad_norm": 0.5522233152721466,
      "learning_rate": 7.021245910400638e-05,
      "loss": 11.9098,
      "step": 22345
    },
    {
      "epoch": 1.2168290076439798,
      "grad_norm": 0.5137823090818378,
      "learning_rate": 7.020404126864794e-05,
      "loss": 11.9703,
      "step": 22346
    },
    {
      "epoch": 1.2168834616405628,
      "grad_norm": 0.5967318474798641,
      "learning_rate": 7.01956236649879e-05,
      "loss": 12.0042,
      "step": 22347
    },
    {
      "epoch": 1.2169379156371458,
      "grad_norm": 0.510429647743093,
      "learning_rate": 7.018720629309167e-05,
      "loss": 12.0425,
      "step": 22348
    },
    {
      "epoch": 1.2169923696337288,
      "grad_norm": 0.6660538355641454,
      "learning_rate": 7.017878915302471e-05,
      "loss": 11.9772,
      "step": 22349
    },
    {
      "epoch": 1.2170468236303118,
      "grad_norm": 0.5156984135756433,
      "learning_rate": 7.017037224485251e-05,
      "loss": 11.9955,
      "step": 22350
    },
    {
      "epoch": 1.2171012776268948,
      "grad_norm": 0.5515831340977696,
      "learning_rate": 7.016195556864049e-05,
      "loss": 12.0198,
      "step": 22351
    },
    {
      "epoch": 1.2171557316234778,
      "grad_norm": 0.5009435038304422,
      "learning_rate": 7.015353912445408e-05,
      "loss": 11.9356,
      "step": 22352
    },
    {
      "epoch": 1.2172101856200608,
      "grad_norm": 0.5027011911111109,
      "learning_rate": 7.01451229123588e-05,
      "loss": 11.8627,
      "step": 22353
    },
    {
      "epoch": 1.2172646396166438,
      "grad_norm": 0.5668826387820898,
      "learning_rate": 7.013670693242005e-05,
      "loss": 12.0749,
      "step": 22354
    },
    {
      "epoch": 1.2173190936132268,
      "grad_norm": 0.5515281227037341,
      "learning_rate": 7.012829118470328e-05,
      "loss": 12.139,
      "step": 22355
    },
    {
      "epoch": 1.21737354760981,
      "grad_norm": 0.4810495926302193,
      "learning_rate": 7.011987566927387e-05,
      "loss": 11.9472,
      "step": 22356
    },
    {
      "epoch": 1.217428001606393,
      "grad_norm": 0.5723272624958875,
      "learning_rate": 7.011146038619735e-05,
      "loss": 11.8951,
      "step": 22357
    },
    {
      "epoch": 1.217482455602976,
      "grad_norm": 0.5549262552776699,
      "learning_rate": 7.010304533553913e-05,
      "loss": 12.0505,
      "step": 22358
    },
    {
      "epoch": 1.217536909599559,
      "grad_norm": 0.5405482614457372,
      "learning_rate": 7.009463051736465e-05,
      "loss": 11.9682,
      "step": 22359
    },
    {
      "epoch": 1.217591363596142,
      "grad_norm": 0.5412814365790568,
      "learning_rate": 7.008621593173932e-05,
      "loss": 11.9465,
      "step": 22360
    },
    {
      "epoch": 1.217645817592725,
      "grad_norm": 0.5467583588804531,
      "learning_rate": 7.00778015787286e-05,
      "loss": 12.0006,
      "step": 22361
    },
    {
      "epoch": 1.217700271589308,
      "grad_norm": 0.4916924452821014,
      "learning_rate": 7.006938745839792e-05,
      "loss": 12.0715,
      "step": 22362
    },
    {
      "epoch": 1.217754725585891,
      "grad_norm": 0.5508482146314773,
      "learning_rate": 7.006097357081269e-05,
      "loss": 12.0779,
      "step": 22363
    },
    {
      "epoch": 1.217809179582474,
      "grad_norm": 0.5041038578527925,
      "learning_rate": 7.005255991603838e-05,
      "loss": 11.9602,
      "step": 22364
    },
    {
      "epoch": 1.217863633579057,
      "grad_norm": 0.5695379925221715,
      "learning_rate": 7.004414649414037e-05,
      "loss": 12.0427,
      "step": 22365
    },
    {
      "epoch": 1.2179180875756401,
      "grad_norm": 0.525114739195111,
      "learning_rate": 7.003573330518409e-05,
      "loss": 11.8928,
      "step": 22366
    },
    {
      "epoch": 1.2179725415722231,
      "grad_norm": 0.5566647811168054,
      "learning_rate": 7.002732034923499e-05,
      "loss": 12.0557,
      "step": 22367
    },
    {
      "epoch": 1.2180269955688061,
      "grad_norm": 0.49262671405552383,
      "learning_rate": 7.001890762635844e-05,
      "loss": 11.9055,
      "step": 22368
    },
    {
      "epoch": 1.218081449565389,
      "grad_norm": 0.5578951026885405,
      "learning_rate": 7.00104951366199e-05,
      "loss": 11.9872,
      "step": 22369
    },
    {
      "epoch": 1.218135903561972,
      "grad_norm": 0.5683924026785109,
      "learning_rate": 7.000208288008478e-05,
      "loss": 11.9625,
      "step": 22370
    },
    {
      "epoch": 1.218190357558555,
      "grad_norm": 0.5086219925298866,
      "learning_rate": 6.999367085681852e-05,
      "loss": 11.9946,
      "step": 22371
    },
    {
      "epoch": 1.218244811555138,
      "grad_norm": 0.515937880076869,
      "learning_rate": 6.998525906688649e-05,
      "loss": 12.0035,
      "step": 22372
    },
    {
      "epoch": 1.218299265551721,
      "grad_norm": 0.5588427097572062,
      "learning_rate": 6.997684751035411e-05,
      "loss": 11.9378,
      "step": 22373
    },
    {
      "epoch": 1.218353719548304,
      "grad_norm": 0.5279204411786923,
      "learning_rate": 6.996843618728684e-05,
      "loss": 11.9883,
      "step": 22374
    },
    {
      "epoch": 1.218408173544887,
      "grad_norm": 0.5212889295302302,
      "learning_rate": 6.996002509775001e-05,
      "loss": 11.9559,
      "step": 22375
    },
    {
      "epoch": 1.21846262754147,
      "grad_norm": 0.5420376501032255,
      "learning_rate": 6.995161424180908e-05,
      "loss": 11.8928,
      "step": 22376
    },
    {
      "epoch": 1.218517081538053,
      "grad_norm": 0.5166400531308509,
      "learning_rate": 6.994320361952943e-05,
      "loss": 11.8835,
      "step": 22377
    },
    {
      "epoch": 1.218571535534636,
      "grad_norm": 0.5292885786067021,
      "learning_rate": 6.993479323097647e-05,
      "loss": 11.993,
      "step": 22378
    },
    {
      "epoch": 1.218625989531219,
      "grad_norm": 0.583737031027247,
      "learning_rate": 6.992638307621557e-05,
      "loss": 11.8614,
      "step": 22379
    },
    {
      "epoch": 1.2186804435278022,
      "grad_norm": 0.561478994551792,
      "learning_rate": 6.99179731553122e-05,
      "loss": 12.0964,
      "step": 22380
    },
    {
      "epoch": 1.2187348975243852,
      "grad_norm": 0.5640655878626899,
      "learning_rate": 6.990956346833168e-05,
      "loss": 11.9127,
      "step": 22381
    },
    {
      "epoch": 1.2187893515209682,
      "grad_norm": 0.5762987678441256,
      "learning_rate": 6.990115401533946e-05,
      "loss": 12.0166,
      "step": 22382
    },
    {
      "epoch": 1.2188438055175512,
      "grad_norm": 0.6597124157320646,
      "learning_rate": 6.98927447964009e-05,
      "loss": 12.0442,
      "step": 22383
    },
    {
      "epoch": 1.2188982595141342,
      "grad_norm": 0.5609973227230155,
      "learning_rate": 6.988433581158145e-05,
      "loss": 11.9126,
      "step": 22384
    },
    {
      "epoch": 1.2189527135107172,
      "grad_norm": 0.5440578199406028,
      "learning_rate": 6.987592706094643e-05,
      "loss": 11.8648,
      "step": 22385
    },
    {
      "epoch": 1.2190071675073002,
      "grad_norm": 0.5859657659305496,
      "learning_rate": 6.986751854456124e-05,
      "loss": 12.0417,
      "step": 22386
    },
    {
      "epoch": 1.2190616215038832,
      "grad_norm": 0.5791757679621246,
      "learning_rate": 6.98591102624913e-05,
      "loss": 11.8141,
      "step": 22387
    },
    {
      "epoch": 1.2191160755004662,
      "grad_norm": 0.580928826714245,
      "learning_rate": 6.985070221480195e-05,
      "loss": 11.9734,
      "step": 22388
    },
    {
      "epoch": 1.2191705294970492,
      "grad_norm": 0.523511068552816,
      "learning_rate": 6.98422944015586e-05,
      "loss": 11.9273,
      "step": 22389
    },
    {
      "epoch": 1.2192249834936324,
      "grad_norm": 0.5280995417782323,
      "learning_rate": 6.98338868228266e-05,
      "loss": 11.9703,
      "step": 22390
    },
    {
      "epoch": 1.2192794374902154,
      "grad_norm": 0.554149458968344,
      "learning_rate": 6.982547947867141e-05,
      "loss": 11.9068,
      "step": 22391
    },
    {
      "epoch": 1.2193338914867984,
      "grad_norm": 0.5710116531873979,
      "learning_rate": 6.981707236915833e-05,
      "loss": 11.856,
      "step": 22392
    },
    {
      "epoch": 1.2193883454833814,
      "grad_norm": 0.5646006857839734,
      "learning_rate": 6.980866549435275e-05,
      "loss": 11.9466,
      "step": 22393
    },
    {
      "epoch": 1.2194427994799644,
      "grad_norm": 0.5457992022543512,
      "learning_rate": 6.980025885432006e-05,
      "loss": 12.0449,
      "step": 22394
    },
    {
      "epoch": 1.2194972534765474,
      "grad_norm": 0.6146281492599752,
      "learning_rate": 6.979185244912563e-05,
      "loss": 11.9747,
      "step": 22395
    },
    {
      "epoch": 1.2195517074731304,
      "grad_norm": 0.589197910788417,
      "learning_rate": 6.978344627883482e-05,
      "loss": 12.0342,
      "step": 22396
    },
    {
      "epoch": 1.2196061614697133,
      "grad_norm": 0.5438373320570249,
      "learning_rate": 6.977504034351297e-05,
      "loss": 12.0001,
      "step": 22397
    },
    {
      "epoch": 1.2196606154662963,
      "grad_norm": 0.5923276252131006,
      "learning_rate": 6.97666346432255e-05,
      "loss": 12.0653,
      "step": 22398
    },
    {
      "epoch": 1.2197150694628793,
      "grad_norm": 0.6069501760450577,
      "learning_rate": 6.975822917803773e-05,
      "loss": 12.0088,
      "step": 22399
    },
    {
      "epoch": 1.2197695234594623,
      "grad_norm": 0.5136480609945389,
      "learning_rate": 6.974982394801505e-05,
      "loss": 12.0383,
      "step": 22400
    },
    {
      "epoch": 1.2198239774560453,
      "grad_norm": 0.5076437212057705,
      "learning_rate": 6.974141895322282e-05,
      "loss": 11.9397,
      "step": 22401
    },
    {
      "epoch": 1.2198784314526283,
      "grad_norm": 0.5724996968219022,
      "learning_rate": 6.973301419372637e-05,
      "loss": 11.965,
      "step": 22402
    },
    {
      "epoch": 1.2199328854492115,
      "grad_norm": 0.4995089364803861,
      "learning_rate": 6.97246096695911e-05,
      "loss": 11.975,
      "step": 22403
    },
    {
      "epoch": 1.2199873394457945,
      "grad_norm": 0.580369141175968,
      "learning_rate": 6.971620538088233e-05,
      "loss": 11.9444,
      "step": 22404
    },
    {
      "epoch": 1.2200417934423775,
      "grad_norm": 0.49372070185979805,
      "learning_rate": 6.970780132766545e-05,
      "loss": 11.7919,
      "step": 22405
    },
    {
      "epoch": 1.2200962474389605,
      "grad_norm": 0.6917361868930043,
      "learning_rate": 6.969939751000577e-05,
      "loss": 12.0882,
      "step": 22406
    },
    {
      "epoch": 1.2201507014355435,
      "grad_norm": 0.6322136547605418,
      "learning_rate": 6.969099392796865e-05,
      "loss": 12.0858,
      "step": 22407
    },
    {
      "epoch": 1.2202051554321265,
      "grad_norm": 0.5223013259575535,
      "learning_rate": 6.968259058161946e-05,
      "loss": 11.8596,
      "step": 22408
    },
    {
      "epoch": 1.2202596094287095,
      "grad_norm": 0.5512881492846489,
      "learning_rate": 6.967418747102351e-05,
      "loss": 12.0115,
      "step": 22409
    },
    {
      "epoch": 1.2203140634252925,
      "grad_norm": 0.5173212222099689,
      "learning_rate": 6.966578459624617e-05,
      "loss": 11.8679,
      "step": 22410
    },
    {
      "epoch": 1.2203685174218755,
      "grad_norm": 0.5708016938914064,
      "learning_rate": 6.965738195735279e-05,
      "loss": 12.0618,
      "step": 22411
    },
    {
      "epoch": 1.2204229714184585,
      "grad_norm": 0.5448068878427165,
      "learning_rate": 6.964897955440868e-05,
      "loss": 11.8162,
      "step": 22412
    },
    {
      "epoch": 1.2204774254150417,
      "grad_norm": 0.5529709578796422,
      "learning_rate": 6.964057738747918e-05,
      "loss": 12.0061,
      "step": 22413
    },
    {
      "epoch": 1.2205318794116247,
      "grad_norm": 0.5358876828902283,
      "learning_rate": 6.963217545662965e-05,
      "loss": 11.9137,
      "step": 22414
    },
    {
      "epoch": 1.2205863334082077,
      "grad_norm": 0.5647036952014872,
      "learning_rate": 6.962377376192548e-05,
      "loss": 12.0191,
      "step": 22415
    },
    {
      "epoch": 1.2206407874047907,
      "grad_norm": 0.5940657194824316,
      "learning_rate": 6.961537230343188e-05,
      "loss": 12.184,
      "step": 22416
    },
    {
      "epoch": 1.2206952414013736,
      "grad_norm": 0.5790464934404309,
      "learning_rate": 6.960697108121423e-05,
      "loss": 12.0073,
      "step": 22417
    },
    {
      "epoch": 1.2207496953979566,
      "grad_norm": 0.5045579535013146,
      "learning_rate": 6.959857009533787e-05,
      "loss": 11.9809,
      "step": 22418
    },
    {
      "epoch": 1.2208041493945396,
      "grad_norm": 0.5530522189101298,
      "learning_rate": 6.959016934586816e-05,
      "loss": 12.039,
      "step": 22419
    },
    {
      "epoch": 1.2208586033911226,
      "grad_norm": 0.5335794066957925,
      "learning_rate": 6.958176883287037e-05,
      "loss": 11.9908,
      "step": 22420
    },
    {
      "epoch": 1.2209130573877056,
      "grad_norm": 0.5344672194234419,
      "learning_rate": 6.957336855640986e-05,
      "loss": 11.9482,
      "step": 22421
    },
    {
      "epoch": 1.2209675113842886,
      "grad_norm": 0.5409941895580455,
      "learning_rate": 6.956496851655194e-05,
      "loss": 12.055,
      "step": 22422
    },
    {
      "epoch": 1.2210219653808716,
      "grad_norm": 0.5162961278211933,
      "learning_rate": 6.955656871336191e-05,
      "loss": 11.9066,
      "step": 22423
    },
    {
      "epoch": 1.2210764193774546,
      "grad_norm": 0.5731886725647957,
      "learning_rate": 6.954816914690509e-05,
      "loss": 11.9437,
      "step": 22424
    },
    {
      "epoch": 1.2211308733740376,
      "grad_norm": 0.5375378154163946,
      "learning_rate": 6.95397698172469e-05,
      "loss": 11.897,
      "step": 22425
    },
    {
      "epoch": 1.2211853273706208,
      "grad_norm": 0.5308922702002112,
      "learning_rate": 6.95313707244525e-05,
      "loss": 11.8796,
      "step": 22426
    },
    {
      "epoch": 1.2212397813672038,
      "grad_norm": 0.5496555211264113,
      "learning_rate": 6.952297186858728e-05,
      "loss": 11.9544,
      "step": 22427
    },
    {
      "epoch": 1.2212942353637868,
      "grad_norm": 0.5110784431805737,
      "learning_rate": 6.951457324971653e-05,
      "loss": 11.971,
      "step": 22428
    },
    {
      "epoch": 1.2213486893603698,
      "grad_norm": 0.549558745358558,
      "learning_rate": 6.950617486790558e-05,
      "loss": 11.9023,
      "step": 22429
    },
    {
      "epoch": 1.2214031433569528,
      "grad_norm": 0.554236279059842,
      "learning_rate": 6.949777672321973e-05,
      "loss": 12.076,
      "step": 22430
    },
    {
      "epoch": 1.2214575973535358,
      "grad_norm": 0.5544054192583099,
      "learning_rate": 6.948937881572428e-05,
      "loss": 11.9806,
      "step": 22431
    },
    {
      "epoch": 1.2215120513501188,
      "grad_norm": 0.5659384279633011,
      "learning_rate": 6.948098114548454e-05,
      "loss": 11.8784,
      "step": 22432
    },
    {
      "epoch": 1.2215665053467017,
      "grad_norm": 0.5623758176488516,
      "learning_rate": 6.947258371256582e-05,
      "loss": 11.9009,
      "step": 22433
    },
    {
      "epoch": 1.2216209593432847,
      "grad_norm": 0.517444801061649,
      "learning_rate": 6.94641865170334e-05,
      "loss": 12.0179,
      "step": 22434
    },
    {
      "epoch": 1.2216754133398677,
      "grad_norm": 0.5516104639008468,
      "learning_rate": 6.945578955895259e-05,
      "loss": 11.8954,
      "step": 22435
    },
    {
      "epoch": 1.221729867336451,
      "grad_norm": 0.5662922139479862,
      "learning_rate": 6.944739283838868e-05,
      "loss": 11.9934,
      "step": 22436
    },
    {
      "epoch": 1.221784321333034,
      "grad_norm": 0.5264966002719521,
      "learning_rate": 6.943899635540697e-05,
      "loss": 11.8817,
      "step": 22437
    },
    {
      "epoch": 1.221838775329617,
      "grad_norm": 0.5127025445740278,
      "learning_rate": 6.943060011007274e-05,
      "loss": 11.8937,
      "step": 22438
    },
    {
      "epoch": 1.2218932293262,
      "grad_norm": 0.5271368637409032,
      "learning_rate": 6.942220410245128e-05,
      "loss": 11.9038,
      "step": 22439
    },
    {
      "epoch": 1.221947683322783,
      "grad_norm": 0.5469080513964029,
      "learning_rate": 6.94138083326079e-05,
      "loss": 12.1692,
      "step": 22440
    },
    {
      "epoch": 1.222002137319366,
      "grad_norm": 0.4922272766274975,
      "learning_rate": 6.940541280060788e-05,
      "loss": 11.9012,
      "step": 22441
    },
    {
      "epoch": 1.222056591315949,
      "grad_norm": 0.5071699516397944,
      "learning_rate": 6.93970175065165e-05,
      "loss": 11.9217,
      "step": 22442
    },
    {
      "epoch": 1.222111045312532,
      "grad_norm": 0.5443291129826455,
      "learning_rate": 6.938862245039904e-05,
      "loss": 12.0067,
      "step": 22443
    },
    {
      "epoch": 1.2221654993091149,
      "grad_norm": 0.5463068505259102,
      "learning_rate": 6.938022763232079e-05,
      "loss": 11.8455,
      "step": 22444
    },
    {
      "epoch": 1.2222199533056979,
      "grad_norm": 0.5480977159865835,
      "learning_rate": 6.937183305234705e-05,
      "loss": 11.9707,
      "step": 22445
    },
    {
      "epoch": 1.2222744073022809,
      "grad_norm": 0.5634515913441669,
      "learning_rate": 6.936343871054304e-05,
      "loss": 11.9748,
      "step": 22446
    },
    {
      "epoch": 1.2223288612988639,
      "grad_norm": 0.571882199790371,
      "learning_rate": 6.935504460697407e-05,
      "loss": 12.0599,
      "step": 22447
    },
    {
      "epoch": 1.2223833152954469,
      "grad_norm": 0.528124344999855,
      "learning_rate": 6.934665074170542e-05,
      "loss": 11.8866,
      "step": 22448
    },
    {
      "epoch": 1.2224377692920299,
      "grad_norm": 0.5836550455259616,
      "learning_rate": 6.933825711480236e-05,
      "loss": 11.9151,
      "step": 22449
    },
    {
      "epoch": 1.222492223288613,
      "grad_norm": 0.5056020976244936,
      "learning_rate": 6.932986372633012e-05,
      "loss": 11.9975,
      "step": 22450
    },
    {
      "epoch": 1.222546677285196,
      "grad_norm": 0.48986824900253506,
      "learning_rate": 6.932147057635403e-05,
      "loss": 12.004,
      "step": 22451
    },
    {
      "epoch": 1.222601131281779,
      "grad_norm": 0.5660171207746736,
      "learning_rate": 6.931307766493933e-05,
      "loss": 12.0485,
      "step": 22452
    },
    {
      "epoch": 1.222655585278362,
      "grad_norm": 0.6416172517087058,
      "learning_rate": 6.930468499215128e-05,
      "loss": 11.9987,
      "step": 22453
    },
    {
      "epoch": 1.222710039274945,
      "grad_norm": 0.5803503295819011,
      "learning_rate": 6.929629255805514e-05,
      "loss": 11.8998,
      "step": 22454
    },
    {
      "epoch": 1.222764493271528,
      "grad_norm": 0.524470433421093,
      "learning_rate": 6.928790036271622e-05,
      "loss": 12.0086,
      "step": 22455
    },
    {
      "epoch": 1.222818947268111,
      "grad_norm": 0.5916510426763025,
      "learning_rate": 6.92795084061997e-05,
      "loss": 11.8913,
      "step": 22456
    },
    {
      "epoch": 1.222873401264694,
      "grad_norm": 0.6499721457360714,
      "learning_rate": 6.927111668857088e-05,
      "loss": 12.0046,
      "step": 22457
    },
    {
      "epoch": 1.222927855261277,
      "grad_norm": 0.5795488762817579,
      "learning_rate": 6.926272520989501e-05,
      "loss": 12.0773,
      "step": 22458
    },
    {
      "epoch": 1.22298230925786,
      "grad_norm": 0.4982720873340564,
      "learning_rate": 6.925433397023734e-05,
      "loss": 11.9238,
      "step": 22459
    },
    {
      "epoch": 1.2230367632544432,
      "grad_norm": 0.7991892105498443,
      "learning_rate": 6.924594296966313e-05,
      "loss": 11.9147,
      "step": 22460
    },
    {
      "epoch": 1.2230912172510262,
      "grad_norm": 0.5443048221822511,
      "learning_rate": 6.923755220823759e-05,
      "loss": 11.9642,
      "step": 22461
    },
    {
      "epoch": 1.2231456712476092,
      "grad_norm": 0.5778090281931294,
      "learning_rate": 6.922916168602604e-05,
      "loss": 12.0044,
      "step": 22462
    },
    {
      "epoch": 1.2232001252441922,
      "grad_norm": 0.644367087982054,
      "learning_rate": 6.922077140309368e-05,
      "loss": 11.9334,
      "step": 22463
    },
    {
      "epoch": 1.2232545792407752,
      "grad_norm": 0.529193243856524,
      "learning_rate": 6.921238135950578e-05,
      "loss": 11.8468,
      "step": 22464
    },
    {
      "epoch": 1.2233090332373582,
      "grad_norm": 0.5449996993842378,
      "learning_rate": 6.92039915553276e-05,
      "loss": 11.8906,
      "step": 22465
    },
    {
      "epoch": 1.2233634872339412,
      "grad_norm": 0.5335279736405514,
      "learning_rate": 6.919560199062426e-05,
      "loss": 12.024,
      "step": 22466
    },
    {
      "epoch": 1.2234179412305242,
      "grad_norm": 0.7067743752537503,
      "learning_rate": 6.918721266546113e-05,
      "loss": 12.007,
      "step": 22467
    },
    {
      "epoch": 1.2234723952271072,
      "grad_norm": 0.6387622983406572,
      "learning_rate": 6.91788235799034e-05,
      "loss": 12.0687,
      "step": 22468
    },
    {
      "epoch": 1.2235268492236901,
      "grad_norm": 0.6232709815148468,
      "learning_rate": 6.917043473401631e-05,
      "loss": 11.9692,
      "step": 22469
    },
    {
      "epoch": 1.2235813032202731,
      "grad_norm": 0.5561707891930123,
      "learning_rate": 6.916204612786508e-05,
      "loss": 11.9515,
      "step": 22470
    },
    {
      "epoch": 1.2236357572168561,
      "grad_norm": 0.4914024925290726,
      "learning_rate": 6.915365776151495e-05,
      "loss": 11.924,
      "step": 22471
    },
    {
      "epoch": 1.2236902112134391,
      "grad_norm": 0.6152761621250927,
      "learning_rate": 6.914526963503116e-05,
      "loss": 12.0788,
      "step": 22472
    },
    {
      "epoch": 1.2237446652100223,
      "grad_norm": 0.5532885451751817,
      "learning_rate": 6.91368817484789e-05,
      "loss": 11.9162,
      "step": 22473
    },
    {
      "epoch": 1.2237991192066053,
      "grad_norm": 0.5519537101500993,
      "learning_rate": 6.912849410192343e-05,
      "loss": 11.8547,
      "step": 22474
    },
    {
      "epoch": 1.2238535732031883,
      "grad_norm": 0.5261638613155791,
      "learning_rate": 6.912010669543003e-05,
      "loss": 11.9965,
      "step": 22475
    },
    {
      "epoch": 1.2239080271997713,
      "grad_norm": 0.5554353324154613,
      "learning_rate": 6.911171952906381e-05,
      "loss": 11.9629,
      "step": 22476
    },
    {
      "epoch": 1.2239624811963543,
      "grad_norm": 0.5097970921595514,
      "learning_rate": 6.910333260289002e-05,
      "loss": 12.0675,
      "step": 22477
    },
    {
      "epoch": 1.2240169351929373,
      "grad_norm": 0.570485128699268,
      "learning_rate": 6.909494591697391e-05,
      "loss": 11.9794,
      "step": 22478
    },
    {
      "epoch": 1.2240713891895203,
      "grad_norm": 0.5508712214127788,
      "learning_rate": 6.908655947138069e-05,
      "loss": 11.9146,
      "step": 22479
    },
    {
      "epoch": 1.2241258431861033,
      "grad_norm": 0.583189624217757,
      "learning_rate": 6.907817326617558e-05,
      "loss": 12.0606,
      "step": 22480
    },
    {
      "epoch": 1.2241802971826863,
      "grad_norm": 0.5304849872975674,
      "learning_rate": 6.906978730142377e-05,
      "loss": 11.942,
      "step": 22481
    },
    {
      "epoch": 1.2242347511792693,
      "grad_norm": 0.5421844094714305,
      "learning_rate": 6.906140157719048e-05,
      "loss": 11.887,
      "step": 22482
    },
    {
      "epoch": 1.2242892051758525,
      "grad_norm": 0.5693422574327641,
      "learning_rate": 6.90530160935409e-05,
      "loss": 11.9978,
      "step": 22483
    },
    {
      "epoch": 1.2243436591724355,
      "grad_norm": 0.5567753428968648,
      "learning_rate": 6.904463085054028e-05,
      "loss": 11.9522,
      "step": 22484
    },
    {
      "epoch": 1.2243981131690185,
      "grad_norm": 0.5357986277554075,
      "learning_rate": 6.903624584825382e-05,
      "loss": 12.0713,
      "step": 22485
    },
    {
      "epoch": 1.2244525671656015,
      "grad_norm": 0.5721509995507145,
      "learning_rate": 6.902786108674668e-05,
      "loss": 12.022,
      "step": 22486
    },
    {
      "epoch": 1.2245070211621845,
      "grad_norm": 0.5412860233253918,
      "learning_rate": 6.901947656608409e-05,
      "loss": 11.8419,
      "step": 22487
    },
    {
      "epoch": 1.2245614751587675,
      "grad_norm": 0.5522774571518956,
      "learning_rate": 6.901109228633123e-05,
      "loss": 11.9114,
      "step": 22488
    },
    {
      "epoch": 1.2246159291553504,
      "grad_norm": 0.5264038505000138,
      "learning_rate": 6.900270824755333e-05,
      "loss": 11.9053,
      "step": 22489
    },
    {
      "epoch": 1.2246703831519334,
      "grad_norm": 0.5416035535578183,
      "learning_rate": 6.899432444981555e-05,
      "loss": 12.0428,
      "step": 22490
    },
    {
      "epoch": 1.2247248371485164,
      "grad_norm": 0.5632192382301364,
      "learning_rate": 6.898594089318313e-05,
      "loss": 11.7867,
      "step": 22491
    },
    {
      "epoch": 1.2247792911450994,
      "grad_norm": 0.5267971404154178,
      "learning_rate": 6.897755757772122e-05,
      "loss": 11.8979,
      "step": 22492
    },
    {
      "epoch": 1.2248337451416824,
      "grad_norm": 0.5483848947228833,
      "learning_rate": 6.896917450349502e-05,
      "loss": 12.0011,
      "step": 22493
    },
    {
      "epoch": 1.2248881991382654,
      "grad_norm": 0.5449668670560219,
      "learning_rate": 6.896079167056973e-05,
      "loss": 11.9352,
      "step": 22494
    },
    {
      "epoch": 1.2249426531348484,
      "grad_norm": 0.6690299927337976,
      "learning_rate": 6.895240907901056e-05,
      "loss": 12.0275,
      "step": 22495
    },
    {
      "epoch": 1.2249971071314316,
      "grad_norm": 0.5341100907560594,
      "learning_rate": 6.894402672888263e-05,
      "loss": 11.8851,
      "step": 22496
    },
    {
      "epoch": 1.2250515611280146,
      "grad_norm": 0.6020366719678959,
      "learning_rate": 6.893564462025116e-05,
      "loss": 11.8098,
      "step": 22497
    },
    {
      "epoch": 1.2251060151245976,
      "grad_norm": 0.6167733876103798,
      "learning_rate": 6.892726275318133e-05,
      "loss": 12.0455,
      "step": 22498
    },
    {
      "epoch": 1.2251604691211806,
      "grad_norm": 0.5388056884271457,
      "learning_rate": 6.891888112773828e-05,
      "loss": 11.9592,
      "step": 22499
    },
    {
      "epoch": 1.2252149231177636,
      "grad_norm": 0.552119657574981,
      "learning_rate": 6.891049974398727e-05,
      "loss": 11.9421,
      "step": 22500
    },
    {
      "epoch": 1.2252693771143466,
      "grad_norm": 0.5919754129219378,
      "learning_rate": 6.89021186019934e-05,
      "loss": 11.9928,
      "step": 22501
    },
    {
      "epoch": 1.2253238311109296,
      "grad_norm": 0.6203640101726013,
      "learning_rate": 6.889373770182189e-05,
      "loss": 12.0415,
      "step": 22502
    },
    {
      "epoch": 1.2253782851075126,
      "grad_norm": 0.6101859476466426,
      "learning_rate": 6.888535704353789e-05,
      "loss": 12.0055,
      "step": 22503
    },
    {
      "epoch": 1.2254327391040956,
      "grad_norm": 0.5680737713519539,
      "learning_rate": 6.887697662720655e-05,
      "loss": 11.971,
      "step": 22504
    },
    {
      "epoch": 1.2254871931006786,
      "grad_norm": 0.5394158768390048,
      "learning_rate": 6.886859645289312e-05,
      "loss": 11.9425,
      "step": 22505
    },
    {
      "epoch": 1.2255416470972618,
      "grad_norm": 0.5569811013637291,
      "learning_rate": 6.886021652066266e-05,
      "loss": 12.0418,
      "step": 22506
    },
    {
      "epoch": 1.2255961010938448,
      "grad_norm": 0.4913095262555458,
      "learning_rate": 6.885183683058037e-05,
      "loss": 11.9588,
      "step": 22507
    },
    {
      "epoch": 1.2256505550904278,
      "grad_norm": 0.5106557013586046,
      "learning_rate": 6.884345738271144e-05,
      "loss": 11.8817,
      "step": 22508
    },
    {
      "epoch": 1.2257050090870107,
      "grad_norm": 0.5293005486103188,
      "learning_rate": 6.883507817712099e-05,
      "loss": 11.9469,
      "step": 22509
    },
    {
      "epoch": 1.2257594630835937,
      "grad_norm": 0.5521016993068111,
      "learning_rate": 6.882669921387419e-05,
      "loss": 12.1155,
      "step": 22510
    },
    {
      "epoch": 1.2258139170801767,
      "grad_norm": 0.5996763306083485,
      "learning_rate": 6.881832049303622e-05,
      "loss": 11.9832,
      "step": 22511
    },
    {
      "epoch": 1.2258683710767597,
      "grad_norm": 0.49781124861363907,
      "learning_rate": 6.880994201467224e-05,
      "loss": 12.0111,
      "step": 22512
    },
    {
      "epoch": 1.2259228250733427,
      "grad_norm": 0.568234994183213,
      "learning_rate": 6.880156377884736e-05,
      "loss": 12.0161,
      "step": 22513
    },
    {
      "epoch": 1.2259772790699257,
      "grad_norm": 0.5972351646468848,
      "learning_rate": 6.879318578562674e-05,
      "loss": 11.9595,
      "step": 22514
    },
    {
      "epoch": 1.2260317330665087,
      "grad_norm": 0.6252271107783179,
      "learning_rate": 6.878480803507559e-05,
      "loss": 12.1492,
      "step": 22515
    },
    {
      "epoch": 1.2260861870630917,
      "grad_norm": 0.6310811557791951,
      "learning_rate": 6.877643052725898e-05,
      "loss": 11.9716,
      "step": 22516
    },
    {
      "epoch": 1.2261406410596747,
      "grad_norm": 0.5824550513346468,
      "learning_rate": 6.876805326224207e-05,
      "loss": 12.0157,
      "step": 22517
    },
    {
      "epoch": 1.2261950950562577,
      "grad_norm": 0.5659324624616336,
      "learning_rate": 6.875967624009002e-05,
      "loss": 11.9627,
      "step": 22518
    },
    {
      "epoch": 1.226249549052841,
      "grad_norm": 0.5534296008852613,
      "learning_rate": 6.875129946086797e-05,
      "loss": 11.8946,
      "step": 22519
    },
    {
      "epoch": 1.2263040030494239,
      "grad_norm": 0.5444844881677916,
      "learning_rate": 6.874292292464105e-05,
      "loss": 11.9176,
      "step": 22520
    },
    {
      "epoch": 1.2263584570460069,
      "grad_norm": 0.5676441010144988,
      "learning_rate": 6.87345466314744e-05,
      "loss": 11.958,
      "step": 22521
    },
    {
      "epoch": 1.2264129110425899,
      "grad_norm": 0.5618143800630517,
      "learning_rate": 6.872617058143315e-05,
      "loss": 11.8106,
      "step": 22522
    },
    {
      "epoch": 1.2264673650391729,
      "grad_norm": 0.5839632123162697,
      "learning_rate": 6.871779477458246e-05,
      "loss": 11.8015,
      "step": 22523
    },
    {
      "epoch": 1.2265218190357559,
      "grad_norm": 0.5790251698032695,
      "learning_rate": 6.870941921098745e-05,
      "loss": 11.9906,
      "step": 22524
    },
    {
      "epoch": 1.2265762730323388,
      "grad_norm": 0.5921096649894817,
      "learning_rate": 6.870104389071328e-05,
      "loss": 12.1944,
      "step": 22525
    },
    {
      "epoch": 1.2266307270289218,
      "grad_norm": 0.549904217895622,
      "learning_rate": 6.869266881382497e-05,
      "loss": 12.0126,
      "step": 22526
    },
    {
      "epoch": 1.2266851810255048,
      "grad_norm": 0.5672576721021367,
      "learning_rate": 6.868429398038774e-05,
      "loss": 11.9439,
      "step": 22527
    },
    {
      "epoch": 1.2267396350220878,
      "grad_norm": 0.5552647166083431,
      "learning_rate": 6.86759193904667e-05,
      "loss": 11.9606,
      "step": 22528
    },
    {
      "epoch": 1.2267940890186708,
      "grad_norm": 0.5234237128835707,
      "learning_rate": 6.866754504412696e-05,
      "loss": 11.9319,
      "step": 22529
    },
    {
      "epoch": 1.226848543015254,
      "grad_norm": 0.5150598652908518,
      "learning_rate": 6.865917094143365e-05,
      "loss": 11.8037,
      "step": 22530
    },
    {
      "epoch": 1.226902997011837,
      "grad_norm": 0.5196617226235651,
      "learning_rate": 6.865079708245188e-05,
      "loss": 12.0576,
      "step": 22531
    },
    {
      "epoch": 1.22695745100842,
      "grad_norm": 0.5557178074468829,
      "learning_rate": 6.864242346724677e-05,
      "loss": 11.9688,
      "step": 22532
    },
    {
      "epoch": 1.227011905005003,
      "grad_norm": 0.5535336575135944,
      "learning_rate": 6.86340500958834e-05,
      "loss": 12.0447,
      "step": 22533
    },
    {
      "epoch": 1.227066359001586,
      "grad_norm": 0.5763363214579711,
      "learning_rate": 6.862567696842694e-05,
      "loss": 12.0194,
      "step": 22534
    },
    {
      "epoch": 1.227120812998169,
      "grad_norm": 0.569721398763426,
      "learning_rate": 6.861730408494254e-05,
      "loss": 12.0445,
      "step": 22535
    },
    {
      "epoch": 1.227175266994752,
      "grad_norm": 0.5694185737185509,
      "learning_rate": 6.860893144549519e-05,
      "loss": 11.8702,
      "step": 22536
    },
    {
      "epoch": 1.227229720991335,
      "grad_norm": 0.524479808326079,
      "learning_rate": 6.860055905015003e-05,
      "loss": 11.8846,
      "step": 22537
    },
    {
      "epoch": 1.227284174987918,
      "grad_norm": 0.5434479798074577,
      "learning_rate": 6.85921868989722e-05,
      "loss": 11.9942,
      "step": 22538
    },
    {
      "epoch": 1.227338628984501,
      "grad_norm": 0.5475471972796995,
      "learning_rate": 6.858381499202682e-05,
      "loss": 11.9149,
      "step": 22539
    },
    {
      "epoch": 1.227393082981084,
      "grad_norm": 0.578268790355292,
      "learning_rate": 6.857544332937894e-05,
      "loss": 11.9532,
      "step": 22540
    },
    {
      "epoch": 1.227447536977667,
      "grad_norm": 0.5972913130155101,
      "learning_rate": 6.85670719110937e-05,
      "loss": 12.0852,
      "step": 22541
    },
    {
      "epoch": 1.22750199097425,
      "grad_norm": 0.5799477358588745,
      "learning_rate": 6.855870073723616e-05,
      "loss": 12.0768,
      "step": 22542
    },
    {
      "epoch": 1.2275564449708332,
      "grad_norm": 0.6316007411233439,
      "learning_rate": 6.855032980787146e-05,
      "loss": 12.0178,
      "step": 22543
    },
    {
      "epoch": 1.2276108989674162,
      "grad_norm": 0.5529893064266413,
      "learning_rate": 6.854195912306463e-05,
      "loss": 11.7676,
      "step": 22544
    },
    {
      "epoch": 1.2276653529639991,
      "grad_norm": 0.5453360472387511,
      "learning_rate": 6.853358868288082e-05,
      "loss": 11.8291,
      "step": 22545
    },
    {
      "epoch": 1.2277198069605821,
      "grad_norm": 0.5922061215016341,
      "learning_rate": 6.852521848738515e-05,
      "loss": 11.9044,
      "step": 22546
    },
    {
      "epoch": 1.2277742609571651,
      "grad_norm": 0.5260442546307095,
      "learning_rate": 6.851684853664262e-05,
      "loss": 11.9625,
      "step": 22547
    },
    {
      "epoch": 1.2278287149537481,
      "grad_norm": 0.6544268062678831,
      "learning_rate": 6.850847883071832e-05,
      "loss": 11.9651,
      "step": 22548
    },
    {
      "epoch": 1.2278831689503311,
      "grad_norm": 0.5514728833294227,
      "learning_rate": 6.850010936967742e-05,
      "loss": 11.8761,
      "step": 22549
    },
    {
      "epoch": 1.227937622946914,
      "grad_norm": 0.534652900630826,
      "learning_rate": 6.849174015358493e-05,
      "loss": 11.9961,
      "step": 22550
    },
    {
      "epoch": 1.227992076943497,
      "grad_norm": 0.5620353073739606,
      "learning_rate": 6.848337118250596e-05,
      "loss": 12.0729,
      "step": 22551
    },
    {
      "epoch": 1.22804653094008,
      "grad_norm": 0.5092306169240086,
      "learning_rate": 6.847500245650558e-05,
      "loss": 11.8961,
      "step": 22552
    },
    {
      "epoch": 1.2281009849366633,
      "grad_norm": 0.5395441372993375,
      "learning_rate": 6.846663397564885e-05,
      "loss": 11.9984,
      "step": 22553
    },
    {
      "epoch": 1.2281554389332463,
      "grad_norm": 0.555298253458449,
      "learning_rate": 6.845826574000089e-05,
      "loss": 11.9878,
      "step": 22554
    },
    {
      "epoch": 1.2282098929298293,
      "grad_norm": 0.5385689252278301,
      "learning_rate": 6.844989774962671e-05,
      "loss": 11.8666,
      "step": 22555
    },
    {
      "epoch": 1.2282643469264123,
      "grad_norm": 0.6204148714636181,
      "learning_rate": 6.844153000459147e-05,
      "loss": 11.9007,
      "step": 22556
    },
    {
      "epoch": 1.2283188009229953,
      "grad_norm": 0.5367444221664323,
      "learning_rate": 6.843316250496017e-05,
      "loss": 11.9166,
      "step": 22557
    },
    {
      "epoch": 1.2283732549195783,
      "grad_norm": 0.5213410503064586,
      "learning_rate": 6.842479525079787e-05,
      "loss": 12.0957,
      "step": 22558
    },
    {
      "epoch": 1.2284277089161613,
      "grad_norm": 0.5080702699495391,
      "learning_rate": 6.841642824216965e-05,
      "loss": 11.8988,
      "step": 22559
    },
    {
      "epoch": 1.2284821629127443,
      "grad_norm": 0.48398772506905424,
      "learning_rate": 6.840806147914058e-05,
      "loss": 11.9082,
      "step": 22560
    },
    {
      "epoch": 1.2285366169093273,
      "grad_norm": 0.5325699704306123,
      "learning_rate": 6.839969496177574e-05,
      "loss": 11.9781,
      "step": 22561
    },
    {
      "epoch": 1.2285910709059102,
      "grad_norm": 0.5933961282585728,
      "learning_rate": 6.839132869014017e-05,
      "loss": 12.106,
      "step": 22562
    },
    {
      "epoch": 1.2286455249024932,
      "grad_norm": 0.5650542432538881,
      "learning_rate": 6.838296266429893e-05,
      "loss": 11.9166,
      "step": 22563
    },
    {
      "epoch": 1.2286999788990762,
      "grad_norm": 0.5643171401506263,
      "learning_rate": 6.837459688431707e-05,
      "loss": 11.9447,
      "step": 22564
    },
    {
      "epoch": 1.2287544328956592,
      "grad_norm": 0.5603946865686427,
      "learning_rate": 6.836623135025964e-05,
      "loss": 11.9263,
      "step": 22565
    },
    {
      "epoch": 1.2288088868922424,
      "grad_norm": 0.5456345787971714,
      "learning_rate": 6.835786606219174e-05,
      "loss": 12.0358,
      "step": 22566
    },
    {
      "epoch": 1.2288633408888254,
      "grad_norm": 0.535565212057959,
      "learning_rate": 6.834950102017834e-05,
      "loss": 11.9318,
      "step": 22567
    },
    {
      "epoch": 1.2289177948854084,
      "grad_norm": 0.610236385075074,
      "learning_rate": 6.834113622428455e-05,
      "loss": 11.9999,
      "step": 22568
    },
    {
      "epoch": 1.2289722488819914,
      "grad_norm": 0.579675211951449,
      "learning_rate": 6.833277167457536e-05,
      "loss": 12.0322,
      "step": 22569
    },
    {
      "epoch": 1.2290267028785744,
      "grad_norm": 0.5672237234485954,
      "learning_rate": 6.832440737111589e-05,
      "loss": 11.8957,
      "step": 22570
    },
    {
      "epoch": 1.2290811568751574,
      "grad_norm": 0.6164650484087583,
      "learning_rate": 6.831604331397109e-05,
      "loss": 11.9038,
      "step": 22571
    },
    {
      "epoch": 1.2291356108717404,
      "grad_norm": 0.5123542532218338,
      "learning_rate": 6.830767950320608e-05,
      "loss": 11.9202,
      "step": 22572
    },
    {
      "epoch": 1.2291900648683234,
      "grad_norm": 0.6149653410400626,
      "learning_rate": 6.829931593888588e-05,
      "loss": 11.9309,
      "step": 22573
    },
    {
      "epoch": 1.2292445188649064,
      "grad_norm": 0.49101752482147454,
      "learning_rate": 6.82909526210755e-05,
      "loss": 11.8955,
      "step": 22574
    },
    {
      "epoch": 1.2292989728614894,
      "grad_norm": 0.5521005653498695,
      "learning_rate": 6.828258954984e-05,
      "loss": 11.9297,
      "step": 22575
    },
    {
      "epoch": 1.2293534268580726,
      "grad_norm": 0.5256936737272673,
      "learning_rate": 6.82742267252444e-05,
      "loss": 11.9761,
      "step": 22576
    },
    {
      "epoch": 1.2294078808546556,
      "grad_norm": 0.5728446845923955,
      "learning_rate": 6.826586414735374e-05,
      "loss": 11.9677,
      "step": 22577
    },
    {
      "epoch": 1.2294623348512386,
      "grad_norm": 0.5371594106111425,
      "learning_rate": 6.825750181623302e-05,
      "loss": 12.0253,
      "step": 22578
    },
    {
      "epoch": 1.2295167888478216,
      "grad_norm": 0.5341225782071433,
      "learning_rate": 6.82491397319473e-05,
      "loss": 11.9273,
      "step": 22579
    },
    {
      "epoch": 1.2295712428444046,
      "grad_norm": 0.5874889172319521,
      "learning_rate": 6.82407778945616e-05,
      "loss": 11.8723,
      "step": 22580
    },
    {
      "epoch": 1.2296256968409875,
      "grad_norm": 0.5269104334742466,
      "learning_rate": 6.823241630414095e-05,
      "loss": 11.9875,
      "step": 22581
    },
    {
      "epoch": 1.2296801508375705,
      "grad_norm": 0.5422815228069731,
      "learning_rate": 6.822405496075031e-05,
      "loss": 12.0433,
      "step": 22582
    },
    {
      "epoch": 1.2297346048341535,
      "grad_norm": 0.568900395477977,
      "learning_rate": 6.821569386445478e-05,
      "loss": 11.9877,
      "step": 22583
    },
    {
      "epoch": 1.2297890588307365,
      "grad_norm": 0.48023974666047436,
      "learning_rate": 6.820733301531935e-05,
      "loss": 11.8652,
      "step": 22584
    },
    {
      "epoch": 1.2298435128273195,
      "grad_norm": 0.5149114235042755,
      "learning_rate": 6.819897241340902e-05,
      "loss": 11.9277,
      "step": 22585
    },
    {
      "epoch": 1.2298979668239025,
      "grad_norm": 0.5066632032578524,
      "learning_rate": 6.819061205878883e-05,
      "loss": 11.9283,
      "step": 22586
    },
    {
      "epoch": 1.2299524208204855,
      "grad_norm": 0.5299972732180512,
      "learning_rate": 6.818225195152376e-05,
      "loss": 11.9157,
      "step": 22587
    },
    {
      "epoch": 1.2300068748170685,
      "grad_norm": 0.4802202635590141,
      "learning_rate": 6.817389209167883e-05,
      "loss": 11.8945,
      "step": 22588
    },
    {
      "epoch": 1.2300613288136517,
      "grad_norm": 0.5332295352909948,
      "learning_rate": 6.816553247931907e-05,
      "loss": 11.9258,
      "step": 22589
    },
    {
      "epoch": 1.2301157828102347,
      "grad_norm": 0.5719248548778227,
      "learning_rate": 6.815717311450947e-05,
      "loss": 11.8852,
      "step": 22590
    },
    {
      "epoch": 1.2301702368068177,
      "grad_norm": 0.6258691862808952,
      "learning_rate": 6.814881399731501e-05,
      "loss": 12.0193,
      "step": 22591
    },
    {
      "epoch": 1.2302246908034007,
      "grad_norm": 0.5201591560747867,
      "learning_rate": 6.814045512780072e-05,
      "loss": 11.9122,
      "step": 22592
    },
    {
      "epoch": 1.2302791447999837,
      "grad_norm": 0.5890835281633671,
      "learning_rate": 6.813209650603158e-05,
      "loss": 11.9035,
      "step": 22593
    },
    {
      "epoch": 1.2303335987965667,
      "grad_norm": 0.6592927364606244,
      "learning_rate": 6.812373813207262e-05,
      "loss": 11.9232,
      "step": 22594
    },
    {
      "epoch": 1.2303880527931497,
      "grad_norm": 0.5377431673788448,
      "learning_rate": 6.81153800059888e-05,
      "loss": 11.8767,
      "step": 22595
    },
    {
      "epoch": 1.2304425067897327,
      "grad_norm": 0.4924899511594738,
      "learning_rate": 6.810702212784521e-05,
      "loss": 11.8827,
      "step": 22596
    },
    {
      "epoch": 1.2304969607863157,
      "grad_norm": 0.5719182262177606,
      "learning_rate": 6.809866449770668e-05,
      "loss": 11.8717,
      "step": 22597
    },
    {
      "epoch": 1.2305514147828986,
      "grad_norm": 0.5529520024374325,
      "learning_rate": 6.809030711563831e-05,
      "loss": 11.9094,
      "step": 22598
    },
    {
      "epoch": 1.2306058687794816,
      "grad_norm": 0.5762758018285343,
      "learning_rate": 6.808194998170505e-05,
      "loss": 11.9475,
      "step": 22599
    },
    {
      "epoch": 1.2306603227760649,
      "grad_norm": 0.4990905240149167,
      "learning_rate": 6.807359309597192e-05,
      "loss": 11.9709,
      "step": 22600
    },
    {
      "epoch": 1.2307147767726478,
      "grad_norm": 0.5124147148330038,
      "learning_rate": 6.806523645850387e-05,
      "loss": 11.9831,
      "step": 22601
    },
    {
      "epoch": 1.2307692307692308,
      "grad_norm": 0.5695992152209984,
      "learning_rate": 6.805688006936591e-05,
      "loss": 11.8572,
      "step": 22602
    },
    {
      "epoch": 1.2308236847658138,
      "grad_norm": 0.6755224310305814,
      "learning_rate": 6.804852392862299e-05,
      "loss": 11.8961,
      "step": 22603
    },
    {
      "epoch": 1.2308781387623968,
      "grad_norm": 0.546045519852647,
      "learning_rate": 6.80401680363401e-05,
      "loss": 12.0278,
      "step": 22604
    },
    {
      "epoch": 1.2309325927589798,
      "grad_norm": 0.5436839481613359,
      "learning_rate": 6.803181239258223e-05,
      "loss": 11.9962,
      "step": 22605
    },
    {
      "epoch": 1.2309870467555628,
      "grad_norm": 0.5958397610471744,
      "learning_rate": 6.80234569974144e-05,
      "loss": 12.0153,
      "step": 22606
    },
    {
      "epoch": 1.2310415007521458,
      "grad_norm": 0.5994537104580747,
      "learning_rate": 6.801510185090148e-05,
      "loss": 12.0225,
      "step": 22607
    },
    {
      "epoch": 1.2310959547487288,
      "grad_norm": 0.5257403281060432,
      "learning_rate": 6.800674695310848e-05,
      "loss": 11.9313,
      "step": 22608
    },
    {
      "epoch": 1.2311504087453118,
      "grad_norm": 0.5503936262250163,
      "learning_rate": 6.79983923041004e-05,
      "loss": 12.1216,
      "step": 22609
    },
    {
      "epoch": 1.2312048627418948,
      "grad_norm": 0.515757488444049,
      "learning_rate": 6.799003790394218e-05,
      "loss": 11.9001,
      "step": 22610
    },
    {
      "epoch": 1.2312593167384778,
      "grad_norm": 0.6445319175777956,
      "learning_rate": 6.798168375269881e-05,
      "loss": 11.8449,
      "step": 22611
    },
    {
      "epoch": 1.2313137707350608,
      "grad_norm": 0.6110454760683547,
      "learning_rate": 6.797332985043524e-05,
      "loss": 12.0362,
      "step": 22612
    },
    {
      "epoch": 1.231368224731644,
      "grad_norm": 0.5364761671587409,
      "learning_rate": 6.796497619721641e-05,
      "loss": 11.9972,
      "step": 22613
    },
    {
      "epoch": 1.231422678728227,
      "grad_norm": 0.6156344582695163,
      "learning_rate": 6.79566227931073e-05,
      "loss": 11.9898,
      "step": 22614
    },
    {
      "epoch": 1.23147713272481,
      "grad_norm": 0.5757783441910294,
      "learning_rate": 6.794826963817284e-05,
      "loss": 12.0149,
      "step": 22615
    },
    {
      "epoch": 1.231531586721393,
      "grad_norm": 0.5549759110624761,
      "learning_rate": 6.793991673247808e-05,
      "loss": 11.9959,
      "step": 22616
    },
    {
      "epoch": 1.231586040717976,
      "grad_norm": 0.4844406136239094,
      "learning_rate": 6.793156407608788e-05,
      "loss": 11.8264,
      "step": 22617
    },
    {
      "epoch": 1.231640494714559,
      "grad_norm": 0.5388303063570763,
      "learning_rate": 6.792321166906718e-05,
      "loss": 11.9343,
      "step": 22618
    },
    {
      "epoch": 1.231694948711142,
      "grad_norm": 0.5448056529738672,
      "learning_rate": 6.791485951148098e-05,
      "loss": 11.9791,
      "step": 22619
    },
    {
      "epoch": 1.231749402707725,
      "grad_norm": 0.5034495799225918,
      "learning_rate": 6.79065076033942e-05,
      "loss": 11.9938,
      "step": 22620
    },
    {
      "epoch": 1.231803856704308,
      "grad_norm": 0.5739551857827964,
      "learning_rate": 6.789815594487181e-05,
      "loss": 11.9997,
      "step": 22621
    },
    {
      "epoch": 1.231858310700891,
      "grad_norm": 0.5570374439432412,
      "learning_rate": 6.788980453597874e-05,
      "loss": 11.9231,
      "step": 22622
    },
    {
      "epoch": 1.2319127646974741,
      "grad_norm": 0.5414628401592906,
      "learning_rate": 6.788145337677996e-05,
      "loss": 12.04,
      "step": 22623
    },
    {
      "epoch": 1.2319672186940571,
      "grad_norm": 0.592957815956549,
      "learning_rate": 6.787310246734036e-05,
      "loss": 12.0072,
      "step": 22624
    },
    {
      "epoch": 1.2320216726906401,
      "grad_norm": 0.5401153412064662,
      "learning_rate": 6.786475180772492e-05,
      "loss": 12.0457,
      "step": 22625
    },
    {
      "epoch": 1.232076126687223,
      "grad_norm": 0.5082280335715815,
      "learning_rate": 6.785640139799856e-05,
      "loss": 11.9912,
      "step": 22626
    },
    {
      "epoch": 1.232130580683806,
      "grad_norm": 0.6033043900846138,
      "learning_rate": 6.78480512382262e-05,
      "loss": 12.0248,
      "step": 22627
    },
    {
      "epoch": 1.232185034680389,
      "grad_norm": 0.5695370069701157,
      "learning_rate": 6.78397013284728e-05,
      "loss": 11.861,
      "step": 22628
    },
    {
      "epoch": 1.232239488676972,
      "grad_norm": 0.5128287328816599,
      "learning_rate": 6.783135166880326e-05,
      "loss": 11.9754,
      "step": 22629
    },
    {
      "epoch": 1.232293942673555,
      "grad_norm": 0.5195913340414557,
      "learning_rate": 6.782300225928253e-05,
      "loss": 11.8701,
      "step": 22630
    },
    {
      "epoch": 1.232348396670138,
      "grad_norm": 0.5027984525122968,
      "learning_rate": 6.78146530999755e-05,
      "loss": 11.8952,
      "step": 22631
    },
    {
      "epoch": 1.232402850666721,
      "grad_norm": 0.5218699466640082,
      "learning_rate": 6.780630419094715e-05,
      "loss": 11.858,
      "step": 22632
    },
    {
      "epoch": 1.232457304663304,
      "grad_norm": 0.5555496894314489,
      "learning_rate": 6.779795553226239e-05,
      "loss": 11.9742,
      "step": 22633
    },
    {
      "epoch": 1.232511758659887,
      "grad_norm": 0.5483631381595321,
      "learning_rate": 6.778960712398612e-05,
      "loss": 11.8262,
      "step": 22634
    },
    {
      "epoch": 1.23256621265647,
      "grad_norm": 0.6204895092534077,
      "learning_rate": 6.778125896618326e-05,
      "loss": 12.127,
      "step": 22635
    },
    {
      "epoch": 1.2326206666530533,
      "grad_norm": 0.5007419609052064,
      "learning_rate": 6.777291105891876e-05,
      "loss": 11.8728,
      "step": 22636
    },
    {
      "epoch": 1.2326751206496362,
      "grad_norm": 0.5545802228894495,
      "learning_rate": 6.77645634022575e-05,
      "loss": 11.7296,
      "step": 22637
    },
    {
      "epoch": 1.2327295746462192,
      "grad_norm": 0.5471509780195176,
      "learning_rate": 6.775621599626438e-05,
      "loss": 12.0449,
      "step": 22638
    },
    {
      "epoch": 1.2327840286428022,
      "grad_norm": 0.5134534463126355,
      "learning_rate": 6.774786884100435e-05,
      "loss": 11.8806,
      "step": 22639
    },
    {
      "epoch": 1.2328384826393852,
      "grad_norm": 0.5137325719406078,
      "learning_rate": 6.773952193654228e-05,
      "loss": 11.828,
      "step": 22640
    },
    {
      "epoch": 1.2328929366359682,
      "grad_norm": 0.5495636242292773,
      "learning_rate": 6.77311752829431e-05,
      "loss": 12.1211,
      "step": 22641
    },
    {
      "epoch": 1.2329473906325512,
      "grad_norm": 0.4992174580434379,
      "learning_rate": 6.772282888027171e-05,
      "loss": 11.7942,
      "step": 22642
    },
    {
      "epoch": 1.2330018446291342,
      "grad_norm": 0.5063635264465519,
      "learning_rate": 6.771448272859302e-05,
      "loss": 11.9813,
      "step": 22643
    },
    {
      "epoch": 1.2330562986257172,
      "grad_norm": 0.5161442399417954,
      "learning_rate": 6.770613682797193e-05,
      "loss": 11.9526,
      "step": 22644
    },
    {
      "epoch": 1.2331107526223002,
      "grad_norm": 0.5498698779672171,
      "learning_rate": 6.769779117847335e-05,
      "loss": 11.9349,
      "step": 22645
    },
    {
      "epoch": 1.2331652066188834,
      "grad_norm": 0.5908461002668517,
      "learning_rate": 6.768944578016217e-05,
      "loss": 12.0561,
      "step": 22646
    },
    {
      "epoch": 1.2332196606154664,
      "grad_norm": 0.5286359439660577,
      "learning_rate": 6.768110063310325e-05,
      "loss": 11.79,
      "step": 22647
    },
    {
      "epoch": 1.2332741146120494,
      "grad_norm": 0.5591648237871142,
      "learning_rate": 6.767275573736152e-05,
      "loss": 11.9224,
      "step": 22648
    },
    {
      "epoch": 1.2333285686086324,
      "grad_norm": 0.5916860885641544,
      "learning_rate": 6.766441109300187e-05,
      "loss": 12.0007,
      "step": 22649
    },
    {
      "epoch": 1.2333830226052154,
      "grad_norm": 0.5530588051323196,
      "learning_rate": 6.765606670008917e-05,
      "loss": 11.9313,
      "step": 22650
    },
    {
      "epoch": 1.2334374766017984,
      "grad_norm": 0.5291355069703004,
      "learning_rate": 6.764772255868834e-05,
      "loss": 12.0127,
      "step": 22651
    },
    {
      "epoch": 1.2334919305983814,
      "grad_norm": 0.5586532283395358,
      "learning_rate": 6.763937866886422e-05,
      "loss": 11.9683,
      "step": 22652
    },
    {
      "epoch": 1.2335463845949644,
      "grad_norm": 0.631528253563068,
      "learning_rate": 6.763103503068171e-05,
      "loss": 11.8984,
      "step": 22653
    },
    {
      "epoch": 1.2336008385915473,
      "grad_norm": 0.5216279564163208,
      "learning_rate": 6.762269164420572e-05,
      "loss": 11.8951,
      "step": 22654
    },
    {
      "epoch": 1.2336552925881303,
      "grad_norm": 0.5698621996284772,
      "learning_rate": 6.76143485095011e-05,
      "loss": 11.7775,
      "step": 22655
    },
    {
      "epoch": 1.2337097465847133,
      "grad_norm": 0.5700367434440197,
      "learning_rate": 6.760600562663279e-05,
      "loss": 11.982,
      "step": 22656
    },
    {
      "epoch": 1.2337642005812963,
      "grad_norm": 0.5284410018572115,
      "learning_rate": 6.759766299566554e-05,
      "loss": 11.8915,
      "step": 22657
    },
    {
      "epoch": 1.2338186545778793,
      "grad_norm": 0.587121349461119,
      "learning_rate": 6.758932061666431e-05,
      "loss": 12.0293,
      "step": 22658
    },
    {
      "epoch": 1.2338731085744625,
      "grad_norm": 0.5086158498825752,
      "learning_rate": 6.758097848969397e-05,
      "loss": 11.8235,
      "step": 22659
    },
    {
      "epoch": 1.2339275625710455,
      "grad_norm": 0.5458134683974727,
      "learning_rate": 6.75726366148194e-05,
      "loss": 11.7725,
      "step": 22660
    },
    {
      "epoch": 1.2339820165676285,
      "grad_norm": 0.516343793279754,
      "learning_rate": 6.756429499210541e-05,
      "loss": 11.9901,
      "step": 22661
    },
    {
      "epoch": 1.2340364705642115,
      "grad_norm": 0.5301177127299678,
      "learning_rate": 6.755595362161693e-05,
      "loss": 11.984,
      "step": 22662
    },
    {
      "epoch": 1.2340909245607945,
      "grad_norm": 0.5508190832135074,
      "learning_rate": 6.754761250341879e-05,
      "loss": 11.9713,
      "step": 22663
    },
    {
      "epoch": 1.2341453785573775,
      "grad_norm": 0.5372543443815975,
      "learning_rate": 6.753927163757581e-05,
      "loss": 11.9204,
      "step": 22664
    },
    {
      "epoch": 1.2341998325539605,
      "grad_norm": 0.5476867763403248,
      "learning_rate": 6.753093102415295e-05,
      "loss": 11.9879,
      "step": 22665
    },
    {
      "epoch": 1.2342542865505435,
      "grad_norm": 0.5621986188289442,
      "learning_rate": 6.752259066321504e-05,
      "loss": 12.0198,
      "step": 22666
    },
    {
      "epoch": 1.2343087405471265,
      "grad_norm": 0.5251636266109724,
      "learning_rate": 6.751425055482688e-05,
      "loss": 12.0163,
      "step": 22667
    },
    {
      "epoch": 1.2343631945437095,
      "grad_norm": 0.5277914289651588,
      "learning_rate": 6.750591069905334e-05,
      "loss": 11.8987,
      "step": 22668
    },
    {
      "epoch": 1.2344176485402927,
      "grad_norm": 0.5784950361547624,
      "learning_rate": 6.74975710959593e-05,
      "loss": 11.8538,
      "step": 22669
    },
    {
      "epoch": 1.2344721025368757,
      "grad_norm": 0.5752232794584748,
      "learning_rate": 6.748923174560958e-05,
      "loss": 11.9443,
      "step": 22670
    },
    {
      "epoch": 1.2345265565334587,
      "grad_norm": 0.4871636448689952,
      "learning_rate": 6.748089264806907e-05,
      "loss": 11.984,
      "step": 22671
    },
    {
      "epoch": 1.2345810105300417,
      "grad_norm": 0.5971699616649676,
      "learning_rate": 6.747255380340257e-05,
      "loss": 11.897,
      "step": 22672
    },
    {
      "epoch": 1.2346354645266246,
      "grad_norm": 0.5349004302448748,
      "learning_rate": 6.746421521167497e-05,
      "loss": 12.0665,
      "step": 22673
    },
    {
      "epoch": 1.2346899185232076,
      "grad_norm": 0.5500605269688773,
      "learning_rate": 6.745587687295108e-05,
      "loss": 11.8755,
      "step": 22674
    },
    {
      "epoch": 1.2347443725197906,
      "grad_norm": 0.590098038149082,
      "learning_rate": 6.744753878729574e-05,
      "loss": 11.855,
      "step": 22675
    },
    {
      "epoch": 1.2347988265163736,
      "grad_norm": 0.524694545292558,
      "learning_rate": 6.743920095477383e-05,
      "loss": 11.8882,
      "step": 22676
    },
    {
      "epoch": 1.2348532805129566,
      "grad_norm": 0.5093109039751735,
      "learning_rate": 6.743086337545012e-05,
      "loss": 11.7996,
      "step": 22677
    },
    {
      "epoch": 1.2349077345095396,
      "grad_norm": 0.5393763795827721,
      "learning_rate": 6.742252604938949e-05,
      "loss": 11.9541,
      "step": 22678
    },
    {
      "epoch": 1.2349621885061226,
      "grad_norm": 0.5886730619525234,
      "learning_rate": 6.741418897665675e-05,
      "loss": 11.9881,
      "step": 22679
    },
    {
      "epoch": 1.2350166425027056,
      "grad_norm": 0.5932752272406046,
      "learning_rate": 6.740585215731674e-05,
      "loss": 12.0575,
      "step": 22680
    },
    {
      "epoch": 1.2350710964992886,
      "grad_norm": 0.5251342499524592,
      "learning_rate": 6.73975155914343e-05,
      "loss": 12.0279,
      "step": 22681
    },
    {
      "epoch": 1.2351255504958716,
      "grad_norm": 0.5798823011980077,
      "learning_rate": 6.738917927907424e-05,
      "loss": 11.9904,
      "step": 22682
    },
    {
      "epoch": 1.2351800044924548,
      "grad_norm": 0.5514062392490326,
      "learning_rate": 6.73808432203014e-05,
      "loss": 11.9444,
      "step": 22683
    },
    {
      "epoch": 1.2352344584890378,
      "grad_norm": 0.5154261983976824,
      "learning_rate": 6.737250741518058e-05,
      "loss": 12.0411,
      "step": 22684
    },
    {
      "epoch": 1.2352889124856208,
      "grad_norm": 0.5192587201413797,
      "learning_rate": 6.736417186377663e-05,
      "loss": 11.9051,
      "step": 22685
    },
    {
      "epoch": 1.2353433664822038,
      "grad_norm": 0.5458591462569395,
      "learning_rate": 6.735583656615434e-05,
      "loss": 11.8986,
      "step": 22686
    },
    {
      "epoch": 1.2353978204787868,
      "grad_norm": 0.5108083682756132,
      "learning_rate": 6.734750152237856e-05,
      "loss": 11.988,
      "step": 22687
    },
    {
      "epoch": 1.2354522744753698,
      "grad_norm": 0.5506524263833593,
      "learning_rate": 6.733916673251408e-05,
      "loss": 12.0789,
      "step": 22688
    },
    {
      "epoch": 1.2355067284719528,
      "grad_norm": 0.5131917728864864,
      "learning_rate": 6.73308321966257e-05,
      "loss": 11.963,
      "step": 22689
    },
    {
      "epoch": 1.2355611824685357,
      "grad_norm": 0.5439987044445266,
      "learning_rate": 6.732249791477825e-05,
      "loss": 11.9902,
      "step": 22690
    },
    {
      "epoch": 1.2356156364651187,
      "grad_norm": 0.5099899001935355,
      "learning_rate": 6.731416388703652e-05,
      "loss": 12.0014,
      "step": 22691
    },
    {
      "epoch": 1.2356700904617017,
      "grad_norm": 0.5273773552932229,
      "learning_rate": 6.730583011346536e-05,
      "loss": 11.9103,
      "step": 22692
    },
    {
      "epoch": 1.235724544458285,
      "grad_norm": 0.530539161167646,
      "learning_rate": 6.729749659412954e-05,
      "loss": 11.9131,
      "step": 22693
    },
    {
      "epoch": 1.235778998454868,
      "grad_norm": 0.500679073640153,
      "learning_rate": 6.728916332909387e-05,
      "loss": 11.9508,
      "step": 22694
    },
    {
      "epoch": 1.235833452451451,
      "grad_norm": 0.5657946855188445,
      "learning_rate": 6.728083031842315e-05,
      "loss": 11.8095,
      "step": 22695
    },
    {
      "epoch": 1.235887906448034,
      "grad_norm": 0.5454305197094377,
      "learning_rate": 6.727249756218219e-05,
      "loss": 11.8639,
      "step": 22696
    },
    {
      "epoch": 1.235942360444617,
      "grad_norm": 0.5170051698920389,
      "learning_rate": 6.726416506043578e-05,
      "loss": 11.9299,
      "step": 22697
    },
    {
      "epoch": 1.2359968144412,
      "grad_norm": 0.5403322015639688,
      "learning_rate": 6.725583281324871e-05,
      "loss": 11.993,
      "step": 22698
    },
    {
      "epoch": 1.236051268437783,
      "grad_norm": 0.5114468184722819,
      "learning_rate": 6.724750082068576e-05,
      "loss": 11.9085,
      "step": 22699
    },
    {
      "epoch": 1.236105722434366,
      "grad_norm": 0.5425623936775001,
      "learning_rate": 6.723916908281174e-05,
      "loss": 11.9386,
      "step": 22700
    },
    {
      "epoch": 1.2361601764309489,
      "grad_norm": 0.5399263853903555,
      "learning_rate": 6.723083759969145e-05,
      "loss": 11.9577,
      "step": 22701
    },
    {
      "epoch": 1.2362146304275319,
      "grad_norm": 0.5799004326323731,
      "learning_rate": 6.722250637138963e-05,
      "loss": 11.8526,
      "step": 22702
    },
    {
      "epoch": 1.2362690844241149,
      "grad_norm": 0.5418963677157963,
      "learning_rate": 6.721417539797113e-05,
      "loss": 11.9665,
      "step": 22703
    },
    {
      "epoch": 1.2363235384206979,
      "grad_norm": 0.5320204572101112,
      "learning_rate": 6.720584467950068e-05,
      "loss": 11.8885,
      "step": 22704
    },
    {
      "epoch": 1.2363779924172809,
      "grad_norm": 0.5013538756593339,
      "learning_rate": 6.719751421604309e-05,
      "loss": 12.0037,
      "step": 22705
    },
    {
      "epoch": 1.236432446413864,
      "grad_norm": 0.6402306975556139,
      "learning_rate": 6.718918400766312e-05,
      "loss": 11.9172,
      "step": 22706
    },
    {
      "epoch": 1.236486900410447,
      "grad_norm": 0.5376631435166885,
      "learning_rate": 6.71808540544256e-05,
      "loss": 11.9102,
      "step": 22707
    },
    {
      "epoch": 1.23654135440703,
      "grad_norm": 0.5384022363175526,
      "learning_rate": 6.717252435639523e-05,
      "loss": 11.887,
      "step": 22708
    },
    {
      "epoch": 1.236595808403613,
      "grad_norm": 0.5905386507610483,
      "learning_rate": 6.716419491363681e-05,
      "loss": 12.0708,
      "step": 22709
    },
    {
      "epoch": 1.236650262400196,
      "grad_norm": 0.5622731857639514,
      "learning_rate": 6.715586572621512e-05,
      "loss": 11.9543,
      "step": 22710
    },
    {
      "epoch": 1.236704716396779,
      "grad_norm": 0.5143368789981935,
      "learning_rate": 6.714753679419495e-05,
      "loss": 11.9082,
      "step": 22711
    },
    {
      "epoch": 1.236759170393362,
      "grad_norm": 0.5170015509220544,
      "learning_rate": 6.713920811764101e-05,
      "loss": 11.7931,
      "step": 22712
    },
    {
      "epoch": 1.236813624389945,
      "grad_norm": 0.5738164528129192,
      "learning_rate": 6.713087969661808e-05,
      "loss": 11.9441,
      "step": 22713
    },
    {
      "epoch": 1.236868078386528,
      "grad_norm": 0.5163665854909195,
      "learning_rate": 6.712255153119098e-05,
      "loss": 12.0418,
      "step": 22714
    },
    {
      "epoch": 1.236922532383111,
      "grad_norm": 0.5418832478842345,
      "learning_rate": 6.711422362142443e-05,
      "loss": 11.9186,
      "step": 22715
    },
    {
      "epoch": 1.2369769863796942,
      "grad_norm": 0.5630243743266894,
      "learning_rate": 6.710589596738319e-05,
      "loss": 11.8954,
      "step": 22716
    },
    {
      "epoch": 1.2370314403762772,
      "grad_norm": 0.5540732424339679,
      "learning_rate": 6.709756856913203e-05,
      "loss": 12.0347,
      "step": 22717
    },
    {
      "epoch": 1.2370858943728602,
      "grad_norm": 0.5835943891047622,
      "learning_rate": 6.70892414267357e-05,
      "loss": 11.9687,
      "step": 22718
    },
    {
      "epoch": 1.2371403483694432,
      "grad_norm": 0.638789111867418,
      "learning_rate": 6.708091454025891e-05,
      "loss": 12.0823,
      "step": 22719
    },
    {
      "epoch": 1.2371948023660262,
      "grad_norm": 0.5060120714308841,
      "learning_rate": 6.707258790976647e-05,
      "loss": 11.8976,
      "step": 22720
    },
    {
      "epoch": 1.2372492563626092,
      "grad_norm": 0.5404010664963349,
      "learning_rate": 6.706426153532311e-05,
      "loss": 11.9189,
      "step": 22721
    },
    {
      "epoch": 1.2373037103591922,
      "grad_norm": 0.5313752114208501,
      "learning_rate": 6.705593541699358e-05,
      "loss": 11.8907,
      "step": 22722
    },
    {
      "epoch": 1.2373581643557752,
      "grad_norm": 0.5405744362055556,
      "learning_rate": 6.704760955484262e-05,
      "loss": 11.9977,
      "step": 22723
    },
    {
      "epoch": 1.2374126183523582,
      "grad_norm": 0.5053750280482343,
      "learning_rate": 6.703928394893496e-05,
      "loss": 11.916,
      "step": 22724
    },
    {
      "epoch": 1.2374670723489412,
      "grad_norm": 0.6332462231421013,
      "learning_rate": 6.703095859933534e-05,
      "loss": 12.0191,
      "step": 22725
    },
    {
      "epoch": 1.2375215263455241,
      "grad_norm": 0.5620715208301164,
      "learning_rate": 6.702263350610853e-05,
      "loss": 11.9745,
      "step": 22726
    },
    {
      "epoch": 1.2375759803421071,
      "grad_norm": 0.523419185547234,
      "learning_rate": 6.70143086693193e-05,
      "loss": 12.0299,
      "step": 22727
    },
    {
      "epoch": 1.2376304343386901,
      "grad_norm": 0.536929624789274,
      "learning_rate": 6.700598408903231e-05,
      "loss": 12.017,
      "step": 22728
    },
    {
      "epoch": 1.2376848883352733,
      "grad_norm": 0.5240904329169529,
      "learning_rate": 6.69976597653123e-05,
      "loss": 12.0377,
      "step": 22729
    },
    {
      "epoch": 1.2377393423318563,
      "grad_norm": 0.5417193831271322,
      "learning_rate": 6.698933569822401e-05,
      "loss": 11.9908,
      "step": 22730
    },
    {
      "epoch": 1.2377937963284393,
      "grad_norm": 0.5075968817592331,
      "learning_rate": 6.698101188783222e-05,
      "loss": 12.0386,
      "step": 22731
    },
    {
      "epoch": 1.2378482503250223,
      "grad_norm": 0.5439818920805917,
      "learning_rate": 6.697268833420159e-05,
      "loss": 11.9398,
      "step": 22732
    },
    {
      "epoch": 1.2379027043216053,
      "grad_norm": 0.540981612610113,
      "learning_rate": 6.69643650373969e-05,
      "loss": 11.966,
      "step": 22733
    },
    {
      "epoch": 1.2379571583181883,
      "grad_norm": 0.5410089949216403,
      "learning_rate": 6.695604199748282e-05,
      "loss": 11.9297,
      "step": 22734
    },
    {
      "epoch": 1.2380116123147713,
      "grad_norm": 0.5080515709221153,
      "learning_rate": 6.694771921452411e-05,
      "loss": 11.9432,
      "step": 22735
    },
    {
      "epoch": 1.2380660663113543,
      "grad_norm": 0.5701795735195642,
      "learning_rate": 6.693939668858547e-05,
      "loss": 11.9448,
      "step": 22736
    },
    {
      "epoch": 1.2381205203079373,
      "grad_norm": 0.5625459367180433,
      "learning_rate": 6.693107441973166e-05,
      "loss": 12.0541,
      "step": 22737
    },
    {
      "epoch": 1.2381749743045203,
      "grad_norm": 0.5421907134881676,
      "learning_rate": 6.692275240802734e-05,
      "loss": 11.8146,
      "step": 22738
    },
    {
      "epoch": 1.2382294283011035,
      "grad_norm": 0.5157314531188918,
      "learning_rate": 6.691443065353724e-05,
      "loss": 11.9038,
      "step": 22739
    },
    {
      "epoch": 1.2382838822976865,
      "grad_norm": 0.5451684293546573,
      "learning_rate": 6.690610915632605e-05,
      "loss": 11.9046,
      "step": 22740
    },
    {
      "epoch": 1.2383383362942695,
      "grad_norm": 0.5315778944147992,
      "learning_rate": 6.689778791645854e-05,
      "loss": 11.9785,
      "step": 22741
    },
    {
      "epoch": 1.2383927902908525,
      "grad_norm": 0.6178763847409804,
      "learning_rate": 6.688946693399938e-05,
      "loss": 11.9117,
      "step": 22742
    },
    {
      "epoch": 1.2384472442874355,
      "grad_norm": 0.6453867300968816,
      "learning_rate": 6.688114620901327e-05,
      "loss": 11.9758,
      "step": 22743
    },
    {
      "epoch": 1.2385016982840185,
      "grad_norm": 0.5725942618275108,
      "learning_rate": 6.687282574156492e-05,
      "loss": 11.9988,
      "step": 22744
    },
    {
      "epoch": 1.2385561522806015,
      "grad_norm": 0.5849737479201819,
      "learning_rate": 6.686450553171904e-05,
      "loss": 11.944,
      "step": 22745
    },
    {
      "epoch": 1.2386106062771844,
      "grad_norm": 0.5283609630593522,
      "learning_rate": 6.685618557954031e-05,
      "loss": 11.8822,
      "step": 22746
    },
    {
      "epoch": 1.2386650602737674,
      "grad_norm": 0.5740219832501973,
      "learning_rate": 6.684786588509346e-05,
      "loss": 11.9405,
      "step": 22747
    },
    {
      "epoch": 1.2387195142703504,
      "grad_norm": 0.4982715936990555,
      "learning_rate": 6.683954644844316e-05,
      "loss": 11.8322,
      "step": 22748
    },
    {
      "epoch": 1.2387739682669334,
      "grad_norm": 0.6233930846331157,
      "learning_rate": 6.683122726965409e-05,
      "loss": 11.988,
      "step": 22749
    },
    {
      "epoch": 1.2388284222635164,
      "grad_norm": 0.5632371456358553,
      "learning_rate": 6.682290834879096e-05,
      "loss": 11.8357,
      "step": 22750
    },
    {
      "epoch": 1.2388828762600994,
      "grad_norm": 0.5964121805149151,
      "learning_rate": 6.681458968591846e-05,
      "loss": 11.9755,
      "step": 22751
    },
    {
      "epoch": 1.2389373302566826,
      "grad_norm": 0.5756759926454542,
      "learning_rate": 6.680627128110129e-05,
      "loss": 11.9904,
      "step": 22752
    },
    {
      "epoch": 1.2389917842532656,
      "grad_norm": 0.5681878427286643,
      "learning_rate": 6.679795313440412e-05,
      "loss": 11.8841,
      "step": 22753
    },
    {
      "epoch": 1.2390462382498486,
      "grad_norm": 0.5422053691231631,
      "learning_rate": 6.678963524589162e-05,
      "loss": 11.959,
      "step": 22754
    },
    {
      "epoch": 1.2391006922464316,
      "grad_norm": 0.6214782310646312,
      "learning_rate": 6.67813176156285e-05,
      "loss": 12.027,
      "step": 22755
    },
    {
      "epoch": 1.2391551462430146,
      "grad_norm": 0.5894579910660742,
      "learning_rate": 6.67730002436794e-05,
      "loss": 11.9687,
      "step": 22756
    },
    {
      "epoch": 1.2392096002395976,
      "grad_norm": 0.48326707282042,
      "learning_rate": 6.676468313010907e-05,
      "loss": 11.9699,
      "step": 22757
    },
    {
      "epoch": 1.2392640542361806,
      "grad_norm": 0.5369041904412026,
      "learning_rate": 6.67563662749821e-05,
      "loss": 12.0289,
      "step": 22758
    },
    {
      "epoch": 1.2393185082327636,
      "grad_norm": 0.5745372613026676,
      "learning_rate": 6.674804967836321e-05,
      "loss": 11.8012,
      "step": 22759
    },
    {
      "epoch": 1.2393729622293466,
      "grad_norm": 0.5116956700294779,
      "learning_rate": 6.673973334031707e-05,
      "loss": 11.855,
      "step": 22760
    },
    {
      "epoch": 1.2394274162259296,
      "grad_norm": 0.5551668780314658,
      "learning_rate": 6.673141726090833e-05,
      "loss": 12.1515,
      "step": 22761
    },
    {
      "epoch": 1.2394818702225125,
      "grad_norm": 0.5104546891451653,
      "learning_rate": 6.672310144020163e-05,
      "loss": 11.8827,
      "step": 22762
    },
    {
      "epoch": 1.2395363242190958,
      "grad_norm": 0.5674099124673071,
      "learning_rate": 6.671478587826173e-05,
      "loss": 11.9037,
      "step": 22763
    },
    {
      "epoch": 1.2395907782156788,
      "grad_norm": 0.5461280995964353,
      "learning_rate": 6.670647057515322e-05,
      "loss": 11.9623,
      "step": 22764
    },
    {
      "epoch": 1.2396452322122617,
      "grad_norm": 0.5709435991393212,
      "learning_rate": 6.669815553094079e-05,
      "loss": 11.8487,
      "step": 22765
    },
    {
      "epoch": 1.2396996862088447,
      "grad_norm": 0.48453856223587394,
      "learning_rate": 6.668984074568908e-05,
      "loss": 11.881,
      "step": 22766
    },
    {
      "epoch": 1.2397541402054277,
      "grad_norm": 0.6409868095410967,
      "learning_rate": 6.668152621946276e-05,
      "loss": 11.9496,
      "step": 22767
    },
    {
      "epoch": 1.2398085942020107,
      "grad_norm": 0.5477782554657079,
      "learning_rate": 6.667321195232648e-05,
      "loss": 12.0103,
      "step": 22768
    },
    {
      "epoch": 1.2398630481985937,
      "grad_norm": 0.5040437193973005,
      "learning_rate": 6.666489794434487e-05,
      "loss": 11.8298,
      "step": 22769
    },
    {
      "epoch": 1.2399175021951767,
      "grad_norm": 0.5680760203946319,
      "learning_rate": 6.665658419558262e-05,
      "loss": 11.9437,
      "step": 22770
    },
    {
      "epoch": 1.2399719561917597,
      "grad_norm": 0.528085507850227,
      "learning_rate": 6.664827070610436e-05,
      "loss": 11.9805,
      "step": 22771
    },
    {
      "epoch": 1.2400264101883427,
      "grad_norm": 0.5444655791242297,
      "learning_rate": 6.663995747597475e-05,
      "loss": 11.8826,
      "step": 22772
    },
    {
      "epoch": 1.2400808641849257,
      "grad_norm": 0.503196129850154,
      "learning_rate": 6.66316445052584e-05,
      "loss": 11.9569,
      "step": 22773
    },
    {
      "epoch": 1.2401353181815087,
      "grad_norm": 0.4988060771487552,
      "learning_rate": 6.662333179401998e-05,
      "loss": 11.804,
      "step": 22774
    },
    {
      "epoch": 1.2401897721780917,
      "grad_norm": 0.5450478647484439,
      "learning_rate": 6.661501934232414e-05,
      "loss": 12.018,
      "step": 22775
    },
    {
      "epoch": 1.240244226174675,
      "grad_norm": 0.5498181916634691,
      "learning_rate": 6.660670715023551e-05,
      "loss": 11.8347,
      "step": 22776
    },
    {
      "epoch": 1.2402986801712579,
      "grad_norm": 0.5582544321055958,
      "learning_rate": 6.659839521781879e-05,
      "loss": 12.0097,
      "step": 22777
    },
    {
      "epoch": 1.2403531341678409,
      "grad_norm": 0.5152176707698568,
      "learning_rate": 6.659008354513844e-05,
      "loss": 11.7026,
      "step": 22778
    },
    {
      "epoch": 1.2404075881644239,
      "grad_norm": 0.5728070721846744,
      "learning_rate": 6.658177213225927e-05,
      "loss": 11.87,
      "step": 22779
    },
    {
      "epoch": 1.2404620421610069,
      "grad_norm": 0.5975612798629394,
      "learning_rate": 6.657346097924581e-05,
      "loss": 11.8497,
      "step": 22780
    },
    {
      "epoch": 1.2405164961575899,
      "grad_norm": 0.5670380370392075,
      "learning_rate": 6.656515008616275e-05,
      "loss": 12.0412,
      "step": 22781
    },
    {
      "epoch": 1.2405709501541728,
      "grad_norm": 0.5194776893845515,
      "learning_rate": 6.655683945307467e-05,
      "loss": 11.9909,
      "step": 22782
    },
    {
      "epoch": 1.2406254041507558,
      "grad_norm": 0.546511889568762,
      "learning_rate": 6.654852908004623e-05,
      "loss": 11.9428,
      "step": 22783
    },
    {
      "epoch": 1.2406798581473388,
      "grad_norm": 0.5292786096158895,
      "learning_rate": 6.654021896714204e-05,
      "loss": 11.9456,
      "step": 22784
    },
    {
      "epoch": 1.2407343121439218,
      "grad_norm": 0.5195446486772057,
      "learning_rate": 6.653190911442669e-05,
      "loss": 12.0358,
      "step": 22785
    },
    {
      "epoch": 1.240788766140505,
      "grad_norm": 0.5578130700424351,
      "learning_rate": 6.652359952196483e-05,
      "loss": 12.1051,
      "step": 22786
    },
    {
      "epoch": 1.240843220137088,
      "grad_norm": 0.5627174905235722,
      "learning_rate": 6.651529018982113e-05,
      "loss": 11.8365,
      "step": 22787
    },
    {
      "epoch": 1.240897674133671,
      "grad_norm": 0.5522797222027749,
      "learning_rate": 6.65069811180601e-05,
      "loss": 12.0269,
      "step": 22788
    },
    {
      "epoch": 1.240952128130254,
      "grad_norm": 0.5400125516162491,
      "learning_rate": 6.64986723067464e-05,
      "loss": 12.0219,
      "step": 22789
    },
    {
      "epoch": 1.241006582126837,
      "grad_norm": 0.5535811517532464,
      "learning_rate": 6.649036375594466e-05,
      "loss": 11.9918,
      "step": 22790
    },
    {
      "epoch": 1.24106103612342,
      "grad_norm": 0.5981970822688325,
      "learning_rate": 6.648205546571947e-05,
      "loss": 11.8923,
      "step": 22791
    },
    {
      "epoch": 1.241115490120003,
      "grad_norm": 0.49991588153224736,
      "learning_rate": 6.647374743613542e-05,
      "loss": 11.9389,
      "step": 22792
    },
    {
      "epoch": 1.241169944116586,
      "grad_norm": 0.5064208515861376,
      "learning_rate": 6.646543966725715e-05,
      "loss": 12.0333,
      "step": 22793
    },
    {
      "epoch": 1.241224398113169,
      "grad_norm": 0.5021847292174838,
      "learning_rate": 6.645713215914924e-05,
      "loss": 11.8361,
      "step": 22794
    },
    {
      "epoch": 1.241278852109752,
      "grad_norm": 0.5778181717137493,
      "learning_rate": 6.64488249118763e-05,
      "loss": 11.8868,
      "step": 22795
    },
    {
      "epoch": 1.241333306106335,
      "grad_norm": 0.47529978052258676,
      "learning_rate": 6.644051792550288e-05,
      "loss": 11.9687,
      "step": 22796
    },
    {
      "epoch": 1.241387760102918,
      "grad_norm": 0.5428177435372803,
      "learning_rate": 6.643221120009371e-05,
      "loss": 11.9814,
      "step": 22797
    },
    {
      "epoch": 1.241442214099501,
      "grad_norm": 0.5526512328224183,
      "learning_rate": 6.642390473571324e-05,
      "loss": 11.892,
      "step": 22798
    },
    {
      "epoch": 1.2414966680960842,
      "grad_norm": 0.5800933044027282,
      "learning_rate": 6.641559853242612e-05,
      "loss": 12.0973,
      "step": 22799
    },
    {
      "epoch": 1.2415511220926672,
      "grad_norm": 0.6170253827013614,
      "learning_rate": 6.640729259029692e-05,
      "loss": 11.9002,
      "step": 22800
    },
    {
      "epoch": 1.2416055760892502,
      "grad_norm": 0.5343101617382512,
      "learning_rate": 6.639898690939025e-05,
      "loss": 11.9431,
      "step": 22801
    },
    {
      "epoch": 1.2416600300858331,
      "grad_norm": 0.516671138329763,
      "learning_rate": 6.639068148977072e-05,
      "loss": 11.9978,
      "step": 22802
    },
    {
      "epoch": 1.2417144840824161,
      "grad_norm": 0.5553064455594036,
      "learning_rate": 6.638237633150288e-05,
      "loss": 12.0286,
      "step": 22803
    },
    {
      "epoch": 1.2417689380789991,
      "grad_norm": 0.5912290720511462,
      "learning_rate": 6.637407143465131e-05,
      "loss": 11.8816,
      "step": 22804
    },
    {
      "epoch": 1.2418233920755821,
      "grad_norm": 0.5343535316534611,
      "learning_rate": 6.63657667992806e-05,
      "loss": 11.7998,
      "step": 22805
    },
    {
      "epoch": 1.2418778460721651,
      "grad_norm": 0.5206141984023502,
      "learning_rate": 6.635746242545532e-05,
      "loss": 11.9438,
      "step": 22806
    },
    {
      "epoch": 1.241932300068748,
      "grad_norm": 0.6804279007851276,
      "learning_rate": 6.63491583132401e-05,
      "loss": 11.9349,
      "step": 22807
    },
    {
      "epoch": 1.241986754065331,
      "grad_norm": 0.5287261857381383,
      "learning_rate": 6.634085446269944e-05,
      "loss": 11.8697,
      "step": 22808
    },
    {
      "epoch": 1.2420412080619143,
      "grad_norm": 0.5142994957197141,
      "learning_rate": 6.633255087389793e-05,
      "loss": 11.9309,
      "step": 22809
    },
    {
      "epoch": 1.2420956620584973,
      "grad_norm": 0.6111271483717513,
      "learning_rate": 6.632424754690017e-05,
      "loss": 11.9755,
      "step": 22810
    },
    {
      "epoch": 1.2421501160550803,
      "grad_norm": 0.5223875794535703,
      "learning_rate": 6.631594448177066e-05,
      "loss": 11.9082,
      "step": 22811
    },
    {
      "epoch": 1.2422045700516633,
      "grad_norm": 0.5995704030578055,
      "learning_rate": 6.630764167857405e-05,
      "loss": 11.9782,
      "step": 22812
    },
    {
      "epoch": 1.2422590240482463,
      "grad_norm": 0.505948768567176,
      "learning_rate": 6.629933913737486e-05,
      "loss": 11.9421,
      "step": 22813
    },
    {
      "epoch": 1.2423134780448293,
      "grad_norm": 0.54299871443734,
      "learning_rate": 6.629103685823767e-05,
      "loss": 11.9526,
      "step": 22814
    },
    {
      "epoch": 1.2423679320414123,
      "grad_norm": 0.493709313312357,
      "learning_rate": 6.628273484122703e-05,
      "loss": 11.8863,
      "step": 22815
    },
    {
      "epoch": 1.2424223860379953,
      "grad_norm": 0.5821463873061075,
      "learning_rate": 6.62744330864075e-05,
      "loss": 11.8817,
      "step": 22816
    },
    {
      "epoch": 1.2424768400345783,
      "grad_norm": 0.5618459142456248,
      "learning_rate": 6.626613159384366e-05,
      "loss": 11.9891,
      "step": 22817
    },
    {
      "epoch": 1.2425312940311612,
      "grad_norm": 0.5702148586567803,
      "learning_rate": 6.62578303636e-05,
      "loss": 11.9193,
      "step": 22818
    },
    {
      "epoch": 1.2425857480277442,
      "grad_norm": 0.5470756108060282,
      "learning_rate": 6.624952939574111e-05,
      "loss": 12.0095,
      "step": 22819
    },
    {
      "epoch": 1.2426402020243272,
      "grad_norm": 0.5709138327143086,
      "learning_rate": 6.624122869033154e-05,
      "loss": 11.872,
      "step": 22820
    },
    {
      "epoch": 1.2426946560209102,
      "grad_norm": 0.5554455275669883,
      "learning_rate": 6.623292824743585e-05,
      "loss": 11.9505,
      "step": 22821
    },
    {
      "epoch": 1.2427491100174934,
      "grad_norm": 0.5779973836898594,
      "learning_rate": 6.622462806711857e-05,
      "loss": 12.0125,
      "step": 22822
    },
    {
      "epoch": 1.2428035640140764,
      "grad_norm": 0.5528729451071411,
      "learning_rate": 6.621632814944421e-05,
      "loss": 11.9637,
      "step": 22823
    },
    {
      "epoch": 1.2428580180106594,
      "grad_norm": 0.6124236927248429,
      "learning_rate": 6.620802849447738e-05,
      "loss": 12.025,
      "step": 22824
    },
    {
      "epoch": 1.2429124720072424,
      "grad_norm": 0.5262678844303744,
      "learning_rate": 6.61997291022826e-05,
      "loss": 11.8004,
      "step": 22825
    },
    {
      "epoch": 1.2429669260038254,
      "grad_norm": 0.5208419089742101,
      "learning_rate": 6.619142997292437e-05,
      "loss": 11.86,
      "step": 22826
    },
    {
      "epoch": 1.2430213800004084,
      "grad_norm": 0.5465260354290734,
      "learning_rate": 6.618313110646728e-05,
      "loss": 11.9544,
      "step": 22827
    },
    {
      "epoch": 1.2430758339969914,
      "grad_norm": 0.5774287225303286,
      "learning_rate": 6.617483250297582e-05,
      "loss": 12.0348,
      "step": 22828
    },
    {
      "epoch": 1.2431302879935744,
      "grad_norm": 0.6067544016343792,
      "learning_rate": 6.616653416251454e-05,
      "loss": 11.9491,
      "step": 22829
    },
    {
      "epoch": 1.2431847419901574,
      "grad_norm": 0.49681862261512644,
      "learning_rate": 6.615823608514794e-05,
      "loss": 12.0043,
      "step": 22830
    },
    {
      "epoch": 1.2432391959867404,
      "grad_norm": 0.5031150136579183,
      "learning_rate": 6.61499382709406e-05,
      "loss": 11.9967,
      "step": 22831
    },
    {
      "epoch": 1.2432936499833234,
      "grad_norm": 0.5081850415140976,
      "learning_rate": 6.6141640719957e-05,
      "loss": 11.9372,
      "step": 22832
    },
    {
      "epoch": 1.2433481039799066,
      "grad_norm": 0.6067099171071557,
      "learning_rate": 6.61333434322617e-05,
      "loss": 12.0013,
      "step": 22833
    },
    {
      "epoch": 1.2434025579764896,
      "grad_norm": 0.4989131758117803,
      "learning_rate": 6.612504640791914e-05,
      "loss": 11.8733,
      "step": 22834
    },
    {
      "epoch": 1.2434570119730726,
      "grad_norm": 0.5458558129514194,
      "learning_rate": 6.611674964699396e-05,
      "loss": 12.0634,
      "step": 22835
    },
    {
      "epoch": 1.2435114659696556,
      "grad_norm": 0.5371725161108272,
      "learning_rate": 6.61084531495506e-05,
      "loss": 11.7361,
      "step": 22836
    },
    {
      "epoch": 1.2435659199662386,
      "grad_norm": 0.5092776703288111,
      "learning_rate": 6.610015691565359e-05,
      "loss": 12.0028,
      "step": 22837
    },
    {
      "epoch": 1.2436203739628215,
      "grad_norm": 0.555539266294269,
      "learning_rate": 6.609186094536746e-05,
      "loss": 12.0336,
      "step": 22838
    },
    {
      "epoch": 1.2436748279594045,
      "grad_norm": 0.5857264935503956,
      "learning_rate": 6.60835652387567e-05,
      "loss": 12.0261,
      "step": 22839
    },
    {
      "epoch": 1.2437292819559875,
      "grad_norm": 0.5883447922003333,
      "learning_rate": 6.607526979588583e-05,
      "loss": 12.0262,
      "step": 22840
    },
    {
      "epoch": 1.2437837359525705,
      "grad_norm": 0.5495328691077903,
      "learning_rate": 6.606697461681934e-05,
      "loss": 11.9726,
      "step": 22841
    },
    {
      "epoch": 1.2438381899491535,
      "grad_norm": 0.6093131901575297,
      "learning_rate": 6.605867970162174e-05,
      "loss": 11.9135,
      "step": 22842
    },
    {
      "epoch": 1.2438926439457365,
      "grad_norm": 0.542856367570383,
      "learning_rate": 6.605038505035754e-05,
      "loss": 12.0268,
      "step": 22843
    },
    {
      "epoch": 1.2439470979423195,
      "grad_norm": 0.5403828018426662,
      "learning_rate": 6.604209066309124e-05,
      "loss": 11.9622,
      "step": 22844
    },
    {
      "epoch": 1.2440015519389025,
      "grad_norm": 0.5417773340777237,
      "learning_rate": 6.603379653988732e-05,
      "loss": 12.0037,
      "step": 22845
    },
    {
      "epoch": 1.2440560059354857,
      "grad_norm": 0.6233297521403925,
      "learning_rate": 6.602550268081031e-05,
      "loss": 12.0848,
      "step": 22846
    },
    {
      "epoch": 1.2441104599320687,
      "grad_norm": 0.5563424335303409,
      "learning_rate": 6.601720908592471e-05,
      "loss": 11.9075,
      "step": 22847
    },
    {
      "epoch": 1.2441649139286517,
      "grad_norm": 0.5498742965493167,
      "learning_rate": 6.600891575529501e-05,
      "loss": 11.9492,
      "step": 22848
    },
    {
      "epoch": 1.2442193679252347,
      "grad_norm": 0.5495562028184575,
      "learning_rate": 6.600062268898563e-05,
      "loss": 11.9896,
      "step": 22849
    },
    {
      "epoch": 1.2442738219218177,
      "grad_norm": 0.5960210216426902,
      "learning_rate": 6.599232988706112e-05,
      "loss": 12.032,
      "step": 22850
    },
    {
      "epoch": 1.2443282759184007,
      "grad_norm": 0.5405426947554883,
      "learning_rate": 6.598403734958596e-05,
      "loss": 11.881,
      "step": 22851
    },
    {
      "epoch": 1.2443827299149837,
      "grad_norm": 0.6180780284103252,
      "learning_rate": 6.597574507662463e-05,
      "loss": 12.0612,
      "step": 22852
    },
    {
      "epoch": 1.2444371839115667,
      "grad_norm": 0.5053292031113816,
      "learning_rate": 6.596745306824162e-05,
      "loss": 11.999,
      "step": 22853
    },
    {
      "epoch": 1.2444916379081497,
      "grad_norm": 0.6042403860038301,
      "learning_rate": 6.595916132450139e-05,
      "loss": 11.9975,
      "step": 22854
    },
    {
      "epoch": 1.2445460919047326,
      "grad_norm": 0.6292996176273324,
      "learning_rate": 6.595086984546844e-05,
      "loss": 11.965,
      "step": 22855
    },
    {
      "epoch": 1.2446005459013159,
      "grad_norm": 0.5859200357053135,
      "learning_rate": 6.594257863120721e-05,
      "loss": 12.0571,
      "step": 22856
    },
    {
      "epoch": 1.2446549998978989,
      "grad_norm": 0.5907369634070254,
      "learning_rate": 6.593428768178223e-05,
      "loss": 12.0495,
      "step": 22857
    },
    {
      "epoch": 1.2447094538944818,
      "grad_norm": 0.9430568370294133,
      "learning_rate": 6.592599699725796e-05,
      "loss": 11.966,
      "step": 22858
    },
    {
      "epoch": 1.2447639078910648,
      "grad_norm": 0.5372757406434971,
      "learning_rate": 6.591770657769884e-05,
      "loss": 12.0155,
      "step": 22859
    },
    {
      "epoch": 1.2448183618876478,
      "grad_norm": 0.5540848802763377,
      "learning_rate": 6.590941642316931e-05,
      "loss": 12.0618,
      "step": 22860
    },
    {
      "epoch": 1.2448728158842308,
      "grad_norm": 0.5110306054951284,
      "learning_rate": 6.590112653373391e-05,
      "loss": 11.9085,
      "step": 22861
    },
    {
      "epoch": 1.2449272698808138,
      "grad_norm": 0.5478302796397301,
      "learning_rate": 6.589283690945704e-05,
      "loss": 12.1195,
      "step": 22862
    },
    {
      "epoch": 1.2449817238773968,
      "grad_norm": 0.5953012209816627,
      "learning_rate": 6.588454755040322e-05,
      "loss": 11.8825,
      "step": 22863
    },
    {
      "epoch": 1.2450361778739798,
      "grad_norm": 0.6720573249958113,
      "learning_rate": 6.587625845663687e-05,
      "loss": 11.8381,
      "step": 22864
    },
    {
      "epoch": 1.2450906318705628,
      "grad_norm": 0.5689001755365769,
      "learning_rate": 6.586796962822247e-05,
      "loss": 11.9777,
      "step": 22865
    },
    {
      "epoch": 1.2451450858671458,
      "grad_norm": 0.5345723488025051,
      "learning_rate": 6.585968106522443e-05,
      "loss": 11.9733,
      "step": 22866
    },
    {
      "epoch": 1.2451995398637288,
      "grad_norm": 0.5711652832127669,
      "learning_rate": 6.585139276770724e-05,
      "loss": 11.9361,
      "step": 22867
    },
    {
      "epoch": 1.2452539938603118,
      "grad_norm": 0.6009085892003865,
      "learning_rate": 6.58431047357354e-05,
      "loss": 11.9393,
      "step": 22868
    },
    {
      "epoch": 1.245308447856895,
      "grad_norm": 0.5404850062028494,
      "learning_rate": 6.583481696937326e-05,
      "loss": 11.9827,
      "step": 22869
    },
    {
      "epoch": 1.245362901853478,
      "grad_norm": 0.5653755541851823,
      "learning_rate": 6.582652946868532e-05,
      "loss": 11.9453,
      "step": 22870
    },
    {
      "epoch": 1.245417355850061,
      "grad_norm": 0.5379684103837694,
      "learning_rate": 6.5818242233736e-05,
      "loss": 12.0339,
      "step": 22871
    },
    {
      "epoch": 1.245471809846644,
      "grad_norm": 0.5248870771654717,
      "learning_rate": 6.580995526458977e-05,
      "loss": 12.0152,
      "step": 22872
    },
    {
      "epoch": 1.245526263843227,
      "grad_norm": 0.5737092984332973,
      "learning_rate": 6.580166856131106e-05,
      "loss": 11.9311,
      "step": 22873
    },
    {
      "epoch": 1.24558071783981,
      "grad_norm": 0.48237775427526935,
      "learning_rate": 6.579338212396432e-05,
      "loss": 11.9401,
      "step": 22874
    },
    {
      "epoch": 1.245635171836393,
      "grad_norm": 0.5559850193376921,
      "learning_rate": 6.578509595261397e-05,
      "loss": 11.8752,
      "step": 22875
    },
    {
      "epoch": 1.245689625832976,
      "grad_norm": 0.4930299259343834,
      "learning_rate": 6.577681004732445e-05,
      "loss": 11.7788,
      "step": 22876
    },
    {
      "epoch": 1.245744079829559,
      "grad_norm": 0.5179040222610088,
      "learning_rate": 6.576852440816022e-05,
      "loss": 11.9134,
      "step": 22877
    },
    {
      "epoch": 1.245798533826142,
      "grad_norm": 0.510126945991866,
      "learning_rate": 6.576023903518568e-05,
      "loss": 11.8957,
      "step": 22878
    },
    {
      "epoch": 1.2458529878227251,
      "grad_norm": 0.5671174775740991,
      "learning_rate": 6.575195392846525e-05,
      "loss": 11.969,
      "step": 22879
    },
    {
      "epoch": 1.2459074418193081,
      "grad_norm": 0.5147618181834822,
      "learning_rate": 6.574366908806337e-05,
      "loss": 11.9845,
      "step": 22880
    },
    {
      "epoch": 1.2459618958158911,
      "grad_norm": 0.507930474647851,
      "learning_rate": 6.573538451404446e-05,
      "loss": 11.9498,
      "step": 22881
    },
    {
      "epoch": 1.2460163498124741,
      "grad_norm": 0.5310853360787483,
      "learning_rate": 6.572710020647295e-05,
      "loss": 11.9354,
      "step": 22882
    },
    {
      "epoch": 1.246070803809057,
      "grad_norm": 0.5177735012089201,
      "learning_rate": 6.571881616541323e-05,
      "loss": 11.8733,
      "step": 22883
    },
    {
      "epoch": 1.24612525780564,
      "grad_norm": 0.5360896782995199,
      "learning_rate": 6.571053239092977e-05,
      "loss": 12.0134,
      "step": 22884
    },
    {
      "epoch": 1.246179711802223,
      "grad_norm": 0.5017990803176361,
      "learning_rate": 6.570224888308695e-05,
      "loss": 11.938,
      "step": 22885
    },
    {
      "epoch": 1.246234165798806,
      "grad_norm": 0.5250885074961562,
      "learning_rate": 6.569396564194921e-05,
      "loss": 11.9224,
      "step": 22886
    },
    {
      "epoch": 1.246288619795389,
      "grad_norm": 0.5691015680339959,
      "learning_rate": 6.568568266758094e-05,
      "loss": 11.8836,
      "step": 22887
    },
    {
      "epoch": 1.246343073791972,
      "grad_norm": 0.5500882464683317,
      "learning_rate": 6.567739996004658e-05,
      "loss": 12.0345,
      "step": 22888
    },
    {
      "epoch": 1.246397527788555,
      "grad_norm": 0.5518441169966526,
      "learning_rate": 6.56691175194105e-05,
      "loss": 11.9901,
      "step": 22889
    },
    {
      "epoch": 1.246451981785138,
      "grad_norm": 0.5058913121367379,
      "learning_rate": 6.56608353457371e-05,
      "loss": 11.9948,
      "step": 22890
    },
    {
      "epoch": 1.246506435781721,
      "grad_norm": 0.5543399473024002,
      "learning_rate": 6.565255343909081e-05,
      "loss": 11.9444,
      "step": 22891
    },
    {
      "epoch": 1.2465608897783043,
      "grad_norm": 0.4886695931854353,
      "learning_rate": 6.564427179953603e-05,
      "loss": 11.8646,
      "step": 22892
    },
    {
      "epoch": 1.2466153437748873,
      "grad_norm": 0.5083409658185977,
      "learning_rate": 6.563599042713715e-05,
      "loss": 11.9316,
      "step": 22893
    },
    {
      "epoch": 1.2466697977714702,
      "grad_norm": 0.509851838688398,
      "learning_rate": 6.562770932195856e-05,
      "loss": 12.0863,
      "step": 22894
    },
    {
      "epoch": 1.2467242517680532,
      "grad_norm": 0.5570191696101516,
      "learning_rate": 6.561942848406469e-05,
      "loss": 12.0591,
      "step": 22895
    },
    {
      "epoch": 1.2467787057646362,
      "grad_norm": 0.5679042787084743,
      "learning_rate": 6.561114791351987e-05,
      "loss": 12.0477,
      "step": 22896
    },
    {
      "epoch": 1.2468331597612192,
      "grad_norm": 0.5306661509851105,
      "learning_rate": 6.560286761038858e-05,
      "loss": 11.938,
      "step": 22897
    },
    {
      "epoch": 1.2468876137578022,
      "grad_norm": 0.5204155469272607,
      "learning_rate": 6.559458757473516e-05,
      "loss": 11.8639,
      "step": 22898
    },
    {
      "epoch": 1.2469420677543852,
      "grad_norm": 0.4769915772493378,
      "learning_rate": 6.558630780662397e-05,
      "loss": 11.9881,
      "step": 22899
    },
    {
      "epoch": 1.2469965217509682,
      "grad_norm": 0.552021418415857,
      "learning_rate": 6.557802830611943e-05,
      "loss": 11.9603,
      "step": 22900
    },
    {
      "epoch": 1.2470509757475512,
      "grad_norm": 0.6544919271195204,
      "learning_rate": 6.556974907328591e-05,
      "loss": 12.048,
      "step": 22901
    },
    {
      "epoch": 1.2471054297441344,
      "grad_norm": 0.5398390438960115,
      "learning_rate": 6.55614701081878e-05,
      "loss": 11.8334,
      "step": 22902
    },
    {
      "epoch": 1.2471598837407174,
      "grad_norm": 0.5215423436765481,
      "learning_rate": 6.555319141088947e-05,
      "loss": 11.908,
      "step": 22903
    },
    {
      "epoch": 1.2472143377373004,
      "grad_norm": 0.5968230781626701,
      "learning_rate": 6.554491298145531e-05,
      "loss": 11.9538,
      "step": 22904
    },
    {
      "epoch": 1.2472687917338834,
      "grad_norm": 0.5614518888979998,
      "learning_rate": 6.553663481994965e-05,
      "loss": 11.9059,
      "step": 22905
    },
    {
      "epoch": 1.2473232457304664,
      "grad_norm": 0.5533441516707198,
      "learning_rate": 6.552835692643693e-05,
      "loss": 11.9552,
      "step": 22906
    },
    {
      "epoch": 1.2473776997270494,
      "grad_norm": 0.49192685322037805,
      "learning_rate": 6.552007930098147e-05,
      "loss": 11.905,
      "step": 22907
    },
    {
      "epoch": 1.2474321537236324,
      "grad_norm": 0.5075322530471847,
      "learning_rate": 6.551180194364772e-05,
      "loss": 11.9061,
      "step": 22908
    },
    {
      "epoch": 1.2474866077202154,
      "grad_norm": 0.5567091267623645,
      "learning_rate": 6.550352485449991e-05,
      "loss": 11.9821,
      "step": 22909
    },
    {
      "epoch": 1.2475410617167983,
      "grad_norm": 0.543098851357432,
      "learning_rate": 6.549524803360248e-05,
      "loss": 11.9562,
      "step": 22910
    },
    {
      "epoch": 1.2475955157133813,
      "grad_norm": 0.6312837386891761,
      "learning_rate": 6.54869714810198e-05,
      "loss": 11.8395,
      "step": 22911
    },
    {
      "epoch": 1.2476499697099643,
      "grad_norm": 0.736972893975821,
      "learning_rate": 6.547869519681622e-05,
      "loss": 12.0243,
      "step": 22912
    },
    {
      "epoch": 1.2477044237065473,
      "grad_norm": 0.5898809579991051,
      "learning_rate": 6.54704191810561e-05,
      "loss": 12.0607,
      "step": 22913
    },
    {
      "epoch": 1.2477588777031303,
      "grad_norm": 0.5352698872447551,
      "learning_rate": 6.546214343380379e-05,
      "loss": 11.928,
      "step": 22914
    },
    {
      "epoch": 1.2478133316997133,
      "grad_norm": 0.5886613853761997,
      "learning_rate": 6.545386795512364e-05,
      "loss": 11.8943,
      "step": 22915
    },
    {
      "epoch": 1.2478677856962965,
      "grad_norm": 0.5940386684191961,
      "learning_rate": 6.544559274507998e-05,
      "loss": 11.8677,
      "step": 22916
    },
    {
      "epoch": 1.2479222396928795,
      "grad_norm": 0.6104365727146253,
      "learning_rate": 6.543731780373721e-05,
      "loss": 11.9247,
      "step": 22917
    },
    {
      "epoch": 1.2479766936894625,
      "grad_norm": 0.5376032921813544,
      "learning_rate": 6.542904313115971e-05,
      "loss": 12.0488,
      "step": 22918
    },
    {
      "epoch": 1.2480311476860455,
      "grad_norm": 0.5762837312993669,
      "learning_rate": 6.542076872741172e-05,
      "loss": 12.0017,
      "step": 22919
    },
    {
      "epoch": 1.2480856016826285,
      "grad_norm": 0.5363121824057173,
      "learning_rate": 6.54124945925576e-05,
      "loss": 11.9132,
      "step": 22920
    },
    {
      "epoch": 1.2481400556792115,
      "grad_norm": 0.5791488075391481,
      "learning_rate": 6.540422072666175e-05,
      "loss": 11.7995,
      "step": 22921
    },
    {
      "epoch": 1.2481945096757945,
      "grad_norm": 0.5395242927333442,
      "learning_rate": 6.539594712978848e-05,
      "loss": 11.955,
      "step": 22922
    },
    {
      "epoch": 1.2482489636723775,
      "grad_norm": 0.5487681436831426,
      "learning_rate": 6.538767380200212e-05,
      "loss": 11.9998,
      "step": 22923
    },
    {
      "epoch": 1.2483034176689605,
      "grad_norm": 0.5203020553522513,
      "learning_rate": 6.537940074336701e-05,
      "loss": 11.9544,
      "step": 22924
    },
    {
      "epoch": 1.2483578716655435,
      "grad_norm": 0.5889912883526086,
      "learning_rate": 6.537112795394751e-05,
      "loss": 11.9886,
      "step": 22925
    },
    {
      "epoch": 1.2484123256621267,
      "grad_norm": 0.5290888907053558,
      "learning_rate": 6.536285543380791e-05,
      "loss": 11.8611,
      "step": 22926
    },
    {
      "epoch": 1.2484667796587097,
      "grad_norm": 0.528132850681092,
      "learning_rate": 6.535458318301255e-05,
      "loss": 11.9693,
      "step": 22927
    },
    {
      "epoch": 1.2485212336552927,
      "grad_norm": 0.555364238149745,
      "learning_rate": 6.53463112016258e-05,
      "loss": 11.9385,
      "step": 22928
    },
    {
      "epoch": 1.2485756876518757,
      "grad_norm": 0.49825336231402734,
      "learning_rate": 6.53380394897119e-05,
      "loss": 11.8099,
      "step": 22929
    },
    {
      "epoch": 1.2486301416484586,
      "grad_norm": 0.5050365700015226,
      "learning_rate": 6.532976804733524e-05,
      "loss": 11.9691,
      "step": 22930
    },
    {
      "epoch": 1.2486845956450416,
      "grad_norm": 0.5237797915712181,
      "learning_rate": 6.532149687456011e-05,
      "loss": 11.8852,
      "step": 22931
    },
    {
      "epoch": 1.2487390496416246,
      "grad_norm": 0.5184891802825093,
      "learning_rate": 6.531322597145081e-05,
      "loss": 12.0183,
      "step": 22932
    },
    {
      "epoch": 1.2487935036382076,
      "grad_norm": 0.5059313801881709,
      "learning_rate": 6.530495533807171e-05,
      "loss": 11.9459,
      "step": 22933
    },
    {
      "epoch": 1.2488479576347906,
      "grad_norm": 0.5441516367093713,
      "learning_rate": 6.529668497448709e-05,
      "loss": 12.0306,
      "step": 22934
    },
    {
      "epoch": 1.2489024116313736,
      "grad_norm": 0.5734294262571994,
      "learning_rate": 6.528841488076127e-05,
      "loss": 11.9553,
      "step": 22935
    },
    {
      "epoch": 1.2489568656279566,
      "grad_norm": 0.5639771296590832,
      "learning_rate": 6.528014505695856e-05,
      "loss": 11.9432,
      "step": 22936
    },
    {
      "epoch": 1.2490113196245396,
      "grad_norm": 0.549302656599705,
      "learning_rate": 6.527187550314327e-05,
      "loss": 12.0203,
      "step": 22937
    },
    {
      "epoch": 1.2490657736211226,
      "grad_norm": 0.5319682895932197,
      "learning_rate": 6.52636062193797e-05,
      "loss": 11.8934,
      "step": 22938
    },
    {
      "epoch": 1.2491202276177058,
      "grad_norm": 0.5681112030552473,
      "learning_rate": 6.525533720573214e-05,
      "loss": 11.8764,
      "step": 22939
    },
    {
      "epoch": 1.2491746816142888,
      "grad_norm": 0.5331700864501003,
      "learning_rate": 6.524706846226492e-05,
      "loss": 11.8594,
      "step": 22940
    },
    {
      "epoch": 1.2492291356108718,
      "grad_norm": 0.5368385453299201,
      "learning_rate": 6.52387999890423e-05,
      "loss": 12.0113,
      "step": 22941
    },
    {
      "epoch": 1.2492835896074548,
      "grad_norm": 0.561377539879999,
      "learning_rate": 6.523053178612861e-05,
      "loss": 11.8856,
      "step": 22942
    },
    {
      "epoch": 1.2493380436040378,
      "grad_norm": 0.5632375007956382,
      "learning_rate": 6.522226385358813e-05,
      "loss": 12.0217,
      "step": 22943
    },
    {
      "epoch": 1.2493924976006208,
      "grad_norm": 0.5442323860301268,
      "learning_rate": 6.521399619148517e-05,
      "loss": 11.972,
      "step": 22944
    },
    {
      "epoch": 1.2494469515972038,
      "grad_norm": 0.524902231818379,
      "learning_rate": 6.5205728799884e-05,
      "loss": 11.9334,
      "step": 22945
    },
    {
      "epoch": 1.2495014055937868,
      "grad_norm": 0.7055924414692931,
      "learning_rate": 6.519746167884892e-05,
      "loss": 12.0036,
      "step": 22946
    },
    {
      "epoch": 1.2495558595903697,
      "grad_norm": 0.590997549458098,
      "learning_rate": 6.518919482844423e-05,
      "loss": 11.7153,
      "step": 22947
    },
    {
      "epoch": 1.2496103135869527,
      "grad_norm": 0.5579959947996137,
      "learning_rate": 6.518092824873421e-05,
      "loss": 11.8822,
      "step": 22948
    },
    {
      "epoch": 1.249664767583536,
      "grad_norm": 0.5963325373914566,
      "learning_rate": 6.51726619397831e-05,
      "loss": 11.8778,
      "step": 22949
    },
    {
      "epoch": 1.249719221580119,
      "grad_norm": 0.585942340167395,
      "learning_rate": 6.516439590165522e-05,
      "loss": 11.9765,
      "step": 22950
    },
    {
      "epoch": 1.249773675576702,
      "grad_norm": 0.5401228626095188,
      "learning_rate": 6.515613013441485e-05,
      "loss": 11.9633,
      "step": 22951
    },
    {
      "epoch": 1.249828129573285,
      "grad_norm": 0.5686170809381587,
      "learning_rate": 6.514786463812624e-05,
      "loss": 11.8138,
      "step": 22952
    },
    {
      "epoch": 1.249882583569868,
      "grad_norm": 0.5419393697444189,
      "learning_rate": 6.513959941285368e-05,
      "loss": 11.8725,
      "step": 22953
    },
    {
      "epoch": 1.249937037566451,
      "grad_norm": 0.5487208114712231,
      "learning_rate": 6.513133445866143e-05,
      "loss": 11.8953,
      "step": 22954
    },
    {
      "epoch": 1.249991491563034,
      "grad_norm": 0.5181968881859137,
      "learning_rate": 6.512306977561379e-05,
      "loss": 11.8141,
      "step": 22955
    },
    {
      "epoch": 1.250045945559617,
      "grad_norm": 0.5156325427064138,
      "learning_rate": 6.5114805363775e-05,
      "loss": 11.8736,
      "step": 22956
    },
    {
      "epoch": 1.2501003995562,
      "grad_norm": 0.5216489790880355,
      "learning_rate": 6.510654122320934e-05,
      "loss": 11.9813,
      "step": 22957
    },
    {
      "epoch": 1.2501548535527829,
      "grad_norm": 0.5144461249999991,
      "learning_rate": 6.509827735398108e-05,
      "loss": 11.9433,
      "step": 22958
    },
    {
      "epoch": 1.2502093075493659,
      "grad_norm": 0.5668710856366023,
      "learning_rate": 6.509001375615446e-05,
      "loss": 12.0624,
      "step": 22959
    },
    {
      "epoch": 1.2502637615459489,
      "grad_norm": 0.5463624945385009,
      "learning_rate": 6.508175042979374e-05,
      "loss": 11.9766,
      "step": 22960
    },
    {
      "epoch": 1.2503182155425319,
      "grad_norm": 0.489156747425386,
      "learning_rate": 6.507348737496319e-05,
      "loss": 11.9146,
      "step": 22961
    },
    {
      "epoch": 1.2503726695391149,
      "grad_norm": 0.5718373374121186,
      "learning_rate": 6.506522459172705e-05,
      "loss": 12.0009,
      "step": 22962
    },
    {
      "epoch": 1.250427123535698,
      "grad_norm": 0.52519505403957,
      "learning_rate": 6.505696208014959e-05,
      "loss": 11.9269,
      "step": 22963
    },
    {
      "epoch": 1.250481577532281,
      "grad_norm": 0.6130330578240278,
      "learning_rate": 6.504869984029504e-05,
      "loss": 11.9778,
      "step": 22964
    },
    {
      "epoch": 1.250536031528864,
      "grad_norm": 0.5223733110557156,
      "learning_rate": 6.504043787222767e-05,
      "loss": 11.8907,
      "step": 22965
    },
    {
      "epoch": 1.250590485525447,
      "grad_norm": 0.5227047815433213,
      "learning_rate": 6.503217617601171e-05,
      "loss": 12.0034,
      "step": 22966
    },
    {
      "epoch": 1.25064493952203,
      "grad_norm": 0.5840137043222837,
      "learning_rate": 6.502391475171142e-05,
      "loss": 11.9016,
      "step": 22967
    },
    {
      "epoch": 1.250699393518613,
      "grad_norm": 0.6283189772930491,
      "learning_rate": 6.501565359939108e-05,
      "loss": 12.0354,
      "step": 22968
    },
    {
      "epoch": 1.250753847515196,
      "grad_norm": 0.5581576060407885,
      "learning_rate": 6.500739271911482e-05,
      "loss": 11.9421,
      "step": 22969
    },
    {
      "epoch": 1.250808301511779,
      "grad_norm": 0.5156956675751961,
      "learning_rate": 6.499913211094697e-05,
      "loss": 11.9535,
      "step": 22970
    },
    {
      "epoch": 1.250862755508362,
      "grad_norm": 0.5931791196620244,
      "learning_rate": 6.499087177495173e-05,
      "loss": 11.9979,
      "step": 22971
    },
    {
      "epoch": 1.2509172095049452,
      "grad_norm": 0.4876625214204427,
      "learning_rate": 6.498261171119333e-05,
      "loss": 11.7944,
      "step": 22972
    },
    {
      "epoch": 1.2509716635015282,
      "grad_norm": 0.5471450507710717,
      "learning_rate": 6.497435191973605e-05,
      "loss": 11.9792,
      "step": 22973
    },
    {
      "epoch": 1.2510261174981112,
      "grad_norm": 0.5485752705540134,
      "learning_rate": 6.496609240064404e-05,
      "loss": 12.1258,
      "step": 22974
    },
    {
      "epoch": 1.2510805714946942,
      "grad_norm": 0.5385059712319659,
      "learning_rate": 6.495783315398159e-05,
      "loss": 11.9627,
      "step": 22975
    },
    {
      "epoch": 1.2511350254912772,
      "grad_norm": 0.5762942386677303,
      "learning_rate": 6.494957417981291e-05,
      "loss": 12.0554,
      "step": 22976
    },
    {
      "epoch": 1.2511894794878602,
      "grad_norm": 0.530072852836688,
      "learning_rate": 6.49413154782022e-05,
      "loss": 11.8623,
      "step": 22977
    },
    {
      "epoch": 1.2512439334844432,
      "grad_norm": 0.5243166244014196,
      "learning_rate": 6.493305704921371e-05,
      "loss": 11.972,
      "step": 22978
    },
    {
      "epoch": 1.2512983874810262,
      "grad_norm": 0.5025990193059422,
      "learning_rate": 6.49247988929117e-05,
      "loss": 11.9838,
      "step": 22979
    },
    {
      "epoch": 1.2513528414776092,
      "grad_norm": 0.5799394634592401,
      "learning_rate": 6.491654100936029e-05,
      "loss": 12.1679,
      "step": 22980
    },
    {
      "epoch": 1.2514072954741922,
      "grad_norm": 0.5624798675759112,
      "learning_rate": 6.490828339862372e-05,
      "loss": 11.9008,
      "step": 22981
    },
    {
      "epoch": 1.2514617494707752,
      "grad_norm": 0.5181483739429259,
      "learning_rate": 6.490002606076625e-05,
      "loss": 11.9532,
      "step": 22982
    },
    {
      "epoch": 1.2515162034673581,
      "grad_norm": 0.6048615842854305,
      "learning_rate": 6.489176899585205e-05,
      "loss": 12.0913,
      "step": 22983
    },
    {
      "epoch": 1.2515706574639411,
      "grad_norm": 0.5168002705232881,
      "learning_rate": 6.488351220394536e-05,
      "loss": 11.9505,
      "step": 22984
    },
    {
      "epoch": 1.2516251114605241,
      "grad_norm": 0.5534263194047239,
      "learning_rate": 6.487525568511036e-05,
      "loss": 11.9204,
      "step": 22985
    },
    {
      "epoch": 1.2516795654571073,
      "grad_norm": 0.5006750619220163,
      "learning_rate": 6.486699943941126e-05,
      "loss": 11.9048,
      "step": 22986
    },
    {
      "epoch": 1.2517340194536903,
      "grad_norm": 0.5385279523589465,
      "learning_rate": 6.485874346691227e-05,
      "loss": 11.9173,
      "step": 22987
    },
    {
      "epoch": 1.2517884734502733,
      "grad_norm": 0.5664077290372833,
      "learning_rate": 6.485048776767754e-05,
      "loss": 11.9994,
      "step": 22988
    },
    {
      "epoch": 1.2518429274468563,
      "grad_norm": 0.6190079342912472,
      "learning_rate": 6.48422323417714e-05,
      "loss": 11.8936,
      "step": 22989
    },
    {
      "epoch": 1.2518973814434393,
      "grad_norm": 0.5514339130229768,
      "learning_rate": 6.483397718925792e-05,
      "loss": 11.9804,
      "step": 22990
    },
    {
      "epoch": 1.2519518354400223,
      "grad_norm": 0.5127695023833323,
      "learning_rate": 6.482572231020132e-05,
      "loss": 11.8427,
      "step": 22991
    },
    {
      "epoch": 1.2520062894366053,
      "grad_norm": 0.5620487969975653,
      "learning_rate": 6.481746770466577e-05,
      "loss": 11.9001,
      "step": 22992
    },
    {
      "epoch": 1.2520607434331883,
      "grad_norm": 0.49963670589846776,
      "learning_rate": 6.480921337271553e-05,
      "loss": 11.8697,
      "step": 22993
    },
    {
      "epoch": 1.2521151974297713,
      "grad_norm": 0.5557818379722875,
      "learning_rate": 6.480095931441475e-05,
      "loss": 11.8505,
      "step": 22994
    },
    {
      "epoch": 1.2521696514263545,
      "grad_norm": 0.5568201837159433,
      "learning_rate": 6.47927055298276e-05,
      "loss": 12.027,
      "step": 22995
    },
    {
      "epoch": 1.2522241054229375,
      "grad_norm": 0.5674997406215048,
      "learning_rate": 6.478445201901827e-05,
      "loss": 12.0055,
      "step": 22996
    },
    {
      "epoch": 1.2522785594195205,
      "grad_norm": 0.5519263533010362,
      "learning_rate": 6.477619878205095e-05,
      "loss": 11.9743,
      "step": 22997
    },
    {
      "epoch": 1.2523330134161035,
      "grad_norm": 0.49660196651081107,
      "learning_rate": 6.476794581898983e-05,
      "loss": 11.9608,
      "step": 22998
    },
    {
      "epoch": 1.2523874674126865,
      "grad_norm": 0.5599330299498402,
      "learning_rate": 6.475969312989909e-05,
      "loss": 11.9485,
      "step": 22999
    },
    {
      "epoch": 1.2524419214092695,
      "grad_norm": 0.5344015870015618,
      "learning_rate": 6.475144071484285e-05,
      "loss": 11.8989,
      "step": 23000
    },
    {
      "epoch": 1.2524963754058525,
      "grad_norm": 0.5197967280637122,
      "learning_rate": 6.474318857388532e-05,
      "loss": 11.8861,
      "step": 23001
    },
    {
      "epoch": 1.2525508294024355,
      "grad_norm": 0.5200498758350924,
      "learning_rate": 6.473493670709068e-05,
      "loss": 11.8961,
      "step": 23002
    },
    {
      "epoch": 1.2526052833990184,
      "grad_norm": 0.5206830469370257,
      "learning_rate": 6.472668511452307e-05,
      "loss": 11.9804,
      "step": 23003
    },
    {
      "epoch": 1.2526597373956014,
      "grad_norm": 0.5131759205830891,
      "learning_rate": 6.471843379624669e-05,
      "loss": 11.9574,
      "step": 23004
    },
    {
      "epoch": 1.2527141913921844,
      "grad_norm": 0.513270765710931,
      "learning_rate": 6.471018275232568e-05,
      "loss": 11.9623,
      "step": 23005
    },
    {
      "epoch": 1.2527686453887674,
      "grad_norm": 0.5124940921030762,
      "learning_rate": 6.47019319828242e-05,
      "loss": 11.8931,
      "step": 23006
    },
    {
      "epoch": 1.2528230993853504,
      "grad_norm": 0.5672971005079075,
      "learning_rate": 6.469368148780641e-05,
      "loss": 11.9584,
      "step": 23007
    },
    {
      "epoch": 1.2528775533819334,
      "grad_norm": 0.6092767991292602,
      "learning_rate": 6.468543126733651e-05,
      "loss": 12.1172,
      "step": 23008
    },
    {
      "epoch": 1.2529320073785164,
      "grad_norm": 0.5113985421572147,
      "learning_rate": 6.46771813214786e-05,
      "loss": 11.9211,
      "step": 23009
    },
    {
      "epoch": 1.2529864613750996,
      "grad_norm": 0.5015182630609952,
      "learning_rate": 6.466893165029685e-05,
      "loss": 11.9121,
      "step": 23010
    },
    {
      "epoch": 1.2530409153716826,
      "grad_norm": 0.5314409037250363,
      "learning_rate": 6.466068225385542e-05,
      "loss": 11.8639,
      "step": 23011
    },
    {
      "epoch": 1.2530953693682656,
      "grad_norm": 0.545839570387426,
      "learning_rate": 6.465243313221842e-05,
      "loss": 11.9892,
      "step": 23012
    },
    {
      "epoch": 1.2531498233648486,
      "grad_norm": 0.4882250111572915,
      "learning_rate": 6.464418428545006e-05,
      "loss": 11.9823,
      "step": 23013
    },
    {
      "epoch": 1.2532042773614316,
      "grad_norm": 0.5030383229180341,
      "learning_rate": 6.463593571361441e-05,
      "loss": 11.9142,
      "step": 23014
    },
    {
      "epoch": 1.2532587313580146,
      "grad_norm": 0.5949180837371855,
      "learning_rate": 6.46276874167757e-05,
      "loss": 11.9972,
      "step": 23015
    },
    {
      "epoch": 1.2533131853545976,
      "grad_norm": 0.5167607919930672,
      "learning_rate": 6.4619439394998e-05,
      "loss": 11.7229,
      "step": 23016
    },
    {
      "epoch": 1.2533676393511806,
      "grad_norm": 0.5033872352171551,
      "learning_rate": 6.461119164834548e-05,
      "loss": 11.8811,
      "step": 23017
    },
    {
      "epoch": 1.2534220933477638,
      "grad_norm": 0.5670707557320989,
      "learning_rate": 6.460294417688227e-05,
      "loss": 12.1498,
      "step": 23018
    },
    {
      "epoch": 1.2534765473443468,
      "grad_norm": 0.5405924893289286,
      "learning_rate": 6.459469698067253e-05,
      "loss": 11.9884,
      "step": 23019
    },
    {
      "epoch": 1.2535310013409298,
      "grad_norm": 0.604217689453398,
      "learning_rate": 6.458645005978033e-05,
      "loss": 12.097,
      "step": 23020
    },
    {
      "epoch": 1.2535854553375128,
      "grad_norm": 0.49637233258769686,
      "learning_rate": 6.457820341426985e-05,
      "loss": 11.8859,
      "step": 23021
    },
    {
      "epoch": 1.2536399093340957,
      "grad_norm": 0.5607773866657347,
      "learning_rate": 6.456995704420518e-05,
      "loss": 12.0768,
      "step": 23022
    },
    {
      "epoch": 1.2536943633306787,
      "grad_norm": 0.5030846259925406,
      "learning_rate": 6.456171094965049e-05,
      "loss": 11.9584,
      "step": 23023
    },
    {
      "epoch": 1.2537488173272617,
      "grad_norm": 0.571124339836725,
      "learning_rate": 6.455346513066985e-05,
      "loss": 11.9904,
      "step": 23024
    },
    {
      "epoch": 1.2538032713238447,
      "grad_norm": 0.515989313493274,
      "learning_rate": 6.454521958732743e-05,
      "loss": 12.0152,
      "step": 23025
    },
    {
      "epoch": 1.2538577253204277,
      "grad_norm": 0.5371035254206465,
      "learning_rate": 6.45369743196873e-05,
      "loss": 11.8703,
      "step": 23026
    },
    {
      "epoch": 1.2539121793170107,
      "grad_norm": 0.5229543350872914,
      "learning_rate": 6.452872932781363e-05,
      "loss": 11.9173,
      "step": 23027
    },
    {
      "epoch": 1.2539666333135937,
      "grad_norm": 0.5932582677642733,
      "learning_rate": 6.45204846117705e-05,
      "loss": 11.8756,
      "step": 23028
    },
    {
      "epoch": 1.2540210873101767,
      "grad_norm": 0.5891383987686949,
      "learning_rate": 6.451224017162209e-05,
      "loss": 11.965,
      "step": 23029
    },
    {
      "epoch": 1.2540755413067597,
      "grad_norm": 0.4903832737201092,
      "learning_rate": 6.450399600743238e-05,
      "loss": 11.9513,
      "step": 23030
    },
    {
      "epoch": 1.2541299953033427,
      "grad_norm": 0.5674561792441308,
      "learning_rate": 6.449575211926556e-05,
      "loss": 11.8607,
      "step": 23031
    },
    {
      "epoch": 1.2541844492999257,
      "grad_norm": 0.5579231146428775,
      "learning_rate": 6.448750850718575e-05,
      "loss": 12.0236,
      "step": 23032
    },
    {
      "epoch": 1.254238903296509,
      "grad_norm": 0.5181541147940405,
      "learning_rate": 6.4479265171257e-05,
      "loss": 11.8999,
      "step": 23033
    },
    {
      "epoch": 1.2542933572930919,
      "grad_norm": 0.5489469574332925,
      "learning_rate": 6.447102211154346e-05,
      "loss": 11.9517,
      "step": 23034
    },
    {
      "epoch": 1.2543478112896749,
      "grad_norm": 0.5369419671946504,
      "learning_rate": 6.44627793281092e-05,
      "loss": 12.0059,
      "step": 23035
    },
    {
      "epoch": 1.2544022652862579,
      "grad_norm": 0.5083090277377812,
      "learning_rate": 6.445453682101835e-05,
      "loss": 12.0114,
      "step": 23036
    },
    {
      "epoch": 1.2544567192828409,
      "grad_norm": 0.506253196690573,
      "learning_rate": 6.444629459033493e-05,
      "loss": 12.0002,
      "step": 23037
    },
    {
      "epoch": 1.2545111732794239,
      "grad_norm": 0.5636252841187461,
      "learning_rate": 6.443805263612313e-05,
      "loss": 12.0145,
      "step": 23038
    },
    {
      "epoch": 1.2545656272760068,
      "grad_norm": 0.5842933617248807,
      "learning_rate": 6.442981095844702e-05,
      "loss": 11.812,
      "step": 23039
    },
    {
      "epoch": 1.2546200812725898,
      "grad_norm": 0.5633163761859511,
      "learning_rate": 6.442156955737064e-05,
      "loss": 11.9461,
      "step": 23040
    },
    {
      "epoch": 1.2546745352691728,
      "grad_norm": 0.5499040542405221,
      "learning_rate": 6.441332843295807e-05,
      "loss": 12.1107,
      "step": 23041
    },
    {
      "epoch": 1.254728989265756,
      "grad_norm": 0.5445402995069029,
      "learning_rate": 6.440508758527344e-05,
      "loss": 11.9543,
      "step": 23042
    },
    {
      "epoch": 1.254783443262339,
      "grad_norm": 0.5453971556779926,
      "learning_rate": 6.439684701438085e-05,
      "loss": 12.0705,
      "step": 23043
    },
    {
      "epoch": 1.254837897258922,
      "grad_norm": 0.5205463045709341,
      "learning_rate": 6.438860672034433e-05,
      "loss": 11.8273,
      "step": 23044
    },
    {
      "epoch": 1.254892351255505,
      "grad_norm": 0.5060498204368865,
      "learning_rate": 6.438036670322795e-05,
      "loss": 11.8163,
      "step": 23045
    },
    {
      "epoch": 1.254946805252088,
      "grad_norm": 0.5179264361560095,
      "learning_rate": 6.437212696309585e-05,
      "loss": 11.8962,
      "step": 23046
    },
    {
      "epoch": 1.255001259248671,
      "grad_norm": 0.5623083269078556,
      "learning_rate": 6.436388750001205e-05,
      "loss": 12.0951,
      "step": 23047
    },
    {
      "epoch": 1.255055713245254,
      "grad_norm": 0.523609816078576,
      "learning_rate": 6.435564831404061e-05,
      "loss": 11.7966,
      "step": 23048
    },
    {
      "epoch": 1.255110167241837,
      "grad_norm": 0.5472175230259042,
      "learning_rate": 6.434740940524569e-05,
      "loss": 11.8966,
      "step": 23049
    },
    {
      "epoch": 1.25516462123842,
      "grad_norm": 0.5651524739962391,
      "learning_rate": 6.433917077369127e-05,
      "loss": 11.9129,
      "step": 23050
    },
    {
      "epoch": 1.255219075235003,
      "grad_norm": 0.6087141538496955,
      "learning_rate": 6.433093241944141e-05,
      "loss": 11.8583,
      "step": 23051
    },
    {
      "epoch": 1.255273529231586,
      "grad_norm": 0.5504794378725869,
      "learning_rate": 6.432269434256021e-05,
      "loss": 11.9824,
      "step": 23052
    },
    {
      "epoch": 1.255327983228169,
      "grad_norm": 0.5703693768570628,
      "learning_rate": 6.431445654311173e-05,
      "loss": 11.7509,
      "step": 23053
    },
    {
      "epoch": 1.255382437224752,
      "grad_norm": 0.5131923723032333,
      "learning_rate": 6.430621902116e-05,
      "loss": 12.0155,
      "step": 23054
    },
    {
      "epoch": 1.255436891221335,
      "grad_norm": 0.5824598217731434,
      "learning_rate": 6.429798177676913e-05,
      "loss": 12.092,
      "step": 23055
    },
    {
      "epoch": 1.2554913452179182,
      "grad_norm": 0.5023407872327654,
      "learning_rate": 6.428974481000312e-05,
      "loss": 11.9702,
      "step": 23056
    },
    {
      "epoch": 1.2555457992145012,
      "grad_norm": 0.5230683133346614,
      "learning_rate": 6.428150812092606e-05,
      "loss": 12.0129,
      "step": 23057
    },
    {
      "epoch": 1.2556002532110841,
      "grad_norm": 0.5839902902678337,
      "learning_rate": 6.427327170960197e-05,
      "loss": 11.9333,
      "step": 23058
    },
    {
      "epoch": 1.2556547072076671,
      "grad_norm": 0.5607862559971485,
      "learning_rate": 6.426503557609494e-05,
      "loss": 11.8152,
      "step": 23059
    },
    {
      "epoch": 1.2557091612042501,
      "grad_norm": 0.5186679038061481,
      "learning_rate": 6.425679972046895e-05,
      "loss": 11.9656,
      "step": 23060
    },
    {
      "epoch": 1.2557636152008331,
      "grad_norm": 0.5383825335095569,
      "learning_rate": 6.424856414278809e-05,
      "loss": 11.9613,
      "step": 23061
    },
    {
      "epoch": 1.2558180691974161,
      "grad_norm": 0.5256951533910921,
      "learning_rate": 6.424032884311639e-05,
      "loss": 11.9378,
      "step": 23062
    },
    {
      "epoch": 1.2558725231939991,
      "grad_norm": 0.5566260473565764,
      "learning_rate": 6.423209382151787e-05,
      "loss": 11.9813,
      "step": 23063
    },
    {
      "epoch": 1.255926977190582,
      "grad_norm": 0.5047489211992726,
      "learning_rate": 6.422385907805661e-05,
      "loss": 11.7673,
      "step": 23064
    },
    {
      "epoch": 1.2559814311871653,
      "grad_norm": 0.5319156804588298,
      "learning_rate": 6.421562461279662e-05,
      "loss": 11.9302,
      "step": 23065
    },
    {
      "epoch": 1.2560358851837483,
      "grad_norm": 0.5469902777045331,
      "learning_rate": 6.420739042580192e-05,
      "loss": 11.8669,
      "step": 23066
    },
    {
      "epoch": 1.2560903391803313,
      "grad_norm": 0.5895496587336508,
      "learning_rate": 6.419915651713657e-05,
      "loss": 11.8581,
      "step": 23067
    },
    {
      "epoch": 1.2561447931769143,
      "grad_norm": 0.4797947992699836,
      "learning_rate": 6.419092288686458e-05,
      "loss": 11.9102,
      "step": 23068
    },
    {
      "epoch": 1.2561992471734973,
      "grad_norm": 0.5577967038474159,
      "learning_rate": 6.418268953505e-05,
      "loss": 11.9772,
      "step": 23069
    },
    {
      "epoch": 1.2562537011700803,
      "grad_norm": 0.5434598292455445,
      "learning_rate": 6.417445646175682e-05,
      "loss": 11.9336,
      "step": 23070
    },
    {
      "epoch": 1.2563081551666633,
      "grad_norm": 0.5360937510040483,
      "learning_rate": 6.416622366704906e-05,
      "loss": 11.8435,
      "step": 23071
    },
    {
      "epoch": 1.2563626091632463,
      "grad_norm": 0.5290102951434025,
      "learning_rate": 6.415799115099075e-05,
      "loss": 12.0287,
      "step": 23072
    },
    {
      "epoch": 1.2564170631598293,
      "grad_norm": 0.4780939803977824,
      "learning_rate": 6.414975891364591e-05,
      "loss": 11.8835,
      "step": 23073
    },
    {
      "epoch": 1.2564715171564123,
      "grad_norm": 0.5697258236466961,
      "learning_rate": 6.414152695507855e-05,
      "loss": 11.9116,
      "step": 23074
    },
    {
      "epoch": 1.2565259711529952,
      "grad_norm": 0.546484044969463,
      "learning_rate": 6.413329527535272e-05,
      "loss": 11.888,
      "step": 23075
    },
    {
      "epoch": 1.2565804251495782,
      "grad_norm": 0.6539474260009979,
      "learning_rate": 6.412506387453239e-05,
      "loss": 12.0349,
      "step": 23076
    },
    {
      "epoch": 1.2566348791461612,
      "grad_norm": 0.5893862166396721,
      "learning_rate": 6.411683275268157e-05,
      "loss": 11.8892,
      "step": 23077
    },
    {
      "epoch": 1.2566893331427442,
      "grad_norm": 0.5086572468603361,
      "learning_rate": 6.410860190986428e-05,
      "loss": 11.8784,
      "step": 23078
    },
    {
      "epoch": 1.2567437871393274,
      "grad_norm": 0.5472500138255146,
      "learning_rate": 6.410037134614455e-05,
      "loss": 11.8455,
      "step": 23079
    },
    {
      "epoch": 1.2567982411359104,
      "grad_norm": 0.544630527948792,
      "learning_rate": 6.409214106158633e-05,
      "loss": 12.0407,
      "step": 23080
    },
    {
      "epoch": 1.2568526951324934,
      "grad_norm": 0.5492984380849422,
      "learning_rate": 6.408391105625365e-05,
      "loss": 11.9997,
      "step": 23081
    },
    {
      "epoch": 1.2569071491290764,
      "grad_norm": 0.5511564523785917,
      "learning_rate": 6.407568133021048e-05,
      "loss": 12.0376,
      "step": 23082
    },
    {
      "epoch": 1.2569616031256594,
      "grad_norm": 0.6279874531789499,
      "learning_rate": 6.406745188352085e-05,
      "loss": 11.9121,
      "step": 23083
    },
    {
      "epoch": 1.2570160571222424,
      "grad_norm": 0.6220941261245012,
      "learning_rate": 6.405922271624874e-05,
      "loss": 11.9591,
      "step": 23084
    },
    {
      "epoch": 1.2570705111188254,
      "grad_norm": 0.5370646846868051,
      "learning_rate": 6.405099382845814e-05,
      "loss": 11.8795,
      "step": 23085
    },
    {
      "epoch": 1.2571249651154084,
      "grad_norm": 0.5544988273277008,
      "learning_rate": 6.404276522021301e-05,
      "loss": 12.0876,
      "step": 23086
    },
    {
      "epoch": 1.2571794191119914,
      "grad_norm": 0.6192562302073945,
      "learning_rate": 6.40345368915774e-05,
      "loss": 11.9662,
      "step": 23087
    },
    {
      "epoch": 1.2572338731085746,
      "grad_norm": 0.575416522526096,
      "learning_rate": 6.402630884261526e-05,
      "loss": 11.9524,
      "step": 23088
    },
    {
      "epoch": 1.2572883271051576,
      "grad_norm": 0.5161518379421237,
      "learning_rate": 6.401808107339062e-05,
      "loss": 11.9907,
      "step": 23089
    },
    {
      "epoch": 1.2573427811017406,
      "grad_norm": 0.5524182548409718,
      "learning_rate": 6.400985358396733e-05,
      "loss": 11.9012,
      "step": 23090
    },
    {
      "epoch": 1.2573972350983236,
      "grad_norm": 0.5543905883301882,
      "learning_rate": 6.40016263744095e-05,
      "loss": 11.9165,
      "step": 23091
    },
    {
      "epoch": 1.2574516890949066,
      "grad_norm": 0.5261536283933002,
      "learning_rate": 6.399339944478107e-05,
      "loss": 11.7749,
      "step": 23092
    },
    {
      "epoch": 1.2575061430914896,
      "grad_norm": 0.5374076723199579,
      "learning_rate": 6.398517279514598e-05,
      "loss": 11.9491,
      "step": 23093
    },
    {
      "epoch": 1.2575605970880726,
      "grad_norm": 0.5399215177823272,
      "learning_rate": 6.397694642556824e-05,
      "loss": 11.9369,
      "step": 23094
    },
    {
      "epoch": 1.2576150510846555,
      "grad_norm": 0.5332425503379244,
      "learning_rate": 6.396872033611179e-05,
      "loss": 11.9261,
      "step": 23095
    },
    {
      "epoch": 1.2576695050812385,
      "grad_norm": 0.5709318959269863,
      "learning_rate": 6.396049452684062e-05,
      "loss": 11.9361,
      "step": 23096
    },
    {
      "epoch": 1.2577239590778215,
      "grad_norm": 0.6037910756924191,
      "learning_rate": 6.395226899781868e-05,
      "loss": 11.9538,
      "step": 23097
    },
    {
      "epoch": 1.2577784130744045,
      "grad_norm": 0.5358524572601534,
      "learning_rate": 6.394404374910996e-05,
      "loss": 11.9424,
      "step": 23098
    },
    {
      "epoch": 1.2578328670709875,
      "grad_norm": 0.6004955658989394,
      "learning_rate": 6.393581878077844e-05,
      "loss": 11.9896,
      "step": 23099
    },
    {
      "epoch": 1.2578873210675705,
      "grad_norm": 0.5485408023650694,
      "learning_rate": 6.392759409288799e-05,
      "loss": 11.9138,
      "step": 23100
    },
    {
      "epoch": 1.2579417750641535,
      "grad_norm": 0.5666292944739462,
      "learning_rate": 6.391936968550261e-05,
      "loss": 11.931,
      "step": 23101
    },
    {
      "epoch": 1.2579962290607365,
      "grad_norm": 0.5597404353871231,
      "learning_rate": 6.391114555868627e-05,
      "loss": 11.9859,
      "step": 23102
    },
    {
      "epoch": 1.2580506830573197,
      "grad_norm": 0.5670522544599553,
      "learning_rate": 6.390292171250291e-05,
      "loss": 11.8759,
      "step": 23103
    },
    {
      "epoch": 1.2581051370539027,
      "grad_norm": 0.5995999828066928,
      "learning_rate": 6.389469814701651e-05,
      "loss": 11.9792,
      "step": 23104
    },
    {
      "epoch": 1.2581595910504857,
      "grad_norm": 0.5782929205834054,
      "learning_rate": 6.388647486229097e-05,
      "loss": 11.9888,
      "step": 23105
    },
    {
      "epoch": 1.2582140450470687,
      "grad_norm": 0.5484559404524344,
      "learning_rate": 6.387825185839026e-05,
      "loss": 11.8832,
      "step": 23106
    },
    {
      "epoch": 1.2582684990436517,
      "grad_norm": 0.598937853919029,
      "learning_rate": 6.387002913537834e-05,
      "loss": 11.9739,
      "step": 23107
    },
    {
      "epoch": 1.2583229530402347,
      "grad_norm": 0.4995317441061127,
      "learning_rate": 6.38618066933191e-05,
      "loss": 11.9355,
      "step": 23108
    },
    {
      "epoch": 1.2583774070368177,
      "grad_norm": 0.5618095843570149,
      "learning_rate": 6.385358453227657e-05,
      "loss": 12.1109,
      "step": 23109
    },
    {
      "epoch": 1.2584318610334007,
      "grad_norm": 0.5526007604359351,
      "learning_rate": 6.384536265231457e-05,
      "loss": 11.8914,
      "step": 23110
    },
    {
      "epoch": 1.2584863150299836,
      "grad_norm": 0.5531099195759396,
      "learning_rate": 6.383714105349712e-05,
      "loss": 12.0769,
      "step": 23111
    },
    {
      "epoch": 1.2585407690265669,
      "grad_norm": 0.4985342167243614,
      "learning_rate": 6.382891973588809e-05,
      "loss": 11.9721,
      "step": 23112
    },
    {
      "epoch": 1.2585952230231499,
      "grad_norm": 0.556810224128595,
      "learning_rate": 6.382069869955149e-05,
      "loss": 11.8717,
      "step": 23113
    },
    {
      "epoch": 1.2586496770197328,
      "grad_norm": 0.6270485080778043,
      "learning_rate": 6.381247794455118e-05,
      "loss": 11.9444,
      "step": 23114
    },
    {
      "epoch": 1.2587041310163158,
      "grad_norm": 0.5523639002205809,
      "learning_rate": 6.380425747095111e-05,
      "loss": 11.9966,
      "step": 23115
    },
    {
      "epoch": 1.2587585850128988,
      "grad_norm": 0.5353121475579924,
      "learning_rate": 6.379603727881522e-05,
      "loss": 11.8434,
      "step": 23116
    },
    {
      "epoch": 1.2588130390094818,
      "grad_norm": 0.5446101326768436,
      "learning_rate": 6.37878173682074e-05,
      "loss": 11.8856,
      "step": 23117
    },
    {
      "epoch": 1.2588674930060648,
      "grad_norm": 0.6029760601144726,
      "learning_rate": 6.37795977391916e-05,
      "loss": 12.0094,
      "step": 23118
    },
    {
      "epoch": 1.2589219470026478,
      "grad_norm": 0.5865087141597802,
      "learning_rate": 6.37713783918317e-05,
      "loss": 11.9017,
      "step": 23119
    },
    {
      "epoch": 1.2589764009992308,
      "grad_norm": 0.5477665226348177,
      "learning_rate": 6.376315932619169e-05,
      "loss": 11.9262,
      "step": 23120
    },
    {
      "epoch": 1.2590308549958138,
      "grad_norm": 0.5282841499264371,
      "learning_rate": 6.37549405423354e-05,
      "loss": 11.9528,
      "step": 23121
    },
    {
      "epoch": 1.2590853089923968,
      "grad_norm": 0.5412124065498328,
      "learning_rate": 6.374672204032675e-05,
      "loss": 12.0454,
      "step": 23122
    },
    {
      "epoch": 1.2591397629889798,
      "grad_norm": 0.5553462550966848,
      "learning_rate": 6.373850382022965e-05,
      "loss": 11.959,
      "step": 23123
    },
    {
      "epoch": 1.2591942169855628,
      "grad_norm": 0.6113007116293697,
      "learning_rate": 6.373028588210808e-05,
      "loss": 12.0259,
      "step": 23124
    },
    {
      "epoch": 1.2592486709821458,
      "grad_norm": 0.5830883374819321,
      "learning_rate": 6.372206822602586e-05,
      "loss": 11.8463,
      "step": 23125
    },
    {
      "epoch": 1.259303124978729,
      "grad_norm": 0.5504992518560115,
      "learning_rate": 6.371385085204693e-05,
      "loss": 11.9425,
      "step": 23126
    },
    {
      "epoch": 1.259357578975312,
      "grad_norm": 0.576981204386988,
      "learning_rate": 6.370563376023517e-05,
      "loss": 11.8694,
      "step": 23127
    },
    {
      "epoch": 1.259412032971895,
      "grad_norm": 0.5528066597285911,
      "learning_rate": 6.36974169506545e-05,
      "loss": 11.9209,
      "step": 23128
    },
    {
      "epoch": 1.259466486968478,
      "grad_norm": 0.5578907904421315,
      "learning_rate": 6.36892004233688e-05,
      "loss": 12.0281,
      "step": 23129
    },
    {
      "epoch": 1.259520940965061,
      "grad_norm": 0.49908705921764723,
      "learning_rate": 6.368098417844199e-05,
      "loss": 11.9853,
      "step": 23130
    },
    {
      "epoch": 1.259575394961644,
      "grad_norm": 0.6456895948428514,
      "learning_rate": 6.367276821593791e-05,
      "loss": 11.9007,
      "step": 23131
    },
    {
      "epoch": 1.259629848958227,
      "grad_norm": 0.5393688064228819,
      "learning_rate": 6.366455253592048e-05,
      "loss": 11.8418,
      "step": 23132
    },
    {
      "epoch": 1.25968430295481,
      "grad_norm": 0.5367421126036882,
      "learning_rate": 6.365633713845358e-05,
      "loss": 11.9011,
      "step": 23133
    },
    {
      "epoch": 1.259738756951393,
      "grad_norm": 0.5310563904804384,
      "learning_rate": 6.364812202360111e-05,
      "loss": 11.8744,
      "step": 23134
    },
    {
      "epoch": 1.2597932109479761,
      "grad_norm": 0.545635062163532,
      "learning_rate": 6.363990719142691e-05,
      "loss": 12.103,
      "step": 23135
    },
    {
      "epoch": 1.2598476649445591,
      "grad_norm": 0.529964804335486,
      "learning_rate": 6.363169264199491e-05,
      "loss": 12.0641,
      "step": 23136
    },
    {
      "epoch": 1.2599021189411421,
      "grad_norm": 0.5454530713208232,
      "learning_rate": 6.362347837536898e-05,
      "loss": 11.8868,
      "step": 23137
    },
    {
      "epoch": 1.2599565729377251,
      "grad_norm": 0.5138547367416849,
      "learning_rate": 6.361526439161297e-05,
      "loss": 11.9207,
      "step": 23138
    },
    {
      "epoch": 1.260011026934308,
      "grad_norm": 0.5379980430562593,
      "learning_rate": 6.360705069079076e-05,
      "loss": 11.9224,
      "step": 23139
    },
    {
      "epoch": 1.260065480930891,
      "grad_norm": 0.5570539254887212,
      "learning_rate": 6.359883727296625e-05,
      "loss": 12.0006,
      "step": 23140
    },
    {
      "epoch": 1.260119934927474,
      "grad_norm": 0.5660768674787063,
      "learning_rate": 6.359062413820327e-05,
      "loss": 12.0522,
      "step": 23141
    },
    {
      "epoch": 1.260174388924057,
      "grad_norm": 0.49642938404356757,
      "learning_rate": 6.35824112865657e-05,
      "loss": 11.9169,
      "step": 23142
    },
    {
      "epoch": 1.26022884292064,
      "grad_norm": 0.5354280412353992,
      "learning_rate": 6.357419871811741e-05,
      "loss": 11.9543,
      "step": 23143
    },
    {
      "epoch": 1.260283296917223,
      "grad_norm": 0.5997684021188936,
      "learning_rate": 6.356598643292225e-05,
      "loss": 11.9438,
      "step": 23144
    },
    {
      "epoch": 1.260337750913806,
      "grad_norm": 0.5336866194677516,
      "learning_rate": 6.355777443104409e-05,
      "loss": 11.9164,
      "step": 23145
    },
    {
      "epoch": 1.260392204910389,
      "grad_norm": 0.5803367431008287,
      "learning_rate": 6.354956271254678e-05,
      "loss": 12.1357,
      "step": 23146
    },
    {
      "epoch": 1.260446658906972,
      "grad_norm": 0.5217981905140658,
      "learning_rate": 6.35413512774942e-05,
      "loss": 11.9656,
      "step": 23147
    },
    {
      "epoch": 1.260501112903555,
      "grad_norm": 0.5283137024171627,
      "learning_rate": 6.353314012595018e-05,
      "loss": 11.9016,
      "step": 23148
    },
    {
      "epoch": 1.2605555669001383,
      "grad_norm": 0.47655219752417416,
      "learning_rate": 6.352492925797859e-05,
      "loss": 11.9532,
      "step": 23149
    },
    {
      "epoch": 1.2606100208967213,
      "grad_norm": 0.517452982016707,
      "learning_rate": 6.351671867364327e-05,
      "loss": 11.9848,
      "step": 23150
    },
    {
      "epoch": 1.2606644748933042,
      "grad_norm": 0.518384238667633,
      "learning_rate": 6.350850837300805e-05,
      "loss": 11.9969,
      "step": 23151
    },
    {
      "epoch": 1.2607189288898872,
      "grad_norm": 0.5072064806068445,
      "learning_rate": 6.35002983561368e-05,
      "loss": 12.031,
      "step": 23152
    },
    {
      "epoch": 1.2607733828864702,
      "grad_norm": 0.49020800588185137,
      "learning_rate": 6.349208862309334e-05,
      "loss": 11.8534,
      "step": 23153
    },
    {
      "epoch": 1.2608278368830532,
      "grad_norm": 0.46912186730601757,
      "learning_rate": 6.348387917394152e-05,
      "loss": 11.8691,
      "step": 23154
    },
    {
      "epoch": 1.2608822908796362,
      "grad_norm": 0.5524909289000592,
      "learning_rate": 6.347567000874519e-05,
      "loss": 12.024,
      "step": 23155
    },
    {
      "epoch": 1.2609367448762192,
      "grad_norm": 0.5420559822279098,
      "learning_rate": 6.346746112756816e-05,
      "loss": 11.7939,
      "step": 23156
    },
    {
      "epoch": 1.2609911988728022,
      "grad_norm": 0.6005430952423993,
      "learning_rate": 6.345925253047426e-05,
      "loss": 11.8845,
      "step": 23157
    },
    {
      "epoch": 1.2610456528693854,
      "grad_norm": 0.584496432963144,
      "learning_rate": 6.345104421752737e-05,
      "loss": 12.0256,
      "step": 23158
    },
    {
      "epoch": 1.2611001068659684,
      "grad_norm": 0.5289213329822015,
      "learning_rate": 6.344283618879128e-05,
      "loss": 11.8286,
      "step": 23159
    },
    {
      "epoch": 1.2611545608625514,
      "grad_norm": 0.5215239341627839,
      "learning_rate": 6.343462844432988e-05,
      "loss": 11.9474,
      "step": 23160
    },
    {
      "epoch": 1.2612090148591344,
      "grad_norm": 0.5349334059457926,
      "learning_rate": 6.342642098420688e-05,
      "loss": 12.0584,
      "step": 23161
    },
    {
      "epoch": 1.2612634688557174,
      "grad_norm": 0.5446407112239008,
      "learning_rate": 6.341821380848618e-05,
      "loss": 12.0434,
      "step": 23162
    },
    {
      "epoch": 1.2613179228523004,
      "grad_norm": 0.564088965609529,
      "learning_rate": 6.341000691723158e-05,
      "loss": 11.9337,
      "step": 23163
    },
    {
      "epoch": 1.2613723768488834,
      "grad_norm": 0.5340390154873355,
      "learning_rate": 6.340180031050691e-05,
      "loss": 11.8474,
      "step": 23164
    },
    {
      "epoch": 1.2614268308454664,
      "grad_norm": 0.47720631400330055,
      "learning_rate": 6.339359398837596e-05,
      "loss": 11.8776,
      "step": 23165
    },
    {
      "epoch": 1.2614812848420494,
      "grad_norm": 0.5736225944813493,
      "learning_rate": 6.338538795090258e-05,
      "loss": 12.0683,
      "step": 23166
    },
    {
      "epoch": 1.2615357388386323,
      "grad_norm": 0.5591859961628993,
      "learning_rate": 6.337718219815057e-05,
      "loss": 12.0328,
      "step": 23167
    },
    {
      "epoch": 1.2615901928352153,
      "grad_norm": 0.539813660015235,
      "learning_rate": 6.336897673018369e-05,
      "loss": 12.006,
      "step": 23168
    },
    {
      "epoch": 1.2616446468317983,
      "grad_norm": 0.47370333657589964,
      "learning_rate": 6.336077154706581e-05,
      "loss": 11.9422,
      "step": 23169
    },
    {
      "epoch": 1.2616991008283813,
      "grad_norm": 0.5175435505840981,
      "learning_rate": 6.335256664886078e-05,
      "loss": 11.8468,
      "step": 23170
    },
    {
      "epoch": 1.2617535548249643,
      "grad_norm": 0.5364831212928655,
      "learning_rate": 6.334436203563228e-05,
      "loss": 11.9393,
      "step": 23171
    },
    {
      "epoch": 1.2618080088215473,
      "grad_norm": 0.5821503420255064,
      "learning_rate": 6.333615770744414e-05,
      "loss": 11.9126,
      "step": 23172
    },
    {
      "epoch": 1.2618624628181305,
      "grad_norm": 0.5733777960952744,
      "learning_rate": 6.332795366436024e-05,
      "loss": 12.055,
      "step": 23173
    },
    {
      "epoch": 1.2619169168147135,
      "grad_norm": 0.5389332093998159,
      "learning_rate": 6.33197499064443e-05,
      "loss": 11.9563,
      "step": 23174
    },
    {
      "epoch": 1.2619713708112965,
      "grad_norm": 0.5275144965712444,
      "learning_rate": 6.331154643376012e-05,
      "loss": 11.7789,
      "step": 23175
    },
    {
      "epoch": 1.2620258248078795,
      "grad_norm": 0.5537466378450545,
      "learning_rate": 6.330334324637153e-05,
      "loss": 11.935,
      "step": 23176
    },
    {
      "epoch": 1.2620802788044625,
      "grad_norm": 0.5470867083695272,
      "learning_rate": 6.329514034434229e-05,
      "loss": 11.9791,
      "step": 23177
    },
    {
      "epoch": 1.2621347328010455,
      "grad_norm": 0.5287713218484725,
      "learning_rate": 6.328693772773619e-05,
      "loss": 12.1183,
      "step": 23178
    },
    {
      "epoch": 1.2621891867976285,
      "grad_norm": 0.5675339930776699,
      "learning_rate": 6.327873539661701e-05,
      "loss": 11.777,
      "step": 23179
    },
    {
      "epoch": 1.2622436407942115,
      "grad_norm": 0.5937655479464573,
      "learning_rate": 6.327053335104858e-05,
      "loss": 11.9364,
      "step": 23180
    },
    {
      "epoch": 1.2622980947907947,
      "grad_norm": 0.5768170937460548,
      "learning_rate": 6.326233159109462e-05,
      "loss": 11.8478,
      "step": 23181
    },
    {
      "epoch": 1.2623525487873777,
      "grad_norm": 0.5116243289974295,
      "learning_rate": 6.325413011681893e-05,
      "loss": 11.8969,
      "step": 23182
    },
    {
      "epoch": 1.2624070027839607,
      "grad_norm": 0.5474243620780744,
      "learning_rate": 6.32459289282853e-05,
      "loss": 11.9277,
      "step": 23183
    },
    {
      "epoch": 1.2624614567805437,
      "grad_norm": 0.5677173261618479,
      "learning_rate": 6.323772802555745e-05,
      "loss": 11.9387,
      "step": 23184
    },
    {
      "epoch": 1.2625159107771267,
      "grad_norm": 0.5112451267947925,
      "learning_rate": 6.32295274086992e-05,
      "loss": 11.8689,
      "step": 23185
    },
    {
      "epoch": 1.2625703647737097,
      "grad_norm": 0.512827670829193,
      "learning_rate": 6.322132707777433e-05,
      "loss": 11.971,
      "step": 23186
    },
    {
      "epoch": 1.2626248187702926,
      "grad_norm": 0.5207663536707516,
      "learning_rate": 6.321312703284658e-05,
      "loss": 11.8929,
      "step": 23187
    },
    {
      "epoch": 1.2626792727668756,
      "grad_norm": 0.5155169795566987,
      "learning_rate": 6.320492727397974e-05,
      "loss": 11.9738,
      "step": 23188
    },
    {
      "epoch": 1.2627337267634586,
      "grad_norm": 0.5314580142077816,
      "learning_rate": 6.319672780123755e-05,
      "loss": 11.9145,
      "step": 23189
    },
    {
      "epoch": 1.2627881807600416,
      "grad_norm": 0.5421368749543253,
      "learning_rate": 6.318852861468378e-05,
      "loss": 12.0622,
      "step": 23190
    },
    {
      "epoch": 1.2628426347566246,
      "grad_norm": 0.5179067735758253,
      "learning_rate": 6.31803297143822e-05,
      "loss": 11.9269,
      "step": 23191
    },
    {
      "epoch": 1.2628970887532076,
      "grad_norm": 0.5736250693930064,
      "learning_rate": 6.317213110039651e-05,
      "loss": 11.7545,
      "step": 23192
    },
    {
      "epoch": 1.2629515427497906,
      "grad_norm": 0.5697848875943401,
      "learning_rate": 6.316393277279053e-05,
      "loss": 12.0713,
      "step": 23193
    },
    {
      "epoch": 1.2630059967463736,
      "grad_norm": 0.4869568018442,
      "learning_rate": 6.315573473162797e-05,
      "loss": 11.9277,
      "step": 23194
    },
    {
      "epoch": 1.2630604507429566,
      "grad_norm": 0.5616724562357337,
      "learning_rate": 6.314753697697258e-05,
      "loss": 12.0319,
      "step": 23195
    },
    {
      "epoch": 1.2631149047395398,
      "grad_norm": 0.5267218874973129,
      "learning_rate": 6.313933950888815e-05,
      "loss": 12.0089,
      "step": 23196
    },
    {
      "epoch": 1.2631693587361228,
      "grad_norm": 0.5520135585587579,
      "learning_rate": 6.31311423274384e-05,
      "loss": 12.0212,
      "step": 23197
    },
    {
      "epoch": 1.2632238127327058,
      "grad_norm": 0.5351385629875579,
      "learning_rate": 6.312294543268706e-05,
      "loss": 11.7936,
      "step": 23198
    },
    {
      "epoch": 1.2632782667292888,
      "grad_norm": 0.5610677188971568,
      "learning_rate": 6.31147488246979e-05,
      "loss": 11.9184,
      "step": 23199
    },
    {
      "epoch": 1.2633327207258718,
      "grad_norm": 0.5747318834051146,
      "learning_rate": 6.310655250353464e-05,
      "loss": 11.9075,
      "step": 23200
    },
    {
      "epoch": 1.2633871747224548,
      "grad_norm": 0.6582293748444622,
      "learning_rate": 6.3098356469261e-05,
      "loss": 12.006,
      "step": 23201
    },
    {
      "epoch": 1.2634416287190378,
      "grad_norm": 0.5305480900275891,
      "learning_rate": 6.309016072194071e-05,
      "loss": 11.9112,
      "step": 23202
    },
    {
      "epoch": 1.2634960827156207,
      "grad_norm": 0.5357069806269253,
      "learning_rate": 6.308196526163755e-05,
      "loss": 11.9746,
      "step": 23203
    },
    {
      "epoch": 1.2635505367122037,
      "grad_norm": 0.5747193317278483,
      "learning_rate": 6.30737700884152e-05,
      "loss": 12.0809,
      "step": 23204
    },
    {
      "epoch": 1.263604990708787,
      "grad_norm": 0.5651868305719783,
      "learning_rate": 6.306557520233741e-05,
      "loss": 11.8314,
      "step": 23205
    },
    {
      "epoch": 1.26365944470537,
      "grad_norm": 0.5116818528225225,
      "learning_rate": 6.305738060346788e-05,
      "loss": 11.9867,
      "step": 23206
    },
    {
      "epoch": 1.263713898701953,
      "grad_norm": 0.6191881391424359,
      "learning_rate": 6.304918629187037e-05,
      "loss": 12.0998,
      "step": 23207
    },
    {
      "epoch": 1.263768352698536,
      "grad_norm": 0.5084801937641094,
      "learning_rate": 6.30409922676086e-05,
      "loss": 11.8947,
      "step": 23208
    },
    {
      "epoch": 1.263822806695119,
      "grad_norm": 0.5234002530945163,
      "learning_rate": 6.303279853074626e-05,
      "loss": 11.805,
      "step": 23209
    },
    {
      "epoch": 1.263877260691702,
      "grad_norm": 0.5706221736923324,
      "learning_rate": 6.302460508134711e-05,
      "loss": 11.9956,
      "step": 23210
    },
    {
      "epoch": 1.263931714688285,
      "grad_norm": 0.5396563639855002,
      "learning_rate": 6.30164119194748e-05,
      "loss": 11.9397,
      "step": 23211
    },
    {
      "epoch": 1.263986168684868,
      "grad_norm": 0.6354073308334479,
      "learning_rate": 6.300821904519308e-05,
      "loss": 12.1244,
      "step": 23212
    },
    {
      "epoch": 1.264040622681451,
      "grad_norm": 0.5845475176699747,
      "learning_rate": 6.300002645856566e-05,
      "loss": 11.9064,
      "step": 23213
    },
    {
      "epoch": 1.264095076678034,
      "grad_norm": 0.512881774459477,
      "learning_rate": 6.299183415965622e-05,
      "loss": 11.9788,
      "step": 23214
    },
    {
      "epoch": 1.2641495306746169,
      "grad_norm": 0.5672581737919049,
      "learning_rate": 6.298364214852849e-05,
      "loss": 11.9306,
      "step": 23215
    },
    {
      "epoch": 1.2642039846711999,
      "grad_norm": 0.6185248005041978,
      "learning_rate": 6.297545042524617e-05,
      "loss": 11.9911,
      "step": 23216
    },
    {
      "epoch": 1.2642584386677829,
      "grad_norm": 0.579008583626923,
      "learning_rate": 6.296725898987292e-05,
      "loss": 11.9292,
      "step": 23217
    },
    {
      "epoch": 1.2643128926643659,
      "grad_norm": 0.5554296521119939,
      "learning_rate": 6.295906784247252e-05,
      "loss": 12.0232,
      "step": 23218
    },
    {
      "epoch": 1.264367346660949,
      "grad_norm": 0.6114492731324929,
      "learning_rate": 6.295087698310861e-05,
      "loss": 11.9058,
      "step": 23219
    },
    {
      "epoch": 1.264421800657532,
      "grad_norm": 0.5274101672933743,
      "learning_rate": 6.294268641184493e-05,
      "loss": 11.9656,
      "step": 23220
    },
    {
      "epoch": 1.264476254654115,
      "grad_norm": 0.5238667204154219,
      "learning_rate": 6.293449612874508e-05,
      "loss": 11.8131,
      "step": 23221
    },
    {
      "epoch": 1.264530708650698,
      "grad_norm": 0.5434777512564144,
      "learning_rate": 6.292630613387282e-05,
      "loss": 11.9128,
      "step": 23222
    },
    {
      "epoch": 1.264585162647281,
      "grad_norm": 0.5432222398151255,
      "learning_rate": 6.291811642729182e-05,
      "loss": 11.8331,
      "step": 23223
    },
    {
      "epoch": 1.264639616643864,
      "grad_norm": 0.5524449538973034,
      "learning_rate": 6.290992700906577e-05,
      "loss": 11.972,
      "step": 23224
    },
    {
      "epoch": 1.264694070640447,
      "grad_norm": 0.5445788443924797,
      "learning_rate": 6.290173787925835e-05,
      "loss": 11.9811,
      "step": 23225
    },
    {
      "epoch": 1.26474852463703,
      "grad_norm": 0.5215192081338217,
      "learning_rate": 6.289354903793324e-05,
      "loss": 11.9049,
      "step": 23226
    },
    {
      "epoch": 1.264802978633613,
      "grad_norm": 0.5496417411242858,
      "learning_rate": 6.28853604851541e-05,
      "loss": 11.9841,
      "step": 23227
    },
    {
      "epoch": 1.2648574326301962,
      "grad_norm": 0.5736958780394876,
      "learning_rate": 6.287717222098464e-05,
      "loss": 12.0088,
      "step": 23228
    },
    {
      "epoch": 1.2649118866267792,
      "grad_norm": 0.5553454283394794,
      "learning_rate": 6.286898424548848e-05,
      "loss": 11.9605,
      "step": 23229
    },
    {
      "epoch": 1.2649663406233622,
      "grad_norm": 0.49554090269372214,
      "learning_rate": 6.286079655872938e-05,
      "loss": 11.9359,
      "step": 23230
    },
    {
      "epoch": 1.2650207946199452,
      "grad_norm": 0.5995344148429932,
      "learning_rate": 6.285260916077093e-05,
      "loss": 11.8937,
      "step": 23231
    },
    {
      "epoch": 1.2650752486165282,
      "grad_norm": 0.5470788013975341,
      "learning_rate": 6.284442205167681e-05,
      "loss": 12.0121,
      "step": 23232
    },
    {
      "epoch": 1.2651297026131112,
      "grad_norm": 0.5481015507073892,
      "learning_rate": 6.283623523151068e-05,
      "loss": 11.9989,
      "step": 23233
    },
    {
      "epoch": 1.2651841566096942,
      "grad_norm": 0.6058346904021367,
      "learning_rate": 6.282804870033623e-05,
      "loss": 11.7442,
      "step": 23234
    },
    {
      "epoch": 1.2652386106062772,
      "grad_norm": 0.5694201142122483,
      "learning_rate": 6.281986245821712e-05,
      "loss": 12.051,
      "step": 23235
    },
    {
      "epoch": 1.2652930646028602,
      "grad_norm": 0.5331388845783178,
      "learning_rate": 6.281167650521699e-05,
      "loss": 11.9646,
      "step": 23236
    },
    {
      "epoch": 1.2653475185994432,
      "grad_norm": 0.5898842206056572,
      "learning_rate": 6.28034908413995e-05,
      "loss": 12.0569,
      "step": 23237
    },
    {
      "epoch": 1.2654019725960262,
      "grad_norm": 0.5033871143616334,
      "learning_rate": 6.27953054668283e-05,
      "loss": 11.8935,
      "step": 23238
    },
    {
      "epoch": 1.2654564265926092,
      "grad_norm": 0.5025364947010356,
      "learning_rate": 6.278712038156704e-05,
      "loss": 11.8969,
      "step": 23239
    },
    {
      "epoch": 1.2655108805891921,
      "grad_norm": 0.5339284807788707,
      "learning_rate": 6.27789355856794e-05,
      "loss": 11.727,
      "step": 23240
    },
    {
      "epoch": 1.2655653345857751,
      "grad_norm": 0.5831331266534203,
      "learning_rate": 6.277075107922899e-05,
      "loss": 11.9637,
      "step": 23241
    },
    {
      "epoch": 1.2656197885823581,
      "grad_norm": 0.5602101162288189,
      "learning_rate": 6.276256686227944e-05,
      "loss": 11.963,
      "step": 23242
    },
    {
      "epoch": 1.2656742425789413,
      "grad_norm": 0.6348367389306971,
      "learning_rate": 6.275438293489442e-05,
      "loss": 12.0841,
      "step": 23243
    },
    {
      "epoch": 1.2657286965755243,
      "grad_norm": 0.5662549810181848,
      "learning_rate": 6.274619929713755e-05,
      "loss": 11.9062,
      "step": 23244
    },
    {
      "epoch": 1.2657831505721073,
      "grad_norm": 0.5880688167587791,
      "learning_rate": 6.273801594907249e-05,
      "loss": 12.0878,
      "step": 23245
    },
    {
      "epoch": 1.2658376045686903,
      "grad_norm": 0.5658460549330208,
      "learning_rate": 6.272983289076288e-05,
      "loss": 12.0015,
      "step": 23246
    },
    {
      "epoch": 1.2658920585652733,
      "grad_norm": 0.5288370428761704,
      "learning_rate": 6.272165012227235e-05,
      "loss": 11.8035,
      "step": 23247
    },
    {
      "epoch": 1.2659465125618563,
      "grad_norm": 0.5091781328094663,
      "learning_rate": 6.271346764366451e-05,
      "loss": 12.032,
      "step": 23248
    },
    {
      "epoch": 1.2660009665584393,
      "grad_norm": 0.6019770426441612,
      "learning_rate": 6.270528545500298e-05,
      "loss": 11.9276,
      "step": 23249
    },
    {
      "epoch": 1.2660554205550223,
      "grad_norm": 0.5592074457787859,
      "learning_rate": 6.269710355635145e-05,
      "loss": 11.8501,
      "step": 23250
    },
    {
      "epoch": 1.2661098745516055,
      "grad_norm": 0.5581442322272826,
      "learning_rate": 6.268892194777348e-05,
      "loss": 11.9014,
      "step": 23251
    },
    {
      "epoch": 1.2661643285481885,
      "grad_norm": 0.5455366191873362,
      "learning_rate": 6.268074062933269e-05,
      "loss": 11.9663,
      "step": 23252
    },
    {
      "epoch": 1.2662187825447715,
      "grad_norm": 0.6696918915359905,
      "learning_rate": 6.267255960109273e-05,
      "loss": 11.9712,
      "step": 23253
    },
    {
      "epoch": 1.2662732365413545,
      "grad_norm": 0.5379251392148051,
      "learning_rate": 6.26643788631172e-05,
      "loss": 12.0345,
      "step": 23254
    },
    {
      "epoch": 1.2663276905379375,
      "grad_norm": 0.5931605972582454,
      "learning_rate": 6.26561984154697e-05,
      "loss": 12.0742,
      "step": 23255
    },
    {
      "epoch": 1.2663821445345205,
      "grad_norm": 0.5017334888256585,
      "learning_rate": 6.26480182582139e-05,
      "loss": 11.9426,
      "step": 23256
    },
    {
      "epoch": 1.2664365985311035,
      "grad_norm": 0.5068977268219558,
      "learning_rate": 6.263983839141335e-05,
      "loss": 11.9662,
      "step": 23257
    },
    {
      "epoch": 1.2664910525276865,
      "grad_norm": 0.5356184965527556,
      "learning_rate": 6.26316588151317e-05,
      "loss": 11.7663,
      "step": 23258
    },
    {
      "epoch": 1.2665455065242694,
      "grad_norm": 0.6047985115412942,
      "learning_rate": 6.262347952943253e-05,
      "loss": 11.9585,
      "step": 23259
    },
    {
      "epoch": 1.2665999605208524,
      "grad_norm": 0.5935450485177922,
      "learning_rate": 6.261530053437946e-05,
      "loss": 12.0218,
      "step": 23260
    },
    {
      "epoch": 1.2666544145174354,
      "grad_norm": 0.5792616729946286,
      "learning_rate": 6.26071218300361e-05,
      "loss": 11.942,
      "step": 23261
    },
    {
      "epoch": 1.2667088685140184,
      "grad_norm": 0.4911537276124451,
      "learning_rate": 6.2598943416466e-05,
      "loss": 11.8469,
      "step": 23262
    },
    {
      "epoch": 1.2667633225106014,
      "grad_norm": 0.5888120572545132,
      "learning_rate": 6.25907652937328e-05,
      "loss": 12.0768,
      "step": 23263
    },
    {
      "epoch": 1.2668177765071844,
      "grad_norm": 0.5842572999363864,
      "learning_rate": 6.258258746190008e-05,
      "loss": 11.8555,
      "step": 23264
    },
    {
      "epoch": 1.2668722305037674,
      "grad_norm": 0.5770417131764191,
      "learning_rate": 6.257440992103143e-05,
      "loss": 11.9088,
      "step": 23265
    },
    {
      "epoch": 1.2669266845003506,
      "grad_norm": 0.5123296797440976,
      "learning_rate": 6.256623267119043e-05,
      "loss": 11.9103,
      "step": 23266
    },
    {
      "epoch": 1.2669811384969336,
      "grad_norm": 0.5641928636700514,
      "learning_rate": 6.25580557124407e-05,
      "loss": 11.9741,
      "step": 23267
    },
    {
      "epoch": 1.2670355924935166,
      "grad_norm": 0.5655096911419298,
      "learning_rate": 6.25498790448458e-05,
      "loss": 12.0109,
      "step": 23268
    },
    {
      "epoch": 1.2670900464900996,
      "grad_norm": 0.5428229007565369,
      "learning_rate": 6.254170266846933e-05,
      "loss": 11.9479,
      "step": 23269
    },
    {
      "epoch": 1.2671445004866826,
      "grad_norm": 0.5706404677280447,
      "learning_rate": 6.253352658337487e-05,
      "loss": 12.0651,
      "step": 23270
    },
    {
      "epoch": 1.2671989544832656,
      "grad_norm": 0.5262681530829951,
      "learning_rate": 6.2525350789626e-05,
      "loss": 11.8401,
      "step": 23271
    },
    {
      "epoch": 1.2672534084798486,
      "grad_norm": 0.6176674300786359,
      "learning_rate": 6.251717528728627e-05,
      "loss": 11.9482,
      "step": 23272
    },
    {
      "epoch": 1.2673078624764316,
      "grad_norm": 0.5983821018764575,
      "learning_rate": 6.250900007641927e-05,
      "loss": 11.9959,
      "step": 23273
    },
    {
      "epoch": 1.2673623164730146,
      "grad_norm": 0.5273196446127435,
      "learning_rate": 6.250082515708857e-05,
      "loss": 11.8718,
      "step": 23274
    },
    {
      "epoch": 1.2674167704695978,
      "grad_norm": 0.5228535750345598,
      "learning_rate": 6.249265052935774e-05,
      "loss": 11.754,
      "step": 23275
    },
    {
      "epoch": 1.2674712244661808,
      "grad_norm": 0.5031082126493903,
      "learning_rate": 6.248447619329036e-05,
      "loss": 11.9304,
      "step": 23276
    },
    {
      "epoch": 1.2675256784627638,
      "grad_norm": 0.5775344306743875,
      "learning_rate": 6.247630214894995e-05,
      "loss": 11.9764,
      "step": 23277
    },
    {
      "epoch": 1.2675801324593468,
      "grad_norm": 0.5511462710066333,
      "learning_rate": 6.246812839640013e-05,
      "loss": 12.0393,
      "step": 23278
    },
    {
      "epoch": 1.2676345864559297,
      "grad_norm": 0.5546271233692556,
      "learning_rate": 6.245995493570445e-05,
      "loss": 11.8554,
      "step": 23279
    },
    {
      "epoch": 1.2676890404525127,
      "grad_norm": 0.5542476229734516,
      "learning_rate": 6.245178176692645e-05,
      "loss": 11.9874,
      "step": 23280
    },
    {
      "epoch": 1.2677434944490957,
      "grad_norm": 0.5601960937182989,
      "learning_rate": 6.244360889012973e-05,
      "loss": 11.974,
      "step": 23281
    },
    {
      "epoch": 1.2677979484456787,
      "grad_norm": 0.536459073035721,
      "learning_rate": 6.243543630537775e-05,
      "loss": 12.0649,
      "step": 23282
    },
    {
      "epoch": 1.2678524024422617,
      "grad_norm": 0.5326910175065769,
      "learning_rate": 6.242726401273414e-05,
      "loss": 11.9427,
      "step": 23283
    },
    {
      "epoch": 1.2679068564388447,
      "grad_norm": 0.6847770682281887,
      "learning_rate": 6.241909201226242e-05,
      "loss": 12.0452,
      "step": 23284
    },
    {
      "epoch": 1.2679613104354277,
      "grad_norm": 0.4851762226565081,
      "learning_rate": 6.241092030402614e-05,
      "loss": 11.9293,
      "step": 23285
    },
    {
      "epoch": 1.2680157644320107,
      "grad_norm": 0.5076819760948074,
      "learning_rate": 6.240274888808883e-05,
      "loss": 12.0396,
      "step": 23286
    },
    {
      "epoch": 1.2680702184285937,
      "grad_norm": 0.5597168326451334,
      "learning_rate": 6.239457776451409e-05,
      "loss": 12.0031,
      "step": 23287
    },
    {
      "epoch": 1.2681246724251767,
      "grad_norm": 0.6254650905558556,
      "learning_rate": 6.238640693336539e-05,
      "loss": 11.8565,
      "step": 23288
    },
    {
      "epoch": 1.26817912642176,
      "grad_norm": 0.5411474727450071,
      "learning_rate": 6.237823639470628e-05,
      "loss": 11.9987,
      "step": 23289
    },
    {
      "epoch": 1.2682335804183429,
      "grad_norm": 0.5686613732412673,
      "learning_rate": 6.237006614860035e-05,
      "loss": 11.9573,
      "step": 23290
    },
    {
      "epoch": 1.2682880344149259,
      "grad_norm": 0.5273454910767795,
      "learning_rate": 6.236189619511113e-05,
      "loss": 12.0214,
      "step": 23291
    },
    {
      "epoch": 1.2683424884115089,
      "grad_norm": 0.5251569727347625,
      "learning_rate": 6.235372653430207e-05,
      "loss": 11.8177,
      "step": 23292
    },
    {
      "epoch": 1.2683969424080919,
      "grad_norm": 0.5271534590038707,
      "learning_rate": 6.234555716623672e-05,
      "loss": 11.8267,
      "step": 23293
    },
    {
      "epoch": 1.2684513964046749,
      "grad_norm": 0.6089298643908899,
      "learning_rate": 6.233738809097866e-05,
      "loss": 12.0401,
      "step": 23294
    },
    {
      "epoch": 1.2685058504012579,
      "grad_norm": 0.5535058687639198,
      "learning_rate": 6.23292193085914e-05,
      "loss": 12.0141,
      "step": 23295
    },
    {
      "epoch": 1.2685603043978408,
      "grad_norm": 0.57154260724941,
      "learning_rate": 6.232105081913841e-05,
      "loss": 11.9038,
      "step": 23296
    },
    {
      "epoch": 1.2686147583944238,
      "grad_norm": 0.5788602074094982,
      "learning_rate": 6.231288262268328e-05,
      "loss": 12.0593,
      "step": 23297
    },
    {
      "epoch": 1.268669212391007,
      "grad_norm": 0.5300962947667739,
      "learning_rate": 6.23047147192895e-05,
      "loss": 11.9953,
      "step": 23298
    },
    {
      "epoch": 1.26872366638759,
      "grad_norm": 0.6186935749669186,
      "learning_rate": 6.229654710902055e-05,
      "loss": 11.9587,
      "step": 23299
    },
    {
      "epoch": 1.268778120384173,
      "grad_norm": 0.5494811757785866,
      "learning_rate": 6.228837979193997e-05,
      "loss": 12.0647,
      "step": 23300
    },
    {
      "epoch": 1.268832574380756,
      "grad_norm": 0.5447030181196109,
      "learning_rate": 6.228021276811134e-05,
      "loss": 12.0203,
      "step": 23301
    },
    {
      "epoch": 1.268887028377339,
      "grad_norm": 0.5637450092619816,
      "learning_rate": 6.227204603759805e-05,
      "loss": 11.9965,
      "step": 23302
    },
    {
      "epoch": 1.268941482373922,
      "grad_norm": 0.5638522258794966,
      "learning_rate": 6.226387960046367e-05,
      "loss": 11.7628,
      "step": 23303
    },
    {
      "epoch": 1.268995936370505,
      "grad_norm": 0.5467926557255012,
      "learning_rate": 6.225571345677165e-05,
      "loss": 11.9061,
      "step": 23304
    },
    {
      "epoch": 1.269050390367088,
      "grad_norm": 0.5593070649001599,
      "learning_rate": 6.224754760658558e-05,
      "loss": 12.0445,
      "step": 23305
    },
    {
      "epoch": 1.269104844363671,
      "grad_norm": 0.5343540127281311,
      "learning_rate": 6.223938204996889e-05,
      "loss": 11.9997,
      "step": 23306
    },
    {
      "epoch": 1.269159298360254,
      "grad_norm": 0.5801851478240748,
      "learning_rate": 6.223121678698509e-05,
      "loss": 12.0018,
      "step": 23307
    },
    {
      "epoch": 1.269213752356837,
      "grad_norm": 0.5533776677930815,
      "learning_rate": 6.222305181769769e-05,
      "loss": 11.923,
      "step": 23308
    },
    {
      "epoch": 1.26926820635342,
      "grad_norm": 0.5789373933804217,
      "learning_rate": 6.221488714217019e-05,
      "loss": 12.0229,
      "step": 23309
    },
    {
      "epoch": 1.269322660350003,
      "grad_norm": 0.5536744500147756,
      "learning_rate": 6.220672276046604e-05,
      "loss": 11.883,
      "step": 23310
    },
    {
      "epoch": 1.269377114346586,
      "grad_norm": 0.5783446621668239,
      "learning_rate": 6.219855867264878e-05,
      "loss": 11.9313,
      "step": 23311
    },
    {
      "epoch": 1.269431568343169,
      "grad_norm": 0.5404254904073282,
      "learning_rate": 6.219039487878187e-05,
      "loss": 12.0112,
      "step": 23312
    },
    {
      "epoch": 1.2694860223397522,
      "grad_norm": 0.5262018222211831,
      "learning_rate": 6.218223137892876e-05,
      "loss": 11.8799,
      "step": 23313
    },
    {
      "epoch": 1.2695404763363352,
      "grad_norm": 0.540015923190593,
      "learning_rate": 6.217406817315297e-05,
      "loss": 11.9449,
      "step": 23314
    },
    {
      "epoch": 1.2695949303329181,
      "grad_norm": 0.5976045275009173,
      "learning_rate": 6.216590526151795e-05,
      "loss": 12.1262,
      "step": 23315
    },
    {
      "epoch": 1.2696493843295011,
      "grad_norm": 0.5368482700177268,
      "learning_rate": 6.215774264408723e-05,
      "loss": 11.9444,
      "step": 23316
    },
    {
      "epoch": 1.2697038383260841,
      "grad_norm": 0.5558777228267541,
      "learning_rate": 6.214958032092423e-05,
      "loss": 12.0168,
      "step": 23317
    },
    {
      "epoch": 1.2697582923226671,
      "grad_norm": 0.5077506313370604,
      "learning_rate": 6.214141829209245e-05,
      "loss": 11.9611,
      "step": 23318
    },
    {
      "epoch": 1.2698127463192501,
      "grad_norm": 0.5487927148954824,
      "learning_rate": 6.213325655765534e-05,
      "loss": 11.9458,
      "step": 23319
    },
    {
      "epoch": 1.2698672003158331,
      "grad_norm": 0.595899878814489,
      "learning_rate": 6.21250951176764e-05,
      "loss": 11.9533,
      "step": 23320
    },
    {
      "epoch": 1.2699216543124163,
      "grad_norm": 0.5506273459385893,
      "learning_rate": 6.211693397221908e-05,
      "loss": 11.7498,
      "step": 23321
    },
    {
      "epoch": 1.2699761083089993,
      "grad_norm": 0.538278177967009,
      "learning_rate": 6.210877312134679e-05,
      "loss": 11.9185,
      "step": 23322
    },
    {
      "epoch": 1.2700305623055823,
      "grad_norm": 0.5227067430337492,
      "learning_rate": 6.210061256512306e-05,
      "loss": 11.9251,
      "step": 23323
    },
    {
      "epoch": 1.2700850163021653,
      "grad_norm": 0.5872058249802069,
      "learning_rate": 6.209245230361131e-05,
      "loss": 11.9939,
      "step": 23324
    },
    {
      "epoch": 1.2701394702987483,
      "grad_norm": 0.524340618375271,
      "learning_rate": 6.208429233687503e-05,
      "loss": 12.0238,
      "step": 23325
    },
    {
      "epoch": 1.2701939242953313,
      "grad_norm": 0.5671513504042198,
      "learning_rate": 6.20761326649776e-05,
      "loss": 11.8688,
      "step": 23326
    },
    {
      "epoch": 1.2702483782919143,
      "grad_norm": 0.5589026914595644,
      "learning_rate": 6.206797328798257e-05,
      "loss": 12.0015,
      "step": 23327
    },
    {
      "epoch": 1.2703028322884973,
      "grad_norm": 0.5103027645540704,
      "learning_rate": 6.205981420595332e-05,
      "loss": 11.9565,
      "step": 23328
    },
    {
      "epoch": 1.2703572862850803,
      "grad_norm": 0.5709819143252606,
      "learning_rate": 6.205165541895334e-05,
      "loss": 11.9735,
      "step": 23329
    },
    {
      "epoch": 1.2704117402816633,
      "grad_norm": 0.6095574741860947,
      "learning_rate": 6.204349692704604e-05,
      "loss": 12.035,
      "step": 23330
    },
    {
      "epoch": 1.2704661942782463,
      "grad_norm": 0.5646974626640744,
      "learning_rate": 6.20353387302949e-05,
      "loss": 11.8901,
      "step": 23331
    },
    {
      "epoch": 1.2705206482748292,
      "grad_norm": 0.5210950415666447,
      "learning_rate": 6.20271808287633e-05,
      "loss": 12.0456,
      "step": 23332
    },
    {
      "epoch": 1.2705751022714122,
      "grad_norm": 0.5811755441916826,
      "learning_rate": 6.201902322251471e-05,
      "loss": 11.9576,
      "step": 23333
    },
    {
      "epoch": 1.2706295562679952,
      "grad_norm": 0.5897526476423185,
      "learning_rate": 6.201086591161255e-05,
      "loss": 11.8529,
      "step": 23334
    },
    {
      "epoch": 1.2706840102645782,
      "grad_norm": 0.6421405117499367,
      "learning_rate": 6.200270889612029e-05,
      "loss": 12.0423,
      "step": 23335
    },
    {
      "epoch": 1.2707384642611614,
      "grad_norm": 0.5818935058785432,
      "learning_rate": 6.199455217610135e-05,
      "loss": 11.9691,
      "step": 23336
    },
    {
      "epoch": 1.2707929182577444,
      "grad_norm": 0.6004240477887385,
      "learning_rate": 6.198639575161914e-05,
      "loss": 11.946,
      "step": 23337
    },
    {
      "epoch": 1.2708473722543274,
      "grad_norm": 0.539171698349238,
      "learning_rate": 6.197823962273705e-05,
      "loss": 12.0062,
      "step": 23338
    },
    {
      "epoch": 1.2709018262509104,
      "grad_norm": 0.6258640458409044,
      "learning_rate": 6.197008378951858e-05,
      "loss": 11.7681,
      "step": 23339
    },
    {
      "epoch": 1.2709562802474934,
      "grad_norm": 0.5848841771074273,
      "learning_rate": 6.196192825202711e-05,
      "loss": 12.032,
      "step": 23340
    },
    {
      "epoch": 1.2710107342440764,
      "grad_norm": 0.5291610932441859,
      "learning_rate": 6.195377301032611e-05,
      "loss": 11.7731,
      "step": 23341
    },
    {
      "epoch": 1.2710651882406594,
      "grad_norm": 0.4886445273656309,
      "learning_rate": 6.19456180644789e-05,
      "loss": 11.9044,
      "step": 23342
    },
    {
      "epoch": 1.2711196422372424,
      "grad_norm": 0.5475517712648182,
      "learning_rate": 6.193746341454894e-05,
      "loss": 12.0123,
      "step": 23343
    },
    {
      "epoch": 1.2711740962338254,
      "grad_norm": 0.5115837751985072,
      "learning_rate": 6.192930906059966e-05,
      "loss": 11.8949,
      "step": 23344
    },
    {
      "epoch": 1.2712285502304086,
      "grad_norm": 0.5539929387810759,
      "learning_rate": 6.192115500269447e-05,
      "loss": 11.9995,
      "step": 23345
    },
    {
      "epoch": 1.2712830042269916,
      "grad_norm": 0.4909185350024668,
      "learning_rate": 6.191300124089675e-05,
      "loss": 11.916,
      "step": 23346
    },
    {
      "epoch": 1.2713374582235746,
      "grad_norm": 0.49260331961665,
      "learning_rate": 6.190484777526993e-05,
      "loss": 11.8449,
      "step": 23347
    },
    {
      "epoch": 1.2713919122201576,
      "grad_norm": 0.631466451668857,
      "learning_rate": 6.189669460587739e-05,
      "loss": 11.8119,
      "step": 23348
    },
    {
      "epoch": 1.2714463662167406,
      "grad_norm": 0.6089354795289456,
      "learning_rate": 6.188854173278254e-05,
      "loss": 12.0137,
      "step": 23349
    },
    {
      "epoch": 1.2715008202133236,
      "grad_norm": 0.6371413460409984,
      "learning_rate": 6.188038915604877e-05,
      "loss": 11.9834,
      "step": 23350
    },
    {
      "epoch": 1.2715552742099065,
      "grad_norm": 0.5769881919737412,
      "learning_rate": 6.187223687573956e-05,
      "loss": 11.9342,
      "step": 23351
    },
    {
      "epoch": 1.2716097282064895,
      "grad_norm": 0.6152241490133966,
      "learning_rate": 6.186408489191818e-05,
      "loss": 11.9962,
      "step": 23352
    },
    {
      "epoch": 1.2716641822030725,
      "grad_norm": 0.6116610121964597,
      "learning_rate": 6.185593320464805e-05,
      "loss": 11.973,
      "step": 23353
    },
    {
      "epoch": 1.2717186361996555,
      "grad_norm": 0.5256129458254312,
      "learning_rate": 6.184778181399258e-05,
      "loss": 11.8692,
      "step": 23354
    },
    {
      "epoch": 1.2717730901962385,
      "grad_norm": 0.4920780804838193,
      "learning_rate": 6.183963072001517e-05,
      "loss": 11.9996,
      "step": 23355
    },
    {
      "epoch": 1.2718275441928215,
      "grad_norm": 0.5127263926711845,
      "learning_rate": 6.18314799227792e-05,
      "loss": 11.9083,
      "step": 23356
    },
    {
      "epoch": 1.2718819981894045,
      "grad_norm": 0.6217008392542402,
      "learning_rate": 6.182332942234804e-05,
      "loss": 12.036,
      "step": 23357
    },
    {
      "epoch": 1.2719364521859875,
      "grad_norm": 0.5130976735521793,
      "learning_rate": 6.181517921878508e-05,
      "loss": 11.8585,
      "step": 23358
    },
    {
      "epoch": 1.2719909061825707,
      "grad_norm": 0.5202591226152474,
      "learning_rate": 6.180702931215367e-05,
      "loss": 11.9987,
      "step": 23359
    },
    {
      "epoch": 1.2720453601791537,
      "grad_norm": 0.6111920306799264,
      "learning_rate": 6.17988797025172e-05,
      "loss": 12.0882,
      "step": 23360
    },
    {
      "epoch": 1.2720998141757367,
      "grad_norm": 0.6293271982113089,
      "learning_rate": 6.17907303899391e-05,
      "loss": 11.9575,
      "step": 23361
    },
    {
      "epoch": 1.2721542681723197,
      "grad_norm": 0.5099706333577905,
      "learning_rate": 6.178258137448265e-05,
      "loss": 12.0043,
      "step": 23362
    },
    {
      "epoch": 1.2722087221689027,
      "grad_norm": 0.6005071181171349,
      "learning_rate": 6.177443265621127e-05,
      "loss": 11.9809,
      "step": 23363
    },
    {
      "epoch": 1.2722631761654857,
      "grad_norm": 0.48955023102144846,
      "learning_rate": 6.176628423518827e-05,
      "loss": 11.8855,
      "step": 23364
    },
    {
      "epoch": 1.2723176301620687,
      "grad_norm": 0.5513402720154954,
      "learning_rate": 6.17581361114771e-05,
      "loss": 11.9205,
      "step": 23365
    },
    {
      "epoch": 1.2723720841586517,
      "grad_norm": 0.5271349303137202,
      "learning_rate": 6.174998828514106e-05,
      "loss": 11.9024,
      "step": 23366
    },
    {
      "epoch": 1.2724265381552347,
      "grad_norm": 0.5440387096455964,
      "learning_rate": 6.174184075624352e-05,
      "loss": 11.8205,
      "step": 23367
    },
    {
      "epoch": 1.2724809921518179,
      "grad_norm": 0.5316353551943467,
      "learning_rate": 6.173369352484786e-05,
      "loss": 11.9881,
      "step": 23368
    },
    {
      "epoch": 1.2725354461484009,
      "grad_norm": 0.5195114242919787,
      "learning_rate": 6.17255465910174e-05,
      "loss": 11.8855,
      "step": 23369
    },
    {
      "epoch": 1.2725899001449839,
      "grad_norm": 0.6067601046513386,
      "learning_rate": 6.171739995481551e-05,
      "loss": 12.0578,
      "step": 23370
    },
    {
      "epoch": 1.2726443541415668,
      "grad_norm": 0.5922350874741872,
      "learning_rate": 6.170925361630557e-05,
      "loss": 11.955,
      "step": 23371
    },
    {
      "epoch": 1.2726988081381498,
      "grad_norm": 0.523548089866187,
      "learning_rate": 6.170110757555088e-05,
      "loss": 11.9278,
      "step": 23372
    },
    {
      "epoch": 1.2727532621347328,
      "grad_norm": 0.5592176843160478,
      "learning_rate": 6.169296183261477e-05,
      "loss": 11.8571,
      "step": 23373
    },
    {
      "epoch": 1.2728077161313158,
      "grad_norm": 0.6185745649402246,
      "learning_rate": 6.168481638756064e-05,
      "loss": 12.0033,
      "step": 23374
    },
    {
      "epoch": 1.2728621701278988,
      "grad_norm": 0.5580615287086415,
      "learning_rate": 6.167667124045178e-05,
      "loss": 11.9415,
      "step": 23375
    },
    {
      "epoch": 1.2729166241244818,
      "grad_norm": 0.5315801320232145,
      "learning_rate": 6.166852639135156e-05,
      "loss": 11.8717,
      "step": 23376
    },
    {
      "epoch": 1.2729710781210648,
      "grad_norm": 0.5221864854326923,
      "learning_rate": 6.16603818403233e-05,
      "loss": 11.9166,
      "step": 23377
    },
    {
      "epoch": 1.2730255321176478,
      "grad_norm": 0.49489873083951413,
      "learning_rate": 6.165223758743037e-05,
      "loss": 11.9756,
      "step": 23378
    },
    {
      "epoch": 1.2730799861142308,
      "grad_norm": 0.5284182535982023,
      "learning_rate": 6.164409363273604e-05,
      "loss": 11.8718,
      "step": 23379
    },
    {
      "epoch": 1.2731344401108138,
      "grad_norm": 0.5700556655297687,
      "learning_rate": 6.163594997630369e-05,
      "loss": 11.7334,
      "step": 23380
    },
    {
      "epoch": 1.2731888941073968,
      "grad_norm": 0.5452737708328596,
      "learning_rate": 6.162780661819665e-05,
      "loss": 12.0551,
      "step": 23381
    },
    {
      "epoch": 1.27324334810398,
      "grad_norm": 0.5649806032124066,
      "learning_rate": 6.161966355847819e-05,
      "loss": 12.0111,
      "step": 23382
    },
    {
      "epoch": 1.273297802100563,
      "grad_norm": 0.5035865394764981,
      "learning_rate": 6.161152079721166e-05,
      "loss": 12.0424,
      "step": 23383
    },
    {
      "epoch": 1.273352256097146,
      "grad_norm": 0.5401260476633886,
      "learning_rate": 6.16033783344604e-05,
      "loss": 11.9739,
      "step": 23384
    },
    {
      "epoch": 1.273406710093729,
      "grad_norm": 0.5715111184080219,
      "learning_rate": 6.159523617028768e-05,
      "loss": 12.0213,
      "step": 23385
    },
    {
      "epoch": 1.273461164090312,
      "grad_norm": 0.656240407456829,
      "learning_rate": 6.158709430475687e-05,
      "loss": 11.9437,
      "step": 23386
    },
    {
      "epoch": 1.273515618086895,
      "grad_norm": 0.5331603550570894,
      "learning_rate": 6.157895273793123e-05,
      "loss": 11.8895,
      "step": 23387
    },
    {
      "epoch": 1.273570072083478,
      "grad_norm": 0.5571948358317733,
      "learning_rate": 6.15708114698741e-05,
      "loss": 11.9225,
      "step": 23388
    },
    {
      "epoch": 1.273624526080061,
      "grad_norm": 0.576153893707623,
      "learning_rate": 6.156267050064881e-05,
      "loss": 11.9145,
      "step": 23389
    },
    {
      "epoch": 1.273678980076644,
      "grad_norm": 0.580075660793452,
      "learning_rate": 6.155452983031862e-05,
      "loss": 11.9521,
      "step": 23390
    },
    {
      "epoch": 1.2737334340732271,
      "grad_norm": 0.5433488848620138,
      "learning_rate": 6.154638945894689e-05,
      "loss": 11.9118,
      "step": 23391
    },
    {
      "epoch": 1.2737878880698101,
      "grad_norm": 0.5541799106192558,
      "learning_rate": 6.153824938659684e-05,
      "loss": 12.0682,
      "step": 23392
    },
    {
      "epoch": 1.2738423420663931,
      "grad_norm": 0.8937971582716784,
      "learning_rate": 6.153010961333184e-05,
      "loss": 12.0163,
      "step": 23393
    },
    {
      "epoch": 1.2738967960629761,
      "grad_norm": 0.5796018149701783,
      "learning_rate": 6.152197013921515e-05,
      "loss": 11.8705,
      "step": 23394
    },
    {
      "epoch": 1.2739512500595591,
      "grad_norm": 0.5156819839981832,
      "learning_rate": 6.151383096431008e-05,
      "loss": 11.792,
      "step": 23395
    },
    {
      "epoch": 1.274005704056142,
      "grad_norm": 0.5247084374996511,
      "learning_rate": 6.150569208867989e-05,
      "loss": 11.9977,
      "step": 23396
    },
    {
      "epoch": 1.274060158052725,
      "grad_norm": 0.5045128890404776,
      "learning_rate": 6.149755351238791e-05,
      "loss": 11.734,
      "step": 23397
    },
    {
      "epoch": 1.274114612049308,
      "grad_norm": 0.5428727538934164,
      "learning_rate": 6.148941523549739e-05,
      "loss": 11.7835,
      "step": 23398
    },
    {
      "epoch": 1.274169066045891,
      "grad_norm": 0.5639829723044368,
      "learning_rate": 6.148127725807166e-05,
      "loss": 11.9327,
      "step": 23399
    },
    {
      "epoch": 1.274223520042474,
      "grad_norm": 0.5416806279693785,
      "learning_rate": 6.147313958017398e-05,
      "loss": 11.9288,
      "step": 23400
    },
    {
      "epoch": 1.274277974039057,
      "grad_norm": 0.5457388335316171,
      "learning_rate": 6.146500220186766e-05,
      "loss": 12.0199,
      "step": 23401
    },
    {
      "epoch": 1.27433242803564,
      "grad_norm": 0.5179846459378248,
      "learning_rate": 6.145686512321588e-05,
      "loss": 11.9786,
      "step": 23402
    },
    {
      "epoch": 1.274386882032223,
      "grad_norm": 0.5599616162200108,
      "learning_rate": 6.1448728344282e-05,
      "loss": 11.7671,
      "step": 23403
    },
    {
      "epoch": 1.274441336028806,
      "grad_norm": 0.5396860211024469,
      "learning_rate": 6.144059186512928e-05,
      "loss": 11.9588,
      "step": 23404
    },
    {
      "epoch": 1.274495790025389,
      "grad_norm": 0.5213496629497232,
      "learning_rate": 6.143245568582099e-05,
      "loss": 11.9472,
      "step": 23405
    },
    {
      "epoch": 1.2745502440219723,
      "grad_norm": 0.5279588003393272,
      "learning_rate": 6.142431980642039e-05,
      "loss": 11.8882,
      "step": 23406
    },
    {
      "epoch": 1.2746046980185552,
      "grad_norm": 0.555959952393999,
      "learning_rate": 6.141618422699074e-05,
      "loss": 11.8884,
      "step": 23407
    },
    {
      "epoch": 1.2746591520151382,
      "grad_norm": 0.566584123832805,
      "learning_rate": 6.14080489475953e-05,
      "loss": 11.8963,
      "step": 23408
    },
    {
      "epoch": 1.2747136060117212,
      "grad_norm": 0.5402824653181659,
      "learning_rate": 6.139991396829735e-05,
      "loss": 12.0018,
      "step": 23409
    },
    {
      "epoch": 1.2747680600083042,
      "grad_norm": 0.557910135170613,
      "learning_rate": 6.139177928916016e-05,
      "loss": 11.9609,
      "step": 23410
    },
    {
      "epoch": 1.2748225140048872,
      "grad_norm": 0.5543819070305888,
      "learning_rate": 6.138364491024696e-05,
      "loss": 11.8824,
      "step": 23411
    },
    {
      "epoch": 1.2748769680014702,
      "grad_norm": 0.5733588974557553,
      "learning_rate": 6.137551083162105e-05,
      "loss": 12.017,
      "step": 23412
    },
    {
      "epoch": 1.2749314219980532,
      "grad_norm": 0.48920246956431995,
      "learning_rate": 6.136737705334559e-05,
      "loss": 11.9517,
      "step": 23413
    },
    {
      "epoch": 1.2749858759946364,
      "grad_norm": 0.552769775479179,
      "learning_rate": 6.13592435754839e-05,
      "loss": 11.9836,
      "step": 23414
    },
    {
      "epoch": 1.2750403299912194,
      "grad_norm": 0.5824338554355197,
      "learning_rate": 6.135111039809922e-05,
      "loss": 11.9281,
      "step": 23415
    },
    {
      "epoch": 1.2750947839878024,
      "grad_norm": 0.5714441455621442,
      "learning_rate": 6.134297752125477e-05,
      "loss": 11.9724,
      "step": 23416
    },
    {
      "epoch": 1.2751492379843854,
      "grad_norm": 0.4878833198075495,
      "learning_rate": 6.133484494501382e-05,
      "loss": 11.9296,
      "step": 23417
    },
    {
      "epoch": 1.2752036919809684,
      "grad_norm": 0.5395389807250964,
      "learning_rate": 6.132671266943962e-05,
      "loss": 11.9294,
      "step": 23418
    },
    {
      "epoch": 1.2752581459775514,
      "grad_norm": 0.6499568272746115,
      "learning_rate": 6.131858069459537e-05,
      "loss": 12.0162,
      "step": 23419
    },
    {
      "epoch": 1.2753125999741344,
      "grad_norm": 0.5111275170025651,
      "learning_rate": 6.13104490205443e-05,
      "loss": 11.9132,
      "step": 23420
    },
    {
      "epoch": 1.2753670539707174,
      "grad_norm": 0.5512976507197156,
      "learning_rate": 6.130231764734968e-05,
      "loss": 11.8927,
      "step": 23421
    },
    {
      "epoch": 1.2754215079673004,
      "grad_norm": 0.5757738157915058,
      "learning_rate": 6.129418657507478e-05,
      "loss": 11.9277,
      "step": 23422
    },
    {
      "epoch": 1.2754759619638834,
      "grad_norm": 0.5713838120879103,
      "learning_rate": 6.128605580378273e-05,
      "loss": 11.8534,
      "step": 23423
    },
    {
      "epoch": 1.2755304159604663,
      "grad_norm": 0.5810350740042042,
      "learning_rate": 6.127792533353681e-05,
      "loss": 11.9139,
      "step": 23424
    },
    {
      "epoch": 1.2755848699570493,
      "grad_norm": 0.5520984846109018,
      "learning_rate": 6.126979516440021e-05,
      "loss": 11.8544,
      "step": 23425
    },
    {
      "epoch": 1.2756393239536323,
      "grad_norm": 0.6195542466138565,
      "learning_rate": 6.12616652964362e-05,
      "loss": 12.0248,
      "step": 23426
    },
    {
      "epoch": 1.2756937779502153,
      "grad_norm": 0.6065463181157293,
      "learning_rate": 6.125353572970798e-05,
      "loss": 12.0263,
      "step": 23427
    },
    {
      "epoch": 1.2757482319467983,
      "grad_norm": 0.6022890840270189,
      "learning_rate": 6.124540646427877e-05,
      "loss": 11.9128,
      "step": 23428
    },
    {
      "epoch": 1.2758026859433815,
      "grad_norm": 0.5505979578393485,
      "learning_rate": 6.123727750021177e-05,
      "loss": 11.9789,
      "step": 23429
    },
    {
      "epoch": 1.2758571399399645,
      "grad_norm": 0.5313454778922818,
      "learning_rate": 6.122914883757019e-05,
      "loss": 11.8778,
      "step": 23430
    },
    {
      "epoch": 1.2759115939365475,
      "grad_norm": 0.5165311898873544,
      "learning_rate": 6.122102047641725e-05,
      "loss": 11.7385,
      "step": 23431
    },
    {
      "epoch": 1.2759660479331305,
      "grad_norm": 0.5619170786324981,
      "learning_rate": 6.121289241681621e-05,
      "loss": 12.0393,
      "step": 23432
    },
    {
      "epoch": 1.2760205019297135,
      "grad_norm": 0.5371501925842642,
      "learning_rate": 6.120476465883018e-05,
      "loss": 11.9275,
      "step": 23433
    },
    {
      "epoch": 1.2760749559262965,
      "grad_norm": 0.5539806629730009,
      "learning_rate": 6.11966372025224e-05,
      "loss": 11.9057,
      "step": 23434
    },
    {
      "epoch": 1.2761294099228795,
      "grad_norm": 0.5254273132834162,
      "learning_rate": 6.11885100479561e-05,
      "loss": 11.9598,
      "step": 23435
    },
    {
      "epoch": 1.2761838639194625,
      "grad_norm": 0.5256322934788941,
      "learning_rate": 6.118038319519441e-05,
      "loss": 11.8457,
      "step": 23436
    },
    {
      "epoch": 1.2762383179160455,
      "grad_norm": 0.5742182452294878,
      "learning_rate": 6.11722566443006e-05,
      "loss": 12.0103,
      "step": 23437
    },
    {
      "epoch": 1.2762927719126287,
      "grad_norm": 0.5619782065172815,
      "learning_rate": 6.116413039533784e-05,
      "loss": 11.9429,
      "step": 23438
    },
    {
      "epoch": 1.2763472259092117,
      "grad_norm": 0.5912072790665202,
      "learning_rate": 6.11560044483693e-05,
      "loss": 11.9045,
      "step": 23439
    },
    {
      "epoch": 1.2764016799057947,
      "grad_norm": 0.640642372825217,
      "learning_rate": 6.114787880345818e-05,
      "loss": 11.9004,
      "step": 23440
    },
    {
      "epoch": 1.2764561339023777,
      "grad_norm": 0.529558915606228,
      "learning_rate": 6.113975346066769e-05,
      "loss": 11.945,
      "step": 23441
    },
    {
      "epoch": 1.2765105878989607,
      "grad_norm": 0.5491331705510023,
      "learning_rate": 6.1131628420061e-05,
      "loss": 11.865,
      "step": 23442
    },
    {
      "epoch": 1.2765650418955437,
      "grad_norm": 0.5521629736406414,
      "learning_rate": 6.112350368170128e-05,
      "loss": 11.9299,
      "step": 23443
    },
    {
      "epoch": 1.2766194958921266,
      "grad_norm": 0.5413581163713285,
      "learning_rate": 6.11153792456517e-05,
      "loss": 11.999,
      "step": 23444
    },
    {
      "epoch": 1.2766739498887096,
      "grad_norm": 0.5127209632085405,
      "learning_rate": 6.110725511197546e-05,
      "loss": 11.9061,
      "step": 23445
    },
    {
      "epoch": 1.2767284038852926,
      "grad_norm": 0.5368385954467395,
      "learning_rate": 6.109913128073572e-05,
      "loss": 11.9514,
      "step": 23446
    },
    {
      "epoch": 1.2767828578818756,
      "grad_norm": 0.5673990511629734,
      "learning_rate": 6.109100775199565e-05,
      "loss": 12.0544,
      "step": 23447
    },
    {
      "epoch": 1.2768373118784586,
      "grad_norm": 0.5432429153901958,
      "learning_rate": 6.108288452581844e-05,
      "loss": 11.9729,
      "step": 23448
    },
    {
      "epoch": 1.2768917658750416,
      "grad_norm": 0.531720712228885,
      "learning_rate": 6.107476160226725e-05,
      "loss": 12.0446,
      "step": 23449
    },
    {
      "epoch": 1.2769462198716246,
      "grad_norm": 0.5586792284313403,
      "learning_rate": 6.106663898140524e-05,
      "loss": 12.2062,
      "step": 23450
    },
    {
      "epoch": 1.2770006738682076,
      "grad_norm": 0.5147979941718789,
      "learning_rate": 6.105851666329557e-05,
      "loss": 11.9341,
      "step": 23451
    },
    {
      "epoch": 1.2770551278647908,
      "grad_norm": 0.5359541791723327,
      "learning_rate": 6.105039464800143e-05,
      "loss": 11.8957,
      "step": 23452
    },
    {
      "epoch": 1.2771095818613738,
      "grad_norm": 0.5456329212624745,
      "learning_rate": 6.104227293558593e-05,
      "loss": 11.9419,
      "step": 23453
    },
    {
      "epoch": 1.2771640358579568,
      "grad_norm": 0.692260373025588,
      "learning_rate": 6.103415152611225e-05,
      "loss": 12.0249,
      "step": 23454
    },
    {
      "epoch": 1.2772184898545398,
      "grad_norm": 0.5121736932820726,
      "learning_rate": 6.1026030419643544e-05,
      "loss": 11.8515,
      "step": 23455
    },
    {
      "epoch": 1.2772729438511228,
      "grad_norm": 0.534931595755875,
      "learning_rate": 6.1017909616242966e-05,
      "loss": 11.9126,
      "step": 23456
    },
    {
      "epoch": 1.2773273978477058,
      "grad_norm": 0.6172589535524825,
      "learning_rate": 6.100978911597366e-05,
      "loss": 11.936,
      "step": 23457
    },
    {
      "epoch": 1.2773818518442888,
      "grad_norm": 0.5175136620150501,
      "learning_rate": 6.1001668918898734e-05,
      "loss": 11.8191,
      "step": 23458
    },
    {
      "epoch": 1.2774363058408718,
      "grad_norm": 0.5067862787806349,
      "learning_rate": 6.099354902508141e-05,
      "loss": 11.9255,
      "step": 23459
    },
    {
      "epoch": 1.2774907598374547,
      "grad_norm": 0.5599109908899864,
      "learning_rate": 6.098542943458478e-05,
      "loss": 12.0136,
      "step": 23460
    },
    {
      "epoch": 1.277545213834038,
      "grad_norm": 0.5299940958146621,
      "learning_rate": 6.0977310147472e-05,
      "loss": 11.8644,
      "step": 23461
    },
    {
      "epoch": 1.277599667830621,
      "grad_norm": 0.5700600588235133,
      "learning_rate": 6.096919116380622e-05,
      "loss": 11.9795,
      "step": 23462
    },
    {
      "epoch": 1.277654121827204,
      "grad_norm": 0.5638360644329294,
      "learning_rate": 6.0961072483650526e-05,
      "loss": 11.9326,
      "step": 23463
    },
    {
      "epoch": 1.277708575823787,
      "grad_norm": 0.6578693654611429,
      "learning_rate": 6.095295410706809e-05,
      "loss": 12.0909,
      "step": 23464
    },
    {
      "epoch": 1.27776302982037,
      "grad_norm": 0.533684078825207,
      "learning_rate": 6.094483603412203e-05,
      "loss": 11.9293,
      "step": 23465
    },
    {
      "epoch": 1.277817483816953,
      "grad_norm": 0.47515698786075217,
      "learning_rate": 6.093671826487547e-05,
      "loss": 11.9598,
      "step": 23466
    },
    {
      "epoch": 1.277871937813536,
      "grad_norm": 0.5637803768239379,
      "learning_rate": 6.092860079939154e-05,
      "loss": 12.0752,
      "step": 23467
    },
    {
      "epoch": 1.277926391810119,
      "grad_norm": 0.574273764631386,
      "learning_rate": 6.092048363773337e-05,
      "loss": 11.9568,
      "step": 23468
    },
    {
      "epoch": 1.277980845806702,
      "grad_norm": 0.522232082491125,
      "learning_rate": 6.0912366779964035e-05,
      "loss": 11.9929,
      "step": 23469
    },
    {
      "epoch": 1.278035299803285,
      "grad_norm": 0.5076968821147885,
      "learning_rate": 6.0904250226146734e-05,
      "loss": 11.9391,
      "step": 23470
    },
    {
      "epoch": 1.278089753799868,
      "grad_norm": 0.5842175089414539,
      "learning_rate": 6.0896133976344526e-05,
      "loss": 11.8999,
      "step": 23471
    },
    {
      "epoch": 1.2781442077964509,
      "grad_norm": 0.5444197913072716,
      "learning_rate": 6.0888018030620586e-05,
      "loss": 11.8363,
      "step": 23472
    },
    {
      "epoch": 1.2781986617930339,
      "grad_norm": 0.6741627689583077,
      "learning_rate": 6.0879902389037915e-05,
      "loss": 12.0643,
      "step": 23473
    },
    {
      "epoch": 1.2782531157896169,
      "grad_norm": 0.5012801147988735,
      "learning_rate": 6.087178705165969e-05,
      "loss": 11.8394,
      "step": 23474
    },
    {
      "epoch": 1.2783075697861999,
      "grad_norm": 0.5118881877419978,
      "learning_rate": 6.086367201854902e-05,
      "loss": 11.9598,
      "step": 23475
    },
    {
      "epoch": 1.278362023782783,
      "grad_norm": 0.564274403607142,
      "learning_rate": 6.085555728976899e-05,
      "loss": 11.8439,
      "step": 23476
    },
    {
      "epoch": 1.278416477779366,
      "grad_norm": 0.5169678361466552,
      "learning_rate": 6.084744286538273e-05,
      "loss": 11.8106,
      "step": 23477
    },
    {
      "epoch": 1.278470931775949,
      "grad_norm": 0.5286873537986246,
      "learning_rate": 6.08393287454533e-05,
      "loss": 11.986,
      "step": 23478
    },
    {
      "epoch": 1.278525385772532,
      "grad_norm": 0.5170143293378469,
      "learning_rate": 6.083121493004384e-05,
      "loss": 11.962,
      "step": 23479
    },
    {
      "epoch": 1.278579839769115,
      "grad_norm": 0.5290861828173461,
      "learning_rate": 6.082310141921739e-05,
      "loss": 11.9349,
      "step": 23480
    },
    {
      "epoch": 1.278634293765698,
      "grad_norm": 0.5574460050941807,
      "learning_rate": 6.0814988213037094e-05,
      "loss": 11.9356,
      "step": 23481
    },
    {
      "epoch": 1.278688747762281,
      "grad_norm": 0.7228777916774161,
      "learning_rate": 6.080687531156606e-05,
      "loss": 11.964,
      "step": 23482
    },
    {
      "epoch": 1.278743201758864,
      "grad_norm": 0.5450852058212383,
      "learning_rate": 6.079876271486729e-05,
      "loss": 11.9233,
      "step": 23483
    },
    {
      "epoch": 1.2787976557554472,
      "grad_norm": 0.5047807597759899,
      "learning_rate": 6.079065042300393e-05,
      "loss": 11.7822,
      "step": 23484
    },
    {
      "epoch": 1.2788521097520302,
      "grad_norm": 0.5826234685577726,
      "learning_rate": 6.0782538436039e-05,
      "loss": 11.8873,
      "step": 23485
    },
    {
      "epoch": 1.2789065637486132,
      "grad_norm": 0.4916929922109629,
      "learning_rate": 6.0774426754035685e-05,
      "loss": 11.9222,
      "step": 23486
    },
    {
      "epoch": 1.2789610177451962,
      "grad_norm": 0.5386901618287323,
      "learning_rate": 6.076631537705698e-05,
      "loss": 11.9255,
      "step": 23487
    },
    {
      "epoch": 1.2790154717417792,
      "grad_norm": 0.5870239611376995,
      "learning_rate": 6.075820430516599e-05,
      "loss": 12.0858,
      "step": 23488
    },
    {
      "epoch": 1.2790699257383622,
      "grad_norm": 0.4760090413465526,
      "learning_rate": 6.0750093538425804e-05,
      "loss": 11.8747,
      "step": 23489
    },
    {
      "epoch": 1.2791243797349452,
      "grad_norm": 0.5780090479800983,
      "learning_rate": 6.074198307689945e-05,
      "loss": 11.9545,
      "step": 23490
    },
    {
      "epoch": 1.2791788337315282,
      "grad_norm": 0.5252310941109722,
      "learning_rate": 6.0733872920650026e-05,
      "loss": 11.9644,
      "step": 23491
    },
    {
      "epoch": 1.2792332877281112,
      "grad_norm": 0.5351756410748505,
      "learning_rate": 6.0725763069740606e-05,
      "loss": 11.8312,
      "step": 23492
    },
    {
      "epoch": 1.2792877417246942,
      "grad_norm": 0.661062772247031,
      "learning_rate": 6.071765352423422e-05,
      "loss": 11.9972,
      "step": 23493
    },
    {
      "epoch": 1.2793421957212772,
      "grad_norm": 0.49745495292634456,
      "learning_rate": 6.070954428419395e-05,
      "loss": 11.8081,
      "step": 23494
    },
    {
      "epoch": 1.2793966497178602,
      "grad_norm": 0.5352121245957874,
      "learning_rate": 6.070143534968286e-05,
      "loss": 12.01,
      "step": 23495
    },
    {
      "epoch": 1.2794511037144431,
      "grad_norm": 0.5423569846107363,
      "learning_rate": 6.069332672076398e-05,
      "loss": 12.0703,
      "step": 23496
    },
    {
      "epoch": 1.2795055577110261,
      "grad_norm": 0.5854606957132008,
      "learning_rate": 6.068521839750039e-05,
      "loss": 11.911,
      "step": 23497
    },
    {
      "epoch": 1.2795600117076091,
      "grad_norm": 0.5446942177758455,
      "learning_rate": 6.067711037995514e-05,
      "loss": 11.9375,
      "step": 23498
    },
    {
      "epoch": 1.2796144657041923,
      "grad_norm": 0.6168174619334406,
      "learning_rate": 6.066900266819127e-05,
      "loss": 12.051,
      "step": 23499
    },
    {
      "epoch": 1.2796689197007753,
      "grad_norm": 0.6167618173501145,
      "learning_rate": 6.066089526227183e-05,
      "loss": 12.0398,
      "step": 23500
    },
    {
      "epoch": 1.2797233736973583,
      "grad_norm": 0.5456757129491338,
      "learning_rate": 6.065278816225988e-05,
      "loss": 11.9841,
      "step": 23501
    },
    {
      "epoch": 1.2797778276939413,
      "grad_norm": 0.5606503968145746,
      "learning_rate": 6.0644681368218457e-05,
      "loss": 11.976,
      "step": 23502
    },
    {
      "epoch": 1.2798322816905243,
      "grad_norm": 0.6417888688006007,
      "learning_rate": 6.0636574880210574e-05,
      "loss": 11.9576,
      "step": 23503
    },
    {
      "epoch": 1.2798867356871073,
      "grad_norm": 0.536462312518082,
      "learning_rate": 6.062846869829929e-05,
      "loss": 11.868,
      "step": 23504
    },
    {
      "epoch": 1.2799411896836903,
      "grad_norm": 0.4967739744330492,
      "learning_rate": 6.062036282254764e-05,
      "loss": 11.857,
      "step": 23505
    },
    {
      "epoch": 1.2799956436802733,
      "grad_norm": 0.5456281349254622,
      "learning_rate": 6.0612257253018646e-05,
      "loss": 12.0076,
      "step": 23506
    },
    {
      "epoch": 1.2800500976768563,
      "grad_norm": 0.5479946047060017,
      "learning_rate": 6.0604151989775316e-05,
      "loss": 11.9893,
      "step": 23507
    },
    {
      "epoch": 1.2801045516734395,
      "grad_norm": 0.5749185320388244,
      "learning_rate": 6.059604703288073e-05,
      "loss": 11.8828,
      "step": 23508
    },
    {
      "epoch": 1.2801590056700225,
      "grad_norm": 0.5171287018473631,
      "learning_rate": 6.05879423823979e-05,
      "loss": 11.8257,
      "step": 23509
    },
    {
      "epoch": 1.2802134596666055,
      "grad_norm": 0.6543870163652833,
      "learning_rate": 6.0579838038389826e-05,
      "loss": 11.9345,
      "step": 23510
    },
    {
      "epoch": 1.2802679136631885,
      "grad_norm": 0.49841729897115733,
      "learning_rate": 6.0571734000919554e-05,
      "loss": 11.9877,
      "step": 23511
    },
    {
      "epoch": 1.2803223676597715,
      "grad_norm": 0.5220794510578373,
      "learning_rate": 6.056363027005011e-05,
      "loss": 11.9932,
      "step": 23512
    },
    {
      "epoch": 1.2803768216563545,
      "grad_norm": 0.6025757275996647,
      "learning_rate": 6.055552684584447e-05,
      "loss": 11.9358,
      "step": 23513
    },
    {
      "epoch": 1.2804312756529375,
      "grad_norm": 0.5372976783892274,
      "learning_rate": 6.054742372836566e-05,
      "loss": 11.9075,
      "step": 23514
    },
    {
      "epoch": 1.2804857296495205,
      "grad_norm": 0.5521924338851903,
      "learning_rate": 6.0539320917676714e-05,
      "loss": 11.9101,
      "step": 23515
    },
    {
      "epoch": 1.2805401836461034,
      "grad_norm": 0.5861845568906207,
      "learning_rate": 6.0531218413840616e-05,
      "loss": 11.8921,
      "step": 23516
    },
    {
      "epoch": 1.2805946376426864,
      "grad_norm": 0.558562405022393,
      "learning_rate": 6.0523116216920374e-05,
      "loss": 11.8053,
      "step": 23517
    },
    {
      "epoch": 1.2806490916392694,
      "grad_norm": 0.517342550968619,
      "learning_rate": 6.0515014326978994e-05,
      "loss": 11.9603,
      "step": 23518
    },
    {
      "epoch": 1.2807035456358524,
      "grad_norm": 0.5778837259537967,
      "learning_rate": 6.05069127440795e-05,
      "loss": 11.9863,
      "step": 23519
    },
    {
      "epoch": 1.2807579996324354,
      "grad_norm": 0.5400045352346342,
      "learning_rate": 6.0498811468284876e-05,
      "loss": 11.9889,
      "step": 23520
    },
    {
      "epoch": 1.2808124536290184,
      "grad_norm": 0.5218498590662102,
      "learning_rate": 6.049071049965811e-05,
      "loss": 11.7868,
      "step": 23521
    },
    {
      "epoch": 1.2808669076256016,
      "grad_norm": 0.5186413862549286,
      "learning_rate": 6.048260983826224e-05,
      "loss": 12.0402,
      "step": 23522
    },
    {
      "epoch": 1.2809213616221846,
      "grad_norm": 0.5169947844750505,
      "learning_rate": 6.04745094841602e-05,
      "loss": 11.9166,
      "step": 23523
    },
    {
      "epoch": 1.2809758156187676,
      "grad_norm": 0.5430486455984835,
      "learning_rate": 6.0466409437414996e-05,
      "loss": 11.91,
      "step": 23524
    },
    {
      "epoch": 1.2810302696153506,
      "grad_norm": 0.5519218148989543,
      "learning_rate": 6.045830969808963e-05,
      "loss": 11.9521,
      "step": 23525
    },
    {
      "epoch": 1.2810847236119336,
      "grad_norm": 0.6091077701215787,
      "learning_rate": 6.045021026624707e-05,
      "loss": 11.9696,
      "step": 23526
    },
    {
      "epoch": 1.2811391776085166,
      "grad_norm": 0.5080581539476212,
      "learning_rate": 6.0442111141950306e-05,
      "loss": 11.9114,
      "step": 23527
    },
    {
      "epoch": 1.2811936316050996,
      "grad_norm": 0.5620295368624172,
      "learning_rate": 6.0434012325262336e-05,
      "loss": 11.9139,
      "step": 23528
    },
    {
      "epoch": 1.2812480856016826,
      "grad_norm": 0.5600735699797295,
      "learning_rate": 6.042591381624608e-05,
      "loss": 11.8936,
      "step": 23529
    },
    {
      "epoch": 1.2813025395982656,
      "grad_norm": 0.5644602481955032,
      "learning_rate": 6.041781561496458e-05,
      "loss": 11.8653,
      "step": 23530
    },
    {
      "epoch": 1.2813569935948488,
      "grad_norm": 0.5071749127778168,
      "learning_rate": 6.0409717721480796e-05,
      "loss": 11.93,
      "step": 23531
    },
    {
      "epoch": 1.2814114475914318,
      "grad_norm": 0.5515682328891715,
      "learning_rate": 6.040162013585772e-05,
      "loss": 11.8772,
      "step": 23532
    },
    {
      "epoch": 1.2814659015880148,
      "grad_norm": 0.5675425006480583,
      "learning_rate": 6.039352285815823e-05,
      "loss": 11.881,
      "step": 23533
    },
    {
      "epoch": 1.2815203555845978,
      "grad_norm": 0.5471934437434559,
      "learning_rate": 6.038542588844536e-05,
      "loss": 12.0435,
      "step": 23534
    },
    {
      "epoch": 1.2815748095811808,
      "grad_norm": 0.5565368557901031,
      "learning_rate": 6.037732922678206e-05,
      "loss": 11.98,
      "step": 23535
    },
    {
      "epoch": 1.2816292635777637,
      "grad_norm": 0.5070680059614715,
      "learning_rate": 6.036923287323131e-05,
      "loss": 11.7727,
      "step": 23536
    },
    {
      "epoch": 1.2816837175743467,
      "grad_norm": 0.553493808862405,
      "learning_rate": 6.0361136827856025e-05,
      "loss": 12.0505,
      "step": 23537
    },
    {
      "epoch": 1.2817381715709297,
      "grad_norm": 0.5608070685866844,
      "learning_rate": 6.0353041090719196e-05,
      "loss": 12.0352,
      "step": 23538
    },
    {
      "epoch": 1.2817926255675127,
      "grad_norm": 0.5491546897844471,
      "learning_rate": 6.034494566188378e-05,
      "loss": 11.9842,
      "step": 23539
    },
    {
      "epoch": 1.2818470795640957,
      "grad_norm": 0.5405475520046354,
      "learning_rate": 6.033685054141272e-05,
      "loss": 11.9077,
      "step": 23540
    },
    {
      "epoch": 1.2819015335606787,
      "grad_norm": 0.5240052472356858,
      "learning_rate": 6.0328755729368925e-05,
      "loss": 11.8298,
      "step": 23541
    },
    {
      "epoch": 1.2819559875572617,
      "grad_norm": 0.4917994796912182,
      "learning_rate": 6.032066122581545e-05,
      "loss": 11.8881,
      "step": 23542
    },
    {
      "epoch": 1.2820104415538447,
      "grad_norm": 0.5027360717309111,
      "learning_rate": 6.031256703081511e-05,
      "loss": 11.9357,
      "step": 23543
    },
    {
      "epoch": 1.2820648955504277,
      "grad_norm": 0.5170751853146255,
      "learning_rate": 6.0304473144430926e-05,
      "loss": 11.8487,
      "step": 23544
    },
    {
      "epoch": 1.2821193495470107,
      "grad_norm": 0.5896824757083037,
      "learning_rate": 6.0296379566725794e-05,
      "loss": 11.8997,
      "step": 23545
    },
    {
      "epoch": 1.282173803543594,
      "grad_norm": 0.5906842332292388,
      "learning_rate": 6.028828629776269e-05,
      "loss": 11.8785,
      "step": 23546
    },
    {
      "epoch": 1.2822282575401769,
      "grad_norm": 0.5480852917853959,
      "learning_rate": 6.028019333760452e-05,
      "loss": 11.8501,
      "step": 23547
    },
    {
      "epoch": 1.2822827115367599,
      "grad_norm": 0.5572639996001882,
      "learning_rate": 6.0272100686314234e-05,
      "loss": 11.9747,
      "step": 23548
    },
    {
      "epoch": 1.2823371655333429,
      "grad_norm": 0.5369151321080103,
      "learning_rate": 6.0264008343954757e-05,
      "loss": 11.894,
      "step": 23549
    },
    {
      "epoch": 1.2823916195299259,
      "grad_norm": 0.5091076563718924,
      "learning_rate": 6.025591631058901e-05,
      "loss": 11.925,
      "step": 23550
    },
    {
      "epoch": 1.2824460735265089,
      "grad_norm": 0.6046272320357335,
      "learning_rate": 6.0247824586279934e-05,
      "loss": 11.9582,
      "step": 23551
    },
    {
      "epoch": 1.2825005275230918,
      "grad_norm": 0.5397308357040135,
      "learning_rate": 6.023973317109042e-05,
      "loss": 11.9748,
      "step": 23552
    },
    {
      "epoch": 1.2825549815196748,
      "grad_norm": 0.5633565540476361,
      "learning_rate": 6.0231642065083446e-05,
      "loss": 11.9608,
      "step": 23553
    },
    {
      "epoch": 1.282609435516258,
      "grad_norm": 0.5766835774345143,
      "learning_rate": 6.0223551268321874e-05,
      "loss": 12.1016,
      "step": 23554
    },
    {
      "epoch": 1.282663889512841,
      "grad_norm": 0.5828712749179817,
      "learning_rate": 6.021546078086864e-05,
      "loss": 11.944,
      "step": 23555
    },
    {
      "epoch": 1.282718343509424,
      "grad_norm": 0.5719930660253305,
      "learning_rate": 6.0207370602786626e-05,
      "loss": 11.9348,
      "step": 23556
    },
    {
      "epoch": 1.282772797506007,
      "grad_norm": 0.4884218997155032,
      "learning_rate": 6.019928073413879e-05,
      "loss": 11.8262,
      "step": 23557
    },
    {
      "epoch": 1.28282725150259,
      "grad_norm": 0.5204375143460239,
      "learning_rate": 6.019119117498802e-05,
      "loss": 11.9146,
      "step": 23558
    },
    {
      "epoch": 1.282881705499173,
      "grad_norm": 0.5678449367458687,
      "learning_rate": 6.018310192539722e-05,
      "loss": 11.979,
      "step": 23559
    },
    {
      "epoch": 1.282936159495756,
      "grad_norm": 0.5339692036640276,
      "learning_rate": 6.0175012985429313e-05,
      "loss": 11.9012,
      "step": 23560
    },
    {
      "epoch": 1.282990613492339,
      "grad_norm": 0.5594405997241506,
      "learning_rate": 6.0166924355147166e-05,
      "loss": 11.9855,
      "step": 23561
    },
    {
      "epoch": 1.283045067488922,
      "grad_norm": 0.5583463392382488,
      "learning_rate": 6.015883603461372e-05,
      "loss": 11.9826,
      "step": 23562
    },
    {
      "epoch": 1.283099521485505,
      "grad_norm": 0.529926138611232,
      "learning_rate": 6.015074802389184e-05,
      "loss": 11.8863,
      "step": 23563
    },
    {
      "epoch": 1.283153975482088,
      "grad_norm": 0.5203733800981577,
      "learning_rate": 6.0142660323044434e-05,
      "loss": 11.8016,
      "step": 23564
    },
    {
      "epoch": 1.283208429478671,
      "grad_norm": 0.5748523056640842,
      "learning_rate": 6.0134572932134356e-05,
      "loss": 11.8061,
      "step": 23565
    },
    {
      "epoch": 1.283262883475254,
      "grad_norm": 0.5779499015108761,
      "learning_rate": 6.0126485851224534e-05,
      "loss": 11.9713,
      "step": 23566
    },
    {
      "epoch": 1.283317337471837,
      "grad_norm": 0.5401640476135271,
      "learning_rate": 6.011839908037783e-05,
      "loss": 11.9529,
      "step": 23567
    },
    {
      "epoch": 1.28337179146842,
      "grad_norm": 0.538671360492073,
      "learning_rate": 6.011031261965716e-05,
      "loss": 11.8232,
      "step": 23568
    },
    {
      "epoch": 1.2834262454650032,
      "grad_norm": 0.5492423700879108,
      "learning_rate": 6.010222646912539e-05,
      "loss": 11.7709,
      "step": 23569
    },
    {
      "epoch": 1.2834806994615862,
      "grad_norm": 0.5787543441007559,
      "learning_rate": 6.00941406288454e-05,
      "loss": 11.8725,
      "step": 23570
    },
    {
      "epoch": 1.2835351534581692,
      "grad_norm": 0.6230576628149644,
      "learning_rate": 6.0086055098880055e-05,
      "loss": 11.9372,
      "step": 23571
    },
    {
      "epoch": 1.2835896074547521,
      "grad_norm": 0.5521267877334615,
      "learning_rate": 6.007796987929225e-05,
      "loss": 11.9578,
      "step": 23572
    },
    {
      "epoch": 1.2836440614513351,
      "grad_norm": 0.5963668936277664,
      "learning_rate": 6.0069884970144865e-05,
      "loss": 11.9397,
      "step": 23573
    },
    {
      "epoch": 1.2836985154479181,
      "grad_norm": 0.5510056906815245,
      "learning_rate": 6.006180037150073e-05,
      "loss": 11.6585,
      "step": 23574
    },
    {
      "epoch": 1.2837529694445011,
      "grad_norm": 0.5483257883956336,
      "learning_rate": 6.005371608342272e-05,
      "loss": 11.8992,
      "step": 23575
    },
    {
      "epoch": 1.2838074234410841,
      "grad_norm": 0.5198488024735248,
      "learning_rate": 6.004563210597372e-05,
      "loss": 11.9707,
      "step": 23576
    },
    {
      "epoch": 1.283861877437667,
      "grad_norm": 0.5488059841016077,
      "learning_rate": 6.0037548439216596e-05,
      "loss": 11.9824,
      "step": 23577
    },
    {
      "epoch": 1.2839163314342503,
      "grad_norm": 0.5917836802938782,
      "learning_rate": 6.0029465083214166e-05,
      "loss": 11.8574,
      "step": 23578
    },
    {
      "epoch": 1.2839707854308333,
      "grad_norm": 0.5013777123048397,
      "learning_rate": 6.002138203802934e-05,
      "loss": 11.8769,
      "step": 23579
    },
    {
      "epoch": 1.2840252394274163,
      "grad_norm": 0.5305015348335042,
      "learning_rate": 6.0013299303724955e-05,
      "loss": 11.9521,
      "step": 23580
    },
    {
      "epoch": 1.2840796934239993,
      "grad_norm": 0.5501595583170282,
      "learning_rate": 6.0005216880363866e-05,
      "loss": 11.9786,
      "step": 23581
    },
    {
      "epoch": 1.2841341474205823,
      "grad_norm": 0.4893357817304148,
      "learning_rate": 5.99971347680089e-05,
      "loss": 11.8602,
      "step": 23582
    },
    {
      "epoch": 1.2841886014171653,
      "grad_norm": 0.5931966388465176,
      "learning_rate": 5.998905296672296e-05,
      "loss": 12.0242,
      "step": 23583
    },
    {
      "epoch": 1.2842430554137483,
      "grad_norm": 0.502447089109366,
      "learning_rate": 5.9980971476568825e-05,
      "loss": 11.9041,
      "step": 23584
    },
    {
      "epoch": 1.2842975094103313,
      "grad_norm": 0.5075812170370226,
      "learning_rate": 5.997289029760937e-05,
      "loss": 11.8359,
      "step": 23585
    },
    {
      "epoch": 1.2843519634069143,
      "grad_norm": 0.5448849983072174,
      "learning_rate": 5.996480942990743e-05,
      "loss": 11.8758,
      "step": 23586
    },
    {
      "epoch": 1.2844064174034973,
      "grad_norm": 0.553796465616994,
      "learning_rate": 5.995672887352586e-05,
      "loss": 11.9681,
      "step": 23587
    },
    {
      "epoch": 1.2844608714000803,
      "grad_norm": 0.5487255265970418,
      "learning_rate": 5.994864862852746e-05,
      "loss": 11.8359,
      "step": 23588
    },
    {
      "epoch": 1.2845153253966632,
      "grad_norm": 0.573686709985358,
      "learning_rate": 5.9940568694975096e-05,
      "loss": 11.9695,
      "step": 23589
    },
    {
      "epoch": 1.2845697793932462,
      "grad_norm": 0.5637612380220137,
      "learning_rate": 5.993248907293156e-05,
      "loss": 11.7809,
      "step": 23590
    },
    {
      "epoch": 1.2846242333898292,
      "grad_norm": 0.4951625100799867,
      "learning_rate": 5.9924409762459746e-05,
      "loss": 11.815,
      "step": 23591
    },
    {
      "epoch": 1.2846786873864124,
      "grad_norm": 0.5436142416323838,
      "learning_rate": 5.991633076362242e-05,
      "loss": 12.0503,
      "step": 23592
    },
    {
      "epoch": 1.2847331413829954,
      "grad_norm": 0.5323256001549563,
      "learning_rate": 5.990825207648247e-05,
      "loss": 12.0098,
      "step": 23593
    },
    {
      "epoch": 1.2847875953795784,
      "grad_norm": 0.5898551135137177,
      "learning_rate": 5.9900173701102634e-05,
      "loss": 11.9454,
      "step": 23594
    },
    {
      "epoch": 1.2848420493761614,
      "grad_norm": 0.4723549336831795,
      "learning_rate": 5.9892095637545784e-05,
      "loss": 11.9479,
      "step": 23595
    },
    {
      "epoch": 1.2848965033727444,
      "grad_norm": 0.4998330178812302,
      "learning_rate": 5.988401788587472e-05,
      "loss": 11.88,
      "step": 23596
    },
    {
      "epoch": 1.2849509573693274,
      "grad_norm": 0.5051779881653149,
      "learning_rate": 5.987594044615225e-05,
      "loss": 11.9079,
      "step": 23597
    },
    {
      "epoch": 1.2850054113659104,
      "grad_norm": 0.5245486970219934,
      "learning_rate": 5.986786331844122e-05,
      "loss": 11.889,
      "step": 23598
    },
    {
      "epoch": 1.2850598653624934,
      "grad_norm": 0.5615451894349467,
      "learning_rate": 5.98597865028044e-05,
      "loss": 11.7919,
      "step": 23599
    },
    {
      "epoch": 1.2851143193590764,
      "grad_norm": 0.4951813094482244,
      "learning_rate": 5.9851709999304615e-05,
      "loss": 12.0349,
      "step": 23600
    },
    {
      "epoch": 1.2851687733556596,
      "grad_norm": 0.5609478608796767,
      "learning_rate": 5.984363380800465e-05,
      "loss": 12.0273,
      "step": 23601
    },
    {
      "epoch": 1.2852232273522426,
      "grad_norm": 0.5654504929409612,
      "learning_rate": 5.9835557928967335e-05,
      "loss": 11.9728,
      "step": 23602
    },
    {
      "epoch": 1.2852776813488256,
      "grad_norm": 0.5180004123328328,
      "learning_rate": 5.982748236225551e-05,
      "loss": 11.9134,
      "step": 23603
    },
    {
      "epoch": 1.2853321353454086,
      "grad_norm": 0.5284728078042654,
      "learning_rate": 5.9819407107931856e-05,
      "loss": 11.8346,
      "step": 23604
    },
    {
      "epoch": 1.2853865893419916,
      "grad_norm": 0.5193660021771949,
      "learning_rate": 5.981133216605923e-05,
      "loss": 11.9533,
      "step": 23605
    },
    {
      "epoch": 1.2854410433385746,
      "grad_norm": 0.5128027865249251,
      "learning_rate": 5.9803257536700444e-05,
      "loss": 11.9465,
      "step": 23606
    },
    {
      "epoch": 1.2854954973351576,
      "grad_norm": 0.5292023147523693,
      "learning_rate": 5.979518321991826e-05,
      "loss": 11.9244,
      "step": 23607
    },
    {
      "epoch": 1.2855499513317405,
      "grad_norm": 0.5840133205707407,
      "learning_rate": 5.9787109215775484e-05,
      "loss": 11.9686,
      "step": 23608
    },
    {
      "epoch": 1.2856044053283235,
      "grad_norm": 0.5453332728022575,
      "learning_rate": 5.977903552433488e-05,
      "loss": 11.8874,
      "step": 23609
    },
    {
      "epoch": 1.2856588593249065,
      "grad_norm": 0.5694304844855177,
      "learning_rate": 5.9770962145659245e-05,
      "loss": 12.0776,
      "step": 23610
    },
    {
      "epoch": 1.2857133133214895,
      "grad_norm": 0.5776448964429606,
      "learning_rate": 5.9762889079811354e-05,
      "loss": 11.9809,
      "step": 23611
    },
    {
      "epoch": 1.2857677673180725,
      "grad_norm": 0.6017741977456825,
      "learning_rate": 5.975481632685396e-05,
      "loss": 11.8373,
      "step": 23612
    },
    {
      "epoch": 1.2858222213146555,
      "grad_norm": 0.5445144755282323,
      "learning_rate": 5.974674388684993e-05,
      "loss": 11.8613,
      "step": 23613
    },
    {
      "epoch": 1.2858766753112385,
      "grad_norm": 0.5497804284953607,
      "learning_rate": 5.973867175986193e-05,
      "loss": 11.9679,
      "step": 23614
    },
    {
      "epoch": 1.2859311293078217,
      "grad_norm": 0.5528742445411363,
      "learning_rate": 5.973059994595277e-05,
      "loss": 11.9977,
      "step": 23615
    },
    {
      "epoch": 1.2859855833044047,
      "grad_norm": 0.6365557765121772,
      "learning_rate": 5.97225284451852e-05,
      "loss": 11.8794,
      "step": 23616
    },
    {
      "epoch": 1.2860400373009877,
      "grad_norm": 0.5608041996493477,
      "learning_rate": 5.9714457257622016e-05,
      "loss": 12.0025,
      "step": 23617
    },
    {
      "epoch": 1.2860944912975707,
      "grad_norm": 0.4649658802703738,
      "learning_rate": 5.9706386383325976e-05,
      "loss": 11.8543,
      "step": 23618
    },
    {
      "epoch": 1.2861489452941537,
      "grad_norm": 0.5246046723573995,
      "learning_rate": 5.969831582235983e-05,
      "loss": 11.7959,
      "step": 23619
    },
    {
      "epoch": 1.2862033992907367,
      "grad_norm": 0.481864440579084,
      "learning_rate": 5.969024557478633e-05,
      "loss": 11.9564,
      "step": 23620
    },
    {
      "epoch": 1.2862578532873197,
      "grad_norm": 0.5037851902097632,
      "learning_rate": 5.9682175640668244e-05,
      "loss": 11.7368,
      "step": 23621
    },
    {
      "epoch": 1.2863123072839027,
      "grad_norm": 0.5100480487418915,
      "learning_rate": 5.967410602006833e-05,
      "loss": 11.8962,
      "step": 23622
    },
    {
      "epoch": 1.2863667612804857,
      "grad_norm": 0.5635853340470395,
      "learning_rate": 5.9666036713049356e-05,
      "loss": 11.7806,
      "step": 23623
    },
    {
      "epoch": 1.2864212152770689,
      "grad_norm": 0.5291879304710763,
      "learning_rate": 5.9657967719674015e-05,
      "loss": 11.9973,
      "step": 23624
    },
    {
      "epoch": 1.2864756692736519,
      "grad_norm": 0.5879224653664377,
      "learning_rate": 5.9649899040005085e-05,
      "loss": 11.9562,
      "step": 23625
    },
    {
      "epoch": 1.2865301232702349,
      "grad_norm": 0.4965507218558734,
      "learning_rate": 5.9641830674105294e-05,
      "loss": 11.877,
      "step": 23626
    },
    {
      "epoch": 1.2865845772668179,
      "grad_norm": 0.6443914294268037,
      "learning_rate": 5.9633762622037396e-05,
      "loss": 12.0468,
      "step": 23627
    },
    {
      "epoch": 1.2866390312634008,
      "grad_norm": 0.5358332881599275,
      "learning_rate": 5.962569488386413e-05,
      "loss": 11.9733,
      "step": 23628
    },
    {
      "epoch": 1.2866934852599838,
      "grad_norm": 0.4888460672702682,
      "learning_rate": 5.9617627459648253e-05,
      "loss": 11.755,
      "step": 23629
    },
    {
      "epoch": 1.2867479392565668,
      "grad_norm": 0.5055719708645405,
      "learning_rate": 5.960956034945245e-05,
      "loss": 11.8508,
      "step": 23630
    },
    {
      "epoch": 1.2868023932531498,
      "grad_norm": 0.5385345014820144,
      "learning_rate": 5.96014935533395e-05,
      "loss": 11.8001,
      "step": 23631
    },
    {
      "epoch": 1.2868568472497328,
      "grad_norm": 0.5515374747980404,
      "learning_rate": 5.95934270713721e-05,
      "loss": 11.989,
      "step": 23632
    },
    {
      "epoch": 1.2869113012463158,
      "grad_norm": 0.5364470059610076,
      "learning_rate": 5.958536090361302e-05,
      "loss": 11.8877,
      "step": 23633
    },
    {
      "epoch": 1.2869657552428988,
      "grad_norm": 0.6145460880505714,
      "learning_rate": 5.9577295050124926e-05,
      "loss": 12.0312,
      "step": 23634
    },
    {
      "epoch": 1.2870202092394818,
      "grad_norm": 0.6416609088364478,
      "learning_rate": 5.956922951097055e-05,
      "loss": 11.8751,
      "step": 23635
    },
    {
      "epoch": 1.2870746632360648,
      "grad_norm": 0.5836040120019552,
      "learning_rate": 5.956116428621263e-05,
      "loss": 11.9816,
      "step": 23636
    },
    {
      "epoch": 1.2871291172326478,
      "grad_norm": 0.5161969708347284,
      "learning_rate": 5.955309937591389e-05,
      "loss": 11.936,
      "step": 23637
    },
    {
      "epoch": 1.2871835712292308,
      "grad_norm": 0.5620302967688622,
      "learning_rate": 5.954503478013703e-05,
      "loss": 11.9001,
      "step": 23638
    },
    {
      "epoch": 1.287238025225814,
      "grad_norm": 0.6406360918722982,
      "learning_rate": 5.9536970498944745e-05,
      "loss": 11.8731,
      "step": 23639
    },
    {
      "epoch": 1.287292479222397,
      "grad_norm": 0.5256318177011917,
      "learning_rate": 5.9528906532399776e-05,
      "loss": 11.9375,
      "step": 23640
    },
    {
      "epoch": 1.28734693321898,
      "grad_norm": 0.5221750789978684,
      "learning_rate": 5.952084288056482e-05,
      "loss": 11.951,
      "step": 23641
    },
    {
      "epoch": 1.287401387215563,
      "grad_norm": 0.5805291395444275,
      "learning_rate": 5.951277954350259e-05,
      "loss": 11.9437,
      "step": 23642
    },
    {
      "epoch": 1.287455841212146,
      "grad_norm": 0.5745203857240743,
      "learning_rate": 5.950471652127578e-05,
      "loss": 11.9151,
      "step": 23643
    },
    {
      "epoch": 1.287510295208729,
      "grad_norm": 0.5976754697185809,
      "learning_rate": 5.949665381394707e-05,
      "loss": 12.0609,
      "step": 23644
    },
    {
      "epoch": 1.287564749205312,
      "grad_norm": 0.5771545026760481,
      "learning_rate": 5.948859142157917e-05,
      "loss": 11.8742,
      "step": 23645
    },
    {
      "epoch": 1.287619203201895,
      "grad_norm": 0.5279876863213004,
      "learning_rate": 5.948052934423478e-05,
      "loss": 11.8626,
      "step": 23646
    },
    {
      "epoch": 1.287673657198478,
      "grad_norm": 0.48604035686339353,
      "learning_rate": 5.947246758197658e-05,
      "loss": 11.9474,
      "step": 23647
    },
    {
      "epoch": 1.2877281111950611,
      "grad_norm": 0.6153703921357986,
      "learning_rate": 5.946440613486728e-05,
      "loss": 11.9791,
      "step": 23648
    },
    {
      "epoch": 1.2877825651916441,
      "grad_norm": 0.49441011159873605,
      "learning_rate": 5.945634500296955e-05,
      "loss": 11.8251,
      "step": 23649
    },
    {
      "epoch": 1.2878370191882271,
      "grad_norm": 0.5557404908758066,
      "learning_rate": 5.944828418634607e-05,
      "loss": 11.8788,
      "step": 23650
    },
    {
      "epoch": 1.2878914731848101,
      "grad_norm": 0.5616918914548961,
      "learning_rate": 5.9440223685059536e-05,
      "loss": 11.9618,
      "step": 23651
    },
    {
      "epoch": 1.2879459271813931,
      "grad_norm": 0.5308384410030225,
      "learning_rate": 5.9432163499172645e-05,
      "loss": 11.8682,
      "step": 23652
    },
    {
      "epoch": 1.288000381177976,
      "grad_norm": 0.5100486469580383,
      "learning_rate": 5.942410362874807e-05,
      "loss": 11.93,
      "step": 23653
    },
    {
      "epoch": 1.288054835174559,
      "grad_norm": 0.5627076991753851,
      "learning_rate": 5.941604407384842e-05,
      "loss": 11.8506,
      "step": 23654
    },
    {
      "epoch": 1.288109289171142,
      "grad_norm": 0.5809962828595369,
      "learning_rate": 5.940798483453645e-05,
      "loss": 12.1106,
      "step": 23655
    },
    {
      "epoch": 1.288163743167725,
      "grad_norm": 0.5761211825251635,
      "learning_rate": 5.939992591087478e-05,
      "loss": 11.9726,
      "step": 23656
    },
    {
      "epoch": 1.288218197164308,
      "grad_norm": 0.5176080305630002,
      "learning_rate": 5.939186730292611e-05,
      "loss": 11.832,
      "step": 23657
    },
    {
      "epoch": 1.288272651160891,
      "grad_norm": 0.5168934673239702,
      "learning_rate": 5.938380901075308e-05,
      "loss": 11.9764,
      "step": 23658
    },
    {
      "epoch": 1.288327105157474,
      "grad_norm": 0.5240142885989021,
      "learning_rate": 5.937575103441836e-05,
      "loss": 11.9576,
      "step": 23659
    },
    {
      "epoch": 1.288381559154057,
      "grad_norm": 0.5473455566039885,
      "learning_rate": 5.936769337398462e-05,
      "loss": 12.0432,
      "step": 23660
    },
    {
      "epoch": 1.28843601315064,
      "grad_norm": 0.533416332353287,
      "learning_rate": 5.935963602951449e-05,
      "loss": 11.8169,
      "step": 23661
    },
    {
      "epoch": 1.2884904671472233,
      "grad_norm": 0.5651284474850365,
      "learning_rate": 5.9351579001070655e-05,
      "loss": 12.0078,
      "step": 23662
    },
    {
      "epoch": 1.2885449211438063,
      "grad_norm": 0.6008388714869841,
      "learning_rate": 5.93435222887158e-05,
      "loss": 11.9991,
      "step": 23663
    },
    {
      "epoch": 1.2885993751403892,
      "grad_norm": 0.5592068078442981,
      "learning_rate": 5.933546589251251e-05,
      "loss": 11.9556,
      "step": 23664
    },
    {
      "epoch": 1.2886538291369722,
      "grad_norm": 0.6156076208952148,
      "learning_rate": 5.9327409812523424e-05,
      "loss": 11.9975,
      "step": 23665
    },
    {
      "epoch": 1.2887082831335552,
      "grad_norm": 0.5952681026970403,
      "learning_rate": 5.9319354048811237e-05,
      "loss": 11.9256,
      "step": 23666
    },
    {
      "epoch": 1.2887627371301382,
      "grad_norm": 0.5675388295157449,
      "learning_rate": 5.931129860143858e-05,
      "loss": 11.8337,
      "step": 23667
    },
    {
      "epoch": 1.2888171911267212,
      "grad_norm": 0.5549848730510045,
      "learning_rate": 5.9303243470468095e-05,
      "loss": 11.9471,
      "step": 23668
    },
    {
      "epoch": 1.2888716451233042,
      "grad_norm": 0.5394915762639385,
      "learning_rate": 5.929518865596241e-05,
      "loss": 11.9956,
      "step": 23669
    },
    {
      "epoch": 1.2889260991198872,
      "grad_norm": 0.5430149223289498,
      "learning_rate": 5.928713415798416e-05,
      "loss": 11.9668,
      "step": 23670
    },
    {
      "epoch": 1.2889805531164704,
      "grad_norm": 0.5864954645272095,
      "learning_rate": 5.927907997659598e-05,
      "loss": 11.9191,
      "step": 23671
    },
    {
      "epoch": 1.2890350071130534,
      "grad_norm": 0.5578452271628115,
      "learning_rate": 5.9271026111860486e-05,
      "loss": 11.8341,
      "step": 23672
    },
    {
      "epoch": 1.2890894611096364,
      "grad_norm": 0.5312293951369268,
      "learning_rate": 5.926297256384039e-05,
      "loss": 11.844,
      "step": 23673
    },
    {
      "epoch": 1.2891439151062194,
      "grad_norm": 0.5332665994177112,
      "learning_rate": 5.92549193325982e-05,
      "loss": 11.982,
      "step": 23674
    },
    {
      "epoch": 1.2891983691028024,
      "grad_norm": 0.5135015601352471,
      "learning_rate": 5.92468664181966e-05,
      "loss": 12.0161,
      "step": 23675
    },
    {
      "epoch": 1.2892528230993854,
      "grad_norm": 0.5983764378190438,
      "learning_rate": 5.923881382069818e-05,
      "loss": 11.7917,
      "step": 23676
    },
    {
      "epoch": 1.2893072770959684,
      "grad_norm": 0.5557263800671538,
      "learning_rate": 5.923076154016559e-05,
      "loss": 11.9336,
      "step": 23677
    },
    {
      "epoch": 1.2893617310925514,
      "grad_norm": 0.5779450898848552,
      "learning_rate": 5.922270957666145e-05,
      "loss": 12.007,
      "step": 23678
    },
    {
      "epoch": 1.2894161850891344,
      "grad_norm": 0.5638874342151753,
      "learning_rate": 5.921465793024834e-05,
      "loss": 11.9898,
      "step": 23679
    },
    {
      "epoch": 1.2894706390857174,
      "grad_norm": 0.5401848903416356,
      "learning_rate": 5.9206606600988904e-05,
      "loss": 12.068,
      "step": 23680
    },
    {
      "epoch": 1.2895250930823003,
      "grad_norm": 0.5063761932526026,
      "learning_rate": 5.919855558894574e-05,
      "loss": 11.9837,
      "step": 23681
    },
    {
      "epoch": 1.2895795470788833,
      "grad_norm": 0.5055177441847434,
      "learning_rate": 5.919050489418143e-05,
      "loss": 11.9499,
      "step": 23682
    },
    {
      "epoch": 1.2896340010754663,
      "grad_norm": 0.572799586197494,
      "learning_rate": 5.918245451675863e-05,
      "loss": 12.0118,
      "step": 23683
    },
    {
      "epoch": 1.2896884550720493,
      "grad_norm": 0.5243849526515628,
      "learning_rate": 5.9174404456739896e-05,
      "loss": 11.9704,
      "step": 23684
    },
    {
      "epoch": 1.2897429090686325,
      "grad_norm": 0.5067721466066325,
      "learning_rate": 5.9166354714187834e-05,
      "loss": 11.9171,
      "step": 23685
    },
    {
      "epoch": 1.2897973630652155,
      "grad_norm": 0.5837874081527563,
      "learning_rate": 5.915830528916504e-05,
      "loss": 11.9104,
      "step": 23686
    },
    {
      "epoch": 1.2898518170617985,
      "grad_norm": 0.6207793209349238,
      "learning_rate": 5.9150256181734085e-05,
      "loss": 11.9239,
      "step": 23687
    },
    {
      "epoch": 1.2899062710583815,
      "grad_norm": 0.5254442650792589,
      "learning_rate": 5.914220739195763e-05,
      "loss": 11.9123,
      "step": 23688
    },
    {
      "epoch": 1.2899607250549645,
      "grad_norm": 0.49713421448790956,
      "learning_rate": 5.913415891989821e-05,
      "loss": 11.9195,
      "step": 23689
    },
    {
      "epoch": 1.2900151790515475,
      "grad_norm": 0.534137301964781,
      "learning_rate": 5.912611076561843e-05,
      "loss": 11.7454,
      "step": 23690
    },
    {
      "epoch": 1.2900696330481305,
      "grad_norm": 0.4969835572591292,
      "learning_rate": 5.9118062929180854e-05,
      "loss": 11.985,
      "step": 23691
    },
    {
      "epoch": 1.2901240870447135,
      "grad_norm": 0.5544418764078004,
      "learning_rate": 5.911001541064809e-05,
      "loss": 11.9663,
      "step": 23692
    },
    {
      "epoch": 1.2901785410412965,
      "grad_norm": 0.601756892019909,
      "learning_rate": 5.9101968210082714e-05,
      "loss": 11.9447,
      "step": 23693
    },
    {
      "epoch": 1.2902329950378797,
      "grad_norm": 0.5287257994580373,
      "learning_rate": 5.909392132754727e-05,
      "loss": 11.8107,
      "step": 23694
    },
    {
      "epoch": 1.2902874490344627,
      "grad_norm": 0.5607373362066157,
      "learning_rate": 5.908587476310437e-05,
      "loss": 11.9615,
      "step": 23695
    },
    {
      "epoch": 1.2903419030310457,
      "grad_norm": 0.5465967799122933,
      "learning_rate": 5.907782851681656e-05,
      "loss": 11.9678,
      "step": 23696
    },
    {
      "epoch": 1.2903963570276287,
      "grad_norm": 0.572141024917695,
      "learning_rate": 5.9069782588746424e-05,
      "loss": 11.9509,
      "step": 23697
    },
    {
      "epoch": 1.2904508110242117,
      "grad_norm": 0.5471142728851983,
      "learning_rate": 5.9061736978956515e-05,
      "loss": 11.8992,
      "step": 23698
    },
    {
      "epoch": 1.2905052650207947,
      "grad_norm": 0.5580420481713758,
      "learning_rate": 5.9053691687509385e-05,
      "loss": 11.8991,
      "step": 23699
    },
    {
      "epoch": 1.2905597190173776,
      "grad_norm": 0.6032433260272864,
      "learning_rate": 5.904564671446763e-05,
      "loss": 11.9893,
      "step": 23700
    },
    {
      "epoch": 1.2906141730139606,
      "grad_norm": 0.5273241292063136,
      "learning_rate": 5.903760205989379e-05,
      "loss": 11.8925,
      "step": 23701
    },
    {
      "epoch": 1.2906686270105436,
      "grad_norm": 0.5173306857448301,
      "learning_rate": 5.902955772385045e-05,
      "loss": 12.0058,
      "step": 23702
    },
    {
      "epoch": 1.2907230810071266,
      "grad_norm": 0.6058203988463428,
      "learning_rate": 5.902151370640012e-05,
      "loss": 11.9569,
      "step": 23703
    },
    {
      "epoch": 1.2907775350037096,
      "grad_norm": 0.5844213336884051,
      "learning_rate": 5.90134700076054e-05,
      "loss": 12.0116,
      "step": 23704
    },
    {
      "epoch": 1.2908319890002926,
      "grad_norm": 0.5443874397600219,
      "learning_rate": 5.900542662752879e-05,
      "loss": 11.9553,
      "step": 23705
    },
    {
      "epoch": 1.2908864429968756,
      "grad_norm": 0.5647633404967047,
      "learning_rate": 5.899738356623286e-05,
      "loss": 11.9905,
      "step": 23706
    },
    {
      "epoch": 1.2909408969934586,
      "grad_norm": 0.501919395815344,
      "learning_rate": 5.8989340823780135e-05,
      "loss": 11.9122,
      "step": 23707
    },
    {
      "epoch": 1.2909953509900416,
      "grad_norm": 0.48842413756295855,
      "learning_rate": 5.898129840023319e-05,
      "loss": 11.8905,
      "step": 23708
    },
    {
      "epoch": 1.2910498049866248,
      "grad_norm": 0.6030129318770379,
      "learning_rate": 5.8973256295654546e-05,
      "loss": 11.9723,
      "step": 23709
    },
    {
      "epoch": 1.2911042589832078,
      "grad_norm": 0.5126006741755619,
      "learning_rate": 5.8965214510106715e-05,
      "loss": 11.9802,
      "step": 23710
    },
    {
      "epoch": 1.2911587129797908,
      "grad_norm": 0.6057647348014951,
      "learning_rate": 5.895717304365228e-05,
      "loss": 11.811,
      "step": 23711
    },
    {
      "epoch": 1.2912131669763738,
      "grad_norm": 0.5512818310963955,
      "learning_rate": 5.894913189635375e-05,
      "loss": 11.9218,
      "step": 23712
    },
    {
      "epoch": 1.2912676209729568,
      "grad_norm": 0.5559498739751749,
      "learning_rate": 5.894109106827366e-05,
      "loss": 11.8989,
      "step": 23713
    },
    {
      "epoch": 1.2913220749695398,
      "grad_norm": 0.5551262936257649,
      "learning_rate": 5.8933050559474535e-05,
      "loss": 12.003,
      "step": 23714
    },
    {
      "epoch": 1.2913765289661228,
      "grad_norm": 0.5136157492964187,
      "learning_rate": 5.89250103700189e-05,
      "loss": 12.0017,
      "step": 23715
    },
    {
      "epoch": 1.2914309829627058,
      "grad_norm": 0.531313460374404,
      "learning_rate": 5.891697049996925e-05,
      "loss": 11.9446,
      "step": 23716
    },
    {
      "epoch": 1.291485436959289,
      "grad_norm": 0.48825477336853895,
      "learning_rate": 5.890893094938814e-05,
      "loss": 11.9323,
      "step": 23717
    },
    {
      "epoch": 1.291539890955872,
      "grad_norm": 0.5101215598549473,
      "learning_rate": 5.890089171833808e-05,
      "loss": 11.972,
      "step": 23718
    },
    {
      "epoch": 1.291594344952455,
      "grad_norm": 0.5541828303361535,
      "learning_rate": 5.889285280688156e-05,
      "loss": 11.953,
      "step": 23719
    },
    {
      "epoch": 1.291648798949038,
      "grad_norm": 0.5143235778991876,
      "learning_rate": 5.888481421508112e-05,
      "loss": 11.9354,
      "step": 23720
    },
    {
      "epoch": 1.291703252945621,
      "grad_norm": 0.5063198474722517,
      "learning_rate": 5.887677594299924e-05,
      "loss": 11.9112,
      "step": 23721
    },
    {
      "epoch": 1.291757706942204,
      "grad_norm": 0.5603941868171852,
      "learning_rate": 5.8868737990698465e-05,
      "loss": 11.8923,
      "step": 23722
    },
    {
      "epoch": 1.291812160938787,
      "grad_norm": 0.5081302596193433,
      "learning_rate": 5.8860700358241286e-05,
      "loss": 11.8388,
      "step": 23723
    },
    {
      "epoch": 1.29186661493537,
      "grad_norm": 0.5332331312804888,
      "learning_rate": 5.8852663045690236e-05,
      "loss": 11.9585,
      "step": 23724
    },
    {
      "epoch": 1.291921068931953,
      "grad_norm": 0.6407102059965988,
      "learning_rate": 5.884462605310772e-05,
      "loss": 12.1426,
      "step": 23725
    },
    {
      "epoch": 1.291975522928536,
      "grad_norm": 0.5282749642282629,
      "learning_rate": 5.8836589380556306e-05,
      "loss": 11.9303,
      "step": 23726
    },
    {
      "epoch": 1.292029976925119,
      "grad_norm": 0.5010477551716277,
      "learning_rate": 5.882855302809849e-05,
      "loss": 11.8868,
      "step": 23727
    },
    {
      "epoch": 1.2920844309217019,
      "grad_norm": 0.5263521332847538,
      "learning_rate": 5.8820516995796735e-05,
      "loss": 11.686,
      "step": 23728
    },
    {
      "epoch": 1.2921388849182849,
      "grad_norm": 0.5315598524409277,
      "learning_rate": 5.881248128371355e-05,
      "loss": 11.8769,
      "step": 23729
    },
    {
      "epoch": 1.2921933389148679,
      "grad_norm": 0.5113617335711895,
      "learning_rate": 5.880444589191143e-05,
      "loss": 11.9031,
      "step": 23730
    },
    {
      "epoch": 1.2922477929114509,
      "grad_norm": 0.5513426784972582,
      "learning_rate": 5.879641082045284e-05,
      "loss": 11.9405,
      "step": 23731
    },
    {
      "epoch": 1.292302246908034,
      "grad_norm": 0.5547387914133479,
      "learning_rate": 5.8788376069400244e-05,
      "loss": 11.8762,
      "step": 23732
    },
    {
      "epoch": 1.292356700904617,
      "grad_norm": 0.5883632763509766,
      "learning_rate": 5.878034163881616e-05,
      "loss": 11.9423,
      "step": 23733
    },
    {
      "epoch": 1.2924111549012,
      "grad_norm": 0.5147257702675143,
      "learning_rate": 5.87723075287631e-05,
      "loss": 11.9103,
      "step": 23734
    },
    {
      "epoch": 1.292465608897783,
      "grad_norm": 0.5376391013574864,
      "learning_rate": 5.8764273739303464e-05,
      "loss": 11.7991,
      "step": 23735
    },
    {
      "epoch": 1.292520062894366,
      "grad_norm": 0.547459391107265,
      "learning_rate": 5.8756240270499715e-05,
      "loss": 11.882,
      "step": 23736
    },
    {
      "epoch": 1.292574516890949,
      "grad_norm": 0.5494695018731429,
      "learning_rate": 5.874820712241438e-05,
      "loss": 11.9814,
      "step": 23737
    },
    {
      "epoch": 1.292628970887532,
      "grad_norm": 0.5471057445038715,
      "learning_rate": 5.874017429510992e-05,
      "loss": 11.9549,
      "step": 23738
    },
    {
      "epoch": 1.292683424884115,
      "grad_norm": 0.6807750576173539,
      "learning_rate": 5.873214178864877e-05,
      "loss": 11.8596,
      "step": 23739
    },
    {
      "epoch": 1.292737878880698,
      "grad_norm": 0.5740114418723041,
      "learning_rate": 5.87241096030934e-05,
      "loss": 11.9223,
      "step": 23740
    },
    {
      "epoch": 1.2927923328772812,
      "grad_norm": 0.5288712231651108,
      "learning_rate": 5.8716077738506284e-05,
      "loss": 11.8839,
      "step": 23741
    },
    {
      "epoch": 1.2928467868738642,
      "grad_norm": 0.5266905118122426,
      "learning_rate": 5.870804619494987e-05,
      "loss": 11.9494,
      "step": 23742
    },
    {
      "epoch": 1.2929012408704472,
      "grad_norm": 0.5449159386048019,
      "learning_rate": 5.8700014972486606e-05,
      "loss": 11.8494,
      "step": 23743
    },
    {
      "epoch": 1.2929556948670302,
      "grad_norm": 0.5882369463915113,
      "learning_rate": 5.869198407117897e-05,
      "loss": 11.9339,
      "step": 23744
    },
    {
      "epoch": 1.2930101488636132,
      "grad_norm": 0.5461902077565499,
      "learning_rate": 5.8683953491089395e-05,
      "loss": 11.988,
      "step": 23745
    },
    {
      "epoch": 1.2930646028601962,
      "grad_norm": 0.5408887115112508,
      "learning_rate": 5.867592323228031e-05,
      "loss": 11.9151,
      "step": 23746
    },
    {
      "epoch": 1.2931190568567792,
      "grad_norm": 0.5891116599950841,
      "learning_rate": 5.866789329481418e-05,
      "loss": 11.9295,
      "step": 23747
    },
    {
      "epoch": 1.2931735108533622,
      "grad_norm": 0.5673280691471253,
      "learning_rate": 5.865986367875342e-05,
      "loss": 11.9891,
      "step": 23748
    },
    {
      "epoch": 1.2932279648499452,
      "grad_norm": 0.5540971147823334,
      "learning_rate": 5.865183438416051e-05,
      "loss": 11.9701,
      "step": 23749
    },
    {
      "epoch": 1.2932824188465282,
      "grad_norm": 0.5066621558352484,
      "learning_rate": 5.864380541109788e-05,
      "loss": 11.9131,
      "step": 23750
    },
    {
      "epoch": 1.2933368728431112,
      "grad_norm": 0.5779754619780152,
      "learning_rate": 5.863577675962795e-05,
      "loss": 11.9878,
      "step": 23751
    },
    {
      "epoch": 1.2933913268396942,
      "grad_norm": 0.5818797433262762,
      "learning_rate": 5.862774842981315e-05,
      "loss": 11.9,
      "step": 23752
    },
    {
      "epoch": 1.2934457808362771,
      "grad_norm": 0.5821332362538245,
      "learning_rate": 5.861972042171593e-05,
      "loss": 11.967,
      "step": 23753
    },
    {
      "epoch": 1.2935002348328601,
      "grad_norm": 0.5601375137464693,
      "learning_rate": 5.861169273539871e-05,
      "loss": 11.81,
      "step": 23754
    },
    {
      "epoch": 1.2935546888294434,
      "grad_norm": 0.5738173837500221,
      "learning_rate": 5.86036653709239e-05,
      "loss": 11.8963,
      "step": 23755
    },
    {
      "epoch": 1.2936091428260263,
      "grad_norm": 0.6331948374525299,
      "learning_rate": 5.859563832835393e-05,
      "loss": 11.9184,
      "step": 23756
    },
    {
      "epoch": 1.2936635968226093,
      "grad_norm": 0.5269610859734193,
      "learning_rate": 5.858761160775121e-05,
      "loss": 11.9363,
      "step": 23757
    },
    {
      "epoch": 1.2937180508191923,
      "grad_norm": 0.5604407469730615,
      "learning_rate": 5.8579585209178164e-05,
      "loss": 12.0051,
      "step": 23758
    },
    {
      "epoch": 1.2937725048157753,
      "grad_norm": 0.529801134839837,
      "learning_rate": 5.8571559132697206e-05,
      "loss": 11.9784,
      "step": 23759
    },
    {
      "epoch": 1.2938269588123583,
      "grad_norm": 0.5561562143215054,
      "learning_rate": 5.856353337837076e-05,
      "loss": 11.9694,
      "step": 23760
    },
    {
      "epoch": 1.2938814128089413,
      "grad_norm": 0.4916798146514856,
      "learning_rate": 5.8555507946261235e-05,
      "loss": 11.8015,
      "step": 23761
    },
    {
      "epoch": 1.2939358668055243,
      "grad_norm": 0.5454294032558833,
      "learning_rate": 5.8547482836431014e-05,
      "loss": 11.9613,
      "step": 23762
    },
    {
      "epoch": 1.2939903208021073,
      "grad_norm": 0.566705039358307,
      "learning_rate": 5.853945804894254e-05,
      "loss": 11.8301,
      "step": 23763
    },
    {
      "epoch": 1.2940447747986905,
      "grad_norm": 0.5164305283152058,
      "learning_rate": 5.8531433583858195e-05,
      "loss": 11.8817,
      "step": 23764
    },
    {
      "epoch": 1.2940992287952735,
      "grad_norm": 0.5587180282690472,
      "learning_rate": 5.8523409441240375e-05,
      "loss": 11.9838,
      "step": 23765
    },
    {
      "epoch": 1.2941536827918565,
      "grad_norm": 0.5042120080470417,
      "learning_rate": 5.8515385621151464e-05,
      "loss": 12.0209,
      "step": 23766
    },
    {
      "epoch": 1.2942081367884395,
      "grad_norm": 0.5293438762147187,
      "learning_rate": 5.8507362123653865e-05,
      "loss": 11.926,
      "step": 23767
    },
    {
      "epoch": 1.2942625907850225,
      "grad_norm": 0.5307943430357466,
      "learning_rate": 5.849933894880999e-05,
      "loss": 11.8703,
      "step": 23768
    },
    {
      "epoch": 1.2943170447816055,
      "grad_norm": 0.5962872100297739,
      "learning_rate": 5.8491316096682215e-05,
      "loss": 11.9902,
      "step": 23769
    },
    {
      "epoch": 1.2943714987781885,
      "grad_norm": 0.4928160541912051,
      "learning_rate": 5.848329356733291e-05,
      "loss": 11.823,
      "step": 23770
    },
    {
      "epoch": 1.2944259527747715,
      "grad_norm": 0.47542862553071874,
      "learning_rate": 5.847527136082449e-05,
      "loss": 11.8341,
      "step": 23771
    },
    {
      "epoch": 1.2944804067713545,
      "grad_norm": 0.5182900495334072,
      "learning_rate": 5.8467249477219313e-05,
      "loss": 11.8392,
      "step": 23772
    },
    {
      "epoch": 1.2945348607679374,
      "grad_norm": 0.5317771082088423,
      "learning_rate": 5.8459227916579785e-05,
      "loss": 11.8587,
      "step": 23773
    },
    {
      "epoch": 1.2945893147645204,
      "grad_norm": 0.5421895251981557,
      "learning_rate": 5.8451206678968285e-05,
      "loss": 11.8498,
      "step": 23774
    },
    {
      "epoch": 1.2946437687611034,
      "grad_norm": 0.48099212544232717,
      "learning_rate": 5.844318576444714e-05,
      "loss": 11.7839,
      "step": 23775
    },
    {
      "epoch": 1.2946982227576864,
      "grad_norm": 0.6457133927383553,
      "learning_rate": 5.8435165173078765e-05,
      "loss": 11.9891,
      "step": 23776
    },
    {
      "epoch": 1.2947526767542694,
      "grad_norm": 0.581678108408296,
      "learning_rate": 5.842714490492551e-05,
      "loss": 12.0298,
      "step": 23777
    },
    {
      "epoch": 1.2948071307508524,
      "grad_norm": 0.5619604634688051,
      "learning_rate": 5.8419124960049745e-05,
      "loss": 12.0587,
      "step": 23778
    },
    {
      "epoch": 1.2948615847474356,
      "grad_norm": 0.5902723998282311,
      "learning_rate": 5.841110533851384e-05,
      "loss": 11.9813,
      "step": 23779
    },
    {
      "epoch": 1.2949160387440186,
      "grad_norm": 0.5290632394977075,
      "learning_rate": 5.840308604038015e-05,
      "loss": 11.9145,
      "step": 23780
    },
    {
      "epoch": 1.2949704927406016,
      "grad_norm": 0.5630728018540572,
      "learning_rate": 5.839506706571103e-05,
      "loss": 11.872,
      "step": 23781
    },
    {
      "epoch": 1.2950249467371846,
      "grad_norm": 0.49309045151418357,
      "learning_rate": 5.838704841456886e-05,
      "loss": 11.8864,
      "step": 23782
    },
    {
      "epoch": 1.2950794007337676,
      "grad_norm": 0.5205485808217177,
      "learning_rate": 5.837903008701599e-05,
      "loss": 11.8321,
      "step": 23783
    },
    {
      "epoch": 1.2951338547303506,
      "grad_norm": 0.5734610177021967,
      "learning_rate": 5.8371012083114776e-05,
      "loss": 11.9272,
      "step": 23784
    },
    {
      "epoch": 1.2951883087269336,
      "grad_norm": 0.61766529693796,
      "learning_rate": 5.836299440292751e-05,
      "loss": 12.018,
      "step": 23785
    },
    {
      "epoch": 1.2952427627235166,
      "grad_norm": 0.5309879938683016,
      "learning_rate": 5.835497704651662e-05,
      "loss": 11.8067,
      "step": 23786
    },
    {
      "epoch": 1.2952972167200998,
      "grad_norm": 0.5115278956316054,
      "learning_rate": 5.834696001394437e-05,
      "loss": 11.9181,
      "step": 23787
    },
    {
      "epoch": 1.2953516707166828,
      "grad_norm": 0.5730103687587895,
      "learning_rate": 5.8338943305273184e-05,
      "loss": 12.0184,
      "step": 23788
    },
    {
      "epoch": 1.2954061247132658,
      "grad_norm": 0.5128052120920888,
      "learning_rate": 5.833092692056532e-05,
      "loss": 11.7312,
      "step": 23789
    },
    {
      "epoch": 1.2954605787098488,
      "grad_norm": 0.5364698494166938,
      "learning_rate": 5.832291085988316e-05,
      "loss": 11.9624,
      "step": 23790
    },
    {
      "epoch": 1.2955150327064318,
      "grad_norm": 0.5467671699510431,
      "learning_rate": 5.8314895123289074e-05,
      "loss": 12.015,
      "step": 23791
    },
    {
      "epoch": 1.2955694867030147,
      "grad_norm": 0.5578199902118518,
      "learning_rate": 5.830687971084532e-05,
      "loss": 11.8048,
      "step": 23792
    },
    {
      "epoch": 1.2956239406995977,
      "grad_norm": 0.5275092571653156,
      "learning_rate": 5.829886462261429e-05,
      "loss": 11.8615,
      "step": 23793
    },
    {
      "epoch": 1.2956783946961807,
      "grad_norm": 0.545961655763849,
      "learning_rate": 5.829084985865829e-05,
      "loss": 11.9292,
      "step": 23794
    },
    {
      "epoch": 1.2957328486927637,
      "grad_norm": 0.5126942762884821,
      "learning_rate": 5.828283541903959e-05,
      "loss": 11.8929,
      "step": 23795
    },
    {
      "epoch": 1.2957873026893467,
      "grad_norm": 0.6318212840121846,
      "learning_rate": 5.8274821303820605e-05,
      "loss": 12.0266,
      "step": 23796
    },
    {
      "epoch": 1.2958417566859297,
      "grad_norm": 0.5517809070225351,
      "learning_rate": 5.826680751306357e-05,
      "loss": 11.8974,
      "step": 23797
    },
    {
      "epoch": 1.2958962106825127,
      "grad_norm": 0.5450332140338545,
      "learning_rate": 5.825879404683087e-05,
      "loss": 12.0884,
      "step": 23798
    },
    {
      "epoch": 1.2959506646790957,
      "grad_norm": 0.5179500438583599,
      "learning_rate": 5.825078090518474e-05,
      "loss": 11.829,
      "step": 23799
    },
    {
      "epoch": 1.2960051186756787,
      "grad_norm": 0.4967420919153505,
      "learning_rate": 5.824276808818756e-05,
      "loss": 11.9634,
      "step": 23800
    },
    {
      "epoch": 1.2960595726722617,
      "grad_norm": 0.5318850930840494,
      "learning_rate": 5.8234755595901635e-05,
      "loss": 11.9274,
      "step": 23801
    },
    {
      "epoch": 1.296114026668845,
      "grad_norm": 0.5272746189101266,
      "learning_rate": 5.822674342838923e-05,
      "loss": 11.8998,
      "step": 23802
    },
    {
      "epoch": 1.296168480665428,
      "grad_norm": 0.504146223444168,
      "learning_rate": 5.8218731585712716e-05,
      "loss": 11.9453,
      "step": 23803
    },
    {
      "epoch": 1.2962229346620109,
      "grad_norm": 0.5549635152298085,
      "learning_rate": 5.821072006793434e-05,
      "loss": 12.0533,
      "step": 23804
    },
    {
      "epoch": 1.2962773886585939,
      "grad_norm": 0.5818619237050583,
      "learning_rate": 5.8202708875116385e-05,
      "loss": 11.8858,
      "step": 23805
    },
    {
      "epoch": 1.2963318426551769,
      "grad_norm": 0.6054106256274053,
      "learning_rate": 5.8194698007321205e-05,
      "loss": 11.9465,
      "step": 23806
    },
    {
      "epoch": 1.2963862966517599,
      "grad_norm": 0.5968692808356221,
      "learning_rate": 5.818668746461102e-05,
      "loss": 12.0636,
      "step": 23807
    },
    {
      "epoch": 1.2964407506483429,
      "grad_norm": 0.5005143996104131,
      "learning_rate": 5.817867724704822e-05,
      "loss": 11.8417,
      "step": 23808
    },
    {
      "epoch": 1.2964952046449258,
      "grad_norm": 0.5110394529088859,
      "learning_rate": 5.8170667354694984e-05,
      "loss": 11.887,
      "step": 23809
    },
    {
      "epoch": 1.2965496586415088,
      "grad_norm": 0.5864714130318646,
      "learning_rate": 5.8162657787613715e-05,
      "loss": 11.8843,
      "step": 23810
    },
    {
      "epoch": 1.296604112638092,
      "grad_norm": 0.5275926335063034,
      "learning_rate": 5.815464854586657e-05,
      "loss": 11.9233,
      "step": 23811
    },
    {
      "epoch": 1.296658566634675,
      "grad_norm": 0.5586000244429157,
      "learning_rate": 5.814663962951592e-05,
      "loss": 11.8799,
      "step": 23812
    },
    {
      "epoch": 1.296713020631258,
      "grad_norm": 0.530092297639588,
      "learning_rate": 5.8138631038624036e-05,
      "loss": 11.8257,
      "step": 23813
    },
    {
      "epoch": 1.296767474627841,
      "grad_norm": 0.5328622060712425,
      "learning_rate": 5.813062277325319e-05,
      "loss": 11.7601,
      "step": 23814
    },
    {
      "epoch": 1.296821928624424,
      "grad_norm": 0.5578334577050502,
      "learning_rate": 5.812261483346564e-05,
      "loss": 11.9954,
      "step": 23815
    },
    {
      "epoch": 1.296876382621007,
      "grad_norm": 0.5591524005561552,
      "learning_rate": 5.8114607219323623e-05,
      "loss": 11.8942,
      "step": 23816
    },
    {
      "epoch": 1.29693083661759,
      "grad_norm": 0.5014435898895022,
      "learning_rate": 5.810659993088944e-05,
      "loss": 11.9737,
      "step": 23817
    },
    {
      "epoch": 1.296985290614173,
      "grad_norm": 0.5155678054343346,
      "learning_rate": 5.8098592968225396e-05,
      "loss": 11.8323,
      "step": 23818
    },
    {
      "epoch": 1.297039744610756,
      "grad_norm": 0.5822572248388822,
      "learning_rate": 5.809058633139369e-05,
      "loss": 12.0131,
      "step": 23819
    },
    {
      "epoch": 1.297094198607339,
      "grad_norm": 0.5669873017065186,
      "learning_rate": 5.8082580020456656e-05,
      "loss": 11.9765,
      "step": 23820
    },
    {
      "epoch": 1.297148652603922,
      "grad_norm": 0.565166966704871,
      "learning_rate": 5.807457403547646e-05,
      "loss": 11.9164,
      "step": 23821
    },
    {
      "epoch": 1.297203106600505,
      "grad_norm": 0.6252835077462645,
      "learning_rate": 5.806656837651545e-05,
      "loss": 11.9195,
      "step": 23822
    },
    {
      "epoch": 1.297257560597088,
      "grad_norm": 0.5209635888907312,
      "learning_rate": 5.8058563043635796e-05,
      "loss": 11.868,
      "step": 23823
    },
    {
      "epoch": 1.297312014593671,
      "grad_norm": 0.5397714261569933,
      "learning_rate": 5.8050558036899826e-05,
      "loss": 11.922,
      "step": 23824
    },
    {
      "epoch": 1.2973664685902542,
      "grad_norm": 0.5615187512027605,
      "learning_rate": 5.804255335636976e-05,
      "loss": 12.0561,
      "step": 23825
    },
    {
      "epoch": 1.2974209225868372,
      "grad_norm": 0.5356667127732464,
      "learning_rate": 5.8034549002107786e-05,
      "loss": 11.9817,
      "step": 23826
    },
    {
      "epoch": 1.2974753765834202,
      "grad_norm": 0.5363407231181678,
      "learning_rate": 5.802654497417621e-05,
      "loss": 11.8955,
      "step": 23827
    },
    {
      "epoch": 1.2975298305800032,
      "grad_norm": 0.5562681698490525,
      "learning_rate": 5.8018541272637286e-05,
      "loss": 11.9367,
      "step": 23828
    },
    {
      "epoch": 1.2975842845765861,
      "grad_norm": 0.5507851988821135,
      "learning_rate": 5.8010537897553195e-05,
      "loss": 11.8209,
      "step": 23829
    },
    {
      "epoch": 1.2976387385731691,
      "grad_norm": 0.5723448092638935,
      "learning_rate": 5.800253484898623e-05,
      "loss": 11.934,
      "step": 23830
    },
    {
      "epoch": 1.2976931925697521,
      "grad_norm": 0.5014592242239699,
      "learning_rate": 5.7994532126998566e-05,
      "loss": 12.0612,
      "step": 23831
    },
    {
      "epoch": 1.2977476465663351,
      "grad_norm": 0.4921722432223852,
      "learning_rate": 5.798652973165251e-05,
      "loss": 11.8723,
      "step": 23832
    },
    {
      "epoch": 1.2978021005629181,
      "grad_norm": 0.5835628798465374,
      "learning_rate": 5.7978527663010196e-05,
      "loss": 12.0561,
      "step": 23833
    },
    {
      "epoch": 1.2978565545595013,
      "grad_norm": 0.5568174068152698,
      "learning_rate": 5.797052592113394e-05,
      "loss": 11.8702,
      "step": 23834
    },
    {
      "epoch": 1.2979110085560843,
      "grad_norm": 0.5443639620474775,
      "learning_rate": 5.796252450608592e-05,
      "loss": 11.9998,
      "step": 23835
    },
    {
      "epoch": 1.2979654625526673,
      "grad_norm": 0.5243651156283624,
      "learning_rate": 5.795452341792833e-05,
      "loss": 11.8549,
      "step": 23836
    },
    {
      "epoch": 1.2980199165492503,
      "grad_norm": 0.5552627226889993,
      "learning_rate": 5.7946522656723447e-05,
      "loss": 11.9421,
      "step": 23837
    },
    {
      "epoch": 1.2980743705458333,
      "grad_norm": 0.5577937602476003,
      "learning_rate": 5.793852222253342e-05,
      "loss": 11.9975,
      "step": 23838
    },
    {
      "epoch": 1.2981288245424163,
      "grad_norm": 0.5406367919214983,
      "learning_rate": 5.7930522115420495e-05,
      "loss": 11.9579,
      "step": 23839
    },
    {
      "epoch": 1.2981832785389993,
      "grad_norm": 0.5159964460721704,
      "learning_rate": 5.792252233544693e-05,
      "loss": 11.9685,
      "step": 23840
    },
    {
      "epoch": 1.2982377325355823,
      "grad_norm": 0.5752622332443564,
      "learning_rate": 5.791452288267483e-05,
      "loss": 12.0587,
      "step": 23841
    },
    {
      "epoch": 1.2982921865321653,
      "grad_norm": 0.5143233796983403,
      "learning_rate": 5.790652375716652e-05,
      "loss": 11.828,
      "step": 23842
    },
    {
      "epoch": 1.2983466405287483,
      "grad_norm": 0.5817519099039418,
      "learning_rate": 5.789852495898409e-05,
      "loss": 11.948,
      "step": 23843
    },
    {
      "epoch": 1.2984010945253313,
      "grad_norm": 0.5981241894383442,
      "learning_rate": 5.789052648818982e-05,
      "loss": 12.0364,
      "step": 23844
    },
    {
      "epoch": 1.2984555485219142,
      "grad_norm": 0.55008481899298,
      "learning_rate": 5.78825283448459e-05,
      "loss": 11.9335,
      "step": 23845
    },
    {
      "epoch": 1.2985100025184972,
      "grad_norm": 0.5705241028715168,
      "learning_rate": 5.787453052901446e-05,
      "loss": 11.9181,
      "step": 23846
    },
    {
      "epoch": 1.2985644565150802,
      "grad_norm": 0.5941401412513208,
      "learning_rate": 5.786653304075775e-05,
      "loss": 12.0423,
      "step": 23847
    },
    {
      "epoch": 1.2986189105116632,
      "grad_norm": 0.5608612963629043,
      "learning_rate": 5.785853588013793e-05,
      "loss": 11.9988,
      "step": 23848
    },
    {
      "epoch": 1.2986733645082464,
      "grad_norm": 0.5342991395931518,
      "learning_rate": 5.78505390472172e-05,
      "loss": 11.8275,
      "step": 23849
    },
    {
      "epoch": 1.2987278185048294,
      "grad_norm": 0.592120336158684,
      "learning_rate": 5.784254254205779e-05,
      "loss": 11.9297,
      "step": 23850
    },
    {
      "epoch": 1.2987822725014124,
      "grad_norm": 0.5739279250310847,
      "learning_rate": 5.7834546364721785e-05,
      "loss": 12.0212,
      "step": 23851
    },
    {
      "epoch": 1.2988367264979954,
      "grad_norm": 0.5394185905729064,
      "learning_rate": 5.782655051527146e-05,
      "loss": 11.9065,
      "step": 23852
    },
    {
      "epoch": 1.2988911804945784,
      "grad_norm": 0.5220110030213705,
      "learning_rate": 5.781855499376891e-05,
      "loss": 11.8859,
      "step": 23853
    },
    {
      "epoch": 1.2989456344911614,
      "grad_norm": 0.7403391556608028,
      "learning_rate": 5.7810559800276396e-05,
      "loss": 11.9985,
      "step": 23854
    },
    {
      "epoch": 1.2990000884877444,
      "grad_norm": 0.5237367908716927,
      "learning_rate": 5.780256493485605e-05,
      "loss": 11.9219,
      "step": 23855
    },
    {
      "epoch": 1.2990545424843274,
      "grad_norm": 0.5028134170372256,
      "learning_rate": 5.779457039756999e-05,
      "loss": 11.9866,
      "step": 23856
    },
    {
      "epoch": 1.2991089964809106,
      "grad_norm": 0.5407719520119866,
      "learning_rate": 5.7786576188480465e-05,
      "loss": 11.8646,
      "step": 23857
    },
    {
      "epoch": 1.2991634504774936,
      "grad_norm": 0.5432444500769634,
      "learning_rate": 5.777858230764956e-05,
      "loss": 11.799,
      "step": 23858
    },
    {
      "epoch": 1.2992179044740766,
      "grad_norm": 0.556584499302641,
      "learning_rate": 5.777058875513952e-05,
      "loss": 11.9791,
      "step": 23859
    },
    {
      "epoch": 1.2992723584706596,
      "grad_norm": 0.5299547991900019,
      "learning_rate": 5.7762595531012444e-05,
      "loss": 11.9898,
      "step": 23860
    },
    {
      "epoch": 1.2993268124672426,
      "grad_norm": 0.5567155647967426,
      "learning_rate": 5.775460263533048e-05,
      "loss": 11.829,
      "step": 23861
    },
    {
      "epoch": 1.2993812664638256,
      "grad_norm": 0.5485088855415148,
      "learning_rate": 5.774661006815587e-05,
      "loss": 12.0006,
      "step": 23862
    },
    {
      "epoch": 1.2994357204604086,
      "grad_norm": 0.6038910920864603,
      "learning_rate": 5.773861782955066e-05,
      "loss": 11.9628,
      "step": 23863
    },
    {
      "epoch": 1.2994901744569916,
      "grad_norm": 0.5907919696422618,
      "learning_rate": 5.773062591957708e-05,
      "loss": 12.1342,
      "step": 23864
    },
    {
      "epoch": 1.2995446284535745,
      "grad_norm": 0.56804388009057,
      "learning_rate": 5.7722634338297256e-05,
      "loss": 11.9145,
      "step": 23865
    },
    {
      "epoch": 1.2995990824501575,
      "grad_norm": 0.5934006202192055,
      "learning_rate": 5.7714643085773256e-05,
      "loss": 11.9192,
      "step": 23866
    },
    {
      "epoch": 1.2996535364467405,
      "grad_norm": 0.521596306869903,
      "learning_rate": 5.770665216206734e-05,
      "loss": 11.9393,
      "step": 23867
    },
    {
      "epoch": 1.2997079904433235,
      "grad_norm": 0.5476409909677536,
      "learning_rate": 5.769866156724154e-05,
      "loss": 11.8477,
      "step": 23868
    },
    {
      "epoch": 1.2997624444399065,
      "grad_norm": 0.5489940155779739,
      "learning_rate": 5.769067130135808e-05,
      "loss": 11.8936,
      "step": 23869
    },
    {
      "epoch": 1.2998168984364895,
      "grad_norm": 0.5687623535481318,
      "learning_rate": 5.7682681364479027e-05,
      "loss": 11.9452,
      "step": 23870
    },
    {
      "epoch": 1.2998713524330725,
      "grad_norm": 0.49442109312059335,
      "learning_rate": 5.7674691756666574e-05,
      "loss": 11.8604,
      "step": 23871
    },
    {
      "epoch": 1.2999258064296557,
      "grad_norm": 0.5268642892099454,
      "learning_rate": 5.766670247798277e-05,
      "loss": 11.9767,
      "step": 23872
    },
    {
      "epoch": 1.2999802604262387,
      "grad_norm": 0.5900355269341435,
      "learning_rate": 5.7658713528489795e-05,
      "loss": 11.9823,
      "step": 23873
    },
    {
      "epoch": 1.3000347144228217,
      "grad_norm": 0.5524735410150464,
      "learning_rate": 5.76507249082498e-05,
      "loss": 12.0053,
      "step": 23874
    },
    {
      "epoch": 1.3000891684194047,
      "grad_norm": 0.5957451169779749,
      "learning_rate": 5.7642736617324864e-05,
      "loss": 11.9088,
      "step": 23875
    },
    {
      "epoch": 1.3001436224159877,
      "grad_norm": 0.5603659799437561,
      "learning_rate": 5.763474865577708e-05,
      "loss": 11.9087,
      "step": 23876
    },
    {
      "epoch": 1.3001980764125707,
      "grad_norm": 0.5687007741173038,
      "learning_rate": 5.7626761023668644e-05,
      "loss": 11.9719,
      "step": 23877
    },
    {
      "epoch": 1.3002525304091537,
      "grad_norm": 0.523834443595407,
      "learning_rate": 5.761877372106157e-05,
      "loss": 11.9838,
      "step": 23878
    },
    {
      "epoch": 1.3003069844057367,
      "grad_norm": 0.5162018241912517,
      "learning_rate": 5.761078674801805e-05,
      "loss": 11.8398,
      "step": 23879
    },
    {
      "epoch": 1.3003614384023197,
      "grad_norm": 0.5216713171190042,
      "learning_rate": 5.760280010460013e-05,
      "loss": 12.0186,
      "step": 23880
    },
    {
      "epoch": 1.3004158923989029,
      "grad_norm": 0.5385634257076888,
      "learning_rate": 5.759481379086998e-05,
      "loss": 11.9314,
      "step": 23881
    },
    {
      "epoch": 1.3004703463954859,
      "grad_norm": 0.5015222462913387,
      "learning_rate": 5.758682780688964e-05,
      "loss": 11.8739,
      "step": 23882
    },
    {
      "epoch": 1.3005248003920689,
      "grad_norm": 0.5829663844393825,
      "learning_rate": 5.757884215272123e-05,
      "loss": 11.9699,
      "step": 23883
    },
    {
      "epoch": 1.3005792543886519,
      "grad_norm": 0.5987181288521144,
      "learning_rate": 5.7570856828426914e-05,
      "loss": 11.7919,
      "step": 23884
    },
    {
      "epoch": 1.3006337083852348,
      "grad_norm": 0.5477114358282926,
      "learning_rate": 5.756287183406872e-05,
      "loss": 11.9544,
      "step": 23885
    },
    {
      "epoch": 1.3006881623818178,
      "grad_norm": 0.5637247972403945,
      "learning_rate": 5.755488716970874e-05,
      "loss": 11.9266,
      "step": 23886
    },
    {
      "epoch": 1.3007426163784008,
      "grad_norm": 0.5424718020020158,
      "learning_rate": 5.754690283540904e-05,
      "loss": 11.864,
      "step": 23887
    },
    {
      "epoch": 1.3007970703749838,
      "grad_norm": 0.5247746464249025,
      "learning_rate": 5.753891883123176e-05,
      "loss": 11.9365,
      "step": 23888
    },
    {
      "epoch": 1.3008515243715668,
      "grad_norm": 0.5233613628141781,
      "learning_rate": 5.7530935157239e-05,
      "loss": 12.0639,
      "step": 23889
    },
    {
      "epoch": 1.3009059783681498,
      "grad_norm": 0.5323720067503869,
      "learning_rate": 5.7522951813492764e-05,
      "loss": 11.9468,
      "step": 23890
    },
    {
      "epoch": 1.3009604323647328,
      "grad_norm": 0.481733645939565,
      "learning_rate": 5.7514968800055226e-05,
      "loss": 11.8236,
      "step": 23891
    },
    {
      "epoch": 1.3010148863613158,
      "grad_norm": 0.5137549636562518,
      "learning_rate": 5.750698611698837e-05,
      "loss": 11.92,
      "step": 23892
    },
    {
      "epoch": 1.3010693403578988,
      "grad_norm": 0.5909328738816363,
      "learning_rate": 5.7499003764354354e-05,
      "loss": 11.988,
      "step": 23893
    },
    {
      "epoch": 1.3011237943544818,
      "grad_norm": 0.576355762077326,
      "learning_rate": 5.749102174221517e-05,
      "loss": 11.999,
      "step": 23894
    },
    {
      "epoch": 1.301178248351065,
      "grad_norm": 0.5370711057197125,
      "learning_rate": 5.7483040050632974e-05,
      "loss": 11.8079,
      "step": 23895
    },
    {
      "epoch": 1.301232702347648,
      "grad_norm": 0.5246186221614892,
      "learning_rate": 5.747505868966979e-05,
      "loss": 11.9394,
      "step": 23896
    },
    {
      "epoch": 1.301287156344231,
      "grad_norm": 0.5614223842972061,
      "learning_rate": 5.746707765938764e-05,
      "loss": 11.8673,
      "step": 23897
    },
    {
      "epoch": 1.301341610340814,
      "grad_norm": 0.5842289670745541,
      "learning_rate": 5.745909695984864e-05,
      "loss": 11.9609,
      "step": 23898
    },
    {
      "epoch": 1.301396064337397,
      "grad_norm": 0.5484661025083538,
      "learning_rate": 5.745111659111485e-05,
      "loss": 11.8219,
      "step": 23899
    },
    {
      "epoch": 1.30145051833398,
      "grad_norm": 0.485387211382524,
      "learning_rate": 5.744313655324829e-05,
      "loss": 11.9354,
      "step": 23900
    },
    {
      "epoch": 1.301504972330563,
      "grad_norm": 0.5524606598815619,
      "learning_rate": 5.743515684631107e-05,
      "loss": 11.8696,
      "step": 23901
    },
    {
      "epoch": 1.301559426327146,
      "grad_norm": 0.563018059080123,
      "learning_rate": 5.742717747036517e-05,
      "loss": 12.0242,
      "step": 23902
    },
    {
      "epoch": 1.301613880323729,
      "grad_norm": 0.5186934091779591,
      "learning_rate": 5.741919842547272e-05,
      "loss": 11.9867,
      "step": 23903
    },
    {
      "epoch": 1.3016683343203121,
      "grad_norm": 0.5546046245961251,
      "learning_rate": 5.741121971169567e-05,
      "loss": 11.9718,
      "step": 23904
    },
    {
      "epoch": 1.3017227883168951,
      "grad_norm": 0.5722461246795527,
      "learning_rate": 5.740324132909618e-05,
      "loss": 11.9853,
      "step": 23905
    },
    {
      "epoch": 1.3017772423134781,
      "grad_norm": 0.56459150025772,
      "learning_rate": 5.739526327773622e-05,
      "loss": 12.0055,
      "step": 23906
    },
    {
      "epoch": 1.3018316963100611,
      "grad_norm": 0.5408282463553108,
      "learning_rate": 5.73872855576778e-05,
      "loss": 11.9205,
      "step": 23907
    },
    {
      "epoch": 1.3018861503066441,
      "grad_norm": 0.5474122335268758,
      "learning_rate": 5.737930816898303e-05,
      "loss": 12.0301,
      "step": 23908
    },
    {
      "epoch": 1.3019406043032271,
      "grad_norm": 0.5187639001840174,
      "learning_rate": 5.737133111171386e-05,
      "loss": 11.9641,
      "step": 23909
    },
    {
      "epoch": 1.30199505829981,
      "grad_norm": 0.6272055705797348,
      "learning_rate": 5.736335438593239e-05,
      "loss": 12.128,
      "step": 23910
    },
    {
      "epoch": 1.302049512296393,
      "grad_norm": 0.5272254647433033,
      "learning_rate": 5.7355377991700645e-05,
      "loss": 11.9563,
      "step": 23911
    },
    {
      "epoch": 1.302103966292976,
      "grad_norm": 0.5549107036377859,
      "learning_rate": 5.734740192908059e-05,
      "loss": 11.8946,
      "step": 23912
    },
    {
      "epoch": 1.302158420289559,
      "grad_norm": 0.5343595563170288,
      "learning_rate": 5.733942619813435e-05,
      "loss": 11.9768,
      "step": 23913
    },
    {
      "epoch": 1.302212874286142,
      "grad_norm": 0.49984534310555945,
      "learning_rate": 5.7331450798923835e-05,
      "loss": 11.8715,
      "step": 23914
    },
    {
      "epoch": 1.302267328282725,
      "grad_norm": 0.5395153645029583,
      "learning_rate": 5.732347573151116e-05,
      "loss": 11.958,
      "step": 23915
    },
    {
      "epoch": 1.302321782279308,
      "grad_norm": 0.48031310327344234,
      "learning_rate": 5.7315500995958284e-05,
      "loss": 11.91,
      "step": 23916
    },
    {
      "epoch": 1.302376236275891,
      "grad_norm": 0.4942641907324273,
      "learning_rate": 5.73075265923272e-05,
      "loss": 11.8832,
      "step": 23917
    },
    {
      "epoch": 1.3024306902724743,
      "grad_norm": 0.5386913802218604,
      "learning_rate": 5.729955252067998e-05,
      "loss": 11.9891,
      "step": 23918
    },
    {
      "epoch": 1.3024851442690573,
      "grad_norm": 0.5376796334453726,
      "learning_rate": 5.7291578781078556e-05,
      "loss": 11.9751,
      "step": 23919
    },
    {
      "epoch": 1.3025395982656403,
      "grad_norm": 0.5565333856111213,
      "learning_rate": 5.7283605373585035e-05,
      "loss": 11.8746,
      "step": 23920
    },
    {
      "epoch": 1.3025940522622232,
      "grad_norm": 0.5851019026703895,
      "learning_rate": 5.727563229826132e-05,
      "loss": 11.941,
      "step": 23921
    },
    {
      "epoch": 1.3026485062588062,
      "grad_norm": 0.4855587166916286,
      "learning_rate": 5.726765955516944e-05,
      "loss": 11.9242,
      "step": 23922
    },
    {
      "epoch": 1.3027029602553892,
      "grad_norm": 0.5296173452222003,
      "learning_rate": 5.7259687144371454e-05,
      "loss": 11.8905,
      "step": 23923
    },
    {
      "epoch": 1.3027574142519722,
      "grad_norm": 0.581536295566947,
      "learning_rate": 5.725171506592927e-05,
      "loss": 11.7452,
      "step": 23924
    },
    {
      "epoch": 1.3028118682485552,
      "grad_norm": 0.5704231691825496,
      "learning_rate": 5.724374331990495e-05,
      "loss": 11.8212,
      "step": 23925
    },
    {
      "epoch": 1.3028663222451382,
      "grad_norm": 0.6077926864510211,
      "learning_rate": 5.723577190636047e-05,
      "loss": 11.9739,
      "step": 23926
    },
    {
      "epoch": 1.3029207762417214,
      "grad_norm": 0.539981475425696,
      "learning_rate": 5.722780082535775e-05,
      "loss": 11.9746,
      "step": 23927
    },
    {
      "epoch": 1.3029752302383044,
      "grad_norm": 0.5626468766958902,
      "learning_rate": 5.721983007695886e-05,
      "loss": 11.9676,
      "step": 23928
    },
    {
      "epoch": 1.3030296842348874,
      "grad_norm": 0.5107973326796912,
      "learning_rate": 5.72118596612257e-05,
      "loss": 11.9145,
      "step": 23929
    },
    {
      "epoch": 1.3030841382314704,
      "grad_norm": 0.5337323240886189,
      "learning_rate": 5.720388957822034e-05,
      "loss": 11.822,
      "step": 23930
    },
    {
      "epoch": 1.3031385922280534,
      "grad_norm": 0.6045486064607628,
      "learning_rate": 5.7195919828004675e-05,
      "loss": 11.9822,
      "step": 23931
    },
    {
      "epoch": 1.3031930462246364,
      "grad_norm": 0.5117673129782551,
      "learning_rate": 5.718795041064071e-05,
      "loss": 11.9986,
      "step": 23932
    },
    {
      "epoch": 1.3032475002212194,
      "grad_norm": 0.5175239048840924,
      "learning_rate": 5.717998132619047e-05,
      "loss": 11.9173,
      "step": 23933
    },
    {
      "epoch": 1.3033019542178024,
      "grad_norm": 0.5182460916005857,
      "learning_rate": 5.7172012574715826e-05,
      "loss": 11.8938,
      "step": 23934
    },
    {
      "epoch": 1.3033564082143854,
      "grad_norm": 0.5171902548628802,
      "learning_rate": 5.716404415627887e-05,
      "loss": 11.8766,
      "step": 23935
    },
    {
      "epoch": 1.3034108622109684,
      "grad_norm": 0.6326446928776895,
      "learning_rate": 5.715607607094142e-05,
      "loss": 11.9336,
      "step": 23936
    },
    {
      "epoch": 1.3034653162075513,
      "grad_norm": 0.5356317795730359,
      "learning_rate": 5.714810831876551e-05,
      "loss": 11.9468,
      "step": 23937
    },
    {
      "epoch": 1.3035197702041343,
      "grad_norm": 0.5355944319998637,
      "learning_rate": 5.7140140899813144e-05,
      "loss": 11.8743,
      "step": 23938
    },
    {
      "epoch": 1.3035742242007173,
      "grad_norm": 0.576354179545349,
      "learning_rate": 5.7132173814146176e-05,
      "loss": 11.9397,
      "step": 23939
    },
    {
      "epoch": 1.3036286781973003,
      "grad_norm": 0.7450549003939333,
      "learning_rate": 5.712420706182666e-05,
      "loss": 12.009,
      "step": 23940
    },
    {
      "epoch": 1.3036831321938833,
      "grad_norm": 0.8013479299433881,
      "learning_rate": 5.7116240642916455e-05,
      "loss": 12.0488,
      "step": 23941
    },
    {
      "epoch": 1.3037375861904665,
      "grad_norm": 0.5770703440886297,
      "learning_rate": 5.7108274557477606e-05,
      "loss": 11.8351,
      "step": 23942
    },
    {
      "epoch": 1.3037920401870495,
      "grad_norm": 0.5448431251972765,
      "learning_rate": 5.7100308805571946e-05,
      "loss": 11.942,
      "step": 23943
    },
    {
      "epoch": 1.3038464941836325,
      "grad_norm": 0.56651186123102,
      "learning_rate": 5.709234338726149e-05,
      "loss": 11.9969,
      "step": 23944
    },
    {
      "epoch": 1.3039009481802155,
      "grad_norm": 0.5477211928099325,
      "learning_rate": 5.7084378302608245e-05,
      "loss": 11.869,
      "step": 23945
    },
    {
      "epoch": 1.3039554021767985,
      "grad_norm": 0.5792967807923594,
      "learning_rate": 5.707641355167399e-05,
      "loss": 11.993,
      "step": 23946
    },
    {
      "epoch": 1.3040098561733815,
      "grad_norm": 0.6211499004231313,
      "learning_rate": 5.706844913452075e-05,
      "loss": 11.8483,
      "step": 23947
    },
    {
      "epoch": 1.3040643101699645,
      "grad_norm": 0.5835205538992685,
      "learning_rate": 5.7060485051210486e-05,
      "loss": 11.8844,
      "step": 23948
    },
    {
      "epoch": 1.3041187641665475,
      "grad_norm": 0.5406301439325446,
      "learning_rate": 5.705252130180505e-05,
      "loss": 11.8945,
      "step": 23949
    },
    {
      "epoch": 1.3041732181631307,
      "grad_norm": 0.510877086515905,
      "learning_rate": 5.7044557886366455e-05,
      "loss": 11.9438,
      "step": 23950
    },
    {
      "epoch": 1.3042276721597137,
      "grad_norm": 0.6818922851058039,
      "learning_rate": 5.703659480495653e-05,
      "loss": 12.0,
      "step": 23951
    },
    {
      "epoch": 1.3042821261562967,
      "grad_norm": 0.5291035509334728,
      "learning_rate": 5.702863205763731e-05,
      "loss": 11.9178,
      "step": 23952
    },
    {
      "epoch": 1.3043365801528797,
      "grad_norm": 0.6095401079017516,
      "learning_rate": 5.702066964447059e-05,
      "loss": 11.9598,
      "step": 23953
    },
    {
      "epoch": 1.3043910341494627,
      "grad_norm": 0.5731255467056443,
      "learning_rate": 5.7012707565518365e-05,
      "loss": 11.964,
      "step": 23954
    },
    {
      "epoch": 1.3044454881460457,
      "grad_norm": 0.5698033834226252,
      "learning_rate": 5.700474582084261e-05,
      "loss": 11.9513,
      "step": 23955
    },
    {
      "epoch": 1.3044999421426287,
      "grad_norm": 0.5418096243886533,
      "learning_rate": 5.699678441050508e-05,
      "loss": 11.9278,
      "step": 23956
    },
    {
      "epoch": 1.3045543961392116,
      "grad_norm": 0.5285863962434048,
      "learning_rate": 5.698882333456781e-05,
      "loss": 11.8176,
      "step": 23957
    },
    {
      "epoch": 1.3046088501357946,
      "grad_norm": 0.5437655748493991,
      "learning_rate": 5.698086259309261e-05,
      "loss": 11.8714,
      "step": 23958
    },
    {
      "epoch": 1.3046633041323776,
      "grad_norm": 0.5249223565079,
      "learning_rate": 5.697290218614145e-05,
      "loss": 11.9625,
      "step": 23959
    },
    {
      "epoch": 1.3047177581289606,
      "grad_norm": 0.5897176494325972,
      "learning_rate": 5.696494211377625e-05,
      "loss": 11.9968,
      "step": 23960
    },
    {
      "epoch": 1.3047722121255436,
      "grad_norm": 0.5744789564895469,
      "learning_rate": 5.695698237605885e-05,
      "loss": 11.9869,
      "step": 23961
    },
    {
      "epoch": 1.3048266661221266,
      "grad_norm": 0.5416320966784032,
      "learning_rate": 5.69490229730512e-05,
      "loss": 11.9474,
      "step": 23962
    },
    {
      "epoch": 1.3048811201187096,
      "grad_norm": 0.5158342676255641,
      "learning_rate": 5.694106390481514e-05,
      "loss": 11.7993,
      "step": 23963
    },
    {
      "epoch": 1.3049355741152926,
      "grad_norm": 0.5253323981802934,
      "learning_rate": 5.693310517141261e-05,
      "loss": 12.033,
      "step": 23964
    },
    {
      "epoch": 1.3049900281118758,
      "grad_norm": 0.5401077524287262,
      "learning_rate": 5.69251467729055e-05,
      "loss": 12.0019,
      "step": 23965
    },
    {
      "epoch": 1.3050444821084588,
      "grad_norm": 0.575819176008194,
      "learning_rate": 5.691718870935562e-05,
      "loss": 11.9024,
      "step": 23966
    },
    {
      "epoch": 1.3050989361050418,
      "grad_norm": 0.5009735122171262,
      "learning_rate": 5.690923098082495e-05,
      "loss": 12.007,
      "step": 23967
    },
    {
      "epoch": 1.3051533901016248,
      "grad_norm": 0.5469346786144591,
      "learning_rate": 5.690127358737527e-05,
      "loss": 12.0547,
      "step": 23968
    },
    {
      "epoch": 1.3052078440982078,
      "grad_norm": 0.6110871644336969,
      "learning_rate": 5.689331652906857e-05,
      "loss": 11.9572,
      "step": 23969
    },
    {
      "epoch": 1.3052622980947908,
      "grad_norm": 0.4965294515940343,
      "learning_rate": 5.688535980596663e-05,
      "loss": 11.8601,
      "step": 23970
    },
    {
      "epoch": 1.3053167520913738,
      "grad_norm": 0.5301775928037475,
      "learning_rate": 5.687740341813135e-05,
      "loss": 11.8822,
      "step": 23971
    },
    {
      "epoch": 1.3053712060879568,
      "grad_norm": 0.9010601067024929,
      "learning_rate": 5.6869447365624675e-05,
      "loss": 11.9948,
      "step": 23972
    },
    {
      "epoch": 1.3054256600845398,
      "grad_norm": 0.5239675187616208,
      "learning_rate": 5.6861491648508346e-05,
      "loss": 11.7696,
      "step": 23973
    },
    {
      "epoch": 1.305480114081123,
      "grad_norm": 0.6069009527012706,
      "learning_rate": 5.6853536266844346e-05,
      "loss": 12.0514,
      "step": 23974
    },
    {
      "epoch": 1.305534568077706,
      "grad_norm": 0.5187344931404261,
      "learning_rate": 5.684558122069448e-05,
      "loss": 11.8903,
      "step": 23975
    },
    {
      "epoch": 1.305589022074289,
      "grad_norm": 0.545560513923747,
      "learning_rate": 5.683762651012057e-05,
      "loss": 11.8777,
      "step": 23976
    },
    {
      "epoch": 1.305643476070872,
      "grad_norm": 0.5995544898189384,
      "learning_rate": 5.682967213518455e-05,
      "loss": 11.9604,
      "step": 23977
    },
    {
      "epoch": 1.305697930067455,
      "grad_norm": 0.5321000437618817,
      "learning_rate": 5.682171809594821e-05,
      "loss": 11.8635,
      "step": 23978
    },
    {
      "epoch": 1.305752384064038,
      "grad_norm": 0.5049934564198645,
      "learning_rate": 5.6813764392473464e-05,
      "loss": 12.0295,
      "step": 23979
    },
    {
      "epoch": 1.305806838060621,
      "grad_norm": 0.4867020493912554,
      "learning_rate": 5.68058110248221e-05,
      "loss": 11.889,
      "step": 23980
    },
    {
      "epoch": 1.305861292057204,
      "grad_norm": 0.5488352103955727,
      "learning_rate": 5.679785799305597e-05,
      "loss": 11.9764,
      "step": 23981
    },
    {
      "epoch": 1.305915746053787,
      "grad_norm": 0.5984157365757806,
      "learning_rate": 5.6789905297237e-05,
      "loss": 12.0128,
      "step": 23982
    },
    {
      "epoch": 1.30597020005037,
      "grad_norm": 0.5676370800152911,
      "learning_rate": 5.678195293742693e-05,
      "loss": 11.9651,
      "step": 23983
    },
    {
      "epoch": 1.306024654046953,
      "grad_norm": 0.5449524338822279,
      "learning_rate": 5.677400091368767e-05,
      "loss": 11.8785,
      "step": 23984
    },
    {
      "epoch": 1.3060791080435359,
      "grad_norm": 0.5984049980477618,
      "learning_rate": 5.6766049226081e-05,
      "loss": 11.8982,
      "step": 23985
    },
    {
      "epoch": 1.3061335620401189,
      "grad_norm": 0.5300111201265589,
      "learning_rate": 5.675809787466883e-05,
      "loss": 11.7741,
      "step": 23986
    },
    {
      "epoch": 1.3061880160367019,
      "grad_norm": 0.4962737380666011,
      "learning_rate": 5.675014685951292e-05,
      "loss": 11.8057,
      "step": 23987
    },
    {
      "epoch": 1.306242470033285,
      "grad_norm": 0.6006448553922471,
      "learning_rate": 5.67421961806751e-05,
      "loss": 12.1342,
      "step": 23988
    },
    {
      "epoch": 1.306296924029868,
      "grad_norm": 0.5084978759646408,
      "learning_rate": 5.673424583821725e-05,
      "loss": 11.9022,
      "step": 23989
    },
    {
      "epoch": 1.306351378026451,
      "grad_norm": 0.5722836251395635,
      "learning_rate": 5.6726295832201115e-05,
      "loss": 11.922,
      "step": 23990
    },
    {
      "epoch": 1.306405832023034,
      "grad_norm": 0.5830201612688393,
      "learning_rate": 5.671834616268861e-05,
      "loss": 11.972,
      "step": 23991
    },
    {
      "epoch": 1.306460286019617,
      "grad_norm": 0.5242589673921513,
      "learning_rate": 5.6710396829741455e-05,
      "loss": 11.9899,
      "step": 23992
    },
    {
      "epoch": 1.3065147400162,
      "grad_norm": 0.5522634693503423,
      "learning_rate": 5.670244783342151e-05,
      "loss": 11.9227,
      "step": 23993
    },
    {
      "epoch": 1.306569194012783,
      "grad_norm": 0.5975844373618137,
      "learning_rate": 5.669449917379064e-05,
      "loss": 11.9588,
      "step": 23994
    },
    {
      "epoch": 1.306623648009366,
      "grad_norm": 0.5658906806418614,
      "learning_rate": 5.6686550850910566e-05,
      "loss": 11.9769,
      "step": 23995
    },
    {
      "epoch": 1.306678102005949,
      "grad_norm": 0.576029591485249,
      "learning_rate": 5.6678602864843186e-05,
      "loss": 11.9018,
      "step": 23996
    },
    {
      "epoch": 1.3067325560025322,
      "grad_norm": 0.5397198944961555,
      "learning_rate": 5.667065521565025e-05,
      "loss": 11.9782,
      "step": 23997
    },
    {
      "epoch": 1.3067870099991152,
      "grad_norm": 0.5318343941425525,
      "learning_rate": 5.666270790339352e-05,
      "loss": 11.8726,
      "step": 23998
    },
    {
      "epoch": 1.3068414639956982,
      "grad_norm": 0.4897478737466349,
      "learning_rate": 5.665476092813488e-05,
      "loss": 11.8749,
      "step": 23999
    },
    {
      "epoch": 1.3068959179922812,
      "grad_norm": 0.5184081515081808,
      "learning_rate": 5.664681428993604e-05,
      "loss": 11.8335,
      "step": 24000
    },
    {
      "epoch": 1.3069503719888642,
      "grad_norm": 0.5645208913521276,
      "learning_rate": 5.663886798885889e-05,
      "loss": 11.8736,
      "step": 24001
    },
    {
      "epoch": 1.3070048259854472,
      "grad_norm": 0.5654316704582217,
      "learning_rate": 5.663092202496514e-05,
      "loss": 11.983,
      "step": 24002
    },
    {
      "epoch": 1.3070592799820302,
      "grad_norm": 0.5317254218554905,
      "learning_rate": 5.662297639831662e-05,
      "loss": 11.9351,
      "step": 24003
    },
    {
      "epoch": 1.3071137339786132,
      "grad_norm": 0.5035033916689757,
      "learning_rate": 5.661503110897512e-05,
      "loss": 11.8796,
      "step": 24004
    },
    {
      "epoch": 1.3071681879751962,
      "grad_norm": 0.5551355566986512,
      "learning_rate": 5.660708615700241e-05,
      "loss": 11.9708,
      "step": 24005
    },
    {
      "epoch": 1.3072226419717792,
      "grad_norm": 0.5316190494277161,
      "learning_rate": 5.6599141542460334e-05,
      "loss": 11.9273,
      "step": 24006
    },
    {
      "epoch": 1.3072770959683622,
      "grad_norm": 0.5489959422403061,
      "learning_rate": 5.659119726541054e-05,
      "loss": 11.9331,
      "step": 24007
    },
    {
      "epoch": 1.3073315499649452,
      "grad_norm": 0.5314437492337467,
      "learning_rate": 5.658325332591486e-05,
      "loss": 11.9833,
      "step": 24008
    },
    {
      "epoch": 1.3073860039615282,
      "grad_norm": 0.5465914577428189,
      "learning_rate": 5.657530972403514e-05,
      "loss": 11.7488,
      "step": 24009
    },
    {
      "epoch": 1.3074404579581111,
      "grad_norm": 0.5306735013351034,
      "learning_rate": 5.656736645983305e-05,
      "loss": 11.9382,
      "step": 24010
    },
    {
      "epoch": 1.3074949119546941,
      "grad_norm": 0.5726449187442675,
      "learning_rate": 5.6559423533370435e-05,
      "loss": 11.819,
      "step": 24011
    },
    {
      "epoch": 1.3075493659512774,
      "grad_norm": 0.5486158402846787,
      "learning_rate": 5.655148094470899e-05,
      "loss": 11.8961,
      "step": 24012
    },
    {
      "epoch": 1.3076038199478603,
      "grad_norm": 0.5426016652965927,
      "learning_rate": 5.654353869391057e-05,
      "loss": 11.9445,
      "step": 24013
    },
    {
      "epoch": 1.3076582739444433,
      "grad_norm": 0.5245954095774398,
      "learning_rate": 5.6535596781036826e-05,
      "loss": 11.8985,
      "step": 24014
    },
    {
      "epoch": 1.3077127279410263,
      "grad_norm": 0.5068428886140711,
      "learning_rate": 5.652765520614958e-05,
      "loss": 12.0234,
      "step": 24015
    },
    {
      "epoch": 1.3077671819376093,
      "grad_norm": 0.5968747370915668,
      "learning_rate": 5.651971396931064e-05,
      "loss": 11.9745,
      "step": 24016
    },
    {
      "epoch": 1.3078216359341923,
      "grad_norm": 0.5904562558238576,
      "learning_rate": 5.651177307058163e-05,
      "loss": 11.829,
      "step": 24017
    },
    {
      "epoch": 1.3078760899307753,
      "grad_norm": 0.6144528212683565,
      "learning_rate": 5.650383251002437e-05,
      "loss": 12.0584,
      "step": 24018
    },
    {
      "epoch": 1.3079305439273583,
      "grad_norm": 0.5637354895837392,
      "learning_rate": 5.649589228770062e-05,
      "loss": 12.0088,
      "step": 24019
    },
    {
      "epoch": 1.3079849979239415,
      "grad_norm": 0.5174112427094429,
      "learning_rate": 5.64879524036721e-05,
      "loss": 11.859,
      "step": 24020
    },
    {
      "epoch": 1.3080394519205245,
      "grad_norm": 0.5722180502695624,
      "learning_rate": 5.6480012858000575e-05,
      "loss": 11.8917,
      "step": 24021
    },
    {
      "epoch": 1.3080939059171075,
      "grad_norm": 0.5278915893715153,
      "learning_rate": 5.6472073650747736e-05,
      "loss": 11.841,
      "step": 24022
    },
    {
      "epoch": 1.3081483599136905,
      "grad_norm": 0.5434381151015685,
      "learning_rate": 5.646413478197539e-05,
      "loss": 12.0007,
      "step": 24023
    },
    {
      "epoch": 1.3082028139102735,
      "grad_norm": 0.57750000746015,
      "learning_rate": 5.64561962517452e-05,
      "loss": 11.8534,
      "step": 24024
    },
    {
      "epoch": 1.3082572679068565,
      "grad_norm": 0.5459021214604409,
      "learning_rate": 5.6448258060118954e-05,
      "loss": 11.9986,
      "step": 24025
    },
    {
      "epoch": 1.3083117219034395,
      "grad_norm": 0.5416026288585469,
      "learning_rate": 5.644032020715837e-05,
      "loss": 12.0973,
      "step": 24026
    },
    {
      "epoch": 1.3083661759000225,
      "grad_norm": 0.5158199941024781,
      "learning_rate": 5.6432382692925124e-05,
      "loss": 11.9428,
      "step": 24027
    },
    {
      "epoch": 1.3084206298966055,
      "grad_norm": 0.4821028621826143,
      "learning_rate": 5.642444551748101e-05,
      "loss": 11.8582,
      "step": 24028
    },
    {
      "epoch": 1.3084750838931885,
      "grad_norm": 0.5817373794363259,
      "learning_rate": 5.641650868088768e-05,
      "loss": 11.84,
      "step": 24029
    },
    {
      "epoch": 1.3085295378897714,
      "grad_norm": 0.5238270562827404,
      "learning_rate": 5.6408572183206874e-05,
      "loss": 11.8774,
      "step": 24030
    },
    {
      "epoch": 1.3085839918863544,
      "grad_norm": 0.5662106271145493,
      "learning_rate": 5.6400636024500376e-05,
      "loss": 11.8612,
      "step": 24031
    },
    {
      "epoch": 1.3086384458829374,
      "grad_norm": 0.4658411699306566,
      "learning_rate": 5.63927002048298e-05,
      "loss": 11.8203,
      "step": 24032
    },
    {
      "epoch": 1.3086928998795204,
      "grad_norm": 0.5369833044130452,
      "learning_rate": 5.6384764724256935e-05,
      "loss": 11.7989,
      "step": 24033
    },
    {
      "epoch": 1.3087473538761034,
      "grad_norm": 0.6215086146378541,
      "learning_rate": 5.6376829582843416e-05,
      "loss": 11.9815,
      "step": 24034
    },
    {
      "epoch": 1.3088018078726866,
      "grad_norm": 0.575460996600585,
      "learning_rate": 5.636889478065103e-05,
      "loss": 11.8371,
      "step": 24035
    },
    {
      "epoch": 1.3088562618692696,
      "grad_norm": 0.5267694407632315,
      "learning_rate": 5.636096031774143e-05,
      "loss": 11.8948,
      "step": 24036
    },
    {
      "epoch": 1.3089107158658526,
      "grad_norm": 0.9051984971751704,
      "learning_rate": 5.6353026194176284e-05,
      "loss": 11.8355,
      "step": 24037
    },
    {
      "epoch": 1.3089651698624356,
      "grad_norm": 0.5451482067920245,
      "learning_rate": 5.6345092410017366e-05,
      "loss": 11.8367,
      "step": 24038
    },
    {
      "epoch": 1.3090196238590186,
      "grad_norm": 0.556870168449486,
      "learning_rate": 5.633715896532629e-05,
      "loss": 12.0332,
      "step": 24039
    },
    {
      "epoch": 1.3090740778556016,
      "grad_norm": 0.5546790452992627,
      "learning_rate": 5.632922586016482e-05,
      "loss": 11.9491,
      "step": 24040
    },
    {
      "epoch": 1.3091285318521846,
      "grad_norm": 0.548344602851646,
      "learning_rate": 5.6321293094594586e-05,
      "loss": 11.9903,
      "step": 24041
    },
    {
      "epoch": 1.3091829858487676,
      "grad_norm": 0.5352260763084332,
      "learning_rate": 5.63133606686773e-05,
      "loss": 11.7799,
      "step": 24042
    },
    {
      "epoch": 1.3092374398453506,
      "grad_norm": 0.5511868697841315,
      "learning_rate": 5.6305428582474676e-05,
      "loss": 12.0075,
      "step": 24043
    },
    {
      "epoch": 1.3092918938419338,
      "grad_norm": 0.5807556727173098,
      "learning_rate": 5.6297496836048336e-05,
      "loss": 11.9363,
      "step": 24044
    },
    {
      "epoch": 1.3093463478385168,
      "grad_norm": 0.5416019456363944,
      "learning_rate": 5.6289565429460024e-05,
      "loss": 11.9437,
      "step": 24045
    },
    {
      "epoch": 1.3094008018350998,
      "grad_norm": 0.5568288851944784,
      "learning_rate": 5.6281634362771383e-05,
      "loss": 11.9437,
      "step": 24046
    },
    {
      "epoch": 1.3094552558316828,
      "grad_norm": 0.5165909000398059,
      "learning_rate": 5.627370363604405e-05,
      "loss": 11.9533,
      "step": 24047
    },
    {
      "epoch": 1.3095097098282658,
      "grad_norm": 0.4648065800283373,
      "learning_rate": 5.626577324933977e-05,
      "loss": 11.9777,
      "step": 24048
    },
    {
      "epoch": 1.3095641638248487,
      "grad_norm": 0.5402151427444775,
      "learning_rate": 5.625784320272014e-05,
      "loss": 11.9627,
      "step": 24049
    },
    {
      "epoch": 1.3096186178214317,
      "grad_norm": 0.5151015954640977,
      "learning_rate": 5.6249913496246884e-05,
      "loss": 11.8636,
      "step": 24050
    },
    {
      "epoch": 1.3096730718180147,
      "grad_norm": 0.7071352397373822,
      "learning_rate": 5.6241984129981594e-05,
      "loss": 11.9924,
      "step": 24051
    },
    {
      "epoch": 1.3097275258145977,
      "grad_norm": 0.5532174252587672,
      "learning_rate": 5.623405510398598e-05,
      "loss": 11.8778,
      "step": 24052
    },
    {
      "epoch": 1.3097819798111807,
      "grad_norm": 0.5289146410485642,
      "learning_rate": 5.6226126418321734e-05,
      "loss": 11.926,
      "step": 24053
    },
    {
      "epoch": 1.3098364338077637,
      "grad_norm": 0.5334503954840796,
      "learning_rate": 5.621819807305043e-05,
      "loss": 11.9385,
      "step": 24054
    },
    {
      "epoch": 1.3098908878043467,
      "grad_norm": 0.5675098184871193,
      "learning_rate": 5.621027006823381e-05,
      "loss": 11.91,
      "step": 24055
    },
    {
      "epoch": 1.3099453418009297,
      "grad_norm": 0.5520081124621085,
      "learning_rate": 5.620234240393345e-05,
      "loss": 11.8502,
      "step": 24056
    },
    {
      "epoch": 1.3099997957975127,
      "grad_norm": 0.5103938780111638,
      "learning_rate": 5.619441508021099e-05,
      "loss": 11.9141,
      "step": 24057
    },
    {
      "epoch": 1.310054249794096,
      "grad_norm": 0.6273218532787944,
      "learning_rate": 5.618648809712815e-05,
      "loss": 12.0362,
      "step": 24058
    },
    {
      "epoch": 1.310108703790679,
      "grad_norm": 0.6317001937888972,
      "learning_rate": 5.6178561454746495e-05,
      "loss": 11.962,
      "step": 24059
    },
    {
      "epoch": 1.310163157787262,
      "grad_norm": 0.5852638614045123,
      "learning_rate": 5.6170635153127703e-05,
      "loss": 11.9447,
      "step": 24060
    },
    {
      "epoch": 1.3102176117838449,
      "grad_norm": 0.5165607266868081,
      "learning_rate": 5.6162709192333394e-05,
      "loss": 11.9877,
      "step": 24061
    },
    {
      "epoch": 1.3102720657804279,
      "grad_norm": 0.541180216936752,
      "learning_rate": 5.615478357242523e-05,
      "loss": 11.7278,
      "step": 24062
    },
    {
      "epoch": 1.3103265197770109,
      "grad_norm": 0.6027574576790281,
      "learning_rate": 5.61468582934648e-05,
      "loss": 11.7352,
      "step": 24063
    },
    {
      "epoch": 1.3103809737735939,
      "grad_norm": 0.561972172242614,
      "learning_rate": 5.6138933355513744e-05,
      "loss": 11.9482,
      "step": 24064
    },
    {
      "epoch": 1.3104354277701769,
      "grad_norm": 0.540328313228767,
      "learning_rate": 5.613100875863373e-05,
      "loss": 11.8044,
      "step": 24065
    },
    {
      "epoch": 1.3104898817667598,
      "grad_norm": 0.5062809733534139,
      "learning_rate": 5.6123084502886355e-05,
      "loss": 11.4273,
      "step": 24066
    },
    {
      "epoch": 1.310544335763343,
      "grad_norm": 0.5369482911576979,
      "learning_rate": 5.611516058833319e-05,
      "loss": 11.9504,
      "step": 24067
    },
    {
      "epoch": 1.310598789759926,
      "grad_norm": 0.5616688717766474,
      "learning_rate": 5.610723701503595e-05,
      "loss": 11.8698,
      "step": 24068
    },
    {
      "epoch": 1.310653243756509,
      "grad_norm": 0.5599362798095469,
      "learning_rate": 5.609931378305615e-05,
      "loss": 11.7793,
      "step": 24069
    },
    {
      "epoch": 1.310707697753092,
      "grad_norm": 0.5818904319739762,
      "learning_rate": 5.609139089245548e-05,
      "loss": 12.0357,
      "step": 24070
    },
    {
      "epoch": 1.310762151749675,
      "grad_norm": 0.5087136123233187,
      "learning_rate": 5.608346834329547e-05,
      "loss": 11.8868,
      "step": 24071
    },
    {
      "epoch": 1.310816605746258,
      "grad_norm": 0.5162082398924985,
      "learning_rate": 5.607554613563783e-05,
      "loss": 11.928,
      "step": 24072
    },
    {
      "epoch": 1.310871059742841,
      "grad_norm": 0.5629835155002205,
      "learning_rate": 5.606762426954408e-05,
      "loss": 11.8736,
      "step": 24073
    },
    {
      "epoch": 1.310925513739424,
      "grad_norm": 0.5890715222779773,
      "learning_rate": 5.605970274507588e-05,
      "loss": 11.9634,
      "step": 24074
    },
    {
      "epoch": 1.310979967736007,
      "grad_norm": 0.5558526509435125,
      "learning_rate": 5.605178156229476e-05,
      "loss": 11.8771,
      "step": 24075
    },
    {
      "epoch": 1.31103442173259,
      "grad_norm": 0.5009418577530341,
      "learning_rate": 5.604386072126239e-05,
      "loss": 11.9972,
      "step": 24076
    },
    {
      "epoch": 1.311088875729173,
      "grad_norm": 0.5493961153888932,
      "learning_rate": 5.603594022204033e-05,
      "loss": 11.7846,
      "step": 24077
    },
    {
      "epoch": 1.311143329725756,
      "grad_norm": 0.5293464153718507,
      "learning_rate": 5.602802006469014e-05,
      "loss": 11.8235,
      "step": 24078
    },
    {
      "epoch": 1.311197783722339,
      "grad_norm": 0.47494824965195975,
      "learning_rate": 5.602010024927343e-05,
      "loss": 11.9343,
      "step": 24079
    },
    {
      "epoch": 1.311252237718922,
      "grad_norm": 0.5400570253824287,
      "learning_rate": 5.601218077585184e-05,
      "loss": 11.9424,
      "step": 24080
    },
    {
      "epoch": 1.311306691715505,
      "grad_norm": 0.6292446865449202,
      "learning_rate": 5.600426164448688e-05,
      "loss": 11.9883,
      "step": 24081
    },
    {
      "epoch": 1.3113611457120882,
      "grad_norm": 0.5384956763460513,
      "learning_rate": 5.5996342855240194e-05,
      "loss": 12.0204,
      "step": 24082
    },
    {
      "epoch": 1.3114155997086712,
      "grad_norm": 0.4949942337977581,
      "learning_rate": 5.598842440817328e-05,
      "loss": 11.8607,
      "step": 24083
    },
    {
      "epoch": 1.3114700537052542,
      "grad_norm": 0.5495955408059275,
      "learning_rate": 5.59805063033478e-05,
      "loss": 11.8031,
      "step": 24084
    },
    {
      "epoch": 1.3115245077018372,
      "grad_norm": 0.5227917745529473,
      "learning_rate": 5.5972588540825245e-05,
      "loss": 11.9676,
      "step": 24085
    },
    {
      "epoch": 1.3115789616984201,
      "grad_norm": 0.5493167528344649,
      "learning_rate": 5.596467112066727e-05,
      "loss": 12.0129,
      "step": 24086
    },
    {
      "epoch": 1.3116334156950031,
      "grad_norm": 0.5369586038681113,
      "learning_rate": 5.595675404293539e-05,
      "loss": 11.934,
      "step": 24087
    },
    {
      "epoch": 1.3116878696915861,
      "grad_norm": 0.5692549213598999,
      "learning_rate": 5.594883730769114e-05,
      "loss": 12.0284,
      "step": 24088
    },
    {
      "epoch": 1.3117423236881691,
      "grad_norm": 0.5246548461312474,
      "learning_rate": 5.594092091499617e-05,
      "loss": 12.104,
      "step": 24089
    },
    {
      "epoch": 1.3117967776847523,
      "grad_norm": 0.5422912381640886,
      "learning_rate": 5.593300486491193e-05,
      "loss": 11.8926,
      "step": 24090
    },
    {
      "epoch": 1.3118512316813353,
      "grad_norm": 0.5507037008770228,
      "learning_rate": 5.592508915750003e-05,
      "loss": 11.8453,
      "step": 24091
    },
    {
      "epoch": 1.3119056856779183,
      "grad_norm": 0.6109536311951776,
      "learning_rate": 5.5917173792822086e-05,
      "loss": 11.8903,
      "step": 24092
    },
    {
      "epoch": 1.3119601396745013,
      "grad_norm": 0.5667929815034383,
      "learning_rate": 5.590925877093954e-05,
      "loss": 11.925,
      "step": 24093
    },
    {
      "epoch": 1.3120145936710843,
      "grad_norm": 0.5151352515239541,
      "learning_rate": 5.590134409191403e-05,
      "loss": 11.7802,
      "step": 24094
    },
    {
      "epoch": 1.3120690476676673,
      "grad_norm": 0.5312198409149739,
      "learning_rate": 5.589342975580703e-05,
      "loss": 11.8378,
      "step": 24095
    },
    {
      "epoch": 1.3121235016642503,
      "grad_norm": 0.51001384620566,
      "learning_rate": 5.588551576268016e-05,
      "loss": 11.7902,
      "step": 24096
    },
    {
      "epoch": 1.3121779556608333,
      "grad_norm": 0.5474836460471255,
      "learning_rate": 5.587760211259492e-05,
      "loss": 12.0006,
      "step": 24097
    },
    {
      "epoch": 1.3122324096574163,
      "grad_norm": 0.5813697004632432,
      "learning_rate": 5.5869688805612784e-05,
      "loss": 11.9648,
      "step": 24098
    },
    {
      "epoch": 1.3122868636539993,
      "grad_norm": 0.5495130797855269,
      "learning_rate": 5.5861775841795417e-05,
      "loss": 11.8442,
      "step": 24099
    },
    {
      "epoch": 1.3123413176505823,
      "grad_norm": 0.5563958357476311,
      "learning_rate": 5.585386322120423e-05,
      "loss": 11.9653,
      "step": 24100
    },
    {
      "epoch": 1.3123957716471653,
      "grad_norm": 0.5601210651888593,
      "learning_rate": 5.584595094390079e-05,
      "loss": 11.9562,
      "step": 24101
    },
    {
      "epoch": 1.3124502256437482,
      "grad_norm": 0.5085020324736595,
      "learning_rate": 5.583803900994671e-05,
      "loss": 11.974,
      "step": 24102
    },
    {
      "epoch": 1.3125046796403312,
      "grad_norm": 0.5350322377035194,
      "learning_rate": 5.583012741940339e-05,
      "loss": 12.0051,
      "step": 24103
    },
    {
      "epoch": 1.3125591336369142,
      "grad_norm": 0.5307092152711169,
      "learning_rate": 5.5822216172332444e-05,
      "loss": 11.9561,
      "step": 24104
    },
    {
      "epoch": 1.3126135876334974,
      "grad_norm": 0.5560314482779806,
      "learning_rate": 5.581430526879532e-05,
      "loss": 11.8514,
      "step": 24105
    },
    {
      "epoch": 1.3126680416300804,
      "grad_norm": 0.5246609139397009,
      "learning_rate": 5.58063947088536e-05,
      "loss": 11.9932,
      "step": 24106
    },
    {
      "epoch": 1.3127224956266634,
      "grad_norm": 0.5446374354321504,
      "learning_rate": 5.579848449256877e-05,
      "loss": 11.8102,
      "step": 24107
    },
    {
      "epoch": 1.3127769496232464,
      "grad_norm": 0.557710650956272,
      "learning_rate": 5.57905746200023e-05,
      "loss": 11.9275,
      "step": 24108
    },
    {
      "epoch": 1.3128314036198294,
      "grad_norm": 0.569629619957673,
      "learning_rate": 5.5782665091215766e-05,
      "loss": 11.9469,
      "step": 24109
    },
    {
      "epoch": 1.3128858576164124,
      "grad_norm": 0.5452258413061813,
      "learning_rate": 5.5774755906270613e-05,
      "loss": 12.0616,
      "step": 24110
    },
    {
      "epoch": 1.3129403116129954,
      "grad_norm": 0.549681639306198,
      "learning_rate": 5.576684706522841e-05,
      "loss": 11.814,
      "step": 24111
    },
    {
      "epoch": 1.3129947656095784,
      "grad_norm": 0.8968357936170647,
      "learning_rate": 5.5758938568150585e-05,
      "loss": 12.0231,
      "step": 24112
    },
    {
      "epoch": 1.3130492196061614,
      "grad_norm": 0.6114355622916864,
      "learning_rate": 5.575103041509867e-05,
      "loss": 11.9277,
      "step": 24113
    },
    {
      "epoch": 1.3131036736027446,
      "grad_norm": 0.5632441548114183,
      "learning_rate": 5.5743122606134214e-05,
      "loss": 11.8432,
      "step": 24114
    },
    {
      "epoch": 1.3131581275993276,
      "grad_norm": 0.6229341802493858,
      "learning_rate": 5.5735215141318607e-05,
      "loss": 12.0,
      "step": 24115
    },
    {
      "epoch": 1.3132125815959106,
      "grad_norm": 0.5325063396721057,
      "learning_rate": 5.572730802071343e-05,
      "loss": 11.888,
      "step": 24116
    },
    {
      "epoch": 1.3132670355924936,
      "grad_norm": 0.570325299049395,
      "learning_rate": 5.571940124438013e-05,
      "loss": 11.9914,
      "step": 24117
    },
    {
      "epoch": 1.3133214895890766,
      "grad_norm": 0.5685258266474903,
      "learning_rate": 5.571149481238016e-05,
      "loss": 11.9793,
      "step": 24118
    },
    {
      "epoch": 1.3133759435856596,
      "grad_norm": 0.5074342320921148,
      "learning_rate": 5.570358872477507e-05,
      "loss": 11.8027,
      "step": 24119
    },
    {
      "epoch": 1.3134303975822426,
      "grad_norm": 0.5158155993323239,
      "learning_rate": 5.5695682981626266e-05,
      "loss": 11.8288,
      "step": 24120
    },
    {
      "epoch": 1.3134848515788256,
      "grad_norm": 0.5378069571240769,
      "learning_rate": 5.5687777582995284e-05,
      "loss": 11.7794,
      "step": 24121
    },
    {
      "epoch": 1.3135393055754085,
      "grad_norm": 0.5895099475242283,
      "learning_rate": 5.567987252894355e-05,
      "loss": 12.0278,
      "step": 24122
    },
    {
      "epoch": 1.3135937595719915,
      "grad_norm": 0.5474999267889047,
      "learning_rate": 5.56719678195326e-05,
      "loss": 11.9065,
      "step": 24123
    },
    {
      "epoch": 1.3136482135685745,
      "grad_norm": 0.6210183064545841,
      "learning_rate": 5.566406345482382e-05,
      "loss": 12.0531,
      "step": 24124
    },
    {
      "epoch": 1.3137026675651575,
      "grad_norm": 0.5584809577507116,
      "learning_rate": 5.565615943487872e-05,
      "loss": 11.88,
      "step": 24125
    },
    {
      "epoch": 1.3137571215617405,
      "grad_norm": 0.5620921521487785,
      "learning_rate": 5.564825575975883e-05,
      "loss": 11.8425,
      "step": 24126
    },
    {
      "epoch": 1.3138115755583235,
      "grad_norm": 0.6000775354646204,
      "learning_rate": 5.5640352429525466e-05,
      "loss": 11.9881,
      "step": 24127
    },
    {
      "epoch": 1.3138660295549067,
      "grad_norm": 0.5725438406253455,
      "learning_rate": 5.5632449444240165e-05,
      "loss": 11.8342,
      "step": 24128
    },
    {
      "epoch": 1.3139204835514897,
      "grad_norm": 0.5747252928413819,
      "learning_rate": 5.5624546803964414e-05,
      "loss": 11.9359,
      "step": 24129
    },
    {
      "epoch": 1.3139749375480727,
      "grad_norm": 0.5580723502020485,
      "learning_rate": 5.5616644508759585e-05,
      "loss": 11.8348,
      "step": 24130
    },
    {
      "epoch": 1.3140293915446557,
      "grad_norm": 0.5206431814318879,
      "learning_rate": 5.560874255868722e-05,
      "loss": 11.994,
      "step": 24131
    },
    {
      "epoch": 1.3140838455412387,
      "grad_norm": 0.5139602020182866,
      "learning_rate": 5.5600840953808675e-05,
      "loss": 11.8856,
      "step": 24132
    },
    {
      "epoch": 1.3141382995378217,
      "grad_norm": 0.5273997701362135,
      "learning_rate": 5.5592939694185473e-05,
      "loss": 11.9026,
      "step": 24133
    },
    {
      "epoch": 1.3141927535344047,
      "grad_norm": 0.5255371617969886,
      "learning_rate": 5.5585038779878984e-05,
      "loss": 11.869,
      "step": 24134
    },
    {
      "epoch": 1.3142472075309877,
      "grad_norm": 0.5605092161408242,
      "learning_rate": 5.557713821095067e-05,
      "loss": 11.9114,
      "step": 24135
    },
    {
      "epoch": 1.3143016615275707,
      "grad_norm": 0.5323861289247597,
      "learning_rate": 5.556923798746204e-05,
      "loss": 11.8343,
      "step": 24136
    },
    {
      "epoch": 1.3143561155241539,
      "grad_norm": 0.5361199374183426,
      "learning_rate": 5.556133810947446e-05,
      "loss": 11.9019,
      "step": 24137
    },
    {
      "epoch": 1.3144105695207369,
      "grad_norm": 0.5515330296315245,
      "learning_rate": 5.5553438577049354e-05,
      "loss": 11.9666,
      "step": 24138
    },
    {
      "epoch": 1.3144650235173199,
      "grad_norm": 0.532408575148832,
      "learning_rate": 5.5545539390248135e-05,
      "loss": 12.028,
      "step": 24139
    },
    {
      "epoch": 1.3145194775139029,
      "grad_norm": 0.5354823885175315,
      "learning_rate": 5.553764054913225e-05,
      "loss": 11.9947,
      "step": 24140
    },
    {
      "epoch": 1.3145739315104858,
      "grad_norm": 0.5659012633566877,
      "learning_rate": 5.552974205376318e-05,
      "loss": 11.8438,
      "step": 24141
    },
    {
      "epoch": 1.3146283855070688,
      "grad_norm": 0.5160737526667496,
      "learning_rate": 5.552184390420223e-05,
      "loss": 11.8927,
      "step": 24142
    },
    {
      "epoch": 1.3146828395036518,
      "grad_norm": 0.523950389511685,
      "learning_rate": 5.551394610051095e-05,
      "loss": 11.8478,
      "step": 24143
    },
    {
      "epoch": 1.3147372935002348,
      "grad_norm": 0.5525784865107607,
      "learning_rate": 5.550604864275062e-05,
      "loss": 11.9292,
      "step": 24144
    },
    {
      "epoch": 1.3147917474968178,
      "grad_norm": 0.5705633641822888,
      "learning_rate": 5.5498151530982765e-05,
      "loss": 11.9413,
      "step": 24145
    },
    {
      "epoch": 1.3148462014934008,
      "grad_norm": 0.5416032511796919,
      "learning_rate": 5.549025476526872e-05,
      "loss": 11.989,
      "step": 24146
    },
    {
      "epoch": 1.3149006554899838,
      "grad_norm": 0.5547615693651402,
      "learning_rate": 5.548235834566995e-05,
      "loss": 11.8103,
      "step": 24147
    },
    {
      "epoch": 1.3149551094865668,
      "grad_norm": 0.529193120232034,
      "learning_rate": 5.5474462272247816e-05,
      "loss": 11.8657,
      "step": 24148
    },
    {
      "epoch": 1.3150095634831498,
      "grad_norm": 0.5883229261148182,
      "learning_rate": 5.5466566545063705e-05,
      "loss": 11.8602,
      "step": 24149
    },
    {
      "epoch": 1.3150640174797328,
      "grad_norm": 0.5435475930209137,
      "learning_rate": 5.545867116417903e-05,
      "loss": 11.9026,
      "step": 24150
    },
    {
      "epoch": 1.315118471476316,
      "grad_norm": 0.5343514848972449,
      "learning_rate": 5.545077612965525e-05,
      "loss": 11.9724,
      "step": 24151
    },
    {
      "epoch": 1.315172925472899,
      "grad_norm": 0.5512682603081,
      "learning_rate": 5.544288144155365e-05,
      "loss": 11.9269,
      "step": 24152
    },
    {
      "epoch": 1.315227379469482,
      "grad_norm": 0.5353583863282051,
      "learning_rate": 5.543498709993573e-05,
      "loss": 11.957,
      "step": 24153
    },
    {
      "epoch": 1.315281833466065,
      "grad_norm": 0.5511459412465701,
      "learning_rate": 5.5427093104862785e-05,
      "loss": 11.9519,
      "step": 24154
    },
    {
      "epoch": 1.315336287462648,
      "grad_norm": 0.5174468989461876,
      "learning_rate": 5.541919945639626e-05,
      "loss": 11.8569,
      "step": 24155
    },
    {
      "epoch": 1.315390741459231,
      "grad_norm": 0.4994135320521054,
      "learning_rate": 5.54113061545975e-05,
      "loss": 11.927,
      "step": 24156
    },
    {
      "epoch": 1.315445195455814,
      "grad_norm": 0.509100283757051,
      "learning_rate": 5.540341319952793e-05,
      "loss": 11.8701,
      "step": 24157
    },
    {
      "epoch": 1.315499649452397,
      "grad_norm": 0.5681110353943868,
      "learning_rate": 5.5395520591248896e-05,
      "loss": 11.9094,
      "step": 24158
    },
    {
      "epoch": 1.31555410344898,
      "grad_norm": 0.5072899762004406,
      "learning_rate": 5.5387628329821736e-05,
      "loss": 11.9164,
      "step": 24159
    },
    {
      "epoch": 1.3156085574455632,
      "grad_norm": 0.5677614542523862,
      "learning_rate": 5.537973641530789e-05,
      "loss": 12.0422,
      "step": 24160
    },
    {
      "epoch": 1.3156630114421461,
      "grad_norm": 0.5821230787364737,
      "learning_rate": 5.537184484776867e-05,
      "loss": 11.8967,
      "step": 24161
    },
    {
      "epoch": 1.3157174654387291,
      "grad_norm": 0.5055793912300793,
      "learning_rate": 5.536395362726548e-05,
      "loss": 11.9621,
      "step": 24162
    },
    {
      "epoch": 1.3157719194353121,
      "grad_norm": 0.5134850537248695,
      "learning_rate": 5.535606275385968e-05,
      "loss": 11.7587,
      "step": 24163
    },
    {
      "epoch": 1.3158263734318951,
      "grad_norm": 0.5419693543723557,
      "learning_rate": 5.53481722276126e-05,
      "loss": 11.9077,
      "step": 24164
    },
    {
      "epoch": 1.3158808274284781,
      "grad_norm": 0.5173375709901956,
      "learning_rate": 5.5340282048585654e-05,
      "loss": 12.0389,
      "step": 24165
    },
    {
      "epoch": 1.315935281425061,
      "grad_norm": 0.5504550053957293,
      "learning_rate": 5.533239221684015e-05,
      "loss": 11.9779,
      "step": 24166
    },
    {
      "epoch": 1.315989735421644,
      "grad_norm": 0.5451182986736287,
      "learning_rate": 5.5324502732437454e-05,
      "loss": 11.9775,
      "step": 24167
    },
    {
      "epoch": 1.316044189418227,
      "grad_norm": 0.503295735563272,
      "learning_rate": 5.531661359543895e-05,
      "loss": 11.9263,
      "step": 24168
    },
    {
      "epoch": 1.31609864341481,
      "grad_norm": 0.5789356449436505,
      "learning_rate": 5.53087248059059e-05,
      "loss": 11.9394,
      "step": 24169
    },
    {
      "epoch": 1.316153097411393,
      "grad_norm": 0.5325233102245421,
      "learning_rate": 5.5300836363899736e-05,
      "loss": 12.0002,
      "step": 24170
    },
    {
      "epoch": 1.316207551407976,
      "grad_norm": 0.5495782431769719,
      "learning_rate": 5.529294826948173e-05,
      "loss": 11.6923,
      "step": 24171
    },
    {
      "epoch": 1.316262005404559,
      "grad_norm": 0.512139927840057,
      "learning_rate": 5.528506052271325e-05,
      "loss": 11.8159,
      "step": 24172
    },
    {
      "epoch": 1.316316459401142,
      "grad_norm": 0.48019392667061545,
      "learning_rate": 5.527717312365569e-05,
      "loss": 11.8032,
      "step": 24173
    },
    {
      "epoch": 1.316370913397725,
      "grad_norm": 0.539613227120157,
      "learning_rate": 5.526928607237028e-05,
      "loss": 11.8105,
      "step": 24174
    },
    {
      "epoch": 1.3164253673943083,
      "grad_norm": 0.5114423431699913,
      "learning_rate": 5.5261399368918454e-05,
      "loss": 11.7604,
      "step": 24175
    },
    {
      "epoch": 1.3164798213908913,
      "grad_norm": 0.5524076486521812,
      "learning_rate": 5.5253513013361434e-05,
      "loss": 11.8275,
      "step": 24176
    },
    {
      "epoch": 1.3165342753874743,
      "grad_norm": 0.5331113835805055,
      "learning_rate": 5.5245627005760656e-05,
      "loss": 11.7874,
      "step": 24177
    },
    {
      "epoch": 1.3165887293840572,
      "grad_norm": 0.5454035226612589,
      "learning_rate": 5.5237741346177385e-05,
      "loss": 11.9775,
      "step": 24178
    },
    {
      "epoch": 1.3166431833806402,
      "grad_norm": 0.5283015966927866,
      "learning_rate": 5.52298560346729e-05,
      "loss": 11.8976,
      "step": 24179
    },
    {
      "epoch": 1.3166976373772232,
      "grad_norm": 0.5871850154321271,
      "learning_rate": 5.5221971071308595e-05,
      "loss": 11.9353,
      "step": 24180
    },
    {
      "epoch": 1.3167520913738062,
      "grad_norm": 0.6136044105531907,
      "learning_rate": 5.521408645614572e-05,
      "loss": 11.8291,
      "step": 24181
    },
    {
      "epoch": 1.3168065453703892,
      "grad_norm": 0.5487668700359158,
      "learning_rate": 5.520620218924566e-05,
      "loss": 12.0251,
      "step": 24182
    },
    {
      "epoch": 1.3168609993669722,
      "grad_norm": 0.5327122282694071,
      "learning_rate": 5.519831827066965e-05,
      "loss": 11.9015,
      "step": 24183
    },
    {
      "epoch": 1.3169154533635554,
      "grad_norm": 0.5624584878840289,
      "learning_rate": 5.519043470047902e-05,
      "loss": 11.9414,
      "step": 24184
    },
    {
      "epoch": 1.3169699073601384,
      "grad_norm": 0.5313704514202723,
      "learning_rate": 5.518255147873512e-05,
      "loss": 11.8792,
      "step": 24185
    },
    {
      "epoch": 1.3170243613567214,
      "grad_norm": 0.6087527962046092,
      "learning_rate": 5.517466860549919e-05,
      "loss": 11.7326,
      "step": 24186
    },
    {
      "epoch": 1.3170788153533044,
      "grad_norm": 0.5545308368584826,
      "learning_rate": 5.5166786080832635e-05,
      "loss": 11.9461,
      "step": 24187
    },
    {
      "epoch": 1.3171332693498874,
      "grad_norm": 0.5577385464793357,
      "learning_rate": 5.5158903904796576e-05,
      "loss": 11.8884,
      "step": 24188
    },
    {
      "epoch": 1.3171877233464704,
      "grad_norm": 0.5656530028652127,
      "learning_rate": 5.515102207745241e-05,
      "loss": 11.9084,
      "step": 24189
    },
    {
      "epoch": 1.3172421773430534,
      "grad_norm": 0.5910234974929589,
      "learning_rate": 5.5143140598861455e-05,
      "loss": 11.8402,
      "step": 24190
    },
    {
      "epoch": 1.3172966313396364,
      "grad_norm": 0.5865373586103833,
      "learning_rate": 5.513525946908491e-05,
      "loss": 11.9717,
      "step": 24191
    },
    {
      "epoch": 1.3173510853362194,
      "grad_norm": 0.5158602067307513,
      "learning_rate": 5.512737868818416e-05,
      "loss": 12.0789,
      "step": 24192
    },
    {
      "epoch": 1.3174055393328024,
      "grad_norm": 0.5260099951371222,
      "learning_rate": 5.51194982562204e-05,
      "loss": 11.8863,
      "step": 24193
    },
    {
      "epoch": 1.3174599933293853,
      "grad_norm": 0.5960360865287289,
      "learning_rate": 5.5111618173254984e-05,
      "loss": 11.8579,
      "step": 24194
    },
    {
      "epoch": 1.3175144473259683,
      "grad_norm": 0.5974298738793148,
      "learning_rate": 5.510373843934913e-05,
      "loss": 11.9543,
      "step": 24195
    },
    {
      "epoch": 1.3175689013225513,
      "grad_norm": 0.5251629446111129,
      "learning_rate": 5.509585905456411e-05,
      "loss": 11.8436,
      "step": 24196
    },
    {
      "epoch": 1.3176233553191343,
      "grad_norm": 0.5550405248967265,
      "learning_rate": 5.508798001896132e-05,
      "loss": 11.9545,
      "step": 24197
    },
    {
      "epoch": 1.3176778093157175,
      "grad_norm": 0.48351700828475436,
      "learning_rate": 5.508010133260184e-05,
      "loss": 11.8354,
      "step": 24198
    },
    {
      "epoch": 1.3177322633123005,
      "grad_norm": 0.5464638364304195,
      "learning_rate": 5.5072222995547026e-05,
      "loss": 11.8961,
      "step": 24199
    },
    {
      "epoch": 1.3177867173088835,
      "grad_norm": 0.5637109969518448,
      "learning_rate": 5.506434500785817e-05,
      "loss": 12.0174,
      "step": 24200
    },
    {
      "epoch": 1.3178411713054665,
      "grad_norm": 0.5533901495275418,
      "learning_rate": 5.5056467369596484e-05,
      "loss": 11.8048,
      "step": 24201
    },
    {
      "epoch": 1.3178956253020495,
      "grad_norm": 0.5677288863780411,
      "learning_rate": 5.504859008082328e-05,
      "loss": 11.8226,
      "step": 24202
    },
    {
      "epoch": 1.3179500792986325,
      "grad_norm": 0.6358236306367222,
      "learning_rate": 5.504071314159973e-05,
      "loss": 11.9235,
      "step": 24203
    },
    {
      "epoch": 1.3180045332952155,
      "grad_norm": 0.5985184796954829,
      "learning_rate": 5.503283655198718e-05,
      "loss": 11.8566,
      "step": 24204
    },
    {
      "epoch": 1.3180589872917985,
      "grad_norm": 0.5934497309754219,
      "learning_rate": 5.502496031204678e-05,
      "loss": 11.9266,
      "step": 24205
    },
    {
      "epoch": 1.3181134412883815,
      "grad_norm": 0.5245202950791581,
      "learning_rate": 5.501708442183985e-05,
      "loss": 11.8096,
      "step": 24206
    },
    {
      "epoch": 1.3181678952849647,
      "grad_norm": 0.5235175834459708,
      "learning_rate": 5.5009208881427686e-05,
      "loss": 11.9094,
      "step": 24207
    },
    {
      "epoch": 1.3182223492815477,
      "grad_norm": 0.5299367809455927,
      "learning_rate": 5.500133369087137e-05,
      "loss": 11.9188,
      "step": 24208
    },
    {
      "epoch": 1.3182768032781307,
      "grad_norm": 0.5295516934968243,
      "learning_rate": 5.4993458850232285e-05,
      "loss": 11.8589,
      "step": 24209
    },
    {
      "epoch": 1.3183312572747137,
      "grad_norm": 0.5885540068868155,
      "learning_rate": 5.4985584359571565e-05,
      "loss": 12.0495,
      "step": 24210
    },
    {
      "epoch": 1.3183857112712967,
      "grad_norm": 0.6380428176961217,
      "learning_rate": 5.497771021895048e-05,
      "loss": 11.8852,
      "step": 24211
    },
    {
      "epoch": 1.3184401652678797,
      "grad_norm": 0.5672198964296776,
      "learning_rate": 5.496983642843033e-05,
      "loss": 12.014,
      "step": 24212
    },
    {
      "epoch": 1.3184946192644627,
      "grad_norm": 0.8291922617570414,
      "learning_rate": 5.4961962988072237e-05,
      "loss": 11.9553,
      "step": 24213
    },
    {
      "epoch": 1.3185490732610456,
      "grad_norm": 0.5886373816221026,
      "learning_rate": 5.495408989793751e-05,
      "loss": 11.9391,
      "step": 24214
    },
    {
      "epoch": 1.3186035272576286,
      "grad_norm": 0.5320349527412451,
      "learning_rate": 5.49462171580873e-05,
      "loss": 11.9512,
      "step": 24215
    },
    {
      "epoch": 1.3186579812542116,
      "grad_norm": 0.4912981698863273,
      "learning_rate": 5.4938344768582886e-05,
      "loss": 11.8482,
      "step": 24216
    },
    {
      "epoch": 1.3187124352507946,
      "grad_norm": 0.5738559075083566,
      "learning_rate": 5.493047272948547e-05,
      "loss": 11.9387,
      "step": 24217
    },
    {
      "epoch": 1.3187668892473776,
      "grad_norm": 0.6180322146989785,
      "learning_rate": 5.492260104085621e-05,
      "loss": 11.9237,
      "step": 24218
    },
    {
      "epoch": 1.3188213432439606,
      "grad_norm": 0.5092424770447668,
      "learning_rate": 5.491472970275642e-05,
      "loss": 11.8272,
      "step": 24219
    },
    {
      "epoch": 1.3188757972405436,
      "grad_norm": 0.6367962032492092,
      "learning_rate": 5.490685871524719e-05,
      "loss": 12.0051,
      "step": 24220
    },
    {
      "epoch": 1.3189302512371268,
      "grad_norm": 0.537058284768866,
      "learning_rate": 5.48989880783898e-05,
      "loss": 11.8841,
      "step": 24221
    },
    {
      "epoch": 1.3189847052337098,
      "grad_norm": 0.5582860930165967,
      "learning_rate": 5.4891117792245494e-05,
      "loss": 11.9276,
      "step": 24222
    },
    {
      "epoch": 1.3190391592302928,
      "grad_norm": 0.6171494646409815,
      "learning_rate": 5.488324785687537e-05,
      "loss": 11.9321,
      "step": 24223
    },
    {
      "epoch": 1.3190936132268758,
      "grad_norm": 0.5364331402214078,
      "learning_rate": 5.48753782723407e-05,
      "loss": 11.8419,
      "step": 24224
    },
    {
      "epoch": 1.3191480672234588,
      "grad_norm": 0.5715325938987341,
      "learning_rate": 5.4867509038702644e-05,
      "loss": 12.0285,
      "step": 24225
    },
    {
      "epoch": 1.3192025212200418,
      "grad_norm": 0.5696662353364319,
      "learning_rate": 5.485964015602243e-05,
      "loss": 11.8765,
      "step": 24226
    },
    {
      "epoch": 1.3192569752166248,
      "grad_norm": 0.553330927101105,
      "learning_rate": 5.4851771624361236e-05,
      "loss": 11.9422,
      "step": 24227
    },
    {
      "epoch": 1.3193114292132078,
      "grad_norm": 0.6435706481748136,
      "learning_rate": 5.4843903443780185e-05,
      "loss": 12.1121,
      "step": 24228
    },
    {
      "epoch": 1.3193658832097908,
      "grad_norm": 0.5227215737264612,
      "learning_rate": 5.4836035614340566e-05,
      "loss": 11.9436,
      "step": 24229
    },
    {
      "epoch": 1.319420337206374,
      "grad_norm": 0.7070280218748599,
      "learning_rate": 5.482816813610345e-05,
      "loss": 11.7973,
      "step": 24230
    },
    {
      "epoch": 1.319474791202957,
      "grad_norm": 0.5108586704723379,
      "learning_rate": 5.482030100913014e-05,
      "loss": 11.7991,
      "step": 24231
    },
    {
      "epoch": 1.31952924519954,
      "grad_norm": 0.6067175139272265,
      "learning_rate": 5.481243423348168e-05,
      "loss": 11.8569,
      "step": 24232
    },
    {
      "epoch": 1.319583699196123,
      "grad_norm": 0.5096432546957752,
      "learning_rate": 5.480456780921932e-05,
      "loss": 11.9371,
      "step": 24233
    },
    {
      "epoch": 1.319638153192706,
      "grad_norm": 0.568550815802961,
      "learning_rate": 5.479670173640426e-05,
      "loss": 11.9072,
      "step": 24234
    },
    {
      "epoch": 1.319692607189289,
      "grad_norm": 0.5512807238669505,
      "learning_rate": 5.478883601509759e-05,
      "loss": 11.9034,
      "step": 24235
    },
    {
      "epoch": 1.319747061185872,
      "grad_norm": 0.5151034126730991,
      "learning_rate": 5.478097064536054e-05,
      "loss": 11.9168,
      "step": 24236
    },
    {
      "epoch": 1.319801515182455,
      "grad_norm": 0.5536047071293094,
      "learning_rate": 5.477310562725426e-05,
      "loss": 11.9195,
      "step": 24237
    },
    {
      "epoch": 1.319855969179038,
      "grad_norm": 0.5049946864835259,
      "learning_rate": 5.476524096083986e-05,
      "loss": 11.7928,
      "step": 24238
    },
    {
      "epoch": 1.319910423175621,
      "grad_norm": 0.5610898251218176,
      "learning_rate": 5.475737664617856e-05,
      "loss": 11.7958,
      "step": 24239
    },
    {
      "epoch": 1.319964877172204,
      "grad_norm": 0.5647761995622971,
      "learning_rate": 5.474951268333145e-05,
      "loss": 11.8977,
      "step": 24240
    },
    {
      "epoch": 1.320019331168787,
      "grad_norm": 0.5679213557007524,
      "learning_rate": 5.474164907235976e-05,
      "loss": 11.83,
      "step": 24241
    },
    {
      "epoch": 1.3200737851653699,
      "grad_norm": 0.5218434301393386,
      "learning_rate": 5.473378581332456e-05,
      "loss": 11.928,
      "step": 24242
    },
    {
      "epoch": 1.3201282391619529,
      "grad_norm": 0.5110787526488226,
      "learning_rate": 5.472592290628707e-05,
      "loss": 11.8448,
      "step": 24243
    },
    {
      "epoch": 1.3201826931585359,
      "grad_norm": 0.5731643963526519,
      "learning_rate": 5.471806035130834e-05,
      "loss": 11.8394,
      "step": 24244
    },
    {
      "epoch": 1.320237147155119,
      "grad_norm": 0.5569903728328777,
      "learning_rate": 5.4710198148449576e-05,
      "loss": 12.0095,
      "step": 24245
    },
    {
      "epoch": 1.320291601151702,
      "grad_norm": 0.5234051767866967,
      "learning_rate": 5.470233629777195e-05,
      "loss": 11.9206,
      "step": 24246
    },
    {
      "epoch": 1.320346055148285,
      "grad_norm": 0.5198547765554428,
      "learning_rate": 5.469447479933656e-05,
      "loss": 11.9631,
      "step": 24247
    },
    {
      "epoch": 1.320400509144868,
      "grad_norm": 0.5586794934699644,
      "learning_rate": 5.468661365320449e-05,
      "loss": 12.0107,
      "step": 24248
    },
    {
      "epoch": 1.320454963141451,
      "grad_norm": 0.5489873444126732,
      "learning_rate": 5.467875285943693e-05,
      "loss": 11.9004,
      "step": 24249
    },
    {
      "epoch": 1.320509417138034,
      "grad_norm": 0.5008034757135255,
      "learning_rate": 5.467089241809498e-05,
      "loss": 11.891,
      "step": 24250
    },
    {
      "epoch": 1.320563871134617,
      "grad_norm": 0.532421244376691,
      "learning_rate": 5.466303232923979e-05,
      "loss": 11.8025,
      "step": 24251
    },
    {
      "epoch": 1.3206183251312,
      "grad_norm": 0.5069687974078494,
      "learning_rate": 5.465517259293242e-05,
      "loss": 11.7175,
      "step": 24252
    },
    {
      "epoch": 1.3206727791277832,
      "grad_norm": 0.5436007210662951,
      "learning_rate": 5.464731320923409e-05,
      "loss": 11.9452,
      "step": 24253
    },
    {
      "epoch": 1.3207272331243662,
      "grad_norm": 0.537046876651317,
      "learning_rate": 5.463945417820582e-05,
      "loss": 11.7874,
      "step": 24254
    },
    {
      "epoch": 1.3207816871209492,
      "grad_norm": 0.5993244268052655,
      "learning_rate": 5.463159549990875e-05,
      "loss": 11.9102,
      "step": 24255
    },
    {
      "epoch": 1.3208361411175322,
      "grad_norm": 0.617724157337032,
      "learning_rate": 5.4623737174404034e-05,
      "loss": 11.8869,
      "step": 24256
    },
    {
      "epoch": 1.3208905951141152,
      "grad_norm": 0.549806880415778,
      "learning_rate": 5.461587920175276e-05,
      "loss": 11.9112,
      "step": 24257
    },
    {
      "epoch": 1.3209450491106982,
      "grad_norm": 0.5788189893060056,
      "learning_rate": 5.460802158201601e-05,
      "loss": 11.9908,
      "step": 24258
    },
    {
      "epoch": 1.3209995031072812,
      "grad_norm": 0.57018458798926,
      "learning_rate": 5.4600164315254876e-05,
      "loss": 11.8401,
      "step": 24259
    },
    {
      "epoch": 1.3210539571038642,
      "grad_norm": 0.5149251024593897,
      "learning_rate": 5.459230740153046e-05,
      "loss": 11.9464,
      "step": 24260
    },
    {
      "epoch": 1.3211084111004472,
      "grad_norm": 0.6838468707258352,
      "learning_rate": 5.458445084090392e-05,
      "loss": 11.927,
      "step": 24261
    },
    {
      "epoch": 1.3211628650970302,
      "grad_norm": 0.5518764799362267,
      "learning_rate": 5.4576594633436275e-05,
      "loss": 11.803,
      "step": 24262
    },
    {
      "epoch": 1.3212173190936132,
      "grad_norm": 0.5660078351865124,
      "learning_rate": 5.456873877918868e-05,
      "loss": 11.8962,
      "step": 24263
    },
    {
      "epoch": 1.3212717730901962,
      "grad_norm": 0.5327815934793932,
      "learning_rate": 5.4560883278222156e-05,
      "loss": 11.8649,
      "step": 24264
    },
    {
      "epoch": 1.3213262270867792,
      "grad_norm": 0.575616894847649,
      "learning_rate": 5.455302813059786e-05,
      "loss": 12.0352,
      "step": 24265
    },
    {
      "epoch": 1.3213806810833622,
      "grad_norm": 0.5414418027524431,
      "learning_rate": 5.454517333637679e-05,
      "loss": 11.9338,
      "step": 24266
    },
    {
      "epoch": 1.3214351350799451,
      "grad_norm": 0.5228913502066164,
      "learning_rate": 5.4537318895620126e-05,
      "loss": 11.9759,
      "step": 24267
    },
    {
      "epoch": 1.3214895890765284,
      "grad_norm": 0.5037217239524571,
      "learning_rate": 5.4529464808388894e-05,
      "loss": 11.9221,
      "step": 24268
    },
    {
      "epoch": 1.3215440430731114,
      "grad_norm": 0.6567086658448069,
      "learning_rate": 5.4521611074744125e-05,
      "loss": 11.8346,
      "step": 24269
    },
    {
      "epoch": 1.3215984970696943,
      "grad_norm": 0.5419176707417476,
      "learning_rate": 5.4513757694746935e-05,
      "loss": 11.9521,
      "step": 24270
    },
    {
      "epoch": 1.3216529510662773,
      "grad_norm": 0.47978489373554256,
      "learning_rate": 5.450590466845844e-05,
      "loss": 11.9456,
      "step": 24271
    },
    {
      "epoch": 1.3217074050628603,
      "grad_norm": 0.5356943236231767,
      "learning_rate": 5.449805199593962e-05,
      "loss": 11.9803,
      "step": 24272
    },
    {
      "epoch": 1.3217618590594433,
      "grad_norm": 0.5273473014510761,
      "learning_rate": 5.449019967725161e-05,
      "loss": 11.8619,
      "step": 24273
    },
    {
      "epoch": 1.3218163130560263,
      "grad_norm": 0.5501787690953788,
      "learning_rate": 5.44823477124554e-05,
      "loss": 12.0075,
      "step": 24274
    },
    {
      "epoch": 1.3218707670526093,
      "grad_norm": 0.5408211223098347,
      "learning_rate": 5.447449610161213e-05,
      "loss": 11.9897,
      "step": 24275
    },
    {
      "epoch": 1.3219252210491923,
      "grad_norm": 0.5398280967216086,
      "learning_rate": 5.4466644844782765e-05,
      "loss": 11.9916,
      "step": 24276
    },
    {
      "epoch": 1.3219796750457755,
      "grad_norm": 0.5126606641286382,
      "learning_rate": 5.445879394202846e-05,
      "loss": 11.9482,
      "step": 24277
    },
    {
      "epoch": 1.3220341290423585,
      "grad_norm": 0.5223280025625218,
      "learning_rate": 5.44509433934102e-05,
      "loss": 11.8561,
      "step": 24278
    },
    {
      "epoch": 1.3220885830389415,
      "grad_norm": 0.5009270176350861,
      "learning_rate": 5.444309319898901e-05,
      "loss": 11.7228,
      "step": 24279
    },
    {
      "epoch": 1.3221430370355245,
      "grad_norm": 0.5563132453475934,
      "learning_rate": 5.443524335882601e-05,
      "loss": 11.9657,
      "step": 24280
    },
    {
      "epoch": 1.3221974910321075,
      "grad_norm": 0.5256210233397686,
      "learning_rate": 5.442739387298215e-05,
      "loss": 11.9307,
      "step": 24281
    },
    {
      "epoch": 1.3222519450286905,
      "grad_norm": 0.5373238434806185,
      "learning_rate": 5.441954474151851e-05,
      "loss": 11.9033,
      "step": 24282
    },
    {
      "epoch": 1.3223063990252735,
      "grad_norm": 0.5461130735811333,
      "learning_rate": 5.44116959644962e-05,
      "loss": 11.9797,
      "step": 24283
    },
    {
      "epoch": 1.3223608530218565,
      "grad_norm": 0.5347686060063062,
      "learning_rate": 5.440384754197613e-05,
      "loss": 11.8984,
      "step": 24284
    },
    {
      "epoch": 1.3224153070184395,
      "grad_norm": 0.5468334467073818,
      "learning_rate": 5.4395999474019434e-05,
      "loss": 11.8378,
      "step": 24285
    },
    {
      "epoch": 1.3224697610150224,
      "grad_norm": 0.5826262109080343,
      "learning_rate": 5.4388151760687033e-05,
      "loss": 12.0424,
      "step": 24286
    },
    {
      "epoch": 1.3225242150116054,
      "grad_norm": 0.5224605350270002,
      "learning_rate": 5.438030440204007e-05,
      "loss": 11.9409,
      "step": 24287
    },
    {
      "epoch": 1.3225786690081884,
      "grad_norm": 0.4771661678720687,
      "learning_rate": 5.437245739813952e-05,
      "loss": 11.8821,
      "step": 24288
    },
    {
      "epoch": 1.3226331230047714,
      "grad_norm": 0.5457161085290173,
      "learning_rate": 5.436461074904634e-05,
      "loss": 11.8993,
      "step": 24289
    },
    {
      "epoch": 1.3226875770013544,
      "grad_norm": 0.5260466621327258,
      "learning_rate": 5.435676445482165e-05,
      "loss": 12.0321,
      "step": 24290
    },
    {
      "epoch": 1.3227420309979376,
      "grad_norm": 0.5487822149381587,
      "learning_rate": 5.4348918515526356e-05,
      "loss": 11.9641,
      "step": 24291
    },
    {
      "epoch": 1.3227964849945206,
      "grad_norm": 0.5397681819089222,
      "learning_rate": 5.434107293122158e-05,
      "loss": 11.8594,
      "step": 24292
    },
    {
      "epoch": 1.3228509389911036,
      "grad_norm": 0.5507672832446914,
      "learning_rate": 5.4333227701968246e-05,
      "loss": 11.9825,
      "step": 24293
    },
    {
      "epoch": 1.3229053929876866,
      "grad_norm": 0.7380789884234731,
      "learning_rate": 5.4325382827827374e-05,
      "loss": 12.0395,
      "step": 24294
    },
    {
      "epoch": 1.3229598469842696,
      "grad_norm": 0.5656365995695363,
      "learning_rate": 5.431753830886004e-05,
      "loss": 11.9359,
      "step": 24295
    },
    {
      "epoch": 1.3230143009808526,
      "grad_norm": 0.5887100710632452,
      "learning_rate": 5.4309694145127145e-05,
      "loss": 11.9868,
      "step": 24296
    },
    {
      "epoch": 1.3230687549774356,
      "grad_norm": 0.5049209463935754,
      "learning_rate": 5.4301850336689755e-05,
      "loss": 11.8749,
      "step": 24297
    },
    {
      "epoch": 1.3231232089740186,
      "grad_norm": 0.544689104187778,
      "learning_rate": 5.429400688360886e-05,
      "loss": 11.9015,
      "step": 24298
    },
    {
      "epoch": 1.3231776629706016,
      "grad_norm": 0.5438795568555299,
      "learning_rate": 5.428616378594538e-05,
      "loss": 11.82,
      "step": 24299
    },
    {
      "epoch": 1.3232321169671848,
      "grad_norm": 0.5328578054970738,
      "learning_rate": 5.427832104376041e-05,
      "loss": 11.7133,
      "step": 24300
    },
    {
      "epoch": 1.3232865709637678,
      "grad_norm": 0.5500810269527809,
      "learning_rate": 5.427047865711483e-05,
      "loss": 11.9908,
      "step": 24301
    },
    {
      "epoch": 1.3233410249603508,
      "grad_norm": 0.6258240133814889,
      "learning_rate": 5.4262636626069715e-05,
      "loss": 11.9089,
      "step": 24302
    },
    {
      "epoch": 1.3233954789569338,
      "grad_norm": 0.5670219671761658,
      "learning_rate": 5.425479495068597e-05,
      "loss": 11.8998,
      "step": 24303
    },
    {
      "epoch": 1.3234499329535168,
      "grad_norm": 0.5993249992658796,
      "learning_rate": 5.4246953631024633e-05,
      "loss": 12.1534,
      "step": 24304
    },
    {
      "epoch": 1.3235043869500998,
      "grad_norm": 0.5772207235736112,
      "learning_rate": 5.4239112667146676e-05,
      "loss": 11.945,
      "step": 24305
    },
    {
      "epoch": 1.3235588409466827,
      "grad_norm": 0.5207581196920541,
      "learning_rate": 5.4231272059113014e-05,
      "loss": 11.925,
      "step": 24306
    },
    {
      "epoch": 1.3236132949432657,
      "grad_norm": 0.6237959400309706,
      "learning_rate": 5.4223431806984706e-05,
      "loss": 11.9534,
      "step": 24307
    },
    {
      "epoch": 1.3236677489398487,
      "grad_norm": 0.6336839598475581,
      "learning_rate": 5.421559191082266e-05,
      "loss": 12.0141,
      "step": 24308
    },
    {
      "epoch": 1.3237222029364317,
      "grad_norm": 0.5943396579206478,
      "learning_rate": 5.42077523706878e-05,
      "loss": 12.0129,
      "step": 24309
    },
    {
      "epoch": 1.3237766569330147,
      "grad_norm": 0.5404487076541026,
      "learning_rate": 5.4199913186641194e-05,
      "loss": 11.9076,
      "step": 24310
    },
    {
      "epoch": 1.3238311109295977,
      "grad_norm": 0.5890314787145994,
      "learning_rate": 5.4192074358743696e-05,
      "loss": 11.961,
      "step": 24311
    },
    {
      "epoch": 1.3238855649261807,
      "grad_norm": 0.5241183180791975,
      "learning_rate": 5.418423588705634e-05,
      "loss": 11.8418,
      "step": 24312
    },
    {
      "epoch": 1.3239400189227637,
      "grad_norm": 0.5657800702247222,
      "learning_rate": 5.417639777164003e-05,
      "loss": 11.9448,
      "step": 24313
    },
    {
      "epoch": 1.3239944729193467,
      "grad_norm": 0.5704038597285788,
      "learning_rate": 5.416856001255577e-05,
      "loss": 11.9572,
      "step": 24314
    },
    {
      "epoch": 1.32404892691593,
      "grad_norm": 0.5161054332427804,
      "learning_rate": 5.4160722609864425e-05,
      "loss": 11.8315,
      "step": 24315
    },
    {
      "epoch": 1.324103380912513,
      "grad_norm": 0.5530415399281876,
      "learning_rate": 5.4152885563626985e-05,
      "loss": 11.7823,
      "step": 24316
    },
    {
      "epoch": 1.3241578349090959,
      "grad_norm": 0.6341196816585852,
      "learning_rate": 5.414504887390446e-05,
      "loss": 12.083,
      "step": 24317
    },
    {
      "epoch": 1.3242122889056789,
      "grad_norm": 0.5866654115954333,
      "learning_rate": 5.41372125407577e-05,
      "loss": 12.0122,
      "step": 24318
    },
    {
      "epoch": 1.3242667429022619,
      "grad_norm": 0.5358296736449896,
      "learning_rate": 5.4129376564247635e-05,
      "loss": 11.8467,
      "step": 24319
    },
    {
      "epoch": 1.3243211968988449,
      "grad_norm": 0.5614961187626117,
      "learning_rate": 5.412154094443527e-05,
      "loss": 12.1306,
      "step": 24320
    },
    {
      "epoch": 1.3243756508954279,
      "grad_norm": 0.5591502289964349,
      "learning_rate": 5.4113705681381455e-05,
      "loss": 11.8171,
      "step": 24321
    },
    {
      "epoch": 1.3244301048920109,
      "grad_norm": 0.5498758234731576,
      "learning_rate": 5.41058707751472e-05,
      "loss": 11.8002,
      "step": 24322
    },
    {
      "epoch": 1.324484558888594,
      "grad_norm": 0.5196810290051165,
      "learning_rate": 5.409803622579335e-05,
      "loss": 11.9014,
      "step": 24323
    },
    {
      "epoch": 1.324539012885177,
      "grad_norm": 0.4875341647826073,
      "learning_rate": 5.409020203338091e-05,
      "loss": 11.9946,
      "step": 24324
    },
    {
      "epoch": 1.32459346688176,
      "grad_norm": 0.5962898438983585,
      "learning_rate": 5.4082368197970724e-05,
      "loss": 11.8129,
      "step": 24325
    },
    {
      "epoch": 1.324647920878343,
      "grad_norm": 0.5242946635270267,
      "learning_rate": 5.407453471962377e-05,
      "loss": 12.0071,
      "step": 24326
    },
    {
      "epoch": 1.324702374874926,
      "grad_norm": 0.5151795242568236,
      "learning_rate": 5.4066701598400904e-05,
      "loss": 11.8575,
      "step": 24327
    },
    {
      "epoch": 1.324756828871509,
      "grad_norm": 0.508533860739261,
      "learning_rate": 5.405886883436311e-05,
      "loss": 11.9134,
      "step": 24328
    },
    {
      "epoch": 1.324811282868092,
      "grad_norm": 0.5317603989694534,
      "learning_rate": 5.4051036427571254e-05,
      "loss": 11.8262,
      "step": 24329
    },
    {
      "epoch": 1.324865736864675,
      "grad_norm": 0.5590911167033477,
      "learning_rate": 5.404320437808621e-05,
      "loss": 11.9181,
      "step": 24330
    },
    {
      "epoch": 1.324920190861258,
      "grad_norm": 0.5598679141628443,
      "learning_rate": 5.40353726859689e-05,
      "loss": 11.9615,
      "step": 24331
    },
    {
      "epoch": 1.324974644857841,
      "grad_norm": 0.5136271129954619,
      "learning_rate": 5.40275413512803e-05,
      "loss": 11.8343,
      "step": 24332
    },
    {
      "epoch": 1.325029098854424,
      "grad_norm": 0.5734043586137014,
      "learning_rate": 5.401971037408119e-05,
      "loss": 11.9757,
      "step": 24333
    },
    {
      "epoch": 1.325083552851007,
      "grad_norm": 0.5598120208934344,
      "learning_rate": 5.401187975443257e-05,
      "loss": 11.9283,
      "step": 24334
    },
    {
      "epoch": 1.32513800684759,
      "grad_norm": 0.5448110291910014,
      "learning_rate": 5.400404949239525e-05,
      "loss": 11.8949,
      "step": 24335
    },
    {
      "epoch": 1.325192460844173,
      "grad_norm": 0.5642133229271152,
      "learning_rate": 5.399621958803018e-05,
      "loss": 11.9411,
      "step": 24336
    },
    {
      "epoch": 1.325246914840756,
      "grad_norm": 0.5177282489550481,
      "learning_rate": 5.39883900413982e-05,
      "loss": 11.9441,
      "step": 24337
    },
    {
      "epoch": 1.3253013688373392,
      "grad_norm": 0.5626101259603494,
      "learning_rate": 5.3980560852560244e-05,
      "loss": 11.9748,
      "step": 24338
    },
    {
      "epoch": 1.3253558228339222,
      "grad_norm": 0.524196397718115,
      "learning_rate": 5.3972732021577175e-05,
      "loss": 11.857,
      "step": 24339
    },
    {
      "epoch": 1.3254102768305052,
      "grad_norm": 0.5694045290272125,
      "learning_rate": 5.3964903548509816e-05,
      "loss": 11.9094,
      "step": 24340
    },
    {
      "epoch": 1.3254647308270882,
      "grad_norm": 0.5012759947368628,
      "learning_rate": 5.3957075433419126e-05,
      "loss": 11.9684,
      "step": 24341
    },
    {
      "epoch": 1.3255191848236711,
      "grad_norm": 0.5424842903304663,
      "learning_rate": 5.394924767636589e-05,
      "loss": 11.8677,
      "step": 24342
    },
    {
      "epoch": 1.3255736388202541,
      "grad_norm": 0.5094963885154813,
      "learning_rate": 5.3941420277411035e-05,
      "loss": 11.9124,
      "step": 24343
    },
    {
      "epoch": 1.3256280928168371,
      "grad_norm": 0.5695293612183323,
      "learning_rate": 5.3933593236615465e-05,
      "loss": 11.7505,
      "step": 24344
    },
    {
      "epoch": 1.3256825468134201,
      "grad_norm": 0.5244664486741112,
      "learning_rate": 5.392576655403996e-05,
      "loss": 11.8237,
      "step": 24345
    },
    {
      "epoch": 1.3257370008100031,
      "grad_norm": 0.5433440505139582,
      "learning_rate": 5.3917940229745446e-05,
      "loss": 11.9967,
      "step": 24346
    },
    {
      "epoch": 1.3257914548065863,
      "grad_norm": 0.5418608453655283,
      "learning_rate": 5.3910114263792734e-05,
      "loss": 11.9847,
      "step": 24347
    },
    {
      "epoch": 1.3258459088031693,
      "grad_norm": 0.5146203601815554,
      "learning_rate": 5.3902288656242714e-05,
      "loss": 12.0123,
      "step": 24348
    },
    {
      "epoch": 1.3259003627997523,
      "grad_norm": 0.6105695683271954,
      "learning_rate": 5.389446340715626e-05,
      "loss": 11.8639,
      "step": 24349
    },
    {
      "epoch": 1.3259548167963353,
      "grad_norm": 0.5043270843253015,
      "learning_rate": 5.3886638516594136e-05,
      "loss": 11.9484,
      "step": 24350
    },
    {
      "epoch": 1.3260092707929183,
      "grad_norm": 0.6502463641638829,
      "learning_rate": 5.3878813984617294e-05,
      "loss": 12.0225,
      "step": 24351
    },
    {
      "epoch": 1.3260637247895013,
      "grad_norm": 0.5707764957435898,
      "learning_rate": 5.387098981128648e-05,
      "loss": 11.9455,
      "step": 24352
    },
    {
      "epoch": 1.3261181787860843,
      "grad_norm": 0.5853274073453641,
      "learning_rate": 5.386316599666259e-05,
      "loss": 11.9512,
      "step": 24353
    },
    {
      "epoch": 1.3261726327826673,
      "grad_norm": 0.5331348033249079,
      "learning_rate": 5.385534254080649e-05,
      "loss": 11.8601,
      "step": 24354
    },
    {
      "epoch": 1.3262270867792503,
      "grad_norm": 0.535285651695918,
      "learning_rate": 5.3847519443778946e-05,
      "loss": 11.7711,
      "step": 24355
    },
    {
      "epoch": 1.3262815407758333,
      "grad_norm": 0.5094922793687584,
      "learning_rate": 5.3839696705640876e-05,
      "loss": 11.8656,
      "step": 24356
    },
    {
      "epoch": 1.3263359947724163,
      "grad_norm": 0.549080849299552,
      "learning_rate": 5.383187432645303e-05,
      "loss": 11.8211,
      "step": 24357
    },
    {
      "epoch": 1.3263904487689993,
      "grad_norm": 0.5365808510333939,
      "learning_rate": 5.382405230627629e-05,
      "loss": 11.9175,
      "step": 24358
    },
    {
      "epoch": 1.3264449027655822,
      "grad_norm": 0.6795797989768289,
      "learning_rate": 5.3816230645171494e-05,
      "loss": 12.0065,
      "step": 24359
    },
    {
      "epoch": 1.3264993567621652,
      "grad_norm": 0.6255954293734473,
      "learning_rate": 5.3808409343199376e-05,
      "loss": 11.967,
      "step": 24360
    },
    {
      "epoch": 1.3265538107587485,
      "grad_norm": 0.6366357279444079,
      "learning_rate": 5.380058840042085e-05,
      "loss": 11.8996,
      "step": 24361
    },
    {
      "epoch": 1.3266082647553314,
      "grad_norm": 0.6546017019244861,
      "learning_rate": 5.379276781689666e-05,
      "loss": 12.0854,
      "step": 24362
    },
    {
      "epoch": 1.3266627187519144,
      "grad_norm": 0.5896448603649662,
      "learning_rate": 5.3784947592687706e-05,
      "loss": 11.9614,
      "step": 24363
    },
    {
      "epoch": 1.3267171727484974,
      "grad_norm": 0.5935185979095843,
      "learning_rate": 5.3777127727854704e-05,
      "loss": 11.9514,
      "step": 24364
    },
    {
      "epoch": 1.3267716267450804,
      "grad_norm": 0.5664400696177851,
      "learning_rate": 5.3769308222458495e-05,
      "loss": 11.9728,
      "step": 24365
    },
    {
      "epoch": 1.3268260807416634,
      "grad_norm": 0.6030275831908156,
      "learning_rate": 5.3761489076559954e-05,
      "loss": 11.9886,
      "step": 24366
    },
    {
      "epoch": 1.3268805347382464,
      "grad_norm": 0.5607611995842933,
      "learning_rate": 5.3753670290219784e-05,
      "loss": 11.8956,
      "step": 24367
    },
    {
      "epoch": 1.3269349887348294,
      "grad_norm": 0.6031829565266845,
      "learning_rate": 5.374585186349888e-05,
      "loss": 12.0342,
      "step": 24368
    },
    {
      "epoch": 1.3269894427314124,
      "grad_norm": 0.5511490707745242,
      "learning_rate": 5.373803379645797e-05,
      "loss": 11.9302,
      "step": 24369
    },
    {
      "epoch": 1.3270438967279956,
      "grad_norm": 0.5238988219315068,
      "learning_rate": 5.373021608915783e-05,
      "loss": 11.92,
      "step": 24370
    },
    {
      "epoch": 1.3270983507245786,
      "grad_norm": 0.6318560091882891,
      "learning_rate": 5.372239874165934e-05,
      "loss": 11.8764,
      "step": 24371
    },
    {
      "epoch": 1.3271528047211616,
      "grad_norm": 0.5305198440793832,
      "learning_rate": 5.3714581754023184e-05,
      "loss": 11.9477,
      "step": 24372
    },
    {
      "epoch": 1.3272072587177446,
      "grad_norm": 0.5510571124139215,
      "learning_rate": 5.3706765126310254e-05,
      "loss": 12.007,
      "step": 24373
    },
    {
      "epoch": 1.3272617127143276,
      "grad_norm": 0.5449207675209723,
      "learning_rate": 5.3698948858581245e-05,
      "loss": 11.8913,
      "step": 24374
    },
    {
      "epoch": 1.3273161667109106,
      "grad_norm": 0.5610527251170293,
      "learning_rate": 5.369113295089696e-05,
      "loss": 11.868,
      "step": 24375
    },
    {
      "epoch": 1.3273706207074936,
      "grad_norm": 0.4940865860536345,
      "learning_rate": 5.368331740331824e-05,
      "loss": 11.8369,
      "step": 24376
    },
    {
      "epoch": 1.3274250747040766,
      "grad_norm": 0.5277221793351025,
      "learning_rate": 5.367550221590576e-05,
      "loss": 11.8139,
      "step": 24377
    },
    {
      "epoch": 1.3274795287006596,
      "grad_norm": 0.5952348535984634,
      "learning_rate": 5.3667687388720434e-05,
      "loss": 11.9857,
      "step": 24378
    },
    {
      "epoch": 1.3275339826972425,
      "grad_norm": 0.5445620574401022,
      "learning_rate": 5.365987292182286e-05,
      "loss": 11.8433,
      "step": 24379
    },
    {
      "epoch": 1.3275884366938255,
      "grad_norm": 0.5296148059847346,
      "learning_rate": 5.365205881527389e-05,
      "loss": 11.8902,
      "step": 24380
    },
    {
      "epoch": 1.3276428906904085,
      "grad_norm": 0.5251909452213419,
      "learning_rate": 5.364424506913433e-05,
      "loss": 11.9295,
      "step": 24381
    },
    {
      "epoch": 1.3276973446869915,
      "grad_norm": 0.530573351604225,
      "learning_rate": 5.363643168346484e-05,
      "loss": 11.9302,
      "step": 24382
    },
    {
      "epoch": 1.3277517986835745,
      "grad_norm": 0.54045335631268,
      "learning_rate": 5.362861865832627e-05,
      "loss": 11.9627,
      "step": 24383
    },
    {
      "epoch": 1.3278062526801575,
      "grad_norm": 0.612712364336341,
      "learning_rate": 5.36208059937793e-05,
      "loss": 12.139,
      "step": 24384
    },
    {
      "epoch": 1.3278607066767407,
      "grad_norm": 0.5417724530265201,
      "learning_rate": 5.361299368988477e-05,
      "loss": 12.0772,
      "step": 24385
    },
    {
      "epoch": 1.3279151606733237,
      "grad_norm": 0.6751105267403107,
      "learning_rate": 5.3605181746703325e-05,
      "loss": 11.9898,
      "step": 24386
    },
    {
      "epoch": 1.3279696146699067,
      "grad_norm": 0.615979215688675,
      "learning_rate": 5.3597370164295776e-05,
      "loss": 11.9471,
      "step": 24387
    },
    {
      "epoch": 1.3280240686664897,
      "grad_norm": 0.5665303733879382,
      "learning_rate": 5.358955894272295e-05,
      "loss": 11.9218,
      "step": 24388
    },
    {
      "epoch": 1.3280785226630727,
      "grad_norm": 0.5351548006752035,
      "learning_rate": 5.3581748082045394e-05,
      "loss": 11.9376,
      "step": 24389
    },
    {
      "epoch": 1.3281329766596557,
      "grad_norm": 0.5269036829983818,
      "learning_rate": 5.3573937582324004e-05,
      "loss": 11.9498,
      "step": 24390
    },
    {
      "epoch": 1.3281874306562387,
      "grad_norm": 0.5086667444009628,
      "learning_rate": 5.356612744361942e-05,
      "loss": 12.0167,
      "step": 24391
    },
    {
      "epoch": 1.3282418846528217,
      "grad_norm": 0.5647739110462783,
      "learning_rate": 5.355831766599242e-05,
      "loss": 11.8642,
      "step": 24392
    },
    {
      "epoch": 1.3282963386494049,
      "grad_norm": 0.5607725796111824,
      "learning_rate": 5.355050824950376e-05,
      "loss": 11.9575,
      "step": 24393
    },
    {
      "epoch": 1.3283507926459879,
      "grad_norm": 0.5592590286363692,
      "learning_rate": 5.35426991942141e-05,
      "loss": 12.0412,
      "step": 24394
    },
    {
      "epoch": 1.3284052466425709,
      "grad_norm": 0.5414799946184244,
      "learning_rate": 5.353489050018423e-05,
      "loss": 11.8695,
      "step": 24395
    },
    {
      "epoch": 1.3284597006391539,
      "grad_norm": 0.5355594770259369,
      "learning_rate": 5.352708216747482e-05,
      "loss": 11.9433,
      "step": 24396
    },
    {
      "epoch": 1.3285141546357369,
      "grad_norm": 0.5112299339077214,
      "learning_rate": 5.351927419614665e-05,
      "loss": 11.8644,
      "step": 24397
    },
    {
      "epoch": 1.3285686086323198,
      "grad_norm": 0.6137036904474334,
      "learning_rate": 5.351146658626041e-05,
      "loss": 11.9186,
      "step": 24398
    },
    {
      "epoch": 1.3286230626289028,
      "grad_norm": 0.6907624157770516,
      "learning_rate": 5.3503659337876735e-05,
      "loss": 11.9553,
      "step": 24399
    },
    {
      "epoch": 1.3286775166254858,
      "grad_norm": 0.5539291518076767,
      "learning_rate": 5.349585245105646e-05,
      "loss": 11.9558,
      "step": 24400
    },
    {
      "epoch": 1.3287319706220688,
      "grad_norm": 0.51189330897039,
      "learning_rate": 5.348804592586019e-05,
      "loss": 11.8123,
      "step": 24401
    },
    {
      "epoch": 1.3287864246186518,
      "grad_norm": 0.5444129357563464,
      "learning_rate": 5.3480239762348684e-05,
      "loss": 11.8758,
      "step": 24402
    },
    {
      "epoch": 1.3288408786152348,
      "grad_norm": 0.6037881277554985,
      "learning_rate": 5.347243396058266e-05,
      "loss": 12.0076,
      "step": 24403
    },
    {
      "epoch": 1.3288953326118178,
      "grad_norm": 0.546067934881867,
      "learning_rate": 5.346462852062277e-05,
      "loss": 12.0116,
      "step": 24404
    },
    {
      "epoch": 1.3289497866084008,
      "grad_norm": 0.48186776905170187,
      "learning_rate": 5.345682344252976e-05,
      "loss": 11.8478,
      "step": 24405
    },
    {
      "epoch": 1.3290042406049838,
      "grad_norm": 0.5434736762770173,
      "learning_rate": 5.344901872636425e-05,
      "loss": 11.8911,
      "step": 24406
    },
    {
      "epoch": 1.3290586946015668,
      "grad_norm": 0.6060622428524457,
      "learning_rate": 5.3441214372187033e-05,
      "loss": 11.9495,
      "step": 24407
    },
    {
      "epoch": 1.32911314859815,
      "grad_norm": 0.5304292446815432,
      "learning_rate": 5.343341038005873e-05,
      "loss": 12.0005,
      "step": 24408
    },
    {
      "epoch": 1.329167602594733,
      "grad_norm": 0.677423001758076,
      "learning_rate": 5.342560675004e-05,
      "loss": 12.0136,
      "step": 24409
    },
    {
      "epoch": 1.329222056591316,
      "grad_norm": 0.5518555001602644,
      "learning_rate": 5.341780348219161e-05,
      "loss": 12.0136,
      "step": 24410
    },
    {
      "epoch": 1.329276510587899,
      "grad_norm": 0.500025501783482,
      "learning_rate": 5.341000057657415e-05,
      "loss": 11.8849,
      "step": 24411
    },
    {
      "epoch": 1.329330964584482,
      "grad_norm": 0.5974749784099989,
      "learning_rate": 5.340219803324838e-05,
      "loss": 11.8318,
      "step": 24412
    },
    {
      "epoch": 1.329385418581065,
      "grad_norm": 0.5510432659653153,
      "learning_rate": 5.339439585227488e-05,
      "loss": 11.807,
      "step": 24413
    },
    {
      "epoch": 1.329439872577648,
      "grad_norm": 0.5955238957795417,
      "learning_rate": 5.3386594033714376e-05,
      "loss": 11.9354,
      "step": 24414
    },
    {
      "epoch": 1.329494326574231,
      "grad_norm": 0.5705872812703834,
      "learning_rate": 5.337879257762758e-05,
      "loss": 11.9356,
      "step": 24415
    },
    {
      "epoch": 1.329548780570814,
      "grad_norm": 0.5612735588298833,
      "learning_rate": 5.337099148407507e-05,
      "loss": 11.856,
      "step": 24416
    },
    {
      "epoch": 1.3296032345673972,
      "grad_norm": 0.5406940615655401,
      "learning_rate": 5.336319075311759e-05,
      "loss": 11.9474,
      "step": 24417
    },
    {
      "epoch": 1.3296576885639801,
      "grad_norm": 0.6178418823145054,
      "learning_rate": 5.335539038481575e-05,
      "loss": 12.0239,
      "step": 24418
    },
    {
      "epoch": 1.3297121425605631,
      "grad_norm": 0.5790966834105574,
      "learning_rate": 5.3347590379230186e-05,
      "loss": 11.9535,
      "step": 24419
    },
    {
      "epoch": 1.3297665965571461,
      "grad_norm": 0.5336972842510217,
      "learning_rate": 5.3339790736421625e-05,
      "loss": 11.9026,
      "step": 24420
    },
    {
      "epoch": 1.3298210505537291,
      "grad_norm": 0.5112016277825095,
      "learning_rate": 5.333199145645064e-05,
      "loss": 11.9398,
      "step": 24421
    },
    {
      "epoch": 1.3298755045503121,
      "grad_norm": 0.5677668095965251,
      "learning_rate": 5.332419253937795e-05,
      "loss": 11.8973,
      "step": 24422
    },
    {
      "epoch": 1.329929958546895,
      "grad_norm": 0.5846641795364168,
      "learning_rate": 5.3316393985264114e-05,
      "loss": 11.867,
      "step": 24423
    },
    {
      "epoch": 1.329984412543478,
      "grad_norm": 0.5615385904841783,
      "learning_rate": 5.3308595794169844e-05,
      "loss": 11.9397,
      "step": 24424
    },
    {
      "epoch": 1.330038866540061,
      "grad_norm": 0.5154131337766664,
      "learning_rate": 5.330079796615579e-05,
      "loss": 11.8674,
      "step": 24425
    },
    {
      "epoch": 1.330093320536644,
      "grad_norm": 0.5112974569783938,
      "learning_rate": 5.329300050128254e-05,
      "loss": 11.8693,
      "step": 24426
    },
    {
      "epoch": 1.330147774533227,
      "grad_norm": 0.547558573878345,
      "learning_rate": 5.328520339961078e-05,
      "loss": 11.8623,
      "step": 24427
    },
    {
      "epoch": 1.33020222852981,
      "grad_norm": 0.562825292170743,
      "learning_rate": 5.327740666120107e-05,
      "loss": 11.8996,
      "step": 24428
    },
    {
      "epoch": 1.330256682526393,
      "grad_norm": 0.5582074920459502,
      "learning_rate": 5.3269610286114126e-05,
      "loss": 11.7927,
      "step": 24429
    },
    {
      "epoch": 1.330311136522976,
      "grad_norm": 0.5180893200089595,
      "learning_rate": 5.3261814274410523e-05,
      "loss": 11.8159,
      "step": 24430
    },
    {
      "epoch": 1.3303655905195593,
      "grad_norm": 0.6306627129274233,
      "learning_rate": 5.325401862615087e-05,
      "loss": 11.8643,
      "step": 24431
    },
    {
      "epoch": 1.3304200445161423,
      "grad_norm": 0.5500720777771656,
      "learning_rate": 5.324622334139583e-05,
      "loss": 11.8704,
      "step": 24432
    },
    {
      "epoch": 1.3304744985127253,
      "grad_norm": 0.53183988508379,
      "learning_rate": 5.323842842020597e-05,
      "loss": 11.8969,
      "step": 24433
    },
    {
      "epoch": 1.3305289525093082,
      "grad_norm": 0.5738645551365829,
      "learning_rate": 5.323063386264196e-05,
      "loss": 11.8925,
      "step": 24434
    },
    {
      "epoch": 1.3305834065058912,
      "grad_norm": 0.5545923361344481,
      "learning_rate": 5.3222839668764335e-05,
      "loss": 11.8049,
      "step": 24435
    },
    {
      "epoch": 1.3306378605024742,
      "grad_norm": 0.534822476559011,
      "learning_rate": 5.321504583863377e-05,
      "loss": 11.8954,
      "step": 24436
    },
    {
      "epoch": 1.3306923144990572,
      "grad_norm": 0.5072876371324885,
      "learning_rate": 5.320725237231089e-05,
      "loss": 11.9209,
      "step": 24437
    },
    {
      "epoch": 1.3307467684956402,
      "grad_norm": 0.5403614508507687,
      "learning_rate": 5.319945926985622e-05,
      "loss": 11.897,
      "step": 24438
    },
    {
      "epoch": 1.3308012224922232,
      "grad_norm": 0.5495669396723336,
      "learning_rate": 5.319166653133049e-05,
      "loss": 11.9898,
      "step": 24439
    },
    {
      "epoch": 1.3308556764888064,
      "grad_norm": 0.49418161704933866,
      "learning_rate": 5.318387415679413e-05,
      "loss": 11.7318,
      "step": 24440
    },
    {
      "epoch": 1.3309101304853894,
      "grad_norm": 0.5258276724310946,
      "learning_rate": 5.31760821463078e-05,
      "loss": 11.7775,
      "step": 24441
    },
    {
      "epoch": 1.3309645844819724,
      "grad_norm": 0.5363831551349146,
      "learning_rate": 5.316829049993217e-05,
      "loss": 11.9428,
      "step": 24442
    },
    {
      "epoch": 1.3310190384785554,
      "grad_norm": 0.5348364804923785,
      "learning_rate": 5.316049921772772e-05,
      "loss": 11.85,
      "step": 24443
    },
    {
      "epoch": 1.3310734924751384,
      "grad_norm": 0.5216952354541312,
      "learning_rate": 5.315270829975512e-05,
      "loss": 11.9201,
      "step": 24444
    },
    {
      "epoch": 1.3311279464717214,
      "grad_norm": 0.6610065133816551,
      "learning_rate": 5.314491774607487e-05,
      "loss": 11.9976,
      "step": 24445
    },
    {
      "epoch": 1.3311824004683044,
      "grad_norm": 0.5636125963597513,
      "learning_rate": 5.3137127556747645e-05,
      "loss": 12.0046,
      "step": 24446
    },
    {
      "epoch": 1.3312368544648874,
      "grad_norm": 0.5023090693827555,
      "learning_rate": 5.3129337731833926e-05,
      "loss": 11.9616,
      "step": 24447
    },
    {
      "epoch": 1.3312913084614704,
      "grad_norm": 0.6430240866603419,
      "learning_rate": 5.312154827139435e-05,
      "loss": 11.9319,
      "step": 24448
    },
    {
      "epoch": 1.3313457624580534,
      "grad_norm": 0.5633246444295514,
      "learning_rate": 5.3113759175489554e-05,
      "loss": 11.9111,
      "step": 24449
    },
    {
      "epoch": 1.3314002164546364,
      "grad_norm": 0.5242721961133227,
      "learning_rate": 5.310597044417995e-05,
      "loss": 11.9042,
      "step": 24450
    },
    {
      "epoch": 1.3314546704512193,
      "grad_norm": 0.5296018505627887,
      "learning_rate": 5.309818207752617e-05,
      "loss": 11.9196,
      "step": 24451
    },
    {
      "epoch": 1.3315091244478023,
      "grad_norm": 0.6046831151338544,
      "learning_rate": 5.309039407558885e-05,
      "loss": 11.9763,
      "step": 24452
    },
    {
      "epoch": 1.3315635784443853,
      "grad_norm": 0.5651034819341412,
      "learning_rate": 5.3082606438428437e-05,
      "loss": 11.8477,
      "step": 24453
    },
    {
      "epoch": 1.3316180324409685,
      "grad_norm": 0.5141442487505451,
      "learning_rate": 5.3074819166105594e-05,
      "loss": 11.6833,
      "step": 24454
    },
    {
      "epoch": 1.3316724864375515,
      "grad_norm": 0.49895548784066357,
      "learning_rate": 5.306703225868079e-05,
      "loss": 11.7769,
      "step": 24455
    },
    {
      "epoch": 1.3317269404341345,
      "grad_norm": 0.5204968497412266,
      "learning_rate": 5.305924571621464e-05,
      "loss": 11.8639,
      "step": 24456
    },
    {
      "epoch": 1.3317813944307175,
      "grad_norm": 0.5857602081241274,
      "learning_rate": 5.3051459538767645e-05,
      "loss": 11.9054,
      "step": 24457
    },
    {
      "epoch": 1.3318358484273005,
      "grad_norm": 0.5659551784629318,
      "learning_rate": 5.304367372640035e-05,
      "loss": 12.0636,
      "step": 24458
    },
    {
      "epoch": 1.3318903024238835,
      "grad_norm": 0.5773041733355975,
      "learning_rate": 5.303588827917343e-05,
      "loss": 12.0934,
      "step": 24459
    },
    {
      "epoch": 1.3319447564204665,
      "grad_norm": 0.5280070943189968,
      "learning_rate": 5.3028103197147226e-05,
      "loss": 11.9716,
      "step": 24460
    },
    {
      "epoch": 1.3319992104170495,
      "grad_norm": 0.5735275222348536,
      "learning_rate": 5.3020318480382404e-05,
      "loss": 11.8713,
      "step": 24461
    },
    {
      "epoch": 1.3320536644136325,
      "grad_norm": 0.48462536419252333,
      "learning_rate": 5.301253412893943e-05,
      "loss": 11.8588,
      "step": 24462
    },
    {
      "epoch": 1.3321081184102157,
      "grad_norm": 0.5338053619619285,
      "learning_rate": 5.300475014287887e-05,
      "loss": 11.8625,
      "step": 24463
    },
    {
      "epoch": 1.3321625724067987,
      "grad_norm": 0.5613163297630709,
      "learning_rate": 5.299696652226129e-05,
      "loss": 11.9371,
      "step": 24464
    },
    {
      "epoch": 1.3322170264033817,
      "grad_norm": 0.5626807947280761,
      "learning_rate": 5.298918326714715e-05,
      "loss": 11.8772,
      "step": 24465
    },
    {
      "epoch": 1.3322714803999647,
      "grad_norm": 0.601793737070652,
      "learning_rate": 5.2981400377597036e-05,
      "loss": 11.9692,
      "step": 24466
    },
    {
      "epoch": 1.3323259343965477,
      "grad_norm": 0.5741701224189307,
      "learning_rate": 5.29736178536714e-05,
      "loss": 11.901,
      "step": 24467
    },
    {
      "epoch": 1.3323803883931307,
      "grad_norm": 0.5736657088276385,
      "learning_rate": 5.296583569543083e-05,
      "loss": 11.9024,
      "step": 24468
    },
    {
      "epoch": 1.3324348423897137,
      "grad_norm": 0.5978380892139845,
      "learning_rate": 5.295805390293582e-05,
      "loss": 11.9477,
      "step": 24469
    },
    {
      "epoch": 1.3324892963862967,
      "grad_norm": 0.5819745812226959,
      "learning_rate": 5.295027247624683e-05,
      "loss": 12.0241,
      "step": 24470
    },
    {
      "epoch": 1.3325437503828796,
      "grad_norm": 0.5016597715091647,
      "learning_rate": 5.294249141542444e-05,
      "loss": 11.8833,
      "step": 24471
    },
    {
      "epoch": 1.3325982043794626,
      "grad_norm": 0.5297508849514344,
      "learning_rate": 5.293471072052911e-05,
      "loss": 11.8493,
      "step": 24472
    },
    {
      "epoch": 1.3326526583760456,
      "grad_norm": 0.5317063762016192,
      "learning_rate": 5.292693039162135e-05,
      "loss": 11.8721,
      "step": 24473
    },
    {
      "epoch": 1.3327071123726286,
      "grad_norm": 0.5432951166550877,
      "learning_rate": 5.291915042876172e-05,
      "loss": 11.8806,
      "step": 24474
    },
    {
      "epoch": 1.3327615663692116,
      "grad_norm": 0.5898378762569593,
      "learning_rate": 5.291137083201062e-05,
      "loss": 11.969,
      "step": 24475
    },
    {
      "epoch": 1.3328160203657946,
      "grad_norm": 0.571938426440639,
      "learning_rate": 5.290359160142864e-05,
      "loss": 11.8397,
      "step": 24476
    },
    {
      "epoch": 1.3328704743623776,
      "grad_norm": 0.5815525163283235,
      "learning_rate": 5.28958127370762e-05,
      "loss": 11.9098,
      "step": 24477
    },
    {
      "epoch": 1.3329249283589608,
      "grad_norm": 0.5968767462739033,
      "learning_rate": 5.288803423901385e-05,
      "loss": 11.9672,
      "step": 24478
    },
    {
      "epoch": 1.3329793823555438,
      "grad_norm": 0.5754187193247214,
      "learning_rate": 5.288025610730205e-05,
      "loss": 11.9174,
      "step": 24479
    },
    {
      "epoch": 1.3330338363521268,
      "grad_norm": 0.5898023123021086,
      "learning_rate": 5.287247834200124e-05,
      "loss": 11.9645,
      "step": 24480
    },
    {
      "epoch": 1.3330882903487098,
      "grad_norm": 0.5287901927630678,
      "learning_rate": 5.2864700943171975e-05,
      "loss": 11.9177,
      "step": 24481
    },
    {
      "epoch": 1.3331427443452928,
      "grad_norm": 0.6187442586926514,
      "learning_rate": 5.285692391087467e-05,
      "loss": 11.9769,
      "step": 24482
    },
    {
      "epoch": 1.3331971983418758,
      "grad_norm": 0.5660317088857969,
      "learning_rate": 5.2849147245169875e-05,
      "loss": 12.0203,
      "step": 24483
    },
    {
      "epoch": 1.3332516523384588,
      "grad_norm": 0.5753054490269349,
      "learning_rate": 5.2841370946117963e-05,
      "loss": 11.9575,
      "step": 24484
    },
    {
      "epoch": 1.3333061063350418,
      "grad_norm": 0.4944736485581836,
      "learning_rate": 5.2833595013779466e-05,
      "loss": 11.9116,
      "step": 24485
    },
    {
      "epoch": 1.333360560331625,
      "grad_norm": 0.4940920220088117,
      "learning_rate": 5.282581944821489e-05,
      "loss": 11.8072,
      "step": 24486
    },
    {
      "epoch": 1.333415014328208,
      "grad_norm": 0.5380199073446132,
      "learning_rate": 5.28180442494846e-05,
      "loss": 11.9904,
      "step": 24487
    },
    {
      "epoch": 1.333469468324791,
      "grad_norm": 0.6578582201421894,
      "learning_rate": 5.281026941764916e-05,
      "loss": 12.122,
      "step": 24488
    },
    {
      "epoch": 1.333523922321374,
      "grad_norm": 0.5837883037327447,
      "learning_rate": 5.280249495276896e-05,
      "loss": 11.8773,
      "step": 24489
    },
    {
      "epoch": 1.333578376317957,
      "grad_norm": 0.5433843728532075,
      "learning_rate": 5.2794720854904464e-05,
      "loss": 11.9688,
      "step": 24490
    },
    {
      "epoch": 1.33363283031454,
      "grad_norm": 0.560986084298655,
      "learning_rate": 5.278694712411616e-05,
      "loss": 12.0565,
      "step": 24491
    },
    {
      "epoch": 1.333687284311123,
      "grad_norm": 0.515790404395713,
      "learning_rate": 5.277917376046444e-05,
      "loss": 11.9032,
      "step": 24492
    },
    {
      "epoch": 1.333741738307706,
      "grad_norm": 0.55056638242684,
      "learning_rate": 5.2771400764009814e-05,
      "loss": 11.9398,
      "step": 24493
    },
    {
      "epoch": 1.333796192304289,
      "grad_norm": 0.5499523177854556,
      "learning_rate": 5.276362813481266e-05,
      "loss": 11.8566,
      "step": 24494
    },
    {
      "epoch": 1.333850646300872,
      "grad_norm": 0.5444431428316789,
      "learning_rate": 5.2755855872933505e-05,
      "loss": 11.9437,
      "step": 24495
    },
    {
      "epoch": 1.333905100297455,
      "grad_norm": 0.5484030889653387,
      "learning_rate": 5.274808397843267e-05,
      "loss": 11.78,
      "step": 24496
    },
    {
      "epoch": 1.333959554294038,
      "grad_norm": 0.5904625367617514,
      "learning_rate": 5.274031245137068e-05,
      "loss": 11.8786,
      "step": 24497
    },
    {
      "epoch": 1.334014008290621,
      "grad_norm": 0.5597318931763078,
      "learning_rate": 5.273254129180798e-05,
      "loss": 11.9085,
      "step": 24498
    },
    {
      "epoch": 1.3340684622872039,
      "grad_norm": 0.5368922913995148,
      "learning_rate": 5.2724770499804975e-05,
      "loss": 11.8652,
      "step": 24499
    },
    {
      "epoch": 1.3341229162837869,
      "grad_norm": 0.5684830044239768,
      "learning_rate": 5.2717000075422026e-05,
      "loss": 11.9427,
      "step": 24500
    },
    {
      "epoch": 1.33417737028037,
      "grad_norm": 0.5291246322300477,
      "learning_rate": 5.270923001871967e-05,
      "loss": 11.8035,
      "step": 24501
    },
    {
      "epoch": 1.334231824276953,
      "grad_norm": 0.7121779788468687,
      "learning_rate": 5.270146032975822e-05,
      "loss": 12.0363,
      "step": 24502
    },
    {
      "epoch": 1.334286278273536,
      "grad_norm": 0.6311172161278445,
      "learning_rate": 5.26936910085982e-05,
      "loss": 11.7624,
      "step": 24503
    },
    {
      "epoch": 1.334340732270119,
      "grad_norm": 0.5395628372007741,
      "learning_rate": 5.268592205529992e-05,
      "loss": 11.9868,
      "step": 24504
    },
    {
      "epoch": 1.334395186266702,
      "grad_norm": 0.5465804213449359,
      "learning_rate": 5.267815346992388e-05,
      "loss": 11.9089,
      "step": 24505
    },
    {
      "epoch": 1.334449640263285,
      "grad_norm": 0.5342797119889956,
      "learning_rate": 5.267038525253043e-05,
      "loss": 11.8063,
      "step": 24506
    },
    {
      "epoch": 1.334504094259868,
      "grad_norm": 0.5615334252366635,
      "learning_rate": 5.266261740317999e-05,
      "loss": 11.9759,
      "step": 24507
    },
    {
      "epoch": 1.334558548256451,
      "grad_norm": 0.5544772304893303,
      "learning_rate": 5.265484992193301e-05,
      "loss": 11.7509,
      "step": 24508
    },
    {
      "epoch": 1.334613002253034,
      "grad_norm": 0.6759601781059053,
      "learning_rate": 5.264708280884988e-05,
      "loss": 12.0361,
      "step": 24509
    },
    {
      "epoch": 1.3346674562496172,
      "grad_norm": 0.5188243734877167,
      "learning_rate": 5.2639316063990954e-05,
      "loss": 11.8241,
      "step": 24510
    },
    {
      "epoch": 1.3347219102462002,
      "grad_norm": 0.6528429173010141,
      "learning_rate": 5.2631549687416615e-05,
      "loss": 11.9826,
      "step": 24511
    },
    {
      "epoch": 1.3347763642427832,
      "grad_norm": 0.55452836762981,
      "learning_rate": 5.26237836791873e-05,
      "loss": 11.8725,
      "step": 24512
    },
    {
      "epoch": 1.3348308182393662,
      "grad_norm": 0.5459149363313589,
      "learning_rate": 5.261601803936341e-05,
      "loss": 11.8963,
      "step": 24513
    },
    {
      "epoch": 1.3348852722359492,
      "grad_norm": 0.5294156458948727,
      "learning_rate": 5.2608252768005286e-05,
      "loss": 11.869,
      "step": 24514
    },
    {
      "epoch": 1.3349397262325322,
      "grad_norm": 0.602535210597049,
      "learning_rate": 5.260048786517337e-05,
      "loss": 11.9743,
      "step": 24515
    },
    {
      "epoch": 1.3349941802291152,
      "grad_norm": 0.6139798693690568,
      "learning_rate": 5.2592723330927965e-05,
      "loss": 12.0177,
      "step": 24516
    },
    {
      "epoch": 1.3350486342256982,
      "grad_norm": 0.5571415481736457,
      "learning_rate": 5.258495916532953e-05,
      "loss": 11.8239,
      "step": 24517
    },
    {
      "epoch": 1.3351030882222812,
      "grad_norm": 0.6025306949219089,
      "learning_rate": 5.257719536843837e-05,
      "loss": 12.0201,
      "step": 24518
    },
    {
      "epoch": 1.3351575422188642,
      "grad_norm": 0.5926428550507084,
      "learning_rate": 5.256943194031494e-05,
      "loss": 11.9947,
      "step": 24519
    },
    {
      "epoch": 1.3352119962154472,
      "grad_norm": 0.547831804283872,
      "learning_rate": 5.256166888101954e-05,
      "loss": 11.8195,
      "step": 24520
    },
    {
      "epoch": 1.3352664502120302,
      "grad_norm": 0.5741966033625567,
      "learning_rate": 5.2553906190612535e-05,
      "loss": 11.9355,
      "step": 24521
    },
    {
      "epoch": 1.3353209042086132,
      "grad_norm": 0.5688103106950454,
      "learning_rate": 5.254614386915431e-05,
      "loss": 11.8386,
      "step": 24522
    },
    {
      "epoch": 1.3353753582051962,
      "grad_norm": 0.5438147681305743,
      "learning_rate": 5.253838191670527e-05,
      "loss": 11.9718,
      "step": 24523
    },
    {
      "epoch": 1.3354298122017794,
      "grad_norm": 0.5413384242111259,
      "learning_rate": 5.2530620333325696e-05,
      "loss": 11.7993,
      "step": 24524
    },
    {
      "epoch": 1.3354842661983624,
      "grad_norm": 0.5437024314337768,
      "learning_rate": 5.2522859119076006e-05,
      "loss": 11.9408,
      "step": 24525
    },
    {
      "epoch": 1.3355387201949454,
      "grad_norm": 0.5641061542872094,
      "learning_rate": 5.251509827401648e-05,
      "loss": 11.9104,
      "step": 24526
    },
    {
      "epoch": 1.3355931741915283,
      "grad_norm": 0.5531141663828754,
      "learning_rate": 5.250733779820758e-05,
      "loss": 11.9219,
      "step": 24527
    },
    {
      "epoch": 1.3356476281881113,
      "grad_norm": 0.584910086915897,
      "learning_rate": 5.2499577691709526e-05,
      "loss": 11.9262,
      "step": 24528
    },
    {
      "epoch": 1.3357020821846943,
      "grad_norm": 0.4837542733785465,
      "learning_rate": 5.249181795458276e-05,
      "loss": 11.7622,
      "step": 24529
    },
    {
      "epoch": 1.3357565361812773,
      "grad_norm": 0.5236927571332141,
      "learning_rate": 5.248405858688759e-05,
      "loss": 12.0457,
      "step": 24530
    },
    {
      "epoch": 1.3358109901778603,
      "grad_norm": 0.5643534582442321,
      "learning_rate": 5.247629958868432e-05,
      "loss": 11.8375,
      "step": 24531
    },
    {
      "epoch": 1.3358654441744433,
      "grad_norm": 0.6102868601131803,
      "learning_rate": 5.246854096003333e-05,
      "loss": 11.8655,
      "step": 24532
    },
    {
      "epoch": 1.3359198981710265,
      "grad_norm": 0.5241006028585222,
      "learning_rate": 5.2460782700994914e-05,
      "loss": 11.9341,
      "step": 24533
    },
    {
      "epoch": 1.3359743521676095,
      "grad_norm": 0.5393948666480911,
      "learning_rate": 5.2453024811629416e-05,
      "loss": 11.897,
      "step": 24534
    },
    {
      "epoch": 1.3360288061641925,
      "grad_norm": 0.5685925549442461,
      "learning_rate": 5.2445267291997214e-05,
      "loss": 11.9586,
      "step": 24535
    },
    {
      "epoch": 1.3360832601607755,
      "grad_norm": 0.5566302805644933,
      "learning_rate": 5.243751014215855e-05,
      "loss": 11.9093,
      "step": 24536
    },
    {
      "epoch": 1.3361377141573585,
      "grad_norm": 0.5696849007858308,
      "learning_rate": 5.2429753362173814e-05,
      "loss": 11.7861,
      "step": 24537
    },
    {
      "epoch": 1.3361921681539415,
      "grad_norm": 0.5224432631826207,
      "learning_rate": 5.242199695210328e-05,
      "loss": 11.8364,
      "step": 24538
    },
    {
      "epoch": 1.3362466221505245,
      "grad_norm": 0.5013398649851833,
      "learning_rate": 5.2414240912007286e-05,
      "loss": 11.9152,
      "step": 24539
    },
    {
      "epoch": 1.3363010761471075,
      "grad_norm": 0.6143408768650449,
      "learning_rate": 5.240648524194616e-05,
      "loss": 12.0068,
      "step": 24540
    },
    {
      "epoch": 1.3363555301436905,
      "grad_norm": 0.5643609576473816,
      "learning_rate": 5.239872994198012e-05,
      "loss": 11.8676,
      "step": 24541
    },
    {
      "epoch": 1.3364099841402735,
      "grad_norm": 0.4948973319808279,
      "learning_rate": 5.23909750121696e-05,
      "loss": 11.9777,
      "step": 24542
    },
    {
      "epoch": 1.3364644381368564,
      "grad_norm": 0.5487869439216826,
      "learning_rate": 5.238322045257479e-05,
      "loss": 12.0445,
      "step": 24543
    },
    {
      "epoch": 1.3365188921334394,
      "grad_norm": 0.5121733353384509,
      "learning_rate": 5.237546626325609e-05,
      "loss": 11.8944,
      "step": 24544
    },
    {
      "epoch": 1.3365733461300224,
      "grad_norm": 0.5654297358283139,
      "learning_rate": 5.2367712444273697e-05,
      "loss": 11.9342,
      "step": 24545
    },
    {
      "epoch": 1.3366278001266054,
      "grad_norm": 0.6073104836902603,
      "learning_rate": 5.2359958995687973e-05,
      "loss": 11.9483,
      "step": 24546
    },
    {
      "epoch": 1.3366822541231884,
      "grad_norm": 0.5533297583586704,
      "learning_rate": 5.235220591755925e-05,
      "loss": 11.8096,
      "step": 24547
    },
    {
      "epoch": 1.3367367081197716,
      "grad_norm": 0.5467902640031804,
      "learning_rate": 5.23444532099477e-05,
      "loss": 11.8904,
      "step": 24548
    },
    {
      "epoch": 1.3367911621163546,
      "grad_norm": 0.523885299781789,
      "learning_rate": 5.2336700872913724e-05,
      "loss": 11.8243,
      "step": 24549
    },
    {
      "epoch": 1.3368456161129376,
      "grad_norm": 0.5399534978432189,
      "learning_rate": 5.232894890651756e-05,
      "loss": 11.8559,
      "step": 24550
    },
    {
      "epoch": 1.3369000701095206,
      "grad_norm": 0.5301881372557181,
      "learning_rate": 5.232119731081944e-05,
      "loss": 11.9029,
      "step": 24551
    },
    {
      "epoch": 1.3369545241061036,
      "grad_norm": 0.5376157825698882,
      "learning_rate": 5.2313446085879714e-05,
      "loss": 11.9368,
      "step": 24552
    },
    {
      "epoch": 1.3370089781026866,
      "grad_norm": 0.602887367693488,
      "learning_rate": 5.23056952317586e-05,
      "loss": 11.889,
      "step": 24553
    },
    {
      "epoch": 1.3370634320992696,
      "grad_norm": 0.5694006198900216,
      "learning_rate": 5.2297944748516436e-05,
      "loss": 11.9248,
      "step": 24554
    },
    {
      "epoch": 1.3371178860958526,
      "grad_norm": 0.5864370331877468,
      "learning_rate": 5.2290194636213406e-05,
      "loss": 11.8373,
      "step": 24555
    },
    {
      "epoch": 1.3371723400924358,
      "grad_norm": 0.656334369441371,
      "learning_rate": 5.228244489490983e-05,
      "loss": 12.0397,
      "step": 24556
    },
    {
      "epoch": 1.3372267940890188,
      "grad_norm": 0.5181307474081975,
      "learning_rate": 5.227469552466602e-05,
      "loss": 11.9691,
      "step": 24557
    },
    {
      "epoch": 1.3372812480856018,
      "grad_norm": 0.6116110610278155,
      "learning_rate": 5.226694652554211e-05,
      "loss": 11.8194,
      "step": 24558
    },
    {
      "epoch": 1.3373357020821848,
      "grad_norm": 0.5458583824834689,
      "learning_rate": 5.225919789759853e-05,
      "loss": 11.8655,
      "step": 24559
    },
    {
      "epoch": 1.3373901560787678,
      "grad_norm": 0.5168599012507781,
      "learning_rate": 5.225144964089536e-05,
      "loss": 11.7887,
      "step": 24560
    },
    {
      "epoch": 1.3374446100753508,
      "grad_norm": 0.5947057044254064,
      "learning_rate": 5.2243701755492916e-05,
      "loss": 12.0628,
      "step": 24561
    },
    {
      "epoch": 1.3374990640719338,
      "grad_norm": 0.5383409990700289,
      "learning_rate": 5.223595424145149e-05,
      "loss": 11.8655,
      "step": 24562
    },
    {
      "epoch": 1.3375535180685167,
      "grad_norm": 0.5026192462405705,
      "learning_rate": 5.222820709883125e-05,
      "loss": 11.9324,
      "step": 24563
    },
    {
      "epoch": 1.3376079720650997,
      "grad_norm": 0.5398532142020857,
      "learning_rate": 5.2220460327692534e-05,
      "loss": 11.8372,
      "step": 24564
    },
    {
      "epoch": 1.3376624260616827,
      "grad_norm": 0.5478062871579805,
      "learning_rate": 5.221271392809548e-05,
      "loss": 11.8825,
      "step": 24565
    },
    {
      "epoch": 1.3377168800582657,
      "grad_norm": 0.5492161498882498,
      "learning_rate": 5.220496790010043e-05,
      "loss": 11.8732,
      "step": 24566
    },
    {
      "epoch": 1.3377713340548487,
      "grad_norm": 0.5711176807493772,
      "learning_rate": 5.219722224376752e-05,
      "loss": 12.0181,
      "step": 24567
    },
    {
      "epoch": 1.3378257880514317,
      "grad_norm": 0.5198765195672812,
      "learning_rate": 5.218947695915702e-05,
      "loss": 12.0444,
      "step": 24568
    },
    {
      "epoch": 1.3378802420480147,
      "grad_norm": 0.56607808304362,
      "learning_rate": 5.2181732046329216e-05,
      "loss": 11.9568,
      "step": 24569
    },
    {
      "epoch": 1.3379346960445977,
      "grad_norm": 0.5950939842943866,
      "learning_rate": 5.217398750534428e-05,
      "loss": 11.8325,
      "step": 24570
    },
    {
      "epoch": 1.337989150041181,
      "grad_norm": 0.5148668023660602,
      "learning_rate": 5.216624333626239e-05,
      "loss": 11.9365,
      "step": 24571
    },
    {
      "epoch": 1.338043604037764,
      "grad_norm": 0.5832434549780464,
      "learning_rate": 5.215849953914386e-05,
      "loss": 11.9594,
      "step": 24572
    },
    {
      "epoch": 1.338098058034347,
      "grad_norm": 0.5323926007781992,
      "learning_rate": 5.2150756114048814e-05,
      "loss": 11.8446,
      "step": 24573
    },
    {
      "epoch": 1.3381525120309299,
      "grad_norm": 0.6715059857263452,
      "learning_rate": 5.214301306103756e-05,
      "loss": 11.7965,
      "step": 24574
    },
    {
      "epoch": 1.3382069660275129,
      "grad_norm": 0.5972905943475129,
      "learning_rate": 5.213527038017022e-05,
      "loss": 11.9489,
      "step": 24575
    },
    {
      "epoch": 1.3382614200240959,
      "grad_norm": 0.6272140179190515,
      "learning_rate": 5.212752807150708e-05,
      "loss": 11.8413,
      "step": 24576
    },
    {
      "epoch": 1.3383158740206789,
      "grad_norm": 0.5213794721947207,
      "learning_rate": 5.2119786135108284e-05,
      "loss": 11.8303,
      "step": 24577
    },
    {
      "epoch": 1.3383703280172619,
      "grad_norm": 0.6199835343926031,
      "learning_rate": 5.211204457103404e-05,
      "loss": 11.9136,
      "step": 24578
    },
    {
      "epoch": 1.3384247820138448,
      "grad_norm": 0.6456791515026141,
      "learning_rate": 5.2104303379344624e-05,
      "loss": 11.9072,
      "step": 24579
    },
    {
      "epoch": 1.338479236010428,
      "grad_norm": 0.49702494056767493,
      "learning_rate": 5.209656256010016e-05,
      "loss": 11.7111,
      "step": 24580
    },
    {
      "epoch": 1.338533690007011,
      "grad_norm": 0.5736092594306474,
      "learning_rate": 5.208882211336088e-05,
      "loss": 11.918,
      "step": 24581
    },
    {
      "epoch": 1.338588144003594,
      "grad_norm": 0.5551138604712236,
      "learning_rate": 5.2081082039186904e-05,
      "loss": 11.782,
      "step": 24582
    },
    {
      "epoch": 1.338642598000177,
      "grad_norm": 0.532812278004759,
      "learning_rate": 5.2073342337638476e-05,
      "loss": 11.829,
      "step": 24583
    },
    {
      "epoch": 1.33869705199676,
      "grad_norm": 0.6295566813729274,
      "learning_rate": 5.206560300877581e-05,
      "loss": 11.8954,
      "step": 24584
    },
    {
      "epoch": 1.338751505993343,
      "grad_norm": 0.5912823516384009,
      "learning_rate": 5.2057864052659e-05,
      "loss": 11.9551,
      "step": 24585
    },
    {
      "epoch": 1.338805959989926,
      "grad_norm": 0.5743786285938248,
      "learning_rate": 5.205012546934833e-05,
      "loss": 11.971,
      "step": 24586
    },
    {
      "epoch": 1.338860413986509,
      "grad_norm": 0.5508310161518909,
      "learning_rate": 5.2042387258903894e-05,
      "loss": 12.0477,
      "step": 24587
    },
    {
      "epoch": 1.338914867983092,
      "grad_norm": 0.5812276817649266,
      "learning_rate": 5.203464942138594e-05,
      "loss": 11.8536,
      "step": 24588
    },
    {
      "epoch": 1.338969321979675,
      "grad_norm": 0.5663361545925268,
      "learning_rate": 5.202691195685454e-05,
      "loss": 11.8048,
      "step": 24589
    },
    {
      "epoch": 1.339023775976258,
      "grad_norm": 0.5225567662720166,
      "learning_rate": 5.201917486536996e-05,
      "loss": 11.7475,
      "step": 24590
    },
    {
      "epoch": 1.339078229972841,
      "grad_norm": 0.5407205251730083,
      "learning_rate": 5.201143814699233e-05,
      "loss": 11.9106,
      "step": 24591
    },
    {
      "epoch": 1.339132683969424,
      "grad_norm": 0.53190134231552,
      "learning_rate": 5.200370180178176e-05,
      "loss": 11.9709,
      "step": 24592
    },
    {
      "epoch": 1.339187137966007,
      "grad_norm": 0.5828162426608439,
      "learning_rate": 5.1995965829798496e-05,
      "loss": 11.9461,
      "step": 24593
    },
    {
      "epoch": 1.3392415919625902,
      "grad_norm": 0.5606408360367208,
      "learning_rate": 5.198823023110262e-05,
      "loss": 11.9083,
      "step": 24594
    },
    {
      "epoch": 1.3392960459591732,
      "grad_norm": 0.5499153178025608,
      "learning_rate": 5.19804950057543e-05,
      "loss": 12.0599,
      "step": 24595
    },
    {
      "epoch": 1.3393504999557562,
      "grad_norm": 0.6272253214515026,
      "learning_rate": 5.197276015381376e-05,
      "loss": 12.0533,
      "step": 24596
    },
    {
      "epoch": 1.3394049539523392,
      "grad_norm": 0.5874620400508778,
      "learning_rate": 5.196502567534105e-05,
      "loss": 11.934,
      "step": 24597
    },
    {
      "epoch": 1.3394594079489222,
      "grad_norm": 0.5537663911405211,
      "learning_rate": 5.1957291570396385e-05,
      "loss": 11.8273,
      "step": 24598
    },
    {
      "epoch": 1.3395138619455051,
      "grad_norm": 0.5634268633033013,
      "learning_rate": 5.194955783903984e-05,
      "loss": 11.9272,
      "step": 24599
    },
    {
      "epoch": 1.3395683159420881,
      "grad_norm": 0.5515821492661436,
      "learning_rate": 5.1941824481331626e-05,
      "loss": 11.9599,
      "step": 24600
    },
    {
      "epoch": 1.3396227699386711,
      "grad_norm": 0.5012151590987814,
      "learning_rate": 5.193409149733186e-05,
      "loss": 11.928,
      "step": 24601
    },
    {
      "epoch": 1.3396772239352541,
      "grad_norm": 0.5688783355733719,
      "learning_rate": 5.1926358887100604e-05,
      "loss": 11.8969,
      "step": 24602
    },
    {
      "epoch": 1.3397316779318373,
      "grad_norm": 0.4964984174350957,
      "learning_rate": 5.191862665069809e-05,
      "loss": 11.7818,
      "step": 24603
    },
    {
      "epoch": 1.3397861319284203,
      "grad_norm": 0.5021339874537967,
      "learning_rate": 5.191089478818436e-05,
      "loss": 11.96,
      "step": 24604
    },
    {
      "epoch": 1.3398405859250033,
      "grad_norm": 0.5401406439005793,
      "learning_rate": 5.1903163299619575e-05,
      "loss": 11.8579,
      "step": 24605
    },
    {
      "epoch": 1.3398950399215863,
      "grad_norm": 0.5401028643326528,
      "learning_rate": 5.18954321850639e-05,
      "loss": 11.9232,
      "step": 24606
    },
    {
      "epoch": 1.3399494939181693,
      "grad_norm": 0.5118698735519258,
      "learning_rate": 5.1887701444577376e-05,
      "loss": 11.9735,
      "step": 24607
    },
    {
      "epoch": 1.3400039479147523,
      "grad_norm": 0.5270790401729122,
      "learning_rate": 5.187997107822018e-05,
      "loss": 11.9389,
      "step": 24608
    },
    {
      "epoch": 1.3400584019113353,
      "grad_norm": 0.5570540128271331,
      "learning_rate": 5.1872241086052374e-05,
      "loss": 11.8259,
      "step": 24609
    },
    {
      "epoch": 1.3401128559079183,
      "grad_norm": 0.5356794580221736,
      "learning_rate": 5.186451146813411e-05,
      "loss": 12.009,
      "step": 24610
    },
    {
      "epoch": 1.3401673099045013,
      "grad_norm": 0.6214559074683523,
      "learning_rate": 5.18567822245255e-05,
      "loss": 11.8368,
      "step": 24611
    },
    {
      "epoch": 1.3402217639010843,
      "grad_norm": 0.5544359717823504,
      "learning_rate": 5.1849053355286583e-05,
      "loss": 11.8034,
      "step": 24612
    },
    {
      "epoch": 1.3402762178976673,
      "grad_norm": 0.5700299153115338,
      "learning_rate": 5.184132486047752e-05,
      "loss": 11.9758,
      "step": 24613
    },
    {
      "epoch": 1.3403306718942503,
      "grad_norm": 0.5951283432976049,
      "learning_rate": 5.1833596740158376e-05,
      "loss": 11.8935,
      "step": 24614
    },
    {
      "epoch": 1.3403851258908333,
      "grad_norm": 0.5615266933171653,
      "learning_rate": 5.1825868994389304e-05,
      "loss": 11.8921,
      "step": 24615
    },
    {
      "epoch": 1.3404395798874162,
      "grad_norm": 0.4976954408349177,
      "learning_rate": 5.1818141623230285e-05,
      "loss": 11.7621,
      "step": 24616
    },
    {
      "epoch": 1.3404940338839992,
      "grad_norm": 0.5422445813314399,
      "learning_rate": 5.18104146267415e-05,
      "loss": 11.9329,
      "step": 24617
    },
    {
      "epoch": 1.3405484878805825,
      "grad_norm": 0.580614160854788,
      "learning_rate": 5.1802688004983046e-05,
      "loss": 12.011,
      "step": 24618
    },
    {
      "epoch": 1.3406029418771654,
      "grad_norm": 0.5532611306209777,
      "learning_rate": 5.1794961758014936e-05,
      "loss": 11.9552,
      "step": 24619
    },
    {
      "epoch": 1.3406573958737484,
      "grad_norm": 0.5203892391062399,
      "learning_rate": 5.178723588589732e-05,
      "loss": 11.9495,
      "step": 24620
    },
    {
      "epoch": 1.3407118498703314,
      "grad_norm": 0.6274415942007482,
      "learning_rate": 5.177951038869024e-05,
      "loss": 11.9656,
      "step": 24621
    },
    {
      "epoch": 1.3407663038669144,
      "grad_norm": 0.5545298279893404,
      "learning_rate": 5.177178526645375e-05,
      "loss": 11.8779,
      "step": 24622
    },
    {
      "epoch": 1.3408207578634974,
      "grad_norm": 0.5475076096938358,
      "learning_rate": 5.176406051924798e-05,
      "loss": 11.8929,
      "step": 24623
    },
    {
      "epoch": 1.3408752118600804,
      "grad_norm": 0.5193635150414153,
      "learning_rate": 5.175633614713292e-05,
      "loss": 11.8949,
      "step": 24624
    },
    {
      "epoch": 1.3409296658566634,
      "grad_norm": 0.5310161217107139,
      "learning_rate": 5.174861215016873e-05,
      "loss": 11.8812,
      "step": 24625
    },
    {
      "epoch": 1.3409841198532466,
      "grad_norm": 0.5776940149513465,
      "learning_rate": 5.174088852841536e-05,
      "loss": 12.0292,
      "step": 24626
    },
    {
      "epoch": 1.3410385738498296,
      "grad_norm": 0.5795239369137751,
      "learning_rate": 5.173316528193295e-05,
      "loss": 11.8541,
      "step": 24627
    },
    {
      "epoch": 1.3410930278464126,
      "grad_norm": 0.6115089736970947,
      "learning_rate": 5.172544241078159e-05,
      "loss": 11.9109,
      "step": 24628
    },
    {
      "epoch": 1.3411474818429956,
      "grad_norm": 0.5783162807753064,
      "learning_rate": 5.171771991502123e-05,
      "loss": 11.7701,
      "step": 24629
    },
    {
      "epoch": 1.3412019358395786,
      "grad_norm": 0.546458977271874,
      "learning_rate": 5.1709997794712064e-05,
      "loss": 11.7898,
      "step": 24630
    },
    {
      "epoch": 1.3412563898361616,
      "grad_norm": 0.5056918842436856,
      "learning_rate": 5.170227604991398e-05,
      "loss": 11.8176,
      "step": 24631
    },
    {
      "epoch": 1.3413108438327446,
      "grad_norm": 0.47401625281647586,
      "learning_rate": 5.169455468068709e-05,
      "loss": 11.878,
      "step": 24632
    },
    {
      "epoch": 1.3413652978293276,
      "grad_norm": 0.5009397159064994,
      "learning_rate": 5.1686833687091475e-05,
      "loss": 11.758,
      "step": 24633
    },
    {
      "epoch": 1.3414197518259106,
      "grad_norm": 0.5628233913388982,
      "learning_rate": 5.1679113069187115e-05,
      "loss": 12.0128,
      "step": 24634
    },
    {
      "epoch": 1.3414742058224935,
      "grad_norm": 0.498647738732315,
      "learning_rate": 5.1671392827034106e-05,
      "loss": 11.9058,
      "step": 24635
    },
    {
      "epoch": 1.3415286598190765,
      "grad_norm": 0.49257969311586597,
      "learning_rate": 5.166367296069241e-05,
      "loss": 11.9904,
      "step": 24636
    },
    {
      "epoch": 1.3415831138156595,
      "grad_norm": 0.5506269605856507,
      "learning_rate": 5.1655953470222154e-05,
      "loss": 11.857,
      "step": 24637
    },
    {
      "epoch": 1.3416375678122425,
      "grad_norm": 0.559868816327295,
      "learning_rate": 5.164823435568327e-05,
      "loss": 12.0076,
      "step": 24638
    },
    {
      "epoch": 1.3416920218088255,
      "grad_norm": 0.5731561395593245,
      "learning_rate": 5.164051561713581e-05,
      "loss": 11.9248,
      "step": 24639
    },
    {
      "epoch": 1.3417464758054085,
      "grad_norm": 0.5540462796840516,
      "learning_rate": 5.16327972546399e-05,
      "loss": 11.8274,
      "step": 24640
    },
    {
      "epoch": 1.3418009298019917,
      "grad_norm": 0.5320036454013334,
      "learning_rate": 5.1625079268255386e-05,
      "loss": 11.9924,
      "step": 24641
    },
    {
      "epoch": 1.3418553837985747,
      "grad_norm": 0.5241473322774366,
      "learning_rate": 5.1617361658042405e-05,
      "loss": 11.8711,
      "step": 24642
    },
    {
      "epoch": 1.3419098377951577,
      "grad_norm": 0.5055339413936103,
      "learning_rate": 5.16096444240609e-05,
      "loss": 11.9069,
      "step": 24643
    },
    {
      "epoch": 1.3419642917917407,
      "grad_norm": 0.5048298227752752,
      "learning_rate": 5.160192756637091e-05,
      "loss": 11.9758,
      "step": 24644
    },
    {
      "epoch": 1.3420187457883237,
      "grad_norm": 0.5124961733519724,
      "learning_rate": 5.1594211085032484e-05,
      "loss": 11.7453,
      "step": 24645
    },
    {
      "epoch": 1.3420731997849067,
      "grad_norm": 0.5360412781099567,
      "learning_rate": 5.158649498010557e-05,
      "loss": 11.8183,
      "step": 24646
    },
    {
      "epoch": 1.3421276537814897,
      "grad_norm": 0.5217600581621155,
      "learning_rate": 5.157877925165021e-05,
      "loss": 11.8677,
      "step": 24647
    },
    {
      "epoch": 1.3421821077780727,
      "grad_norm": 0.5520672183749687,
      "learning_rate": 5.157106389972635e-05,
      "loss": 11.852,
      "step": 24648
    },
    {
      "epoch": 1.3422365617746557,
      "grad_norm": 0.5344003240530755,
      "learning_rate": 5.156334892439405e-05,
      "loss": 11.978,
      "step": 24649
    },
    {
      "epoch": 1.3422910157712389,
      "grad_norm": 0.5756585787262845,
      "learning_rate": 5.1555634325713284e-05,
      "loss": 11.8156,
      "step": 24650
    },
    {
      "epoch": 1.3423454697678219,
      "grad_norm": 0.5104893397762217,
      "learning_rate": 5.154792010374399e-05,
      "loss": 11.8454,
      "step": 24651
    },
    {
      "epoch": 1.3423999237644049,
      "grad_norm": 0.5863480953662468,
      "learning_rate": 5.154020625854623e-05,
      "loss": 11.9262,
      "step": 24652
    },
    {
      "epoch": 1.3424543777609879,
      "grad_norm": 0.492592261822517,
      "learning_rate": 5.153249279017991e-05,
      "loss": 11.9287,
      "step": 24653
    },
    {
      "epoch": 1.3425088317575709,
      "grad_norm": 0.5872998914003171,
      "learning_rate": 5.152477969870506e-05,
      "loss": 11.908,
      "step": 24654
    },
    {
      "epoch": 1.3425632857541538,
      "grad_norm": 0.5605485116233468,
      "learning_rate": 5.1517066984181685e-05,
      "loss": 11.8801,
      "step": 24655
    },
    {
      "epoch": 1.3426177397507368,
      "grad_norm": 0.5990806711376991,
      "learning_rate": 5.1509354646669695e-05,
      "loss": 11.9292,
      "step": 24656
    },
    {
      "epoch": 1.3426721937473198,
      "grad_norm": 0.7279063077741952,
      "learning_rate": 5.1501642686229126e-05,
      "loss": 12.01,
      "step": 24657
    },
    {
      "epoch": 1.3427266477439028,
      "grad_norm": 0.5294510462086448,
      "learning_rate": 5.1493931102919893e-05,
      "loss": 11.9327,
      "step": 24658
    },
    {
      "epoch": 1.3427811017404858,
      "grad_norm": 0.5012113477080508,
      "learning_rate": 5.1486219896802e-05,
      "loss": 11.979,
      "step": 24659
    },
    {
      "epoch": 1.3428355557370688,
      "grad_norm": 0.5208283316466333,
      "learning_rate": 5.147850906793542e-05,
      "loss": 11.8707,
      "step": 24660
    },
    {
      "epoch": 1.3428900097336518,
      "grad_norm": 0.582335091795844,
      "learning_rate": 5.1470798616380045e-05,
      "loss": 12.0025,
      "step": 24661
    },
    {
      "epoch": 1.3429444637302348,
      "grad_norm": 0.6343594526751137,
      "learning_rate": 5.146308854219591e-05,
      "loss": 12.1038,
      "step": 24662
    },
    {
      "epoch": 1.3429989177268178,
      "grad_norm": 0.5244064185971522,
      "learning_rate": 5.145537884544291e-05,
      "loss": 11.9433,
      "step": 24663
    },
    {
      "epoch": 1.343053371723401,
      "grad_norm": 0.7054565994001688,
      "learning_rate": 5.1447669526181055e-05,
      "loss": 12.0494,
      "step": 24664
    },
    {
      "epoch": 1.343107825719984,
      "grad_norm": 0.570649830549098,
      "learning_rate": 5.143996058447023e-05,
      "loss": 11.9568,
      "step": 24665
    },
    {
      "epoch": 1.343162279716567,
      "grad_norm": 0.6754946316357943,
      "learning_rate": 5.143225202037042e-05,
      "loss": 12.0417,
      "step": 24666
    },
    {
      "epoch": 1.34321673371315,
      "grad_norm": 0.5069186991077531,
      "learning_rate": 5.14245438339416e-05,
      "loss": 11.9761,
      "step": 24667
    },
    {
      "epoch": 1.343271187709733,
      "grad_norm": 0.5415640747473932,
      "learning_rate": 5.141683602524363e-05,
      "loss": 11.8968,
      "step": 24668
    },
    {
      "epoch": 1.343325641706316,
      "grad_norm": 0.5194004817624808,
      "learning_rate": 5.140912859433653e-05,
      "loss": 11.9379,
      "step": 24669
    },
    {
      "epoch": 1.343380095702899,
      "grad_norm": 0.5227581353768905,
      "learning_rate": 5.140142154128018e-05,
      "loss": 11.8972,
      "step": 24670
    },
    {
      "epoch": 1.343434549699482,
      "grad_norm": 0.6213626815070582,
      "learning_rate": 5.139371486613451e-05,
      "loss": 11.9509,
      "step": 24671
    },
    {
      "epoch": 1.343489003696065,
      "grad_norm": 0.5427038923186286,
      "learning_rate": 5.13860085689595e-05,
      "loss": 11.887,
      "step": 24672
    },
    {
      "epoch": 1.3435434576926482,
      "grad_norm": 0.536390506451965,
      "learning_rate": 5.1378302649814994e-05,
      "loss": 11.7878,
      "step": 24673
    },
    {
      "epoch": 1.3435979116892312,
      "grad_norm": 0.5814214936671521,
      "learning_rate": 5.1370597108761e-05,
      "loss": 12.0254,
      "step": 24674
    },
    {
      "epoch": 1.3436523656858141,
      "grad_norm": 0.6034499203431405,
      "learning_rate": 5.136289194585736e-05,
      "loss": 12.001,
      "step": 24675
    },
    {
      "epoch": 1.3437068196823971,
      "grad_norm": 0.5879422515014885,
      "learning_rate": 5.1355187161164034e-05,
      "loss": 11.9203,
      "step": 24676
    },
    {
      "epoch": 1.3437612736789801,
      "grad_norm": 0.5096279975774792,
      "learning_rate": 5.134748275474096e-05,
      "loss": 11.9085,
      "step": 24677
    },
    {
      "epoch": 1.3438157276755631,
      "grad_norm": 0.5210584951397081,
      "learning_rate": 5.133977872664799e-05,
      "loss": 11.9083,
      "step": 24678
    },
    {
      "epoch": 1.3438701816721461,
      "grad_norm": 0.5695970056479077,
      "learning_rate": 5.1332075076945084e-05,
      "loss": 11.7609,
      "step": 24679
    },
    {
      "epoch": 1.343924635668729,
      "grad_norm": 0.5433632179172042,
      "learning_rate": 5.132437180569214e-05,
      "loss": 11.8395,
      "step": 24680
    },
    {
      "epoch": 1.343979089665312,
      "grad_norm": 0.5408495754745256,
      "learning_rate": 5.131666891294899e-05,
      "loss": 11.8961,
      "step": 24681
    },
    {
      "epoch": 1.344033543661895,
      "grad_norm": 0.5367184083183594,
      "learning_rate": 5.130896639877565e-05,
      "loss": 11.92,
      "step": 24682
    },
    {
      "epoch": 1.344087997658478,
      "grad_norm": 0.6222809867914364,
      "learning_rate": 5.13012642632319e-05,
      "loss": 11.8826,
      "step": 24683
    },
    {
      "epoch": 1.344142451655061,
      "grad_norm": 0.5244961821012735,
      "learning_rate": 5.129356250637771e-05,
      "loss": 11.989,
      "step": 24684
    },
    {
      "epoch": 1.344196905651644,
      "grad_norm": 0.548225762241452,
      "learning_rate": 5.1285861128272924e-05,
      "loss": 11.8568,
      "step": 24685
    },
    {
      "epoch": 1.344251359648227,
      "grad_norm": 0.5935253842941441,
      "learning_rate": 5.1278160128977494e-05,
      "loss": 11.9319,
      "step": 24686
    },
    {
      "epoch": 1.3443058136448103,
      "grad_norm": 0.5671982583908727,
      "learning_rate": 5.1270459508551207e-05,
      "loss": 11.9201,
      "step": 24687
    },
    {
      "epoch": 1.3443602676413933,
      "grad_norm": 0.5610219671353242,
      "learning_rate": 5.126275926705402e-05,
      "loss": 11.9223,
      "step": 24688
    },
    {
      "epoch": 1.3444147216379763,
      "grad_norm": 0.5069111973494803,
      "learning_rate": 5.12550594045458e-05,
      "loss": 11.8724,
      "step": 24689
    },
    {
      "epoch": 1.3444691756345593,
      "grad_norm": 0.5620002209813412,
      "learning_rate": 5.1247359921086444e-05,
      "loss": 12.0272,
      "step": 24690
    },
    {
      "epoch": 1.3445236296311422,
      "grad_norm": 0.6137102404689122,
      "learning_rate": 5.1239660816735765e-05,
      "loss": 12.0232,
      "step": 24691
    },
    {
      "epoch": 1.3445780836277252,
      "grad_norm": 0.5901212641170603,
      "learning_rate": 5.123196209155364e-05,
      "loss": 11.9088,
      "step": 24692
    },
    {
      "epoch": 1.3446325376243082,
      "grad_norm": 0.5769018309674903,
      "learning_rate": 5.1224263745599955e-05,
      "loss": 11.8818,
      "step": 24693
    },
    {
      "epoch": 1.3446869916208912,
      "grad_norm": 0.5005582799986129,
      "learning_rate": 5.121656577893461e-05,
      "loss": 11.7601,
      "step": 24694
    },
    {
      "epoch": 1.3447414456174742,
      "grad_norm": 0.5932355413713493,
      "learning_rate": 5.120886819161739e-05,
      "loss": 12.0082,
      "step": 24695
    },
    {
      "epoch": 1.3447958996140574,
      "grad_norm": 0.5305838490879435,
      "learning_rate": 5.120117098370824e-05,
      "loss": 12.0842,
      "step": 24696
    },
    {
      "epoch": 1.3448503536106404,
      "grad_norm": 0.5598975906879371,
      "learning_rate": 5.119347415526692e-05,
      "loss": 11.8764,
      "step": 24697
    },
    {
      "epoch": 1.3449048076072234,
      "grad_norm": 0.5366234881540021,
      "learning_rate": 5.118577770635337e-05,
      "loss": 11.9488,
      "step": 24698
    },
    {
      "epoch": 1.3449592616038064,
      "grad_norm": 0.5326388231089194,
      "learning_rate": 5.1178081637027356e-05,
      "loss": 11.8126,
      "step": 24699
    },
    {
      "epoch": 1.3450137156003894,
      "grad_norm": 0.5264614507788572,
      "learning_rate": 5.117038594734882e-05,
      "loss": 11.9151,
      "step": 24700
    },
    {
      "epoch": 1.3450681695969724,
      "grad_norm": 0.5568658150801595,
      "learning_rate": 5.116269063737753e-05,
      "loss": 12.0364,
      "step": 24701
    },
    {
      "epoch": 1.3451226235935554,
      "grad_norm": 0.5426277076062659,
      "learning_rate": 5.115499570717333e-05,
      "loss": 11.7796,
      "step": 24702
    },
    {
      "epoch": 1.3451770775901384,
      "grad_norm": 0.5405251899144996,
      "learning_rate": 5.1147301156796065e-05,
      "loss": 12.0124,
      "step": 24703
    },
    {
      "epoch": 1.3452315315867214,
      "grad_norm": 0.5650397200967617,
      "learning_rate": 5.113960698630562e-05,
      "loss": 11.94,
      "step": 24704
    },
    {
      "epoch": 1.3452859855833044,
      "grad_norm": 0.6118572302047037,
      "learning_rate": 5.113191319576174e-05,
      "loss": 11.9214,
      "step": 24705
    },
    {
      "epoch": 1.3453404395798874,
      "grad_norm": 0.5286389829934046,
      "learning_rate": 5.112421978522435e-05,
      "loss": 11.743,
      "step": 24706
    },
    {
      "epoch": 1.3453948935764704,
      "grad_norm": 0.5937446151677802,
      "learning_rate": 5.111652675475317e-05,
      "loss": 11.9297,
      "step": 24707
    },
    {
      "epoch": 1.3454493475730533,
      "grad_norm": 0.540039344359391,
      "learning_rate": 5.110883410440812e-05,
      "loss": 11.9706,
      "step": 24708
    },
    {
      "epoch": 1.3455038015696363,
      "grad_norm": 0.5318002732872339,
      "learning_rate": 5.110114183424893e-05,
      "loss": 11.8151,
      "step": 24709
    },
    {
      "epoch": 1.3455582555662193,
      "grad_norm": 0.5701680828712842,
      "learning_rate": 5.109344994433547e-05,
      "loss": 11.9632,
      "step": 24710
    },
    {
      "epoch": 1.3456127095628025,
      "grad_norm": 0.5322747869200056,
      "learning_rate": 5.108575843472763e-05,
      "loss": 11.953,
      "step": 24711
    },
    {
      "epoch": 1.3456671635593855,
      "grad_norm": 0.6386986840153218,
      "learning_rate": 5.107806730548503e-05,
      "loss": 11.9134,
      "step": 24712
    },
    {
      "epoch": 1.3457216175559685,
      "grad_norm": 0.5787294346188318,
      "learning_rate": 5.107037655666765e-05,
      "loss": 11.8885,
      "step": 24713
    },
    {
      "epoch": 1.3457760715525515,
      "grad_norm": 0.48484950700769913,
      "learning_rate": 5.1062686188335184e-05,
      "loss": 11.911,
      "step": 24714
    },
    {
      "epoch": 1.3458305255491345,
      "grad_norm": 0.560911470456335,
      "learning_rate": 5.1054996200547465e-05,
      "loss": 11.9997,
      "step": 24715
    },
    {
      "epoch": 1.3458849795457175,
      "grad_norm": 0.5227067169613693,
      "learning_rate": 5.104730659336435e-05,
      "loss": 11.8588,
      "step": 24716
    },
    {
      "epoch": 1.3459394335423005,
      "grad_norm": 0.5328178749192024,
      "learning_rate": 5.103961736684555e-05,
      "loss": 11.8337,
      "step": 24717
    },
    {
      "epoch": 1.3459938875388835,
      "grad_norm": 0.574303286229096,
      "learning_rate": 5.1031928521050933e-05,
      "loss": 11.9587,
      "step": 24718
    },
    {
      "epoch": 1.3460483415354665,
      "grad_norm": 0.5469256303813594,
      "learning_rate": 5.102424005604021e-05,
      "loss": 11.8126,
      "step": 24719
    },
    {
      "epoch": 1.3461027955320497,
      "grad_norm": 0.5502648367970351,
      "learning_rate": 5.101655197187325e-05,
      "loss": 11.8713,
      "step": 24720
    },
    {
      "epoch": 1.3461572495286327,
      "grad_norm": 0.5825838781402476,
      "learning_rate": 5.10088642686098e-05,
      "loss": 11.9476,
      "step": 24721
    },
    {
      "epoch": 1.3462117035252157,
      "grad_norm": 0.5312492793536634,
      "learning_rate": 5.100117694630959e-05,
      "loss": 11.9215,
      "step": 24722
    },
    {
      "epoch": 1.3462661575217987,
      "grad_norm": 0.5925804114597782,
      "learning_rate": 5.0993490005032506e-05,
      "loss": 11.9656,
      "step": 24723
    },
    {
      "epoch": 1.3463206115183817,
      "grad_norm": 0.597512351118672,
      "learning_rate": 5.098580344483821e-05,
      "loss": 11.857,
      "step": 24724
    },
    {
      "epoch": 1.3463750655149647,
      "grad_norm": 0.5527618332523438,
      "learning_rate": 5.097811726578652e-05,
      "loss": 12.0091,
      "step": 24725
    },
    {
      "epoch": 1.3464295195115477,
      "grad_norm": 0.5815921715331781,
      "learning_rate": 5.0970431467937254e-05,
      "loss": 11.9153,
      "step": 24726
    },
    {
      "epoch": 1.3464839735081306,
      "grad_norm": 0.49469894164678824,
      "learning_rate": 5.09627460513501e-05,
      "loss": 11.874,
      "step": 24727
    },
    {
      "epoch": 1.3465384275047136,
      "grad_norm": 0.5202688863510311,
      "learning_rate": 5.095506101608488e-05,
      "loss": 11.915,
      "step": 24728
    },
    {
      "epoch": 1.3465928815012966,
      "grad_norm": 0.5437772142738702,
      "learning_rate": 5.0947376362201304e-05,
      "loss": 11.7232,
      "step": 24729
    },
    {
      "epoch": 1.3466473354978796,
      "grad_norm": 0.538512507411147,
      "learning_rate": 5.0939692089759195e-05,
      "loss": 11.9361,
      "step": 24730
    },
    {
      "epoch": 1.3467017894944626,
      "grad_norm": 0.5852451636107295,
      "learning_rate": 5.093200819881827e-05,
      "loss": 11.8578,
      "step": 24731
    },
    {
      "epoch": 1.3467562434910456,
      "grad_norm": 0.5744333870393465,
      "learning_rate": 5.092432468943823e-05,
      "loss": 11.9063,
      "step": 24732
    },
    {
      "epoch": 1.3468106974876286,
      "grad_norm": 0.5660255027197925,
      "learning_rate": 5.091664156167892e-05,
      "loss": 11.9843,
      "step": 24733
    },
    {
      "epoch": 1.3468651514842118,
      "grad_norm": 0.8385179784822178,
      "learning_rate": 5.090895881559998e-05,
      "loss": 11.9528,
      "step": 24734
    },
    {
      "epoch": 1.3469196054807948,
      "grad_norm": 0.5644388973902699,
      "learning_rate": 5.090127645126127e-05,
      "loss": 11.9393,
      "step": 24735
    },
    {
      "epoch": 1.3469740594773778,
      "grad_norm": 0.5061500219352385,
      "learning_rate": 5.089359446872242e-05,
      "loss": 11.8778,
      "step": 24736
    },
    {
      "epoch": 1.3470285134739608,
      "grad_norm": 0.6064606475805742,
      "learning_rate": 5.088591286804322e-05,
      "loss": 12.0032,
      "step": 24737
    },
    {
      "epoch": 1.3470829674705438,
      "grad_norm": 0.6085706595767844,
      "learning_rate": 5.0878231649283424e-05,
      "loss": 12.0709,
      "step": 24738
    },
    {
      "epoch": 1.3471374214671268,
      "grad_norm": 0.5112846789243923,
      "learning_rate": 5.0870550812502704e-05,
      "loss": 11.8724,
      "step": 24739
    },
    {
      "epoch": 1.3471918754637098,
      "grad_norm": 0.5642057804212833,
      "learning_rate": 5.086287035776086e-05,
      "loss": 11.9766,
      "step": 24740
    },
    {
      "epoch": 1.3472463294602928,
      "grad_norm": 0.538498722557541,
      "learning_rate": 5.0855190285117585e-05,
      "loss": 11.9649,
      "step": 24741
    },
    {
      "epoch": 1.3473007834568758,
      "grad_norm": 0.5167847240960812,
      "learning_rate": 5.084751059463253e-05,
      "loss": 11.85,
      "step": 24742
    },
    {
      "epoch": 1.347355237453459,
      "grad_norm": 0.5688071269020584,
      "learning_rate": 5.0839831286365535e-05,
      "loss": 11.9503,
      "step": 24743
    },
    {
      "epoch": 1.347409691450042,
      "grad_norm": 0.6078205857455574,
      "learning_rate": 5.083215236037622e-05,
      "loss": 11.8844,
      "step": 24744
    },
    {
      "epoch": 1.347464145446625,
      "grad_norm": 0.5504748636150579,
      "learning_rate": 5.082447381672435e-05,
      "loss": 11.9087,
      "step": 24745
    },
    {
      "epoch": 1.347518599443208,
      "grad_norm": 0.556827012989121,
      "learning_rate": 5.081679565546959e-05,
      "loss": 11.8018,
      "step": 24746
    },
    {
      "epoch": 1.347573053439791,
      "grad_norm": 0.5326846761653032,
      "learning_rate": 5.080911787667173e-05,
      "loss": 11.8898,
      "step": 24747
    },
    {
      "epoch": 1.347627507436374,
      "grad_norm": 0.4771338407625155,
      "learning_rate": 5.080144048039036e-05,
      "loss": 11.9173,
      "step": 24748
    },
    {
      "epoch": 1.347681961432957,
      "grad_norm": 0.5514342083191643,
      "learning_rate": 5.079376346668525e-05,
      "loss": 11.9127,
      "step": 24749
    },
    {
      "epoch": 1.34773641542954,
      "grad_norm": 0.49419029608704157,
      "learning_rate": 5.078608683561612e-05,
      "loss": 11.8876,
      "step": 24750
    },
    {
      "epoch": 1.347790869426123,
      "grad_norm": 0.5668823733875155,
      "learning_rate": 5.077841058724263e-05,
      "loss": 11.8694,
      "step": 24751
    },
    {
      "epoch": 1.347845323422706,
      "grad_norm": 0.532912246970959,
      "learning_rate": 5.077073472162444e-05,
      "loss": 12.0189,
      "step": 24752
    },
    {
      "epoch": 1.347899777419289,
      "grad_norm": 0.5671877456159174,
      "learning_rate": 5.076305923882131e-05,
      "loss": 11.974,
      "step": 24753
    },
    {
      "epoch": 1.347954231415872,
      "grad_norm": 0.5616592638243587,
      "learning_rate": 5.0755384138892846e-05,
      "loss": 11.8129,
      "step": 24754
    },
    {
      "epoch": 1.3480086854124549,
      "grad_norm": 0.570931293347498,
      "learning_rate": 5.074770942189882e-05,
      "loss": 11.9228,
      "step": 24755
    },
    {
      "epoch": 1.3480631394090379,
      "grad_norm": 0.5067208853009946,
      "learning_rate": 5.074003508789882e-05,
      "loss": 11.9231,
      "step": 24756
    },
    {
      "epoch": 1.348117593405621,
      "grad_norm": 0.48252121944302095,
      "learning_rate": 5.073236113695261e-05,
      "loss": 11.7249,
      "step": 24757
    },
    {
      "epoch": 1.348172047402204,
      "grad_norm": 0.517549680877548,
      "learning_rate": 5.072468756911979e-05,
      "loss": 11.9628,
      "step": 24758
    },
    {
      "epoch": 1.348226501398787,
      "grad_norm": 0.6033185712884095,
      "learning_rate": 5.0717014384460036e-05,
      "loss": 11.9381,
      "step": 24759
    },
    {
      "epoch": 1.34828095539537,
      "grad_norm": 0.5521904788131673,
      "learning_rate": 5.070934158303311e-05,
      "loss": 11.8499,
      "step": 24760
    },
    {
      "epoch": 1.348335409391953,
      "grad_norm": 0.5511601673095677,
      "learning_rate": 5.0701669164898603e-05,
      "loss": 11.8565,
      "step": 24761
    },
    {
      "epoch": 1.348389863388536,
      "grad_norm": 0.5607754924702814,
      "learning_rate": 5.069399713011618e-05,
      "loss": 11.8313,
      "step": 24762
    },
    {
      "epoch": 1.348444317385119,
      "grad_norm": 0.5987504656166703,
      "learning_rate": 5.0686325478745466e-05,
      "loss": 11.9435,
      "step": 24763
    },
    {
      "epoch": 1.348498771381702,
      "grad_norm": 0.5803073742557056,
      "learning_rate": 5.067865421084615e-05,
      "loss": 12.02,
      "step": 24764
    },
    {
      "epoch": 1.348553225378285,
      "grad_norm": 0.5411527019282852,
      "learning_rate": 5.0670983326477936e-05,
      "loss": 12.0168,
      "step": 24765
    },
    {
      "epoch": 1.3486076793748683,
      "grad_norm": 0.5220430512587575,
      "learning_rate": 5.0663312825700395e-05,
      "loss": 11.9947,
      "step": 24766
    },
    {
      "epoch": 1.3486621333714512,
      "grad_norm": 0.5189351798939571,
      "learning_rate": 5.065564270857323e-05,
      "loss": 11.868,
      "step": 24767
    },
    {
      "epoch": 1.3487165873680342,
      "grad_norm": 0.6134265449723086,
      "learning_rate": 5.064797297515604e-05,
      "loss": 12.0574,
      "step": 24768
    },
    {
      "epoch": 1.3487710413646172,
      "grad_norm": 0.5799543714779022,
      "learning_rate": 5.0640303625508514e-05,
      "loss": 11.7574,
      "step": 24769
    },
    {
      "epoch": 1.3488254953612002,
      "grad_norm": 0.47761785916124516,
      "learning_rate": 5.063263465969024e-05,
      "loss": 11.93,
      "step": 24770
    },
    {
      "epoch": 1.3488799493577832,
      "grad_norm": 0.532400116807748,
      "learning_rate": 5.06249660777609e-05,
      "loss": 11.9961,
      "step": 24771
    },
    {
      "epoch": 1.3489344033543662,
      "grad_norm": 0.5422298339313437,
      "learning_rate": 5.0617297879780114e-05,
      "loss": 11.8649,
      "step": 24772
    },
    {
      "epoch": 1.3489888573509492,
      "grad_norm": 0.5578831490030823,
      "learning_rate": 5.060963006580746e-05,
      "loss": 11.8773,
      "step": 24773
    },
    {
      "epoch": 1.3490433113475322,
      "grad_norm": 0.5969063133631078,
      "learning_rate": 5.06019626359026e-05,
      "loss": 12.0546,
      "step": 24774
    },
    {
      "epoch": 1.3490977653441152,
      "grad_norm": 0.5065166455132252,
      "learning_rate": 5.059429559012521e-05,
      "loss": 11.8178,
      "step": 24775
    },
    {
      "epoch": 1.3491522193406982,
      "grad_norm": 0.531827767695221,
      "learning_rate": 5.058662892853483e-05,
      "loss": 12.0171,
      "step": 24776
    },
    {
      "epoch": 1.3492066733372812,
      "grad_norm": 0.502000655939235,
      "learning_rate": 5.057896265119113e-05,
      "loss": 11.8882,
      "step": 24777
    },
    {
      "epoch": 1.3492611273338642,
      "grad_norm": 0.5363887226520877,
      "learning_rate": 5.057129675815368e-05,
      "loss": 11.7719,
      "step": 24778
    },
    {
      "epoch": 1.3493155813304472,
      "grad_norm": 0.5046527835178513,
      "learning_rate": 5.056363124948216e-05,
      "loss": 11.9025,
      "step": 24779
    },
    {
      "epoch": 1.3493700353270301,
      "grad_norm": 0.5511538817038009,
      "learning_rate": 5.055596612523607e-05,
      "loss": 11.9255,
      "step": 24780
    },
    {
      "epoch": 1.3494244893236134,
      "grad_norm": 0.508313966078211,
      "learning_rate": 5.054830138547514e-05,
      "loss": 11.8887,
      "step": 24781
    },
    {
      "epoch": 1.3494789433201964,
      "grad_norm": 0.5247460253492393,
      "learning_rate": 5.054063703025891e-05,
      "loss": 11.8504,
      "step": 24782
    },
    {
      "epoch": 1.3495333973167793,
      "grad_norm": 0.5344680484674968,
      "learning_rate": 5.053297305964693e-05,
      "loss": 11.851,
      "step": 24783
    },
    {
      "epoch": 1.3495878513133623,
      "grad_norm": 0.5471176781384924,
      "learning_rate": 5.0525309473698887e-05,
      "loss": 11.8643,
      "step": 24784
    },
    {
      "epoch": 1.3496423053099453,
      "grad_norm": 0.5463936967650441,
      "learning_rate": 5.051764627247431e-05,
      "loss": 12.0349,
      "step": 24785
    },
    {
      "epoch": 1.3496967593065283,
      "grad_norm": 0.526339112385976,
      "learning_rate": 5.05099834560328e-05,
      "loss": 11.8445,
      "step": 24786
    },
    {
      "epoch": 1.3497512133031113,
      "grad_norm": 0.5630772598984861,
      "learning_rate": 5.0502321024434e-05,
      "loss": 11.9418,
      "step": 24787
    },
    {
      "epoch": 1.3498056672996943,
      "grad_norm": 0.551988855827672,
      "learning_rate": 5.049465897773742e-05,
      "loss": 11.9497,
      "step": 24788
    },
    {
      "epoch": 1.3498601212962775,
      "grad_norm": 0.5761189634907042,
      "learning_rate": 5.04869973160027e-05,
      "loss": 11.9864,
      "step": 24789
    },
    {
      "epoch": 1.3499145752928605,
      "grad_norm": 0.5447156986558133,
      "learning_rate": 5.047933603928936e-05,
      "loss": 11.9061,
      "step": 24790
    },
    {
      "epoch": 1.3499690292894435,
      "grad_norm": 0.5088697949684696,
      "learning_rate": 5.047167514765705e-05,
      "loss": 11.9659,
      "step": 24791
    },
    {
      "epoch": 1.3500234832860265,
      "grad_norm": 0.5166478617367919,
      "learning_rate": 5.0464014641165304e-05,
      "loss": 11.9017,
      "step": 24792
    },
    {
      "epoch": 1.3500779372826095,
      "grad_norm": 0.49868874735038105,
      "learning_rate": 5.045635451987365e-05,
      "loss": 11.9322,
      "step": 24793
    },
    {
      "epoch": 1.3501323912791925,
      "grad_norm": 0.5614526172360104,
      "learning_rate": 5.044869478384171e-05,
      "loss": 12.0084,
      "step": 24794
    },
    {
      "epoch": 1.3501868452757755,
      "grad_norm": 0.5314844398684915,
      "learning_rate": 5.044103543312901e-05,
      "loss": 11.7961,
      "step": 24795
    },
    {
      "epoch": 1.3502412992723585,
      "grad_norm": 0.5552772230831415,
      "learning_rate": 5.043337646779516e-05,
      "loss": 11.9476,
      "step": 24796
    },
    {
      "epoch": 1.3502957532689415,
      "grad_norm": 0.5023316306806409,
      "learning_rate": 5.042571788789965e-05,
      "loss": 11.8229,
      "step": 24797
    },
    {
      "epoch": 1.3503502072655245,
      "grad_norm": 0.5011525564571289,
      "learning_rate": 5.0418059693502065e-05,
      "loss": 11.8788,
      "step": 24798
    },
    {
      "epoch": 1.3504046612621075,
      "grad_norm": 0.5200250815765821,
      "learning_rate": 5.041040188466202e-05,
      "loss": 11.9296,
      "step": 24799
    },
    {
      "epoch": 1.3504591152586904,
      "grad_norm": 0.6615102104803817,
      "learning_rate": 5.040274446143896e-05,
      "loss": 12.0247,
      "step": 24800
    },
    {
      "epoch": 1.3505135692552734,
      "grad_norm": 0.595575885434631,
      "learning_rate": 5.0395087423892494e-05,
      "loss": 11.8824,
      "step": 24801
    },
    {
      "epoch": 1.3505680232518564,
      "grad_norm": 0.5113084742992144,
      "learning_rate": 5.0387430772082166e-05,
      "loss": 11.819,
      "step": 24802
    },
    {
      "epoch": 1.3506224772484394,
      "grad_norm": 0.5640317571968053,
      "learning_rate": 5.0379774506067455e-05,
      "loss": 11.7749,
      "step": 24803
    },
    {
      "epoch": 1.3506769312450226,
      "grad_norm": 0.661235023433339,
      "learning_rate": 5.0372118625907984e-05,
      "loss": 12.0037,
      "step": 24804
    },
    {
      "epoch": 1.3507313852416056,
      "grad_norm": 0.5357440654739187,
      "learning_rate": 5.036446313166319e-05,
      "loss": 11.9756,
      "step": 24805
    },
    {
      "epoch": 1.3507858392381886,
      "grad_norm": 0.5308581111306975,
      "learning_rate": 5.03568080233927e-05,
      "loss": 11.8394,
      "step": 24806
    },
    {
      "epoch": 1.3508402932347716,
      "grad_norm": 0.4946856917540988,
      "learning_rate": 5.034915330115595e-05,
      "loss": 11.8253,
      "step": 24807
    },
    {
      "epoch": 1.3508947472313546,
      "grad_norm": 0.5021633837792272,
      "learning_rate": 5.034149896501252e-05,
      "loss": 11.9361,
      "step": 24808
    },
    {
      "epoch": 1.3509492012279376,
      "grad_norm": 0.5608247897623255,
      "learning_rate": 5.0333845015021966e-05,
      "loss": 11.9191,
      "step": 24809
    },
    {
      "epoch": 1.3510036552245206,
      "grad_norm": 0.5198315437772559,
      "learning_rate": 5.032619145124371e-05,
      "loss": 11.9982,
      "step": 24810
    },
    {
      "epoch": 1.3510581092211036,
      "grad_norm": 0.6440230274453169,
      "learning_rate": 5.0318538273737406e-05,
      "loss": 11.9214,
      "step": 24811
    },
    {
      "epoch": 1.3511125632176866,
      "grad_norm": 0.5506894777910234,
      "learning_rate": 5.031088548256239e-05,
      "loss": 11.9466,
      "step": 24812
    },
    {
      "epoch": 1.3511670172142698,
      "grad_norm": 0.5293822800229081,
      "learning_rate": 5.0303233077778265e-05,
      "loss": 11.9223,
      "step": 24813
    },
    {
      "epoch": 1.3512214712108528,
      "grad_norm": 0.5591197450370005,
      "learning_rate": 5.0295581059444584e-05,
      "loss": 11.8849,
      "step": 24814
    },
    {
      "epoch": 1.3512759252074358,
      "grad_norm": 0.5050360193004697,
      "learning_rate": 5.028792942762075e-05,
      "loss": 11.8091,
      "step": 24815
    },
    {
      "epoch": 1.3513303792040188,
      "grad_norm": 0.49635674747590736,
      "learning_rate": 5.0280278182366356e-05,
      "loss": 11.9033,
      "step": 24816
    },
    {
      "epoch": 1.3513848332006018,
      "grad_norm": 0.5362261060313038,
      "learning_rate": 5.0272627323740815e-05,
      "loss": 11.9254,
      "step": 24817
    },
    {
      "epoch": 1.3514392871971848,
      "grad_norm": 0.5232223570308987,
      "learning_rate": 5.0264976851803716e-05,
      "loss": 11.8598,
      "step": 24818
    },
    {
      "epoch": 1.3514937411937678,
      "grad_norm": 0.5556135801606398,
      "learning_rate": 5.025732676661444e-05,
      "loss": 11.9191,
      "step": 24819
    },
    {
      "epoch": 1.3515481951903507,
      "grad_norm": 0.5240704904942667,
      "learning_rate": 5.0249677068232534e-05,
      "loss": 11.9336,
      "step": 24820
    },
    {
      "epoch": 1.3516026491869337,
      "grad_norm": 0.5573539217851793,
      "learning_rate": 5.024202775671758e-05,
      "loss": 11.9374,
      "step": 24821
    },
    {
      "epoch": 1.3516571031835167,
      "grad_norm": 0.5501697310148302,
      "learning_rate": 5.023437883212887e-05,
      "loss": 11.9706,
      "step": 24822
    },
    {
      "epoch": 1.3517115571800997,
      "grad_norm": 0.5063270064138293,
      "learning_rate": 5.022673029452598e-05,
      "loss": 11.9344,
      "step": 24823
    },
    {
      "epoch": 1.3517660111766827,
      "grad_norm": 0.49663763397517213,
      "learning_rate": 5.021908214396842e-05,
      "loss": 11.9286,
      "step": 24824
    },
    {
      "epoch": 1.3518204651732657,
      "grad_norm": 0.670876281443096,
      "learning_rate": 5.021143438051559e-05,
      "loss": 11.8892,
      "step": 24825
    },
    {
      "epoch": 1.3518749191698487,
      "grad_norm": 0.5606586921665635,
      "learning_rate": 5.020378700422702e-05,
      "loss": 12.0507,
      "step": 24826
    },
    {
      "epoch": 1.351929373166432,
      "grad_norm": 0.5107979228761818,
      "learning_rate": 5.019614001516213e-05,
      "loss": 11.947,
      "step": 24827
    },
    {
      "epoch": 1.351983827163015,
      "grad_norm": 0.5142782902402404,
      "learning_rate": 5.018849341338043e-05,
      "loss": 11.9033,
      "step": 24828
    },
    {
      "epoch": 1.352038281159598,
      "grad_norm": 0.5959443943528853,
      "learning_rate": 5.018084719894134e-05,
      "loss": 11.8944,
      "step": 24829
    },
    {
      "epoch": 1.352092735156181,
      "grad_norm": 0.5487928972675663,
      "learning_rate": 5.0173201371904336e-05,
      "loss": 11.912,
      "step": 24830
    },
    {
      "epoch": 1.3521471891527639,
      "grad_norm": 0.5716065146438394,
      "learning_rate": 5.016555593232893e-05,
      "loss": 11.9627,
      "step": 24831
    },
    {
      "epoch": 1.3522016431493469,
      "grad_norm": 0.5838625403119779,
      "learning_rate": 5.015791088027446e-05,
      "loss": 11.8686,
      "step": 24832
    },
    {
      "epoch": 1.3522560971459299,
      "grad_norm": 0.5147912815954611,
      "learning_rate": 5.0150266215800454e-05,
      "loss": 11.9449,
      "step": 24833
    },
    {
      "epoch": 1.3523105511425129,
      "grad_norm": 0.5475568976877669,
      "learning_rate": 5.014262193896632e-05,
      "loss": 11.887,
      "step": 24834
    },
    {
      "epoch": 1.3523650051390959,
      "grad_norm": 0.5511623497369028,
      "learning_rate": 5.013497804983152e-05,
      "loss": 11.9286,
      "step": 24835
    },
    {
      "epoch": 1.352419459135679,
      "grad_norm": 0.5407169117373076,
      "learning_rate": 5.0127334548455504e-05,
      "loss": 11.8254,
      "step": 24836
    },
    {
      "epoch": 1.352473913132262,
      "grad_norm": 0.5617974496305375,
      "learning_rate": 5.0119691434897685e-05,
      "loss": 11.8755,
      "step": 24837
    },
    {
      "epoch": 1.352528367128845,
      "grad_norm": 0.5490719035581467,
      "learning_rate": 5.011204870921754e-05,
      "loss": 11.8921,
      "step": 24838
    },
    {
      "epoch": 1.352582821125428,
      "grad_norm": 0.5316405625088215,
      "learning_rate": 5.0104406371474435e-05,
      "loss": 11.9682,
      "step": 24839
    },
    {
      "epoch": 1.352637275122011,
      "grad_norm": 0.5493005504319712,
      "learning_rate": 5.009676442172787e-05,
      "loss": 11.8559,
      "step": 24840
    },
    {
      "epoch": 1.352691729118594,
      "grad_norm": 0.5042634568948384,
      "learning_rate": 5.008912286003723e-05,
      "loss": 11.8346,
      "step": 24841
    },
    {
      "epoch": 1.352746183115177,
      "grad_norm": 0.6517204678807134,
      "learning_rate": 5.008148168646191e-05,
      "loss": 11.9003,
      "step": 24842
    },
    {
      "epoch": 1.35280063711176,
      "grad_norm": 0.5750013340571146,
      "learning_rate": 5.0073840901061395e-05,
      "loss": 11.9398,
      "step": 24843
    },
    {
      "epoch": 1.352855091108343,
      "grad_norm": 0.5844993004803447,
      "learning_rate": 5.006620050389502e-05,
      "loss": 11.9265,
      "step": 24844
    },
    {
      "epoch": 1.352909545104926,
      "grad_norm": 0.5277049449750013,
      "learning_rate": 5.005856049502229e-05,
      "loss": 11.9073,
      "step": 24845
    },
    {
      "epoch": 1.352963999101509,
      "grad_norm": 0.4876604087834955,
      "learning_rate": 5.005092087450254e-05,
      "loss": 11.8304,
      "step": 24846
    },
    {
      "epoch": 1.353018453098092,
      "grad_norm": 0.5619475806379481,
      "learning_rate": 5.004328164239518e-05,
      "loss": 11.9532,
      "step": 24847
    },
    {
      "epoch": 1.353072907094675,
      "grad_norm": 0.53215760304349,
      "learning_rate": 5.003564279875969e-05,
      "loss": 11.8062,
      "step": 24848
    },
    {
      "epoch": 1.353127361091258,
      "grad_norm": 0.5662691808049553,
      "learning_rate": 5.00280043436554e-05,
      "loss": 11.9444,
      "step": 24849
    },
    {
      "epoch": 1.353181815087841,
      "grad_norm": 0.5046826535678688,
      "learning_rate": 5.0020366277141754e-05,
      "loss": 11.8574,
      "step": 24850
    },
    {
      "epoch": 1.3532362690844242,
      "grad_norm": 0.5465399341865931,
      "learning_rate": 5.0012728599278114e-05,
      "loss": 11.9395,
      "step": 24851
    },
    {
      "epoch": 1.3532907230810072,
      "grad_norm": 0.5409542996884615,
      "learning_rate": 5.000509131012384e-05,
      "loss": 11.8603,
      "step": 24852
    },
    {
      "epoch": 1.3533451770775902,
      "grad_norm": 0.587445380148168,
      "learning_rate": 4.999745440973841e-05,
      "loss": 11.8569,
      "step": 24853
    },
    {
      "epoch": 1.3533996310741732,
      "grad_norm": 0.5277849863970896,
      "learning_rate": 4.998981789818111e-05,
      "loss": 11.8868,
      "step": 24854
    },
    {
      "epoch": 1.3534540850707562,
      "grad_norm": 0.5301040362510373,
      "learning_rate": 4.998218177551143e-05,
      "loss": 11.8316,
      "step": 24855
    },
    {
      "epoch": 1.3535085390673391,
      "grad_norm": 0.5613774863843418,
      "learning_rate": 4.997454604178864e-05,
      "loss": 11.9655,
      "step": 24856
    },
    {
      "epoch": 1.3535629930639221,
      "grad_norm": 0.544793573707874,
      "learning_rate": 4.9966910697072166e-05,
      "loss": 11.9052,
      "step": 24857
    },
    {
      "epoch": 1.3536174470605051,
      "grad_norm": 0.5252444204166443,
      "learning_rate": 4.995927574142143e-05,
      "loss": 11.898,
      "step": 24858
    },
    {
      "epoch": 1.3536719010570883,
      "grad_norm": 0.5068669077648342,
      "learning_rate": 4.995164117489571e-05,
      "loss": 11.7729,
      "step": 24859
    },
    {
      "epoch": 1.3537263550536713,
      "grad_norm": 0.502792964628515,
      "learning_rate": 4.9944006997554474e-05,
      "loss": 11.9333,
      "step": 24860
    },
    {
      "epoch": 1.3537808090502543,
      "grad_norm": 0.5630809332803862,
      "learning_rate": 4.993637320945699e-05,
      "loss": 11.9314,
      "step": 24861
    },
    {
      "epoch": 1.3538352630468373,
      "grad_norm": 0.5063112968886009,
      "learning_rate": 4.9928739810662696e-05,
      "loss": 11.8642,
      "step": 24862
    },
    {
      "epoch": 1.3538897170434203,
      "grad_norm": 0.6370058690335387,
      "learning_rate": 4.992110680123092e-05,
      "loss": 11.953,
      "step": 24863
    },
    {
      "epoch": 1.3539441710400033,
      "grad_norm": 0.611175174799429,
      "learning_rate": 4.991347418122099e-05,
      "loss": 11.8584,
      "step": 24864
    },
    {
      "epoch": 1.3539986250365863,
      "grad_norm": 0.562780777127195,
      "learning_rate": 4.990584195069231e-05,
      "loss": 11.8797,
      "step": 24865
    },
    {
      "epoch": 1.3540530790331693,
      "grad_norm": 0.5177942871062793,
      "learning_rate": 4.989821010970417e-05,
      "loss": 11.8644,
      "step": 24866
    },
    {
      "epoch": 1.3541075330297523,
      "grad_norm": 0.5616015574802201,
      "learning_rate": 4.989057865831599e-05,
      "loss": 11.9686,
      "step": 24867
    },
    {
      "epoch": 1.3541619870263353,
      "grad_norm": 0.5276231084759514,
      "learning_rate": 4.988294759658703e-05,
      "loss": 11.8057,
      "step": 24868
    },
    {
      "epoch": 1.3542164410229183,
      "grad_norm": 0.550768685905666,
      "learning_rate": 4.9875316924576676e-05,
      "loss": 11.9847,
      "step": 24869
    },
    {
      "epoch": 1.3542708950195013,
      "grad_norm": 0.5744872109531516,
      "learning_rate": 4.9867686642344304e-05,
      "loss": 11.9256,
      "step": 24870
    },
    {
      "epoch": 1.3543253490160843,
      "grad_norm": 0.5274101657090788,
      "learning_rate": 4.986005674994917e-05,
      "loss": 11.9319,
      "step": 24871
    },
    {
      "epoch": 1.3543798030126672,
      "grad_norm": 0.5618023876300211,
      "learning_rate": 4.985242724745067e-05,
      "loss": 11.9709,
      "step": 24872
    },
    {
      "epoch": 1.3544342570092502,
      "grad_norm": 0.5078380684729866,
      "learning_rate": 4.984479813490812e-05,
      "loss": 11.7467,
      "step": 24873
    },
    {
      "epoch": 1.3544887110058335,
      "grad_norm": 0.5867966059024414,
      "learning_rate": 4.983716941238079e-05,
      "loss": 11.7997,
      "step": 24874
    },
    {
      "epoch": 1.3545431650024164,
      "grad_norm": 0.5338787637271132,
      "learning_rate": 4.9829541079928075e-05,
      "loss": 11.7774,
      "step": 24875
    },
    {
      "epoch": 1.3545976189989994,
      "grad_norm": 0.5902886660661296,
      "learning_rate": 4.9821913137609224e-05,
      "loss": 11.956,
      "step": 24876
    },
    {
      "epoch": 1.3546520729955824,
      "grad_norm": 0.5348857228702834,
      "learning_rate": 4.9814285585483634e-05,
      "loss": 11.9083,
      "step": 24877
    },
    {
      "epoch": 1.3547065269921654,
      "grad_norm": 0.5916120896647244,
      "learning_rate": 4.980665842361054e-05,
      "loss": 11.9277,
      "step": 24878
    },
    {
      "epoch": 1.3547609809887484,
      "grad_norm": 0.5844802679823214,
      "learning_rate": 4.9799031652049277e-05,
      "loss": 11.8835,
      "step": 24879
    },
    {
      "epoch": 1.3548154349853314,
      "grad_norm": 0.5590944803749194,
      "learning_rate": 4.9791405270859215e-05,
      "loss": 11.9702,
      "step": 24880
    },
    {
      "epoch": 1.3548698889819144,
      "grad_norm": 0.6653605419238408,
      "learning_rate": 4.9783779280099564e-05,
      "loss": 12.0233,
      "step": 24881
    },
    {
      "epoch": 1.3549243429784974,
      "grad_norm": 0.5345790397368297,
      "learning_rate": 4.977615367982973e-05,
      "loss": 11.8585,
      "step": 24882
    },
    {
      "epoch": 1.3549787969750806,
      "grad_norm": 0.4899892796727021,
      "learning_rate": 4.976852847010889e-05,
      "loss": 11.6916,
      "step": 24883
    },
    {
      "epoch": 1.3550332509716636,
      "grad_norm": 0.5972743588836793,
      "learning_rate": 4.976090365099638e-05,
      "loss": 11.9908,
      "step": 24884
    },
    {
      "epoch": 1.3550877049682466,
      "grad_norm": 0.6000639830695981,
      "learning_rate": 4.975327922255155e-05,
      "loss": 11.8087,
      "step": 24885
    },
    {
      "epoch": 1.3551421589648296,
      "grad_norm": 0.5487570979590168,
      "learning_rate": 4.974565518483362e-05,
      "loss": 11.9484,
      "step": 24886
    },
    {
      "epoch": 1.3551966129614126,
      "grad_norm": 0.5623411375132161,
      "learning_rate": 4.973803153790193e-05,
      "loss": 11.8157,
      "step": 24887
    },
    {
      "epoch": 1.3552510669579956,
      "grad_norm": 0.606979887834055,
      "learning_rate": 4.97304082818157e-05,
      "loss": 11.8969,
      "step": 24888
    },
    {
      "epoch": 1.3553055209545786,
      "grad_norm": 0.5377885831843022,
      "learning_rate": 4.972278541663429e-05,
      "loss": 11.8924,
      "step": 24889
    },
    {
      "epoch": 1.3553599749511616,
      "grad_norm": 0.5205384980636952,
      "learning_rate": 4.9715162942416884e-05,
      "loss": 11.928,
      "step": 24890
    },
    {
      "epoch": 1.3554144289477446,
      "grad_norm": 0.614298542184706,
      "learning_rate": 4.97075408592228e-05,
      "loss": 12.0008,
      "step": 24891
    },
    {
      "epoch": 1.3554688829443275,
      "grad_norm": 0.5766063853548465,
      "learning_rate": 4.9699919167111395e-05,
      "loss": 11.9444,
      "step": 24892
    },
    {
      "epoch": 1.3555233369409105,
      "grad_norm": 0.5616804705502019,
      "learning_rate": 4.969229786614179e-05,
      "loss": 11.9267,
      "step": 24893
    },
    {
      "epoch": 1.3555777909374935,
      "grad_norm": 0.5489564411038799,
      "learning_rate": 4.968467695637333e-05,
      "loss": 11.9348,
      "step": 24894
    },
    {
      "epoch": 1.3556322449340765,
      "grad_norm": 0.6474272376209784,
      "learning_rate": 4.9677056437865244e-05,
      "loss": 11.5415,
      "step": 24895
    },
    {
      "epoch": 1.3556866989306595,
      "grad_norm": 0.5156865616846723,
      "learning_rate": 4.9669436310676785e-05,
      "loss": 11.8739,
      "step": 24896
    },
    {
      "epoch": 1.3557411529272427,
      "grad_norm": 0.5824334064447036,
      "learning_rate": 4.966181657486728e-05,
      "loss": 12.044,
      "step": 24897
    },
    {
      "epoch": 1.3557956069238257,
      "grad_norm": 0.5951441337550208,
      "learning_rate": 4.9654197230495895e-05,
      "loss": 11.8968,
      "step": 24898
    },
    {
      "epoch": 1.3558500609204087,
      "grad_norm": 0.5427233998533754,
      "learning_rate": 4.9646578277621947e-05,
      "loss": 11.7835,
      "step": 24899
    },
    {
      "epoch": 1.3559045149169917,
      "grad_norm": 0.5475569697171353,
      "learning_rate": 4.963895971630461e-05,
      "loss": 11.8824,
      "step": 24900
    },
    {
      "epoch": 1.3559589689135747,
      "grad_norm": 0.5679944901373409,
      "learning_rate": 4.9631341546603215e-05,
      "loss": 11.8764,
      "step": 24901
    },
    {
      "epoch": 1.3560134229101577,
      "grad_norm": 0.5977108933058346,
      "learning_rate": 4.962372376857695e-05,
      "loss": 12.0388,
      "step": 24902
    },
    {
      "epoch": 1.3560678769067407,
      "grad_norm": 0.5859425763958728,
      "learning_rate": 4.961610638228502e-05,
      "loss": 11.8136,
      "step": 24903
    },
    {
      "epoch": 1.3561223309033237,
      "grad_norm": 0.5281229218351752,
      "learning_rate": 4.960848938778673e-05,
      "loss": 11.867,
      "step": 24904
    },
    {
      "epoch": 1.3561767848999067,
      "grad_norm": 0.5318152722254618,
      "learning_rate": 4.960087278514124e-05,
      "loss": 11.8967,
      "step": 24905
    },
    {
      "epoch": 1.3562312388964899,
      "grad_norm": 0.5261775820020697,
      "learning_rate": 4.959325657440782e-05,
      "loss": 11.8881,
      "step": 24906
    },
    {
      "epoch": 1.3562856928930729,
      "grad_norm": 0.6243009110779695,
      "learning_rate": 4.958564075564573e-05,
      "loss": 11.922,
      "step": 24907
    },
    {
      "epoch": 1.3563401468896559,
      "grad_norm": 0.5097660088059349,
      "learning_rate": 4.95780253289141e-05,
      "loss": 11.9009,
      "step": 24908
    },
    {
      "epoch": 1.3563946008862389,
      "grad_norm": 0.5416975746395439,
      "learning_rate": 4.957041029427225e-05,
      "loss": 11.917,
      "step": 24909
    },
    {
      "epoch": 1.3564490548828219,
      "grad_norm": 0.5849221474800707,
      "learning_rate": 4.9562795651779316e-05,
      "loss": 11.9707,
      "step": 24910
    },
    {
      "epoch": 1.3565035088794049,
      "grad_norm": 0.5897685853806195,
      "learning_rate": 4.955518140149456e-05,
      "loss": 11.9131,
      "step": 24911
    },
    {
      "epoch": 1.3565579628759878,
      "grad_norm": 0.5436011638671914,
      "learning_rate": 4.954756754347719e-05,
      "loss": 11.7322,
      "step": 24912
    },
    {
      "epoch": 1.3566124168725708,
      "grad_norm": 0.6011848819011538,
      "learning_rate": 4.9539954077786353e-05,
      "loss": 11.8981,
      "step": 24913
    },
    {
      "epoch": 1.3566668708691538,
      "grad_norm": 0.5648134955862548,
      "learning_rate": 4.953234100448133e-05,
      "loss": 12.0249,
      "step": 24914
    },
    {
      "epoch": 1.3567213248657368,
      "grad_norm": 0.5168500769792862,
      "learning_rate": 4.952472832362125e-05,
      "loss": 11.8804,
      "step": 24915
    },
    {
      "epoch": 1.3567757788623198,
      "grad_norm": 0.6142474939863247,
      "learning_rate": 4.951711603526539e-05,
      "loss": 11.8979,
      "step": 24916
    },
    {
      "epoch": 1.3568302328589028,
      "grad_norm": 0.5684425158241186,
      "learning_rate": 4.9509504139472854e-05,
      "loss": 11.9123,
      "step": 24917
    },
    {
      "epoch": 1.3568846868554858,
      "grad_norm": 0.5667423344177543,
      "learning_rate": 4.950189263630288e-05,
      "loss": 11.9643,
      "step": 24918
    },
    {
      "epoch": 1.3569391408520688,
      "grad_norm": 0.5771960361685287,
      "learning_rate": 4.949428152581471e-05,
      "loss": 11.859,
      "step": 24919
    },
    {
      "epoch": 1.3569935948486518,
      "grad_norm": 0.5392846440373192,
      "learning_rate": 4.9486670808067426e-05,
      "loss": 11.9325,
      "step": 24920
    },
    {
      "epoch": 1.357048048845235,
      "grad_norm": 0.5666340357531485,
      "learning_rate": 4.947906048312031e-05,
      "loss": 12.0368,
      "step": 24921
    },
    {
      "epoch": 1.357102502841818,
      "grad_norm": 0.5274722567663747,
      "learning_rate": 4.9471450551032485e-05,
      "loss": 11.851,
      "step": 24922
    },
    {
      "epoch": 1.357156956838401,
      "grad_norm": 0.5371877229027102,
      "learning_rate": 4.94638410118631e-05,
      "loss": 11.8442,
      "step": 24923
    },
    {
      "epoch": 1.357211410834984,
      "grad_norm": 0.5163432852791142,
      "learning_rate": 4.9456231865671386e-05,
      "loss": 12.0214,
      "step": 24924
    },
    {
      "epoch": 1.357265864831567,
      "grad_norm": 0.5429702860581987,
      "learning_rate": 4.944862311251647e-05,
      "loss": 11.8162,
      "step": 24925
    },
    {
      "epoch": 1.35732031882815,
      "grad_norm": 0.49390150469090427,
      "learning_rate": 4.944101475245756e-05,
      "loss": 11.7471,
      "step": 24926
    },
    {
      "epoch": 1.357374772824733,
      "grad_norm": 0.5073739712088102,
      "learning_rate": 4.943340678555377e-05,
      "loss": 11.9097,
      "step": 24927
    },
    {
      "epoch": 1.357429226821316,
      "grad_norm": 0.5357613198551402,
      "learning_rate": 4.9425799211864286e-05,
      "loss": 11.8786,
      "step": 24928
    },
    {
      "epoch": 1.3574836808178992,
      "grad_norm": 0.5262640462521961,
      "learning_rate": 4.9418192031448306e-05,
      "loss": 11.9509,
      "step": 24929
    },
    {
      "epoch": 1.3575381348144822,
      "grad_norm": 0.595498353326288,
      "learning_rate": 4.9410585244364916e-05,
      "loss": 11.9361,
      "step": 24930
    },
    {
      "epoch": 1.3575925888110651,
      "grad_norm": 0.6056796594259529,
      "learning_rate": 4.940297885067333e-05,
      "loss": 11.8909,
      "step": 24931
    },
    {
      "epoch": 1.3576470428076481,
      "grad_norm": 0.5782904163323127,
      "learning_rate": 4.939537285043266e-05,
      "loss": 12.0208,
      "step": 24932
    },
    {
      "epoch": 1.3577014968042311,
      "grad_norm": 0.49495929222631285,
      "learning_rate": 4.938776724370203e-05,
      "loss": 11.7721,
      "step": 24933
    },
    {
      "epoch": 1.3577559508008141,
      "grad_norm": 0.5133983141652042,
      "learning_rate": 4.938016203054064e-05,
      "loss": 11.861,
      "step": 24934
    },
    {
      "epoch": 1.3578104047973971,
      "grad_norm": 0.4951558290143576,
      "learning_rate": 4.9372557211007574e-05,
      "loss": 11.9239,
      "step": 24935
    },
    {
      "epoch": 1.3578648587939801,
      "grad_norm": 0.5733830050411346,
      "learning_rate": 4.936495278516202e-05,
      "loss": 11.8048,
      "step": 24936
    },
    {
      "epoch": 1.357919312790563,
      "grad_norm": 0.5362533987137719,
      "learning_rate": 4.935734875306305e-05,
      "loss": 11.9157,
      "step": 24937
    },
    {
      "epoch": 1.357973766787146,
      "grad_norm": 0.6084958145610728,
      "learning_rate": 4.9349745114769863e-05,
      "loss": 11.7731,
      "step": 24938
    },
    {
      "epoch": 1.358028220783729,
      "grad_norm": 0.5651082642116758,
      "learning_rate": 4.934214187034152e-05,
      "loss": 11.86,
      "step": 24939
    },
    {
      "epoch": 1.358082674780312,
      "grad_norm": 0.5388066316508631,
      "learning_rate": 4.933453901983718e-05,
      "loss": 11.8609,
      "step": 24940
    },
    {
      "epoch": 1.358137128776895,
      "grad_norm": 0.550791723540091,
      "learning_rate": 4.9326936563315984e-05,
      "loss": 11.9139,
      "step": 24941
    },
    {
      "epoch": 1.358191582773478,
      "grad_norm": 0.5582596246482534,
      "learning_rate": 4.931933450083704e-05,
      "loss": 11.908,
      "step": 24942
    },
    {
      "epoch": 1.358246036770061,
      "grad_norm": 0.5569203634323466,
      "learning_rate": 4.9311732832459446e-05,
      "loss": 11.9891,
      "step": 24943
    },
    {
      "epoch": 1.3583004907666443,
      "grad_norm": 0.5156216176427671,
      "learning_rate": 4.930413155824228e-05,
      "loss": 11.7841,
      "step": 24944
    },
    {
      "epoch": 1.3583549447632273,
      "grad_norm": 0.5196163341381659,
      "learning_rate": 4.929653067824469e-05,
      "loss": 11.9316,
      "step": 24945
    },
    {
      "epoch": 1.3584093987598103,
      "grad_norm": 0.5008788675494064,
      "learning_rate": 4.9288930192525815e-05,
      "loss": 11.9397,
      "step": 24946
    },
    {
      "epoch": 1.3584638527563933,
      "grad_norm": 0.5955207579212103,
      "learning_rate": 4.928133010114469e-05,
      "loss": 11.9151,
      "step": 24947
    },
    {
      "epoch": 1.3585183067529762,
      "grad_norm": 0.6049074577657653,
      "learning_rate": 4.927373040416049e-05,
      "loss": 11.7953,
      "step": 24948
    },
    {
      "epoch": 1.3585727607495592,
      "grad_norm": 0.5651218456736897,
      "learning_rate": 4.926613110163222e-05,
      "loss": 11.9555,
      "step": 24949
    },
    {
      "epoch": 1.3586272147461422,
      "grad_norm": 0.5609675285351408,
      "learning_rate": 4.9258532193619064e-05,
      "loss": 11.8516,
      "step": 24950
    },
    {
      "epoch": 1.3586816687427252,
      "grad_norm": 0.5203702149403598,
      "learning_rate": 4.925093368018004e-05,
      "loss": 11.8554,
      "step": 24951
    },
    {
      "epoch": 1.3587361227393082,
      "grad_norm": 0.5292728328043207,
      "learning_rate": 4.924333556137429e-05,
      "loss": 11.9143,
      "step": 24952
    },
    {
      "epoch": 1.3587905767358914,
      "grad_norm": 0.5660082640547254,
      "learning_rate": 4.9235737837260885e-05,
      "loss": 12.0576,
      "step": 24953
    },
    {
      "epoch": 1.3588450307324744,
      "grad_norm": 0.5166251485491004,
      "learning_rate": 4.922814050789886e-05,
      "loss": 11.8077,
      "step": 24954
    },
    {
      "epoch": 1.3588994847290574,
      "grad_norm": 0.5195651561494627,
      "learning_rate": 4.9220543573347324e-05,
      "loss": 11.809,
      "step": 24955
    },
    {
      "epoch": 1.3589539387256404,
      "grad_norm": 0.56371074239359,
      "learning_rate": 4.921294703366539e-05,
      "loss": 11.8172,
      "step": 24956
    },
    {
      "epoch": 1.3590083927222234,
      "grad_norm": 0.5142443462087007,
      "learning_rate": 4.920535088891207e-05,
      "loss": 11.727,
      "step": 24957
    },
    {
      "epoch": 1.3590628467188064,
      "grad_norm": 0.5345036061525329,
      "learning_rate": 4.9197755139146495e-05,
      "loss": 11.8688,
      "step": 24958
    },
    {
      "epoch": 1.3591173007153894,
      "grad_norm": 0.613562091302831,
      "learning_rate": 4.919015978442765e-05,
      "loss": 11.9773,
      "step": 24959
    },
    {
      "epoch": 1.3591717547119724,
      "grad_norm": 0.5397915387932382,
      "learning_rate": 4.9182564824814684e-05,
      "loss": 11.7113,
      "step": 24960
    },
    {
      "epoch": 1.3592262087085554,
      "grad_norm": 0.549886462065191,
      "learning_rate": 4.9174970260366585e-05,
      "loss": 11.946,
      "step": 24961
    },
    {
      "epoch": 1.3592806627051384,
      "grad_norm": 0.550303995844349,
      "learning_rate": 4.916737609114247e-05,
      "loss": 11.8847,
      "step": 24962
    },
    {
      "epoch": 1.3593351167017214,
      "grad_norm": 0.5478981988353996,
      "learning_rate": 4.9159782317201364e-05,
      "loss": 11.9462,
      "step": 24963
    },
    {
      "epoch": 1.3593895706983044,
      "grad_norm": 0.5773689655458489,
      "learning_rate": 4.9152188938602285e-05,
      "loss": 11.759,
      "step": 24964
    },
    {
      "epoch": 1.3594440246948873,
      "grad_norm": 0.5539399400396439,
      "learning_rate": 4.914459595540435e-05,
      "loss": 11.8204,
      "step": 24965
    },
    {
      "epoch": 1.3594984786914703,
      "grad_norm": 0.5791466590168848,
      "learning_rate": 4.9137003367666525e-05,
      "loss": 12.0109,
      "step": 24966
    },
    {
      "epoch": 1.3595529326880536,
      "grad_norm": 0.5757685910387154,
      "learning_rate": 4.912941117544788e-05,
      "loss": 11.9751,
      "step": 24967
    },
    {
      "epoch": 1.3596073866846365,
      "grad_norm": 0.545043819119108,
      "learning_rate": 4.912181937880752e-05,
      "loss": 11.8309,
      "step": 24968
    },
    {
      "epoch": 1.3596618406812195,
      "grad_norm": 0.527299133856043,
      "learning_rate": 4.9114227977804384e-05,
      "loss": 11.8493,
      "step": 24969
    },
    {
      "epoch": 1.3597162946778025,
      "grad_norm": 0.5551793271739268,
      "learning_rate": 4.9106636972497575e-05,
      "loss": 12.0106,
      "step": 24970
    },
    {
      "epoch": 1.3597707486743855,
      "grad_norm": 0.5471731565250786,
      "learning_rate": 4.9099046362946053e-05,
      "loss": 11.8008,
      "step": 24971
    },
    {
      "epoch": 1.3598252026709685,
      "grad_norm": 0.5184502680272319,
      "learning_rate": 4.909145614920893e-05,
      "loss": 11.9077,
      "step": 24972
    },
    {
      "epoch": 1.3598796566675515,
      "grad_norm": 0.5701095237909425,
      "learning_rate": 4.9083866331345174e-05,
      "loss": 11.8327,
      "step": 24973
    },
    {
      "epoch": 1.3599341106641345,
      "grad_norm": 0.6190660343408305,
      "learning_rate": 4.907627690941378e-05,
      "loss": 11.9276,
      "step": 24974
    },
    {
      "epoch": 1.3599885646607175,
      "grad_norm": 0.5447161710312252,
      "learning_rate": 4.906868788347383e-05,
      "loss": 11.9611,
      "step": 24975
    },
    {
      "epoch": 1.3600430186573007,
      "grad_norm": 0.58208473872352,
      "learning_rate": 4.9061099253584264e-05,
      "loss": 11.8794,
      "step": 24976
    },
    {
      "epoch": 1.3600974726538837,
      "grad_norm": 0.5814764443183652,
      "learning_rate": 4.9053511019804136e-05,
      "loss": 11.9842,
      "step": 24977
    },
    {
      "epoch": 1.3601519266504667,
      "grad_norm": 0.5274231067978153,
      "learning_rate": 4.904592318219249e-05,
      "loss": 11.9158,
      "step": 24978
    },
    {
      "epoch": 1.3602063806470497,
      "grad_norm": 0.5660687616682143,
      "learning_rate": 4.903833574080825e-05,
      "loss": 11.879,
      "step": 24979
    },
    {
      "epoch": 1.3602608346436327,
      "grad_norm": 0.5969691407107678,
      "learning_rate": 4.9030748695710494e-05,
      "loss": 11.8296,
      "step": 24980
    },
    {
      "epoch": 1.3603152886402157,
      "grad_norm": 0.532669920349359,
      "learning_rate": 4.902316204695815e-05,
      "loss": 11.8981,
      "step": 24981
    },
    {
      "epoch": 1.3603697426367987,
      "grad_norm": 0.5229960340859549,
      "learning_rate": 4.901557579461028e-05,
      "loss": 11.9678,
      "step": 24982
    },
    {
      "epoch": 1.3604241966333817,
      "grad_norm": 0.5684134918572549,
      "learning_rate": 4.900798993872584e-05,
      "loss": 11.9296,
      "step": 24983
    },
    {
      "epoch": 1.3604786506299646,
      "grad_norm": 0.5987188117018615,
      "learning_rate": 4.900040447936379e-05,
      "loss": 11.9985,
      "step": 24984
    },
    {
      "epoch": 1.3605331046265476,
      "grad_norm": 0.7762637705718533,
      "learning_rate": 4.899281941658319e-05,
      "loss": 11.8949,
      "step": 24985
    },
    {
      "epoch": 1.3605875586231306,
      "grad_norm": 0.5754549163370596,
      "learning_rate": 4.8985234750442934e-05,
      "loss": 11.9371,
      "step": 24986
    },
    {
      "epoch": 1.3606420126197136,
      "grad_norm": 0.5003681898766952,
      "learning_rate": 4.897765048100209e-05,
      "loss": 11.7886,
      "step": 24987
    },
    {
      "epoch": 1.3606964666162966,
      "grad_norm": 0.4882107238229703,
      "learning_rate": 4.897006660831954e-05,
      "loss": 11.8665,
      "step": 24988
    },
    {
      "epoch": 1.3607509206128796,
      "grad_norm": 0.716362334980051,
      "learning_rate": 4.896248313245433e-05,
      "loss": 11.7667,
      "step": 24989
    },
    {
      "epoch": 1.3608053746094628,
      "grad_norm": 0.5407693854731415,
      "learning_rate": 4.895490005346543e-05,
      "loss": 11.914,
      "step": 24990
    },
    {
      "epoch": 1.3608598286060458,
      "grad_norm": 0.5903030712307776,
      "learning_rate": 4.894731737141177e-05,
      "loss": 11.8986,
      "step": 24991
    },
    {
      "epoch": 1.3609142826026288,
      "grad_norm": 0.5072839435812865,
      "learning_rate": 4.89397350863524e-05,
      "loss": 11.892,
      "step": 24992
    },
    {
      "epoch": 1.3609687365992118,
      "grad_norm": 0.501903336905804,
      "learning_rate": 4.893215319834613e-05,
      "loss": 11.8784,
      "step": 24993
    },
    {
      "epoch": 1.3610231905957948,
      "grad_norm": 0.5636339992920806,
      "learning_rate": 4.8924571707452004e-05,
      "loss": 12.0286,
      "step": 24994
    },
    {
      "epoch": 1.3610776445923778,
      "grad_norm": 0.5393615922258653,
      "learning_rate": 4.891699061372901e-05,
      "loss": 11.8979,
      "step": 24995
    },
    {
      "epoch": 1.3611320985889608,
      "grad_norm": 0.6009854715854602,
      "learning_rate": 4.890940991723603e-05,
      "loss": 12.016,
      "step": 24996
    },
    {
      "epoch": 1.3611865525855438,
      "grad_norm": 0.5298412273781102,
      "learning_rate": 4.890182961803208e-05,
      "loss": 11.9466,
      "step": 24997
    },
    {
      "epoch": 1.3612410065821268,
      "grad_norm": 0.568971366230619,
      "learning_rate": 4.8894249716176044e-05,
      "loss": 11.9406,
      "step": 24998
    },
    {
      "epoch": 1.36129546057871,
      "grad_norm": 0.4855817437367864,
      "learning_rate": 4.8886670211726916e-05,
      "loss": 11.8565,
      "step": 24999
    },
    {
      "epoch": 1.361349914575293,
      "grad_norm": 0.5258545939766616,
      "learning_rate": 4.887909110474358e-05,
      "loss": 11.778,
      "step": 25000
    },
    {
      "epoch": 1.361404368571876,
      "grad_norm": 0.5644727999056377,
      "learning_rate": 4.8871512395285015e-05,
      "loss": 11.906,
      "step": 25001
    },
    {
      "epoch": 1.361458822568459,
      "grad_norm": 0.5631330583033027,
      "learning_rate": 4.8863934083410165e-05,
      "loss": 11.9083,
      "step": 25002
    },
    {
      "epoch": 1.361513276565042,
      "grad_norm": 0.5928746067366175,
      "learning_rate": 4.885635616917795e-05,
      "loss": 11.8918,
      "step": 25003
    },
    {
      "epoch": 1.361567730561625,
      "grad_norm": 0.5687408818579682,
      "learning_rate": 4.884877865264724e-05,
      "loss": 11.764,
      "step": 25004
    },
    {
      "epoch": 1.361622184558208,
      "grad_norm": 0.5547182565620317,
      "learning_rate": 4.884120153387706e-05,
      "loss": 11.7977,
      "step": 25005
    },
    {
      "epoch": 1.361676638554791,
      "grad_norm": 0.5406958966746962,
      "learning_rate": 4.8833624812926215e-05,
      "loss": 11.9221,
      "step": 25006
    },
    {
      "epoch": 1.361731092551374,
      "grad_norm": 0.6347223954139799,
      "learning_rate": 4.882604848985374e-05,
      "loss": 11.9636,
      "step": 25007
    },
    {
      "epoch": 1.361785546547957,
      "grad_norm": 0.5340752072264334,
      "learning_rate": 4.881847256471844e-05,
      "loss": 11.8655,
      "step": 25008
    },
    {
      "epoch": 1.36184000054454,
      "grad_norm": 0.5216907951302637,
      "learning_rate": 4.881089703757934e-05,
      "loss": 11.9513,
      "step": 25009
    },
    {
      "epoch": 1.361894454541123,
      "grad_norm": 0.5954455984659678,
      "learning_rate": 4.8803321908495235e-05,
      "loss": 11.9156,
      "step": 25010
    },
    {
      "epoch": 1.361948908537706,
      "grad_norm": 0.5497393167262156,
      "learning_rate": 4.879574717752509e-05,
      "loss": 11.8963,
      "step": 25011
    },
    {
      "epoch": 1.3620033625342889,
      "grad_norm": 0.6359159377486772,
      "learning_rate": 4.878817284472783e-05,
      "loss": 11.9282,
      "step": 25012
    },
    {
      "epoch": 1.3620578165308719,
      "grad_norm": 0.5206324241828418,
      "learning_rate": 4.8780598910162346e-05,
      "loss": 11.8689,
      "step": 25013
    },
    {
      "epoch": 1.362112270527455,
      "grad_norm": 0.5272454399438127,
      "learning_rate": 4.8773025373887496e-05,
      "loss": 11.8483,
      "step": 25014
    },
    {
      "epoch": 1.362166724524038,
      "grad_norm": 0.5282543903076978,
      "learning_rate": 4.8765452235962164e-05,
      "loss": 11.9176,
      "step": 25015
    },
    {
      "epoch": 1.362221178520621,
      "grad_norm": 0.559768472267573,
      "learning_rate": 4.875787949644526e-05,
      "loss": 11.8522,
      "step": 25016
    },
    {
      "epoch": 1.362275632517204,
      "grad_norm": 0.4938271457560536,
      "learning_rate": 4.875030715539572e-05,
      "loss": 11.9746,
      "step": 25017
    },
    {
      "epoch": 1.362330086513787,
      "grad_norm": 0.5901155786033425,
      "learning_rate": 4.874273521287235e-05,
      "loss": 12.0275,
      "step": 25018
    },
    {
      "epoch": 1.36238454051037,
      "grad_norm": 0.5128795830427856,
      "learning_rate": 4.8735163668934105e-05,
      "loss": 11.8649,
      "step": 25019
    },
    {
      "epoch": 1.362438994506953,
      "grad_norm": 0.5243891263339125,
      "learning_rate": 4.872759252363979e-05,
      "loss": 12.0431,
      "step": 25020
    },
    {
      "epoch": 1.362493448503536,
      "grad_norm": 0.5440153012281209,
      "learning_rate": 4.872002177704834e-05,
      "loss": 11.8967,
      "step": 25021
    },
    {
      "epoch": 1.3625479025001193,
      "grad_norm": 0.5309573831848331,
      "learning_rate": 4.871245142921856e-05,
      "loss": 11.8478,
      "step": 25022
    },
    {
      "epoch": 1.3626023564967022,
      "grad_norm": 0.6028129225695973,
      "learning_rate": 4.870488148020941e-05,
      "loss": 12.0187,
      "step": 25023
    },
    {
      "epoch": 1.3626568104932852,
      "grad_norm": 0.5359101509285282,
      "learning_rate": 4.8697311930079705e-05,
      "loss": 11.8899,
      "step": 25024
    },
    {
      "epoch": 1.3627112644898682,
      "grad_norm": 0.5397611360516368,
      "learning_rate": 4.8689742778888256e-05,
      "loss": 11.7515,
      "step": 25025
    },
    {
      "epoch": 1.3627657184864512,
      "grad_norm": 0.524039195851762,
      "learning_rate": 4.868217402669398e-05,
      "loss": 11.8701,
      "step": 25026
    },
    {
      "epoch": 1.3628201724830342,
      "grad_norm": 0.5795503373074804,
      "learning_rate": 4.8674605673555765e-05,
      "loss": 11.9561,
      "step": 25027
    },
    {
      "epoch": 1.3628746264796172,
      "grad_norm": 0.5932548255088791,
      "learning_rate": 4.866703771953238e-05,
      "loss": 11.9275,
      "step": 25028
    },
    {
      "epoch": 1.3629290804762002,
      "grad_norm": 0.4853918644768155,
      "learning_rate": 4.865947016468275e-05,
      "loss": 11.9024,
      "step": 25029
    },
    {
      "epoch": 1.3629835344727832,
      "grad_norm": 0.5411828980209702,
      "learning_rate": 4.865190300906567e-05,
      "loss": 11.8984,
      "step": 25030
    },
    {
      "epoch": 1.3630379884693662,
      "grad_norm": 0.5102095294681734,
      "learning_rate": 4.864433625274003e-05,
      "loss": 11.8875,
      "step": 25031
    },
    {
      "epoch": 1.3630924424659492,
      "grad_norm": 0.5811649139961952,
      "learning_rate": 4.86367698957646e-05,
      "loss": 11.9619,
      "step": 25032
    },
    {
      "epoch": 1.3631468964625322,
      "grad_norm": 0.5633498008262142,
      "learning_rate": 4.862920393819831e-05,
      "loss": 11.889,
      "step": 25033
    },
    {
      "epoch": 1.3632013504591152,
      "grad_norm": 0.5285322694474356,
      "learning_rate": 4.862163838009993e-05,
      "loss": 11.8194,
      "step": 25034
    },
    {
      "epoch": 1.3632558044556982,
      "grad_norm": 0.5347508240447599,
      "learning_rate": 4.8614073221528286e-05,
      "loss": 11.9274,
      "step": 25035
    },
    {
      "epoch": 1.3633102584522812,
      "grad_norm": 0.5788175449431141,
      "learning_rate": 4.8606508462542254e-05,
      "loss": 11.8674,
      "step": 25036
    },
    {
      "epoch": 1.3633647124488644,
      "grad_norm": 0.5147804669991923,
      "learning_rate": 4.85989441032006e-05,
      "loss": 11.824,
      "step": 25037
    },
    {
      "epoch": 1.3634191664454474,
      "grad_norm": 0.5149866176731968,
      "learning_rate": 4.859138014356217e-05,
      "loss": 11.915,
      "step": 25038
    },
    {
      "epoch": 1.3634736204420304,
      "grad_norm": 0.5697637018498767,
      "learning_rate": 4.858381658368583e-05,
      "loss": 11.8232,
      "step": 25039
    },
    {
      "epoch": 1.3635280744386133,
      "grad_norm": 0.5000018076758913,
      "learning_rate": 4.857625342363031e-05,
      "loss": 11.793,
      "step": 25040
    },
    {
      "epoch": 1.3635825284351963,
      "grad_norm": 0.5272847581622794,
      "learning_rate": 4.856869066345452e-05,
      "loss": 11.8191,
      "step": 25041
    },
    {
      "epoch": 1.3636369824317793,
      "grad_norm": 0.5357826017991316,
      "learning_rate": 4.856112830321716e-05,
      "loss": 11.9674,
      "step": 25042
    },
    {
      "epoch": 1.3636914364283623,
      "grad_norm": 0.5568912482017461,
      "learning_rate": 4.855356634297714e-05,
      "loss": 11.8877,
      "step": 25043
    },
    {
      "epoch": 1.3637458904249453,
      "grad_norm": 0.5041668188356421,
      "learning_rate": 4.854600478279322e-05,
      "loss": 11.9672,
      "step": 25044
    },
    {
      "epoch": 1.3638003444215283,
      "grad_norm": 0.5459840938632199,
      "learning_rate": 4.853844362272415e-05,
      "loss": 11.941,
      "step": 25045
    },
    {
      "epoch": 1.3638547984181115,
      "grad_norm": 0.7894134375331131,
      "learning_rate": 4.853088286282882e-05,
      "loss": 12.0943,
      "step": 25046
    },
    {
      "epoch": 1.3639092524146945,
      "grad_norm": 0.572334369362874,
      "learning_rate": 4.8523322503165926e-05,
      "loss": 11.9471,
      "step": 25047
    },
    {
      "epoch": 1.3639637064112775,
      "grad_norm": 0.5214567336702625,
      "learning_rate": 4.851576254379435e-05,
      "loss": 11.8591,
      "step": 25048
    },
    {
      "epoch": 1.3640181604078605,
      "grad_norm": 0.49154675703009915,
      "learning_rate": 4.850820298477281e-05,
      "loss": 11.8707,
      "step": 25049
    },
    {
      "epoch": 1.3640726144044435,
      "grad_norm": 0.5325034004871464,
      "learning_rate": 4.850064382616011e-05,
      "loss": 11.8024,
      "step": 25050
    },
    {
      "epoch": 1.3641270684010265,
      "grad_norm": 0.5189007373633713,
      "learning_rate": 4.849308506801509e-05,
      "loss": 11.8567,
      "step": 25051
    },
    {
      "epoch": 1.3641815223976095,
      "grad_norm": 0.5972202800016274,
      "learning_rate": 4.848552671039642e-05,
      "loss": 11.76,
      "step": 25052
    },
    {
      "epoch": 1.3642359763941925,
      "grad_norm": 0.5665482075263434,
      "learning_rate": 4.847796875336298e-05,
      "loss": 11.8383,
      "step": 25053
    },
    {
      "epoch": 1.3642904303907755,
      "grad_norm": 0.5531479756601952,
      "learning_rate": 4.84704111969735e-05,
      "loss": 11.774,
      "step": 25054
    },
    {
      "epoch": 1.3643448843873585,
      "grad_norm": 0.5419518727756614,
      "learning_rate": 4.84628540412867e-05,
      "loss": 11.8875,
      "step": 25055
    },
    {
      "epoch": 1.3643993383839415,
      "grad_norm": 0.5558341957921653,
      "learning_rate": 4.845529728636143e-05,
      "loss": 11.8634,
      "step": 25056
    },
    {
      "epoch": 1.3644537923805244,
      "grad_norm": 0.614957087424945,
      "learning_rate": 4.844774093225638e-05,
      "loss": 11.7651,
      "step": 25057
    },
    {
      "epoch": 1.3645082463771074,
      "grad_norm": 0.580647493413801,
      "learning_rate": 4.844018497903038e-05,
      "loss": 11.8813,
      "step": 25058
    },
    {
      "epoch": 1.3645627003736904,
      "grad_norm": 0.49358035332314476,
      "learning_rate": 4.8432629426742104e-05,
      "loss": 11.9072,
      "step": 25059
    },
    {
      "epoch": 1.3646171543702736,
      "grad_norm": 0.5582621331307084,
      "learning_rate": 4.8425074275450344e-05,
      "loss": 11.9912,
      "step": 25060
    },
    {
      "epoch": 1.3646716083668566,
      "grad_norm": 0.5544974137118196,
      "learning_rate": 4.8417519525213906e-05,
      "loss": 11.925,
      "step": 25061
    },
    {
      "epoch": 1.3647260623634396,
      "grad_norm": 0.5267663796662333,
      "learning_rate": 4.8409965176091445e-05,
      "loss": 11.8615,
      "step": 25062
    },
    {
      "epoch": 1.3647805163600226,
      "grad_norm": 0.52441552411104,
      "learning_rate": 4.840241122814181e-05,
      "loss": 11.7422,
      "step": 25063
    },
    {
      "epoch": 1.3648349703566056,
      "grad_norm": 0.5639695274470757,
      "learning_rate": 4.839485768142361e-05,
      "loss": 11.8825,
      "step": 25064
    },
    {
      "epoch": 1.3648894243531886,
      "grad_norm": 0.5510800558494531,
      "learning_rate": 4.838730453599566e-05,
      "loss": 11.8495,
      "step": 25065
    },
    {
      "epoch": 1.3649438783497716,
      "grad_norm": 0.5248793378922084,
      "learning_rate": 4.8379751791916716e-05,
      "loss": 11.8267,
      "step": 25066
    },
    {
      "epoch": 1.3649983323463546,
      "grad_norm": 0.5712846925018582,
      "learning_rate": 4.837219944924544e-05,
      "loss": 11.7966,
      "step": 25067
    },
    {
      "epoch": 1.3650527863429376,
      "grad_norm": 0.49709859810052864,
      "learning_rate": 4.836464750804064e-05,
      "loss": 11.929,
      "step": 25068
    },
    {
      "epoch": 1.3651072403395208,
      "grad_norm": 0.587437006672066,
      "learning_rate": 4.8357095968360955e-05,
      "loss": 11.9191,
      "step": 25069
    },
    {
      "epoch": 1.3651616943361038,
      "grad_norm": 0.522421371357169,
      "learning_rate": 4.83495448302652e-05,
      "loss": 11.9529,
      "step": 25070
    },
    {
      "epoch": 1.3652161483326868,
      "grad_norm": 0.5547914272280554,
      "learning_rate": 4.8341994093812005e-05,
      "loss": 11.8742,
      "step": 25071
    },
    {
      "epoch": 1.3652706023292698,
      "grad_norm": 0.5628350573401014,
      "learning_rate": 4.833444375906012e-05,
      "loss": 11.8248,
      "step": 25072
    },
    {
      "epoch": 1.3653250563258528,
      "grad_norm": 0.5454799580843143,
      "learning_rate": 4.8326893826068354e-05,
      "loss": 11.8551,
      "step": 25073
    },
    {
      "epoch": 1.3653795103224358,
      "grad_norm": 0.5680693707193344,
      "learning_rate": 4.831934429489524e-05,
      "loss": 11.8932,
      "step": 25074
    },
    {
      "epoch": 1.3654339643190188,
      "grad_norm": 0.5389331220279573,
      "learning_rate": 4.831179516559958e-05,
      "loss": 11.9218,
      "step": 25075
    },
    {
      "epoch": 1.3654884183156017,
      "grad_norm": 0.6550773298623871,
      "learning_rate": 4.83042464382401e-05,
      "loss": 11.8776,
      "step": 25076
    },
    {
      "epoch": 1.3655428723121847,
      "grad_norm": 0.5385274700697291,
      "learning_rate": 4.829669811287544e-05,
      "loss": 11.9966,
      "step": 25077
    },
    {
      "epoch": 1.3655973263087677,
      "grad_norm": 0.5645934639414184,
      "learning_rate": 4.828915018956435e-05,
      "loss": 12.0673,
      "step": 25078
    },
    {
      "epoch": 1.3656517803053507,
      "grad_norm": 0.526032775194877,
      "learning_rate": 4.828160266836547e-05,
      "loss": 11.8598,
      "step": 25079
    },
    {
      "epoch": 1.3657062343019337,
      "grad_norm": 0.5167559172748148,
      "learning_rate": 4.827405554933756e-05,
      "loss": 11.9208,
      "step": 25080
    },
    {
      "epoch": 1.3657606882985167,
      "grad_norm": 0.5437637163621964,
      "learning_rate": 4.826650883253923e-05,
      "loss": 11.8471,
      "step": 25081
    },
    {
      "epoch": 1.3658151422950997,
      "grad_norm": 0.588792792656736,
      "learning_rate": 4.8258962518029205e-05,
      "loss": 11.7768,
      "step": 25082
    },
    {
      "epoch": 1.3658695962916827,
      "grad_norm": 0.5382521194921016,
      "learning_rate": 4.825141660586623e-05,
      "loss": 11.8772,
      "step": 25083
    },
    {
      "epoch": 1.365924050288266,
      "grad_norm": 0.5069262275811791,
      "learning_rate": 4.824387109610885e-05,
      "loss": 11.9157,
      "step": 25084
    },
    {
      "epoch": 1.365978504284849,
      "grad_norm": 0.6846066968569539,
      "learning_rate": 4.823632598881585e-05,
      "loss": 11.6901,
      "step": 25085
    },
    {
      "epoch": 1.366032958281432,
      "grad_norm": 0.5400538581503838,
      "learning_rate": 4.822878128404581e-05,
      "loss": 11.9139,
      "step": 25086
    },
    {
      "epoch": 1.366087412278015,
      "grad_norm": 0.548696516380634,
      "learning_rate": 4.8221236981857455e-05,
      "loss": 11.8388,
      "step": 25087
    },
    {
      "epoch": 1.3661418662745979,
      "grad_norm": 0.6096702113572322,
      "learning_rate": 4.821369308230949e-05,
      "loss": 12.0497,
      "step": 25088
    },
    {
      "epoch": 1.3661963202711809,
      "grad_norm": 0.5265305841633061,
      "learning_rate": 4.820614958546048e-05,
      "loss": 11.8539,
      "step": 25089
    },
    {
      "epoch": 1.3662507742677639,
      "grad_norm": 0.5527700762439866,
      "learning_rate": 4.819860649136919e-05,
      "loss": 11.9079,
      "step": 25090
    },
    {
      "epoch": 1.3663052282643469,
      "grad_norm": 0.5337874650232489,
      "learning_rate": 4.8191063800094175e-05,
      "loss": 11.9632,
      "step": 25091
    },
    {
      "epoch": 1.36635968226093,
      "grad_norm": 0.5312908667458723,
      "learning_rate": 4.818352151169418e-05,
      "loss": 11.9241,
      "step": 25092
    },
    {
      "epoch": 1.366414136257513,
      "grad_norm": 0.5200527242232768,
      "learning_rate": 4.8175979626227806e-05,
      "loss": 11.8154,
      "step": 25093
    },
    {
      "epoch": 1.366468590254096,
      "grad_norm": 0.6659346802966777,
      "learning_rate": 4.816843814375367e-05,
      "loss": 12.0484,
      "step": 25094
    },
    {
      "epoch": 1.366523044250679,
      "grad_norm": 0.5748616208321881,
      "learning_rate": 4.816089706433048e-05,
      "loss": 11.8186,
      "step": 25095
    },
    {
      "epoch": 1.366577498247262,
      "grad_norm": 0.59562321821412,
      "learning_rate": 4.8153356388016827e-05,
      "loss": 11.965,
      "step": 25096
    },
    {
      "epoch": 1.366631952243845,
      "grad_norm": 0.5705926117797742,
      "learning_rate": 4.814581611487139e-05,
      "loss": 11.8715,
      "step": 25097
    },
    {
      "epoch": 1.366686406240428,
      "grad_norm": 0.5407234790532826,
      "learning_rate": 4.8138276244952754e-05,
      "loss": 11.7471,
      "step": 25098
    },
    {
      "epoch": 1.366740860237011,
      "grad_norm": 0.6501824942662773,
      "learning_rate": 4.813073677831958e-05,
      "loss": 11.9337,
      "step": 25099
    },
    {
      "epoch": 1.366795314233594,
      "grad_norm": 0.5784681159521441,
      "learning_rate": 4.812319771503053e-05,
      "loss": 11.9465,
      "step": 25100
    },
    {
      "epoch": 1.366849768230177,
      "grad_norm": 0.6056400943832342,
      "learning_rate": 4.811565905514416e-05,
      "loss": 12.0192,
      "step": 25101
    },
    {
      "epoch": 1.36690422222676,
      "grad_norm": 0.5396109459761824,
      "learning_rate": 4.810812079871916e-05,
      "loss": 11.9325,
      "step": 25102
    },
    {
      "epoch": 1.366958676223343,
      "grad_norm": 0.5546552113436038,
      "learning_rate": 4.810058294581412e-05,
      "loss": 11.9535,
      "step": 25103
    },
    {
      "epoch": 1.367013130219926,
      "grad_norm": 0.5599864368610015,
      "learning_rate": 4.809304549648761e-05,
      "loss": 11.9501,
      "step": 25104
    },
    {
      "epoch": 1.367067584216509,
      "grad_norm": 0.49451505886383473,
      "learning_rate": 4.808550845079832e-05,
      "loss": 11.9211,
      "step": 25105
    },
    {
      "epoch": 1.367122038213092,
      "grad_norm": 0.55109309273116,
      "learning_rate": 4.807797180880479e-05,
      "loss": 11.9217,
      "step": 25106
    },
    {
      "epoch": 1.3671764922096752,
      "grad_norm": 0.537961764596016,
      "learning_rate": 4.8070435570565685e-05,
      "loss": 11.798,
      "step": 25107
    },
    {
      "epoch": 1.3672309462062582,
      "grad_norm": 0.541574789119973,
      "learning_rate": 4.806289973613956e-05,
      "loss": 11.7621,
      "step": 25108
    },
    {
      "epoch": 1.3672854002028412,
      "grad_norm": 0.5324905199750999,
      "learning_rate": 4.8055364305585014e-05,
      "loss": 11.9363,
      "step": 25109
    },
    {
      "epoch": 1.3673398541994242,
      "grad_norm": 0.5219113301890548,
      "learning_rate": 4.804782927896072e-05,
      "loss": 11.8879,
      "step": 25110
    },
    {
      "epoch": 1.3673943081960072,
      "grad_norm": 0.5669892893288252,
      "learning_rate": 4.804029465632518e-05,
      "loss": 11.9929,
      "step": 25111
    },
    {
      "epoch": 1.3674487621925902,
      "grad_norm": 0.5728257842324002,
      "learning_rate": 4.8032760437737055e-05,
      "loss": 11.8895,
      "step": 25112
    },
    {
      "epoch": 1.3675032161891731,
      "grad_norm": 0.5028461713954278,
      "learning_rate": 4.8025226623254906e-05,
      "loss": 11.8547,
      "step": 25113
    },
    {
      "epoch": 1.3675576701857561,
      "grad_norm": 0.51435943829678,
      "learning_rate": 4.8017693212937266e-05,
      "loss": 11.8178,
      "step": 25114
    },
    {
      "epoch": 1.3676121241823391,
      "grad_norm": 0.5929600916261433,
      "learning_rate": 4.8010160206842803e-05,
      "loss": 11.7993,
      "step": 25115
    },
    {
      "epoch": 1.3676665781789223,
      "grad_norm": 0.5382834120972866,
      "learning_rate": 4.800262760503002e-05,
      "loss": 11.9301,
      "step": 25116
    },
    {
      "epoch": 1.3677210321755053,
      "grad_norm": 0.49008064164726056,
      "learning_rate": 4.799509540755754e-05,
      "loss": 11.935,
      "step": 25117
    },
    {
      "epoch": 1.3677754861720883,
      "grad_norm": 0.5619016132843284,
      "learning_rate": 4.798756361448391e-05,
      "loss": 11.8332,
      "step": 25118
    },
    {
      "epoch": 1.3678299401686713,
      "grad_norm": 0.4842402516284035,
      "learning_rate": 4.798003222586773e-05,
      "loss": 11.9025,
      "step": 25119
    },
    {
      "epoch": 1.3678843941652543,
      "grad_norm": 0.5270994832602205,
      "learning_rate": 4.797250124176751e-05,
      "loss": 11.7359,
      "step": 25120
    },
    {
      "epoch": 1.3679388481618373,
      "grad_norm": 0.5546253579220396,
      "learning_rate": 4.796497066224184e-05,
      "loss": 11.8169,
      "step": 25121
    },
    {
      "epoch": 1.3679933021584203,
      "grad_norm": 0.5895691163162081,
      "learning_rate": 4.795744048734932e-05,
      "loss": 11.8789,
      "step": 25122
    },
    {
      "epoch": 1.3680477561550033,
      "grad_norm": 0.5342975863541174,
      "learning_rate": 4.794991071714848e-05,
      "loss": 11.9294,
      "step": 25123
    },
    {
      "epoch": 1.3681022101515863,
      "grad_norm": 0.5366704933948835,
      "learning_rate": 4.794238135169783e-05,
      "loss": 11.8725,
      "step": 25124
    },
    {
      "epoch": 1.3681566641481693,
      "grad_norm": 0.5003059131680908,
      "learning_rate": 4.7934852391055985e-05,
      "loss": 11.9016,
      "step": 25125
    },
    {
      "epoch": 1.3682111181447523,
      "grad_norm": 0.5766781270961125,
      "learning_rate": 4.792732383528141e-05,
      "loss": 12.0251,
      "step": 25126
    },
    {
      "epoch": 1.3682655721413353,
      "grad_norm": 0.5609779498089513,
      "learning_rate": 4.791979568443274e-05,
      "loss": 11.8133,
      "step": 25127
    },
    {
      "epoch": 1.3683200261379183,
      "grad_norm": 0.5411245033745566,
      "learning_rate": 4.7912267938568445e-05,
      "loss": 12.0362,
      "step": 25128
    },
    {
      "epoch": 1.3683744801345012,
      "grad_norm": 0.5237389445196476,
      "learning_rate": 4.7904740597747124e-05,
      "loss": 11.9776,
      "step": 25129
    },
    {
      "epoch": 1.3684289341310845,
      "grad_norm": 0.7018697693221225,
      "learning_rate": 4.789721366202724e-05,
      "loss": 12.1431,
      "step": 25130
    },
    {
      "epoch": 1.3684833881276675,
      "grad_norm": 0.5315774511077166,
      "learning_rate": 4.7889687131467355e-05,
      "loss": 11.8995,
      "step": 25131
    },
    {
      "epoch": 1.3685378421242504,
      "grad_norm": 0.5699725123374426,
      "learning_rate": 4.7882161006126034e-05,
      "loss": 12.0593,
      "step": 25132
    },
    {
      "epoch": 1.3685922961208334,
      "grad_norm": 0.5555549938187669,
      "learning_rate": 4.7874635286061776e-05,
      "loss": 11.9752,
      "step": 25133
    },
    {
      "epoch": 1.3686467501174164,
      "grad_norm": 0.5380393157728871,
      "learning_rate": 4.78671099713331e-05,
      "loss": 11.9835,
      "step": 25134
    },
    {
      "epoch": 1.3687012041139994,
      "grad_norm": 0.5505625833372509,
      "learning_rate": 4.785958506199848e-05,
      "loss": 11.8858,
      "step": 25135
    },
    {
      "epoch": 1.3687556581105824,
      "grad_norm": 0.5378844181894424,
      "learning_rate": 4.7852060558116466e-05,
      "loss": 11.8883,
      "step": 25136
    },
    {
      "epoch": 1.3688101121071654,
      "grad_norm": 0.5239550100816478,
      "learning_rate": 4.7844536459745615e-05,
      "loss": 11.9211,
      "step": 25137
    },
    {
      "epoch": 1.3688645661037484,
      "grad_norm": 0.5158765324469426,
      "learning_rate": 4.783701276694436e-05,
      "loss": 11.774,
      "step": 25138
    },
    {
      "epoch": 1.3689190201003316,
      "grad_norm": 0.544013191229186,
      "learning_rate": 4.782948947977127e-05,
      "loss": 11.8005,
      "step": 25139
    },
    {
      "epoch": 1.3689734740969146,
      "grad_norm": 0.6405344375573496,
      "learning_rate": 4.7821966598284786e-05,
      "loss": 11.8133,
      "step": 25140
    },
    {
      "epoch": 1.3690279280934976,
      "grad_norm": 0.5154339333499073,
      "learning_rate": 4.7814444122543476e-05,
      "loss": 11.8936,
      "step": 25141
    },
    {
      "epoch": 1.3690823820900806,
      "grad_norm": 0.5362485119681301,
      "learning_rate": 4.7806922052605765e-05,
      "loss": 11.8353,
      "step": 25142
    },
    {
      "epoch": 1.3691368360866636,
      "grad_norm": 0.5109631313283256,
      "learning_rate": 4.779940038853018e-05,
      "loss": 12.0095,
      "step": 25143
    },
    {
      "epoch": 1.3691912900832466,
      "grad_norm": 0.5810936133366243,
      "learning_rate": 4.7791879130375286e-05,
      "loss": 11.9121,
      "step": 25144
    },
    {
      "epoch": 1.3692457440798296,
      "grad_norm": 0.5453383983930444,
      "learning_rate": 4.778435827819943e-05,
      "loss": 11.8733,
      "step": 25145
    },
    {
      "epoch": 1.3693001980764126,
      "grad_norm": 0.6126937632389491,
      "learning_rate": 4.777683783206118e-05,
      "loss": 11.8779,
      "step": 25146
    },
    {
      "epoch": 1.3693546520729956,
      "grad_norm": 0.5910559879078056,
      "learning_rate": 4.7769317792018974e-05,
      "loss": 11.9472,
      "step": 25147
    },
    {
      "epoch": 1.3694091060695786,
      "grad_norm": 0.5774818513328772,
      "learning_rate": 4.77617981581313e-05,
      "loss": 11.918,
      "step": 25148
    },
    {
      "epoch": 1.3694635600661615,
      "grad_norm": 0.5975863467060338,
      "learning_rate": 4.7754278930456685e-05,
      "loss": 12.0273,
      "step": 25149
    },
    {
      "epoch": 1.3695180140627445,
      "grad_norm": 0.5685265374387557,
      "learning_rate": 4.774676010905353e-05,
      "loss": 11.905,
      "step": 25150
    },
    {
      "epoch": 1.3695724680593275,
      "grad_norm": 0.5739833439041315,
      "learning_rate": 4.7739241693980366e-05,
      "loss": 11.9733,
      "step": 25151
    },
    {
      "epoch": 1.3696269220559105,
      "grad_norm": 0.5565029887800292,
      "learning_rate": 4.773172368529558e-05,
      "loss": 11.8743,
      "step": 25152
    },
    {
      "epoch": 1.3696813760524935,
      "grad_norm": 0.5897144964944023,
      "learning_rate": 4.772420608305771e-05,
      "loss": 12.0013,
      "step": 25153
    },
    {
      "epoch": 1.3697358300490767,
      "grad_norm": 0.5215098716775922,
      "learning_rate": 4.771668888732519e-05,
      "loss": 11.794,
      "step": 25154
    },
    {
      "epoch": 1.3697902840456597,
      "grad_norm": 0.5405015588673393,
      "learning_rate": 4.770917209815642e-05,
      "loss": 11.9028,
      "step": 25155
    },
    {
      "epoch": 1.3698447380422427,
      "grad_norm": 0.5862248229033459,
      "learning_rate": 4.770165571560994e-05,
      "loss": 11.9305,
      "step": 25156
    },
    {
      "epoch": 1.3698991920388257,
      "grad_norm": 0.5012201539882242,
      "learning_rate": 4.769413973974412e-05,
      "loss": 11.8819,
      "step": 25157
    },
    {
      "epoch": 1.3699536460354087,
      "grad_norm": 0.6094271616706446,
      "learning_rate": 4.768662417061743e-05,
      "loss": 12.0085,
      "step": 25158
    },
    {
      "epoch": 1.3700081000319917,
      "grad_norm": 0.5649822763892307,
      "learning_rate": 4.767910900828837e-05,
      "loss": 11.9952,
      "step": 25159
    },
    {
      "epoch": 1.3700625540285747,
      "grad_norm": 0.5411521046562873,
      "learning_rate": 4.76715942528153e-05,
      "loss": 11.9949,
      "step": 25160
    },
    {
      "epoch": 1.3701170080251577,
      "grad_norm": 0.5488463848978068,
      "learning_rate": 4.766407990425671e-05,
      "loss": 11.9352,
      "step": 25161
    },
    {
      "epoch": 1.370171462021741,
      "grad_norm": 0.5654767948971301,
      "learning_rate": 4.765656596267099e-05,
      "loss": 11.8449,
      "step": 25162
    },
    {
      "epoch": 1.3702259160183239,
      "grad_norm": 0.5506834463189413,
      "learning_rate": 4.7649052428116614e-05,
      "loss": 11.8977,
      "step": 25163
    },
    {
      "epoch": 1.3702803700149069,
      "grad_norm": 0.5271581165985557,
      "learning_rate": 4.7641539300651994e-05,
      "loss": 11.9711,
      "step": 25164
    },
    {
      "epoch": 1.3703348240114899,
      "grad_norm": 0.568354222805055,
      "learning_rate": 4.763402658033551e-05,
      "loss": 11.7845,
      "step": 25165
    },
    {
      "epoch": 1.3703892780080729,
      "grad_norm": 0.5373291371622491,
      "learning_rate": 4.7626514267225654e-05,
      "loss": 11.9826,
      "step": 25166
    },
    {
      "epoch": 1.3704437320046559,
      "grad_norm": 0.562625605928785,
      "learning_rate": 4.761900236138076e-05,
      "loss": 11.9212,
      "step": 25167
    },
    {
      "epoch": 1.3704981860012388,
      "grad_norm": 0.4849888466500203,
      "learning_rate": 4.7611490862859333e-05,
      "loss": 11.84,
      "step": 25168
    },
    {
      "epoch": 1.3705526399978218,
      "grad_norm": 0.634616885300659,
      "learning_rate": 4.76039797717197e-05,
      "loss": 11.8696,
      "step": 25169
    },
    {
      "epoch": 1.3706070939944048,
      "grad_norm": 0.654571175472763,
      "learning_rate": 4.75964690880203e-05,
      "loss": 11.9599,
      "step": 25170
    },
    {
      "epoch": 1.3706615479909878,
      "grad_norm": 0.5840094561144437,
      "learning_rate": 4.758895881181959e-05,
      "loss": 11.9393,
      "step": 25171
    },
    {
      "epoch": 1.3707160019875708,
      "grad_norm": 0.5412491234691605,
      "learning_rate": 4.7581448943175886e-05,
      "loss": 12.0059,
      "step": 25172
    },
    {
      "epoch": 1.3707704559841538,
      "grad_norm": 0.5806772752957869,
      "learning_rate": 4.757393948214767e-05,
      "loss": 11.7347,
      "step": 25173
    },
    {
      "epoch": 1.3708249099807368,
      "grad_norm": 0.5187746714760103,
      "learning_rate": 4.756643042879329e-05,
      "loss": 11.8312,
      "step": 25174
    },
    {
      "epoch": 1.3708793639773198,
      "grad_norm": 0.526855635336909,
      "learning_rate": 4.7558921783171095e-05,
      "loss": 11.8599,
      "step": 25175
    },
    {
      "epoch": 1.3709338179739028,
      "grad_norm": 0.5579671947534771,
      "learning_rate": 4.755141354533956e-05,
      "loss": 11.8686,
      "step": 25176
    },
    {
      "epoch": 1.370988271970486,
      "grad_norm": 0.5205755938172167,
      "learning_rate": 4.7543905715356984e-05,
      "loss": 11.8199,
      "step": 25177
    },
    {
      "epoch": 1.371042725967069,
      "grad_norm": 0.5795151769743531,
      "learning_rate": 4.753639829328185e-05,
      "loss": 11.8959,
      "step": 25178
    },
    {
      "epoch": 1.371097179963652,
      "grad_norm": 0.5063909628788978,
      "learning_rate": 4.7528891279172424e-05,
      "loss": 11.8671,
      "step": 25179
    },
    {
      "epoch": 1.371151633960235,
      "grad_norm": 0.5889294962063523,
      "learning_rate": 4.752138467308714e-05,
      "loss": 11.9673,
      "step": 25180
    },
    {
      "epoch": 1.371206087956818,
      "grad_norm": 0.49894799623537534,
      "learning_rate": 4.7513878475084414e-05,
      "loss": 11.681,
      "step": 25181
    },
    {
      "epoch": 1.371260541953401,
      "grad_norm": 0.5264034405288752,
      "learning_rate": 4.750637268522254e-05,
      "loss": 11.7979,
      "step": 25182
    },
    {
      "epoch": 1.371314995949984,
      "grad_norm": 0.5741823667048461,
      "learning_rate": 4.749886730355995e-05,
      "loss": 11.9187,
      "step": 25183
    },
    {
      "epoch": 1.371369449946567,
      "grad_norm": 0.5478759188318844,
      "learning_rate": 4.7491362330154964e-05,
      "loss": 11.9207,
      "step": 25184
    },
    {
      "epoch": 1.37142390394315,
      "grad_norm": 0.5566758328287349,
      "learning_rate": 4.748385776506592e-05,
      "loss": 11.8836,
      "step": 25185
    },
    {
      "epoch": 1.3714783579397332,
      "grad_norm": 0.5954818195244415,
      "learning_rate": 4.747635360835124e-05,
      "loss": 11.9352,
      "step": 25186
    },
    {
      "epoch": 1.3715328119363162,
      "grad_norm": 0.5044565346058839,
      "learning_rate": 4.74688498600692e-05,
      "loss": 11.8805,
      "step": 25187
    },
    {
      "epoch": 1.3715872659328991,
      "grad_norm": 0.5042383277607629,
      "learning_rate": 4.7461346520278236e-05,
      "loss": 11.8401,
      "step": 25188
    },
    {
      "epoch": 1.3716417199294821,
      "grad_norm": 0.4840766664645052,
      "learning_rate": 4.745384358903662e-05,
      "loss": 11.9215,
      "step": 25189
    },
    {
      "epoch": 1.3716961739260651,
      "grad_norm": 0.5489749929948173,
      "learning_rate": 4.7446341066402754e-05,
      "loss": 11.908,
      "step": 25190
    },
    {
      "epoch": 1.3717506279226481,
      "grad_norm": 0.526247010486432,
      "learning_rate": 4.7438838952434916e-05,
      "loss": 11.8905,
      "step": 25191
    },
    {
      "epoch": 1.3718050819192311,
      "grad_norm": 0.6014050591841199,
      "learning_rate": 4.743133724719149e-05,
      "loss": 12.0415,
      "step": 25192
    },
    {
      "epoch": 1.371859535915814,
      "grad_norm": 0.5231203855374652,
      "learning_rate": 4.7423835950730834e-05,
      "loss": 11.8379,
      "step": 25193
    },
    {
      "epoch": 1.371913989912397,
      "grad_norm": 0.5238528836274371,
      "learning_rate": 4.741633506311125e-05,
      "loss": 12.0402,
      "step": 25194
    },
    {
      "epoch": 1.37196844390898,
      "grad_norm": 0.4942266262421094,
      "learning_rate": 4.740883458439106e-05,
      "loss": 11.8109,
      "step": 25195
    },
    {
      "epoch": 1.372022897905563,
      "grad_norm": 0.5534674871706498,
      "learning_rate": 4.740133451462855e-05,
      "loss": 11.9346,
      "step": 25196
    },
    {
      "epoch": 1.372077351902146,
      "grad_norm": 0.5036089615406288,
      "learning_rate": 4.739383485388209e-05,
      "loss": 11.8729,
      "step": 25197
    },
    {
      "epoch": 1.372131805898729,
      "grad_norm": 0.5676058382743236,
      "learning_rate": 4.738633560221003e-05,
      "loss": 11.9334,
      "step": 25198
    },
    {
      "epoch": 1.372186259895312,
      "grad_norm": 0.5308956539066537,
      "learning_rate": 4.7378836759670606e-05,
      "loss": 11.9188,
      "step": 25199
    },
    {
      "epoch": 1.3722407138918953,
      "grad_norm": 0.5227422629580815,
      "learning_rate": 4.737133832632221e-05,
      "loss": 11.892,
      "step": 25200
    },
    {
      "epoch": 1.3722951678884783,
      "grad_norm": 0.5778140275861255,
      "learning_rate": 4.736384030222308e-05,
      "loss": 11.9294,
      "step": 25201
    },
    {
      "epoch": 1.3723496218850613,
      "grad_norm": 0.600801053051945,
      "learning_rate": 4.7356342687431585e-05,
      "loss": 11.8202,
      "step": 25202
    },
    {
      "epoch": 1.3724040758816443,
      "grad_norm": 0.6657469160372627,
      "learning_rate": 4.734884548200597e-05,
      "loss": 12.0397,
      "step": 25203
    },
    {
      "epoch": 1.3724585298782273,
      "grad_norm": 0.5863296049708091,
      "learning_rate": 4.7341348686004596e-05,
      "loss": 11.8541,
      "step": 25204
    },
    {
      "epoch": 1.3725129838748102,
      "grad_norm": 0.571748109594584,
      "learning_rate": 4.733385229948572e-05,
      "loss": 11.8864,
      "step": 25205
    },
    {
      "epoch": 1.3725674378713932,
      "grad_norm": 0.5720556581573786,
      "learning_rate": 4.7326356322507606e-05,
      "loss": 11.9563,
      "step": 25206
    },
    {
      "epoch": 1.3726218918679762,
      "grad_norm": 0.6269365995903576,
      "learning_rate": 4.731886075512858e-05,
      "loss": 11.8724,
      "step": 25207
    },
    {
      "epoch": 1.3726763458645592,
      "grad_norm": 0.5166449411134307,
      "learning_rate": 4.7311365597406964e-05,
      "loss": 11.7918,
      "step": 25208
    },
    {
      "epoch": 1.3727307998611424,
      "grad_norm": 0.5363266455847857,
      "learning_rate": 4.7303870849400964e-05,
      "loss": 11.8909,
      "step": 25209
    },
    {
      "epoch": 1.3727852538577254,
      "grad_norm": 0.5368339221215567,
      "learning_rate": 4.729637651116895e-05,
      "loss": 11.9264,
      "step": 25210
    },
    {
      "epoch": 1.3728397078543084,
      "grad_norm": 0.56968156954211,
      "learning_rate": 4.72888825827691e-05,
      "loss": 11.9953,
      "step": 25211
    },
    {
      "epoch": 1.3728941618508914,
      "grad_norm": 0.5571619729912212,
      "learning_rate": 4.728138906425978e-05,
      "loss": 11.9766,
      "step": 25212
    },
    {
      "epoch": 1.3729486158474744,
      "grad_norm": 0.5685060197223936,
      "learning_rate": 4.7273895955699185e-05,
      "loss": 11.6695,
      "step": 25213
    },
    {
      "epoch": 1.3730030698440574,
      "grad_norm": 0.5598016315734076,
      "learning_rate": 4.726640325714565e-05,
      "loss": 11.7932,
      "step": 25214
    },
    {
      "epoch": 1.3730575238406404,
      "grad_norm": 0.5451234135253817,
      "learning_rate": 4.725891096865742e-05,
      "loss": 11.8479,
      "step": 25215
    },
    {
      "epoch": 1.3731119778372234,
      "grad_norm": 0.5907122942209788,
      "learning_rate": 4.7251419090292694e-05,
      "loss": 11.9886,
      "step": 25216
    },
    {
      "epoch": 1.3731664318338064,
      "grad_norm": 0.59937953453529,
      "learning_rate": 4.724392762210982e-05,
      "loss": 11.925,
      "step": 25217
    },
    {
      "epoch": 1.3732208858303894,
      "grad_norm": 0.6053097495226107,
      "learning_rate": 4.723643656416698e-05,
      "loss": 11.8532,
      "step": 25218
    },
    {
      "epoch": 1.3732753398269724,
      "grad_norm": 0.5266506477173108,
      "learning_rate": 4.722894591652244e-05,
      "loss": 11.8468,
      "step": 25219
    },
    {
      "epoch": 1.3733297938235554,
      "grad_norm": 0.5443021348448976,
      "learning_rate": 4.722145567923452e-05,
      "loss": 11.8475,
      "step": 25220
    },
    {
      "epoch": 1.3733842478201383,
      "grad_norm": 0.5543951736863234,
      "learning_rate": 4.7213965852361364e-05,
      "loss": 11.8765,
      "step": 25221
    },
    {
      "epoch": 1.3734387018167213,
      "grad_norm": 0.6131721022895514,
      "learning_rate": 4.72064764359613e-05,
      "loss": 11.8663,
      "step": 25222
    },
    {
      "epoch": 1.3734931558133046,
      "grad_norm": 0.5172387160024928,
      "learning_rate": 4.71989874300925e-05,
      "loss": 11.9415,
      "step": 25223
    },
    {
      "epoch": 1.3735476098098875,
      "grad_norm": 0.5210427710252998,
      "learning_rate": 4.719149883481326e-05,
      "loss": 11.9812,
      "step": 25224
    },
    {
      "epoch": 1.3736020638064705,
      "grad_norm": 0.5658738284479276,
      "learning_rate": 4.718401065018179e-05,
      "loss": 11.9707,
      "step": 25225
    },
    {
      "epoch": 1.3736565178030535,
      "grad_norm": 0.6153394309268462,
      "learning_rate": 4.717652287625626e-05,
      "loss": 11.9621,
      "step": 25226
    },
    {
      "epoch": 1.3737109717996365,
      "grad_norm": 0.50846970213708,
      "learning_rate": 4.716903551309498e-05,
      "loss": 11.8379,
      "step": 25227
    },
    {
      "epoch": 1.3737654257962195,
      "grad_norm": 0.555380332133304,
      "learning_rate": 4.7161548560756116e-05,
      "loss": 11.9199,
      "step": 25228
    },
    {
      "epoch": 1.3738198797928025,
      "grad_norm": 0.5489522499044746,
      "learning_rate": 4.71540620192979e-05,
      "loss": 11.8861,
      "step": 25229
    },
    {
      "epoch": 1.3738743337893855,
      "grad_norm": 0.552731690999242,
      "learning_rate": 4.714657588877861e-05,
      "loss": 11.9407,
      "step": 25230
    },
    {
      "epoch": 1.3739287877859685,
      "grad_norm": 0.6860486499702981,
      "learning_rate": 4.713909016925637e-05,
      "loss": 12.0791,
      "step": 25231
    },
    {
      "epoch": 1.3739832417825517,
      "grad_norm": 0.5805692979565028,
      "learning_rate": 4.713160486078946e-05,
      "loss": 11.894,
      "step": 25232
    },
    {
      "epoch": 1.3740376957791347,
      "grad_norm": 0.5302705322568467,
      "learning_rate": 4.7124119963436034e-05,
      "loss": 11.8835,
      "step": 25233
    },
    {
      "epoch": 1.3740921497757177,
      "grad_norm": 0.5372320335109962,
      "learning_rate": 4.7116635477254336e-05,
      "loss": 11.9224,
      "step": 25234
    },
    {
      "epoch": 1.3741466037723007,
      "grad_norm": 0.6281229387489039,
      "learning_rate": 4.710915140230255e-05,
      "loss": 11.8337,
      "step": 25235
    },
    {
      "epoch": 1.3742010577688837,
      "grad_norm": 0.6253712785976908,
      "learning_rate": 4.710166773863885e-05,
      "loss": 11.819,
      "step": 25236
    },
    {
      "epoch": 1.3742555117654667,
      "grad_norm": 0.5015754396671326,
      "learning_rate": 4.7094184486321476e-05,
      "loss": 11.97,
      "step": 25237
    },
    {
      "epoch": 1.3743099657620497,
      "grad_norm": 0.5246357859823589,
      "learning_rate": 4.708670164540857e-05,
      "loss": 11.921,
      "step": 25238
    },
    {
      "epoch": 1.3743644197586327,
      "grad_norm": 0.5362420472950458,
      "learning_rate": 4.707921921595838e-05,
      "loss": 11.9285,
      "step": 25239
    },
    {
      "epoch": 1.3744188737552157,
      "grad_norm": 0.5114109675130338,
      "learning_rate": 4.707173719802902e-05,
      "loss": 11.9729,
      "step": 25240
    },
    {
      "epoch": 1.3744733277517986,
      "grad_norm": 0.5685769139154091,
      "learning_rate": 4.70642555916787e-05,
      "loss": 12.0068,
      "step": 25241
    },
    {
      "epoch": 1.3745277817483816,
      "grad_norm": 0.5272460377539192,
      "learning_rate": 4.705677439696565e-05,
      "loss": 11.8695,
      "step": 25242
    },
    {
      "epoch": 1.3745822357449646,
      "grad_norm": 0.5174468059830823,
      "learning_rate": 4.704929361394795e-05,
      "loss": 11.8628,
      "step": 25243
    },
    {
      "epoch": 1.3746366897415476,
      "grad_norm": 0.5454333648435192,
      "learning_rate": 4.7041813242683874e-05,
      "loss": 11.8649,
      "step": 25244
    },
    {
      "epoch": 1.3746911437381306,
      "grad_norm": 0.5325871754894,
      "learning_rate": 4.703433328323155e-05,
      "loss": 11.9209,
      "step": 25245
    },
    {
      "epoch": 1.3747455977347136,
      "grad_norm": 0.5632304771927739,
      "learning_rate": 4.702685373564907e-05,
      "loss": 11.95,
      "step": 25246
    },
    {
      "epoch": 1.3748000517312968,
      "grad_norm": 0.5144207844755943,
      "learning_rate": 4.701937459999471e-05,
      "loss": 11.9093,
      "step": 25247
    },
    {
      "epoch": 1.3748545057278798,
      "grad_norm": 0.5334950450649695,
      "learning_rate": 4.701189587632654e-05,
      "loss": 11.9731,
      "step": 25248
    },
    {
      "epoch": 1.3749089597244628,
      "grad_norm": 0.5165965965441954,
      "learning_rate": 4.7004417564702785e-05,
      "loss": 11.7983,
      "step": 25249
    },
    {
      "epoch": 1.3749634137210458,
      "grad_norm": 0.5635104825476014,
      "learning_rate": 4.699693966518154e-05,
      "loss": 11.9499,
      "step": 25250
    },
    {
      "epoch": 1.3750178677176288,
      "grad_norm": 0.5991284988665184,
      "learning_rate": 4.698946217782101e-05,
      "loss": 11.8144,
      "step": 25251
    },
    {
      "epoch": 1.3750723217142118,
      "grad_norm": 0.5224801381198569,
      "learning_rate": 4.698198510267928e-05,
      "loss": 11.6778,
      "step": 25252
    },
    {
      "epoch": 1.3751267757107948,
      "grad_norm": 0.5569192678137207,
      "learning_rate": 4.6974508439814523e-05,
      "loss": 11.8938,
      "step": 25253
    },
    {
      "epoch": 1.3751812297073778,
      "grad_norm": 0.6074234566892095,
      "learning_rate": 4.6967032189284955e-05,
      "loss": 11.9493,
      "step": 25254
    },
    {
      "epoch": 1.3752356837039608,
      "grad_norm": 0.5195477621291372,
      "learning_rate": 4.695955635114856e-05,
      "loss": 11.9062,
      "step": 25255
    },
    {
      "epoch": 1.375290137700544,
      "grad_norm": 0.5788947863090088,
      "learning_rate": 4.695208092546355e-05,
      "loss": 11.9226,
      "step": 25256
    },
    {
      "epoch": 1.375344591697127,
      "grad_norm": 0.5659844363225056,
      "learning_rate": 4.69446059122881e-05,
      "loss": 11.6769,
      "step": 25257
    },
    {
      "epoch": 1.37539904569371,
      "grad_norm": 0.5824028162796828,
      "learning_rate": 4.693713131168024e-05,
      "loss": 11.9505,
      "step": 25258
    },
    {
      "epoch": 1.375453499690293,
      "grad_norm": 0.581703679716766,
      "learning_rate": 4.69296571236982e-05,
      "loss": 11.8626,
      "step": 25259
    },
    {
      "epoch": 1.375507953686876,
      "grad_norm": 0.5528773091074491,
      "learning_rate": 4.6922183348399996e-05,
      "loss": 11.8795,
      "step": 25260
    },
    {
      "epoch": 1.375562407683459,
      "grad_norm": 0.5539793709418636,
      "learning_rate": 4.6914709985843844e-05,
      "loss": 12.0209,
      "step": 25261
    },
    {
      "epoch": 1.375616861680042,
      "grad_norm": 0.557919650522969,
      "learning_rate": 4.6907237036087756e-05,
      "loss": 11.9375,
      "step": 25262
    },
    {
      "epoch": 1.375671315676625,
      "grad_norm": 0.5279663707755494,
      "learning_rate": 4.689976449918991e-05,
      "loss": 11.8639,
      "step": 25263
    },
    {
      "epoch": 1.375725769673208,
      "grad_norm": 0.524774750805883,
      "learning_rate": 4.6892292375208467e-05,
      "loss": 11.8669,
      "step": 25264
    },
    {
      "epoch": 1.375780223669791,
      "grad_norm": 0.5125114864252607,
      "learning_rate": 4.6884820664201404e-05,
      "loss": 11.7979,
      "step": 25265
    },
    {
      "epoch": 1.375834677666374,
      "grad_norm": 0.5581420293347898,
      "learning_rate": 4.6877349366226906e-05,
      "loss": 11.7847,
      "step": 25266
    },
    {
      "epoch": 1.375889131662957,
      "grad_norm": 0.5485033093475268,
      "learning_rate": 4.686987848134301e-05,
      "loss": 11.9629,
      "step": 25267
    },
    {
      "epoch": 1.37594358565954,
      "grad_norm": 0.5842899444584857,
      "learning_rate": 4.686240800960786e-05,
      "loss": 11.8865,
      "step": 25268
    },
    {
      "epoch": 1.3759980396561229,
      "grad_norm": 0.5167122307149978,
      "learning_rate": 4.6854937951079566e-05,
      "loss": 11.9847,
      "step": 25269
    },
    {
      "epoch": 1.376052493652706,
      "grad_norm": 0.5384273414088226,
      "learning_rate": 4.6847468305816144e-05,
      "loss": 11.9743,
      "step": 25270
    },
    {
      "epoch": 1.376106947649289,
      "grad_norm": 0.537810582793952,
      "learning_rate": 4.683999907387577e-05,
      "loss": 11.9524,
      "step": 25271
    },
    {
      "epoch": 1.376161401645872,
      "grad_norm": 0.5046327608039041,
      "learning_rate": 4.683253025531644e-05,
      "loss": 11.8534,
      "step": 25272
    },
    {
      "epoch": 1.376215855642455,
      "grad_norm": 0.529295704030328,
      "learning_rate": 4.6825061850196304e-05,
      "loss": 11.8482,
      "step": 25273
    },
    {
      "epoch": 1.376270309639038,
      "grad_norm": 0.6293232892194328,
      "learning_rate": 4.68175938585734e-05,
      "loss": 12.0959,
      "step": 25274
    },
    {
      "epoch": 1.376324763635621,
      "grad_norm": 0.5299306994355352,
      "learning_rate": 4.681012628050578e-05,
      "loss": 11.8061,
      "step": 25275
    },
    {
      "epoch": 1.376379217632204,
      "grad_norm": 0.5470687913712922,
      "learning_rate": 4.680265911605157e-05,
      "loss": 11.961,
      "step": 25276
    },
    {
      "epoch": 1.376433671628787,
      "grad_norm": 0.6534864760950344,
      "learning_rate": 4.679519236526877e-05,
      "loss": 11.8893,
      "step": 25277
    },
    {
      "epoch": 1.37648812562537,
      "grad_norm": 0.5708609657460811,
      "learning_rate": 4.678772602821547e-05,
      "loss": 11.8014,
      "step": 25278
    },
    {
      "epoch": 1.3765425796219533,
      "grad_norm": 0.515604461607012,
      "learning_rate": 4.678026010494977e-05,
      "loss": 11.9886,
      "step": 25279
    },
    {
      "epoch": 1.3765970336185362,
      "grad_norm": 0.5606195302778902,
      "learning_rate": 4.6772794595529665e-05,
      "loss": 11.8297,
      "step": 25280
    },
    {
      "epoch": 1.3766514876151192,
      "grad_norm": 0.6045793687423382,
      "learning_rate": 4.676532950001327e-05,
      "loss": 11.8623,
      "step": 25281
    },
    {
      "epoch": 1.3767059416117022,
      "grad_norm": 0.5499582869822852,
      "learning_rate": 4.6757864818458565e-05,
      "loss": 11.9649,
      "step": 25282
    },
    {
      "epoch": 1.3767603956082852,
      "grad_norm": 0.586820199174203,
      "learning_rate": 4.675040055092366e-05,
      "loss": 11.8915,
      "step": 25283
    },
    {
      "epoch": 1.3768148496048682,
      "grad_norm": 0.5161861001214637,
      "learning_rate": 4.6742936697466574e-05,
      "loss": 11.8473,
      "step": 25284
    },
    {
      "epoch": 1.3768693036014512,
      "grad_norm": 0.5552569018298428,
      "learning_rate": 4.673547325814531e-05,
      "loss": 11.8344,
      "step": 25285
    },
    {
      "epoch": 1.3769237575980342,
      "grad_norm": 0.5176177753705812,
      "learning_rate": 4.672801023301797e-05,
      "loss": 11.9716,
      "step": 25286
    },
    {
      "epoch": 1.3769782115946172,
      "grad_norm": 0.530667070775309,
      "learning_rate": 4.672054762214253e-05,
      "loss": 11.8369,
      "step": 25287
    },
    {
      "epoch": 1.3770326655912002,
      "grad_norm": 0.5371237937745807,
      "learning_rate": 4.671308542557707e-05,
      "loss": 11.8416,
      "step": 25288
    },
    {
      "epoch": 1.3770871195877832,
      "grad_norm": 0.5502273005808457,
      "learning_rate": 4.670562364337957e-05,
      "loss": 11.7173,
      "step": 25289
    },
    {
      "epoch": 1.3771415735843662,
      "grad_norm": 0.5168596563174641,
      "learning_rate": 4.669816227560807e-05,
      "loss": 11.8472,
      "step": 25290
    },
    {
      "epoch": 1.3771960275809492,
      "grad_norm": 0.5011662809415851,
      "learning_rate": 4.669070132232063e-05,
      "loss": 11.9003,
      "step": 25291
    },
    {
      "epoch": 1.3772504815775322,
      "grad_norm": 0.4688201579640735,
      "learning_rate": 4.668324078357521e-05,
      "loss": 11.8583,
      "step": 25292
    },
    {
      "epoch": 1.3773049355741154,
      "grad_norm": 0.5398657184161583,
      "learning_rate": 4.667578065942989e-05,
      "loss": 11.8276,
      "step": 25293
    },
    {
      "epoch": 1.3773593895706984,
      "grad_norm": 0.5140614422749251,
      "learning_rate": 4.666832094994259e-05,
      "loss": 11.7812,
      "step": 25294
    },
    {
      "epoch": 1.3774138435672814,
      "grad_norm": 0.5605245670159629,
      "learning_rate": 4.666086165517142e-05,
      "loss": 11.984,
      "step": 25295
    },
    {
      "epoch": 1.3774682975638644,
      "grad_norm": 0.5837910327272393,
      "learning_rate": 4.665340277517434e-05,
      "loss": 11.8923,
      "step": 25296
    },
    {
      "epoch": 1.3775227515604473,
      "grad_norm": 0.5181840213998935,
      "learning_rate": 4.6645944310009295e-05,
      "loss": 11.8295,
      "step": 25297
    },
    {
      "epoch": 1.3775772055570303,
      "grad_norm": 0.5549549871853204,
      "learning_rate": 4.663848625973438e-05,
      "loss": 11.8531,
      "step": 25298
    },
    {
      "epoch": 1.3776316595536133,
      "grad_norm": 0.5703559806526367,
      "learning_rate": 4.66310286244075e-05,
      "loss": 11.8954,
      "step": 25299
    },
    {
      "epoch": 1.3776861135501963,
      "grad_norm": 0.5370731923959247,
      "learning_rate": 4.662357140408673e-05,
      "loss": 11.915,
      "step": 25300
    },
    {
      "epoch": 1.3777405675467793,
      "grad_norm": 0.544693483049611,
      "learning_rate": 4.6616114598829994e-05,
      "loss": 11.8466,
      "step": 25301
    },
    {
      "epoch": 1.3777950215433625,
      "grad_norm": 0.5423966923917084,
      "learning_rate": 4.660865820869529e-05,
      "loss": 11.8965,
      "step": 25302
    },
    {
      "epoch": 1.3778494755399455,
      "grad_norm": 0.6431719711641443,
      "learning_rate": 4.660120223374066e-05,
      "loss": 11.924,
      "step": 25303
    },
    {
      "epoch": 1.3779039295365285,
      "grad_norm": 0.5159256627432793,
      "learning_rate": 4.6593746674023994e-05,
      "loss": 11.8755,
      "step": 25304
    },
    {
      "epoch": 1.3779583835331115,
      "grad_norm": 0.5051576987891798,
      "learning_rate": 4.658629152960335e-05,
      "loss": 11.8992,
      "step": 25305
    },
    {
      "epoch": 1.3780128375296945,
      "grad_norm": 0.5555260791708165,
      "learning_rate": 4.657883680053666e-05,
      "loss": 11.9193,
      "step": 25306
    },
    {
      "epoch": 1.3780672915262775,
      "grad_norm": 0.605734829380128,
      "learning_rate": 4.657138248688185e-05,
      "loss": 11.8852,
      "step": 25307
    },
    {
      "epoch": 1.3781217455228605,
      "grad_norm": 0.5981299656884339,
      "learning_rate": 4.656392858869698e-05,
      "loss": 11.737,
      "step": 25308
    },
    {
      "epoch": 1.3781761995194435,
      "grad_norm": 0.5605431575308834,
      "learning_rate": 4.655647510603991e-05,
      "loss": 11.8532,
      "step": 25309
    },
    {
      "epoch": 1.3782306535160265,
      "grad_norm": 0.6098572930186241,
      "learning_rate": 4.6549022038968704e-05,
      "loss": 11.8812,
      "step": 25310
    },
    {
      "epoch": 1.3782851075126095,
      "grad_norm": 0.5475993808226356,
      "learning_rate": 4.654156938754122e-05,
      "loss": 11.8025,
      "step": 25311
    },
    {
      "epoch": 1.3783395615091925,
      "grad_norm": 0.6035442108574312,
      "learning_rate": 4.653411715181546e-05,
      "loss": 11.8948,
      "step": 25312
    },
    {
      "epoch": 1.3783940155057754,
      "grad_norm": 0.6582446198534526,
      "learning_rate": 4.65266653318494e-05,
      "loss": 11.9297,
      "step": 25313
    },
    {
      "epoch": 1.3784484695023584,
      "grad_norm": 0.5647195182100125,
      "learning_rate": 4.651921392770093e-05,
      "loss": 11.7704,
      "step": 25314
    },
    {
      "epoch": 1.3785029234989414,
      "grad_norm": 0.5096170369971617,
      "learning_rate": 4.651176293942811e-05,
      "loss": 11.884,
      "step": 25315
    },
    {
      "epoch": 1.3785573774955244,
      "grad_norm": 0.5158635753513805,
      "learning_rate": 4.65043123670887e-05,
      "loss": 11.9498,
      "step": 25316
    },
    {
      "epoch": 1.3786118314921076,
      "grad_norm": 0.5335481980621415,
      "learning_rate": 4.649686221074072e-05,
      "loss": 11.9132,
      "step": 25317
    },
    {
      "epoch": 1.3786662854886906,
      "grad_norm": 0.6399638169629862,
      "learning_rate": 4.648941247044216e-05,
      "loss": 11.9258,
      "step": 25318
    },
    {
      "epoch": 1.3787207394852736,
      "grad_norm": 0.6079666643576195,
      "learning_rate": 4.648196314625086e-05,
      "loss": 11.9187,
      "step": 25319
    },
    {
      "epoch": 1.3787751934818566,
      "grad_norm": 0.5271068845853558,
      "learning_rate": 4.647451423822484e-05,
      "loss": 11.8874,
      "step": 25320
    },
    {
      "epoch": 1.3788296474784396,
      "grad_norm": 0.5557132555366495,
      "learning_rate": 4.6467065746421925e-05,
      "loss": 11.8037,
      "step": 25321
    },
    {
      "epoch": 1.3788841014750226,
      "grad_norm": 0.6228311442435652,
      "learning_rate": 4.645961767090012e-05,
      "loss": 12.0277,
      "step": 25322
    },
    {
      "epoch": 1.3789385554716056,
      "grad_norm": 0.5514131563069321,
      "learning_rate": 4.645217001171728e-05,
      "loss": 11.9728,
      "step": 25323
    },
    {
      "epoch": 1.3789930094681886,
      "grad_norm": 0.49910688224173033,
      "learning_rate": 4.644472276893134e-05,
      "loss": 11.8547,
      "step": 25324
    },
    {
      "epoch": 1.3790474634647718,
      "grad_norm": 0.5045112355743566,
      "learning_rate": 4.643727594260029e-05,
      "loss": 11.9359,
      "step": 25325
    },
    {
      "epoch": 1.3791019174613548,
      "grad_norm": 0.5641923857059223,
      "learning_rate": 4.642982953278189e-05,
      "loss": 12.0442,
      "step": 25326
    },
    {
      "epoch": 1.3791563714579378,
      "grad_norm": 0.564463834160809,
      "learning_rate": 4.642238353953412e-05,
      "loss": 11.8735,
      "step": 25327
    },
    {
      "epoch": 1.3792108254545208,
      "grad_norm": 0.570900474626188,
      "learning_rate": 4.641493796291492e-05,
      "loss": 11.9399,
      "step": 25328
    },
    {
      "epoch": 1.3792652794511038,
      "grad_norm": 0.5315766014289637,
      "learning_rate": 4.64074928029821e-05,
      "loss": 11.8108,
      "step": 25329
    },
    {
      "epoch": 1.3793197334476868,
      "grad_norm": 0.5351653281513066,
      "learning_rate": 4.6400048059793656e-05,
      "loss": 11.9468,
      "step": 25330
    },
    {
      "epoch": 1.3793741874442698,
      "grad_norm": 0.5941142858822028,
      "learning_rate": 4.639260373340738e-05,
      "loss": 11.8205,
      "step": 25331
    },
    {
      "epoch": 1.3794286414408528,
      "grad_norm": 0.5827751609890155,
      "learning_rate": 4.638515982388125e-05,
      "loss": 11.8947,
      "step": 25332
    },
    {
      "epoch": 1.3794830954374357,
      "grad_norm": 0.5521555797331066,
      "learning_rate": 4.6377716331273066e-05,
      "loss": 11.909,
      "step": 25333
    },
    {
      "epoch": 1.3795375494340187,
      "grad_norm": 0.5263839897161878,
      "learning_rate": 4.637027325564076e-05,
      "loss": 11.8809,
      "step": 25334
    },
    {
      "epoch": 1.3795920034306017,
      "grad_norm": 0.5406684908264028,
      "learning_rate": 4.636283059704227e-05,
      "loss": 11.9414,
      "step": 25335
    },
    {
      "epoch": 1.3796464574271847,
      "grad_norm": 0.5338849962676493,
      "learning_rate": 4.635538835553533e-05,
      "loss": 11.8371,
      "step": 25336
    },
    {
      "epoch": 1.3797009114237677,
      "grad_norm": 0.5699664733897476,
      "learning_rate": 4.6347946531177935e-05,
      "loss": 11.9682,
      "step": 25337
    },
    {
      "epoch": 1.3797553654203507,
      "grad_norm": 0.5638315708518855,
      "learning_rate": 4.634050512402786e-05,
      "loss": 11.8624,
      "step": 25338
    },
    {
      "epoch": 1.3798098194169337,
      "grad_norm": 0.502264483104613,
      "learning_rate": 4.633306413414301e-05,
      "loss": 11.8048,
      "step": 25339
    },
    {
      "epoch": 1.379864273413517,
      "grad_norm": 0.5765177970339285,
      "learning_rate": 4.63256235615813e-05,
      "loss": 11.8838,
      "step": 25340
    },
    {
      "epoch": 1.3799187274101,
      "grad_norm": 0.5291928412229079,
      "learning_rate": 4.631818340640049e-05,
      "loss": 11.7628,
      "step": 25341
    },
    {
      "epoch": 1.379973181406683,
      "grad_norm": 0.5935222709828194,
      "learning_rate": 4.631074366865855e-05,
      "loss": 11.9262,
      "step": 25342
    },
    {
      "epoch": 1.380027635403266,
      "grad_norm": 0.5485942271044094,
      "learning_rate": 4.630330434841321e-05,
      "loss": 11.8891,
      "step": 25343
    },
    {
      "epoch": 1.3800820893998489,
      "grad_norm": 0.6240681136806207,
      "learning_rate": 4.629586544572243e-05,
      "loss": 12.0713,
      "step": 25344
    },
    {
      "epoch": 1.3801365433964319,
      "grad_norm": 0.5251818277570874,
      "learning_rate": 4.6288426960644006e-05,
      "loss": 11.9282,
      "step": 25345
    },
    {
      "epoch": 1.3801909973930149,
      "grad_norm": 0.6202449369655912,
      "learning_rate": 4.628098889323574e-05,
      "loss": 12.031,
      "step": 25346
    },
    {
      "epoch": 1.3802454513895979,
      "grad_norm": 0.5865949498063342,
      "learning_rate": 4.627355124355556e-05,
      "loss": 11.914,
      "step": 25347
    },
    {
      "epoch": 1.3802999053861809,
      "grad_norm": 0.6780010259118175,
      "learning_rate": 4.626611401166121e-05,
      "loss": 11.8229,
      "step": 25348
    },
    {
      "epoch": 1.380354359382764,
      "grad_norm": 0.5327655092422529,
      "learning_rate": 4.62586771976106e-05,
      "loss": 11.9316,
      "step": 25349
    },
    {
      "epoch": 1.380408813379347,
      "grad_norm": 0.6142761650187462,
      "learning_rate": 4.62512408014615e-05,
      "loss": 11.9944,
      "step": 25350
    },
    {
      "epoch": 1.38046326737593,
      "grad_norm": 0.5161875552161428,
      "learning_rate": 4.6243804823271766e-05,
      "loss": 11.8983,
      "step": 25351
    },
    {
      "epoch": 1.380517721372513,
      "grad_norm": 0.5663489760836916,
      "learning_rate": 4.6236369263099254e-05,
      "loss": 11.8626,
      "step": 25352
    },
    {
      "epoch": 1.380572175369096,
      "grad_norm": 0.5621841637946295,
      "learning_rate": 4.622893412100171e-05,
      "loss": 12.046,
      "step": 25353
    },
    {
      "epoch": 1.380626629365679,
      "grad_norm": 0.5500645122406784,
      "learning_rate": 4.622149939703704e-05,
      "loss": 11.7661,
      "step": 25354
    },
    {
      "epoch": 1.380681083362262,
      "grad_norm": 0.5699335645189085,
      "learning_rate": 4.6214065091263e-05,
      "loss": 11.8789,
      "step": 25355
    },
    {
      "epoch": 1.380735537358845,
      "grad_norm": 0.5879388378839406,
      "learning_rate": 4.620663120373738e-05,
      "loss": 11.9825,
      "step": 25356
    },
    {
      "epoch": 1.380789991355428,
      "grad_norm": 0.6442632725935946,
      "learning_rate": 4.619919773451805e-05,
      "loss": 11.9544,
      "step": 25357
    },
    {
      "epoch": 1.380844445352011,
      "grad_norm": 0.5388417844431643,
      "learning_rate": 4.6191764683662744e-05,
      "loss": 11.9933,
      "step": 25358
    },
    {
      "epoch": 1.380898899348594,
      "grad_norm": 0.5419684114350882,
      "learning_rate": 4.618433205122933e-05,
      "loss": 11.987,
      "step": 25359
    },
    {
      "epoch": 1.380953353345177,
      "grad_norm": 0.5144860434373913,
      "learning_rate": 4.617689983727555e-05,
      "loss": 11.9195,
      "step": 25360
    },
    {
      "epoch": 1.38100780734176,
      "grad_norm": 0.5200277033184444,
      "learning_rate": 4.616946804185921e-05,
      "loss": 11.9442,
      "step": 25361
    },
    {
      "epoch": 1.381062261338343,
      "grad_norm": 0.5905218274944044,
      "learning_rate": 4.6162036665038155e-05,
      "loss": 11.8349,
      "step": 25362
    },
    {
      "epoch": 1.3811167153349262,
      "grad_norm": 0.5034152148165404,
      "learning_rate": 4.61546057068701e-05,
      "loss": 11.8981,
      "step": 25363
    },
    {
      "epoch": 1.3811711693315092,
      "grad_norm": 0.5545798446381608,
      "learning_rate": 4.614717516741289e-05,
      "loss": 11.9117,
      "step": 25364
    },
    {
      "epoch": 1.3812256233280922,
      "grad_norm": 0.5808052724832219,
      "learning_rate": 4.6139745046724294e-05,
      "loss": 11.8981,
      "step": 25365
    },
    {
      "epoch": 1.3812800773246752,
      "grad_norm": 0.5312349226885734,
      "learning_rate": 4.613231534486202e-05,
      "loss": 11.8773,
      "step": 25366
    },
    {
      "epoch": 1.3813345313212582,
      "grad_norm": 0.5482902407992518,
      "learning_rate": 4.6124886061883934e-05,
      "loss": 11.963,
      "step": 25367
    },
    {
      "epoch": 1.3813889853178412,
      "grad_norm": 0.5904775818451609,
      "learning_rate": 4.6117457197847736e-05,
      "loss": 11.9922,
      "step": 25368
    },
    {
      "epoch": 1.3814434393144241,
      "grad_norm": 0.5438179334951186,
      "learning_rate": 4.6110028752811266e-05,
      "loss": 11.934,
      "step": 25369
    },
    {
      "epoch": 1.3814978933110071,
      "grad_norm": 0.5615586565282236,
      "learning_rate": 4.6102600726832204e-05,
      "loss": 11.8404,
      "step": 25370
    },
    {
      "epoch": 1.3815523473075901,
      "grad_norm": 0.5757624633721171,
      "learning_rate": 4.609517311996839e-05,
      "loss": 11.9536,
      "step": 25371
    },
    {
      "epoch": 1.3816068013041733,
      "grad_norm": 0.5315687733262756,
      "learning_rate": 4.608774593227753e-05,
      "loss": 11.7931,
      "step": 25372
    },
    {
      "epoch": 1.3816612553007563,
      "grad_norm": 0.510926070633032,
      "learning_rate": 4.608031916381739e-05,
      "loss": 11.7279,
      "step": 25373
    },
    {
      "epoch": 1.3817157092973393,
      "grad_norm": 0.5413068769685286,
      "learning_rate": 4.607289281464578e-05,
      "loss": 11.9082,
      "step": 25374
    },
    {
      "epoch": 1.3817701632939223,
      "grad_norm": 0.5720049337632385,
      "learning_rate": 4.6065466884820376e-05,
      "loss": 11.8959,
      "step": 25375
    },
    {
      "epoch": 1.3818246172905053,
      "grad_norm": 0.5415679879386224,
      "learning_rate": 4.605804137439892e-05,
      "loss": 11.702,
      "step": 25376
    },
    {
      "epoch": 1.3818790712870883,
      "grad_norm": 0.519743435854354,
      "learning_rate": 4.605061628343922e-05,
      "loss": 11.9134,
      "step": 25377
    },
    {
      "epoch": 1.3819335252836713,
      "grad_norm": 0.5362190930944459,
      "learning_rate": 4.604319161199894e-05,
      "loss": 11.9347,
      "step": 25378
    },
    {
      "epoch": 1.3819879792802543,
      "grad_norm": 0.5978464665510768,
      "learning_rate": 4.603576736013587e-05,
      "loss": 11.8299,
      "step": 25379
    },
    {
      "epoch": 1.3820424332768373,
      "grad_norm": 0.601051854487191,
      "learning_rate": 4.60283435279077e-05,
      "loss": 11.8637,
      "step": 25380
    },
    {
      "epoch": 1.3820968872734203,
      "grad_norm": 0.5364435772542022,
      "learning_rate": 4.602092011537222e-05,
      "loss": 11.9527,
      "step": 25381
    },
    {
      "epoch": 1.3821513412700033,
      "grad_norm": 0.5385852669288471,
      "learning_rate": 4.601349712258708e-05,
      "loss": 11.8286,
      "step": 25382
    },
    {
      "epoch": 1.3822057952665863,
      "grad_norm": 0.5261441609401453,
      "learning_rate": 4.600607454961004e-05,
      "loss": 11.8859,
      "step": 25383
    },
    {
      "epoch": 1.3822602492631693,
      "grad_norm": 0.5543163456264253,
      "learning_rate": 4.599865239649885e-05,
      "loss": 11.8436,
      "step": 25384
    },
    {
      "epoch": 1.3823147032597523,
      "grad_norm": 0.5696395854086732,
      "learning_rate": 4.59912306633112e-05,
      "loss": 11.8876,
      "step": 25385
    },
    {
      "epoch": 1.3823691572563352,
      "grad_norm": 0.5372364461671898,
      "learning_rate": 4.59838093501048e-05,
      "loss": 11.7971,
      "step": 25386
    },
    {
      "epoch": 1.3824236112529185,
      "grad_norm": 0.5336524515688087,
      "learning_rate": 4.597638845693733e-05,
      "loss": 11.9275,
      "step": 25387
    },
    {
      "epoch": 1.3824780652495015,
      "grad_norm": 0.5475560823657954,
      "learning_rate": 4.5968967983866495e-05,
      "loss": 11.8402,
      "step": 25388
    },
    {
      "epoch": 1.3825325192460844,
      "grad_norm": 0.5369080152016062,
      "learning_rate": 4.5961547930950086e-05,
      "loss": 11.911,
      "step": 25389
    },
    {
      "epoch": 1.3825869732426674,
      "grad_norm": 0.5314198399045027,
      "learning_rate": 4.595412829824569e-05,
      "loss": 11.819,
      "step": 25390
    },
    {
      "epoch": 1.3826414272392504,
      "grad_norm": 0.5615156119415065,
      "learning_rate": 4.59467090858111e-05,
      "loss": 11.8952,
      "step": 25391
    },
    {
      "epoch": 1.3826958812358334,
      "grad_norm": 0.49312984592688724,
      "learning_rate": 4.5939290293703926e-05,
      "loss": 11.7849,
      "step": 25392
    },
    {
      "epoch": 1.3827503352324164,
      "grad_norm": 0.5322878051254277,
      "learning_rate": 4.593187192198195e-05,
      "loss": 11.9094,
      "step": 25393
    },
    {
      "epoch": 1.3828047892289994,
      "grad_norm": 0.5838399952260754,
      "learning_rate": 4.5924453970702755e-05,
      "loss": 11.9369,
      "step": 25394
    },
    {
      "epoch": 1.3828592432255826,
      "grad_norm": 0.5053029548424226,
      "learning_rate": 4.591703643992411e-05,
      "loss": 11.7684,
      "step": 25395
    },
    {
      "epoch": 1.3829136972221656,
      "grad_norm": 0.5615231519740667,
      "learning_rate": 4.5909619329703655e-05,
      "loss": 11.9489,
      "step": 25396
    },
    {
      "epoch": 1.3829681512187486,
      "grad_norm": 0.5421039509312398,
      "learning_rate": 4.590220264009903e-05,
      "loss": 11.9166,
      "step": 25397
    },
    {
      "epoch": 1.3830226052153316,
      "grad_norm": 0.5251245200652788,
      "learning_rate": 4.589478637116801e-05,
      "loss": 11.7721,
      "step": 25398
    },
    {
      "epoch": 1.3830770592119146,
      "grad_norm": 0.518286402281066,
      "learning_rate": 4.588737052296815e-05,
      "loss": 11.8004,
      "step": 25399
    },
    {
      "epoch": 1.3831315132084976,
      "grad_norm": 0.5281743691607864,
      "learning_rate": 4.587995509555717e-05,
      "loss": 11.8995,
      "step": 25400
    },
    {
      "epoch": 1.3831859672050806,
      "grad_norm": 0.5350429109521715,
      "learning_rate": 4.587254008899278e-05,
      "loss": 11.805,
      "step": 25401
    },
    {
      "epoch": 1.3832404212016636,
      "grad_norm": 0.5528001096401002,
      "learning_rate": 4.586512550333255e-05,
      "loss": 11.9221,
      "step": 25402
    },
    {
      "epoch": 1.3832948751982466,
      "grad_norm": 0.7356239817084284,
      "learning_rate": 4.5857711338634235e-05,
      "loss": 11.975,
      "step": 25403
    },
    {
      "epoch": 1.3833493291948296,
      "grad_norm": 0.5268355168398967,
      "learning_rate": 4.585029759495538e-05,
      "loss": 11.951,
      "step": 25404
    },
    {
      "epoch": 1.3834037831914126,
      "grad_norm": 0.5754927109925986,
      "learning_rate": 4.5842884272353745e-05,
      "loss": 12.0442,
      "step": 25405
    },
    {
      "epoch": 1.3834582371879955,
      "grad_norm": 0.5323192851534655,
      "learning_rate": 4.583547137088692e-05,
      "loss": 11.9776,
      "step": 25406
    },
    {
      "epoch": 1.3835126911845785,
      "grad_norm": 0.6187177938805167,
      "learning_rate": 4.5828058890612516e-05,
      "loss": 11.9483,
      "step": 25407
    },
    {
      "epoch": 1.3835671451811615,
      "grad_norm": 0.5396065121654304,
      "learning_rate": 4.582064683158823e-05,
      "loss": 11.8958,
      "step": 25408
    },
    {
      "epoch": 1.3836215991777445,
      "grad_norm": 0.5310752884032458,
      "learning_rate": 4.5813235193871665e-05,
      "loss": 11.8103,
      "step": 25409
    },
    {
      "epoch": 1.3836760531743277,
      "grad_norm": 0.5781549138458341,
      "learning_rate": 4.580582397752046e-05,
      "loss": 12.0116,
      "step": 25410
    },
    {
      "epoch": 1.3837305071709107,
      "grad_norm": 0.5857054738959128,
      "learning_rate": 4.5798413182592305e-05,
      "loss": 11.7495,
      "step": 25411
    },
    {
      "epoch": 1.3837849611674937,
      "grad_norm": 0.5771792869696822,
      "learning_rate": 4.5791002809144724e-05,
      "loss": 11.9045,
      "step": 25412
    },
    {
      "epoch": 1.3838394151640767,
      "grad_norm": 0.5173491338387444,
      "learning_rate": 4.5783592857235444e-05,
      "loss": 11.8329,
      "step": 25413
    },
    {
      "epoch": 1.3838938691606597,
      "grad_norm": 0.6159445884622708,
      "learning_rate": 4.577618332692199e-05,
      "loss": 12.0177,
      "step": 25414
    },
    {
      "epoch": 1.3839483231572427,
      "grad_norm": 0.6027146149136118,
      "learning_rate": 4.576877421826208e-05,
      "loss": 11.9161,
      "step": 25415
    },
    {
      "epoch": 1.3840027771538257,
      "grad_norm": 0.5776180683335476,
      "learning_rate": 4.576136553131327e-05,
      "loss": 11.8918,
      "step": 25416
    },
    {
      "epoch": 1.3840572311504087,
      "grad_norm": 0.5812378036134579,
      "learning_rate": 4.575395726613314e-05,
      "loss": 11.9817,
      "step": 25417
    },
    {
      "epoch": 1.3841116851469917,
      "grad_norm": 0.5221685414790718,
      "learning_rate": 4.574654942277937e-05,
      "loss": 11.901,
      "step": 25418
    },
    {
      "epoch": 1.384166139143575,
      "grad_norm": 0.5835118401881975,
      "learning_rate": 4.57391420013095e-05,
      "loss": 11.9185,
      "step": 25419
    },
    {
      "epoch": 1.3842205931401579,
      "grad_norm": 0.5535971253639945,
      "learning_rate": 4.573173500178119e-05,
      "loss": 11.8291,
      "step": 25420
    },
    {
      "epoch": 1.3842750471367409,
      "grad_norm": 0.5632179748324755,
      "learning_rate": 4.5724328424251985e-05,
      "loss": 11.932,
      "step": 25421
    },
    {
      "epoch": 1.3843295011333239,
      "grad_norm": 0.565666884742778,
      "learning_rate": 4.5716922268779495e-05,
      "loss": 11.8911,
      "step": 25422
    },
    {
      "epoch": 1.3843839551299069,
      "grad_norm": 0.567112815335102,
      "learning_rate": 4.570951653542136e-05,
      "loss": 12.0588,
      "step": 25423
    },
    {
      "epoch": 1.3844384091264899,
      "grad_norm": 0.5168100326162838,
      "learning_rate": 4.570211122423509e-05,
      "loss": 11.8124,
      "step": 25424
    },
    {
      "epoch": 1.3844928631230728,
      "grad_norm": 0.4863290188853976,
      "learning_rate": 4.5694706335278346e-05,
      "loss": 11.8577,
      "step": 25425
    },
    {
      "epoch": 1.3845473171196558,
      "grad_norm": 0.6330527557283268,
      "learning_rate": 4.568730186860867e-05,
      "loss": 12.1193,
      "step": 25426
    },
    {
      "epoch": 1.3846017711162388,
      "grad_norm": 0.5254447852363948,
      "learning_rate": 4.5679897824283615e-05,
      "loss": 11.9341,
      "step": 25427
    },
    {
      "epoch": 1.3846562251128218,
      "grad_norm": 0.6192901503995081,
      "learning_rate": 4.56724942023608e-05,
      "loss": 11.846,
      "step": 25428
    },
    {
      "epoch": 1.3847106791094048,
      "grad_norm": 0.5601019083569937,
      "learning_rate": 4.566509100289777e-05,
      "loss": 12.0017,
      "step": 25429
    },
    {
      "epoch": 1.3847651331059878,
      "grad_norm": 0.6557075919599537,
      "learning_rate": 4.565768822595213e-05,
      "loss": 11.9712,
      "step": 25430
    },
    {
      "epoch": 1.3848195871025708,
      "grad_norm": 0.5268907779907597,
      "learning_rate": 4.5650285871581376e-05,
      "loss": 11.873,
      "step": 25431
    },
    {
      "epoch": 1.3848740410991538,
      "grad_norm": 0.623219367007664,
      "learning_rate": 4.564288393984313e-05,
      "loss": 11.9674,
      "step": 25432
    },
    {
      "epoch": 1.384928495095737,
      "grad_norm": 0.5609954238970658,
      "learning_rate": 4.563548243079495e-05,
      "loss": 11.8812,
      "step": 25433
    },
    {
      "epoch": 1.38498294909232,
      "grad_norm": 0.5168749717472204,
      "learning_rate": 4.562808134449436e-05,
      "loss": 11.9109,
      "step": 25434
    },
    {
      "epoch": 1.385037403088903,
      "grad_norm": 0.545321160679276,
      "learning_rate": 4.562068068099895e-05,
      "loss": 11.8895,
      "step": 25435
    },
    {
      "epoch": 1.385091857085486,
      "grad_norm": 0.6123168173423389,
      "learning_rate": 4.561328044036625e-05,
      "loss": 11.9163,
      "step": 25436
    },
    {
      "epoch": 1.385146311082069,
      "grad_norm": 0.5544430228111364,
      "learning_rate": 4.5605880622653766e-05,
      "loss": 11.8053,
      "step": 25437
    },
    {
      "epoch": 1.385200765078652,
      "grad_norm": 0.5243571797546914,
      "learning_rate": 4.559848122791911e-05,
      "loss": 12.0372,
      "step": 25438
    },
    {
      "epoch": 1.385255219075235,
      "grad_norm": 0.5226354727460679,
      "learning_rate": 4.559108225621975e-05,
      "loss": 11.8838,
      "step": 25439
    },
    {
      "epoch": 1.385309673071818,
      "grad_norm": 0.5399147634701176,
      "learning_rate": 4.55836837076133e-05,
      "loss": 11.9785,
      "step": 25440
    },
    {
      "epoch": 1.385364127068401,
      "grad_norm": 0.5556679512831352,
      "learning_rate": 4.557628558215722e-05,
      "loss": 11.7897,
      "step": 25441
    },
    {
      "epoch": 1.3854185810649842,
      "grad_norm": 0.5826198731737683,
      "learning_rate": 4.5568887879909096e-05,
      "loss": 11.6895,
      "step": 25442
    },
    {
      "epoch": 1.3854730350615672,
      "grad_norm": 0.5721627565551967,
      "learning_rate": 4.556149060092639e-05,
      "loss": 11.9325,
      "step": 25443
    },
    {
      "epoch": 1.3855274890581502,
      "grad_norm": 0.627853917146858,
      "learning_rate": 4.555409374526668e-05,
      "loss": 11.8464,
      "step": 25444
    },
    {
      "epoch": 1.3855819430547331,
      "grad_norm": 0.5387585089526937,
      "learning_rate": 4.5546697312987484e-05,
      "loss": 11.9004,
      "step": 25445
    },
    {
      "epoch": 1.3856363970513161,
      "grad_norm": 0.5306238776071194,
      "learning_rate": 4.553930130414631e-05,
      "loss": 11.9283,
      "step": 25446
    },
    {
      "epoch": 1.3856908510478991,
      "grad_norm": 0.5280528107632034,
      "learning_rate": 4.5531905718800627e-05,
      "loss": 11.858,
      "step": 25447
    },
    {
      "epoch": 1.3857453050444821,
      "grad_norm": 0.5283147579328331,
      "learning_rate": 4.5524510557008014e-05,
      "loss": 11.7974,
      "step": 25448
    },
    {
      "epoch": 1.3857997590410651,
      "grad_norm": 0.5155778636904704,
      "learning_rate": 4.551711581882591e-05,
      "loss": 11.8523,
      "step": 25449
    },
    {
      "epoch": 1.385854213037648,
      "grad_norm": 0.5445239226906861,
      "learning_rate": 4.5509721504311877e-05,
      "loss": 11.909,
      "step": 25450
    },
    {
      "epoch": 1.385908667034231,
      "grad_norm": 0.5471540714237298,
      "learning_rate": 4.550232761352335e-05,
      "loss": 11.9284,
      "step": 25451
    },
    {
      "epoch": 1.385963121030814,
      "grad_norm": 0.7100869057455694,
      "learning_rate": 4.5494934146517906e-05,
      "loss": 12.0101,
      "step": 25452
    },
    {
      "epoch": 1.386017575027397,
      "grad_norm": 0.5891335356799535,
      "learning_rate": 4.548754110335297e-05,
      "loss": 11.8092,
      "step": 25453
    },
    {
      "epoch": 1.38607202902398,
      "grad_norm": 0.5924728538463313,
      "learning_rate": 4.548014848408607e-05,
      "loss": 11.883,
      "step": 25454
    },
    {
      "epoch": 1.386126483020563,
      "grad_norm": 0.5510637713760189,
      "learning_rate": 4.5472756288774656e-05,
      "loss": 11.8986,
      "step": 25455
    },
    {
      "epoch": 1.386180937017146,
      "grad_norm": 0.5799269518433228,
      "learning_rate": 4.5465364517476275e-05,
      "loss": 12.1175,
      "step": 25456
    },
    {
      "epoch": 1.3862353910137293,
      "grad_norm": 0.500171227649584,
      "learning_rate": 4.545797317024835e-05,
      "loss": 11.8863,
      "step": 25457
    },
    {
      "epoch": 1.3862898450103123,
      "grad_norm": 0.5859140594240135,
      "learning_rate": 4.545058224714834e-05,
      "loss": 11.8102,
      "step": 25458
    },
    {
      "epoch": 1.3863442990068953,
      "grad_norm": 0.5806878009529896,
      "learning_rate": 4.544319174823376e-05,
      "loss": 11.9901,
      "step": 25459
    },
    {
      "epoch": 1.3863987530034783,
      "grad_norm": 0.5696105103622016,
      "learning_rate": 4.5435801673562096e-05,
      "loss": 11.9582,
      "step": 25460
    },
    {
      "epoch": 1.3864532070000612,
      "grad_norm": 0.5470078970069159,
      "learning_rate": 4.542841202319076e-05,
      "loss": 11.8457,
      "step": 25461
    },
    {
      "epoch": 1.3865076609966442,
      "grad_norm": 0.6089801746010085,
      "learning_rate": 4.542102279717727e-05,
      "loss": 11.8616,
      "step": 25462
    },
    {
      "epoch": 1.3865621149932272,
      "grad_norm": 0.5430565653973743,
      "learning_rate": 4.541363399557903e-05,
      "loss": 11.8596,
      "step": 25463
    },
    {
      "epoch": 1.3866165689898102,
      "grad_norm": 0.5840183025092082,
      "learning_rate": 4.540624561845356e-05,
      "loss": 11.7728,
      "step": 25464
    },
    {
      "epoch": 1.3866710229863934,
      "grad_norm": 0.5835431124442011,
      "learning_rate": 4.5398857665858243e-05,
      "loss": 11.8373,
      "step": 25465
    },
    {
      "epoch": 1.3867254769829764,
      "grad_norm": 0.503293458535758,
      "learning_rate": 4.53914701378506e-05,
      "loss": 11.7757,
      "step": 25466
    },
    {
      "epoch": 1.3867799309795594,
      "grad_norm": 0.5566424047691468,
      "learning_rate": 4.538408303448804e-05,
      "loss": 11.7665,
      "step": 25467
    },
    {
      "epoch": 1.3868343849761424,
      "grad_norm": 0.6272897259346766,
      "learning_rate": 4.537669635582799e-05,
      "loss": 11.9857,
      "step": 25468
    },
    {
      "epoch": 1.3868888389727254,
      "grad_norm": 0.6602007550727755,
      "learning_rate": 4.5369310101927933e-05,
      "loss": 11.8863,
      "step": 25469
    },
    {
      "epoch": 1.3869432929693084,
      "grad_norm": 0.5598081238967012,
      "learning_rate": 4.5361924272845246e-05,
      "loss": 11.9909,
      "step": 25470
    },
    {
      "epoch": 1.3869977469658914,
      "grad_norm": 0.5666801433389889,
      "learning_rate": 4.5354538868637395e-05,
      "loss": 11.8528,
      "step": 25471
    },
    {
      "epoch": 1.3870522009624744,
      "grad_norm": 0.5043592237807832,
      "learning_rate": 4.534715388936186e-05,
      "loss": 11.9294,
      "step": 25472
    },
    {
      "epoch": 1.3871066549590574,
      "grad_norm": 0.5636143197585487,
      "learning_rate": 4.5339769335075986e-05,
      "loss": 11.9502,
      "step": 25473
    },
    {
      "epoch": 1.3871611089556404,
      "grad_norm": 0.5240660368865463,
      "learning_rate": 4.533238520583726e-05,
      "loss": 11.9469,
      "step": 25474
    },
    {
      "epoch": 1.3872155629522234,
      "grad_norm": 0.5128105974233724,
      "learning_rate": 4.532500150170305e-05,
      "loss": 11.9357,
      "step": 25475
    },
    {
      "epoch": 1.3872700169488064,
      "grad_norm": 0.5404955144886905,
      "learning_rate": 4.531761822273082e-05,
      "loss": 11.9008,
      "step": 25476
    },
    {
      "epoch": 1.3873244709453894,
      "grad_norm": 0.5251588220746668,
      "learning_rate": 4.531023536897797e-05,
      "loss": 11.8544,
      "step": 25477
    },
    {
      "epoch": 1.3873789249419723,
      "grad_norm": 0.5430300012882553,
      "learning_rate": 4.530285294050186e-05,
      "loss": 11.9175,
      "step": 25478
    },
    {
      "epoch": 1.3874333789385553,
      "grad_norm": 0.5686834287718434,
      "learning_rate": 4.5295470937359976e-05,
      "loss": 12.0042,
      "step": 25479
    },
    {
      "epoch": 1.3874878329351386,
      "grad_norm": 0.5858173784688535,
      "learning_rate": 4.528808935960964e-05,
      "loss": 11.9313,
      "step": 25480
    },
    {
      "epoch": 1.3875422869317215,
      "grad_norm": 0.5593095346896827,
      "learning_rate": 4.528070820730831e-05,
      "loss": 11.8404,
      "step": 25481
    },
    {
      "epoch": 1.3875967409283045,
      "grad_norm": 0.5063867083757851,
      "learning_rate": 4.5273327480513395e-05,
      "loss": 11.8159,
      "step": 25482
    },
    {
      "epoch": 1.3876511949248875,
      "grad_norm": 0.6199157314152283,
      "learning_rate": 4.526594717928223e-05,
      "loss": 11.986,
      "step": 25483
    },
    {
      "epoch": 1.3877056489214705,
      "grad_norm": 0.5319134109511767,
      "learning_rate": 4.5258567303672286e-05,
      "loss": 11.9429,
      "step": 25484
    },
    {
      "epoch": 1.3877601029180535,
      "grad_norm": 0.583256809519084,
      "learning_rate": 4.525118785374085e-05,
      "loss": 11.8272,
      "step": 25485
    },
    {
      "epoch": 1.3878145569146365,
      "grad_norm": 0.5252811924613185,
      "learning_rate": 4.52438088295454e-05,
      "loss": 11.9259,
      "step": 25486
    },
    {
      "epoch": 1.3878690109112195,
      "grad_norm": 0.6129171070906239,
      "learning_rate": 4.523643023114328e-05,
      "loss": 11.8242,
      "step": 25487
    },
    {
      "epoch": 1.3879234649078025,
      "grad_norm": 0.5827912163869879,
      "learning_rate": 4.522905205859182e-05,
      "loss": 12.0187,
      "step": 25488
    },
    {
      "epoch": 1.3879779189043857,
      "grad_norm": 0.48887177592055336,
      "learning_rate": 4.522167431194848e-05,
      "loss": 11.9158,
      "step": 25489
    },
    {
      "epoch": 1.3880323729009687,
      "grad_norm": 0.5907938633320445,
      "learning_rate": 4.521429699127054e-05,
      "loss": 11.9518,
      "step": 25490
    },
    {
      "epoch": 1.3880868268975517,
      "grad_norm": 0.540746337114204,
      "learning_rate": 4.5206920096615455e-05,
      "loss": 11.9923,
      "step": 25491
    },
    {
      "epoch": 1.3881412808941347,
      "grad_norm": 0.5476902998471366,
      "learning_rate": 4.519954362804052e-05,
      "loss": 11.8386,
      "step": 25492
    },
    {
      "epoch": 1.3881957348907177,
      "grad_norm": 0.5363041472656379,
      "learning_rate": 4.519216758560312e-05,
      "loss": 11.8305,
      "step": 25493
    },
    {
      "epoch": 1.3882501888873007,
      "grad_norm": 0.5791260720557126,
      "learning_rate": 4.518479196936064e-05,
      "loss": 11.9111,
      "step": 25494
    },
    {
      "epoch": 1.3883046428838837,
      "grad_norm": 0.6012232839727382,
      "learning_rate": 4.5177416779370386e-05,
      "loss": 11.9442,
      "step": 25495
    },
    {
      "epoch": 1.3883590968804667,
      "grad_norm": 0.559152603766884,
      "learning_rate": 4.5170042015689765e-05,
      "loss": 11.9554,
      "step": 25496
    },
    {
      "epoch": 1.3884135508770497,
      "grad_norm": 0.5294207091331442,
      "learning_rate": 4.5162667678376094e-05,
      "loss": 11.8791,
      "step": 25497
    },
    {
      "epoch": 1.3884680048736326,
      "grad_norm": 0.5144270454198645,
      "learning_rate": 4.5155293767486686e-05,
      "loss": 11.8537,
      "step": 25498
    },
    {
      "epoch": 1.3885224588702156,
      "grad_norm": 0.5470857206995124,
      "learning_rate": 4.5147920283078936e-05,
      "loss": 11.8058,
      "step": 25499
    },
    {
      "epoch": 1.3885769128667986,
      "grad_norm": 0.5702107730197208,
      "learning_rate": 4.514054722521013e-05,
      "loss": 11.9726,
      "step": 25500
    },
    {
      "epoch": 1.3886313668633816,
      "grad_norm": 0.5766661552214912,
      "learning_rate": 4.513317459393765e-05,
      "loss": 11.8671,
      "step": 25501
    },
    {
      "epoch": 1.3886858208599646,
      "grad_norm": 0.5540396004218782,
      "learning_rate": 4.5125802389318785e-05,
      "loss": 11.9179,
      "step": 25502
    },
    {
      "epoch": 1.3887402748565478,
      "grad_norm": 0.5613776022593168,
      "learning_rate": 4.511843061141091e-05,
      "loss": 11.9476,
      "step": 25503
    },
    {
      "epoch": 1.3887947288531308,
      "grad_norm": 0.524968666061809,
      "learning_rate": 4.5111059260271294e-05,
      "loss": 11.9534,
      "step": 25504
    },
    {
      "epoch": 1.3888491828497138,
      "grad_norm": 0.5134529876106015,
      "learning_rate": 4.5103688335957276e-05,
      "loss": 11.8218,
      "step": 25505
    },
    {
      "epoch": 1.3889036368462968,
      "grad_norm": 0.550685330143858,
      "learning_rate": 4.509631783852626e-05,
      "loss": 11.9059,
      "step": 25506
    },
    {
      "epoch": 1.3889580908428798,
      "grad_norm": 0.5379171415541388,
      "learning_rate": 4.508894776803542e-05,
      "loss": 11.9005,
      "step": 25507
    },
    {
      "epoch": 1.3890125448394628,
      "grad_norm": 0.5284340806057991,
      "learning_rate": 4.5081578124542126e-05,
      "loss": 11.952,
      "step": 25508
    },
    {
      "epoch": 1.3890669988360458,
      "grad_norm": 0.5385966255964703,
      "learning_rate": 4.507420890810372e-05,
      "loss": 11.7745,
      "step": 25509
    },
    {
      "epoch": 1.3891214528326288,
      "grad_norm": 0.513349547827117,
      "learning_rate": 4.506684011877744e-05,
      "loss": 11.8494,
      "step": 25510
    },
    {
      "epoch": 1.3891759068292118,
      "grad_norm": 0.5573192302187278,
      "learning_rate": 4.505947175662066e-05,
      "loss": 11.959,
      "step": 25511
    },
    {
      "epoch": 1.389230360825795,
      "grad_norm": 0.5856179498725235,
      "learning_rate": 4.505210382169062e-05,
      "loss": 12.0187,
      "step": 25512
    },
    {
      "epoch": 1.389284814822378,
      "grad_norm": 0.5847391135503383,
      "learning_rate": 4.504473631404465e-05,
      "loss": 11.9218,
      "step": 25513
    },
    {
      "epoch": 1.389339268818961,
      "grad_norm": 0.5553770178542953,
      "learning_rate": 4.503736923374e-05,
      "loss": 11.8982,
      "step": 25514
    },
    {
      "epoch": 1.389393722815544,
      "grad_norm": 0.5086326480257842,
      "learning_rate": 4.5030002580833985e-05,
      "loss": 11.6923,
      "step": 25515
    },
    {
      "epoch": 1.389448176812127,
      "grad_norm": 0.5409528637701126,
      "learning_rate": 4.5022636355383966e-05,
      "loss": 11.9011,
      "step": 25516
    },
    {
      "epoch": 1.38950263080871,
      "grad_norm": 0.5161787951550454,
      "learning_rate": 4.501527055744707e-05,
      "loss": 11.8725,
      "step": 25517
    },
    {
      "epoch": 1.389557084805293,
      "grad_norm": 0.5369030135149553,
      "learning_rate": 4.500790518708068e-05,
      "loss": 11.6291,
      "step": 25518
    },
    {
      "epoch": 1.389611538801876,
      "grad_norm": 0.5515006321898562,
      "learning_rate": 4.5000540244342015e-05,
      "loss": 11.858,
      "step": 25519
    },
    {
      "epoch": 1.389665992798459,
      "grad_norm": 0.6431150513188725,
      "learning_rate": 4.4993175729288374e-05,
      "loss": 11.751,
      "step": 25520
    },
    {
      "epoch": 1.389720446795042,
      "grad_norm": 0.6076053213737065,
      "learning_rate": 4.498581164197705e-05,
      "loss": 12.0461,
      "step": 25521
    },
    {
      "epoch": 1.389774900791625,
      "grad_norm": 0.46667096650716783,
      "learning_rate": 4.4978447982465247e-05,
      "loss": 11.7802,
      "step": 25522
    },
    {
      "epoch": 1.389829354788208,
      "grad_norm": 0.5114626162584534,
      "learning_rate": 4.49710847508103e-05,
      "loss": 12.0273,
      "step": 25523
    },
    {
      "epoch": 1.389883808784791,
      "grad_norm": 0.6418240926117168,
      "learning_rate": 4.4963721947069383e-05,
      "loss": 11.9599,
      "step": 25524
    },
    {
      "epoch": 1.389938262781374,
      "grad_norm": 0.4839788863961615,
      "learning_rate": 4.495635957129983e-05,
      "loss": 11.7584,
      "step": 25525
    },
    {
      "epoch": 1.389992716777957,
      "grad_norm": 0.5026526852372909,
      "learning_rate": 4.494899762355887e-05,
      "loss": 11.8142,
      "step": 25526
    },
    {
      "epoch": 1.39004717077454,
      "grad_norm": 0.5296149944903044,
      "learning_rate": 4.4941636103903684e-05,
      "loss": 11.9144,
      "step": 25527
    },
    {
      "epoch": 1.390101624771123,
      "grad_norm": 0.515495589090639,
      "learning_rate": 4.493427501239161e-05,
      "loss": 11.8679,
      "step": 25528
    },
    {
      "epoch": 1.390156078767706,
      "grad_norm": 0.5398518962699593,
      "learning_rate": 4.492691434907982e-05,
      "loss": 11.8615,
      "step": 25529
    },
    {
      "epoch": 1.390210532764289,
      "grad_norm": 0.581374360757756,
      "learning_rate": 4.491955411402557e-05,
      "loss": 11.9026,
      "step": 25530
    },
    {
      "epoch": 1.390264986760872,
      "grad_norm": 0.5838567757464395,
      "learning_rate": 4.491219430728615e-05,
      "loss": 11.9498,
      "step": 25531
    },
    {
      "epoch": 1.390319440757455,
      "grad_norm": 0.5434824934602308,
      "learning_rate": 4.4904834928918696e-05,
      "loss": 11.8908,
      "step": 25532
    },
    {
      "epoch": 1.390373894754038,
      "grad_norm": 0.5400376119511892,
      "learning_rate": 4.489747597898053e-05,
      "loss": 11.8023,
      "step": 25533
    },
    {
      "epoch": 1.390428348750621,
      "grad_norm": 0.5228692699215824,
      "learning_rate": 4.489011745752879e-05,
      "loss": 11.87,
      "step": 25534
    },
    {
      "epoch": 1.3904828027472043,
      "grad_norm": 0.5350715212943529,
      "learning_rate": 4.4882759364620787e-05,
      "loss": 11.944,
      "step": 25535
    },
    {
      "epoch": 1.3905372567437873,
      "grad_norm": 0.5297107854174508,
      "learning_rate": 4.487540170031368e-05,
      "loss": 11.9025,
      "step": 25536
    },
    {
      "epoch": 1.3905917107403702,
      "grad_norm": 0.5217943928297886,
      "learning_rate": 4.486804446466465e-05,
      "loss": 11.8636,
      "step": 25537
    },
    {
      "epoch": 1.3906461647369532,
      "grad_norm": 0.5646540885623449,
      "learning_rate": 4.4860687657731004e-05,
      "loss": 11.9166,
      "step": 25538
    },
    {
      "epoch": 1.3907006187335362,
      "grad_norm": 0.5172199674405743,
      "learning_rate": 4.4853331279569856e-05,
      "loss": 11.7803,
      "step": 25539
    },
    {
      "epoch": 1.3907550727301192,
      "grad_norm": 0.5196236659981314,
      "learning_rate": 4.484597533023849e-05,
      "loss": 11.8839,
      "step": 25540
    },
    {
      "epoch": 1.3908095267267022,
      "grad_norm": 0.5585562010892676,
      "learning_rate": 4.4838619809794025e-05,
      "loss": 11.8597,
      "step": 25541
    },
    {
      "epoch": 1.3908639807232852,
      "grad_norm": 0.511576594652112,
      "learning_rate": 4.483126471829371e-05,
      "loss": 11.8981,
      "step": 25542
    },
    {
      "epoch": 1.3909184347198682,
      "grad_norm": 0.6139840837053527,
      "learning_rate": 4.482391005579476e-05,
      "loss": 12.0498,
      "step": 25543
    },
    {
      "epoch": 1.3909728887164512,
      "grad_norm": 0.5784039918772768,
      "learning_rate": 4.4816555822354314e-05,
      "loss": 11.9092,
      "step": 25544
    },
    {
      "epoch": 1.3910273427130342,
      "grad_norm": 0.5074897671565245,
      "learning_rate": 4.480920201802961e-05,
      "loss": 11.6981,
      "step": 25545
    },
    {
      "epoch": 1.3910817967096172,
      "grad_norm": 0.4931749930730538,
      "learning_rate": 4.480184864287781e-05,
      "loss": 11.872,
      "step": 25546
    },
    {
      "epoch": 1.3911362507062002,
      "grad_norm": 0.5373058663523753,
      "learning_rate": 4.479449569695606e-05,
      "loss": 11.7687,
      "step": 25547
    },
    {
      "epoch": 1.3911907047027832,
      "grad_norm": 0.5148041142482348,
      "learning_rate": 4.47871431803216e-05,
      "loss": 11.8017,
      "step": 25548
    },
    {
      "epoch": 1.3912451586993662,
      "grad_norm": 0.5340044989149939,
      "learning_rate": 4.477979109303154e-05,
      "loss": 11.9403,
      "step": 25549
    },
    {
      "epoch": 1.3912996126959494,
      "grad_norm": 0.4967906977063719,
      "learning_rate": 4.4772439435143124e-05,
      "loss": 11.8576,
      "step": 25550
    },
    {
      "epoch": 1.3913540666925324,
      "grad_norm": 0.5141829250771999,
      "learning_rate": 4.4765088206713434e-05,
      "loss": 11.969,
      "step": 25551
    },
    {
      "epoch": 1.3914085206891154,
      "grad_norm": 0.5594723020273837,
      "learning_rate": 4.4757737407799724e-05,
      "loss": 11.8893,
      "step": 25552
    },
    {
      "epoch": 1.3914629746856984,
      "grad_norm": 0.5806813743109116,
      "learning_rate": 4.4750387038459086e-05,
      "loss": 11.8356,
      "step": 25553
    },
    {
      "epoch": 1.3915174286822813,
      "grad_norm": 0.568429998278889,
      "learning_rate": 4.4743037098748696e-05,
      "loss": 11.8732,
      "step": 25554
    },
    {
      "epoch": 1.3915718826788643,
      "grad_norm": 0.5419538497127687,
      "learning_rate": 4.473568758872575e-05,
      "loss": 11.7468,
      "step": 25555
    },
    {
      "epoch": 1.3916263366754473,
      "grad_norm": 0.5225619019455237,
      "learning_rate": 4.472833850844736e-05,
      "loss": 11.8665,
      "step": 25556
    },
    {
      "epoch": 1.3916807906720303,
      "grad_norm": 0.5396102597901121,
      "learning_rate": 4.4720989857970654e-05,
      "loss": 12.0524,
      "step": 25557
    },
    {
      "epoch": 1.3917352446686133,
      "grad_norm": 0.5108951221640262,
      "learning_rate": 4.471364163735283e-05,
      "loss": 11.9733,
      "step": 25558
    },
    {
      "epoch": 1.3917896986651965,
      "grad_norm": 0.5130039467516576,
      "learning_rate": 4.470629384665096e-05,
      "loss": 11.9014,
      "step": 25559
    },
    {
      "epoch": 1.3918441526617795,
      "grad_norm": 0.5247712073740278,
      "learning_rate": 4.469894648592227e-05,
      "loss": 11.9558,
      "step": 25560
    },
    {
      "epoch": 1.3918986066583625,
      "grad_norm": 0.5626381865461848,
      "learning_rate": 4.46915995552238e-05,
      "loss": 11.9597,
      "step": 25561
    },
    {
      "epoch": 1.3919530606549455,
      "grad_norm": 0.5176142223681709,
      "learning_rate": 4.468425305461277e-05,
      "loss": 11.8549,
      "step": 25562
    },
    {
      "epoch": 1.3920075146515285,
      "grad_norm": 0.5361655587293035,
      "learning_rate": 4.4676906984146225e-05,
      "loss": 11.8692,
      "step": 25563
    },
    {
      "epoch": 1.3920619686481115,
      "grad_norm": 0.5196121113509619,
      "learning_rate": 4.4669561343881326e-05,
      "loss": 11.7564,
      "step": 25564
    },
    {
      "epoch": 1.3921164226446945,
      "grad_norm": 0.5965802711646296,
      "learning_rate": 4.466221613387524e-05,
      "loss": 11.9678,
      "step": 25565
    },
    {
      "epoch": 1.3921708766412775,
      "grad_norm": 0.5278279479966292,
      "learning_rate": 4.465487135418504e-05,
      "loss": 11.8423,
      "step": 25566
    },
    {
      "epoch": 1.3922253306378605,
      "grad_norm": 0.4899969182386487,
      "learning_rate": 4.464752700486784e-05,
      "loss": 11.8754,
      "step": 25567
    },
    {
      "epoch": 1.3922797846344435,
      "grad_norm": 0.5544441426716407,
      "learning_rate": 4.4640183085980715e-05,
      "loss": 11.9249,
      "step": 25568
    },
    {
      "epoch": 1.3923342386310265,
      "grad_norm": 0.534695084533497,
      "learning_rate": 4.463283959758081e-05,
      "loss": 11.8792,
      "step": 25569
    },
    {
      "epoch": 1.3923886926276094,
      "grad_norm": 0.5690192789312117,
      "learning_rate": 4.4625496539725264e-05,
      "loss": 11.8757,
      "step": 25570
    },
    {
      "epoch": 1.3924431466241924,
      "grad_norm": 0.4945283277132608,
      "learning_rate": 4.461815391247112e-05,
      "loss": 11.809,
      "step": 25571
    },
    {
      "epoch": 1.3924976006207754,
      "grad_norm": 0.6311904744090591,
      "learning_rate": 4.461081171587552e-05,
      "loss": 11.9291,
      "step": 25572
    },
    {
      "epoch": 1.3925520546173586,
      "grad_norm": 0.5131269660543281,
      "learning_rate": 4.4603469949995504e-05,
      "loss": 11.861,
      "step": 25573
    },
    {
      "epoch": 1.3926065086139416,
      "grad_norm": 0.5299962043621316,
      "learning_rate": 4.4596128614888235e-05,
      "loss": 11.7927,
      "step": 25574
    },
    {
      "epoch": 1.3926609626105246,
      "grad_norm": 0.4829217822093533,
      "learning_rate": 4.458878771061072e-05,
      "loss": 11.8651,
      "step": 25575
    },
    {
      "epoch": 1.3927154166071076,
      "grad_norm": 0.5485101424678609,
      "learning_rate": 4.458144723722013e-05,
      "loss": 11.9357,
      "step": 25576
    },
    {
      "epoch": 1.3927698706036906,
      "grad_norm": 0.5766981103581194,
      "learning_rate": 4.4574107194773495e-05,
      "loss": 11.898,
      "step": 25577
    },
    {
      "epoch": 1.3928243246002736,
      "grad_norm": 0.5593629114152067,
      "learning_rate": 4.456676758332785e-05,
      "loss": 12.0337,
      "step": 25578
    },
    {
      "epoch": 1.3928787785968566,
      "grad_norm": 0.5440644568553376,
      "learning_rate": 4.455942840294033e-05,
      "loss": 11.9586,
      "step": 25579
    },
    {
      "epoch": 1.3929332325934396,
      "grad_norm": 0.5471786288865869,
      "learning_rate": 4.455208965366803e-05,
      "loss": 11.9155,
      "step": 25580
    },
    {
      "epoch": 1.3929876865900226,
      "grad_norm": 0.5732036331337723,
      "learning_rate": 4.454475133556794e-05,
      "loss": 11.8596,
      "step": 25581
    },
    {
      "epoch": 1.3930421405866058,
      "grad_norm": 0.510982090028107,
      "learning_rate": 4.45374134486972e-05,
      "loss": 11.9077,
      "step": 25582
    },
    {
      "epoch": 1.3930965945831888,
      "grad_norm": 0.5482976943136708,
      "learning_rate": 4.4530075993112785e-05,
      "loss": 11.9974,
      "step": 25583
    },
    {
      "epoch": 1.3931510485797718,
      "grad_norm": 0.5461209763693049,
      "learning_rate": 4.4522738968871844e-05,
      "loss": 11.9263,
      "step": 25584
    },
    {
      "epoch": 1.3932055025763548,
      "grad_norm": 0.5476284641533394,
      "learning_rate": 4.451540237603137e-05,
      "loss": 11.8504,
      "step": 25585
    },
    {
      "epoch": 1.3932599565729378,
      "grad_norm": 0.5555034223480412,
      "learning_rate": 4.450806621464842e-05,
      "loss": 11.8776,
      "step": 25586
    },
    {
      "epoch": 1.3933144105695208,
      "grad_norm": 0.5415518693860845,
      "learning_rate": 4.450073048478012e-05,
      "loss": 11.9524,
      "step": 25587
    },
    {
      "epoch": 1.3933688645661038,
      "grad_norm": 0.4836607640817616,
      "learning_rate": 4.449339518648338e-05,
      "loss": 11.8876,
      "step": 25588
    },
    {
      "epoch": 1.3934233185626868,
      "grad_norm": 0.5806205706248562,
      "learning_rate": 4.448606031981534e-05,
      "loss": 11.9442,
      "step": 25589
    },
    {
      "epoch": 1.3934777725592697,
      "grad_norm": 0.5245357843627159,
      "learning_rate": 4.447872588483296e-05,
      "loss": 11.884,
      "step": 25590
    },
    {
      "epoch": 1.3935322265558527,
      "grad_norm": 0.5933993341539028,
      "learning_rate": 4.447139188159334e-05,
      "loss": 11.8883,
      "step": 25591
    },
    {
      "epoch": 1.3935866805524357,
      "grad_norm": 0.5626036087049634,
      "learning_rate": 4.446405831015351e-05,
      "loss": 11.9668,
      "step": 25592
    },
    {
      "epoch": 1.3936411345490187,
      "grad_norm": 0.5854899098641337,
      "learning_rate": 4.4456725170570446e-05,
      "loss": 11.9953,
      "step": 25593
    },
    {
      "epoch": 1.3936955885456017,
      "grad_norm": 0.583069753177118,
      "learning_rate": 4.444939246290123e-05,
      "loss": 11.967,
      "step": 25594
    },
    {
      "epoch": 1.3937500425421847,
      "grad_norm": 0.5900050905800227,
      "learning_rate": 4.444206018720283e-05,
      "loss": 11.8586,
      "step": 25595
    },
    {
      "epoch": 1.393804496538768,
      "grad_norm": 0.4992737059483989,
      "learning_rate": 4.443472834353232e-05,
      "loss": 11.8711,
      "step": 25596
    },
    {
      "epoch": 1.393858950535351,
      "grad_norm": 0.5004623240886439,
      "learning_rate": 4.4427396931946675e-05,
      "loss": 11.9301,
      "step": 25597
    },
    {
      "epoch": 1.393913404531934,
      "grad_norm": 0.5464915720685761,
      "learning_rate": 4.442006595250288e-05,
      "loss": 11.7774,
      "step": 25598
    },
    {
      "epoch": 1.393967858528517,
      "grad_norm": 0.5200105712903359,
      "learning_rate": 4.4412735405258e-05,
      "loss": 11.9007,
      "step": 25599
    },
    {
      "epoch": 1.3940223125251,
      "grad_norm": 0.5065329383440792,
      "learning_rate": 4.4405405290268976e-05,
      "loss": 11.952,
      "step": 25600
    },
    {
      "epoch": 1.3940767665216829,
      "grad_norm": 0.5755515418314147,
      "learning_rate": 4.439807560759285e-05,
      "loss": 11.9608,
      "step": 25601
    },
    {
      "epoch": 1.3941312205182659,
      "grad_norm": 0.5944838060345434,
      "learning_rate": 4.439074635728665e-05,
      "loss": 11.9533,
      "step": 25602
    },
    {
      "epoch": 1.3941856745148489,
      "grad_norm": 0.5493298611399842,
      "learning_rate": 4.438341753940729e-05,
      "loss": 11.8226,
      "step": 25603
    },
    {
      "epoch": 1.3942401285114319,
      "grad_norm": 0.5235303374651344,
      "learning_rate": 4.437608915401185e-05,
      "loss": 11.9139,
      "step": 25604
    },
    {
      "epoch": 1.394294582508015,
      "grad_norm": 0.5791529929868455,
      "learning_rate": 4.436876120115723e-05,
      "loss": 11.7544,
      "step": 25605
    },
    {
      "epoch": 1.394349036504598,
      "grad_norm": 0.5406835856521723,
      "learning_rate": 4.4361433680900474e-05,
      "loss": 11.9002,
      "step": 25606
    },
    {
      "epoch": 1.394403490501181,
      "grad_norm": 0.5226489783085283,
      "learning_rate": 4.435410659329855e-05,
      "loss": 11.8162,
      "step": 25607
    },
    {
      "epoch": 1.394457944497764,
      "grad_norm": 0.5366347793195804,
      "learning_rate": 4.434677993840839e-05,
      "loss": 11.8039,
      "step": 25608
    },
    {
      "epoch": 1.394512398494347,
      "grad_norm": 0.5243448582888759,
      "learning_rate": 4.433945371628704e-05,
      "loss": 11.8087,
      "step": 25609
    },
    {
      "epoch": 1.39456685249093,
      "grad_norm": 0.5581672782464008,
      "learning_rate": 4.433212792699138e-05,
      "loss": 11.8108,
      "step": 25610
    },
    {
      "epoch": 1.394621306487513,
      "grad_norm": 0.5777969188109933,
      "learning_rate": 4.432480257057848e-05,
      "loss": 12.0268,
      "step": 25611
    },
    {
      "epoch": 1.394675760484096,
      "grad_norm": 0.5438379840429651,
      "learning_rate": 4.431747764710522e-05,
      "loss": 11.8112,
      "step": 25612
    },
    {
      "epoch": 1.394730214480679,
      "grad_norm": 0.5495210033207965,
      "learning_rate": 4.431015315662859e-05,
      "loss": 11.909,
      "step": 25613
    },
    {
      "epoch": 1.394784668477262,
      "grad_norm": 0.5553615982274764,
      "learning_rate": 4.430282909920558e-05,
      "loss": 11.9504,
      "step": 25614
    },
    {
      "epoch": 1.394839122473845,
      "grad_norm": 0.5296519106752849,
      "learning_rate": 4.429550547489307e-05,
      "loss": 11.948,
      "step": 25615
    },
    {
      "epoch": 1.394893576470428,
      "grad_norm": 0.5662093547186087,
      "learning_rate": 4.4288182283748094e-05,
      "loss": 11.9523,
      "step": 25616
    },
    {
      "epoch": 1.394948030467011,
      "grad_norm": 0.5814211401726735,
      "learning_rate": 4.4280859525827557e-05,
      "loss": 11.9414,
      "step": 25617
    },
    {
      "epoch": 1.395002484463594,
      "grad_norm": 0.5176920390380298,
      "learning_rate": 4.427353720118835e-05,
      "loss": 11.8059,
      "step": 25618
    },
    {
      "epoch": 1.395056938460177,
      "grad_norm": 0.5637400044293797,
      "learning_rate": 4.426621530988751e-05,
      "loss": 11.9364,
      "step": 25619
    },
    {
      "epoch": 1.3951113924567602,
      "grad_norm": 0.5891127385535817,
      "learning_rate": 4.4258893851981895e-05,
      "loss": 11.7332,
      "step": 25620
    },
    {
      "epoch": 1.3951658464533432,
      "grad_norm": 0.5876437351777504,
      "learning_rate": 4.42515728275285e-05,
      "loss": 11.9318,
      "step": 25621
    },
    {
      "epoch": 1.3952203004499262,
      "grad_norm": 0.49966714443086996,
      "learning_rate": 4.424425223658419e-05,
      "loss": 11.9052,
      "step": 25622
    },
    {
      "epoch": 1.3952747544465092,
      "grad_norm": 0.5713599956178447,
      "learning_rate": 4.423693207920596e-05,
      "loss": 11.9294,
      "step": 25623
    },
    {
      "epoch": 1.3953292084430922,
      "grad_norm": 0.5477477694312434,
      "learning_rate": 4.4229612355450655e-05,
      "loss": 11.756,
      "step": 25624
    },
    {
      "epoch": 1.3953836624396752,
      "grad_norm": 0.48363716283238173,
      "learning_rate": 4.422229306537524e-05,
      "loss": 11.7877,
      "step": 25625
    },
    {
      "epoch": 1.3954381164362581,
      "grad_norm": 0.5237849668974416,
      "learning_rate": 4.421497420903668e-05,
      "loss": 11.8695,
      "step": 25626
    },
    {
      "epoch": 1.3954925704328411,
      "grad_norm": 0.5289670155708505,
      "learning_rate": 4.420765578649182e-05,
      "loss": 11.6797,
      "step": 25627
    },
    {
      "epoch": 1.3955470244294244,
      "grad_norm": 0.5385830800928001,
      "learning_rate": 4.420033779779754e-05,
      "loss": 11.8105,
      "step": 25628
    },
    {
      "epoch": 1.3956014784260073,
      "grad_norm": 0.5452827851664649,
      "learning_rate": 4.419302024301084e-05,
      "loss": 11.9292,
      "step": 25629
    },
    {
      "epoch": 1.3956559324225903,
      "grad_norm": 0.5027393070915539,
      "learning_rate": 4.4185703122188536e-05,
      "loss": 11.8437,
      "step": 25630
    },
    {
      "epoch": 1.3957103864191733,
      "grad_norm": 0.5452691469643637,
      "learning_rate": 4.417838643538761e-05,
      "loss": 11.9122,
      "step": 25631
    },
    {
      "epoch": 1.3957648404157563,
      "grad_norm": 0.572446812699789,
      "learning_rate": 4.417107018266486e-05,
      "loss": 11.8799,
      "step": 25632
    },
    {
      "epoch": 1.3958192944123393,
      "grad_norm": 0.5232701785438433,
      "learning_rate": 4.416375436407727e-05,
      "loss": 11.6885,
      "step": 25633
    },
    {
      "epoch": 1.3958737484089223,
      "grad_norm": 0.5650317891925306,
      "learning_rate": 4.415643897968166e-05,
      "loss": 11.8378,
      "step": 25634
    },
    {
      "epoch": 1.3959282024055053,
      "grad_norm": 0.597295783221329,
      "learning_rate": 4.414912402953494e-05,
      "loss": 11.8465,
      "step": 25635
    },
    {
      "epoch": 1.3959826564020883,
      "grad_norm": 0.5439921230625333,
      "learning_rate": 4.4141809513694043e-05,
      "loss": 11.8836,
      "step": 25636
    },
    {
      "epoch": 1.3960371103986713,
      "grad_norm": 0.5133879306520974,
      "learning_rate": 4.4134495432215795e-05,
      "loss": 11.894,
      "step": 25637
    },
    {
      "epoch": 1.3960915643952543,
      "grad_norm": 0.5308147457633371,
      "learning_rate": 4.4127181785157077e-05,
      "loss": 11.8861,
      "step": 25638
    },
    {
      "epoch": 1.3961460183918373,
      "grad_norm": 0.580976142919681,
      "learning_rate": 4.411986857257473e-05,
      "loss": 11.8624,
      "step": 25639
    },
    {
      "epoch": 1.3962004723884203,
      "grad_norm": 0.5221320733890685,
      "learning_rate": 4.411255579452566e-05,
      "loss": 11.8193,
      "step": 25640
    },
    {
      "epoch": 1.3962549263850033,
      "grad_norm": 0.5638440618118854,
      "learning_rate": 4.410524345106676e-05,
      "loss": 11.9277,
      "step": 25641
    },
    {
      "epoch": 1.3963093803815863,
      "grad_norm": 0.5243922082073832,
      "learning_rate": 4.409793154225482e-05,
      "loss": 11.8344,
      "step": 25642
    },
    {
      "epoch": 1.3963638343781695,
      "grad_norm": 0.5442168463416602,
      "learning_rate": 4.4090620068146785e-05,
      "loss": 11.8964,
      "step": 25643
    },
    {
      "epoch": 1.3964182883747525,
      "grad_norm": 0.5371329213804302,
      "learning_rate": 4.408330902879942e-05,
      "loss": 11.7654,
      "step": 25644
    },
    {
      "epoch": 1.3964727423713355,
      "grad_norm": 0.640722081839925,
      "learning_rate": 4.407599842426967e-05,
      "loss": 12.1038,
      "step": 25645
    },
    {
      "epoch": 1.3965271963679184,
      "grad_norm": 0.5837988396414286,
      "learning_rate": 4.406868825461429e-05,
      "loss": 11.963,
      "step": 25646
    },
    {
      "epoch": 1.3965816503645014,
      "grad_norm": 0.515178023263635,
      "learning_rate": 4.4061378519890206e-05,
      "loss": 11.874,
      "step": 25647
    },
    {
      "epoch": 1.3966361043610844,
      "grad_norm": 0.5772878515993738,
      "learning_rate": 4.4054069220154215e-05,
      "loss": 11.7675,
      "step": 25648
    },
    {
      "epoch": 1.3966905583576674,
      "grad_norm": 0.5585453623940094,
      "learning_rate": 4.404676035546313e-05,
      "loss": 11.896,
      "step": 25649
    },
    {
      "epoch": 1.3967450123542504,
      "grad_norm": 0.5446295610336994,
      "learning_rate": 4.403945192587383e-05,
      "loss": 11.9455,
      "step": 25650
    },
    {
      "epoch": 1.3967994663508334,
      "grad_norm": 0.520477903605035,
      "learning_rate": 4.403214393144317e-05,
      "loss": 11.984,
      "step": 25651
    },
    {
      "epoch": 1.3968539203474166,
      "grad_norm": 0.7631798390604274,
      "learning_rate": 4.402483637222791e-05,
      "loss": 11.8636,
      "step": 25652
    },
    {
      "epoch": 1.3969083743439996,
      "grad_norm": 0.5357976761801222,
      "learning_rate": 4.4017529248284936e-05,
      "loss": 11.844,
      "step": 25653
    },
    {
      "epoch": 1.3969628283405826,
      "grad_norm": 0.5375128494981319,
      "learning_rate": 4.4010222559671024e-05,
      "loss": 11.8929,
      "step": 25654
    },
    {
      "epoch": 1.3970172823371656,
      "grad_norm": 0.5684473019634566,
      "learning_rate": 4.400291630644303e-05,
      "loss": 11.8925,
      "step": 25655
    },
    {
      "epoch": 1.3970717363337486,
      "grad_norm": 0.6145092032176454,
      "learning_rate": 4.399561048865772e-05,
      "loss": 11.8634,
      "step": 25656
    },
    {
      "epoch": 1.3971261903303316,
      "grad_norm": 0.5871747327630277,
      "learning_rate": 4.398830510637198e-05,
      "loss": 12.017,
      "step": 25657
    },
    {
      "epoch": 1.3971806443269146,
      "grad_norm": 0.5408892166409712,
      "learning_rate": 4.398100015964256e-05,
      "loss": 11.9226,
      "step": 25658
    },
    {
      "epoch": 1.3972350983234976,
      "grad_norm": 0.6011747196050518,
      "learning_rate": 4.397369564852624e-05,
      "loss": 11.846,
      "step": 25659
    },
    {
      "epoch": 1.3972895523200806,
      "grad_norm": 0.5365215992691857,
      "learning_rate": 4.39663915730799e-05,
      "loss": 11.9564,
      "step": 25660
    },
    {
      "epoch": 1.3973440063166636,
      "grad_norm": 0.5578866734708908,
      "learning_rate": 4.395908793336026e-05,
      "loss": 11.9316,
      "step": 25661
    },
    {
      "epoch": 1.3973984603132465,
      "grad_norm": 0.5684847679081799,
      "learning_rate": 4.395178472942415e-05,
      "loss": 11.9762,
      "step": 25662
    },
    {
      "epoch": 1.3974529143098295,
      "grad_norm": 0.5737354362562226,
      "learning_rate": 4.394448196132839e-05,
      "loss": 11.8986,
      "step": 25663
    },
    {
      "epoch": 1.3975073683064125,
      "grad_norm": 0.5643940556914142,
      "learning_rate": 4.393717962912971e-05,
      "loss": 11.8787,
      "step": 25664
    },
    {
      "epoch": 1.3975618223029955,
      "grad_norm": 0.5163855351759655,
      "learning_rate": 4.392987773288496e-05,
      "loss": 11.9327,
      "step": 25665
    },
    {
      "epoch": 1.3976162762995787,
      "grad_norm": 0.5165131419827805,
      "learning_rate": 4.392257627265084e-05,
      "loss": 11.8745,
      "step": 25666
    },
    {
      "epoch": 1.3976707302961617,
      "grad_norm": 0.5623763370391212,
      "learning_rate": 4.3915275248484224e-05,
      "loss": 11.8753,
      "step": 25667
    },
    {
      "epoch": 1.3977251842927447,
      "grad_norm": 0.5021095954282441,
      "learning_rate": 4.390797466044182e-05,
      "loss": 11.8867,
      "step": 25668
    },
    {
      "epoch": 1.3977796382893277,
      "grad_norm": 0.5343072695118061,
      "learning_rate": 4.3900674508580374e-05,
      "loss": 11.9957,
      "step": 25669
    },
    {
      "epoch": 1.3978340922859107,
      "grad_norm": 0.5840461650399251,
      "learning_rate": 4.389337479295673e-05,
      "loss": 11.9446,
      "step": 25670
    },
    {
      "epoch": 1.3978885462824937,
      "grad_norm": 0.5555837544430504,
      "learning_rate": 4.388607551362758e-05,
      "loss": 11.8412,
      "step": 25671
    },
    {
      "epoch": 1.3979430002790767,
      "grad_norm": 0.5567206681928537,
      "learning_rate": 4.3878776670649746e-05,
      "loss": 11.841,
      "step": 25672
    },
    {
      "epoch": 1.3979974542756597,
      "grad_norm": 0.5194923352163608,
      "learning_rate": 4.3871478264079924e-05,
      "loss": 11.7848,
      "step": 25673
    },
    {
      "epoch": 1.3980519082722427,
      "grad_norm": 0.49589168254954724,
      "learning_rate": 4.38641802939749e-05,
      "loss": 11.8045,
      "step": 25674
    },
    {
      "epoch": 1.398106362268826,
      "grad_norm": 0.5753414999346085,
      "learning_rate": 4.385688276039146e-05,
      "loss": 11.8016,
      "step": 25675
    },
    {
      "epoch": 1.398160816265409,
      "grad_norm": 0.5054072524763868,
      "learning_rate": 4.384958566338627e-05,
      "loss": 11.9053,
      "step": 25676
    },
    {
      "epoch": 1.3982152702619919,
      "grad_norm": 0.5207491949903695,
      "learning_rate": 4.384228900301617e-05,
      "loss": 11.877,
      "step": 25677
    },
    {
      "epoch": 1.3982697242585749,
      "grad_norm": 0.6050903200238807,
      "learning_rate": 4.3834992779337845e-05,
      "loss": 11.9729,
      "step": 25678
    },
    {
      "epoch": 1.3983241782551579,
      "grad_norm": 0.5035147841598213,
      "learning_rate": 4.3827696992407995e-05,
      "loss": 11.8971,
      "step": 25679
    },
    {
      "epoch": 1.3983786322517409,
      "grad_norm": 0.5849917897173539,
      "learning_rate": 4.382040164228343e-05,
      "loss": 11.9739,
      "step": 25680
    },
    {
      "epoch": 1.3984330862483239,
      "grad_norm": 0.5173595634324137,
      "learning_rate": 4.3813106729020806e-05,
      "loss": 11.7193,
      "step": 25681
    },
    {
      "epoch": 1.3984875402449068,
      "grad_norm": 0.5088712814321042,
      "learning_rate": 4.380581225267693e-05,
      "loss": 11.8223,
      "step": 25682
    },
    {
      "epoch": 1.3985419942414898,
      "grad_norm": 0.5513836839263556,
      "learning_rate": 4.379851821330843e-05,
      "loss": 11.8423,
      "step": 25683
    },
    {
      "epoch": 1.3985964482380728,
      "grad_norm": 0.5057797012944087,
      "learning_rate": 4.379122461097208e-05,
      "loss": 11.6728,
      "step": 25684
    },
    {
      "epoch": 1.3986509022346558,
      "grad_norm": 0.5718664348820757,
      "learning_rate": 4.3783931445724645e-05,
      "loss": 11.9907,
      "step": 25685
    },
    {
      "epoch": 1.3987053562312388,
      "grad_norm": 0.5910400993603159,
      "learning_rate": 4.377663871762274e-05,
      "loss": 12.0681,
      "step": 25686
    },
    {
      "epoch": 1.3987598102278218,
      "grad_norm": 0.8504708164787771,
      "learning_rate": 4.376934642672319e-05,
      "loss": 11.9959,
      "step": 25687
    },
    {
      "epoch": 1.3988142642244048,
      "grad_norm": 0.5088988067425663,
      "learning_rate": 4.376205457308257e-05,
      "loss": 11.8203,
      "step": 25688
    },
    {
      "epoch": 1.3988687182209878,
      "grad_norm": 0.5316068582897,
      "learning_rate": 4.375476315675764e-05,
      "loss": 11.8087,
      "step": 25689
    },
    {
      "epoch": 1.398923172217571,
      "grad_norm": 0.5454832715356545,
      "learning_rate": 4.3747472177805135e-05,
      "loss": 11.8985,
      "step": 25690
    },
    {
      "epoch": 1.398977626214154,
      "grad_norm": 0.5422557431517581,
      "learning_rate": 4.37401816362817e-05,
      "loss": 11.9926,
      "step": 25691
    },
    {
      "epoch": 1.399032080210737,
      "grad_norm": 0.5232367087176085,
      "learning_rate": 4.373289153224407e-05,
      "loss": 11.794,
      "step": 25692
    },
    {
      "epoch": 1.39908653420732,
      "grad_norm": 0.5764699835647442,
      "learning_rate": 4.372560186574887e-05,
      "loss": 11.8392,
      "step": 25693
    },
    {
      "epoch": 1.399140988203903,
      "grad_norm": 0.5653583270220415,
      "learning_rate": 4.371831263685288e-05,
      "loss": 11.8948,
      "step": 25694
    },
    {
      "epoch": 1.399195442200486,
      "grad_norm": 0.5332046554670394,
      "learning_rate": 4.3711023845612675e-05,
      "loss": 11.964,
      "step": 25695
    },
    {
      "epoch": 1.399249896197069,
      "grad_norm": 0.5613374131471114,
      "learning_rate": 4.3703735492085e-05,
      "loss": 11.922,
      "step": 25696
    },
    {
      "epoch": 1.399304350193652,
      "grad_norm": 0.5404182515945336,
      "learning_rate": 4.369644757632658e-05,
      "loss": 11.9681,
      "step": 25697
    },
    {
      "epoch": 1.3993588041902352,
      "grad_norm": 0.4801664996309418,
      "learning_rate": 4.368916009839397e-05,
      "loss": 11.8808,
      "step": 25698
    },
    {
      "epoch": 1.3994132581868182,
      "grad_norm": 0.5792402197427226,
      "learning_rate": 4.368187305834388e-05,
      "loss": 11.7865,
      "step": 25699
    },
    {
      "epoch": 1.3994677121834012,
      "grad_norm": 0.5203970169492241,
      "learning_rate": 4.367458645623304e-05,
      "loss": 11.7728,
      "step": 25700
    },
    {
      "epoch": 1.3995221661799842,
      "grad_norm": 0.581598290117175,
      "learning_rate": 4.366730029211801e-05,
      "loss": 11.8598,
      "step": 25701
    },
    {
      "epoch": 1.3995766201765671,
      "grad_norm": 0.6091364758528232,
      "learning_rate": 4.366001456605554e-05,
      "loss": 11.893,
      "step": 25702
    },
    {
      "epoch": 1.3996310741731501,
      "grad_norm": 0.5477307382500277,
      "learning_rate": 4.365272927810221e-05,
      "loss": 11.8992,
      "step": 25703
    },
    {
      "epoch": 1.3996855281697331,
      "grad_norm": 0.5498687874056694,
      "learning_rate": 4.3645444428314754e-05,
      "loss": 11.7799,
      "step": 25704
    },
    {
      "epoch": 1.3997399821663161,
      "grad_norm": 0.5837738167710438,
      "learning_rate": 4.363816001674972e-05,
      "loss": 12.0774,
      "step": 25705
    },
    {
      "epoch": 1.3997944361628991,
      "grad_norm": 0.5492987014768775,
      "learning_rate": 4.363087604346385e-05,
      "loss": 11.9124,
      "step": 25706
    },
    {
      "epoch": 1.399848890159482,
      "grad_norm": 0.51453150083243,
      "learning_rate": 4.362359250851374e-05,
      "loss": 11.8891,
      "step": 25707
    },
    {
      "epoch": 1.399903344156065,
      "grad_norm": 0.5271191344921754,
      "learning_rate": 4.3616309411955994e-05,
      "loss": 11.8725,
      "step": 25708
    },
    {
      "epoch": 1.399957798152648,
      "grad_norm": 0.5335090745817481,
      "learning_rate": 4.3609026753847315e-05,
      "loss": 11.9013,
      "step": 25709
    },
    {
      "epoch": 1.400012252149231,
      "grad_norm": 0.5408831310830207,
      "learning_rate": 4.360174453424427e-05,
      "loss": 11.8668,
      "step": 25710
    },
    {
      "epoch": 1.400066706145814,
      "grad_norm": 0.5873790115469334,
      "learning_rate": 4.3594462753203516e-05,
      "loss": 11.8959,
      "step": 25711
    },
    {
      "epoch": 1.400121160142397,
      "grad_norm": 0.5084114241267758,
      "learning_rate": 4.358718141078171e-05,
      "loss": 11.7836,
      "step": 25712
    },
    {
      "epoch": 1.4001756141389803,
      "grad_norm": 0.5610339364949798,
      "learning_rate": 4.357990050703541e-05,
      "loss": 11.8994,
      "step": 25713
    },
    {
      "epoch": 1.4002300681355633,
      "grad_norm": 0.5326563329470386,
      "learning_rate": 4.3572620042021296e-05,
      "loss": 11.9368,
      "step": 25714
    },
    {
      "epoch": 1.4002845221321463,
      "grad_norm": 0.48618825218399014,
      "learning_rate": 4.356534001579592e-05,
      "loss": 11.7222,
      "step": 25715
    },
    {
      "epoch": 1.4003389761287293,
      "grad_norm": 0.5149899727645597,
      "learning_rate": 4.355806042841596e-05,
      "loss": 11.9233,
      "step": 25716
    },
    {
      "epoch": 1.4003934301253123,
      "grad_norm": 0.5748589060599986,
      "learning_rate": 4.355078127993798e-05,
      "loss": 11.9294,
      "step": 25717
    },
    {
      "epoch": 1.4004478841218952,
      "grad_norm": 0.683915970408038,
      "learning_rate": 4.354350257041857e-05,
      "loss": 11.9369,
      "step": 25718
    },
    {
      "epoch": 1.4005023381184782,
      "grad_norm": 0.5652671862783292,
      "learning_rate": 4.3536224299914374e-05,
      "loss": 11.8888,
      "step": 25719
    },
    {
      "epoch": 1.4005567921150612,
      "grad_norm": 0.5390889528715983,
      "learning_rate": 4.352894646848194e-05,
      "loss": 11.9011,
      "step": 25720
    },
    {
      "epoch": 1.4006112461116442,
      "grad_norm": 0.5569542491738358,
      "learning_rate": 4.352166907617792e-05,
      "loss": 11.8275,
      "step": 25721
    },
    {
      "epoch": 1.4006657001082274,
      "grad_norm": 0.526211258247391,
      "learning_rate": 4.351439212305883e-05,
      "loss": 11.989,
      "step": 25722
    },
    {
      "epoch": 1.4007201541048104,
      "grad_norm": 0.5817649898617785,
      "learning_rate": 4.3507115609181315e-05,
      "loss": 11.7951,
      "step": 25723
    },
    {
      "epoch": 1.4007746081013934,
      "grad_norm": 0.5849301463570367,
      "learning_rate": 4.349983953460197e-05,
      "loss": 11.8035,
      "step": 25724
    },
    {
      "epoch": 1.4008290620979764,
      "grad_norm": 0.5743438742601729,
      "learning_rate": 4.3492563899377316e-05,
      "loss": 11.8936,
      "step": 25725
    },
    {
      "epoch": 1.4008835160945594,
      "grad_norm": 0.5725940876010246,
      "learning_rate": 4.3485288703563986e-05,
      "loss": 11.9187,
      "step": 25726
    },
    {
      "epoch": 1.4009379700911424,
      "grad_norm": 0.533858575078255,
      "learning_rate": 4.347801394721851e-05,
      "loss": 11.8609,
      "step": 25727
    },
    {
      "epoch": 1.4009924240877254,
      "grad_norm": 0.5972171361554862,
      "learning_rate": 4.347073963039749e-05,
      "loss": 11.8956,
      "step": 25728
    },
    {
      "epoch": 1.4010468780843084,
      "grad_norm": 0.5431114513833376,
      "learning_rate": 4.3463465753157505e-05,
      "loss": 11.8938,
      "step": 25729
    },
    {
      "epoch": 1.4011013320808914,
      "grad_norm": 0.5755390994591522,
      "learning_rate": 4.3456192315555034e-05,
      "loss": 11.9397,
      "step": 25730
    },
    {
      "epoch": 1.4011557860774744,
      "grad_norm": 0.5730787357320373,
      "learning_rate": 4.344891931764674e-05,
      "loss": 11.8986,
      "step": 25731
    },
    {
      "epoch": 1.4012102400740574,
      "grad_norm": 0.5007471459092729,
      "learning_rate": 4.34416467594891e-05,
      "loss": 11.8538,
      "step": 25732
    },
    {
      "epoch": 1.4012646940706404,
      "grad_norm": 0.5371171533501721,
      "learning_rate": 4.343437464113869e-05,
      "loss": 11.8889,
      "step": 25733
    },
    {
      "epoch": 1.4013191480672234,
      "grad_norm": 0.5478632832461535,
      "learning_rate": 4.342710296265211e-05,
      "loss": 11.8248,
      "step": 25734
    },
    {
      "epoch": 1.4013736020638063,
      "grad_norm": 0.5631985721648314,
      "learning_rate": 4.341983172408582e-05,
      "loss": 11.875,
      "step": 25735
    },
    {
      "epoch": 1.4014280560603896,
      "grad_norm": 0.5449914502765917,
      "learning_rate": 4.341256092549646e-05,
      "loss": 11.8051,
      "step": 25736
    },
    {
      "epoch": 1.4014825100569726,
      "grad_norm": 0.5375566491885477,
      "learning_rate": 4.340529056694047e-05,
      "loss": 11.9043,
      "step": 25737
    },
    {
      "epoch": 1.4015369640535555,
      "grad_norm": 0.5365139119787555,
      "learning_rate": 4.339802064847447e-05,
      "loss": 11.9476,
      "step": 25738
    },
    {
      "epoch": 1.4015914180501385,
      "grad_norm": 0.5766863127162225,
      "learning_rate": 4.339075117015495e-05,
      "loss": 11.8478,
      "step": 25739
    },
    {
      "epoch": 1.4016458720467215,
      "grad_norm": 0.6210547963280137,
      "learning_rate": 4.338348213203841e-05,
      "loss": 11.8514,
      "step": 25740
    },
    {
      "epoch": 1.4017003260433045,
      "grad_norm": 0.5459567670103419,
      "learning_rate": 4.3376213534181445e-05,
      "loss": 11.7452,
      "step": 25741
    },
    {
      "epoch": 1.4017547800398875,
      "grad_norm": 0.5594405762582951,
      "learning_rate": 4.33689453766405e-05,
      "loss": 11.9725,
      "step": 25742
    },
    {
      "epoch": 1.4018092340364705,
      "grad_norm": 0.5454613069963479,
      "learning_rate": 4.3361677659472164e-05,
      "loss": 11.9589,
      "step": 25743
    },
    {
      "epoch": 1.4018636880330535,
      "grad_norm": 0.5481028587273373,
      "learning_rate": 4.335441038273289e-05,
      "loss": 11.8762,
      "step": 25744
    },
    {
      "epoch": 1.4019181420296367,
      "grad_norm": 0.5606037842650509,
      "learning_rate": 4.334714354647923e-05,
      "loss": 11.8584,
      "step": 25745
    },
    {
      "epoch": 1.4019725960262197,
      "grad_norm": 0.5185782378086978,
      "learning_rate": 4.33398771507677e-05,
      "loss": 11.8792,
      "step": 25746
    },
    {
      "epoch": 1.4020270500228027,
      "grad_norm": 0.5281439600207601,
      "learning_rate": 4.333261119565476e-05,
      "loss": 12.0006,
      "step": 25747
    },
    {
      "epoch": 1.4020815040193857,
      "grad_norm": 0.6277291921330076,
      "learning_rate": 4.332534568119698e-05,
      "loss": 11.8947,
      "step": 25748
    },
    {
      "epoch": 1.4021359580159687,
      "grad_norm": 0.534869803590322,
      "learning_rate": 4.33180806074508e-05,
      "loss": 11.9372,
      "step": 25749
    },
    {
      "epoch": 1.4021904120125517,
      "grad_norm": 0.5720715166334139,
      "learning_rate": 4.331081597447272e-05,
      "loss": 12.0421,
      "step": 25750
    },
    {
      "epoch": 1.4022448660091347,
      "grad_norm": 0.6700032731573184,
      "learning_rate": 4.330355178231926e-05,
      "loss": 11.8688,
      "step": 25751
    },
    {
      "epoch": 1.4022993200057177,
      "grad_norm": 0.6422941064293342,
      "learning_rate": 4.329628803104685e-05,
      "loss": 11.9984,
      "step": 25752
    },
    {
      "epoch": 1.4023537740023007,
      "grad_norm": 0.5218657818654752,
      "learning_rate": 4.3289024720712055e-05,
      "loss": 11.8448,
      "step": 25753
    },
    {
      "epoch": 1.4024082279988836,
      "grad_norm": 0.5567331107952798,
      "learning_rate": 4.3281761851371284e-05,
      "loss": 11.8074,
      "step": 25754
    },
    {
      "epoch": 1.4024626819954666,
      "grad_norm": 0.553640112261936,
      "learning_rate": 4.327449942308107e-05,
      "loss": 11.6889,
      "step": 25755
    },
    {
      "epoch": 1.4025171359920496,
      "grad_norm": 0.5626446887770098,
      "learning_rate": 4.326723743589783e-05,
      "loss": 11.9616,
      "step": 25756
    },
    {
      "epoch": 1.4025715899886326,
      "grad_norm": 0.5404421380628863,
      "learning_rate": 4.325997588987808e-05,
      "loss": 11.9401,
      "step": 25757
    },
    {
      "epoch": 1.4026260439852156,
      "grad_norm": 0.5994161744225048,
      "learning_rate": 4.325271478507833e-05,
      "loss": 11.8843,
      "step": 25758
    },
    {
      "epoch": 1.4026804979817986,
      "grad_norm": 0.507601738856093,
      "learning_rate": 4.324545412155492e-05,
      "loss": 11.8573,
      "step": 25759
    },
    {
      "epoch": 1.4027349519783818,
      "grad_norm": 0.5399060464408709,
      "learning_rate": 4.323819389936437e-05,
      "loss": 11.8914,
      "step": 25760
    },
    {
      "epoch": 1.4027894059749648,
      "grad_norm": 0.4985717580031476,
      "learning_rate": 4.3230934118563185e-05,
      "loss": 11.9473,
      "step": 25761
    },
    {
      "epoch": 1.4028438599715478,
      "grad_norm": 0.5098343717376963,
      "learning_rate": 4.322367477920773e-05,
      "loss": 11.8673,
      "step": 25762
    },
    {
      "epoch": 1.4028983139681308,
      "grad_norm": 0.5031072887679476,
      "learning_rate": 4.321641588135454e-05,
      "loss": 11.9143,
      "step": 25763
    },
    {
      "epoch": 1.4029527679647138,
      "grad_norm": 0.5128637907285493,
      "learning_rate": 4.320915742506e-05,
      "loss": 11.806,
      "step": 25764
    },
    {
      "epoch": 1.4030072219612968,
      "grad_norm": 0.5643419053060134,
      "learning_rate": 4.32018994103806e-05,
      "loss": 11.9542,
      "step": 25765
    },
    {
      "epoch": 1.4030616759578798,
      "grad_norm": 0.5496149552258993,
      "learning_rate": 4.319464183737272e-05,
      "loss": 11.9774,
      "step": 25766
    },
    {
      "epoch": 1.4031161299544628,
      "grad_norm": 0.5385685349043441,
      "learning_rate": 4.318738470609284e-05,
      "loss": 11.7526,
      "step": 25767
    },
    {
      "epoch": 1.403170583951046,
      "grad_norm": 0.5051635585643108,
      "learning_rate": 4.318012801659744e-05,
      "loss": 11.8489,
      "step": 25768
    },
    {
      "epoch": 1.403225037947629,
      "grad_norm": 0.5528945036488232,
      "learning_rate": 4.3172871768942834e-05,
      "loss": 12.0119,
      "step": 25769
    },
    {
      "epoch": 1.403279491944212,
      "grad_norm": 0.5778477693792172,
      "learning_rate": 4.3165615963185545e-05,
      "loss": 11.9029,
      "step": 25770
    },
    {
      "epoch": 1.403333945940795,
      "grad_norm": 0.5397905335023614,
      "learning_rate": 4.315836059938191e-05,
      "loss": 11.828,
      "step": 25771
    },
    {
      "epoch": 1.403388399937378,
      "grad_norm": 0.5438297685566836,
      "learning_rate": 4.31511056775884e-05,
      "loss": 11.8823,
      "step": 25772
    },
    {
      "epoch": 1.403442853933961,
      "grad_norm": 0.5701260681333727,
      "learning_rate": 4.314385119786146e-05,
      "loss": 11.8675,
      "step": 25773
    },
    {
      "epoch": 1.403497307930544,
      "grad_norm": 0.5539240777978596,
      "learning_rate": 4.313659716025744e-05,
      "loss": 11.9092,
      "step": 25774
    },
    {
      "epoch": 1.403551761927127,
      "grad_norm": 0.5696074381941687,
      "learning_rate": 4.312934356483281e-05,
      "loss": 11.845,
      "step": 25775
    },
    {
      "epoch": 1.40360621592371,
      "grad_norm": 0.6578220433315457,
      "learning_rate": 4.31220904116439e-05,
      "loss": 12.0892,
      "step": 25776
    },
    {
      "epoch": 1.403660669920293,
      "grad_norm": 0.5337505459524566,
      "learning_rate": 4.3114837700747203e-05,
      "loss": 11.8102,
      "step": 25777
    },
    {
      "epoch": 1.403715123916876,
      "grad_norm": 0.5252660110524694,
      "learning_rate": 4.310758543219905e-05,
      "loss": 11.8788,
      "step": 25778
    },
    {
      "epoch": 1.403769577913459,
      "grad_norm": 0.5426806737980335,
      "learning_rate": 4.310033360605582e-05,
      "loss": 11.8705,
      "step": 25779
    },
    {
      "epoch": 1.403824031910042,
      "grad_norm": 0.5849980006069722,
      "learning_rate": 4.309308222237398e-05,
      "loss": 11.8774,
      "step": 25780
    },
    {
      "epoch": 1.403878485906625,
      "grad_norm": 0.5298430039427744,
      "learning_rate": 4.308583128120984e-05,
      "loss": 11.9699,
      "step": 25781
    },
    {
      "epoch": 1.4039329399032079,
      "grad_norm": 0.5711868513615855,
      "learning_rate": 4.307858078261983e-05,
      "loss": 11.8563,
      "step": 25782
    },
    {
      "epoch": 1.403987393899791,
      "grad_norm": 0.6034781661497142,
      "learning_rate": 4.307133072666034e-05,
      "loss": 11.9382,
      "step": 25783
    },
    {
      "epoch": 1.404041847896374,
      "grad_norm": 0.5275655011643641,
      "learning_rate": 4.30640811133877e-05,
      "loss": 11.883,
      "step": 25784
    },
    {
      "epoch": 1.404096301892957,
      "grad_norm": 0.6132439151156021,
      "learning_rate": 4.305683194285834e-05,
      "loss": 11.9922,
      "step": 25785
    },
    {
      "epoch": 1.40415075588954,
      "grad_norm": 0.4809888890409843,
      "learning_rate": 4.304958321512858e-05,
      "loss": 11.8524,
      "step": 25786
    },
    {
      "epoch": 1.404205209886123,
      "grad_norm": 0.5915868691239101,
      "learning_rate": 4.3042334930254855e-05,
      "loss": 11.9001,
      "step": 25787
    },
    {
      "epoch": 1.404259663882706,
      "grad_norm": 0.4780354993578049,
      "learning_rate": 4.303508708829347e-05,
      "loss": 11.8547,
      "step": 25788
    },
    {
      "epoch": 1.404314117879289,
      "grad_norm": 0.5796849881484587,
      "learning_rate": 4.302783968930078e-05,
      "loss": 11.8499,
      "step": 25789
    },
    {
      "epoch": 1.404368571875872,
      "grad_norm": 0.5537851104142394,
      "learning_rate": 4.3020592733333185e-05,
      "loss": 11.8688,
      "step": 25790
    },
    {
      "epoch": 1.404423025872455,
      "grad_norm": 0.5442721735305788,
      "learning_rate": 4.301334622044698e-05,
      "loss": 11.8563,
      "step": 25791
    },
    {
      "epoch": 1.4044774798690383,
      "grad_norm": 0.5955860010807744,
      "learning_rate": 4.3006100150698594e-05,
      "loss": 11.9449,
      "step": 25792
    },
    {
      "epoch": 1.4045319338656213,
      "grad_norm": 0.5395442100716481,
      "learning_rate": 4.299885452414429e-05,
      "loss": 11.7314,
      "step": 25793
    },
    {
      "epoch": 1.4045863878622042,
      "grad_norm": 0.5194883904364771,
      "learning_rate": 4.2991609340840455e-05,
      "loss": 11.916,
      "step": 25794
    },
    {
      "epoch": 1.4046408418587872,
      "grad_norm": 0.5643779043933354,
      "learning_rate": 4.298436460084346e-05,
      "loss": 11.814,
      "step": 25795
    },
    {
      "epoch": 1.4046952958553702,
      "grad_norm": 0.6175041382935421,
      "learning_rate": 4.2977120304209586e-05,
      "loss": 11.7237,
      "step": 25796
    },
    {
      "epoch": 1.4047497498519532,
      "grad_norm": 0.5257130784124201,
      "learning_rate": 4.29698764509952e-05,
      "loss": 11.9171,
      "step": 25797
    },
    {
      "epoch": 1.4048042038485362,
      "grad_norm": 0.5466752839396776,
      "learning_rate": 4.296263304125664e-05,
      "loss": 11.8934,
      "step": 25798
    },
    {
      "epoch": 1.4048586578451192,
      "grad_norm": 0.5403470529263875,
      "learning_rate": 4.295539007505016e-05,
      "loss": 11.9722,
      "step": 25799
    },
    {
      "epoch": 1.4049131118417022,
      "grad_norm": 0.5531022681001917,
      "learning_rate": 4.294814755243217e-05,
      "loss": 11.9278,
      "step": 25800
    },
    {
      "epoch": 1.4049675658382852,
      "grad_norm": 0.5067067110711966,
      "learning_rate": 4.2940905473458926e-05,
      "loss": 11.8181,
      "step": 25801
    },
    {
      "epoch": 1.4050220198348682,
      "grad_norm": 0.5813067824610553,
      "learning_rate": 4.29336638381868e-05,
      "loss": 11.9637,
      "step": 25802
    },
    {
      "epoch": 1.4050764738314512,
      "grad_norm": 0.5669991039890107,
      "learning_rate": 4.2926422646672036e-05,
      "loss": 11.8125,
      "step": 25803
    },
    {
      "epoch": 1.4051309278280342,
      "grad_norm": 0.49300588130870826,
      "learning_rate": 4.2919181898970984e-05,
      "loss": 11.8649,
      "step": 25804
    },
    {
      "epoch": 1.4051853818246172,
      "grad_norm": 0.5076407580270612,
      "learning_rate": 4.291194159513998e-05,
      "loss": 11.8819,
      "step": 25805
    },
    {
      "epoch": 1.4052398358212004,
      "grad_norm": 0.5236553023447134,
      "learning_rate": 4.2904701735235256e-05,
      "loss": 11.7725,
      "step": 25806
    },
    {
      "epoch": 1.4052942898177834,
      "grad_norm": 0.5537238593380521,
      "learning_rate": 4.289746231931319e-05,
      "loss": 11.8104,
      "step": 25807
    },
    {
      "epoch": 1.4053487438143664,
      "grad_norm": 0.6155129506674,
      "learning_rate": 4.289022334743001e-05,
      "loss": 11.8991,
      "step": 25808
    },
    {
      "epoch": 1.4054031978109494,
      "grad_norm": 0.5394076255226601,
      "learning_rate": 4.2882984819642015e-05,
      "loss": 11.9354,
      "step": 25809
    },
    {
      "epoch": 1.4054576518075323,
      "grad_norm": 0.5782024250137687,
      "learning_rate": 4.287574673600554e-05,
      "loss": 11.8426,
      "step": 25810
    },
    {
      "epoch": 1.4055121058041153,
      "grad_norm": 0.5313240044500805,
      "learning_rate": 4.28685090965768e-05,
      "loss": 11.8289,
      "step": 25811
    },
    {
      "epoch": 1.4055665598006983,
      "grad_norm": 0.5398556323000993,
      "learning_rate": 4.286127190141216e-05,
      "loss": 11.9861,
      "step": 25812
    },
    {
      "epoch": 1.4056210137972813,
      "grad_norm": 0.5323480588440391,
      "learning_rate": 4.285403515056781e-05,
      "loss": 11.9367,
      "step": 25813
    },
    {
      "epoch": 1.4056754677938643,
      "grad_norm": 0.5714705974394412,
      "learning_rate": 4.28467988441001e-05,
      "loss": 11.8651,
      "step": 25814
    },
    {
      "epoch": 1.4057299217904475,
      "grad_norm": 0.5258662331210137,
      "learning_rate": 4.283956298206523e-05,
      "loss": 12.0163,
      "step": 25815
    },
    {
      "epoch": 1.4057843757870305,
      "grad_norm": 0.5249059355857912,
      "learning_rate": 4.28323275645195e-05,
      "loss": 11.933,
      "step": 25816
    },
    {
      "epoch": 1.4058388297836135,
      "grad_norm": 0.5088191200406901,
      "learning_rate": 4.282509259151922e-05,
      "loss": 11.9727,
      "step": 25817
    },
    {
      "epoch": 1.4058932837801965,
      "grad_norm": 0.6366457523019411,
      "learning_rate": 4.281785806312061e-05,
      "loss": 11.6813,
      "step": 25818
    },
    {
      "epoch": 1.4059477377767795,
      "grad_norm": 0.5581525711871052,
      "learning_rate": 4.281062397937992e-05,
      "loss": 11.836,
      "step": 25819
    },
    {
      "epoch": 1.4060021917733625,
      "grad_norm": 0.6347471647572209,
      "learning_rate": 4.280339034035338e-05,
      "loss": 11.9899,
      "step": 25820
    },
    {
      "epoch": 1.4060566457699455,
      "grad_norm": 0.5325272664915921,
      "learning_rate": 4.279615714609726e-05,
      "loss": 11.829,
      "step": 25821
    },
    {
      "epoch": 1.4061110997665285,
      "grad_norm": 0.48389713357782344,
      "learning_rate": 4.278892439666785e-05,
      "loss": 11.8101,
      "step": 25822
    },
    {
      "epoch": 1.4061655537631115,
      "grad_norm": 0.5360783902279086,
      "learning_rate": 4.278169209212133e-05,
      "loss": 11.8839,
      "step": 25823
    },
    {
      "epoch": 1.4062200077596945,
      "grad_norm": 0.5481278820150643,
      "learning_rate": 4.2774460232514e-05,
      "loss": 11.8488,
      "step": 25824
    },
    {
      "epoch": 1.4062744617562775,
      "grad_norm": 0.5720749791879448,
      "learning_rate": 4.276722881790203e-05,
      "loss": 11.8759,
      "step": 25825
    },
    {
      "epoch": 1.4063289157528605,
      "grad_norm": 0.5443309638373263,
      "learning_rate": 4.275999784834172e-05,
      "loss": 11.7165,
      "step": 25826
    },
    {
      "epoch": 1.4063833697494434,
      "grad_norm": 0.5959614687964457,
      "learning_rate": 4.275276732388923e-05,
      "loss": 11.933,
      "step": 25827
    },
    {
      "epoch": 1.4064378237460264,
      "grad_norm": 0.5179583093528469,
      "learning_rate": 4.274553724460086e-05,
      "loss": 11.8332,
      "step": 25828
    },
    {
      "epoch": 1.4064922777426097,
      "grad_norm": 0.5274418774361285,
      "learning_rate": 4.273830761053279e-05,
      "loss": 11.8372,
      "step": 25829
    },
    {
      "epoch": 1.4065467317391926,
      "grad_norm": 0.5491854188103588,
      "learning_rate": 4.273107842174121e-05,
      "loss": 11.862,
      "step": 25830
    },
    {
      "epoch": 1.4066011857357756,
      "grad_norm": 0.560334495452463,
      "learning_rate": 4.272384967828237e-05,
      "loss": 11.8708,
      "step": 25831
    },
    {
      "epoch": 1.4066556397323586,
      "grad_norm": 0.4878984289510161,
      "learning_rate": 4.2716621380212505e-05,
      "loss": 11.8301,
      "step": 25832
    },
    {
      "epoch": 1.4067100937289416,
      "grad_norm": 0.5708076057091399,
      "learning_rate": 4.270939352758778e-05,
      "loss": 11.7839,
      "step": 25833
    },
    {
      "epoch": 1.4067645477255246,
      "grad_norm": 0.5787300159516624,
      "learning_rate": 4.270216612046445e-05,
      "loss": 11.909,
      "step": 25834
    },
    {
      "epoch": 1.4068190017221076,
      "grad_norm": 0.5941794725826334,
      "learning_rate": 4.269493915889864e-05,
      "loss": 12.0,
      "step": 25835
    },
    {
      "epoch": 1.4068734557186906,
      "grad_norm": 0.5648772748489747,
      "learning_rate": 4.2687712642946634e-05,
      "loss": 11.8915,
      "step": 25836
    },
    {
      "epoch": 1.4069279097152736,
      "grad_norm": 0.5066468622749574,
      "learning_rate": 4.268048657266455e-05,
      "loss": 11.8217,
      "step": 25837
    },
    {
      "epoch": 1.4069823637118568,
      "grad_norm": 0.613763671049065,
      "learning_rate": 4.267326094810865e-05,
      "loss": 12.0097,
      "step": 25838
    },
    {
      "epoch": 1.4070368177084398,
      "grad_norm": 0.5151996896175168,
      "learning_rate": 4.2666035769335086e-05,
      "loss": 11.8486,
      "step": 25839
    },
    {
      "epoch": 1.4070912717050228,
      "grad_norm": 0.5229846377104254,
      "learning_rate": 4.265881103640002e-05,
      "loss": 11.9894,
      "step": 25840
    },
    {
      "epoch": 1.4071457257016058,
      "grad_norm": 0.5552971388629591,
      "learning_rate": 4.265158674935967e-05,
      "loss": 11.8026,
      "step": 25841
    },
    {
      "epoch": 1.4072001796981888,
      "grad_norm": 0.5820783232430011,
      "learning_rate": 4.2644362908270174e-05,
      "loss": 11.8772,
      "step": 25842
    },
    {
      "epoch": 1.4072546336947718,
      "grad_norm": 0.5743819396700995,
      "learning_rate": 4.263713951318773e-05,
      "loss": 11.8406,
      "step": 25843
    },
    {
      "epoch": 1.4073090876913548,
      "grad_norm": 0.5078189515427319,
      "learning_rate": 4.262991656416854e-05,
      "loss": 11.6802,
      "step": 25844
    },
    {
      "epoch": 1.4073635416879378,
      "grad_norm": 0.5683440894347913,
      "learning_rate": 4.262269406126871e-05,
      "loss": 11.8316,
      "step": 25845
    },
    {
      "epoch": 1.4074179956845208,
      "grad_norm": 0.5744332619359597,
      "learning_rate": 4.261547200454446e-05,
      "loss": 11.9266,
      "step": 25846
    },
    {
      "epoch": 1.4074724496811037,
      "grad_norm": 0.5612854370753293,
      "learning_rate": 4.26082503940519e-05,
      "loss": 11.8786,
      "step": 25847
    },
    {
      "epoch": 1.4075269036776867,
      "grad_norm": 0.5525059653758136,
      "learning_rate": 4.2601029229847235e-05,
      "loss": 11.777,
      "step": 25848
    },
    {
      "epoch": 1.4075813576742697,
      "grad_norm": 0.6006601390818188,
      "learning_rate": 4.25938085119866e-05,
      "loss": 11.9093,
      "step": 25849
    },
    {
      "epoch": 1.4076358116708527,
      "grad_norm": 0.5156693231260574,
      "learning_rate": 4.2586588240526095e-05,
      "loss": 11.9375,
      "step": 25850
    },
    {
      "epoch": 1.4076902656674357,
      "grad_norm": 0.5669870649255772,
      "learning_rate": 4.2579368415521946e-05,
      "loss": 11.9091,
      "step": 25851
    },
    {
      "epoch": 1.4077447196640187,
      "grad_norm": 0.5197345446943886,
      "learning_rate": 4.2572149037030226e-05,
      "loss": 11.839,
      "step": 25852
    },
    {
      "epoch": 1.407799173660602,
      "grad_norm": 0.5070070801694782,
      "learning_rate": 4.2564930105107104e-05,
      "loss": 11.9568,
      "step": 25853
    },
    {
      "epoch": 1.407853627657185,
      "grad_norm": 0.5339323222051076,
      "learning_rate": 4.255771161980875e-05,
      "loss": 11.812,
      "step": 25854
    },
    {
      "epoch": 1.407908081653768,
      "grad_norm": 0.5444994776815661,
      "learning_rate": 4.255049358119123e-05,
      "loss": 11.913,
      "step": 25855
    },
    {
      "epoch": 1.407962535650351,
      "grad_norm": 0.5496574992041923,
      "learning_rate": 4.254327598931074e-05,
      "loss": 11.9443,
      "step": 25856
    },
    {
      "epoch": 1.408016989646934,
      "grad_norm": 0.5914788405976005,
      "learning_rate": 4.253605884422334e-05,
      "loss": 11.8381,
      "step": 25857
    },
    {
      "epoch": 1.4080714436435169,
      "grad_norm": 0.5455994167009618,
      "learning_rate": 4.2528842145985214e-05,
      "loss": 11.8733,
      "step": 25858
    },
    {
      "epoch": 1.4081258976400999,
      "grad_norm": 0.5345820288709208,
      "learning_rate": 4.252162589465245e-05,
      "loss": 12.0014,
      "step": 25859
    },
    {
      "epoch": 1.4081803516366829,
      "grad_norm": 0.5798557355366813,
      "learning_rate": 4.251441009028113e-05,
      "loss": 11.958,
      "step": 25860
    },
    {
      "epoch": 1.408234805633266,
      "grad_norm": 0.5207074577882463,
      "learning_rate": 4.2507194732927433e-05,
      "loss": 11.8906,
      "step": 25861
    },
    {
      "epoch": 1.408289259629849,
      "grad_norm": 0.5041419379473353,
      "learning_rate": 4.2499979822647396e-05,
      "loss": 11.8203,
      "step": 25862
    },
    {
      "epoch": 1.408343713626432,
      "grad_norm": 0.5565637241578298,
      "learning_rate": 4.2492765359497187e-05,
      "loss": 11.7416,
      "step": 25863
    },
    {
      "epoch": 1.408398167623015,
      "grad_norm": 0.5591302357801531,
      "learning_rate": 4.248555134353285e-05,
      "loss": 12.0768,
      "step": 25864
    },
    {
      "epoch": 1.408452621619598,
      "grad_norm": 0.5502567109883069,
      "learning_rate": 4.2478337774810504e-05,
      "loss": 11.9551,
      "step": 25865
    },
    {
      "epoch": 1.408507075616181,
      "grad_norm": 0.5295241005744646,
      "learning_rate": 4.2471124653386294e-05,
      "loss": 11.8909,
      "step": 25866
    },
    {
      "epoch": 1.408561529612764,
      "grad_norm": 0.5834154258213458,
      "learning_rate": 4.2463911979316226e-05,
      "loss": 11.6616,
      "step": 25867
    },
    {
      "epoch": 1.408615983609347,
      "grad_norm": 0.5363876180307716,
      "learning_rate": 4.245669975265646e-05,
      "loss": 11.9225,
      "step": 25868
    },
    {
      "epoch": 1.40867043760593,
      "grad_norm": 0.5021256883576487,
      "learning_rate": 4.2449487973463055e-05,
      "loss": 11.9048,
      "step": 25869
    },
    {
      "epoch": 1.408724891602513,
      "grad_norm": 0.5526527474962228,
      "learning_rate": 4.2442276641792046e-05,
      "loss": 11.9494,
      "step": 25870
    },
    {
      "epoch": 1.408779345599096,
      "grad_norm": 0.5518407773921247,
      "learning_rate": 4.2435065757699575e-05,
      "loss": 11.8001,
      "step": 25871
    },
    {
      "epoch": 1.408833799595679,
      "grad_norm": 0.535865054136345,
      "learning_rate": 4.242785532124165e-05,
      "loss": 11.7701,
      "step": 25872
    },
    {
      "epoch": 1.408888253592262,
      "grad_norm": 0.5547459065259486,
      "learning_rate": 4.242064533247442e-05,
      "loss": 11.9321,
      "step": 25873
    },
    {
      "epoch": 1.408942707588845,
      "grad_norm": 0.5226768792334109,
      "learning_rate": 4.241343579145387e-05,
      "loss": 11.8714,
      "step": 25874
    },
    {
      "epoch": 1.408997161585428,
      "grad_norm": 0.5393516411682427,
      "learning_rate": 4.240622669823614e-05,
      "loss": 11.9472,
      "step": 25875
    },
    {
      "epoch": 1.4090516155820112,
      "grad_norm": 0.49869323998245224,
      "learning_rate": 4.2399018052877205e-05,
      "loss": 11.8296,
      "step": 25876
    },
    {
      "epoch": 1.4091060695785942,
      "grad_norm": 0.5525337755186532,
      "learning_rate": 4.2391809855433164e-05,
      "loss": 11.8329,
      "step": 25877
    },
    {
      "epoch": 1.4091605235751772,
      "grad_norm": 0.5040937477148583,
      "learning_rate": 4.2384602105960115e-05,
      "loss": 11.9391,
      "step": 25878
    },
    {
      "epoch": 1.4092149775717602,
      "grad_norm": 0.5958430697037971,
      "learning_rate": 4.237739480451406e-05,
      "loss": 12.0315,
      "step": 25879
    },
    {
      "epoch": 1.4092694315683432,
      "grad_norm": 0.5141700211675545,
      "learning_rate": 4.237018795115101e-05,
      "loss": 11.8403,
      "step": 25880
    },
    {
      "epoch": 1.4093238855649262,
      "grad_norm": 0.5118231921700256,
      "learning_rate": 4.236298154592707e-05,
      "loss": 11.9423,
      "step": 25881
    },
    {
      "epoch": 1.4093783395615092,
      "grad_norm": 0.5595852370792733,
      "learning_rate": 4.235577558889823e-05,
      "loss": 11.9391,
      "step": 25882
    },
    {
      "epoch": 1.4094327935580921,
      "grad_norm": 0.5730760624709044,
      "learning_rate": 4.2348570080120576e-05,
      "loss": 11.8369,
      "step": 25883
    },
    {
      "epoch": 1.4094872475546751,
      "grad_norm": 0.5328544162840606,
      "learning_rate": 4.234136501965007e-05,
      "loss": 11.8821,
      "step": 25884
    },
    {
      "epoch": 1.4095417015512584,
      "grad_norm": 0.49246478701004204,
      "learning_rate": 4.2334160407542824e-05,
      "loss": 11.9199,
      "step": 25885
    },
    {
      "epoch": 1.4095961555478413,
      "grad_norm": 0.598958754766261,
      "learning_rate": 4.232695624385478e-05,
      "loss": 11.7857,
      "step": 25886
    },
    {
      "epoch": 1.4096506095444243,
      "grad_norm": 0.522251487365959,
      "learning_rate": 4.2319752528641986e-05,
      "loss": 11.8622,
      "step": 25887
    },
    {
      "epoch": 1.4097050635410073,
      "grad_norm": 0.5077513731704751,
      "learning_rate": 4.231254926196051e-05,
      "loss": 11.8541,
      "step": 25888
    },
    {
      "epoch": 1.4097595175375903,
      "grad_norm": 0.5783091568956663,
      "learning_rate": 4.230534644386632e-05,
      "loss": 11.8103,
      "step": 25889
    },
    {
      "epoch": 1.4098139715341733,
      "grad_norm": 0.5622060053448952,
      "learning_rate": 4.229814407441544e-05,
      "loss": 11.8969,
      "step": 25890
    },
    {
      "epoch": 1.4098684255307563,
      "grad_norm": 0.6072125634276111,
      "learning_rate": 4.2290942153663825e-05,
      "loss": 11.9061,
      "step": 25891
    },
    {
      "epoch": 1.4099228795273393,
      "grad_norm": 0.6973086392010021,
      "learning_rate": 4.228374068166753e-05,
      "loss": 11.8754,
      "step": 25892
    },
    {
      "epoch": 1.4099773335239223,
      "grad_norm": 0.513304898126486,
      "learning_rate": 4.2276539658482574e-05,
      "loss": 11.8946,
      "step": 25893
    },
    {
      "epoch": 1.4100317875205053,
      "grad_norm": 0.5527366824529903,
      "learning_rate": 4.226933908416489e-05,
      "loss": 11.7546,
      "step": 25894
    },
    {
      "epoch": 1.4100862415170883,
      "grad_norm": 0.5491514441669649,
      "learning_rate": 4.226213895877056e-05,
      "loss": 11.9177,
      "step": 25895
    },
    {
      "epoch": 1.4101406955136713,
      "grad_norm": 0.550730264638003,
      "learning_rate": 4.2254939282355466e-05,
      "loss": 11.8666,
      "step": 25896
    },
    {
      "epoch": 1.4101951495102543,
      "grad_norm": 0.5456382860175364,
      "learning_rate": 4.224774005497568e-05,
      "loss": 11.753,
      "step": 25897
    },
    {
      "epoch": 1.4102496035068373,
      "grad_norm": 0.5516554284156222,
      "learning_rate": 4.224054127668713e-05,
      "loss": 11.8482,
      "step": 25898
    },
    {
      "epoch": 1.4103040575034205,
      "grad_norm": 0.5625259871003233,
      "learning_rate": 4.223334294754585e-05,
      "loss": 11.904,
      "step": 25899
    },
    {
      "epoch": 1.4103585115000035,
      "grad_norm": 0.5618478905726333,
      "learning_rate": 4.222614506760778e-05,
      "loss": 11.9644,
      "step": 25900
    },
    {
      "epoch": 1.4104129654965865,
      "grad_norm": 0.6460636520761004,
      "learning_rate": 4.221894763692886e-05,
      "loss": 12.0559,
      "step": 25901
    },
    {
      "epoch": 1.4104674194931694,
      "grad_norm": 0.5273121768935696,
      "learning_rate": 4.22117506555651e-05,
      "loss": 11.9369,
      "step": 25902
    },
    {
      "epoch": 1.4105218734897524,
      "grad_norm": 0.5162378960358935,
      "learning_rate": 4.2204554123572485e-05,
      "loss": 11.9836,
      "step": 25903
    },
    {
      "epoch": 1.4105763274863354,
      "grad_norm": 0.5196327394712076,
      "learning_rate": 4.219735804100691e-05,
      "loss": 11.7405,
      "step": 25904
    },
    {
      "epoch": 1.4106307814829184,
      "grad_norm": 0.5254282810437462,
      "learning_rate": 4.219016240792442e-05,
      "loss": 11.9094,
      "step": 25905
    },
    {
      "epoch": 1.4106852354795014,
      "grad_norm": 0.5216154572046777,
      "learning_rate": 4.218296722438087e-05,
      "loss": 11.8347,
      "step": 25906
    },
    {
      "epoch": 1.4107396894760844,
      "grad_norm": 0.541167348260566,
      "learning_rate": 4.217577249043231e-05,
      "loss": 11.9124,
      "step": 25907
    },
    {
      "epoch": 1.4107941434726676,
      "grad_norm": 0.5565774346147542,
      "learning_rate": 4.21685782061346e-05,
      "loss": 11.8609,
      "step": 25908
    },
    {
      "epoch": 1.4108485974692506,
      "grad_norm": 0.51847801428641,
      "learning_rate": 4.216138437154377e-05,
      "loss": 11.862,
      "step": 25909
    },
    {
      "epoch": 1.4109030514658336,
      "grad_norm": 0.5120099100789547,
      "learning_rate": 4.2154190986715704e-05,
      "loss": 11.8856,
      "step": 25910
    },
    {
      "epoch": 1.4109575054624166,
      "grad_norm": 0.5283512903397058,
      "learning_rate": 4.214699805170632e-05,
      "loss": 11.864,
      "step": 25911
    },
    {
      "epoch": 1.4110119594589996,
      "grad_norm": 0.5636639989040957,
      "learning_rate": 4.213980556657161e-05,
      "loss": 11.9639,
      "step": 25912
    },
    {
      "epoch": 1.4110664134555826,
      "grad_norm": 0.6077813526368897,
      "learning_rate": 4.213261353136745e-05,
      "loss": 11.934,
      "step": 25913
    },
    {
      "epoch": 1.4111208674521656,
      "grad_norm": 0.5086606259692638,
      "learning_rate": 4.212542194614978e-05,
      "loss": 11.7002,
      "step": 25914
    },
    {
      "epoch": 1.4111753214487486,
      "grad_norm": 0.5502609041887185,
      "learning_rate": 4.211823081097458e-05,
      "loss": 11.8365,
      "step": 25915
    },
    {
      "epoch": 1.4112297754453316,
      "grad_norm": 0.5273037271101029,
      "learning_rate": 4.2111040125897685e-05,
      "loss": 11.8961,
      "step": 25916
    },
    {
      "epoch": 1.4112842294419146,
      "grad_norm": 0.5133770616820434,
      "learning_rate": 4.210384989097508e-05,
      "loss": 11.8726,
      "step": 25917
    },
    {
      "epoch": 1.4113386834384976,
      "grad_norm": 0.5029753219698895,
      "learning_rate": 4.209666010626262e-05,
      "loss": 11.79,
      "step": 25918
    },
    {
      "epoch": 1.4113931374350805,
      "grad_norm": 0.4950874803758819,
      "learning_rate": 4.208947077181626e-05,
      "loss": 11.8046,
      "step": 25919
    },
    {
      "epoch": 1.4114475914316635,
      "grad_norm": 0.6203688283204638,
      "learning_rate": 4.208228188769191e-05,
      "loss": 11.9855,
      "step": 25920
    },
    {
      "epoch": 1.4115020454282465,
      "grad_norm": 0.5162160457790488,
      "learning_rate": 4.20750934539454e-05,
      "loss": 11.8382,
      "step": 25921
    },
    {
      "epoch": 1.4115564994248295,
      "grad_norm": 0.5355800112870308,
      "learning_rate": 4.206790547063272e-05,
      "loss": 11.8719,
      "step": 25922
    },
    {
      "epoch": 1.4116109534214127,
      "grad_norm": 0.6020767351640445,
      "learning_rate": 4.206071793780968e-05,
      "loss": 11.9148,
      "step": 25923
    },
    {
      "epoch": 1.4116654074179957,
      "grad_norm": 0.5947597914235936,
      "learning_rate": 4.2053530855532254e-05,
      "loss": 11.9118,
      "step": 25924
    },
    {
      "epoch": 1.4117198614145787,
      "grad_norm": 0.5909367902531036,
      "learning_rate": 4.204634422385625e-05,
      "loss": 11.84,
      "step": 25925
    },
    {
      "epoch": 1.4117743154111617,
      "grad_norm": 0.5570109662277771,
      "learning_rate": 4.203915804283759e-05,
      "loss": 11.8911,
      "step": 25926
    },
    {
      "epoch": 1.4118287694077447,
      "grad_norm": 0.5541344179528458,
      "learning_rate": 4.2031972312532196e-05,
      "loss": 11.8311,
      "step": 25927
    },
    {
      "epoch": 1.4118832234043277,
      "grad_norm": 0.5637549961939454,
      "learning_rate": 4.202478703299587e-05,
      "loss": 11.8737,
      "step": 25928
    },
    {
      "epoch": 1.4119376774009107,
      "grad_norm": 0.5726764453654672,
      "learning_rate": 4.201760220428457e-05,
      "loss": 11.9924,
      "step": 25929
    },
    {
      "epoch": 1.4119921313974937,
      "grad_norm": 0.5894911664127516,
      "learning_rate": 4.2010417826454106e-05,
      "loss": 11.8918,
      "step": 25930
    },
    {
      "epoch": 1.412046585394077,
      "grad_norm": 0.6004226055445582,
      "learning_rate": 4.200323389956032e-05,
      "loss": 11.8862,
      "step": 25931
    },
    {
      "epoch": 1.41210103939066,
      "grad_norm": 0.5289710695479329,
      "learning_rate": 4.199605042365916e-05,
      "loss": 11.7657,
      "step": 25932
    },
    {
      "epoch": 1.4121554933872429,
      "grad_norm": 0.5841983179479884,
      "learning_rate": 4.19888673988064e-05,
      "loss": 11.9163,
      "step": 25933
    },
    {
      "epoch": 1.4122099473838259,
      "grad_norm": 0.5469696232190195,
      "learning_rate": 4.1981684825057966e-05,
      "loss": 11.9362,
      "step": 25934
    },
    {
      "epoch": 1.4122644013804089,
      "grad_norm": 0.5852481285258428,
      "learning_rate": 4.197450270246964e-05,
      "loss": 11.9623,
      "step": 25935
    },
    {
      "epoch": 1.4123188553769919,
      "grad_norm": 0.5803512784379835,
      "learning_rate": 4.196732103109733e-05,
      "loss": 11.9014,
      "step": 25936
    },
    {
      "epoch": 1.4123733093735749,
      "grad_norm": 0.5560973915181886,
      "learning_rate": 4.196013981099689e-05,
      "loss": 11.9202,
      "step": 25937
    },
    {
      "epoch": 1.4124277633701579,
      "grad_norm": 0.5695512447169855,
      "learning_rate": 4.1952959042224095e-05,
      "loss": 11.9226,
      "step": 25938
    },
    {
      "epoch": 1.4124822173667408,
      "grad_norm": 0.555477549408239,
      "learning_rate": 4.1945778724834906e-05,
      "loss": 11.9163,
      "step": 25939
    },
    {
      "epoch": 1.4125366713633238,
      "grad_norm": 0.5212130144156074,
      "learning_rate": 4.193859885888499e-05,
      "loss": 11.9148,
      "step": 25940
    },
    {
      "epoch": 1.4125911253599068,
      "grad_norm": 0.5884439021591624,
      "learning_rate": 4.1931419444430274e-05,
      "loss": 11.9955,
      "step": 25941
    },
    {
      "epoch": 1.4126455793564898,
      "grad_norm": 0.5246729502820188,
      "learning_rate": 4.192424048152662e-05,
      "loss": 11.9317,
      "step": 25942
    },
    {
      "epoch": 1.4127000333530728,
      "grad_norm": 0.5588712101795598,
      "learning_rate": 4.1917061970229775e-05,
      "loss": 11.8533,
      "step": 25943
    },
    {
      "epoch": 1.4127544873496558,
      "grad_norm": 0.5507920691027859,
      "learning_rate": 4.1909883910595625e-05,
      "loss": 11.9232,
      "step": 25944
    },
    {
      "epoch": 1.4128089413462388,
      "grad_norm": 0.6211357304721636,
      "learning_rate": 4.190270630267994e-05,
      "loss": 11.8887,
      "step": 25945
    },
    {
      "epoch": 1.412863395342822,
      "grad_norm": 0.5255692057499124,
      "learning_rate": 4.189552914653857e-05,
      "loss": 11.8578,
      "step": 25946
    },
    {
      "epoch": 1.412917849339405,
      "grad_norm": 0.5233315789125356,
      "learning_rate": 4.188835244222729e-05,
      "loss": 12.0373,
      "step": 25947
    },
    {
      "epoch": 1.412972303335988,
      "grad_norm": 0.6191500565628775,
      "learning_rate": 4.188117618980192e-05,
      "loss": 11.9274,
      "step": 25948
    },
    {
      "epoch": 1.413026757332571,
      "grad_norm": 0.5176264051522861,
      "learning_rate": 4.187400038931834e-05,
      "loss": 11.8133,
      "step": 25949
    },
    {
      "epoch": 1.413081211329154,
      "grad_norm": 0.5517724031078244,
      "learning_rate": 4.1866825040832205e-05,
      "loss": 11.953,
      "step": 25950
    },
    {
      "epoch": 1.413135665325737,
      "grad_norm": 0.5731983108006287,
      "learning_rate": 4.185965014439939e-05,
      "loss": 11.9859,
      "step": 25951
    },
    {
      "epoch": 1.41319011932232,
      "grad_norm": 0.5820963108441654,
      "learning_rate": 4.1852475700075724e-05,
      "loss": 11.7734,
      "step": 25952
    },
    {
      "epoch": 1.413244573318903,
      "grad_norm": 0.5010189682805626,
      "learning_rate": 4.184530170791692e-05,
      "loss": 11.8814,
      "step": 25953
    },
    {
      "epoch": 1.413299027315486,
      "grad_norm": 0.5401379374152437,
      "learning_rate": 4.183812816797884e-05,
      "loss": 11.9156,
      "step": 25954
    },
    {
      "epoch": 1.4133534813120692,
      "grad_norm": 0.5478793243519098,
      "learning_rate": 4.183095508031719e-05,
      "loss": 11.7476,
      "step": 25955
    },
    {
      "epoch": 1.4134079353086522,
      "grad_norm": 0.5464283418929661,
      "learning_rate": 4.182378244498784e-05,
      "loss": 11.9163,
      "step": 25956
    },
    {
      "epoch": 1.4134623893052352,
      "grad_norm": 0.4842141536250966,
      "learning_rate": 4.1816610262046464e-05,
      "loss": 11.7685,
      "step": 25957
    },
    {
      "epoch": 1.4135168433018181,
      "grad_norm": 0.5477737922724731,
      "learning_rate": 4.180943853154888e-05,
      "loss": 12.0035,
      "step": 25958
    },
    {
      "epoch": 1.4135712972984011,
      "grad_norm": 0.538179849185479,
      "learning_rate": 4.1802267253550945e-05,
      "loss": 11.8145,
      "step": 25959
    },
    {
      "epoch": 1.4136257512949841,
      "grad_norm": 0.5399542451935301,
      "learning_rate": 4.179509642810826e-05,
      "loss": 11.7071,
      "step": 25960
    },
    {
      "epoch": 1.4136802052915671,
      "grad_norm": 0.5188807013594945,
      "learning_rate": 4.178792605527669e-05,
      "loss": 11.9183,
      "step": 25961
    },
    {
      "epoch": 1.4137346592881501,
      "grad_norm": 0.5547543279310141,
      "learning_rate": 4.178075613511195e-05,
      "loss": 11.8354,
      "step": 25962
    },
    {
      "epoch": 1.4137891132847331,
      "grad_norm": 0.5450098727823062,
      "learning_rate": 4.1773586667669815e-05,
      "loss": 11.9785,
      "step": 25963
    },
    {
      "epoch": 1.413843567281316,
      "grad_norm": 0.5382920333937433,
      "learning_rate": 4.1766417653006065e-05,
      "loss": 11.7307,
      "step": 25964
    },
    {
      "epoch": 1.413898021277899,
      "grad_norm": 0.5608936529144182,
      "learning_rate": 4.175924909117638e-05,
      "loss": 11.9767,
      "step": 25965
    },
    {
      "epoch": 1.413952475274482,
      "grad_norm": 0.5389868238399339,
      "learning_rate": 4.175208098223657e-05,
      "loss": 11.7917,
      "step": 25966
    },
    {
      "epoch": 1.414006929271065,
      "grad_norm": 0.5091692630679351,
      "learning_rate": 4.174491332624233e-05,
      "loss": 11.8441,
      "step": 25967
    },
    {
      "epoch": 1.414061383267648,
      "grad_norm": 0.6467762865340976,
      "learning_rate": 4.173774612324943e-05,
      "loss": 11.9332,
      "step": 25968
    },
    {
      "epoch": 1.4141158372642313,
      "grad_norm": 0.5567841750221176,
      "learning_rate": 4.17305793733136e-05,
      "loss": 11.9145,
      "step": 25969
    },
    {
      "epoch": 1.4141702912608143,
      "grad_norm": 0.5222396009735883,
      "learning_rate": 4.1723413076490514e-05,
      "loss": 11.8906,
      "step": 25970
    },
    {
      "epoch": 1.4142247452573973,
      "grad_norm": 0.5485755108908761,
      "learning_rate": 4.171624723283598e-05,
      "loss": 11.913,
      "step": 25971
    },
    {
      "epoch": 1.4142791992539803,
      "grad_norm": 0.5922561876464869,
      "learning_rate": 4.170908184240565e-05,
      "loss": 11.9357,
      "step": 25972
    },
    {
      "epoch": 1.4143336532505633,
      "grad_norm": 0.5856668848009153,
      "learning_rate": 4.17019169052553e-05,
      "loss": 11.8688,
      "step": 25973
    },
    {
      "epoch": 1.4143881072471463,
      "grad_norm": 0.538227228542079,
      "learning_rate": 4.169475242144059e-05,
      "loss": 11.9368,
      "step": 25974
    },
    {
      "epoch": 1.4144425612437292,
      "grad_norm": 0.5515683291958238,
      "learning_rate": 4.168758839101726e-05,
      "loss": 11.8369,
      "step": 25975
    },
    {
      "epoch": 1.4144970152403122,
      "grad_norm": 0.5537332162780446,
      "learning_rate": 4.168042481404106e-05,
      "loss": 11.9316,
      "step": 25976
    },
    {
      "epoch": 1.4145514692368952,
      "grad_norm": 0.5196901118528175,
      "learning_rate": 4.1673261690567625e-05,
      "loss": 11.9138,
      "step": 25977
    },
    {
      "epoch": 1.4146059232334784,
      "grad_norm": 0.5744457087145551,
      "learning_rate": 4.1666099020652725e-05,
      "loss": 11.9619,
      "step": 25978
    },
    {
      "epoch": 1.4146603772300614,
      "grad_norm": 0.5410766570338509,
      "learning_rate": 4.1658936804352014e-05,
      "loss": 11.8747,
      "step": 25979
    },
    {
      "epoch": 1.4147148312266444,
      "grad_norm": 0.4945232811183512,
      "learning_rate": 4.165177504172116e-05,
      "loss": 11.8752,
      "step": 25980
    },
    {
      "epoch": 1.4147692852232274,
      "grad_norm": 0.508991236122,
      "learning_rate": 4.164461373281593e-05,
      "loss": 11.9007,
      "step": 25981
    },
    {
      "epoch": 1.4148237392198104,
      "grad_norm": 0.5330165501366706,
      "learning_rate": 4.1637452877691926e-05,
      "loss": 11.8423,
      "step": 25982
    },
    {
      "epoch": 1.4148781932163934,
      "grad_norm": 0.5652449436507545,
      "learning_rate": 4.163029247640492e-05,
      "loss": 11.6582,
      "step": 25983
    },
    {
      "epoch": 1.4149326472129764,
      "grad_norm": 0.519014294380056,
      "learning_rate": 4.16231325290105e-05,
      "loss": 11.8555,
      "step": 25984
    },
    {
      "epoch": 1.4149871012095594,
      "grad_norm": 0.5537374027393267,
      "learning_rate": 4.16159730355644e-05,
      "loss": 11.7105,
      "step": 25985
    },
    {
      "epoch": 1.4150415552061424,
      "grad_norm": 0.570714286819854,
      "learning_rate": 4.160881399612232e-05,
      "loss": 12.0244,
      "step": 25986
    },
    {
      "epoch": 1.4150960092027254,
      "grad_norm": 0.5083293404096292,
      "learning_rate": 4.160165541073986e-05,
      "loss": 11.8749,
      "step": 25987
    },
    {
      "epoch": 1.4151504631993084,
      "grad_norm": 0.5373861650093892,
      "learning_rate": 4.1594497279472756e-05,
      "loss": 11.9579,
      "step": 25988
    },
    {
      "epoch": 1.4152049171958914,
      "grad_norm": 0.5370248081639699,
      "learning_rate": 4.158733960237664e-05,
      "loss": 11.8867,
      "step": 25989
    },
    {
      "epoch": 1.4152593711924744,
      "grad_norm": 0.5386214998025568,
      "learning_rate": 4.1580182379507125e-05,
      "loss": 11.8309,
      "step": 25990
    },
    {
      "epoch": 1.4153138251890574,
      "grad_norm": 0.5192645532576364,
      "learning_rate": 4.157302561091995e-05,
      "loss": 11.8792,
      "step": 25991
    },
    {
      "epoch": 1.4153682791856403,
      "grad_norm": 0.5721741265348732,
      "learning_rate": 4.156586929667068e-05,
      "loss": 11.7963,
      "step": 25992
    },
    {
      "epoch": 1.4154227331822236,
      "grad_norm": 0.5601051246503809,
      "learning_rate": 4.1558713436815055e-05,
      "loss": 11.9062,
      "step": 25993
    },
    {
      "epoch": 1.4154771871788066,
      "grad_norm": 0.5503199851394696,
      "learning_rate": 4.1551558031408634e-05,
      "loss": 11.7497,
      "step": 25994
    },
    {
      "epoch": 1.4155316411753895,
      "grad_norm": 0.5728285074879068,
      "learning_rate": 4.154440308050713e-05,
      "loss": 11.8929,
      "step": 25995
    },
    {
      "epoch": 1.4155860951719725,
      "grad_norm": 0.5737624825764145,
      "learning_rate": 4.1537248584166136e-05,
      "loss": 11.8988,
      "step": 25996
    },
    {
      "epoch": 1.4156405491685555,
      "grad_norm": 0.5604210694310348,
      "learning_rate": 4.1530094542441276e-05,
      "loss": 11.6098,
      "step": 25997
    },
    {
      "epoch": 1.4156950031651385,
      "grad_norm": 0.5971442269481954,
      "learning_rate": 4.152294095538826e-05,
      "loss": 11.632,
      "step": 25998
    },
    {
      "epoch": 1.4157494571617215,
      "grad_norm": 0.587712392055818,
      "learning_rate": 4.151578782306266e-05,
      "loss": 11.8529,
      "step": 25999
    },
    {
      "epoch": 1.4158039111583045,
      "grad_norm": 0.5199201438969665,
      "learning_rate": 4.150863514552006e-05,
      "loss": 11.8342,
      "step": 26000
    },
    {
      "epoch": 1.4158583651548877,
      "grad_norm": 0.5491790085603228,
      "learning_rate": 4.150148292281615e-05,
      "loss": 11.8861,
      "step": 26001
    },
    {
      "epoch": 1.4159128191514707,
      "grad_norm": 0.5391689353760124,
      "learning_rate": 4.149433115500649e-05,
      "loss": 11.9343,
      "step": 26002
    },
    {
      "epoch": 1.4159672731480537,
      "grad_norm": 0.5539038199046,
      "learning_rate": 4.148717984214676e-05,
      "loss": 11.8852,
      "step": 26003
    },
    {
      "epoch": 1.4160217271446367,
      "grad_norm": 0.5887859534750126,
      "learning_rate": 4.1480028984292494e-05,
      "loss": 11.8545,
      "step": 26004
    },
    {
      "epoch": 1.4160761811412197,
      "grad_norm": 0.4961945883089661,
      "learning_rate": 4.147287858149939e-05,
      "loss": 11.819,
      "step": 26005
    },
    {
      "epoch": 1.4161306351378027,
      "grad_norm": 0.5518306064887253,
      "learning_rate": 4.1465728633822945e-05,
      "loss": 11.9726,
      "step": 26006
    },
    {
      "epoch": 1.4161850891343857,
      "grad_norm": 0.5842209688541331,
      "learning_rate": 4.1458579141318806e-05,
      "loss": 11.8569,
      "step": 26007
    },
    {
      "epoch": 1.4162395431309687,
      "grad_norm": 0.5350962990333932,
      "learning_rate": 4.145143010404261e-05,
      "loss": 11.9312,
      "step": 26008
    },
    {
      "epoch": 1.4162939971275517,
      "grad_norm": 0.5791220006522387,
      "learning_rate": 4.144428152204992e-05,
      "loss": 11.9225,
      "step": 26009
    },
    {
      "epoch": 1.4163484511241347,
      "grad_norm": 0.5465158073386612,
      "learning_rate": 4.143713339539631e-05,
      "loss": 11.9973,
      "step": 26010
    },
    {
      "epoch": 1.4164029051207176,
      "grad_norm": 0.54725218969395,
      "learning_rate": 4.1429985724137344e-05,
      "loss": 11.8318,
      "step": 26011
    },
    {
      "epoch": 1.4164573591173006,
      "grad_norm": 0.5809625208062601,
      "learning_rate": 4.142283850832862e-05,
      "loss": 11.957,
      "step": 26012
    },
    {
      "epoch": 1.4165118131138836,
      "grad_norm": 0.522191329054975,
      "learning_rate": 4.141569174802577e-05,
      "loss": 11.8639,
      "step": 26013
    },
    {
      "epoch": 1.4165662671104666,
      "grad_norm": 0.5063744605868702,
      "learning_rate": 4.1408545443284296e-05,
      "loss": 11.8261,
      "step": 26014
    },
    {
      "epoch": 1.4166207211070496,
      "grad_norm": 0.5846019839427633,
      "learning_rate": 4.140139959415983e-05,
      "loss": 11.8435,
      "step": 26015
    },
    {
      "epoch": 1.4166751751036328,
      "grad_norm": 0.5365235964460565,
      "learning_rate": 4.139425420070786e-05,
      "loss": 11.9106,
      "step": 26016
    },
    {
      "epoch": 1.4167296291002158,
      "grad_norm": 0.5778544788992478,
      "learning_rate": 4.1387109262984045e-05,
      "loss": 11.7469,
      "step": 26017
    },
    {
      "epoch": 1.4167840830967988,
      "grad_norm": 0.553398277724836,
      "learning_rate": 4.137996478104386e-05,
      "loss": 11.8099,
      "step": 26018
    },
    {
      "epoch": 1.4168385370933818,
      "grad_norm": 0.5741702152478732,
      "learning_rate": 4.13728207549429e-05,
      "loss": 11.8734,
      "step": 26019
    },
    {
      "epoch": 1.4168929910899648,
      "grad_norm": 0.550900619627361,
      "learning_rate": 4.136567718473678e-05,
      "loss": 11.8239,
      "step": 26020
    },
    {
      "epoch": 1.4169474450865478,
      "grad_norm": 0.544200301115898,
      "learning_rate": 4.135853407048093e-05,
      "loss": 11.8117,
      "step": 26021
    },
    {
      "epoch": 1.4170018990831308,
      "grad_norm": 0.5174507075260631,
      "learning_rate": 4.135139141223098e-05,
      "loss": 11.908,
      "step": 26022
    },
    {
      "epoch": 1.4170563530797138,
      "grad_norm": 0.6257017394501336,
      "learning_rate": 4.1344249210042416e-05,
      "loss": 12.0665,
      "step": 26023
    },
    {
      "epoch": 1.4171108070762968,
      "grad_norm": 0.533852899698103,
      "learning_rate": 4.13371074639708e-05,
      "loss": 11.7009,
      "step": 26024
    },
    {
      "epoch": 1.41716526107288,
      "grad_norm": 0.5477964478698966,
      "learning_rate": 4.132996617407171e-05,
      "loss": 11.8385,
      "step": 26025
    },
    {
      "epoch": 1.417219715069463,
      "grad_norm": 0.6206393240749204,
      "learning_rate": 4.132282534040062e-05,
      "loss": 11.9334,
      "step": 26026
    },
    {
      "epoch": 1.417274169066046,
      "grad_norm": 0.5392711500843311,
      "learning_rate": 4.13156849630131e-05,
      "loss": 11.5982,
      "step": 26027
    },
    {
      "epoch": 1.417328623062629,
      "grad_norm": 0.5773436095740883,
      "learning_rate": 4.130854504196463e-05,
      "loss": 11.9154,
      "step": 26028
    },
    {
      "epoch": 1.417383077059212,
      "grad_norm": 0.614142876669388,
      "learning_rate": 4.130140557731078e-05,
      "loss": 11.7,
      "step": 26029
    },
    {
      "epoch": 1.417437531055795,
      "grad_norm": 0.6222823179624032,
      "learning_rate": 4.1294266569107054e-05,
      "loss": 11.9589,
      "step": 26030
    },
    {
      "epoch": 1.417491985052378,
      "grad_norm": 0.4657596046349355,
      "learning_rate": 4.1287128017408915e-05,
      "loss": 11.8332,
      "step": 26031
    },
    {
      "epoch": 1.417546439048961,
      "grad_norm": 0.533174547148453,
      "learning_rate": 4.127998992227196e-05,
      "loss": 11.8108,
      "step": 26032
    },
    {
      "epoch": 1.417600893045544,
      "grad_norm": 0.5240046027988714,
      "learning_rate": 4.1272852283751603e-05,
      "loss": 11.9401,
      "step": 26033
    },
    {
      "epoch": 1.417655347042127,
      "grad_norm": 0.5544403069218031,
      "learning_rate": 4.1265715101903405e-05,
      "loss": 11.8395,
      "step": 26034
    },
    {
      "epoch": 1.41770980103871,
      "grad_norm": 0.4970618399510174,
      "learning_rate": 4.1258578376782886e-05,
      "loss": 11.7907,
      "step": 26035
    },
    {
      "epoch": 1.417764255035293,
      "grad_norm": 0.5300108850664907,
      "learning_rate": 4.125144210844547e-05,
      "loss": 11.8877,
      "step": 26036
    },
    {
      "epoch": 1.417818709031876,
      "grad_norm": 0.5711334236326334,
      "learning_rate": 4.1244306296946736e-05,
      "loss": 11.9022,
      "step": 26037
    },
    {
      "epoch": 1.417873163028459,
      "grad_norm": 0.6409867052380667,
      "learning_rate": 4.1237170942342094e-05,
      "loss": 11.9016,
      "step": 26038
    },
    {
      "epoch": 1.417927617025042,
      "grad_norm": 0.5461814100789183,
      "learning_rate": 4.12300360446871e-05,
      "loss": 11.8645,
      "step": 26039
    },
    {
      "epoch": 1.417982071021625,
      "grad_norm": 0.5403477338091139,
      "learning_rate": 4.12229016040372e-05,
      "loss": 11.8089,
      "step": 26040
    },
    {
      "epoch": 1.418036525018208,
      "grad_norm": 0.5330103235941679,
      "learning_rate": 4.121576762044784e-05,
      "loss": 11.8497,
      "step": 26041
    },
    {
      "epoch": 1.418090979014791,
      "grad_norm": 0.5678020052658446,
      "learning_rate": 4.1208634093974575e-05,
      "loss": 11.9499,
      "step": 26042
    },
    {
      "epoch": 1.418145433011374,
      "grad_norm": 0.5869124138748899,
      "learning_rate": 4.120150102467279e-05,
      "loss": 11.8834,
      "step": 26043
    },
    {
      "epoch": 1.418199887007957,
      "grad_norm": 0.5267235374511102,
      "learning_rate": 4.1194368412598015e-05,
      "loss": 11.9181,
      "step": 26044
    },
    {
      "epoch": 1.41825434100454,
      "grad_norm": 0.5408424887123142,
      "learning_rate": 4.118723625780566e-05,
      "loss": 11.907,
      "step": 26045
    },
    {
      "epoch": 1.418308795001123,
      "grad_norm": 0.5820581437565443,
      "learning_rate": 4.118010456035123e-05,
      "loss": 11.8812,
      "step": 26046
    },
    {
      "epoch": 1.418363248997706,
      "grad_norm": 0.5774985774803503,
      "learning_rate": 4.11729733202902e-05,
      "loss": 11.9813,
      "step": 26047
    },
    {
      "epoch": 1.4184177029942893,
      "grad_norm": 0.5343610725050975,
      "learning_rate": 4.1165842537677954e-05,
      "loss": 11.8822,
      "step": 26048
    },
    {
      "epoch": 1.4184721569908723,
      "grad_norm": 0.537262384378862,
      "learning_rate": 4.1158712212570036e-05,
      "loss": 11.9668,
      "step": 26049
    },
    {
      "epoch": 1.4185266109874552,
      "grad_norm": 0.5063228528453032,
      "learning_rate": 4.115158234502182e-05,
      "loss": 11.7703,
      "step": 26050
    },
    {
      "epoch": 1.4185810649840382,
      "grad_norm": 0.5652784886923571,
      "learning_rate": 4.114445293508875e-05,
      "loss": 11.8963,
      "step": 26051
    },
    {
      "epoch": 1.4186355189806212,
      "grad_norm": 0.565569947237846,
      "learning_rate": 4.1137323982826314e-05,
      "loss": 11.8755,
      "step": 26052
    },
    {
      "epoch": 1.4186899729772042,
      "grad_norm": 0.5304698072295083,
      "learning_rate": 4.113019548828988e-05,
      "loss": 11.7958,
      "step": 26053
    },
    {
      "epoch": 1.4187444269737872,
      "grad_norm": 0.6203083568848862,
      "learning_rate": 4.1123067451534944e-05,
      "loss": 11.9796,
      "step": 26054
    },
    {
      "epoch": 1.4187988809703702,
      "grad_norm": 0.5354760052231521,
      "learning_rate": 4.111593987261689e-05,
      "loss": 11.9106,
      "step": 26055
    },
    {
      "epoch": 1.4188533349669532,
      "grad_norm": 0.5340381212757845,
      "learning_rate": 4.110881275159115e-05,
      "loss": 11.9029,
      "step": 26056
    },
    {
      "epoch": 1.4189077889635362,
      "grad_norm": 0.6101841736073859,
      "learning_rate": 4.11016860885132e-05,
      "loss": 11.9062,
      "step": 26057
    },
    {
      "epoch": 1.4189622429601192,
      "grad_norm": 0.5232982904949554,
      "learning_rate": 4.109455988343839e-05,
      "loss": 11.8734,
      "step": 26058
    },
    {
      "epoch": 1.4190166969567022,
      "grad_norm": 0.5467383575943048,
      "learning_rate": 4.108743413642219e-05,
      "loss": 11.7749,
      "step": 26059
    },
    {
      "epoch": 1.4190711509532852,
      "grad_norm": 0.5555725119473592,
      "learning_rate": 4.1080308847519966e-05,
      "loss": 11.7675,
      "step": 26060
    },
    {
      "epoch": 1.4191256049498682,
      "grad_norm": 0.524627134341648,
      "learning_rate": 4.1073184016787124e-05,
      "loss": 11.8669,
      "step": 26061
    },
    {
      "epoch": 1.4191800589464514,
      "grad_norm": 0.5357802924944004,
      "learning_rate": 4.106605964427911e-05,
      "loss": 11.8418,
      "step": 26062
    },
    {
      "epoch": 1.4192345129430344,
      "grad_norm": 0.5722948634309796,
      "learning_rate": 4.105893573005128e-05,
      "loss": 11.8218,
      "step": 26063
    },
    {
      "epoch": 1.4192889669396174,
      "grad_norm": 0.5262608179823776,
      "learning_rate": 4.1051812274159064e-05,
      "loss": 11.9163,
      "step": 26064
    },
    {
      "epoch": 1.4193434209362004,
      "grad_norm": 0.5398488565451185,
      "learning_rate": 4.104468927665782e-05,
      "loss": 11.9241,
      "step": 26065
    },
    {
      "epoch": 1.4193978749327834,
      "grad_norm": 0.5687012735162117,
      "learning_rate": 4.1037566737602994e-05,
      "loss": 11.8976,
      "step": 26066
    },
    {
      "epoch": 1.4194523289293663,
      "grad_norm": 0.5131670006623822,
      "learning_rate": 4.1030444657049894e-05,
      "loss": 11.8278,
      "step": 26067
    },
    {
      "epoch": 1.4195067829259493,
      "grad_norm": 0.5906771309699338,
      "learning_rate": 4.102332303505395e-05,
      "loss": 11.9589,
      "step": 26068
    },
    {
      "epoch": 1.4195612369225323,
      "grad_norm": 0.5221242590418391,
      "learning_rate": 4.1016201871670566e-05,
      "loss": 11.7565,
      "step": 26069
    },
    {
      "epoch": 1.4196156909191153,
      "grad_norm": 0.5080847532759977,
      "learning_rate": 4.10090811669551e-05,
      "loss": 11.8155,
      "step": 26070
    },
    {
      "epoch": 1.4196701449156985,
      "grad_norm": 0.6013787933184735,
      "learning_rate": 4.10019609209629e-05,
      "loss": 12.0543,
      "step": 26071
    },
    {
      "epoch": 1.4197245989122815,
      "grad_norm": 0.5656051014291049,
      "learning_rate": 4.0994841133749316e-05,
      "loss": 11.9374,
      "step": 26072
    },
    {
      "epoch": 1.4197790529088645,
      "grad_norm": 0.5337381619939967,
      "learning_rate": 4.098772180536975e-05,
      "loss": 11.8808,
      "step": 26073
    },
    {
      "epoch": 1.4198335069054475,
      "grad_norm": 0.5726554967655715,
      "learning_rate": 4.098060293587957e-05,
      "loss": 11.8528,
      "step": 26074
    },
    {
      "epoch": 1.4198879609020305,
      "grad_norm": 0.6861360714871024,
      "learning_rate": 4.09734845253341e-05,
      "loss": 11.9746,
      "step": 26075
    },
    {
      "epoch": 1.4199424148986135,
      "grad_norm": 0.5730838285555493,
      "learning_rate": 4.096636657378875e-05,
      "loss": 11.8877,
      "step": 26076
    },
    {
      "epoch": 1.4199968688951965,
      "grad_norm": 0.5645379395460011,
      "learning_rate": 4.09592490812988e-05,
      "loss": 11.9075,
      "step": 26077
    },
    {
      "epoch": 1.4200513228917795,
      "grad_norm": 0.5888163244518616,
      "learning_rate": 4.0952132047919654e-05,
      "loss": 11.8095,
      "step": 26078
    },
    {
      "epoch": 1.4201057768883625,
      "grad_norm": 0.5114499624367852,
      "learning_rate": 4.094501547370661e-05,
      "loss": 11.8477,
      "step": 26079
    },
    {
      "epoch": 1.4201602308849455,
      "grad_norm": 0.5129660993821542,
      "learning_rate": 4.093789935871505e-05,
      "loss": 11.847,
      "step": 26080
    },
    {
      "epoch": 1.4202146848815285,
      "grad_norm": 0.5390772687294157,
      "learning_rate": 4.09307837030003e-05,
      "loss": 11.9161,
      "step": 26081
    },
    {
      "epoch": 1.4202691388781115,
      "grad_norm": 0.5380222504227737,
      "learning_rate": 4.092366850661763e-05,
      "loss": 11.8749,
      "step": 26082
    },
    {
      "epoch": 1.4203235928746945,
      "grad_norm": 0.5877424641458174,
      "learning_rate": 4.0916553769622435e-05,
      "loss": 11.9353,
      "step": 26083
    },
    {
      "epoch": 1.4203780468712774,
      "grad_norm": 0.5670860176310908,
      "learning_rate": 4.0909439492070054e-05,
      "loss": 11.8937,
      "step": 26084
    },
    {
      "epoch": 1.4204325008678604,
      "grad_norm": 0.6083928727948095,
      "learning_rate": 4.090232567401576e-05,
      "loss": 11.9563,
      "step": 26085
    },
    {
      "epoch": 1.4204869548644437,
      "grad_norm": 0.5287215448878694,
      "learning_rate": 4.089521231551492e-05,
      "loss": 11.8916,
      "step": 26086
    },
    {
      "epoch": 1.4205414088610266,
      "grad_norm": 0.509914956110607,
      "learning_rate": 4.088809941662278e-05,
      "loss": 11.7701,
      "step": 26087
    },
    {
      "epoch": 1.4205958628576096,
      "grad_norm": 0.6305841139139768,
      "learning_rate": 4.088098697739473e-05,
      "loss": 11.903,
      "step": 26088
    },
    {
      "epoch": 1.4206503168541926,
      "grad_norm": 0.5456601598513064,
      "learning_rate": 4.087387499788601e-05,
      "loss": 11.8674,
      "step": 26089
    },
    {
      "epoch": 1.4207047708507756,
      "grad_norm": 0.5477823649061792,
      "learning_rate": 4.086676347815198e-05,
      "loss": 11.8448,
      "step": 26090
    },
    {
      "epoch": 1.4207592248473586,
      "grad_norm": 0.56747306327434,
      "learning_rate": 4.0859652418247915e-05,
      "loss": 11.9006,
      "step": 26091
    },
    {
      "epoch": 1.4208136788439416,
      "grad_norm": 0.566420837252433,
      "learning_rate": 4.0852541818229085e-05,
      "loss": 11.8488,
      "step": 26092
    },
    {
      "epoch": 1.4208681328405246,
      "grad_norm": 0.5619605691291363,
      "learning_rate": 4.084543167815084e-05,
      "loss": 11.9379,
      "step": 26093
    },
    {
      "epoch": 1.4209225868371076,
      "grad_norm": 0.5967933155463008,
      "learning_rate": 4.08383219980684e-05,
      "loss": 11.9458,
      "step": 26094
    },
    {
      "epoch": 1.4209770408336908,
      "grad_norm": 0.5697095771538786,
      "learning_rate": 4.083121277803709e-05,
      "loss": 11.9208,
      "step": 26095
    },
    {
      "epoch": 1.4210314948302738,
      "grad_norm": 0.5351231640418428,
      "learning_rate": 4.082410401811222e-05,
      "loss": 11.8725,
      "step": 26096
    },
    {
      "epoch": 1.4210859488268568,
      "grad_norm": 0.6268998248893082,
      "learning_rate": 4.081699571834902e-05,
      "loss": 11.958,
      "step": 26097
    },
    {
      "epoch": 1.4211404028234398,
      "grad_norm": 0.5329125590064157,
      "learning_rate": 4.080988787880281e-05,
      "loss": 11.8002,
      "step": 26098
    },
    {
      "epoch": 1.4211948568200228,
      "grad_norm": 0.6251433177643696,
      "learning_rate": 4.0802780499528804e-05,
      "loss": 12.0663,
      "step": 26099
    },
    {
      "epoch": 1.4212493108166058,
      "grad_norm": 0.5002374462493898,
      "learning_rate": 4.079567358058235e-05,
      "loss": 11.7771,
      "step": 26100
    },
    {
      "epoch": 1.4213037648131888,
      "grad_norm": 0.5244976714655124,
      "learning_rate": 4.078856712201866e-05,
      "loss": 11.8338,
      "step": 26101
    },
    {
      "epoch": 1.4213582188097718,
      "grad_norm": 0.5511202951548435,
      "learning_rate": 4.0781461123892964e-05,
      "loss": 11.8922,
      "step": 26102
    },
    {
      "epoch": 1.4214126728063547,
      "grad_norm": 0.5225968468841047,
      "learning_rate": 4.077435558626059e-05,
      "loss": 11.8962,
      "step": 26103
    },
    {
      "epoch": 1.4214671268029377,
      "grad_norm": 0.5870296286405391,
      "learning_rate": 4.0767250509176734e-05,
      "loss": 11.9209,
      "step": 26104
    },
    {
      "epoch": 1.4215215807995207,
      "grad_norm": 0.5483696163688081,
      "learning_rate": 4.076014589269667e-05,
      "loss": 11.854,
      "step": 26105
    },
    {
      "epoch": 1.4215760347961037,
      "grad_norm": 0.5629190365951552,
      "learning_rate": 4.0753041736875675e-05,
      "loss": 11.9133,
      "step": 26106
    },
    {
      "epoch": 1.4216304887926867,
      "grad_norm": 0.5403040801153737,
      "learning_rate": 4.0745938041768926e-05,
      "loss": 11.9471,
      "step": 26107
    },
    {
      "epoch": 1.4216849427892697,
      "grad_norm": 0.6256786439182609,
      "learning_rate": 4.0738834807431734e-05,
      "loss": 12.0484,
      "step": 26108
    },
    {
      "epoch": 1.421739396785853,
      "grad_norm": 0.5477870347909671,
      "learning_rate": 4.073173203391928e-05,
      "loss": 11.8385,
      "step": 26109
    },
    {
      "epoch": 1.421793850782436,
      "grad_norm": 0.5396897595736808,
      "learning_rate": 4.0724629721286846e-05,
      "loss": 11.9866,
      "step": 26110
    },
    {
      "epoch": 1.421848304779019,
      "grad_norm": 0.548165056983725,
      "learning_rate": 4.071752786958962e-05,
      "loss": 11.9578,
      "step": 26111
    },
    {
      "epoch": 1.421902758775602,
      "grad_norm": 0.5067345279142136,
      "learning_rate": 4.071042647888281e-05,
      "loss": 11.9181,
      "step": 26112
    },
    {
      "epoch": 1.421957212772185,
      "grad_norm": 0.5302774756390238,
      "learning_rate": 4.0703325549221715e-05,
      "loss": 11.7496,
      "step": 26113
    },
    {
      "epoch": 1.422011666768768,
      "grad_norm": 0.5373970293476852,
      "learning_rate": 4.069622508066146e-05,
      "loss": 11.9389,
      "step": 26114
    },
    {
      "epoch": 1.4220661207653509,
      "grad_norm": 0.5824346892358903,
      "learning_rate": 4.068912507325733e-05,
      "loss": 11.7986,
      "step": 26115
    },
    {
      "epoch": 1.4221205747619339,
      "grad_norm": 0.5117474109183592,
      "learning_rate": 4.0682025527064486e-05,
      "loss": 11.8509,
      "step": 26116
    },
    {
      "epoch": 1.4221750287585169,
      "grad_norm": 0.5371351655839012,
      "learning_rate": 4.067492644213815e-05,
      "loss": 11.8704,
      "step": 26117
    },
    {
      "epoch": 1.4222294827551,
      "grad_norm": 0.548480832880369,
      "learning_rate": 4.066782781853358e-05,
      "loss": 11.9652,
      "step": 26118
    },
    {
      "epoch": 1.422283936751683,
      "grad_norm": 0.5413346781788357,
      "learning_rate": 4.066072965630588e-05,
      "loss": 11.8309,
      "step": 26119
    },
    {
      "epoch": 1.422338390748266,
      "grad_norm": 0.5582722841388086,
      "learning_rate": 4.065363195551039e-05,
      "loss": 11.9146,
      "step": 26120
    },
    {
      "epoch": 1.422392844744849,
      "grad_norm": 0.6060756787871406,
      "learning_rate": 4.064653471620213e-05,
      "loss": 12.0039,
      "step": 26121
    },
    {
      "epoch": 1.422447298741432,
      "grad_norm": 0.5455678881802932,
      "learning_rate": 4.0639437938436365e-05,
      "loss": 11.8754,
      "step": 26122
    },
    {
      "epoch": 1.422501752738015,
      "grad_norm": 0.5312607281569431,
      "learning_rate": 4.063234162226832e-05,
      "loss": 11.8576,
      "step": 26123
    },
    {
      "epoch": 1.422556206734598,
      "grad_norm": 0.5362934715754557,
      "learning_rate": 4.062524576775312e-05,
      "loss": 11.8552,
      "step": 26124
    },
    {
      "epoch": 1.422610660731181,
      "grad_norm": 0.5407476284695961,
      "learning_rate": 4.0618150374946e-05,
      "loss": 11.9082,
      "step": 26125
    },
    {
      "epoch": 1.422665114727764,
      "grad_norm": 0.5480359721976515,
      "learning_rate": 4.061105544390205e-05,
      "loss": 11.9509,
      "step": 26126
    },
    {
      "epoch": 1.422719568724347,
      "grad_norm": 0.5530294624497818,
      "learning_rate": 4.060396097467653e-05,
      "loss": 11.952,
      "step": 26127
    },
    {
      "epoch": 1.42277402272093,
      "grad_norm": 0.5398715923136687,
      "learning_rate": 4.0596866967324546e-05,
      "loss": 11.8422,
      "step": 26128
    },
    {
      "epoch": 1.422828476717513,
      "grad_norm": 0.4983930270551854,
      "learning_rate": 4.058977342190128e-05,
      "loss": 11.8372,
      "step": 26129
    },
    {
      "epoch": 1.422882930714096,
      "grad_norm": 0.6456253775489371,
      "learning_rate": 4.058268033846198e-05,
      "loss": 11.8939,
      "step": 26130
    },
    {
      "epoch": 1.422937384710679,
      "grad_norm": 0.5009297295174231,
      "learning_rate": 4.057558771706163e-05,
      "loss": 11.7764,
      "step": 26131
    },
    {
      "epoch": 1.4229918387072622,
      "grad_norm": 0.5131048532860858,
      "learning_rate": 4.056849555775549e-05,
      "loss": 11.8219,
      "step": 26132
    },
    {
      "epoch": 1.4230462927038452,
      "grad_norm": 0.5496043825935185,
      "learning_rate": 4.056140386059873e-05,
      "loss": 11.8435,
      "step": 26133
    },
    {
      "epoch": 1.4231007467004282,
      "grad_norm": 0.5249219292855265,
      "learning_rate": 4.055431262564643e-05,
      "loss": 11.9183,
      "step": 26134
    },
    {
      "epoch": 1.4231552006970112,
      "grad_norm": 0.541743106037878,
      "learning_rate": 4.054722185295379e-05,
      "loss": 11.886,
      "step": 26135
    },
    {
      "epoch": 1.4232096546935942,
      "grad_norm": 0.4990124456180477,
      "learning_rate": 4.0540131542575885e-05,
      "loss": 11.9758,
      "step": 26136
    },
    {
      "epoch": 1.4232641086901772,
      "grad_norm": 0.5689709929035197,
      "learning_rate": 4.0533041694567944e-05,
      "loss": 12.0183,
      "step": 26137
    },
    {
      "epoch": 1.4233185626867602,
      "grad_norm": 0.5319986108079289,
      "learning_rate": 4.0525952308984995e-05,
      "loss": 11.93,
      "step": 26138
    },
    {
      "epoch": 1.4233730166833432,
      "grad_norm": 0.5119422973923216,
      "learning_rate": 4.051886338588221e-05,
      "loss": 11.8446,
      "step": 26139
    },
    {
      "epoch": 1.4234274706799261,
      "grad_norm": 0.529031137841042,
      "learning_rate": 4.051177492531479e-05,
      "loss": 11.8661,
      "step": 26140
    },
    {
      "epoch": 1.4234819246765094,
      "grad_norm": 0.5263093224068561,
      "learning_rate": 4.050468692733772e-05,
      "loss": 11.9608,
      "step": 26141
    },
    {
      "epoch": 1.4235363786730924,
      "grad_norm": 0.5701329276668913,
      "learning_rate": 4.049759939200621e-05,
      "loss": 11.7892,
      "step": 26142
    },
    {
      "epoch": 1.4235908326696753,
      "grad_norm": 0.5224754995736903,
      "learning_rate": 4.049051231937532e-05,
      "loss": 11.8852,
      "step": 26143
    },
    {
      "epoch": 1.4236452866662583,
      "grad_norm": 0.6701563146020101,
      "learning_rate": 4.0483425709500175e-05,
      "loss": 11.981,
      "step": 26144
    },
    {
      "epoch": 1.4236997406628413,
      "grad_norm": 0.5642484403765652,
      "learning_rate": 4.047633956243594e-05,
      "loss": 11.8037,
      "step": 26145
    },
    {
      "epoch": 1.4237541946594243,
      "grad_norm": 0.519367438488522,
      "learning_rate": 4.046925387823762e-05,
      "loss": 11.8752,
      "step": 26146
    },
    {
      "epoch": 1.4238086486560073,
      "grad_norm": 0.6012658325997082,
      "learning_rate": 4.046216865696041e-05,
      "loss": 11.9951,
      "step": 26147
    },
    {
      "epoch": 1.4238631026525903,
      "grad_norm": 0.5626756090497997,
      "learning_rate": 4.045508389865933e-05,
      "loss": 11.9036,
      "step": 26148
    },
    {
      "epoch": 1.4239175566491733,
      "grad_norm": 0.5470869524108306,
      "learning_rate": 4.044799960338953e-05,
      "loss": 11.8585,
      "step": 26149
    },
    {
      "epoch": 1.4239720106457563,
      "grad_norm": 0.5231068677699756,
      "learning_rate": 4.044091577120608e-05,
      "loss": 11.9079,
      "step": 26150
    },
    {
      "epoch": 1.4240264646423393,
      "grad_norm": 0.5193856497954444,
      "learning_rate": 4.043383240216402e-05,
      "loss": 11.8574,
      "step": 26151
    },
    {
      "epoch": 1.4240809186389223,
      "grad_norm": 0.5775055306382821,
      "learning_rate": 4.042674949631849e-05,
      "loss": 11.8931,
      "step": 26152
    },
    {
      "epoch": 1.4241353726355053,
      "grad_norm": 0.5401020807876373,
      "learning_rate": 4.0419667053724533e-05,
      "loss": 11.8396,
      "step": 26153
    },
    {
      "epoch": 1.4241898266320883,
      "grad_norm": 0.5886578061616173,
      "learning_rate": 4.0412585074437214e-05,
      "loss": 11.8322,
      "step": 26154
    },
    {
      "epoch": 1.4242442806286713,
      "grad_norm": 0.630354039105725,
      "learning_rate": 4.040550355851168e-05,
      "loss": 11.8991,
      "step": 26155
    },
    {
      "epoch": 1.4242987346252545,
      "grad_norm": 0.5108081604678174,
      "learning_rate": 4.0398422506002896e-05,
      "loss": 11.8605,
      "step": 26156
    },
    {
      "epoch": 1.4243531886218375,
      "grad_norm": 0.5493105690597134,
      "learning_rate": 4.039134191696602e-05,
      "loss": 11.8864,
      "step": 26157
    },
    {
      "epoch": 1.4244076426184205,
      "grad_norm": 0.5939040435766363,
      "learning_rate": 4.0384261791456024e-05,
      "loss": 11.8893,
      "step": 26158
    },
    {
      "epoch": 1.4244620966150034,
      "grad_norm": 0.5848380633593319,
      "learning_rate": 4.037718212952805e-05,
      "loss": 11.8828,
      "step": 26159
    },
    {
      "epoch": 1.4245165506115864,
      "grad_norm": 0.603612017642678,
      "learning_rate": 4.0370102931237055e-05,
      "loss": 11.9141,
      "step": 26160
    },
    {
      "epoch": 1.4245710046081694,
      "grad_norm": 0.6001998674268149,
      "learning_rate": 4.036302419663819e-05,
      "loss": 11.8094,
      "step": 26161
    },
    {
      "epoch": 1.4246254586047524,
      "grad_norm": 0.5414205918706573,
      "learning_rate": 4.035594592578644e-05,
      "loss": 11.8223,
      "step": 26162
    },
    {
      "epoch": 1.4246799126013354,
      "grad_norm": 0.5063186603290926,
      "learning_rate": 4.034886811873683e-05,
      "loss": 11.7757,
      "step": 26163
    },
    {
      "epoch": 1.4247343665979186,
      "grad_norm": 0.5185572655618436,
      "learning_rate": 4.034179077554445e-05,
      "loss": 11.8154,
      "step": 26164
    },
    {
      "epoch": 1.4247888205945016,
      "grad_norm": 0.5204732265212506,
      "learning_rate": 4.0334713896264285e-05,
      "loss": 11.7669,
      "step": 26165
    },
    {
      "epoch": 1.4248432745910846,
      "grad_norm": 0.5523725407549351,
      "learning_rate": 4.0327637480951384e-05,
      "loss": 11.8258,
      "step": 26166
    },
    {
      "epoch": 1.4248977285876676,
      "grad_norm": 0.5470036729652743,
      "learning_rate": 4.0320561529660816e-05,
      "loss": 11.8053,
      "step": 26167
    },
    {
      "epoch": 1.4249521825842506,
      "grad_norm": 0.5282009126328882,
      "learning_rate": 4.031348604244753e-05,
      "loss": 11.8806,
      "step": 26168
    },
    {
      "epoch": 1.4250066365808336,
      "grad_norm": 0.6393707360575117,
      "learning_rate": 4.030641101936663e-05,
      "loss": 11.7239,
      "step": 26169
    },
    {
      "epoch": 1.4250610905774166,
      "grad_norm": 0.49075736873806214,
      "learning_rate": 4.029933646047306e-05,
      "loss": 11.8649,
      "step": 26170
    },
    {
      "epoch": 1.4251155445739996,
      "grad_norm": 0.587060484404055,
      "learning_rate": 4.0292262365821885e-05,
      "loss": 11.8673,
      "step": 26171
    },
    {
      "epoch": 1.4251699985705826,
      "grad_norm": 0.5549308115617355,
      "learning_rate": 4.0285188735468095e-05,
      "loss": 11.7866,
      "step": 26172
    },
    {
      "epoch": 1.4252244525671656,
      "grad_norm": 0.6082517650070302,
      "learning_rate": 4.0278115569466645e-05,
      "loss": 12.0176,
      "step": 26173
    },
    {
      "epoch": 1.4252789065637486,
      "grad_norm": 0.5425571203776478,
      "learning_rate": 4.027104286787263e-05,
      "loss": 11.8957,
      "step": 26174
    },
    {
      "epoch": 1.4253333605603316,
      "grad_norm": 0.5535651829595233,
      "learning_rate": 4.026397063074097e-05,
      "loss": 11.6618,
      "step": 26175
    },
    {
      "epoch": 1.4253878145569145,
      "grad_norm": 0.5535735271816487,
      "learning_rate": 4.0256898858126726e-05,
      "loss": 11.8162,
      "step": 26176
    },
    {
      "epoch": 1.4254422685534975,
      "grad_norm": 0.5490468139436682,
      "learning_rate": 4.0249827550084815e-05,
      "loss": 11.8025,
      "step": 26177
    },
    {
      "epoch": 1.4254967225500805,
      "grad_norm": 0.5263919424235999,
      "learning_rate": 4.024275670667027e-05,
      "loss": 11.784,
      "step": 26178
    },
    {
      "epoch": 1.4255511765466637,
      "grad_norm": 0.6019793120904708,
      "learning_rate": 4.0235686327938096e-05,
      "loss": 11.7365,
      "step": 26179
    },
    {
      "epoch": 1.4256056305432467,
      "grad_norm": 0.5459217845914554,
      "learning_rate": 4.022861641394322e-05,
      "loss": 11.8549,
      "step": 26180
    },
    {
      "epoch": 1.4256600845398297,
      "grad_norm": 0.5721077614789521,
      "learning_rate": 4.0221546964740684e-05,
      "loss": 11.8439,
      "step": 26181
    },
    {
      "epoch": 1.4257145385364127,
      "grad_norm": 0.5334336493246139,
      "learning_rate": 4.0214477980385427e-05,
      "loss": 11.9349,
      "step": 26182
    },
    {
      "epoch": 1.4257689925329957,
      "grad_norm": 0.5811059264341433,
      "learning_rate": 4.020740946093237e-05,
      "loss": 11.7375,
      "step": 26183
    },
    {
      "epoch": 1.4258234465295787,
      "grad_norm": 0.5033816827030118,
      "learning_rate": 4.0200341406436557e-05,
      "loss": 11.8887,
      "step": 26184
    },
    {
      "epoch": 1.4258779005261617,
      "grad_norm": 0.6284397320310644,
      "learning_rate": 4.019327381695289e-05,
      "loss": 11.6962,
      "step": 26185
    },
    {
      "epoch": 1.4259323545227447,
      "grad_norm": 0.529838968932736,
      "learning_rate": 4.018620669253639e-05,
      "loss": 11.7569,
      "step": 26186
    },
    {
      "epoch": 1.4259868085193277,
      "grad_norm": 0.5636641361792095,
      "learning_rate": 4.0179140033241936e-05,
      "loss": 11.7074,
      "step": 26187
    },
    {
      "epoch": 1.426041262515911,
      "grad_norm": 0.5397557261826808,
      "learning_rate": 4.0172073839124527e-05,
      "loss": 11.9393,
      "step": 26188
    },
    {
      "epoch": 1.426095716512494,
      "grad_norm": 0.7155158038755675,
      "learning_rate": 4.0165008110239144e-05,
      "loss": 12.053,
      "step": 26189
    },
    {
      "epoch": 1.4261501705090769,
      "grad_norm": 0.6202564568682207,
      "learning_rate": 4.015794284664065e-05,
      "loss": 11.8824,
      "step": 26190
    },
    {
      "epoch": 1.4262046245056599,
      "grad_norm": 0.5361782999426571,
      "learning_rate": 4.01508780483841e-05,
      "loss": 11.8593,
      "step": 26191
    },
    {
      "epoch": 1.4262590785022429,
      "grad_norm": 0.51339325970096,
      "learning_rate": 4.0143813715524295e-05,
      "loss": 11.7959,
      "step": 26192
    },
    {
      "epoch": 1.4263135324988259,
      "grad_norm": 0.5247329542683532,
      "learning_rate": 4.013674984811622e-05,
      "loss": 11.8474,
      "step": 26193
    },
    {
      "epoch": 1.4263679864954089,
      "grad_norm": 0.5310358060783806,
      "learning_rate": 4.012968644621487e-05,
      "loss": 11.8446,
      "step": 26194
    },
    {
      "epoch": 1.4264224404919918,
      "grad_norm": 0.5698145778999093,
      "learning_rate": 4.012262350987507e-05,
      "loss": 11.9861,
      "step": 26195
    },
    {
      "epoch": 1.4264768944885748,
      "grad_norm": 0.5218084305810632,
      "learning_rate": 4.011556103915183e-05,
      "loss": 11.7827,
      "step": 26196
    },
    {
      "epoch": 1.4265313484851578,
      "grad_norm": 0.5828391582600716,
      "learning_rate": 4.010849903409999e-05,
      "loss": 11.8251,
      "step": 26197
    },
    {
      "epoch": 1.4265858024817408,
      "grad_norm": 0.5927900006822193,
      "learning_rate": 4.010143749477454e-05,
      "loss": 11.8264,
      "step": 26198
    },
    {
      "epoch": 1.4266402564783238,
      "grad_norm": 0.5523033674032058,
      "learning_rate": 4.0094376421230326e-05,
      "loss": 11.8712,
      "step": 26199
    },
    {
      "epoch": 1.4266947104749068,
      "grad_norm": 0.48796831006239316,
      "learning_rate": 4.0087315813522283e-05,
      "loss": 11.7577,
      "step": 26200
    },
    {
      "epoch": 1.4267491644714898,
      "grad_norm": 0.6614840011013181,
      "learning_rate": 4.00802556717054e-05,
      "loss": 11.8877,
      "step": 26201
    },
    {
      "epoch": 1.426803618468073,
      "grad_norm": 0.509287481742404,
      "learning_rate": 4.0073195995834424e-05,
      "loss": 11.7933,
      "step": 26202
    },
    {
      "epoch": 1.426858072464656,
      "grad_norm": 0.5295338816705573,
      "learning_rate": 4.0066136785964316e-05,
      "loss": 11.8831,
      "step": 26203
    },
    {
      "epoch": 1.426912526461239,
      "grad_norm": 0.5209799435848458,
      "learning_rate": 4.0059078042150024e-05,
      "loss": 11.6572,
      "step": 26204
    },
    {
      "epoch": 1.426966980457822,
      "grad_norm": 0.5310987449037925,
      "learning_rate": 4.0052019764446355e-05,
      "loss": 11.8869,
      "step": 26205
    },
    {
      "epoch": 1.427021434454405,
      "grad_norm": 0.5326269055056083,
      "learning_rate": 4.0044961952908264e-05,
      "loss": 11.7507,
      "step": 26206
    },
    {
      "epoch": 1.427075888450988,
      "grad_norm": 0.5432080036948125,
      "learning_rate": 4.003790460759058e-05,
      "loss": 11.9245,
      "step": 26207
    },
    {
      "epoch": 1.427130342447571,
      "grad_norm": 0.5271225762138729,
      "learning_rate": 4.0030847728548235e-05,
      "loss": 11.9062,
      "step": 26208
    },
    {
      "epoch": 1.427184796444154,
      "grad_norm": 0.5309945912159396,
      "learning_rate": 4.002379131583603e-05,
      "loss": 11.9904,
      "step": 26209
    },
    {
      "epoch": 1.427239250440737,
      "grad_norm": 0.5184897846958649,
      "learning_rate": 4.0016735369508896e-05,
      "loss": 11.8469,
      "step": 26210
    },
    {
      "epoch": 1.4272937044373202,
      "grad_norm": 0.5304836564332092,
      "learning_rate": 4.000967988962176e-05,
      "loss": 11.8576,
      "step": 26211
    },
    {
      "epoch": 1.4273481584339032,
      "grad_norm": 0.5057999406033876,
      "learning_rate": 4.000262487622932e-05,
      "loss": 11.8753,
      "step": 26212
    },
    {
      "epoch": 1.4274026124304862,
      "grad_norm": 0.6491732972027163,
      "learning_rate": 3.999557032938659e-05,
      "loss": 11.9121,
      "step": 26213
    },
    {
      "epoch": 1.4274570664270692,
      "grad_norm": 0.5739909975227664,
      "learning_rate": 3.998851624914832e-05,
      "loss": 11.9992,
      "step": 26214
    },
    {
      "epoch": 1.4275115204236521,
      "grad_norm": 0.4940367374012289,
      "learning_rate": 3.998146263556941e-05,
      "loss": 11.7942,
      "step": 26215
    },
    {
      "epoch": 1.4275659744202351,
      "grad_norm": 0.544345263463131,
      "learning_rate": 3.997440948870476e-05,
      "loss": 11.7832,
      "step": 26216
    },
    {
      "epoch": 1.4276204284168181,
      "grad_norm": 0.5316257964238552,
      "learning_rate": 3.9967356808609126e-05,
      "loss": 11.9227,
      "step": 26217
    },
    {
      "epoch": 1.4276748824134011,
      "grad_norm": 0.49500755455490225,
      "learning_rate": 3.9960304595337414e-05,
      "loss": 11.8202,
      "step": 26218
    },
    {
      "epoch": 1.4277293364099841,
      "grad_norm": 0.5218596838066738,
      "learning_rate": 3.995325284894442e-05,
      "loss": 11.8303,
      "step": 26219
    },
    {
      "epoch": 1.427783790406567,
      "grad_norm": 0.529783229415229,
      "learning_rate": 3.994620156948504e-05,
      "loss": 11.8972,
      "step": 26220
    },
    {
      "epoch": 1.42783824440315,
      "grad_norm": 0.5265848539579019,
      "learning_rate": 3.993915075701407e-05,
      "loss": 11.8772,
      "step": 26221
    },
    {
      "epoch": 1.427892698399733,
      "grad_norm": 0.5954826891122208,
      "learning_rate": 3.9932100411586296e-05,
      "loss": 11.9406,
      "step": 26222
    },
    {
      "epoch": 1.427947152396316,
      "grad_norm": 0.5807564782298249,
      "learning_rate": 3.992505053325662e-05,
      "loss": 11.7988,
      "step": 26223
    },
    {
      "epoch": 1.428001606392899,
      "grad_norm": 0.5196498650807065,
      "learning_rate": 3.9918001122079785e-05,
      "loss": 11.9653,
      "step": 26224
    },
    {
      "epoch": 1.428056060389482,
      "grad_norm": 0.5158191391345669,
      "learning_rate": 3.9910952178110686e-05,
      "loss": 11.8003,
      "step": 26225
    },
    {
      "epoch": 1.4281105143860653,
      "grad_norm": 0.5334710572942586,
      "learning_rate": 3.9903903701404054e-05,
      "loss": 11.8635,
      "step": 26226
    },
    {
      "epoch": 1.4281649683826483,
      "grad_norm": 0.5423864235929465,
      "learning_rate": 3.9896855692014755e-05,
      "loss": 11.8995,
      "step": 26227
    },
    {
      "epoch": 1.4282194223792313,
      "grad_norm": 0.5497611115481867,
      "learning_rate": 3.988980814999763e-05,
      "loss": 11.9004,
      "step": 26228
    },
    {
      "epoch": 1.4282738763758143,
      "grad_norm": 0.611807334997668,
      "learning_rate": 3.9882761075407396e-05,
      "loss": 11.6557,
      "step": 26229
    },
    {
      "epoch": 1.4283283303723973,
      "grad_norm": 0.5336473333260907,
      "learning_rate": 3.9875714468298916e-05,
      "loss": 11.7722,
      "step": 26230
    },
    {
      "epoch": 1.4283827843689803,
      "grad_norm": 0.5725208809146682,
      "learning_rate": 3.9868668328726974e-05,
      "loss": 11.9451,
      "step": 26231
    },
    {
      "epoch": 1.4284372383655632,
      "grad_norm": 0.5715020903163366,
      "learning_rate": 3.986162265674632e-05,
      "loss": 11.9609,
      "step": 26232
    },
    {
      "epoch": 1.4284916923621462,
      "grad_norm": 0.6011111369036927,
      "learning_rate": 3.98545774524118e-05,
      "loss": 11.9879,
      "step": 26233
    },
    {
      "epoch": 1.4285461463587295,
      "grad_norm": 0.5837201780273968,
      "learning_rate": 3.9847532715778147e-05,
      "loss": 11.8455,
      "step": 26234
    },
    {
      "epoch": 1.4286006003553124,
      "grad_norm": 0.5816848722841986,
      "learning_rate": 3.984048844690019e-05,
      "loss": 11.9353,
      "step": 26235
    },
    {
      "epoch": 1.4286550543518954,
      "grad_norm": 0.5700366504916987,
      "learning_rate": 3.983344464583266e-05,
      "loss": 11.9536,
      "step": 26236
    },
    {
      "epoch": 1.4287095083484784,
      "grad_norm": 0.539345981134819,
      "learning_rate": 3.9826401312630345e-05,
      "loss": 11.7942,
      "step": 26237
    },
    {
      "epoch": 1.4287639623450614,
      "grad_norm": 0.5475082507150271,
      "learning_rate": 3.981935844734806e-05,
      "loss": 11.7893,
      "step": 26238
    },
    {
      "epoch": 1.4288184163416444,
      "grad_norm": 0.5207017972419579,
      "learning_rate": 3.981231605004051e-05,
      "loss": 11.8262,
      "step": 26239
    },
    {
      "epoch": 1.4288728703382274,
      "grad_norm": 0.5933324562536302,
      "learning_rate": 3.9805274120762516e-05,
      "loss": 11.9662,
      "step": 26240
    },
    {
      "epoch": 1.4289273243348104,
      "grad_norm": 0.5644267583206891,
      "learning_rate": 3.97982326595688e-05,
      "loss": 11.8991,
      "step": 26241
    },
    {
      "epoch": 1.4289817783313934,
      "grad_norm": 0.5959134431198703,
      "learning_rate": 3.9791191666514093e-05,
      "loss": 11.891,
      "step": 26242
    },
    {
      "epoch": 1.4290362323279764,
      "grad_norm": 0.5069766114074111,
      "learning_rate": 3.9784151141653206e-05,
      "loss": 11.83,
      "step": 26243
    },
    {
      "epoch": 1.4290906863245594,
      "grad_norm": 0.500081980322355,
      "learning_rate": 3.977711108504083e-05,
      "loss": 11.8952,
      "step": 26244
    },
    {
      "epoch": 1.4291451403211424,
      "grad_norm": 0.5509883593483207,
      "learning_rate": 3.977007149673177e-05,
      "loss": 11.9447,
      "step": 26245
    },
    {
      "epoch": 1.4291995943177254,
      "grad_norm": 0.5105273151697763,
      "learning_rate": 3.97630323767807e-05,
      "loss": 11.8086,
      "step": 26246
    },
    {
      "epoch": 1.4292540483143084,
      "grad_norm": 0.5205523131391283,
      "learning_rate": 3.9755993725242414e-05,
      "loss": 11.9408,
      "step": 26247
    },
    {
      "epoch": 1.4293085023108913,
      "grad_norm": 0.499158442176314,
      "learning_rate": 3.9748955542171605e-05,
      "loss": 11.8502,
      "step": 26248
    },
    {
      "epoch": 1.4293629563074746,
      "grad_norm": 0.4944646452886164,
      "learning_rate": 3.9741917827623024e-05,
      "loss": 11.9085,
      "step": 26249
    },
    {
      "epoch": 1.4294174103040576,
      "grad_norm": 0.5931700142739695,
      "learning_rate": 3.973488058165141e-05,
      "loss": 11.7838,
      "step": 26250
    },
    {
      "epoch": 1.4294718643006405,
      "grad_norm": 0.5208407011360756,
      "learning_rate": 3.972784380431149e-05,
      "loss": 11.7544,
      "step": 26251
    },
    {
      "epoch": 1.4295263182972235,
      "grad_norm": 0.5181320276905301,
      "learning_rate": 3.972080749565792e-05,
      "loss": 11.8213,
      "step": 26252
    },
    {
      "epoch": 1.4295807722938065,
      "grad_norm": 0.5528504837602145,
      "learning_rate": 3.9713771655745504e-05,
      "loss": 11.8798,
      "step": 26253
    },
    {
      "epoch": 1.4296352262903895,
      "grad_norm": 0.5721125616895366,
      "learning_rate": 3.970673628462886e-05,
      "loss": 11.8666,
      "step": 26254
    },
    {
      "epoch": 1.4296896802869725,
      "grad_norm": 0.5634194346236857,
      "learning_rate": 3.96997013823628e-05,
      "loss": 11.809,
      "step": 26255
    },
    {
      "epoch": 1.4297441342835555,
      "grad_norm": 0.5339333376747236,
      "learning_rate": 3.9692666949001925e-05,
      "loss": 11.8948,
      "step": 26256
    },
    {
      "epoch": 1.4297985882801385,
      "grad_norm": 0.6087385653395961,
      "learning_rate": 3.968563298460102e-05,
      "loss": 11.8149,
      "step": 26257
    },
    {
      "epoch": 1.4298530422767217,
      "grad_norm": 0.4913557122914008,
      "learning_rate": 3.967859948921472e-05,
      "loss": 11.7974,
      "step": 26258
    },
    {
      "epoch": 1.4299074962733047,
      "grad_norm": 0.5510039420406786,
      "learning_rate": 3.9671566462897734e-05,
      "loss": 11.9316,
      "step": 26259
    },
    {
      "epoch": 1.4299619502698877,
      "grad_norm": 0.6037638777716818,
      "learning_rate": 3.9664533905704815e-05,
      "loss": 11.9781,
      "step": 26260
    },
    {
      "epoch": 1.4300164042664707,
      "grad_norm": 0.5375655881978909,
      "learning_rate": 3.965750181769059e-05,
      "loss": 11.9076,
      "step": 26261
    },
    {
      "epoch": 1.4300708582630537,
      "grad_norm": 0.5605225482339553,
      "learning_rate": 3.965047019890975e-05,
      "loss": 11.898,
      "step": 26262
    },
    {
      "epoch": 1.4301253122596367,
      "grad_norm": 0.5463242407483168,
      "learning_rate": 3.964343904941694e-05,
      "loss": 11.8452,
      "step": 26263
    },
    {
      "epoch": 1.4301797662562197,
      "grad_norm": 0.5270700342324047,
      "learning_rate": 3.9636408369266874e-05,
      "loss": 11.8032,
      "step": 26264
    },
    {
      "epoch": 1.4302342202528027,
      "grad_norm": 0.5304016085082199,
      "learning_rate": 3.962937815851425e-05,
      "loss": 11.7853,
      "step": 26265
    },
    {
      "epoch": 1.4302886742493857,
      "grad_norm": 0.5134334150262553,
      "learning_rate": 3.9622348417213674e-05,
      "loss": 11.8255,
      "step": 26266
    },
    {
      "epoch": 1.4303431282459687,
      "grad_norm": 0.5402625393626279,
      "learning_rate": 3.961531914541987e-05,
      "loss": 11.863,
      "step": 26267
    },
    {
      "epoch": 1.4303975822425516,
      "grad_norm": 0.5673066880455563,
      "learning_rate": 3.960829034318745e-05,
      "loss": 11.7563,
      "step": 26268
    },
    {
      "epoch": 1.4304520362391346,
      "grad_norm": 0.5133852115553672,
      "learning_rate": 3.960126201057112e-05,
      "loss": 11.8093,
      "step": 26269
    },
    {
      "epoch": 1.4305064902357176,
      "grad_norm": 0.5717736604225998,
      "learning_rate": 3.959423414762546e-05,
      "loss": 11.9245,
      "step": 26270
    },
    {
      "epoch": 1.4305609442323006,
      "grad_norm": 0.5379993862072009,
      "learning_rate": 3.9587206754405215e-05,
      "loss": 11.7369,
      "step": 26271
    },
    {
      "epoch": 1.4306153982288838,
      "grad_norm": 0.5560623358336024,
      "learning_rate": 3.958017983096497e-05,
      "loss": 11.9136,
      "step": 26272
    },
    {
      "epoch": 1.4306698522254668,
      "grad_norm": 0.5375427123651487,
      "learning_rate": 3.957315337735935e-05,
      "loss": 11.7954,
      "step": 26273
    },
    {
      "epoch": 1.4307243062220498,
      "grad_norm": 0.5535774098794369,
      "learning_rate": 3.956612739364306e-05,
      "loss": 11.7769,
      "step": 26274
    },
    {
      "epoch": 1.4307787602186328,
      "grad_norm": 0.5391283028540276,
      "learning_rate": 3.955910187987066e-05,
      "loss": 11.9049,
      "step": 26275
    },
    {
      "epoch": 1.4308332142152158,
      "grad_norm": 0.5617764524621167,
      "learning_rate": 3.955207683609682e-05,
      "loss": 11.8965,
      "step": 26276
    },
    {
      "epoch": 1.4308876682117988,
      "grad_norm": 0.561202540887149,
      "learning_rate": 3.9545052262376205e-05,
      "loss": 11.9329,
      "step": 26277
    },
    {
      "epoch": 1.4309421222083818,
      "grad_norm": 0.5495423846866626,
      "learning_rate": 3.953802815876336e-05,
      "loss": 11.8629,
      "step": 26278
    },
    {
      "epoch": 1.4309965762049648,
      "grad_norm": 0.6215470131896877,
      "learning_rate": 3.9531004525312984e-05,
      "loss": 12.0592,
      "step": 26279
    },
    {
      "epoch": 1.4310510302015478,
      "grad_norm": 0.5042639495011494,
      "learning_rate": 3.9523981362079633e-05,
      "loss": 11.8585,
      "step": 26280
    },
    {
      "epoch": 1.431105484198131,
      "grad_norm": 0.5553416594210632,
      "learning_rate": 3.951695866911798e-05,
      "loss": 11.7984,
      "step": 26281
    },
    {
      "epoch": 1.431159938194714,
      "grad_norm": 0.5139239586395046,
      "learning_rate": 3.9509936446482584e-05,
      "loss": 11.9491,
      "step": 26282
    },
    {
      "epoch": 1.431214392191297,
      "grad_norm": 0.5871534059786202,
      "learning_rate": 3.9502914694228043e-05,
      "loss": 11.9945,
      "step": 26283
    },
    {
      "epoch": 1.43126884618788,
      "grad_norm": 0.5534052479716304,
      "learning_rate": 3.9495893412409015e-05,
      "loss": 11.8266,
      "step": 26284
    },
    {
      "epoch": 1.431323300184463,
      "grad_norm": 0.5597620717468003,
      "learning_rate": 3.948887260108003e-05,
      "loss": 11.9313,
      "step": 26285
    },
    {
      "epoch": 1.431377754181046,
      "grad_norm": 0.5563427268608134,
      "learning_rate": 3.948185226029571e-05,
      "loss": 11.8254,
      "step": 26286
    },
    {
      "epoch": 1.431432208177629,
      "grad_norm": 0.5759137649375125,
      "learning_rate": 3.9474832390110705e-05,
      "loss": 11.9367,
      "step": 26287
    },
    {
      "epoch": 1.431486662174212,
      "grad_norm": 0.49215373796281114,
      "learning_rate": 3.9467812990579514e-05,
      "loss": 11.8003,
      "step": 26288
    },
    {
      "epoch": 1.431541116170795,
      "grad_norm": 0.5960469170428271,
      "learning_rate": 3.94607940617568e-05,
      "loss": 11.9812,
      "step": 26289
    },
    {
      "epoch": 1.431595570167378,
      "grad_norm": 0.5400588528318553,
      "learning_rate": 3.945377560369706e-05,
      "loss": 11.8169,
      "step": 26290
    },
    {
      "epoch": 1.431650024163961,
      "grad_norm": 0.5116560040426905,
      "learning_rate": 3.944675761645495e-05,
      "loss": 11.8077,
      "step": 26291
    },
    {
      "epoch": 1.431704478160544,
      "grad_norm": 0.5938105752669841,
      "learning_rate": 3.9439740100085024e-05,
      "loss": 11.8301,
      "step": 26292
    },
    {
      "epoch": 1.431758932157127,
      "grad_norm": 0.5676136162209637,
      "learning_rate": 3.9432723054641786e-05,
      "loss": 11.7587,
      "step": 26293
    },
    {
      "epoch": 1.43181338615371,
      "grad_norm": 0.6049528042386776,
      "learning_rate": 3.942570648017988e-05,
      "loss": 11.9591,
      "step": 26294
    },
    {
      "epoch": 1.431867840150293,
      "grad_norm": 0.5602248595932456,
      "learning_rate": 3.9418690376753806e-05,
      "loss": 11.8574,
      "step": 26295
    },
    {
      "epoch": 1.431922294146876,
      "grad_norm": 0.5735933686420074,
      "learning_rate": 3.94116747444182e-05,
      "loss": 11.9627,
      "step": 26296
    },
    {
      "epoch": 1.431976748143459,
      "grad_norm": 0.5344121507458838,
      "learning_rate": 3.940465958322753e-05,
      "loss": 11.8407,
      "step": 26297
    },
    {
      "epoch": 1.432031202140042,
      "grad_norm": 0.5387036759812672,
      "learning_rate": 3.9397644893236396e-05,
      "loss": 11.941,
      "step": 26298
    },
    {
      "epoch": 1.432085656136625,
      "grad_norm": 0.5520957462342303,
      "learning_rate": 3.939063067449936e-05,
      "loss": 11.83,
      "step": 26299
    },
    {
      "epoch": 1.432140110133208,
      "grad_norm": 0.5798661274176112,
      "learning_rate": 3.9383616927070924e-05,
      "loss": 11.9121,
      "step": 26300
    },
    {
      "epoch": 1.432194564129791,
      "grad_norm": 0.5537868892540694,
      "learning_rate": 3.937660365100567e-05,
      "loss": 12.0007,
      "step": 26301
    },
    {
      "epoch": 1.432249018126374,
      "grad_norm": 0.6438595153869893,
      "learning_rate": 3.9369590846358115e-05,
      "loss": 11.8297,
      "step": 26302
    },
    {
      "epoch": 1.432303472122957,
      "grad_norm": 0.5848865296047472,
      "learning_rate": 3.9362578513182766e-05,
      "loss": 12.022,
      "step": 26303
    },
    {
      "epoch": 1.4323579261195403,
      "grad_norm": 0.5646558160250067,
      "learning_rate": 3.9355566651534206e-05,
      "loss": 11.7747,
      "step": 26304
    },
    {
      "epoch": 1.4324123801161233,
      "grad_norm": 0.5127284651778233,
      "learning_rate": 3.934855526146689e-05,
      "loss": 11.8638,
      "step": 26305
    },
    {
      "epoch": 1.4324668341127063,
      "grad_norm": 0.5319451536490147,
      "learning_rate": 3.934154434303541e-05,
      "loss": 11.9292,
      "step": 26306
    },
    {
      "epoch": 1.4325212881092892,
      "grad_norm": 0.546971837514057,
      "learning_rate": 3.9334533896294226e-05,
      "loss": 11.9256,
      "step": 26307
    },
    {
      "epoch": 1.4325757421058722,
      "grad_norm": 0.5490548675384516,
      "learning_rate": 3.9327523921297884e-05,
      "loss": 11.8369,
      "step": 26308
    },
    {
      "epoch": 1.4326301961024552,
      "grad_norm": 0.5982828939345142,
      "learning_rate": 3.932051441810092e-05,
      "loss": 11.8974,
      "step": 26309
    },
    {
      "epoch": 1.4326846500990382,
      "grad_norm": 0.5310811719517157,
      "learning_rate": 3.931350538675777e-05,
      "loss": 11.8493,
      "step": 26310
    },
    {
      "epoch": 1.4327391040956212,
      "grad_norm": 0.5493572506616808,
      "learning_rate": 3.930649682732302e-05,
      "loss": 11.7759,
      "step": 26311
    },
    {
      "epoch": 1.4327935580922042,
      "grad_norm": 0.5325032859555068,
      "learning_rate": 3.929948873985113e-05,
      "loss": 11.9618,
      "step": 26312
    },
    {
      "epoch": 1.4328480120887872,
      "grad_norm": 0.5283287961537859,
      "learning_rate": 3.9292481124396565e-05,
      "loss": 11.8385,
      "step": 26313
    },
    {
      "epoch": 1.4329024660853702,
      "grad_norm": 0.5244158558104707,
      "learning_rate": 3.9285473981013876e-05,
      "loss": 11.6144,
      "step": 26314
    },
    {
      "epoch": 1.4329569200819532,
      "grad_norm": 0.5108402266768652,
      "learning_rate": 3.9278467309757485e-05,
      "loss": 11.8209,
      "step": 26315
    },
    {
      "epoch": 1.4330113740785362,
      "grad_norm": 0.5963943287252947,
      "learning_rate": 3.927146111068196e-05,
      "loss": 11.8848,
      "step": 26316
    },
    {
      "epoch": 1.4330658280751192,
      "grad_norm": 0.5553531840371618,
      "learning_rate": 3.9264455383841694e-05,
      "loss": 11.8934,
      "step": 26317
    },
    {
      "epoch": 1.4331202820717022,
      "grad_norm": 0.5009744804337983,
      "learning_rate": 3.925745012929125e-05,
      "loss": 11.81,
      "step": 26318
    },
    {
      "epoch": 1.4331747360682854,
      "grad_norm": 0.5576770648042227,
      "learning_rate": 3.925044534708502e-05,
      "loss": 11.8416,
      "step": 26319
    },
    {
      "epoch": 1.4332291900648684,
      "grad_norm": 0.5287430877263224,
      "learning_rate": 3.924344103727752e-05,
      "loss": 11.9291,
      "step": 26320
    },
    {
      "epoch": 1.4332836440614514,
      "grad_norm": 0.5399687620022304,
      "learning_rate": 3.923643719992324e-05,
      "loss": 11.9051,
      "step": 26321
    },
    {
      "epoch": 1.4333380980580344,
      "grad_norm": 0.553287743124179,
      "learning_rate": 3.922943383507662e-05,
      "loss": 11.6551,
      "step": 26322
    },
    {
      "epoch": 1.4333925520546174,
      "grad_norm": 0.5524811046205452,
      "learning_rate": 3.922243094279211e-05,
      "loss": 11.6851,
      "step": 26323
    },
    {
      "epoch": 1.4334470060512003,
      "grad_norm": 0.5337478631145688,
      "learning_rate": 3.9215428523124134e-05,
      "loss": 11.9709,
      "step": 26324
    },
    {
      "epoch": 1.4335014600477833,
      "grad_norm": 0.5352497624881862,
      "learning_rate": 3.920842657612718e-05,
      "loss": 11.8142,
      "step": 26325
    },
    {
      "epoch": 1.4335559140443663,
      "grad_norm": 0.5726735856035453,
      "learning_rate": 3.9201425101855734e-05,
      "loss": 11.9576,
      "step": 26326
    },
    {
      "epoch": 1.4336103680409493,
      "grad_norm": 0.5719626561301574,
      "learning_rate": 3.9194424100364166e-05,
      "loss": 11.8859,
      "step": 26327
    },
    {
      "epoch": 1.4336648220375325,
      "grad_norm": 0.515771477977669,
      "learning_rate": 3.918742357170698e-05,
      "loss": 11.848,
      "step": 26328
    },
    {
      "epoch": 1.4337192760341155,
      "grad_norm": 0.5041864435515397,
      "learning_rate": 3.918042351593857e-05,
      "loss": 11.9247,
      "step": 26329
    },
    {
      "epoch": 1.4337737300306985,
      "grad_norm": 0.5749954366515224,
      "learning_rate": 3.9173423933113405e-05,
      "loss": 11.7697,
      "step": 26330
    },
    {
      "epoch": 1.4338281840272815,
      "grad_norm": 0.5694306716716669,
      "learning_rate": 3.916642482328586e-05,
      "loss": 11.9713,
      "step": 26331
    },
    {
      "epoch": 1.4338826380238645,
      "grad_norm": 0.579256836282674,
      "learning_rate": 3.915942618651045e-05,
      "loss": 11.896,
      "step": 26332
    },
    {
      "epoch": 1.4339370920204475,
      "grad_norm": 0.5563756159888272,
      "learning_rate": 3.915242802284152e-05,
      "loss": 11.9364,
      "step": 26333
    },
    {
      "epoch": 1.4339915460170305,
      "grad_norm": 0.6281289022665371,
      "learning_rate": 3.91454303323335e-05,
      "loss": 11.9363,
      "step": 26334
    },
    {
      "epoch": 1.4340460000136135,
      "grad_norm": 0.5904478348938093,
      "learning_rate": 3.9138433115040804e-05,
      "loss": 11.7967,
      "step": 26335
    },
    {
      "epoch": 1.4341004540101965,
      "grad_norm": 0.5799892327547969,
      "learning_rate": 3.91314363710179e-05,
      "loss": 11.8623,
      "step": 26336
    },
    {
      "epoch": 1.4341549080067795,
      "grad_norm": 0.5405389470553,
      "learning_rate": 3.9124440100319123e-05,
      "loss": 11.9452,
      "step": 26337
    },
    {
      "epoch": 1.4342093620033625,
      "grad_norm": 0.6783720474388546,
      "learning_rate": 3.911744430299895e-05,
      "loss": 11.9873,
      "step": 26338
    },
    {
      "epoch": 1.4342638159999455,
      "grad_norm": 0.6327982039133199,
      "learning_rate": 3.911044897911169e-05,
      "loss": 11.9228,
      "step": 26339
    },
    {
      "epoch": 1.4343182699965284,
      "grad_norm": 0.5590534285869779,
      "learning_rate": 3.910345412871184e-05,
      "loss": 11.891,
      "step": 26340
    },
    {
      "epoch": 1.4343727239931114,
      "grad_norm": 0.5534170392644077,
      "learning_rate": 3.90964597518537e-05,
      "loss": 11.9337,
      "step": 26341
    },
    {
      "epoch": 1.4344271779896947,
      "grad_norm": 0.5288421244190704,
      "learning_rate": 3.9089465848591735e-05,
      "loss": 11.8466,
      "step": 26342
    },
    {
      "epoch": 1.4344816319862776,
      "grad_norm": 0.5962069198937746,
      "learning_rate": 3.90824724189803e-05,
      "loss": 11.8624,
      "step": 26343
    },
    {
      "epoch": 1.4345360859828606,
      "grad_norm": 0.5750638101000574,
      "learning_rate": 3.907547946307374e-05,
      "loss": 11.9572,
      "step": 26344
    },
    {
      "epoch": 1.4345905399794436,
      "grad_norm": 0.5350056698965392,
      "learning_rate": 3.90684869809265e-05,
      "loss": 11.9767,
      "step": 26345
    },
    {
      "epoch": 1.4346449939760266,
      "grad_norm": 0.5163007456316633,
      "learning_rate": 3.906149497259289e-05,
      "loss": 11.7562,
      "step": 26346
    },
    {
      "epoch": 1.4346994479726096,
      "grad_norm": 0.5973698289624069,
      "learning_rate": 3.905450343812732e-05,
      "loss": 11.8655,
      "step": 26347
    },
    {
      "epoch": 1.4347539019691926,
      "grad_norm": 0.5816683959688256,
      "learning_rate": 3.904751237758418e-05,
      "loss": 12.0081,
      "step": 26348
    },
    {
      "epoch": 1.4348083559657756,
      "grad_norm": 0.5248681375402489,
      "learning_rate": 3.904052179101778e-05,
      "loss": 11.8766,
      "step": 26349
    },
    {
      "epoch": 1.4348628099623586,
      "grad_norm": 0.5227799303859438,
      "learning_rate": 3.9033531678482535e-05,
      "loss": 11.8141,
      "step": 26350
    },
    {
      "epoch": 1.4349172639589418,
      "grad_norm": 0.6169144903374221,
      "learning_rate": 3.9026542040032733e-05,
      "loss": 11.8988,
      "step": 26351
    },
    {
      "epoch": 1.4349717179555248,
      "grad_norm": 0.5674696223899558,
      "learning_rate": 3.901955287572281e-05,
      "loss": 11.9535,
      "step": 26352
    },
    {
      "epoch": 1.4350261719521078,
      "grad_norm": 0.5495261823381598,
      "learning_rate": 3.901256418560706e-05,
      "loss": 11.7811,
      "step": 26353
    },
    {
      "epoch": 1.4350806259486908,
      "grad_norm": 0.5634648186678949,
      "learning_rate": 3.900557596973981e-05,
      "loss": 11.9454,
      "step": 26354
    },
    {
      "epoch": 1.4351350799452738,
      "grad_norm": 0.5225187981987457,
      "learning_rate": 3.899858822817546e-05,
      "loss": 11.8261,
      "step": 26355
    },
    {
      "epoch": 1.4351895339418568,
      "grad_norm": 0.4900501069485667,
      "learning_rate": 3.8991600960968285e-05,
      "loss": 11.8859,
      "step": 26356
    },
    {
      "epoch": 1.4352439879384398,
      "grad_norm": 0.5506176697346621,
      "learning_rate": 3.898461416817265e-05,
      "loss": 11.9595,
      "step": 26357
    },
    {
      "epoch": 1.4352984419350228,
      "grad_norm": 0.5510950018951903,
      "learning_rate": 3.8977627849842926e-05,
      "loss": 11.9026,
      "step": 26358
    },
    {
      "epoch": 1.4353528959316058,
      "grad_norm": 0.5596867563777409,
      "learning_rate": 3.8970642006033366e-05,
      "loss": 11.9269,
      "step": 26359
    },
    {
      "epoch": 1.4354073499281887,
      "grad_norm": 0.5554123574041167,
      "learning_rate": 3.896365663679836e-05,
      "loss": 11.6873,
      "step": 26360
    },
    {
      "epoch": 1.4354618039247717,
      "grad_norm": 0.4755923426377247,
      "learning_rate": 3.895667174219216e-05,
      "loss": 11.8354,
      "step": 26361
    },
    {
      "epoch": 1.4355162579213547,
      "grad_norm": 0.5352883198171227,
      "learning_rate": 3.894968732226916e-05,
      "loss": 11.7626,
      "step": 26362
    },
    {
      "epoch": 1.4355707119179377,
      "grad_norm": 0.5174955441146005,
      "learning_rate": 3.8942703377083636e-05,
      "loss": 11.836,
      "step": 26363
    },
    {
      "epoch": 1.4356251659145207,
      "grad_norm": 0.5788548460555466,
      "learning_rate": 3.8935719906689836e-05,
      "loss": 11.9306,
      "step": 26364
    },
    {
      "epoch": 1.435679619911104,
      "grad_norm": 0.5036639279723619,
      "learning_rate": 3.892873691114216e-05,
      "loss": 11.9272,
      "step": 26365
    },
    {
      "epoch": 1.435734073907687,
      "grad_norm": 0.5320876843083453,
      "learning_rate": 3.892175439049484e-05,
      "loss": 11.7054,
      "step": 26366
    },
    {
      "epoch": 1.43578852790427,
      "grad_norm": 0.5157691623536944,
      "learning_rate": 3.891477234480223e-05,
      "loss": 11.9383,
      "step": 26367
    },
    {
      "epoch": 1.435842981900853,
      "grad_norm": 0.5954104878448915,
      "learning_rate": 3.8907790774118555e-05,
      "loss": 11.7849,
      "step": 26368
    },
    {
      "epoch": 1.435897435897436,
      "grad_norm": 0.6063313458480252,
      "learning_rate": 3.8900809678498155e-05,
      "loss": 11.6839,
      "step": 26369
    },
    {
      "epoch": 1.435951889894019,
      "grad_norm": 0.5522728385786745,
      "learning_rate": 3.8893829057995326e-05,
      "loss": 11.861,
      "step": 26370
    },
    {
      "epoch": 1.4360063438906019,
      "grad_norm": 0.5285990004560783,
      "learning_rate": 3.8886848912664306e-05,
      "loss": 11.8667,
      "step": 26371
    },
    {
      "epoch": 1.4360607978871849,
      "grad_norm": 0.5034565597361829,
      "learning_rate": 3.887986924255946e-05,
      "loss": 11.7953,
      "step": 26372
    },
    {
      "epoch": 1.4361152518837679,
      "grad_norm": 0.5746892136454145,
      "learning_rate": 3.887289004773493e-05,
      "loss": 11.9137,
      "step": 26373
    },
    {
      "epoch": 1.436169705880351,
      "grad_norm": 0.6797035662605884,
      "learning_rate": 3.886591132824506e-05,
      "loss": 11.9601,
      "step": 26374
    },
    {
      "epoch": 1.436224159876934,
      "grad_norm": 0.6305808313574126,
      "learning_rate": 3.885893308414417e-05,
      "loss": 11.9089,
      "step": 26375
    },
    {
      "epoch": 1.436278613873517,
      "grad_norm": 0.5573557777157498,
      "learning_rate": 3.885195531548641e-05,
      "loss": 11.8543,
      "step": 26376
    },
    {
      "epoch": 1.4363330678701,
      "grad_norm": 0.6008072377443651,
      "learning_rate": 3.884497802232614e-05,
      "loss": 11.9674,
      "step": 26377
    },
    {
      "epoch": 1.436387521866683,
      "grad_norm": 0.5162928348344781,
      "learning_rate": 3.883800120471754e-05,
      "loss": 11.8639,
      "step": 26378
    },
    {
      "epoch": 1.436441975863266,
      "grad_norm": 0.5573660627832177,
      "learning_rate": 3.883102486271495e-05,
      "loss": 11.9008,
      "step": 26379
    },
    {
      "epoch": 1.436496429859849,
      "grad_norm": 0.6070489743820368,
      "learning_rate": 3.882404899637252e-05,
      "loss": 11.9311,
      "step": 26380
    },
    {
      "epoch": 1.436550883856432,
      "grad_norm": 0.5135783551282166,
      "learning_rate": 3.8817073605744544e-05,
      "loss": 11.8667,
      "step": 26381
    },
    {
      "epoch": 1.436605337853015,
      "grad_norm": 0.5885763562446317,
      "learning_rate": 3.881009869088534e-05,
      "loss": 11.8776,
      "step": 26382
    },
    {
      "epoch": 1.436659791849598,
      "grad_norm": 0.5819358618645047,
      "learning_rate": 3.8803124251849e-05,
      "loss": 11.7464,
      "step": 26383
    },
    {
      "epoch": 1.436714245846181,
      "grad_norm": 0.525154708840969,
      "learning_rate": 3.8796150288689824e-05,
      "loss": 11.8576,
      "step": 26384
    },
    {
      "epoch": 1.436768699842764,
      "grad_norm": 0.5921139619493957,
      "learning_rate": 3.878917680146208e-05,
      "loss": 11.8638,
      "step": 26385
    },
    {
      "epoch": 1.436823153839347,
      "grad_norm": 0.718740564888115,
      "learning_rate": 3.878220379021993e-05,
      "loss": 11.9646,
      "step": 26386
    },
    {
      "epoch": 1.43687760783593,
      "grad_norm": 0.5153265648328553,
      "learning_rate": 3.877523125501767e-05,
      "loss": 11.9061,
      "step": 26387
    },
    {
      "epoch": 1.436932061832513,
      "grad_norm": 0.5316327421807723,
      "learning_rate": 3.876825919590944e-05,
      "loss": 11.9404,
      "step": 26388
    },
    {
      "epoch": 1.4369865158290962,
      "grad_norm": 0.5156205903919038,
      "learning_rate": 3.8761287612949526e-05,
      "loss": 11.9662,
      "step": 26389
    },
    {
      "epoch": 1.4370409698256792,
      "grad_norm": 0.5184465002568762,
      "learning_rate": 3.875431650619208e-05,
      "loss": 11.8669,
      "step": 26390
    },
    {
      "epoch": 1.4370954238222622,
      "grad_norm": 0.5099609041209606,
      "learning_rate": 3.874734587569134e-05,
      "loss": 11.8613,
      "step": 26391
    },
    {
      "epoch": 1.4371498778188452,
      "grad_norm": 0.5144846205085002,
      "learning_rate": 3.874037572150158e-05,
      "loss": 11.7146,
      "step": 26392
    },
    {
      "epoch": 1.4372043318154282,
      "grad_norm": 0.5811415516372281,
      "learning_rate": 3.8733406043676855e-05,
      "loss": 11.8797,
      "step": 26393
    },
    {
      "epoch": 1.4372587858120112,
      "grad_norm": 0.5806101047973045,
      "learning_rate": 3.8726436842271485e-05,
      "loss": 11.9977,
      "step": 26394
    },
    {
      "epoch": 1.4373132398085942,
      "grad_norm": 0.5283532018391751,
      "learning_rate": 3.8719468117339576e-05,
      "loss": 11.8021,
      "step": 26395
    },
    {
      "epoch": 1.4373676938051771,
      "grad_norm": 0.5078686479502453,
      "learning_rate": 3.871249986893536e-05,
      "loss": 11.9617,
      "step": 26396
    },
    {
      "epoch": 1.4374221478017604,
      "grad_norm": 0.6796389836641767,
      "learning_rate": 3.8705532097113064e-05,
      "loss": 11.91,
      "step": 26397
    },
    {
      "epoch": 1.4374766017983434,
      "grad_norm": 0.5635833071322215,
      "learning_rate": 3.869856480192679e-05,
      "loss": 11.8082,
      "step": 26398
    },
    {
      "epoch": 1.4375310557949263,
      "grad_norm": 0.6260701241950422,
      "learning_rate": 3.8691597983430784e-05,
      "loss": 12.012,
      "step": 26399
    },
    {
      "epoch": 1.4375855097915093,
      "grad_norm": 0.5743393146103634,
      "learning_rate": 3.868463164167916e-05,
      "loss": 11.9071,
      "step": 26400
    },
    {
      "epoch": 1.4376399637880923,
      "grad_norm": 0.5868376642187261,
      "learning_rate": 3.867766577672617e-05,
      "loss": 11.9936,
      "step": 26401
    },
    {
      "epoch": 1.4376944177846753,
      "grad_norm": 0.6182174557983814,
      "learning_rate": 3.867070038862592e-05,
      "loss": 11.759,
      "step": 26402
    },
    {
      "epoch": 1.4377488717812583,
      "grad_norm": 0.5351313167747331,
      "learning_rate": 3.866373547743256e-05,
      "loss": 11.7878,
      "step": 26403
    },
    {
      "epoch": 1.4378033257778413,
      "grad_norm": 0.5716894709738916,
      "learning_rate": 3.8656771043200327e-05,
      "loss": 11.9386,
      "step": 26404
    },
    {
      "epoch": 1.4378577797744243,
      "grad_norm": 0.5221652512282031,
      "learning_rate": 3.864980708598328e-05,
      "loss": 11.8306,
      "step": 26405
    },
    {
      "epoch": 1.4379122337710073,
      "grad_norm": 0.7000787175088273,
      "learning_rate": 3.864284360583562e-05,
      "loss": 11.7572,
      "step": 26406
    },
    {
      "epoch": 1.4379666877675903,
      "grad_norm": 0.5853533856279501,
      "learning_rate": 3.8635880602811535e-05,
      "loss": 11.7812,
      "step": 26407
    },
    {
      "epoch": 1.4380211417641733,
      "grad_norm": 0.5571842516686745,
      "learning_rate": 3.86289180769651e-05,
      "loss": 11.8221,
      "step": 26408
    },
    {
      "epoch": 1.4380755957607563,
      "grad_norm": 0.6136674336037315,
      "learning_rate": 3.862195602835053e-05,
      "loss": 11.9586,
      "step": 26409
    },
    {
      "epoch": 1.4381300497573393,
      "grad_norm": 0.492538472415326,
      "learning_rate": 3.861499445702188e-05,
      "loss": 11.771,
      "step": 26410
    },
    {
      "epoch": 1.4381845037539223,
      "grad_norm": 0.6849348467466794,
      "learning_rate": 3.860803336303337e-05,
      "loss": 11.8212,
      "step": 26411
    },
    {
      "epoch": 1.4382389577505055,
      "grad_norm": 0.567441722797438,
      "learning_rate": 3.860107274643908e-05,
      "loss": 11.9102,
      "step": 26412
    },
    {
      "epoch": 1.4382934117470885,
      "grad_norm": 0.5350889633818435,
      "learning_rate": 3.859411260729311e-05,
      "loss": 11.7683,
      "step": 26413
    },
    {
      "epoch": 1.4383478657436715,
      "grad_norm": 0.5794163374270717,
      "learning_rate": 3.8587152945649664e-05,
      "loss": 11.9231,
      "step": 26414
    },
    {
      "epoch": 1.4384023197402545,
      "grad_norm": 0.5222541612633167,
      "learning_rate": 3.8580193761562764e-05,
      "loss": 11.9064,
      "step": 26415
    },
    {
      "epoch": 1.4384567737368374,
      "grad_norm": 0.5449803229054676,
      "learning_rate": 3.857323505508663e-05,
      "loss": 11.8881,
      "step": 26416
    },
    {
      "epoch": 1.4385112277334204,
      "grad_norm": 0.5909731847978851,
      "learning_rate": 3.856627682627527e-05,
      "loss": 11.8117,
      "step": 26417
    },
    {
      "epoch": 1.4385656817300034,
      "grad_norm": 0.5943101138668914,
      "learning_rate": 3.8559319075182855e-05,
      "loss": 11.8565,
      "step": 26418
    },
    {
      "epoch": 1.4386201357265864,
      "grad_norm": 0.48160549429955773,
      "learning_rate": 3.855236180186351e-05,
      "loss": 11.8283,
      "step": 26419
    },
    {
      "epoch": 1.4386745897231694,
      "grad_norm": 0.5934331987570598,
      "learning_rate": 3.854540500637127e-05,
      "loss": 11.9252,
      "step": 26420
    },
    {
      "epoch": 1.4387290437197526,
      "grad_norm": 0.5368310818555759,
      "learning_rate": 3.853844868876031e-05,
      "loss": 11.8327,
      "step": 26421
    },
    {
      "epoch": 1.4387834977163356,
      "grad_norm": 0.5534315381139365,
      "learning_rate": 3.853149284908466e-05,
      "loss": 11.8371,
      "step": 26422
    },
    {
      "epoch": 1.4388379517129186,
      "grad_norm": 0.5518777788488775,
      "learning_rate": 3.852453748739841e-05,
      "loss": 11.9097,
      "step": 26423
    },
    {
      "epoch": 1.4388924057095016,
      "grad_norm": 0.5599390786515479,
      "learning_rate": 3.8517582603755696e-05,
      "loss": 12.0061,
      "step": 26424
    },
    {
      "epoch": 1.4389468597060846,
      "grad_norm": 0.5794946414300994,
      "learning_rate": 3.851062819821054e-05,
      "loss": 11.8319,
      "step": 26425
    },
    {
      "epoch": 1.4390013137026676,
      "grad_norm": 0.5266736447050263,
      "learning_rate": 3.850367427081708e-05,
      "loss": 11.7306,
      "step": 26426
    },
    {
      "epoch": 1.4390557676992506,
      "grad_norm": 0.5496319407583486,
      "learning_rate": 3.8496720821629326e-05,
      "loss": 11.915,
      "step": 26427
    },
    {
      "epoch": 1.4391102216958336,
      "grad_norm": 0.5414984675723962,
      "learning_rate": 3.848976785070143e-05,
      "loss": 11.8731,
      "step": 26428
    },
    {
      "epoch": 1.4391646756924166,
      "grad_norm": 0.5293628663317904,
      "learning_rate": 3.848281535808738e-05,
      "loss": 11.7896,
      "step": 26429
    },
    {
      "epoch": 1.4392191296889996,
      "grad_norm": 0.6275777157507865,
      "learning_rate": 3.8475863343841255e-05,
      "loss": 11.9586,
      "step": 26430
    },
    {
      "epoch": 1.4392735836855826,
      "grad_norm": 0.5526135047547414,
      "learning_rate": 3.8468911808017184e-05,
      "loss": 11.8154,
      "step": 26431
    },
    {
      "epoch": 1.4393280376821656,
      "grad_norm": 0.5011062321628569,
      "learning_rate": 3.846196075066917e-05,
      "loss": 11.9022,
      "step": 26432
    },
    {
      "epoch": 1.4393824916787485,
      "grad_norm": 0.5148144995516717,
      "learning_rate": 3.845501017185123e-05,
      "loss": 11.795,
      "step": 26433
    },
    {
      "epoch": 1.4394369456753315,
      "grad_norm": 0.5343934689759077,
      "learning_rate": 3.8448060071617496e-05,
      "loss": 11.7295,
      "step": 26434
    },
    {
      "epoch": 1.4394913996719148,
      "grad_norm": 0.5310400119548662,
      "learning_rate": 3.844111045002193e-05,
      "loss": 11.7665,
      "step": 26435
    },
    {
      "epoch": 1.4395458536684977,
      "grad_norm": 0.5262723138066365,
      "learning_rate": 3.8434161307118655e-05,
      "loss": 11.796,
      "step": 26436
    },
    {
      "epoch": 1.4396003076650807,
      "grad_norm": 0.5262234443805276,
      "learning_rate": 3.842721264296162e-05,
      "loss": 11.7942,
      "step": 26437
    },
    {
      "epoch": 1.4396547616616637,
      "grad_norm": 0.5552265090171387,
      "learning_rate": 3.8420264457604946e-05,
      "loss": 11.858,
      "step": 26438
    },
    {
      "epoch": 1.4397092156582467,
      "grad_norm": 0.5284576041256092,
      "learning_rate": 3.841331675110259e-05,
      "loss": 11.9226,
      "step": 26439
    },
    {
      "epoch": 1.4397636696548297,
      "grad_norm": 0.528154978750724,
      "learning_rate": 3.84063695235086e-05,
      "loss": 11.8791,
      "step": 26440
    },
    {
      "epoch": 1.4398181236514127,
      "grad_norm": 0.5375114336456656,
      "learning_rate": 3.839942277487706e-05,
      "loss": 11.9747,
      "step": 26441
    },
    {
      "epoch": 1.4398725776479957,
      "grad_norm": 0.6520639507404585,
      "learning_rate": 3.839247650526192e-05,
      "loss": 11.9064,
      "step": 26442
    },
    {
      "epoch": 1.4399270316445787,
      "grad_norm": 0.4965413455352172,
      "learning_rate": 3.8385530714717235e-05,
      "loss": 11.8684,
      "step": 26443
    },
    {
      "epoch": 1.439981485641162,
      "grad_norm": 0.5642819911632675,
      "learning_rate": 3.837858540329694e-05,
      "loss": 11.9912,
      "step": 26444
    },
    {
      "epoch": 1.440035939637745,
      "grad_norm": 0.5705476875390401,
      "learning_rate": 3.837164057105511e-05,
      "loss": 12.0032,
      "step": 26445
    },
    {
      "epoch": 1.440090393634328,
      "grad_norm": 0.516073449236866,
      "learning_rate": 3.836469621804578e-05,
      "loss": 11.6778,
      "step": 26446
    },
    {
      "epoch": 1.4401448476309109,
      "grad_norm": 0.5273706663052106,
      "learning_rate": 3.835775234432286e-05,
      "loss": 11.8991,
      "step": 26447
    },
    {
      "epoch": 1.4401993016274939,
      "grad_norm": 0.516940943447397,
      "learning_rate": 3.8350808949940444e-05,
      "loss": 11.9498,
      "step": 26448
    },
    {
      "epoch": 1.4402537556240769,
      "grad_norm": 0.6129870866541962,
      "learning_rate": 3.8343866034952426e-05,
      "loss": 11.9421,
      "step": 26449
    },
    {
      "epoch": 1.4403082096206599,
      "grad_norm": 0.5000039200400039,
      "learning_rate": 3.8336923599412886e-05,
      "loss": 11.9062,
      "step": 26450
    },
    {
      "epoch": 1.4403626636172429,
      "grad_norm": 0.5895570357586272,
      "learning_rate": 3.832998164337574e-05,
      "loss": 11.8764,
      "step": 26451
    },
    {
      "epoch": 1.4404171176138258,
      "grad_norm": 0.5096744817429743,
      "learning_rate": 3.8323040166894996e-05,
      "loss": 11.9575,
      "step": 26452
    },
    {
      "epoch": 1.4404715716104088,
      "grad_norm": 0.5210919444623549,
      "learning_rate": 3.83160991700247e-05,
      "loss": 11.8493,
      "step": 26453
    },
    {
      "epoch": 1.4405260256069918,
      "grad_norm": 0.566028201220997,
      "learning_rate": 3.83091586528187e-05,
      "loss": 11.9164,
      "step": 26454
    },
    {
      "epoch": 1.4405804796035748,
      "grad_norm": 0.5592746555967717,
      "learning_rate": 3.830221861533102e-05,
      "loss": 11.7384,
      "step": 26455
    },
    {
      "epoch": 1.4406349336001578,
      "grad_norm": 0.49807439105786255,
      "learning_rate": 3.829527905761567e-05,
      "loss": 11.8128,
      "step": 26456
    },
    {
      "epoch": 1.4406893875967408,
      "grad_norm": 0.4993852580956869,
      "learning_rate": 3.828833997972655e-05,
      "loss": 11.9153,
      "step": 26457
    },
    {
      "epoch": 1.4407438415933238,
      "grad_norm": 0.5091481273852589,
      "learning_rate": 3.828140138171767e-05,
      "loss": 11.7583,
      "step": 26458
    },
    {
      "epoch": 1.440798295589907,
      "grad_norm": 0.5316267220699452,
      "learning_rate": 3.827446326364295e-05,
      "loss": 11.8889,
      "step": 26459
    },
    {
      "epoch": 1.44085274958649,
      "grad_norm": 0.5049445355096555,
      "learning_rate": 3.8267525625556363e-05,
      "loss": 11.7802,
      "step": 26460
    },
    {
      "epoch": 1.440907203583073,
      "grad_norm": 0.4884318562691344,
      "learning_rate": 3.8260588467511824e-05,
      "loss": 11.8209,
      "step": 26461
    },
    {
      "epoch": 1.440961657579656,
      "grad_norm": 0.5591009650182845,
      "learning_rate": 3.8253651789563316e-05,
      "loss": 11.7407,
      "step": 26462
    },
    {
      "epoch": 1.441016111576239,
      "grad_norm": 0.5713114332985573,
      "learning_rate": 3.8246715591764825e-05,
      "loss": 11.8233,
      "step": 26463
    },
    {
      "epoch": 1.441070565572822,
      "grad_norm": 0.5529993521007889,
      "learning_rate": 3.823977987417016e-05,
      "loss": 11.793,
      "step": 26464
    },
    {
      "epoch": 1.441125019569405,
      "grad_norm": 0.5700467073177319,
      "learning_rate": 3.8232844636833364e-05,
      "loss": 11.8365,
      "step": 26465
    },
    {
      "epoch": 1.441179473565988,
      "grad_norm": 0.557722646863616,
      "learning_rate": 3.8225909879808285e-05,
      "loss": 11.7904,
      "step": 26466
    },
    {
      "epoch": 1.4412339275625712,
      "grad_norm": 0.5416506466773395,
      "learning_rate": 3.8218975603148885e-05,
      "loss": 11.9212,
      "step": 26467
    },
    {
      "epoch": 1.4412883815591542,
      "grad_norm": 0.4982781364953499,
      "learning_rate": 3.821204180690914e-05,
      "loss": 11.7878,
      "step": 26468
    },
    {
      "epoch": 1.4413428355557372,
      "grad_norm": 0.5143642027104275,
      "learning_rate": 3.820510849114288e-05,
      "loss": 11.8249,
      "step": 26469
    },
    {
      "epoch": 1.4413972895523202,
      "grad_norm": 0.5456973299290206,
      "learning_rate": 3.819817565590409e-05,
      "loss": 11.7678,
      "step": 26470
    },
    {
      "epoch": 1.4414517435489032,
      "grad_norm": 0.4926600743195348,
      "learning_rate": 3.8191243301246616e-05,
      "loss": 11.8152,
      "step": 26471
    },
    {
      "epoch": 1.4415061975454861,
      "grad_norm": 0.5032933729527044,
      "learning_rate": 3.818431142722444e-05,
      "loss": 11.8554,
      "step": 26472
    },
    {
      "epoch": 1.4415606515420691,
      "grad_norm": 0.5550752795060577,
      "learning_rate": 3.817738003389142e-05,
      "loss": 11.8841,
      "step": 26473
    },
    {
      "epoch": 1.4416151055386521,
      "grad_norm": 0.5282346410220012,
      "learning_rate": 3.817044912130143e-05,
      "loss": 11.8373,
      "step": 26474
    },
    {
      "epoch": 1.4416695595352351,
      "grad_norm": 0.6017067112650445,
      "learning_rate": 3.8163518689508425e-05,
      "loss": 11.9226,
      "step": 26475
    },
    {
      "epoch": 1.4417240135318181,
      "grad_norm": 0.4762906312380366,
      "learning_rate": 3.8156588738566245e-05,
      "loss": 11.8273,
      "step": 26476
    },
    {
      "epoch": 1.441778467528401,
      "grad_norm": 0.5889751542683911,
      "learning_rate": 3.8149659268528824e-05,
      "loss": 11.9261,
      "step": 26477
    },
    {
      "epoch": 1.441832921524984,
      "grad_norm": 0.5465175591594423,
      "learning_rate": 3.814273027945e-05,
      "loss": 11.9285,
      "step": 26478
    },
    {
      "epoch": 1.441887375521567,
      "grad_norm": 0.5498044555217565,
      "learning_rate": 3.8135801771383674e-05,
      "loss": 11.845,
      "step": 26479
    },
    {
      "epoch": 1.44194182951815,
      "grad_norm": 0.5569685373284927,
      "learning_rate": 3.812887374438376e-05,
      "loss": 11.8335,
      "step": 26480
    },
    {
      "epoch": 1.441996283514733,
      "grad_norm": 0.5465969630641462,
      "learning_rate": 3.8121946198504066e-05,
      "loss": 11.9757,
      "step": 26481
    },
    {
      "epoch": 1.4420507375113163,
      "grad_norm": 0.49155857069869185,
      "learning_rate": 3.811501913379853e-05,
      "loss": 11.8822,
      "step": 26482
    },
    {
      "epoch": 1.4421051915078993,
      "grad_norm": 0.5603115641034294,
      "learning_rate": 3.8108092550320985e-05,
      "loss": 11.9908,
      "step": 26483
    },
    {
      "epoch": 1.4421596455044823,
      "grad_norm": 0.5073353673802056,
      "learning_rate": 3.810116644812526e-05,
      "loss": 11.9003,
      "step": 26484
    },
    {
      "epoch": 1.4422140995010653,
      "grad_norm": 0.5392727103105533,
      "learning_rate": 3.809424082726528e-05,
      "loss": 11.9213,
      "step": 26485
    },
    {
      "epoch": 1.4422685534976483,
      "grad_norm": 0.5492288241804698,
      "learning_rate": 3.8087315687794824e-05,
      "loss": 11.8974,
      "step": 26486
    },
    {
      "epoch": 1.4423230074942313,
      "grad_norm": 0.5193820078847496,
      "learning_rate": 3.8080391029767825e-05,
      "loss": 11.9137,
      "step": 26487
    },
    {
      "epoch": 1.4423774614908142,
      "grad_norm": 0.5612989517853024,
      "learning_rate": 3.807346685323805e-05,
      "loss": 11.8743,
      "step": 26488
    },
    {
      "epoch": 1.4424319154873972,
      "grad_norm": 0.5516156638909567,
      "learning_rate": 3.806654315825938e-05,
      "loss": 11.9698,
      "step": 26489
    },
    {
      "epoch": 1.4424863694839802,
      "grad_norm": 0.533090466164547,
      "learning_rate": 3.805961994488569e-05,
      "loss": 11.8209,
      "step": 26490
    },
    {
      "epoch": 1.4425408234805634,
      "grad_norm": 0.5513457231363742,
      "learning_rate": 3.8052697213170763e-05,
      "loss": 11.8851,
      "step": 26491
    },
    {
      "epoch": 1.4425952774771464,
      "grad_norm": 0.5166375002092177,
      "learning_rate": 3.8045774963168465e-05,
      "loss": 11.9074,
      "step": 26492
    },
    {
      "epoch": 1.4426497314737294,
      "grad_norm": 0.5589315034614036,
      "learning_rate": 3.8038853194932636e-05,
      "loss": 11.8397,
      "step": 26493
    },
    {
      "epoch": 1.4427041854703124,
      "grad_norm": 0.6089769791437563,
      "learning_rate": 3.803193190851702e-05,
      "loss": 12.0076,
      "step": 26494
    },
    {
      "epoch": 1.4427586394668954,
      "grad_norm": 0.5698836597910661,
      "learning_rate": 3.802501110397553e-05,
      "loss": 11.8266,
      "step": 26495
    },
    {
      "epoch": 1.4428130934634784,
      "grad_norm": 0.5435638151684871,
      "learning_rate": 3.8018090781361914e-05,
      "loss": 11.887,
      "step": 26496
    },
    {
      "epoch": 1.4428675474600614,
      "grad_norm": 0.5469715169926146,
      "learning_rate": 3.8011170940730056e-05,
      "loss": 11.8753,
      "step": 26497
    },
    {
      "epoch": 1.4429220014566444,
      "grad_norm": 0.518215807742917,
      "learning_rate": 3.80042515821337e-05,
      "loss": 11.8149,
      "step": 26498
    },
    {
      "epoch": 1.4429764554532274,
      "grad_norm": 0.5426434446045338,
      "learning_rate": 3.799733270562671e-05,
      "loss": 11.9739,
      "step": 26499
    },
    {
      "epoch": 1.4430309094498104,
      "grad_norm": 0.5954430547312034,
      "learning_rate": 3.7990414311262815e-05,
      "loss": 11.8652,
      "step": 26500
    },
    {
      "epoch": 1.4430853634463934,
      "grad_norm": 0.5290342827964081,
      "learning_rate": 3.7983496399095865e-05,
      "loss": 11.8137,
      "step": 26501
    },
    {
      "epoch": 1.4431398174429764,
      "grad_norm": 0.5430989247657333,
      "learning_rate": 3.797657896917968e-05,
      "loss": 11.9709,
      "step": 26502
    },
    {
      "epoch": 1.4431942714395594,
      "grad_norm": 0.547745476687581,
      "learning_rate": 3.796966202156802e-05,
      "loss": 11.9644,
      "step": 26503
    },
    {
      "epoch": 1.4432487254361424,
      "grad_norm": 0.5779703008891022,
      "learning_rate": 3.7962745556314636e-05,
      "loss": 11.9251,
      "step": 26504
    },
    {
      "epoch": 1.4433031794327256,
      "grad_norm": 0.5508320217686239,
      "learning_rate": 3.795582957347338e-05,
      "loss": 11.8728,
      "step": 26505
    },
    {
      "epoch": 1.4433576334293086,
      "grad_norm": 0.5561825859949718,
      "learning_rate": 3.7948914073097964e-05,
      "loss": 11.9296,
      "step": 26506
    },
    {
      "epoch": 1.4434120874258916,
      "grad_norm": 0.5901528433694065,
      "learning_rate": 3.7941999055242236e-05,
      "loss": 11.757,
      "step": 26507
    },
    {
      "epoch": 1.4434665414224745,
      "grad_norm": 0.5434915409575033,
      "learning_rate": 3.793508451995989e-05,
      "loss": 11.8442,
      "step": 26508
    },
    {
      "epoch": 1.4435209954190575,
      "grad_norm": 0.5755886356664981,
      "learning_rate": 3.792817046730477e-05,
      "loss": 11.9183,
      "step": 26509
    },
    {
      "epoch": 1.4435754494156405,
      "grad_norm": 0.5414810281444997,
      "learning_rate": 3.792125689733057e-05,
      "loss": 11.7676,
      "step": 26510
    },
    {
      "epoch": 1.4436299034122235,
      "grad_norm": 0.5362016108857978,
      "learning_rate": 3.791434381009109e-05,
      "loss": 11.839,
      "step": 26511
    },
    {
      "epoch": 1.4436843574088065,
      "grad_norm": 0.5387616562075546,
      "learning_rate": 3.790743120564012e-05,
      "loss": 11.8144,
      "step": 26512
    },
    {
      "epoch": 1.4437388114053895,
      "grad_norm": 0.5298039663680805,
      "learning_rate": 3.790051908403138e-05,
      "loss": 11.8176,
      "step": 26513
    },
    {
      "epoch": 1.4437932654019727,
      "grad_norm": 0.6147624955791243,
      "learning_rate": 3.789360744531861e-05,
      "loss": 11.812,
      "step": 26514
    },
    {
      "epoch": 1.4438477193985557,
      "grad_norm": 0.5303930286967303,
      "learning_rate": 3.788669628955554e-05,
      "loss": 11.7855,
      "step": 26515
    },
    {
      "epoch": 1.4439021733951387,
      "grad_norm": 0.5383850553182316,
      "learning_rate": 3.787978561679593e-05,
      "loss": 11.8885,
      "step": 26516
    },
    {
      "epoch": 1.4439566273917217,
      "grad_norm": 0.5290483551962385,
      "learning_rate": 3.7872875427093554e-05,
      "loss": 11.9994,
      "step": 26517
    },
    {
      "epoch": 1.4440110813883047,
      "grad_norm": 0.5625168397314703,
      "learning_rate": 3.786596572050209e-05,
      "loss": 11.8345,
      "step": 26518
    },
    {
      "epoch": 1.4440655353848877,
      "grad_norm": 0.5108397956744112,
      "learning_rate": 3.7859056497075326e-05,
      "loss": 11.8848,
      "step": 26519
    },
    {
      "epoch": 1.4441199893814707,
      "grad_norm": 0.5683020133212698,
      "learning_rate": 3.785214775686693e-05,
      "loss": 11.7156,
      "step": 26520
    },
    {
      "epoch": 1.4441744433780537,
      "grad_norm": 0.548609603558456,
      "learning_rate": 3.78452394999307e-05,
      "loss": 11.7671,
      "step": 26521
    },
    {
      "epoch": 1.4442288973746367,
      "grad_norm": 0.5811885727393167,
      "learning_rate": 3.7838331726320254e-05,
      "loss": 11.8624,
      "step": 26522
    },
    {
      "epoch": 1.4442833513712197,
      "grad_norm": 0.5426540756261988,
      "learning_rate": 3.7831424436089415e-05,
      "loss": 11.8728,
      "step": 26523
    },
    {
      "epoch": 1.4443378053678027,
      "grad_norm": 0.5602572778176816,
      "learning_rate": 3.7824517629291835e-05,
      "loss": 11.8197,
      "step": 26524
    },
    {
      "epoch": 1.4443922593643856,
      "grad_norm": 0.5192908691092869,
      "learning_rate": 3.7817611305981205e-05,
      "loss": 11.8315,
      "step": 26525
    },
    {
      "epoch": 1.4444467133609686,
      "grad_norm": 0.534263257596746,
      "learning_rate": 3.781070546621129e-05,
      "loss": 11.889,
      "step": 26526
    },
    {
      "epoch": 1.4445011673575516,
      "grad_norm": 0.5953627627965478,
      "learning_rate": 3.7803800110035725e-05,
      "loss": 11.9301,
      "step": 26527
    },
    {
      "epoch": 1.4445556213541346,
      "grad_norm": 0.6921158756832816,
      "learning_rate": 3.7796895237508245e-05,
      "loss": 12.0371,
      "step": 26528
    },
    {
      "epoch": 1.4446100753507178,
      "grad_norm": 0.5317652565629187,
      "learning_rate": 3.778999084868257e-05,
      "loss": 11.9459,
      "step": 26529
    },
    {
      "epoch": 1.4446645293473008,
      "grad_norm": 0.5907446369727426,
      "learning_rate": 3.7783086943612324e-05,
      "loss": 11.8588,
      "step": 26530
    },
    {
      "epoch": 1.4447189833438838,
      "grad_norm": 0.5225637083776867,
      "learning_rate": 3.777618352235125e-05,
      "loss": 11.8373,
      "step": 26531
    },
    {
      "epoch": 1.4447734373404668,
      "grad_norm": 0.5645536273713637,
      "learning_rate": 3.7769280584952994e-05,
      "loss": 11.9279,
      "step": 26532
    },
    {
      "epoch": 1.4448278913370498,
      "grad_norm": 0.519573391850562,
      "learning_rate": 3.7762378131471266e-05,
      "loss": 11.6787,
      "step": 26533
    },
    {
      "epoch": 1.4448823453336328,
      "grad_norm": 0.5483649445711375,
      "learning_rate": 3.7755476161959736e-05,
      "loss": 11.9166,
      "step": 26534
    },
    {
      "epoch": 1.4449367993302158,
      "grad_norm": 0.5154928719869454,
      "learning_rate": 3.7748574676472016e-05,
      "loss": 11.8932,
      "step": 26535
    },
    {
      "epoch": 1.4449912533267988,
      "grad_norm": 0.581986857093285,
      "learning_rate": 3.7741673675061865e-05,
      "loss": 11.9404,
      "step": 26536
    },
    {
      "epoch": 1.445045707323382,
      "grad_norm": 0.5947792692617778,
      "learning_rate": 3.7734773157782854e-05,
      "loss": 11.8795,
      "step": 26537
    },
    {
      "epoch": 1.445100161319965,
      "grad_norm": 0.5449383458016724,
      "learning_rate": 3.772787312468869e-05,
      "loss": 12.0347,
      "step": 26538
    },
    {
      "epoch": 1.445154615316548,
      "grad_norm": 0.5352143213393263,
      "learning_rate": 3.7720973575833075e-05,
      "loss": 11.8536,
      "step": 26539
    },
    {
      "epoch": 1.445209069313131,
      "grad_norm": 0.545495224596101,
      "learning_rate": 3.771407451126957e-05,
      "loss": 11.8957,
      "step": 26540
    },
    {
      "epoch": 1.445263523309714,
      "grad_norm": 0.5775567875249777,
      "learning_rate": 3.7707175931051896e-05,
      "loss": 11.9274,
      "step": 26541
    },
    {
      "epoch": 1.445317977306297,
      "grad_norm": 0.574280040621081,
      "learning_rate": 3.770027783523364e-05,
      "loss": 11.9951,
      "step": 26542
    },
    {
      "epoch": 1.44537243130288,
      "grad_norm": 0.5640495140559315,
      "learning_rate": 3.769338022386851e-05,
      "loss": 11.8683,
      "step": 26543
    },
    {
      "epoch": 1.445426885299463,
      "grad_norm": 0.5200569374135725,
      "learning_rate": 3.768648309701009e-05,
      "loss": 11.8488,
      "step": 26544
    },
    {
      "epoch": 1.445481339296046,
      "grad_norm": 0.5270282307754426,
      "learning_rate": 3.767958645471201e-05,
      "loss": 11.9137,
      "step": 26545
    },
    {
      "epoch": 1.445535793292629,
      "grad_norm": 0.5133276504647482,
      "learning_rate": 3.767269029702795e-05,
      "loss": 11.8536,
      "step": 26546
    },
    {
      "epoch": 1.445590247289212,
      "grad_norm": 0.5383081779076734,
      "learning_rate": 3.766579462401146e-05,
      "loss": 11.96,
      "step": 26547
    },
    {
      "epoch": 1.445644701285795,
      "grad_norm": 0.5721004487225891,
      "learning_rate": 3.7658899435716245e-05,
      "loss": 11.8655,
      "step": 26548
    },
    {
      "epoch": 1.445699155282378,
      "grad_norm": 0.46913941220405814,
      "learning_rate": 3.7652004732195834e-05,
      "loss": 11.6833,
      "step": 26549
    },
    {
      "epoch": 1.445753609278961,
      "grad_norm": 0.5752380497709699,
      "learning_rate": 3.76451105135039e-05,
      "loss": 11.8896,
      "step": 26550
    },
    {
      "epoch": 1.445808063275544,
      "grad_norm": 0.5650594853578803,
      "learning_rate": 3.763821677969408e-05,
      "loss": 11.835,
      "step": 26551
    },
    {
      "epoch": 1.4458625172721271,
      "grad_norm": 0.5662700850432301,
      "learning_rate": 3.7631323530819905e-05,
      "loss": 11.8166,
      "step": 26552
    },
    {
      "epoch": 1.44591697126871,
      "grad_norm": 0.5807502724892819,
      "learning_rate": 3.762443076693506e-05,
      "loss": 11.746,
      "step": 26553
    },
    {
      "epoch": 1.445971425265293,
      "grad_norm": 0.5565076886180103,
      "learning_rate": 3.7617538488093094e-05,
      "loss": 11.881,
      "step": 26554
    },
    {
      "epoch": 1.446025879261876,
      "grad_norm": 0.5800596085095544,
      "learning_rate": 3.761064669434758e-05,
      "loss": 11.9344,
      "step": 26555
    },
    {
      "epoch": 1.446080333258459,
      "grad_norm": 0.5521810673268931,
      "learning_rate": 3.7603755385752185e-05,
      "loss": 11.94,
      "step": 26556
    },
    {
      "epoch": 1.446134787255042,
      "grad_norm": 0.523797652176005,
      "learning_rate": 3.75968645623604e-05,
      "loss": 11.9031,
      "step": 26557
    },
    {
      "epoch": 1.446189241251625,
      "grad_norm": 0.5597867020012961,
      "learning_rate": 3.75899742242259e-05,
      "loss": 11.8146,
      "step": 26558
    },
    {
      "epoch": 1.446243695248208,
      "grad_norm": 0.5392480061645936,
      "learning_rate": 3.758308437140219e-05,
      "loss": 11.7328,
      "step": 26559
    },
    {
      "epoch": 1.446298149244791,
      "grad_norm": 0.6091671909538555,
      "learning_rate": 3.757619500394289e-05,
      "loss": 12.0145,
      "step": 26560
    },
    {
      "epoch": 1.4463526032413743,
      "grad_norm": 0.5692700371527594,
      "learning_rate": 3.756930612190159e-05,
      "loss": 11.8579,
      "step": 26561
    },
    {
      "epoch": 1.4464070572379573,
      "grad_norm": 0.6063908307754797,
      "learning_rate": 3.7562417725331814e-05,
      "loss": 11.8455,
      "step": 26562
    },
    {
      "epoch": 1.4464615112345403,
      "grad_norm": 0.594514892602684,
      "learning_rate": 3.755552981428722e-05,
      "loss": 11.8091,
      "step": 26563
    },
    {
      "epoch": 1.4465159652311232,
      "grad_norm": 0.5557790256658615,
      "learning_rate": 3.754864238882121e-05,
      "loss": 11.8945,
      "step": 26564
    },
    {
      "epoch": 1.4465704192277062,
      "grad_norm": 0.521261547004267,
      "learning_rate": 3.754175544898745e-05,
      "loss": 11.8221,
      "step": 26565
    },
    {
      "epoch": 1.4466248732242892,
      "grad_norm": 0.5929063166523696,
      "learning_rate": 3.753486899483949e-05,
      "loss": 11.9031,
      "step": 26566
    },
    {
      "epoch": 1.4466793272208722,
      "grad_norm": 0.4929755068670029,
      "learning_rate": 3.7527983026430834e-05,
      "loss": 12.0085,
      "step": 26567
    },
    {
      "epoch": 1.4467337812174552,
      "grad_norm": 0.5439263161014277,
      "learning_rate": 3.752109754381511e-05,
      "loss": 11.9281,
      "step": 26568
    },
    {
      "epoch": 1.4467882352140382,
      "grad_norm": 0.5373762935560514,
      "learning_rate": 3.751421254704576e-05,
      "loss": 11.9186,
      "step": 26569
    },
    {
      "epoch": 1.4468426892106212,
      "grad_norm": 0.5807048933876086,
      "learning_rate": 3.75073280361764e-05,
      "loss": 11.9166,
      "step": 26570
    },
    {
      "epoch": 1.4468971432072042,
      "grad_norm": 0.5481809421062326,
      "learning_rate": 3.7500444011260515e-05,
      "loss": 11.8524,
      "step": 26571
    },
    {
      "epoch": 1.4469515972037872,
      "grad_norm": 0.587755210628643,
      "learning_rate": 3.749356047235165e-05,
      "loss": 12.0914,
      "step": 26572
    },
    {
      "epoch": 1.4470060512003702,
      "grad_norm": 0.5665313473285608,
      "learning_rate": 3.7486677419503427e-05,
      "loss": 11.8461,
      "step": 26573
    },
    {
      "epoch": 1.4470605051969532,
      "grad_norm": 0.5106968799287922,
      "learning_rate": 3.74797948527692e-05,
      "loss": 11.8505,
      "step": 26574
    },
    {
      "epoch": 1.4471149591935364,
      "grad_norm": 0.5304620584865829,
      "learning_rate": 3.7472912772202605e-05,
      "loss": 11.8942,
      "step": 26575
    },
    {
      "epoch": 1.4471694131901194,
      "grad_norm": 0.5264843733801304,
      "learning_rate": 3.74660311778571e-05,
      "loss": 11.8378,
      "step": 26576
    },
    {
      "epoch": 1.4472238671867024,
      "grad_norm": 0.529979963182147,
      "learning_rate": 3.7459150069786216e-05,
      "loss": 11.838,
      "step": 26577
    },
    {
      "epoch": 1.4472783211832854,
      "grad_norm": 0.5467720129741784,
      "learning_rate": 3.745226944804352e-05,
      "loss": 11.8257,
      "step": 26578
    },
    {
      "epoch": 1.4473327751798684,
      "grad_norm": 0.533493764804981,
      "learning_rate": 3.744538931268241e-05,
      "loss": 11.8858,
      "step": 26579
    },
    {
      "epoch": 1.4473872291764514,
      "grad_norm": 0.5378775567100834,
      "learning_rate": 3.7438509663756494e-05,
      "loss": 11.8686,
      "step": 26580
    },
    {
      "epoch": 1.4474416831730343,
      "grad_norm": 0.5566027107870536,
      "learning_rate": 3.743163050131917e-05,
      "loss": 11.8811,
      "step": 26581
    },
    {
      "epoch": 1.4474961371696173,
      "grad_norm": 0.5515944992876728,
      "learning_rate": 3.742475182542403e-05,
      "loss": 11.8411,
      "step": 26582
    },
    {
      "epoch": 1.4475505911662003,
      "grad_norm": 0.552049968058611,
      "learning_rate": 3.74178736361245e-05,
      "loss": 11.8082,
      "step": 26583
    },
    {
      "epoch": 1.4476050451627835,
      "grad_norm": 0.537822660831308,
      "learning_rate": 3.741099593347406e-05,
      "loss": 11.9451,
      "step": 26584
    },
    {
      "epoch": 1.4476594991593665,
      "grad_norm": 0.5853675462032129,
      "learning_rate": 3.740411871752622e-05,
      "loss": 11.9511,
      "step": 26585
    },
    {
      "epoch": 1.4477139531559495,
      "grad_norm": 0.5226096967266199,
      "learning_rate": 3.739724198833444e-05,
      "loss": 11.8439,
      "step": 26586
    },
    {
      "epoch": 1.4477684071525325,
      "grad_norm": 0.5312786117885538,
      "learning_rate": 3.739036574595221e-05,
      "loss": 11.9,
      "step": 26587
    },
    {
      "epoch": 1.4478228611491155,
      "grad_norm": 0.6276620068702384,
      "learning_rate": 3.7383489990433005e-05,
      "loss": 11.8458,
      "step": 26588
    },
    {
      "epoch": 1.4478773151456985,
      "grad_norm": 0.680144299531745,
      "learning_rate": 3.737661472183026e-05,
      "loss": 12.0234,
      "step": 26589
    },
    {
      "epoch": 1.4479317691422815,
      "grad_norm": 0.5355976618891843,
      "learning_rate": 3.73697399401975e-05,
      "loss": 11.8629,
      "step": 26590
    },
    {
      "epoch": 1.4479862231388645,
      "grad_norm": 0.5919818661633061,
      "learning_rate": 3.736286564558811e-05,
      "loss": 11.9697,
      "step": 26591
    },
    {
      "epoch": 1.4480406771354475,
      "grad_norm": 0.5310696870018928,
      "learning_rate": 3.7355991838055606e-05,
      "loss": 11.8087,
      "step": 26592
    },
    {
      "epoch": 1.4480951311320305,
      "grad_norm": 0.5297474909597147,
      "learning_rate": 3.734911851765339e-05,
      "loss": 11.8513,
      "step": 26593
    },
    {
      "epoch": 1.4481495851286135,
      "grad_norm": 0.5349907359447504,
      "learning_rate": 3.7342245684434964e-05,
      "loss": 11.8563,
      "step": 26594
    },
    {
      "epoch": 1.4482040391251965,
      "grad_norm": 0.6776675951913991,
      "learning_rate": 3.733537333845375e-05,
      "loss": 11.9932,
      "step": 26595
    },
    {
      "epoch": 1.4482584931217795,
      "grad_norm": 0.506046570760118,
      "learning_rate": 3.7328501479763144e-05,
      "loss": 11.8887,
      "step": 26596
    },
    {
      "epoch": 1.4483129471183624,
      "grad_norm": 0.5844919760727451,
      "learning_rate": 3.732163010841665e-05,
      "loss": 11.7822,
      "step": 26597
    },
    {
      "epoch": 1.4483674011149457,
      "grad_norm": 0.5726426211448638,
      "learning_rate": 3.7314759224467646e-05,
      "loss": 11.8709,
      "step": 26598
    },
    {
      "epoch": 1.4484218551115287,
      "grad_norm": 0.6322767415967914,
      "learning_rate": 3.730788882796957e-05,
      "loss": 11.7881,
      "step": 26599
    },
    {
      "epoch": 1.4484763091081116,
      "grad_norm": 0.5457326375300916,
      "learning_rate": 3.730101891897592e-05,
      "loss": 11.8803,
      "step": 26600
    },
    {
      "epoch": 1.4485307631046946,
      "grad_norm": 0.5716956568315725,
      "learning_rate": 3.729414949754001e-05,
      "loss": 11.7994,
      "step": 26601
    },
    {
      "epoch": 1.4485852171012776,
      "grad_norm": 0.5718924603303069,
      "learning_rate": 3.728728056371536e-05,
      "loss": 11.8763,
      "step": 26602
    },
    {
      "epoch": 1.4486396710978606,
      "grad_norm": 0.5960681913198139,
      "learning_rate": 3.728041211755529e-05,
      "loss": 11.85,
      "step": 26603
    },
    {
      "epoch": 1.4486941250944436,
      "grad_norm": 0.5632622976026171,
      "learning_rate": 3.7273544159113284e-05,
      "loss": 11.9303,
      "step": 26604
    },
    {
      "epoch": 1.4487485790910266,
      "grad_norm": 0.6080133805513952,
      "learning_rate": 3.726667668844271e-05,
      "loss": 11.8867,
      "step": 26605
    },
    {
      "epoch": 1.4488030330876096,
      "grad_norm": 0.5636445259452803,
      "learning_rate": 3.725980970559696e-05,
      "loss": 11.9338,
      "step": 26606
    },
    {
      "epoch": 1.4488574870841928,
      "grad_norm": 0.5272925428562419,
      "learning_rate": 3.7252943210629475e-05,
      "loss": 11.8475,
      "step": 26607
    },
    {
      "epoch": 1.4489119410807758,
      "grad_norm": 0.5407317561188552,
      "learning_rate": 3.72460772035936e-05,
      "loss": 11.8126,
      "step": 26608
    },
    {
      "epoch": 1.4489663950773588,
      "grad_norm": 0.5934215016958524,
      "learning_rate": 3.723921168454275e-05,
      "loss": 11.9487,
      "step": 26609
    },
    {
      "epoch": 1.4490208490739418,
      "grad_norm": 0.6207775714699713,
      "learning_rate": 3.723234665353035e-05,
      "loss": 11.9526,
      "step": 26610
    },
    {
      "epoch": 1.4490753030705248,
      "grad_norm": 0.5380609919260102,
      "learning_rate": 3.7225482110609725e-05,
      "loss": 11.932,
      "step": 26611
    },
    {
      "epoch": 1.4491297570671078,
      "grad_norm": 0.5668187977897106,
      "learning_rate": 3.72186180558343e-05,
      "loss": 11.7625,
      "step": 26612
    },
    {
      "epoch": 1.4491842110636908,
      "grad_norm": 0.6256438118898571,
      "learning_rate": 3.721175448925739e-05,
      "loss": 11.9884,
      "step": 26613
    },
    {
      "epoch": 1.4492386650602738,
      "grad_norm": 0.5845515425396075,
      "learning_rate": 3.720489141093245e-05,
      "loss": 11.7293,
      "step": 26614
    },
    {
      "epoch": 1.4492931190568568,
      "grad_norm": 0.5545123865862376,
      "learning_rate": 3.71980288209128e-05,
      "loss": 11.6521,
      "step": 26615
    },
    {
      "epoch": 1.4493475730534398,
      "grad_norm": 0.5884598664294155,
      "learning_rate": 3.719116671925178e-05,
      "loss": 11.8231,
      "step": 26616
    },
    {
      "epoch": 1.4494020270500227,
      "grad_norm": 0.5254250350294184,
      "learning_rate": 3.7184305106002815e-05,
      "loss": 11.8132,
      "step": 26617
    },
    {
      "epoch": 1.4494564810466057,
      "grad_norm": 0.5477490901207946,
      "learning_rate": 3.717744398121919e-05,
      "loss": 11.8797,
      "step": 26618
    },
    {
      "epoch": 1.4495109350431887,
      "grad_norm": 0.5908517690865315,
      "learning_rate": 3.717058334495432e-05,
      "loss": 11.8457,
      "step": 26619
    },
    {
      "epoch": 1.4495653890397717,
      "grad_norm": 0.51775420041226,
      "learning_rate": 3.716372319726151e-05,
      "loss": 11.6898,
      "step": 26620
    },
    {
      "epoch": 1.4496198430363547,
      "grad_norm": 0.4816928323939602,
      "learning_rate": 3.715686353819413e-05,
      "loss": 11.718,
      "step": 26621
    },
    {
      "epoch": 1.449674297032938,
      "grad_norm": 0.5499302883225158,
      "learning_rate": 3.7150004367805544e-05,
      "loss": 11.7975,
      "step": 26622
    },
    {
      "epoch": 1.449728751029521,
      "grad_norm": 0.5083234784909617,
      "learning_rate": 3.714314568614904e-05,
      "loss": 11.9855,
      "step": 26623
    },
    {
      "epoch": 1.449783205026104,
      "grad_norm": 0.5703637258080845,
      "learning_rate": 3.713628749327803e-05,
      "loss": 11.9224,
      "step": 26624
    },
    {
      "epoch": 1.449837659022687,
      "grad_norm": 0.521465439116705,
      "learning_rate": 3.712942978924573e-05,
      "loss": 11.911,
      "step": 26625
    },
    {
      "epoch": 1.44989211301927,
      "grad_norm": 0.5457540952526343,
      "learning_rate": 3.712257257410553e-05,
      "loss": 11.8902,
      "step": 26626
    },
    {
      "epoch": 1.449946567015853,
      "grad_norm": 0.6187793111457757,
      "learning_rate": 3.711571584791078e-05,
      "loss": 12.0318,
      "step": 26627
    },
    {
      "epoch": 1.4500010210124359,
      "grad_norm": 0.5475059480060864,
      "learning_rate": 3.710885961071473e-05,
      "loss": 11.7676,
      "step": 26628
    },
    {
      "epoch": 1.4500554750090189,
      "grad_norm": 0.5617799550263257,
      "learning_rate": 3.710200386257078e-05,
      "loss": 11.8396,
      "step": 26629
    },
    {
      "epoch": 1.4501099290056019,
      "grad_norm": 0.5459667987203591,
      "learning_rate": 3.7095148603532145e-05,
      "loss": 11.9124,
      "step": 26630
    },
    {
      "epoch": 1.450164383002185,
      "grad_norm": 0.4810094622553554,
      "learning_rate": 3.7088293833652235e-05,
      "loss": 11.9806,
      "step": 26631
    },
    {
      "epoch": 1.450218836998768,
      "grad_norm": 0.5419299697593373,
      "learning_rate": 3.708143955298427e-05,
      "loss": 11.9082,
      "step": 26632
    },
    {
      "epoch": 1.450273290995351,
      "grad_norm": 0.5519326139380485,
      "learning_rate": 3.707458576158157e-05,
      "loss": 11.843,
      "step": 26633
    },
    {
      "epoch": 1.450327744991934,
      "grad_norm": 0.5701770708516417,
      "learning_rate": 3.706773245949752e-05,
      "loss": 11.872,
      "step": 26634
    },
    {
      "epoch": 1.450382198988517,
      "grad_norm": 0.5318936987815376,
      "learning_rate": 3.7060879646785263e-05,
      "loss": 11.8942,
      "step": 26635
    },
    {
      "epoch": 1.4504366529851,
      "grad_norm": 0.5432286107373698,
      "learning_rate": 3.7054027323498154e-05,
      "loss": 11.9042,
      "step": 26636
    },
    {
      "epoch": 1.450491106981683,
      "grad_norm": 0.5471231654232496,
      "learning_rate": 3.704717548968953e-05,
      "loss": 11.8201,
      "step": 26637
    },
    {
      "epoch": 1.450545560978266,
      "grad_norm": 0.5292156627920321,
      "learning_rate": 3.704032414541258e-05,
      "loss": 11.9441,
      "step": 26638
    },
    {
      "epoch": 1.450600014974849,
      "grad_norm": 0.5307262638375472,
      "learning_rate": 3.703347329072068e-05,
      "loss": 11.7009,
      "step": 26639
    },
    {
      "epoch": 1.450654468971432,
      "grad_norm": 0.5031604460807896,
      "learning_rate": 3.702662292566701e-05,
      "loss": 11.8017,
      "step": 26640
    },
    {
      "epoch": 1.450708922968015,
      "grad_norm": 0.5135695826525555,
      "learning_rate": 3.701977305030492e-05,
      "loss": 11.9195,
      "step": 26641
    },
    {
      "epoch": 1.450763376964598,
      "grad_norm": 0.523123755323666,
      "learning_rate": 3.701292366468759e-05,
      "loss": 11.8516,
      "step": 26642
    },
    {
      "epoch": 1.450817830961181,
      "grad_norm": 0.5273212093046554,
      "learning_rate": 3.700607476886834e-05,
      "loss": 11.7828,
      "step": 26643
    },
    {
      "epoch": 1.450872284957764,
      "grad_norm": 0.5765967455236608,
      "learning_rate": 3.699922636290047e-05,
      "loss": 11.7969,
      "step": 26644
    },
    {
      "epoch": 1.4509267389543472,
      "grad_norm": 0.5088291676098585,
      "learning_rate": 3.699237844683713e-05,
      "loss": 11.7438,
      "step": 26645
    },
    {
      "epoch": 1.4509811929509302,
      "grad_norm": 0.619557520415979,
      "learning_rate": 3.6985531020731645e-05,
      "loss": 11.8212,
      "step": 26646
    },
    {
      "epoch": 1.4510356469475132,
      "grad_norm": 0.563658842852241,
      "learning_rate": 3.697868408463721e-05,
      "loss": 11.8958,
      "step": 26647
    },
    {
      "epoch": 1.4510901009440962,
      "grad_norm": 0.5518264658910694,
      "learning_rate": 3.6971837638607086e-05,
      "loss": 11.829,
      "step": 26648
    },
    {
      "epoch": 1.4511445549406792,
      "grad_norm": 0.5108141829811468,
      "learning_rate": 3.696499168269456e-05,
      "loss": 11.715,
      "step": 26649
    },
    {
      "epoch": 1.4511990089372622,
      "grad_norm": 0.5799726850046848,
      "learning_rate": 3.6958146216952806e-05,
      "loss": 11.8591,
      "step": 26650
    },
    {
      "epoch": 1.4512534629338452,
      "grad_norm": 0.6435227694286262,
      "learning_rate": 3.6951301241435096e-05,
      "loss": 11.8321,
      "step": 26651
    },
    {
      "epoch": 1.4513079169304282,
      "grad_norm": 0.547431749579825,
      "learning_rate": 3.6944456756194625e-05,
      "loss": 11.8445,
      "step": 26652
    },
    {
      "epoch": 1.4513623709270111,
      "grad_norm": 0.5623785425870065,
      "learning_rate": 3.6937612761284654e-05,
      "loss": 11.9257,
      "step": 26653
    },
    {
      "epoch": 1.4514168249235944,
      "grad_norm": 0.6250343284096292,
      "learning_rate": 3.693076925675839e-05,
      "loss": 12.0168,
      "step": 26654
    },
    {
      "epoch": 1.4514712789201774,
      "grad_norm": 0.5355238706229178,
      "learning_rate": 3.6923926242669e-05,
      "loss": 11.8364,
      "step": 26655
    },
    {
      "epoch": 1.4515257329167603,
      "grad_norm": 0.5228373184359539,
      "learning_rate": 3.6917083719069775e-05,
      "loss": 11.7161,
      "step": 26656
    },
    {
      "epoch": 1.4515801869133433,
      "grad_norm": 0.5314041809686707,
      "learning_rate": 3.691024168601386e-05,
      "loss": 11.9017,
      "step": 26657
    },
    {
      "epoch": 1.4516346409099263,
      "grad_norm": 0.504874262963168,
      "learning_rate": 3.6903400143554476e-05,
      "loss": 11.8323,
      "step": 26658
    },
    {
      "epoch": 1.4516890949065093,
      "grad_norm": 0.5325775240786298,
      "learning_rate": 3.6896559091744866e-05,
      "loss": 11.9752,
      "step": 26659
    },
    {
      "epoch": 1.4517435489030923,
      "grad_norm": 0.5333450132962121,
      "learning_rate": 3.6889718530638165e-05,
      "loss": 11.8763,
      "step": 26660
    },
    {
      "epoch": 1.4517980028996753,
      "grad_norm": 0.5229551905724732,
      "learning_rate": 3.688287846028763e-05,
      "loss": 11.8127,
      "step": 26661
    },
    {
      "epoch": 1.4518524568962583,
      "grad_norm": 0.5546493919859037,
      "learning_rate": 3.687603888074638e-05,
      "loss": 11.8888,
      "step": 26662
    },
    {
      "epoch": 1.4519069108928413,
      "grad_norm": 0.6208436977953069,
      "learning_rate": 3.686919979206768e-05,
      "loss": 11.8377,
      "step": 26663
    },
    {
      "epoch": 1.4519613648894243,
      "grad_norm": 0.6029887441206228,
      "learning_rate": 3.6862361194304663e-05,
      "loss": 11.8657,
      "step": 26664
    },
    {
      "epoch": 1.4520158188860073,
      "grad_norm": 0.5307189853439686,
      "learning_rate": 3.685552308751048e-05,
      "loss": 11.8823,
      "step": 26665
    },
    {
      "epoch": 1.4520702728825903,
      "grad_norm": 0.567352885839594,
      "learning_rate": 3.6848685471738375e-05,
      "loss": 11.8497,
      "step": 26666
    },
    {
      "epoch": 1.4521247268791733,
      "grad_norm": 0.5181823060831318,
      "learning_rate": 3.684184834704144e-05,
      "loss": 11.9856,
      "step": 26667
    },
    {
      "epoch": 1.4521791808757565,
      "grad_norm": 0.49518589676387914,
      "learning_rate": 3.683501171347292e-05,
      "loss": 11.9018,
      "step": 26668
    },
    {
      "epoch": 1.4522336348723395,
      "grad_norm": 0.5724251951825899,
      "learning_rate": 3.682817557108592e-05,
      "loss": 11.9484,
      "step": 26669
    },
    {
      "epoch": 1.4522880888689225,
      "grad_norm": 0.4889244917085866,
      "learning_rate": 3.68213399199336e-05,
      "loss": 11.8869,
      "step": 26670
    },
    {
      "epoch": 1.4523425428655055,
      "grad_norm": 0.5889861321989771,
      "learning_rate": 3.681450476006919e-05,
      "loss": 11.8546,
      "step": 26671
    },
    {
      "epoch": 1.4523969968620885,
      "grad_norm": 0.5525937298946881,
      "learning_rate": 3.680767009154574e-05,
      "loss": 11.8557,
      "step": 26672
    },
    {
      "epoch": 1.4524514508586714,
      "grad_norm": 0.5281489612725787,
      "learning_rate": 3.680083591441649e-05,
      "loss": 11.8243,
      "step": 26673
    },
    {
      "epoch": 1.4525059048552544,
      "grad_norm": 0.5640452792677921,
      "learning_rate": 3.679400222873454e-05,
      "loss": 12.0197,
      "step": 26674
    },
    {
      "epoch": 1.4525603588518374,
      "grad_norm": 0.5531250123575686,
      "learning_rate": 3.678716903455298e-05,
      "loss": 11.909,
      "step": 26675
    },
    {
      "epoch": 1.4526148128484204,
      "grad_norm": 0.516898248232284,
      "learning_rate": 3.6780336331925035e-05,
      "loss": 11.9252,
      "step": 26676
    },
    {
      "epoch": 1.4526692668450036,
      "grad_norm": 0.5915812540439752,
      "learning_rate": 3.677350412090377e-05,
      "loss": 11.8243,
      "step": 26677
    },
    {
      "epoch": 1.4527237208415866,
      "grad_norm": 0.5657633122548344,
      "learning_rate": 3.676667240154236e-05,
      "loss": 11.8663,
      "step": 26678
    },
    {
      "epoch": 1.4527781748381696,
      "grad_norm": 0.5015793374178141,
      "learning_rate": 3.6759841173893884e-05,
      "loss": 11.7767,
      "step": 26679
    },
    {
      "epoch": 1.4528326288347526,
      "grad_norm": 0.5368184892830379,
      "learning_rate": 3.6753010438011524e-05,
      "loss": 11.9265,
      "step": 26680
    },
    {
      "epoch": 1.4528870828313356,
      "grad_norm": 0.578257864527884,
      "learning_rate": 3.674618019394832e-05,
      "loss": 12.0579,
      "step": 26681
    },
    {
      "epoch": 1.4529415368279186,
      "grad_norm": 0.550452838930861,
      "learning_rate": 3.6739350441757436e-05,
      "loss": 11.8322,
      "step": 26682
    },
    {
      "epoch": 1.4529959908245016,
      "grad_norm": 0.5818041465059394,
      "learning_rate": 3.6732521181492e-05,
      "loss": 11.957,
      "step": 26683
    },
    {
      "epoch": 1.4530504448210846,
      "grad_norm": 0.49841726471481923,
      "learning_rate": 3.672569241320509e-05,
      "loss": 11.9075,
      "step": 26684
    },
    {
      "epoch": 1.4531048988176676,
      "grad_norm": 0.5062898364607856,
      "learning_rate": 3.671886413694977e-05,
      "loss": 11.7822,
      "step": 26685
    },
    {
      "epoch": 1.4531593528142506,
      "grad_norm": 0.5812687075027585,
      "learning_rate": 3.671203635277921e-05,
      "loss": 11.9392,
      "step": 26686
    },
    {
      "epoch": 1.4532138068108336,
      "grad_norm": 0.5438822429058972,
      "learning_rate": 3.670520906074644e-05,
      "loss": 12.0322,
      "step": 26687
    },
    {
      "epoch": 1.4532682608074166,
      "grad_norm": 0.6587171183868992,
      "learning_rate": 3.6698382260904605e-05,
      "loss": 11.8866,
      "step": 26688
    },
    {
      "epoch": 1.4533227148039995,
      "grad_norm": 0.6229946243171511,
      "learning_rate": 3.669155595330673e-05,
      "loss": 11.8018,
      "step": 26689
    },
    {
      "epoch": 1.4533771688005825,
      "grad_norm": 0.5237312266741596,
      "learning_rate": 3.668473013800599e-05,
      "loss": 11.7315,
      "step": 26690
    },
    {
      "epoch": 1.4534316227971655,
      "grad_norm": 0.5458746431351207,
      "learning_rate": 3.667790481505534e-05,
      "loss": 11.8983,
      "step": 26691
    },
    {
      "epoch": 1.4534860767937487,
      "grad_norm": 0.5293096620609764,
      "learning_rate": 3.667107998450794e-05,
      "loss": 11.8849,
      "step": 26692
    },
    {
      "epoch": 1.4535405307903317,
      "grad_norm": 0.5801932421878874,
      "learning_rate": 3.666425564641687e-05,
      "loss": 11.8451,
      "step": 26693
    },
    {
      "epoch": 1.4535949847869147,
      "grad_norm": 0.5618362862651421,
      "learning_rate": 3.665743180083517e-05,
      "loss": 11.9924,
      "step": 26694
    },
    {
      "epoch": 1.4536494387834977,
      "grad_norm": 0.5225101256088879,
      "learning_rate": 3.6650608447815904e-05,
      "loss": 11.7398,
      "step": 26695
    },
    {
      "epoch": 1.4537038927800807,
      "grad_norm": 0.560091871139057,
      "learning_rate": 3.6643785587412106e-05,
      "loss": 11.764,
      "step": 26696
    },
    {
      "epoch": 1.4537583467766637,
      "grad_norm": 0.49389084916806575,
      "learning_rate": 3.6636963219676843e-05,
      "loss": 11.7378,
      "step": 26697
    },
    {
      "epoch": 1.4538128007732467,
      "grad_norm": 0.5167474388162381,
      "learning_rate": 3.6630141344663214e-05,
      "loss": 11.8895,
      "step": 26698
    },
    {
      "epoch": 1.4538672547698297,
      "grad_norm": 0.5112341266163132,
      "learning_rate": 3.662331996242421e-05,
      "loss": 11.7585,
      "step": 26699
    },
    {
      "epoch": 1.453921708766413,
      "grad_norm": 0.6146450847183496,
      "learning_rate": 3.661649907301292e-05,
      "loss": 11.8389,
      "step": 26700
    },
    {
      "epoch": 1.453976162762996,
      "grad_norm": 0.5976576909184441,
      "learning_rate": 3.660967867648234e-05,
      "loss": 11.934,
      "step": 26701
    },
    {
      "epoch": 1.454030616759579,
      "grad_norm": 0.5418991571391522,
      "learning_rate": 3.660285877288555e-05,
      "loss": 12.0553,
      "step": 26702
    },
    {
      "epoch": 1.454085070756162,
      "grad_norm": 0.515063518740225,
      "learning_rate": 3.659603936227554e-05,
      "loss": 11.8825,
      "step": 26703
    },
    {
      "epoch": 1.4541395247527449,
      "grad_norm": 0.6161357065413106,
      "learning_rate": 3.658922044470538e-05,
      "loss": 11.9571,
      "step": 26704
    },
    {
      "epoch": 1.4541939787493279,
      "grad_norm": 0.5561470406601793,
      "learning_rate": 3.658240202022809e-05,
      "loss": 11.8942,
      "step": 26705
    },
    {
      "epoch": 1.4542484327459109,
      "grad_norm": 0.5795350217089544,
      "learning_rate": 3.657558408889664e-05,
      "loss": 11.9671,
      "step": 26706
    },
    {
      "epoch": 1.4543028867424939,
      "grad_norm": 0.6153799742189493,
      "learning_rate": 3.6568766650764085e-05,
      "loss": 11.8301,
      "step": 26707
    },
    {
      "epoch": 1.4543573407390769,
      "grad_norm": 0.5661994072207569,
      "learning_rate": 3.656194970588347e-05,
      "loss": 11.7437,
      "step": 26708
    },
    {
      "epoch": 1.4544117947356598,
      "grad_norm": 0.5755120456972027,
      "learning_rate": 3.655513325430774e-05,
      "loss": 11.8081,
      "step": 26709
    },
    {
      "epoch": 1.4544662487322428,
      "grad_norm": 0.5819433422622243,
      "learning_rate": 3.6548317296089966e-05,
      "loss": 11.7607,
      "step": 26710
    },
    {
      "epoch": 1.4545207027288258,
      "grad_norm": 0.5758021353347903,
      "learning_rate": 3.6541501831283086e-05,
      "loss": 11.8294,
      "step": 26711
    },
    {
      "epoch": 1.4545751567254088,
      "grad_norm": 0.5337121951230486,
      "learning_rate": 3.653468685994016e-05,
      "loss": 12.0003,
      "step": 26712
    },
    {
      "epoch": 1.4546296107219918,
      "grad_norm": 0.5368404474628399,
      "learning_rate": 3.6527872382114116e-05,
      "loss": 11.5876,
      "step": 26713
    },
    {
      "epoch": 1.4546840647185748,
      "grad_norm": 0.5330001295476007,
      "learning_rate": 3.652105839785802e-05,
      "loss": 11.7357,
      "step": 26714
    },
    {
      "epoch": 1.454738518715158,
      "grad_norm": 0.5481479255822557,
      "learning_rate": 3.651424490722481e-05,
      "loss": 11.8336,
      "step": 26715
    },
    {
      "epoch": 1.454792972711741,
      "grad_norm": 0.5863006525049337,
      "learning_rate": 3.6507431910267454e-05,
      "loss": 12.0167,
      "step": 26716
    },
    {
      "epoch": 1.454847426708324,
      "grad_norm": 0.5850871674472374,
      "learning_rate": 3.650061940703899e-05,
      "loss": 11.8807,
      "step": 26717
    },
    {
      "epoch": 1.454901880704907,
      "grad_norm": 0.5020447523837402,
      "learning_rate": 3.649380739759232e-05,
      "loss": 11.9299,
      "step": 26718
    },
    {
      "epoch": 1.45495633470149,
      "grad_norm": 0.5603533755896191,
      "learning_rate": 3.6486995881980454e-05,
      "loss": 11.7836,
      "step": 26719
    },
    {
      "epoch": 1.455010788698073,
      "grad_norm": 0.49994528514717607,
      "learning_rate": 3.648018486025639e-05,
      "loss": 11.8348,
      "step": 26720
    },
    {
      "epoch": 1.455065242694656,
      "grad_norm": 0.5114811811329905,
      "learning_rate": 3.647337433247304e-05,
      "loss": 11.7676,
      "step": 26721
    },
    {
      "epoch": 1.455119696691239,
      "grad_norm": 0.4767098339040379,
      "learning_rate": 3.64665642986834e-05,
      "loss": 11.8318,
      "step": 26722
    },
    {
      "epoch": 1.455174150687822,
      "grad_norm": 0.5873454360190589,
      "learning_rate": 3.645975475894039e-05,
      "loss": 11.8242,
      "step": 26723
    },
    {
      "epoch": 1.4552286046844052,
      "grad_norm": 0.7365266586157547,
      "learning_rate": 3.645294571329702e-05,
      "loss": 11.8758,
      "step": 26724
    },
    {
      "epoch": 1.4552830586809882,
      "grad_norm": 0.6174881721673229,
      "learning_rate": 3.6446137161806194e-05,
      "loss": 12.0069,
      "step": 26725
    },
    {
      "epoch": 1.4553375126775712,
      "grad_norm": 0.5699336506367176,
      "learning_rate": 3.6439329104520824e-05,
      "loss": 11.9938,
      "step": 26726
    },
    {
      "epoch": 1.4553919666741542,
      "grad_norm": 0.5293799174951075,
      "learning_rate": 3.6432521541493924e-05,
      "loss": 11.9106,
      "step": 26727
    },
    {
      "epoch": 1.4554464206707372,
      "grad_norm": 0.5537992367636208,
      "learning_rate": 3.642571447277837e-05,
      "loss": 11.7465,
      "step": 26728
    },
    {
      "epoch": 1.4555008746673201,
      "grad_norm": 0.4849929869678965,
      "learning_rate": 3.6418907898427156e-05,
      "loss": 11.8867,
      "step": 26729
    },
    {
      "epoch": 1.4555553286639031,
      "grad_norm": 0.585787101802056,
      "learning_rate": 3.641210181849314e-05,
      "loss": 11.8157,
      "step": 26730
    },
    {
      "epoch": 1.4556097826604861,
      "grad_norm": 0.6370423223181031,
      "learning_rate": 3.6405296233029285e-05,
      "loss": 11.9242,
      "step": 26731
    },
    {
      "epoch": 1.4556642366570691,
      "grad_norm": 0.5208407888828812,
      "learning_rate": 3.639849114208854e-05,
      "loss": 11.7121,
      "step": 26732
    },
    {
      "epoch": 1.4557186906536521,
      "grad_norm": 0.5123049212917185,
      "learning_rate": 3.6391686545723766e-05,
      "loss": 11.7233,
      "step": 26733
    },
    {
      "epoch": 1.455773144650235,
      "grad_norm": 0.5498272494814008,
      "learning_rate": 3.638488244398795e-05,
      "loss": 11.8731,
      "step": 26734
    },
    {
      "epoch": 1.455827598646818,
      "grad_norm": 0.5633624620278386,
      "learning_rate": 3.637807883693394e-05,
      "loss": 11.8209,
      "step": 26735
    },
    {
      "epoch": 1.455882052643401,
      "grad_norm": 0.5101358751773256,
      "learning_rate": 3.6371275724614616e-05,
      "loss": 11.6467,
      "step": 26736
    },
    {
      "epoch": 1.455936506639984,
      "grad_norm": 0.6661543826322869,
      "learning_rate": 3.6364473107082976e-05,
      "loss": 11.8434,
      "step": 26737
    },
    {
      "epoch": 1.4559909606365673,
      "grad_norm": 0.5377742590857267,
      "learning_rate": 3.6357670984391825e-05,
      "loss": 11.9791,
      "step": 26738
    },
    {
      "epoch": 1.4560454146331503,
      "grad_norm": 0.6069873937588747,
      "learning_rate": 3.635086935659412e-05,
      "loss": 11.985,
      "step": 26739
    },
    {
      "epoch": 1.4560998686297333,
      "grad_norm": 0.5805923894623883,
      "learning_rate": 3.634406822374271e-05,
      "loss": 11.7486,
      "step": 26740
    },
    {
      "epoch": 1.4561543226263163,
      "grad_norm": 0.5034230426158893,
      "learning_rate": 3.6337267585890486e-05,
      "loss": 11.9536,
      "step": 26741
    },
    {
      "epoch": 1.4562087766228993,
      "grad_norm": 0.5110394749315411,
      "learning_rate": 3.63304674430904e-05,
      "loss": 11.7933,
      "step": 26742
    },
    {
      "epoch": 1.4562632306194823,
      "grad_norm": 0.5787534043271866,
      "learning_rate": 3.632366779539522e-05,
      "loss": 12.0233,
      "step": 26743
    },
    {
      "epoch": 1.4563176846160653,
      "grad_norm": 0.542646975531092,
      "learning_rate": 3.631686864285793e-05,
      "loss": 11.8384,
      "step": 26744
    },
    {
      "epoch": 1.4563721386126482,
      "grad_norm": 0.5042075821464146,
      "learning_rate": 3.631006998553134e-05,
      "loss": 11.8898,
      "step": 26745
    },
    {
      "epoch": 1.4564265926092312,
      "grad_norm": 0.5677960509359683,
      "learning_rate": 3.63032718234683e-05,
      "loss": 11.8293,
      "step": 26746
    },
    {
      "epoch": 1.4564810466058145,
      "grad_norm": 0.49353872529268183,
      "learning_rate": 3.6296474156721725e-05,
      "loss": 11.8039,
      "step": 26747
    },
    {
      "epoch": 1.4565355006023974,
      "grad_norm": 0.5477240079817448,
      "learning_rate": 3.628967698534441e-05,
      "loss": 11.7792,
      "step": 26748
    },
    {
      "epoch": 1.4565899545989804,
      "grad_norm": 0.5096398289991971,
      "learning_rate": 3.62828803093893e-05,
      "loss": 11.8037,
      "step": 26749
    },
    {
      "epoch": 1.4566444085955634,
      "grad_norm": 0.5487537163745176,
      "learning_rate": 3.627608412890916e-05,
      "loss": 12.0059,
      "step": 26750
    },
    {
      "epoch": 1.4566988625921464,
      "grad_norm": 0.5507765976045028,
      "learning_rate": 3.6269288443956906e-05,
      "loss": 11.9403,
      "step": 26751
    },
    {
      "epoch": 1.4567533165887294,
      "grad_norm": 0.5494549162198724,
      "learning_rate": 3.626249325458533e-05,
      "loss": 11.8644,
      "step": 26752
    },
    {
      "epoch": 1.4568077705853124,
      "grad_norm": 0.4974389737999597,
      "learning_rate": 3.625569856084728e-05,
      "loss": 11.7748,
      "step": 26753
    },
    {
      "epoch": 1.4568622245818954,
      "grad_norm": 0.5486646637385234,
      "learning_rate": 3.624890436279565e-05,
      "loss": 11.8773,
      "step": 26754
    },
    {
      "epoch": 1.4569166785784784,
      "grad_norm": 0.5552403145496668,
      "learning_rate": 3.624211066048323e-05,
      "loss": 11.8685,
      "step": 26755
    },
    {
      "epoch": 1.4569711325750614,
      "grad_norm": 0.5307731225341632,
      "learning_rate": 3.623531745396282e-05,
      "loss": 11.8372,
      "step": 26756
    },
    {
      "epoch": 1.4570255865716444,
      "grad_norm": 0.5385312329365134,
      "learning_rate": 3.6228524743287294e-05,
      "loss": 11.742,
      "step": 26757
    },
    {
      "epoch": 1.4570800405682274,
      "grad_norm": 0.5731116936757621,
      "learning_rate": 3.622173252850943e-05,
      "loss": 11.869,
      "step": 26758
    },
    {
      "epoch": 1.4571344945648104,
      "grad_norm": 0.5654345221512659,
      "learning_rate": 3.621494080968211e-05,
      "loss": 11.9001,
      "step": 26759
    },
    {
      "epoch": 1.4571889485613934,
      "grad_norm": 0.5589460771174387,
      "learning_rate": 3.620814958685807e-05,
      "loss": 11.9373,
      "step": 26760
    },
    {
      "epoch": 1.4572434025579764,
      "grad_norm": 0.5212287814667456,
      "learning_rate": 3.6201358860090184e-05,
      "loss": 11.8175,
      "step": 26761
    },
    {
      "epoch": 1.4572978565545596,
      "grad_norm": 0.5560803817828982,
      "learning_rate": 3.61945686294312e-05,
      "loss": 12.0346,
      "step": 26762
    },
    {
      "epoch": 1.4573523105511426,
      "grad_norm": 0.4815342156419329,
      "learning_rate": 3.618777889493394e-05,
      "loss": 11.8207,
      "step": 26763
    },
    {
      "epoch": 1.4574067645477256,
      "grad_norm": 0.5340409393213523,
      "learning_rate": 3.618098965665126e-05,
      "loss": 11.8581,
      "step": 26764
    },
    {
      "epoch": 1.4574612185443085,
      "grad_norm": 0.5808746622743467,
      "learning_rate": 3.6174200914635904e-05,
      "loss": 11.7477,
      "step": 26765
    },
    {
      "epoch": 1.4575156725408915,
      "grad_norm": 0.5066700160259542,
      "learning_rate": 3.616741266894067e-05,
      "loss": 11.899,
      "step": 26766
    },
    {
      "epoch": 1.4575701265374745,
      "grad_norm": 0.5024104816794057,
      "learning_rate": 3.6160624919618304e-05,
      "loss": 11.8207,
      "step": 26767
    },
    {
      "epoch": 1.4576245805340575,
      "grad_norm": 0.6249859002216991,
      "learning_rate": 3.6153837666721616e-05,
      "loss": 11.919,
      "step": 26768
    },
    {
      "epoch": 1.4576790345306405,
      "grad_norm": 0.5393837181351506,
      "learning_rate": 3.614705091030344e-05,
      "loss": 11.675,
      "step": 26769
    },
    {
      "epoch": 1.4577334885272237,
      "grad_norm": 0.5552178948971118,
      "learning_rate": 3.614026465041645e-05,
      "loss": 11.8879,
      "step": 26770
    },
    {
      "epoch": 1.4577879425238067,
      "grad_norm": 0.5082658302359733,
      "learning_rate": 3.6133478887113525e-05,
      "loss": 11.8018,
      "step": 26771
    },
    {
      "epoch": 1.4578423965203897,
      "grad_norm": 0.5895837781272815,
      "learning_rate": 3.612669362044734e-05,
      "loss": 11.9849,
      "step": 26772
    },
    {
      "epoch": 1.4578968505169727,
      "grad_norm": 0.5581416663970848,
      "learning_rate": 3.611990885047073e-05,
      "loss": 11.9058,
      "step": 26773
    },
    {
      "epoch": 1.4579513045135557,
      "grad_norm": 0.5629753655164212,
      "learning_rate": 3.6113124577236376e-05,
      "loss": 11.7372,
      "step": 26774
    },
    {
      "epoch": 1.4580057585101387,
      "grad_norm": 0.6344005842001699,
      "learning_rate": 3.610634080079713e-05,
      "loss": 12.0204,
      "step": 26775
    },
    {
      "epoch": 1.4580602125067217,
      "grad_norm": 0.5377170799900914,
      "learning_rate": 3.609955752120568e-05,
      "loss": 11.9078,
      "step": 26776
    },
    {
      "epoch": 1.4581146665033047,
      "grad_norm": 0.518073274344814,
      "learning_rate": 3.609277473851477e-05,
      "loss": 11.7801,
      "step": 26777
    },
    {
      "epoch": 1.4581691204998877,
      "grad_norm": 0.5436874352859177,
      "learning_rate": 3.6085992452777184e-05,
      "loss": 11.8592,
      "step": 26778
    },
    {
      "epoch": 1.4582235744964707,
      "grad_norm": 0.5310667102839995,
      "learning_rate": 3.6079210664045607e-05,
      "loss": 11.8141,
      "step": 26779
    },
    {
      "epoch": 1.4582780284930537,
      "grad_norm": 0.586173361703436,
      "learning_rate": 3.60724293723728e-05,
      "loss": 11.759,
      "step": 26780
    },
    {
      "epoch": 1.4583324824896367,
      "grad_norm": 0.5070016668348178,
      "learning_rate": 3.606564857781154e-05,
      "loss": 11.8323,
      "step": 26781
    },
    {
      "epoch": 1.4583869364862196,
      "grad_norm": 0.5335407611867968,
      "learning_rate": 3.605886828041449e-05,
      "loss": 11.7087,
      "step": 26782
    },
    {
      "epoch": 1.4584413904828026,
      "grad_norm": 0.6227259365432857,
      "learning_rate": 3.605208848023444e-05,
      "loss": 11.8679,
      "step": 26783
    },
    {
      "epoch": 1.4584958444793856,
      "grad_norm": 0.4915145450358055,
      "learning_rate": 3.6045309177324027e-05,
      "loss": 11.7542,
      "step": 26784
    },
    {
      "epoch": 1.4585502984759688,
      "grad_norm": 0.5394648993560738,
      "learning_rate": 3.6038530371736055e-05,
      "loss": 11.8166,
      "step": 26785
    },
    {
      "epoch": 1.4586047524725518,
      "grad_norm": 0.5687555144585736,
      "learning_rate": 3.603175206352321e-05,
      "loss": 11.9645,
      "step": 26786
    },
    {
      "epoch": 1.4586592064691348,
      "grad_norm": 0.5606517627161273,
      "learning_rate": 3.602497425273813e-05,
      "loss": 11.8242,
      "step": 26787
    },
    {
      "epoch": 1.4587136604657178,
      "grad_norm": 0.5292711113321935,
      "learning_rate": 3.6018196939433626e-05,
      "loss": 11.8848,
      "step": 26788
    },
    {
      "epoch": 1.4587681144623008,
      "grad_norm": 0.5183681193963549,
      "learning_rate": 3.601142012366232e-05,
      "loss": 11.9129,
      "step": 26789
    },
    {
      "epoch": 1.4588225684588838,
      "grad_norm": 0.5621041526731029,
      "learning_rate": 3.6004643805476943e-05,
      "loss": 11.8582,
      "step": 26790
    },
    {
      "epoch": 1.4588770224554668,
      "grad_norm": 0.5960845339392585,
      "learning_rate": 3.599786798493021e-05,
      "loss": 11.9627,
      "step": 26791
    },
    {
      "epoch": 1.4589314764520498,
      "grad_norm": 0.5491270168168505,
      "learning_rate": 3.5991092662074765e-05,
      "loss": 11.7948,
      "step": 26792
    },
    {
      "epoch": 1.4589859304486328,
      "grad_norm": 0.5597381496196345,
      "learning_rate": 3.598431783696335e-05,
      "loss": 11.8733,
      "step": 26793
    },
    {
      "epoch": 1.459040384445216,
      "grad_norm": 0.5240365781155311,
      "learning_rate": 3.597754350964858e-05,
      "loss": 11.8348,
      "step": 26794
    },
    {
      "epoch": 1.459094838441799,
      "grad_norm": 0.6225060645311038,
      "learning_rate": 3.597076968018321e-05,
      "loss": 11.9037,
      "step": 26795
    },
    {
      "epoch": 1.459149292438382,
      "grad_norm": 0.5528228808901301,
      "learning_rate": 3.596399634861987e-05,
      "loss": 11.7331,
      "step": 26796
    },
    {
      "epoch": 1.459203746434965,
      "grad_norm": 0.5245412508166607,
      "learning_rate": 3.5957223515011195e-05,
      "loss": 11.8954,
      "step": 26797
    },
    {
      "epoch": 1.459258200431548,
      "grad_norm": 0.5355427766129638,
      "learning_rate": 3.595045117940991e-05,
      "loss": 11.8718,
      "step": 26798
    },
    {
      "epoch": 1.459312654428131,
      "grad_norm": 0.528865987818189,
      "learning_rate": 3.594367934186865e-05,
      "loss": 11.8158,
      "step": 26799
    },
    {
      "epoch": 1.459367108424714,
      "grad_norm": 0.5227031212702511,
      "learning_rate": 3.5936908002440105e-05,
      "loss": 11.8405,
      "step": 26800
    },
    {
      "epoch": 1.459421562421297,
      "grad_norm": 0.6284403378354587,
      "learning_rate": 3.593013716117687e-05,
      "loss": 11.9101,
      "step": 26801
    },
    {
      "epoch": 1.45947601641788,
      "grad_norm": 0.6028463324801152,
      "learning_rate": 3.592336681813163e-05,
      "loss": 11.8608,
      "step": 26802
    },
    {
      "epoch": 1.459530470414463,
      "grad_norm": 0.5399263416332223,
      "learning_rate": 3.591659697335707e-05,
      "loss": 11.7132,
      "step": 26803
    },
    {
      "epoch": 1.459584924411046,
      "grad_norm": 0.5004537076128194,
      "learning_rate": 3.590982762690578e-05,
      "loss": 11.8438,
      "step": 26804
    },
    {
      "epoch": 1.459639378407629,
      "grad_norm": 0.505222743084213,
      "learning_rate": 3.5903058778830434e-05,
      "loss": 11.9408,
      "step": 26805
    },
    {
      "epoch": 1.459693832404212,
      "grad_norm": 0.6122278623686817,
      "learning_rate": 3.5896290429183656e-05,
      "loss": 11.9357,
      "step": 26806
    },
    {
      "epoch": 1.459748286400795,
      "grad_norm": 0.5553954049374712,
      "learning_rate": 3.5889522578018044e-05,
      "loss": 11.8876,
      "step": 26807
    },
    {
      "epoch": 1.4598027403973781,
      "grad_norm": 0.5401387797359927,
      "learning_rate": 3.588275522538629e-05,
      "loss": 11.9887,
      "step": 26808
    },
    {
      "epoch": 1.459857194393961,
      "grad_norm": 0.5533280896905917,
      "learning_rate": 3.587598837134095e-05,
      "loss": 11.9955,
      "step": 26809
    },
    {
      "epoch": 1.459911648390544,
      "grad_norm": 0.6550990427751122,
      "learning_rate": 3.5869222015934715e-05,
      "loss": 11.8289,
      "step": 26810
    },
    {
      "epoch": 1.459966102387127,
      "grad_norm": 0.587633376029004,
      "learning_rate": 3.5862456159220114e-05,
      "loss": 11.8928,
      "step": 26811
    },
    {
      "epoch": 1.46002055638371,
      "grad_norm": 0.557811769610465,
      "learning_rate": 3.585569080124983e-05,
      "loss": 11.8358,
      "step": 26812
    },
    {
      "epoch": 1.460075010380293,
      "grad_norm": 0.5273554918722372,
      "learning_rate": 3.584892594207647e-05,
      "loss": 11.7675,
      "step": 26813
    },
    {
      "epoch": 1.460129464376876,
      "grad_norm": 0.5055439436397466,
      "learning_rate": 3.5842161581752596e-05,
      "loss": 11.8395,
      "step": 26814
    },
    {
      "epoch": 1.460183918373459,
      "grad_norm": 0.5862420016312873,
      "learning_rate": 3.58353977203309e-05,
      "loss": 11.9083,
      "step": 26815
    },
    {
      "epoch": 1.460238372370042,
      "grad_norm": 0.5139011935152126,
      "learning_rate": 3.5828634357863835e-05,
      "loss": 11.9216,
      "step": 26816
    },
    {
      "epoch": 1.4602928263666253,
      "grad_norm": 0.5439183726164122,
      "learning_rate": 3.582187149440408e-05,
      "loss": 11.8978,
      "step": 26817
    },
    {
      "epoch": 1.4603472803632083,
      "grad_norm": 0.5520476594550109,
      "learning_rate": 3.581510913000424e-05,
      "loss": 11.9112,
      "step": 26818
    },
    {
      "epoch": 1.4604017343597913,
      "grad_norm": 0.5260466044548181,
      "learning_rate": 3.580834726471686e-05,
      "loss": 11.7303,
      "step": 26819
    },
    {
      "epoch": 1.4604561883563743,
      "grad_norm": 0.4896045580720908,
      "learning_rate": 3.5801585898594545e-05,
      "loss": 11.8574,
      "step": 26820
    },
    {
      "epoch": 1.4605106423529572,
      "grad_norm": 0.5525963234266409,
      "learning_rate": 3.579482503168985e-05,
      "loss": 11.9571,
      "step": 26821
    },
    {
      "epoch": 1.4605650963495402,
      "grad_norm": 0.5674735658381522,
      "learning_rate": 3.5788064664055384e-05,
      "loss": 11.9662,
      "step": 26822
    },
    {
      "epoch": 1.4606195503461232,
      "grad_norm": 0.5309966560276173,
      "learning_rate": 3.578130479574367e-05,
      "loss": 11.8387,
      "step": 26823
    },
    {
      "epoch": 1.4606740043427062,
      "grad_norm": 0.6262015901906925,
      "learning_rate": 3.5774545426807296e-05,
      "loss": 11.9699,
      "step": 26824
    },
    {
      "epoch": 1.4607284583392892,
      "grad_norm": 0.5931571144835259,
      "learning_rate": 3.57677865572989e-05,
      "loss": 11.8291,
      "step": 26825
    },
    {
      "epoch": 1.4607829123358722,
      "grad_norm": 0.5761147142916926,
      "learning_rate": 3.5761028187270896e-05,
      "loss": 11.8606,
      "step": 26826
    },
    {
      "epoch": 1.4608373663324552,
      "grad_norm": 0.5143222559023625,
      "learning_rate": 3.575427031677594e-05,
      "loss": 11.8135,
      "step": 26827
    },
    {
      "epoch": 1.4608918203290382,
      "grad_norm": 0.5204746049835881,
      "learning_rate": 3.574751294586652e-05,
      "loss": 11.9793,
      "step": 26828
    },
    {
      "epoch": 1.4609462743256212,
      "grad_norm": 0.5909807716816972,
      "learning_rate": 3.574075607459522e-05,
      "loss": 11.751,
      "step": 26829
    },
    {
      "epoch": 1.4610007283222042,
      "grad_norm": 0.6470226609947656,
      "learning_rate": 3.5733999703014596e-05,
      "loss": 11.9792,
      "step": 26830
    },
    {
      "epoch": 1.4610551823187872,
      "grad_norm": 0.5570834055776428,
      "learning_rate": 3.572724383117715e-05,
      "loss": 11.8176,
      "step": 26831
    },
    {
      "epoch": 1.4611096363153704,
      "grad_norm": 0.5649935117040239,
      "learning_rate": 3.572048845913546e-05,
      "loss": 11.872,
      "step": 26832
    },
    {
      "epoch": 1.4611640903119534,
      "grad_norm": 0.5690616025433386,
      "learning_rate": 3.5713733586942e-05,
      "loss": 11.8906,
      "step": 26833
    },
    {
      "epoch": 1.4612185443085364,
      "grad_norm": 0.5559128719296456,
      "learning_rate": 3.570697921464936e-05,
      "loss": 11.8206,
      "step": 26834
    },
    {
      "epoch": 1.4612729983051194,
      "grad_norm": 0.568355227035462,
      "learning_rate": 3.5700225342310045e-05,
      "loss": 11.8729,
      "step": 26835
    },
    {
      "epoch": 1.4613274523017024,
      "grad_norm": 0.5144879399681113,
      "learning_rate": 3.5693471969976514e-05,
      "loss": 11.8475,
      "step": 26836
    },
    {
      "epoch": 1.4613819062982853,
      "grad_norm": 0.5461364187896164,
      "learning_rate": 3.568671909770136e-05,
      "loss": 11.8737,
      "step": 26837
    },
    {
      "epoch": 1.4614363602948683,
      "grad_norm": 0.5846842650329348,
      "learning_rate": 3.567996672553704e-05,
      "loss": 11.7958,
      "step": 26838
    },
    {
      "epoch": 1.4614908142914513,
      "grad_norm": 0.5830549694521869,
      "learning_rate": 3.5673214853536074e-05,
      "loss": 11.9108,
      "step": 26839
    },
    {
      "epoch": 1.4615452682880345,
      "grad_norm": 0.5879849804474622,
      "learning_rate": 3.566646348175102e-05,
      "loss": 11.8296,
      "step": 26840
    },
    {
      "epoch": 1.4615997222846175,
      "grad_norm": 0.5461265208976671,
      "learning_rate": 3.5659712610234295e-05,
      "loss": 11.9402,
      "step": 26841
    },
    {
      "epoch": 1.4616541762812005,
      "grad_norm": 0.609963121610458,
      "learning_rate": 3.565296223903848e-05,
      "loss": 11.9677,
      "step": 26842
    },
    {
      "epoch": 1.4617086302777835,
      "grad_norm": 0.5175958788029443,
      "learning_rate": 3.564621236821597e-05,
      "loss": 11.8477,
      "step": 26843
    },
    {
      "epoch": 1.4617630842743665,
      "grad_norm": 0.5896009516127384,
      "learning_rate": 3.563946299781935e-05,
      "loss": 11.8115,
      "step": 26844
    },
    {
      "epoch": 1.4618175382709495,
      "grad_norm": 0.5422785192947653,
      "learning_rate": 3.5632714127901054e-05,
      "loss": 11.9648,
      "step": 26845
    },
    {
      "epoch": 1.4618719922675325,
      "grad_norm": 0.4731157840095486,
      "learning_rate": 3.562596575851354e-05,
      "loss": 11.7739,
      "step": 26846
    },
    {
      "epoch": 1.4619264462641155,
      "grad_norm": 0.5296582183190118,
      "learning_rate": 3.5619217889709346e-05,
      "loss": 11.8756,
      "step": 26847
    },
    {
      "epoch": 1.4619809002606985,
      "grad_norm": 0.5278908416265546,
      "learning_rate": 3.561247052154086e-05,
      "loss": 11.7552,
      "step": 26848
    },
    {
      "epoch": 1.4620353542572815,
      "grad_norm": 0.5306894901091312,
      "learning_rate": 3.5605723654060654e-05,
      "loss": 11.8228,
      "step": 26849
    },
    {
      "epoch": 1.4620898082538645,
      "grad_norm": 0.5127144944073668,
      "learning_rate": 3.55989772873211e-05,
      "loss": 11.9017,
      "step": 26850
    },
    {
      "epoch": 1.4621442622504475,
      "grad_norm": 0.5350098128136271,
      "learning_rate": 3.55922314213747e-05,
      "loss": 11.879,
      "step": 26851
    },
    {
      "epoch": 1.4621987162470305,
      "grad_norm": 0.5825422245832014,
      "learning_rate": 3.5585486056273943e-05,
      "loss": 11.9091,
      "step": 26852
    },
    {
      "epoch": 1.4622531702436135,
      "grad_norm": 0.5155547722424547,
      "learning_rate": 3.557874119207121e-05,
      "loss": 11.8226,
      "step": 26853
    },
    {
      "epoch": 1.4623076242401964,
      "grad_norm": 0.5701044575689473,
      "learning_rate": 3.557199682881902e-05,
      "loss": 11.8285,
      "step": 26854
    },
    {
      "epoch": 1.4623620782367797,
      "grad_norm": 0.5585870580704279,
      "learning_rate": 3.556525296656979e-05,
      "loss": 11.8385,
      "step": 26855
    },
    {
      "epoch": 1.4624165322333627,
      "grad_norm": 0.5699723718271553,
      "learning_rate": 3.555850960537593e-05,
      "loss": 11.8206,
      "step": 26856
    },
    {
      "epoch": 1.4624709862299456,
      "grad_norm": 0.5263886877754508,
      "learning_rate": 3.555176674528994e-05,
      "loss": 11.8327,
      "step": 26857
    },
    {
      "epoch": 1.4625254402265286,
      "grad_norm": 0.5653618440115944,
      "learning_rate": 3.554502438636419e-05,
      "loss": 11.8992,
      "step": 26858
    },
    {
      "epoch": 1.4625798942231116,
      "grad_norm": 0.5660923805417012,
      "learning_rate": 3.553828252865117e-05,
      "loss": 11.8532,
      "step": 26859
    },
    {
      "epoch": 1.4626343482196946,
      "grad_norm": 0.5369221715785782,
      "learning_rate": 3.553154117220323e-05,
      "loss": 11.8881,
      "step": 26860
    },
    {
      "epoch": 1.4626888022162776,
      "grad_norm": 0.5552046678845857,
      "learning_rate": 3.552480031707285e-05,
      "loss": 11.8117,
      "step": 26861
    },
    {
      "epoch": 1.4627432562128606,
      "grad_norm": 0.5309263407310856,
      "learning_rate": 3.551805996331247e-05,
      "loss": 11.8196,
      "step": 26862
    },
    {
      "epoch": 1.4627977102094436,
      "grad_norm": 0.5603158701294209,
      "learning_rate": 3.551132011097442e-05,
      "loss": 11.928,
      "step": 26863
    },
    {
      "epoch": 1.4628521642060268,
      "grad_norm": 0.5828384033150691,
      "learning_rate": 3.55045807601112e-05,
      "loss": 11.7837,
      "step": 26864
    },
    {
      "epoch": 1.4629066182026098,
      "grad_norm": 0.5465292314854292,
      "learning_rate": 3.549784191077519e-05,
      "loss": 11.912,
      "step": 26865
    },
    {
      "epoch": 1.4629610721991928,
      "grad_norm": 0.5273812907053981,
      "learning_rate": 3.549110356301873e-05,
      "loss": 11.895,
      "step": 26866
    },
    {
      "epoch": 1.4630155261957758,
      "grad_norm": 0.5416311755046903,
      "learning_rate": 3.548436571689431e-05,
      "loss": 11.9237,
      "step": 26867
    },
    {
      "epoch": 1.4630699801923588,
      "grad_norm": 0.5237906236370793,
      "learning_rate": 3.547762837245424e-05,
      "loss": 11.7981,
      "step": 26868
    },
    {
      "epoch": 1.4631244341889418,
      "grad_norm": 0.5410911256517951,
      "learning_rate": 3.547089152975098e-05,
      "loss": 11.7256,
      "step": 26869
    },
    {
      "epoch": 1.4631788881855248,
      "grad_norm": 0.497770829428523,
      "learning_rate": 3.546415518883687e-05,
      "loss": 11.8333,
      "step": 26870
    },
    {
      "epoch": 1.4632333421821078,
      "grad_norm": 0.5575192326980732,
      "learning_rate": 3.545741934976434e-05,
      "loss": 11.8322,
      "step": 26871
    },
    {
      "epoch": 1.4632877961786908,
      "grad_norm": 0.5396793254369858,
      "learning_rate": 3.545068401258571e-05,
      "loss": 11.8685,
      "step": 26872
    },
    {
      "epoch": 1.4633422501752738,
      "grad_norm": 0.5514470270430754,
      "learning_rate": 3.544394917735337e-05,
      "loss": 11.8436,
      "step": 26873
    },
    {
      "epoch": 1.4633967041718567,
      "grad_norm": 0.5739962542643732,
      "learning_rate": 3.543721484411976e-05,
      "loss": 11.8729,
      "step": 26874
    },
    {
      "epoch": 1.4634511581684397,
      "grad_norm": 0.5346187295383611,
      "learning_rate": 3.543048101293719e-05,
      "loss": 11.9297,
      "step": 26875
    },
    {
      "epoch": 1.4635056121650227,
      "grad_norm": 0.5533855270475371,
      "learning_rate": 3.5423747683857986e-05,
      "loss": 11.8589,
      "step": 26876
    },
    {
      "epoch": 1.4635600661616057,
      "grad_norm": 0.5170854857149493,
      "learning_rate": 3.5417014856934595e-05,
      "loss": 11.9063,
      "step": 26877
    },
    {
      "epoch": 1.463614520158189,
      "grad_norm": 0.5254200164626402,
      "learning_rate": 3.541028253221929e-05,
      "loss": 11.8278,
      "step": 26878
    },
    {
      "epoch": 1.463668974154772,
      "grad_norm": 0.5644971590330566,
      "learning_rate": 3.5403550709764486e-05,
      "loss": 11.9211,
      "step": 26879
    },
    {
      "epoch": 1.463723428151355,
      "grad_norm": 0.554265191206821,
      "learning_rate": 3.539681938962248e-05,
      "loss": 11.8722,
      "step": 26880
    },
    {
      "epoch": 1.463777882147938,
      "grad_norm": 0.5473133514142409,
      "learning_rate": 3.539008857184567e-05,
      "loss": 11.8108,
      "step": 26881
    },
    {
      "epoch": 1.463832336144521,
      "grad_norm": 0.5564749672755559,
      "learning_rate": 3.538335825648633e-05,
      "loss": 11.9022,
      "step": 26882
    },
    {
      "epoch": 1.463886790141104,
      "grad_norm": 0.5924881033935094,
      "learning_rate": 3.537662844359687e-05,
      "loss": 11.88,
      "step": 26883
    },
    {
      "epoch": 1.463941244137687,
      "grad_norm": 0.5988835492623468,
      "learning_rate": 3.5369899133229554e-05,
      "loss": 11.982,
      "step": 26884
    },
    {
      "epoch": 1.4639956981342699,
      "grad_norm": 0.4902957174226832,
      "learning_rate": 3.536317032543673e-05,
      "loss": 11.7601,
      "step": 26885
    },
    {
      "epoch": 1.4640501521308529,
      "grad_norm": 0.5221769566812132,
      "learning_rate": 3.535644202027081e-05,
      "loss": 11.8759,
      "step": 26886
    },
    {
      "epoch": 1.464104606127436,
      "grad_norm": 0.5963895010595072,
      "learning_rate": 3.534971421778397e-05,
      "loss": 11.8333,
      "step": 26887
    },
    {
      "epoch": 1.464159060124019,
      "grad_norm": 0.5724937814045064,
      "learning_rate": 3.5342986918028584e-05,
      "loss": 11.8156,
      "step": 26888
    },
    {
      "epoch": 1.464213514120602,
      "grad_norm": 0.5734674796494019,
      "learning_rate": 3.533626012105702e-05,
      "loss": 11.9431,
      "step": 26889
    },
    {
      "epoch": 1.464267968117185,
      "grad_norm": 0.5768639096487074,
      "learning_rate": 3.532953382692151e-05,
      "loss": 11.8823,
      "step": 26890
    },
    {
      "epoch": 1.464322422113768,
      "grad_norm": 0.5828853823118294,
      "learning_rate": 3.5322808035674414e-05,
      "loss": 11.8061,
      "step": 26891
    },
    {
      "epoch": 1.464376876110351,
      "grad_norm": 0.6427780669610677,
      "learning_rate": 3.5316082747367984e-05,
      "loss": 11.8337,
      "step": 26892
    },
    {
      "epoch": 1.464431330106934,
      "grad_norm": 0.532718269284918,
      "learning_rate": 3.530935796205457e-05,
      "loss": 11.9549,
      "step": 26893
    },
    {
      "epoch": 1.464485784103517,
      "grad_norm": 0.5845266960964168,
      "learning_rate": 3.53026336797864e-05,
      "loss": 11.9907,
      "step": 26894
    },
    {
      "epoch": 1.4645402381001,
      "grad_norm": 0.5820803008122897,
      "learning_rate": 3.529590990061581e-05,
      "loss": 11.9355,
      "step": 26895
    },
    {
      "epoch": 1.464594692096683,
      "grad_norm": 0.553904992996588,
      "learning_rate": 3.5289186624595125e-05,
      "loss": 11.8495,
      "step": 26896
    },
    {
      "epoch": 1.464649146093266,
      "grad_norm": 0.502351866275459,
      "learning_rate": 3.528246385177653e-05,
      "loss": 11.8297,
      "step": 26897
    },
    {
      "epoch": 1.464703600089849,
      "grad_norm": 0.5226287251745798,
      "learning_rate": 3.5275741582212376e-05,
      "loss": 11.8943,
      "step": 26898
    },
    {
      "epoch": 1.464758054086432,
      "grad_norm": 0.5174506947448685,
      "learning_rate": 3.526901981595487e-05,
      "loss": 11.8376,
      "step": 26899
    },
    {
      "epoch": 1.464812508083015,
      "grad_norm": 0.5745438889851867,
      "learning_rate": 3.526229855305633e-05,
      "loss": 11.875,
      "step": 26900
    },
    {
      "epoch": 1.4648669620795982,
      "grad_norm": 0.5942905440575059,
      "learning_rate": 3.525557779356904e-05,
      "loss": 11.9326,
      "step": 26901
    },
    {
      "epoch": 1.4649214160761812,
      "grad_norm": 0.5543504431445094,
      "learning_rate": 3.52488575375452e-05,
      "loss": 11.8438,
      "step": 26902
    },
    {
      "epoch": 1.4649758700727642,
      "grad_norm": 0.6329014310092056,
      "learning_rate": 3.524213778503714e-05,
      "loss": 12.005,
      "step": 26903
    },
    {
      "epoch": 1.4650303240693472,
      "grad_norm": 0.5071093456661097,
      "learning_rate": 3.523541853609704e-05,
      "loss": 11.6483,
      "step": 26904
    },
    {
      "epoch": 1.4650847780659302,
      "grad_norm": 0.5729737739930458,
      "learning_rate": 3.522869979077723e-05,
      "loss": 11.9562,
      "step": 26905
    },
    {
      "epoch": 1.4651392320625132,
      "grad_norm": 0.4999666939485268,
      "learning_rate": 3.52219815491299e-05,
      "loss": 11.8352,
      "step": 26906
    },
    {
      "epoch": 1.4651936860590962,
      "grad_norm": 0.5454996946055474,
      "learning_rate": 3.5215263811207276e-05,
      "loss": 11.8997,
      "step": 26907
    },
    {
      "epoch": 1.4652481400556792,
      "grad_norm": 0.5751158072911432,
      "learning_rate": 3.5208546577061666e-05,
      "loss": 11.7981,
      "step": 26908
    },
    {
      "epoch": 1.4653025940522622,
      "grad_norm": 0.6659713664357423,
      "learning_rate": 3.520182984674522e-05,
      "loss": 11.8795,
      "step": 26909
    },
    {
      "epoch": 1.4653570480488454,
      "grad_norm": 0.5503720926795402,
      "learning_rate": 3.519511362031021e-05,
      "loss": 11.8338,
      "step": 26910
    },
    {
      "epoch": 1.4654115020454284,
      "grad_norm": 0.5594942983638554,
      "learning_rate": 3.518839789780891e-05,
      "loss": 11.9687,
      "step": 26911
    },
    {
      "epoch": 1.4654659560420114,
      "grad_norm": 0.5921855345390399,
      "learning_rate": 3.518168267929345e-05,
      "loss": 11.8966,
      "step": 26912
    },
    {
      "epoch": 1.4655204100385943,
      "grad_norm": 0.5396709006485998,
      "learning_rate": 3.517496796481614e-05,
      "loss": 11.9207,
      "step": 26913
    },
    {
      "epoch": 1.4655748640351773,
      "grad_norm": 0.5722722756055318,
      "learning_rate": 3.516825375442912e-05,
      "loss": 11.943,
      "step": 26914
    },
    {
      "epoch": 1.4656293180317603,
      "grad_norm": 0.541991721209282,
      "learning_rate": 3.516154004818465e-05,
      "loss": 11.8641,
      "step": 26915
    },
    {
      "epoch": 1.4656837720283433,
      "grad_norm": 0.5808604813469752,
      "learning_rate": 3.5154826846134925e-05,
      "loss": 11.777,
      "step": 26916
    },
    {
      "epoch": 1.4657382260249263,
      "grad_norm": 0.5337465519442687,
      "learning_rate": 3.51481141483321e-05,
      "loss": 11.8277,
      "step": 26917
    },
    {
      "epoch": 1.4657926800215093,
      "grad_norm": 0.5614730591651275,
      "learning_rate": 3.514140195482846e-05,
      "loss": 11.8785,
      "step": 26918
    },
    {
      "epoch": 1.4658471340180923,
      "grad_norm": 0.5212476666733246,
      "learning_rate": 3.513469026567612e-05,
      "loss": 11.8603,
      "step": 26919
    },
    {
      "epoch": 1.4659015880146753,
      "grad_norm": 0.5539456935912866,
      "learning_rate": 3.512797908092733e-05,
      "loss": 11.8429,
      "step": 26920
    },
    {
      "epoch": 1.4659560420112583,
      "grad_norm": 0.5525381910997136,
      "learning_rate": 3.5121268400634235e-05,
      "loss": 11.9267,
      "step": 26921
    },
    {
      "epoch": 1.4660104960078413,
      "grad_norm": 0.5257762624339773,
      "learning_rate": 3.5114558224849025e-05,
      "loss": 11.8649,
      "step": 26922
    },
    {
      "epoch": 1.4660649500044243,
      "grad_norm": 0.5861112674124763,
      "learning_rate": 3.510784855362392e-05,
      "loss": 11.9031,
      "step": 26923
    },
    {
      "epoch": 1.4661194040010073,
      "grad_norm": 0.5735585913197729,
      "learning_rate": 3.510113938701105e-05,
      "loss": 11.8264,
      "step": 26924
    },
    {
      "epoch": 1.4661738579975905,
      "grad_norm": 0.5952942463803762,
      "learning_rate": 3.509443072506263e-05,
      "loss": 11.8942,
      "step": 26925
    },
    {
      "epoch": 1.4662283119941735,
      "grad_norm": 0.5060379458745126,
      "learning_rate": 3.5087722567830796e-05,
      "loss": 11.833,
      "step": 26926
    },
    {
      "epoch": 1.4662827659907565,
      "grad_norm": 0.5227254367275919,
      "learning_rate": 3.508101491536769e-05,
      "loss": 11.8399,
      "step": 26927
    },
    {
      "epoch": 1.4663372199873395,
      "grad_norm": 0.5728525605740037,
      "learning_rate": 3.507430776772553e-05,
      "loss": 11.8202,
      "step": 26928
    },
    {
      "epoch": 1.4663916739839225,
      "grad_norm": 0.578307121397152,
      "learning_rate": 3.506760112495642e-05,
      "loss": 11.9659,
      "step": 26929
    },
    {
      "epoch": 1.4664461279805054,
      "grad_norm": 0.5611285752605445,
      "learning_rate": 3.5060894987112544e-05,
      "loss": 11.6997,
      "step": 26930
    },
    {
      "epoch": 1.4665005819770884,
      "grad_norm": 0.6009530812971448,
      "learning_rate": 3.5054189354246015e-05,
      "loss": 11.8917,
      "step": 26931
    },
    {
      "epoch": 1.4665550359736714,
      "grad_norm": 0.49966910990061286,
      "learning_rate": 3.504748422640904e-05,
      "loss": 11.8914,
      "step": 26932
    },
    {
      "epoch": 1.4666094899702546,
      "grad_norm": 0.6037952896568908,
      "learning_rate": 3.5040779603653684e-05,
      "loss": 11.8515,
      "step": 26933
    },
    {
      "epoch": 1.4666639439668376,
      "grad_norm": 0.566280631163193,
      "learning_rate": 3.5034075486032115e-05,
      "loss": 11.7609,
      "step": 26934
    },
    {
      "epoch": 1.4667183979634206,
      "grad_norm": 0.571709024512032,
      "learning_rate": 3.502737187359651e-05,
      "loss": 11.8101,
      "step": 26935
    },
    {
      "epoch": 1.4667728519600036,
      "grad_norm": 0.5679542079821897,
      "learning_rate": 3.5020668766398955e-05,
      "loss": 11.8582,
      "step": 26936
    },
    {
      "epoch": 1.4668273059565866,
      "grad_norm": 0.5674774159137708,
      "learning_rate": 3.501396616449154e-05,
      "loss": 11.8548,
      "step": 26937
    },
    {
      "epoch": 1.4668817599531696,
      "grad_norm": 0.538239445264612,
      "learning_rate": 3.500726406792646e-05,
      "loss": 11.8107,
      "step": 26938
    },
    {
      "epoch": 1.4669362139497526,
      "grad_norm": 0.6377770860059915,
      "learning_rate": 3.500056247675575e-05,
      "loss": 11.749,
      "step": 26939
    },
    {
      "epoch": 1.4669906679463356,
      "grad_norm": 0.563975129858514,
      "learning_rate": 3.499386139103161e-05,
      "loss": 11.7855,
      "step": 26940
    },
    {
      "epoch": 1.4670451219429186,
      "grad_norm": 0.53408722706937,
      "learning_rate": 3.498716081080607e-05,
      "loss": 11.9532,
      "step": 26941
    },
    {
      "epoch": 1.4670995759395016,
      "grad_norm": 0.5466688182286152,
      "learning_rate": 3.49804607361313e-05,
      "loss": 11.8466,
      "step": 26942
    },
    {
      "epoch": 1.4671540299360846,
      "grad_norm": 0.6010772436464253,
      "learning_rate": 3.4973761167059346e-05,
      "loss": 11.9362,
      "step": 26943
    },
    {
      "epoch": 1.4672084839326676,
      "grad_norm": 0.5392786398843316,
      "learning_rate": 3.496706210364232e-05,
      "loss": 11.8418,
      "step": 26944
    },
    {
      "epoch": 1.4672629379292506,
      "grad_norm": 0.6659633762584934,
      "learning_rate": 3.4960363545932376e-05,
      "loss": 11.8143,
      "step": 26945
    },
    {
      "epoch": 1.4673173919258335,
      "grad_norm": 0.6231163485999179,
      "learning_rate": 3.495366549398154e-05,
      "loss": 11.775,
      "step": 26946
    },
    {
      "epoch": 1.4673718459224165,
      "grad_norm": 0.5094187662757288,
      "learning_rate": 3.49469679478419e-05,
      "loss": 11.818,
      "step": 26947
    },
    {
      "epoch": 1.4674262999189998,
      "grad_norm": 0.5494230557980302,
      "learning_rate": 3.494027090756552e-05,
      "loss": 11.9304,
      "step": 26948
    },
    {
      "epoch": 1.4674807539155827,
      "grad_norm": 0.5640995906812241,
      "learning_rate": 3.4933574373204515e-05,
      "loss": 11.9446,
      "step": 26949
    },
    {
      "epoch": 1.4675352079121657,
      "grad_norm": 0.5499026076881367,
      "learning_rate": 3.4926878344810964e-05,
      "loss": 11.9071,
      "step": 26950
    },
    {
      "epoch": 1.4675896619087487,
      "grad_norm": 0.5215088938960376,
      "learning_rate": 3.4920182822436895e-05,
      "loss": 11.8015,
      "step": 26951
    },
    {
      "epoch": 1.4676441159053317,
      "grad_norm": 0.5455027911489024,
      "learning_rate": 3.491348780613444e-05,
      "loss": 11.8753,
      "step": 26952
    },
    {
      "epoch": 1.4676985699019147,
      "grad_norm": 0.6132704465202785,
      "learning_rate": 3.490679329595558e-05,
      "loss": 11.9248,
      "step": 26953
    },
    {
      "epoch": 1.4677530238984977,
      "grad_norm": 0.5588797520801592,
      "learning_rate": 3.4900099291952435e-05,
      "loss": 11.9458,
      "step": 26954
    },
    {
      "epoch": 1.4678074778950807,
      "grad_norm": 0.5575655682178964,
      "learning_rate": 3.4893405794177e-05,
      "loss": 11.7532,
      "step": 26955
    },
    {
      "epoch": 1.4678619318916637,
      "grad_norm": 0.5791931871644307,
      "learning_rate": 3.4886712802681406e-05,
      "loss": 11.888,
      "step": 26956
    },
    {
      "epoch": 1.467916385888247,
      "grad_norm": 0.520258128178175,
      "learning_rate": 3.4880020317517645e-05,
      "loss": 11.8153,
      "step": 26957
    },
    {
      "epoch": 1.46797083988483,
      "grad_norm": 0.5632911075780257,
      "learning_rate": 3.4873328338737746e-05,
      "loss": 11.9586,
      "step": 26958
    },
    {
      "epoch": 1.468025293881413,
      "grad_norm": 0.5242617532428729,
      "learning_rate": 3.4866636866393757e-05,
      "loss": 11.8639,
      "step": 26959
    },
    {
      "epoch": 1.4680797478779959,
      "grad_norm": 0.5768426935427615,
      "learning_rate": 3.4859945900537746e-05,
      "loss": 11.8118,
      "step": 26960
    },
    {
      "epoch": 1.4681342018745789,
      "grad_norm": 0.5871673347753809,
      "learning_rate": 3.485325544122171e-05,
      "loss": 11.9576,
      "step": 26961
    },
    {
      "epoch": 1.4681886558711619,
      "grad_norm": 0.5450289652016302,
      "learning_rate": 3.484656548849771e-05,
      "loss": 11.6982,
      "step": 26962
    },
    {
      "epoch": 1.4682431098677449,
      "grad_norm": 0.5727478440550319,
      "learning_rate": 3.4839876042417695e-05,
      "loss": 11.8777,
      "step": 26963
    },
    {
      "epoch": 1.4682975638643279,
      "grad_norm": 0.547667895974564,
      "learning_rate": 3.483318710303378e-05,
      "loss": 11.9015,
      "step": 26964
    },
    {
      "epoch": 1.4683520178609109,
      "grad_norm": 0.5668453309871445,
      "learning_rate": 3.482649867039789e-05,
      "loss": 11.9573,
      "step": 26965
    },
    {
      "epoch": 1.4684064718574938,
      "grad_norm": 0.5466427199304342,
      "learning_rate": 3.481981074456211e-05,
      "loss": 11.9304,
      "step": 26966
    },
    {
      "epoch": 1.4684609258540768,
      "grad_norm": 0.5118188742957974,
      "learning_rate": 3.4813123325578415e-05,
      "loss": 11.7512,
      "step": 26967
    },
    {
      "epoch": 1.4685153798506598,
      "grad_norm": 0.5745895014674879,
      "learning_rate": 3.480643641349877e-05,
      "loss": 11.9672,
      "step": 26968
    },
    {
      "epoch": 1.4685698338472428,
      "grad_norm": 0.5358808814029702,
      "learning_rate": 3.479975000837524e-05,
      "loss": 11.8356,
      "step": 26969
    },
    {
      "epoch": 1.4686242878438258,
      "grad_norm": 0.5066478546280682,
      "learning_rate": 3.4793064110259755e-05,
      "loss": 11.8727,
      "step": 26970
    },
    {
      "epoch": 1.468678741840409,
      "grad_norm": 0.5347859064300481,
      "learning_rate": 3.478637871920435e-05,
      "loss": 11.884,
      "step": 26971
    },
    {
      "epoch": 1.468733195836992,
      "grad_norm": 0.5496757316401885,
      "learning_rate": 3.4779693835261015e-05,
      "loss": 11.8785,
      "step": 26972
    },
    {
      "epoch": 1.468787649833575,
      "grad_norm": 0.5679918776674071,
      "learning_rate": 3.47730094584817e-05,
      "loss": 11.7205,
      "step": 26973
    },
    {
      "epoch": 1.468842103830158,
      "grad_norm": 0.5752347630178402,
      "learning_rate": 3.476632558891843e-05,
      "loss": 11.7147,
      "step": 26974
    },
    {
      "epoch": 1.468896557826741,
      "grad_norm": 0.5453418953600246,
      "learning_rate": 3.475964222662311e-05,
      "loss": 11.9417,
      "step": 26975
    },
    {
      "epoch": 1.468951011823324,
      "grad_norm": 0.5675759849355251,
      "learning_rate": 3.47529593716478e-05,
      "loss": 11.823,
      "step": 26976
    },
    {
      "epoch": 1.469005465819907,
      "grad_norm": 0.5538553395942424,
      "learning_rate": 3.474627702404441e-05,
      "loss": 11.8549,
      "step": 26977
    },
    {
      "epoch": 1.46905991981649,
      "grad_norm": 0.4988055556439985,
      "learning_rate": 3.473959518386488e-05,
      "loss": 11.8171,
      "step": 26978
    },
    {
      "epoch": 1.469114373813073,
      "grad_norm": 0.5308248569857584,
      "learning_rate": 3.473291385116124e-05,
      "loss": 11.8448,
      "step": 26979
    },
    {
      "epoch": 1.4691688278096562,
      "grad_norm": 0.5654780362187766,
      "learning_rate": 3.4726233025985375e-05,
      "loss": 11.7154,
      "step": 26980
    },
    {
      "epoch": 1.4692232818062392,
      "grad_norm": 0.4706152556619751,
      "learning_rate": 3.47195527083893e-05,
      "loss": 11.8449,
      "step": 26981
    },
    {
      "epoch": 1.4692777358028222,
      "grad_norm": 0.5511450909412954,
      "learning_rate": 3.47128728984249e-05,
      "loss": 11.9013,
      "step": 26982
    },
    {
      "epoch": 1.4693321897994052,
      "grad_norm": 0.5091049474093662,
      "learning_rate": 3.4706193596144144e-05,
      "loss": 11.8955,
      "step": 26983
    },
    {
      "epoch": 1.4693866437959882,
      "grad_norm": 0.5733711417322171,
      "learning_rate": 3.469951480159901e-05,
      "loss": 11.8904,
      "step": 26984
    },
    {
      "epoch": 1.4694410977925711,
      "grad_norm": 0.50403405339265,
      "learning_rate": 3.469283651484136e-05,
      "loss": 11.951,
      "step": 26985
    },
    {
      "epoch": 1.4694955517891541,
      "grad_norm": 0.551467789854702,
      "learning_rate": 3.4686158735923216e-05,
      "loss": 11.9534,
      "step": 26986
    },
    {
      "epoch": 1.4695500057857371,
      "grad_norm": 0.5554771604025316,
      "learning_rate": 3.4679481464896446e-05,
      "loss": 11.8636,
      "step": 26987
    },
    {
      "epoch": 1.4696044597823201,
      "grad_norm": 0.5434516447398801,
      "learning_rate": 3.4672804701812945e-05,
      "loss": 11.9037,
      "step": 26988
    },
    {
      "epoch": 1.4696589137789031,
      "grad_norm": 0.5314844299161825,
      "learning_rate": 3.466612844672471e-05,
      "loss": 11.8527,
      "step": 26989
    },
    {
      "epoch": 1.4697133677754861,
      "grad_norm": 0.5504031324251071,
      "learning_rate": 3.465945269968358e-05,
      "loss": 12.0211,
      "step": 26990
    },
    {
      "epoch": 1.469767821772069,
      "grad_norm": 0.5704168540629599,
      "learning_rate": 3.465277746074154e-05,
      "loss": 11.8902,
      "step": 26991
    },
    {
      "epoch": 1.469822275768652,
      "grad_norm": 0.4879255321437034,
      "learning_rate": 3.464610272995043e-05,
      "loss": 11.7967,
      "step": 26992
    },
    {
      "epoch": 1.469876729765235,
      "grad_norm": 0.5395397518607035,
      "learning_rate": 3.4639428507362174e-05,
      "loss": 11.7737,
      "step": 26993
    },
    {
      "epoch": 1.469931183761818,
      "grad_norm": 0.5583342025955563,
      "learning_rate": 3.463275479302872e-05,
      "loss": 11.7656,
      "step": 26994
    },
    {
      "epoch": 1.4699856377584013,
      "grad_norm": 0.5333749188630824,
      "learning_rate": 3.46260815870019e-05,
      "loss": 11.8344,
      "step": 26995
    },
    {
      "epoch": 1.4700400917549843,
      "grad_norm": 0.5947621372013558,
      "learning_rate": 3.4619408889333696e-05,
      "loss": 11.835,
      "step": 26996
    },
    {
      "epoch": 1.4700945457515673,
      "grad_norm": 0.5218967318383265,
      "learning_rate": 3.461273670007587e-05,
      "loss": 11.835,
      "step": 26997
    },
    {
      "epoch": 1.4701489997481503,
      "grad_norm": 0.5692122610369998,
      "learning_rate": 3.460606501928035e-05,
      "loss": 11.8929,
      "step": 26998
    },
    {
      "epoch": 1.4702034537447333,
      "grad_norm": 0.4741647367532856,
      "learning_rate": 3.4599393846999087e-05,
      "loss": 11.6957,
      "step": 26999
    },
    {
      "epoch": 1.4702579077413163,
      "grad_norm": 0.5489315259413341,
      "learning_rate": 3.459272318328387e-05,
      "loss": 11.8165,
      "step": 27000
    },
    {
      "epoch": 1.4703123617378993,
      "grad_norm": 0.5828263256267341,
      "learning_rate": 3.4586053028186636e-05,
      "loss": 11.8819,
      "step": 27001
    },
    {
      "epoch": 1.4703668157344822,
      "grad_norm": 0.574516434369927,
      "learning_rate": 3.457938338175919e-05,
      "loss": 11.9278,
      "step": 27002
    },
    {
      "epoch": 1.4704212697310655,
      "grad_norm": 0.5907239366505275,
      "learning_rate": 3.457271424405346e-05,
      "loss": 11.928,
      "step": 27003
    },
    {
      "epoch": 1.4704757237276485,
      "grad_norm": 0.5465245700889383,
      "learning_rate": 3.4566045615121246e-05,
      "loss": 11.9369,
      "step": 27004
    },
    {
      "epoch": 1.4705301777242314,
      "grad_norm": 0.5332316149709876,
      "learning_rate": 3.455937749501445e-05,
      "loss": 11.96,
      "step": 27005
    },
    {
      "epoch": 1.4705846317208144,
      "grad_norm": 0.5861251101772974,
      "learning_rate": 3.455270988378495e-05,
      "loss": 11.8555,
      "step": 27006
    },
    {
      "epoch": 1.4706390857173974,
      "grad_norm": 0.5417709300912483,
      "learning_rate": 3.45460427814845e-05,
      "loss": 11.8343,
      "step": 27007
    },
    {
      "epoch": 1.4706935397139804,
      "grad_norm": 0.5464346475311613,
      "learning_rate": 3.4539376188165e-05,
      "loss": 12.0091,
      "step": 27008
    },
    {
      "epoch": 1.4707479937105634,
      "grad_norm": 0.5347709733589778,
      "learning_rate": 3.453271010387832e-05,
      "loss": 11.8109,
      "step": 27009
    },
    {
      "epoch": 1.4708024477071464,
      "grad_norm": 0.5393284119988231,
      "learning_rate": 3.4526044528676225e-05,
      "loss": 11.9394,
      "step": 27010
    },
    {
      "epoch": 1.4708569017037294,
      "grad_norm": 0.5806957972972273,
      "learning_rate": 3.4519379462610625e-05,
      "loss": 11.7509,
      "step": 27011
    },
    {
      "epoch": 1.4709113557003124,
      "grad_norm": 0.5465121406012963,
      "learning_rate": 3.4512714905733277e-05,
      "loss": 11.9348,
      "step": 27012
    },
    {
      "epoch": 1.4709658096968954,
      "grad_norm": 0.5724799408772205,
      "learning_rate": 3.450605085809607e-05,
      "loss": 11.9405,
      "step": 27013
    },
    {
      "epoch": 1.4710202636934784,
      "grad_norm": 0.5955513575638715,
      "learning_rate": 3.449938731975078e-05,
      "loss": 11.9518,
      "step": 27014
    },
    {
      "epoch": 1.4710747176900614,
      "grad_norm": 0.6018328515717035,
      "learning_rate": 3.4492724290749224e-05,
      "loss": 11.9525,
      "step": 27015
    },
    {
      "epoch": 1.4711291716866444,
      "grad_norm": 0.5015215623563867,
      "learning_rate": 3.4486061771143296e-05,
      "loss": 11.7717,
      "step": 27016
    },
    {
      "epoch": 1.4711836256832274,
      "grad_norm": 0.49988841942712686,
      "learning_rate": 3.4479399760984664e-05,
      "loss": 11.8735,
      "step": 27017
    },
    {
      "epoch": 1.4712380796798106,
      "grad_norm": 0.5245592975923902,
      "learning_rate": 3.447273826032525e-05,
      "loss": 11.8615,
      "step": 27018
    },
    {
      "epoch": 1.4712925336763936,
      "grad_norm": 0.649376184336951,
      "learning_rate": 3.446607726921679e-05,
      "loss": 11.9087,
      "step": 27019
    },
    {
      "epoch": 1.4713469876729766,
      "grad_norm": 0.5830823200606501,
      "learning_rate": 3.4459416787711084e-05,
      "loss": 11.947,
      "step": 27020
    },
    {
      "epoch": 1.4714014416695596,
      "grad_norm": 0.5540168823321698,
      "learning_rate": 3.445275681585999e-05,
      "loss": 11.7104,
      "step": 27021
    },
    {
      "epoch": 1.4714558956661425,
      "grad_norm": 0.5830936253241314,
      "learning_rate": 3.444609735371521e-05,
      "loss": 11.8026,
      "step": 27022
    },
    {
      "epoch": 1.4715103496627255,
      "grad_norm": 0.5401627797490386,
      "learning_rate": 3.4439438401328606e-05,
      "loss": 11.8363,
      "step": 27023
    },
    {
      "epoch": 1.4715648036593085,
      "grad_norm": 0.5427904249373643,
      "learning_rate": 3.443277995875189e-05,
      "loss": 11.7379,
      "step": 27024
    },
    {
      "epoch": 1.4716192576558915,
      "grad_norm": 0.5524654577776319,
      "learning_rate": 3.442612202603691e-05,
      "loss": 11.9151,
      "step": 27025
    },
    {
      "epoch": 1.4716737116524745,
      "grad_norm": 0.5471381024249221,
      "learning_rate": 3.4419464603235364e-05,
      "loss": 11.8217,
      "step": 27026
    },
    {
      "epoch": 1.4717281656490577,
      "grad_norm": 0.5546715885633946,
      "learning_rate": 3.4412807690399096e-05,
      "loss": 11.7589,
      "step": 27027
    },
    {
      "epoch": 1.4717826196456407,
      "grad_norm": 0.7117188961593346,
      "learning_rate": 3.440615128757984e-05,
      "loss": 11.8712,
      "step": 27028
    },
    {
      "epoch": 1.4718370736422237,
      "grad_norm": 0.5420285753965889,
      "learning_rate": 3.4399495394829305e-05,
      "loss": 11.9132,
      "step": 27029
    },
    {
      "epoch": 1.4718915276388067,
      "grad_norm": 0.543851358459557,
      "learning_rate": 3.4392840012199304e-05,
      "loss": 11.7777,
      "step": 27030
    },
    {
      "epoch": 1.4719459816353897,
      "grad_norm": 0.49003595741190664,
      "learning_rate": 3.438618513974161e-05,
      "loss": 11.801,
      "step": 27031
    },
    {
      "epoch": 1.4720004356319727,
      "grad_norm": 0.514922000773794,
      "learning_rate": 3.437953077750791e-05,
      "loss": 11.7983,
      "step": 27032
    },
    {
      "epoch": 1.4720548896285557,
      "grad_norm": 0.5555140945535076,
      "learning_rate": 3.437287692555003e-05,
      "loss": 11.7921,
      "step": 27033
    },
    {
      "epoch": 1.4721093436251387,
      "grad_norm": 0.5197018757079596,
      "learning_rate": 3.436622358391961e-05,
      "loss": 11.8424,
      "step": 27034
    },
    {
      "epoch": 1.4721637976217217,
      "grad_norm": 0.5746961787467441,
      "learning_rate": 3.435957075266849e-05,
      "loss": 11.8774,
      "step": 27035
    },
    {
      "epoch": 1.4722182516183047,
      "grad_norm": 0.5523608801389424,
      "learning_rate": 3.4352918431848313e-05,
      "loss": 11.9063,
      "step": 27036
    },
    {
      "epoch": 1.4722727056148877,
      "grad_norm": 0.6098172617612233,
      "learning_rate": 3.434626662151089e-05,
      "loss": 11.9233,
      "step": 27037
    },
    {
      "epoch": 1.4723271596114706,
      "grad_norm": 0.6189974377781735,
      "learning_rate": 3.4339615321707904e-05,
      "loss": 11.8561,
      "step": 27038
    },
    {
      "epoch": 1.4723816136080536,
      "grad_norm": 0.6098684049964824,
      "learning_rate": 3.433296453249105e-05,
      "loss": 12.0428,
      "step": 27039
    },
    {
      "epoch": 1.4724360676046366,
      "grad_norm": 0.5231442378536864,
      "learning_rate": 3.4326314253912114e-05,
      "loss": 11.9368,
      "step": 27040
    },
    {
      "epoch": 1.4724905216012198,
      "grad_norm": 0.5554595187181036,
      "learning_rate": 3.431966448602274e-05,
      "loss": 11.8724,
      "step": 27041
    },
    {
      "epoch": 1.4725449755978028,
      "grad_norm": 0.5093832129644811,
      "learning_rate": 3.431301522887467e-05,
      "loss": 11.8652,
      "step": 27042
    },
    {
      "epoch": 1.4725994295943858,
      "grad_norm": 0.532916989178289,
      "learning_rate": 3.4306366482519624e-05,
      "loss": 11.7608,
      "step": 27043
    },
    {
      "epoch": 1.4726538835909688,
      "grad_norm": 0.5299826967679284,
      "learning_rate": 3.429971824700927e-05,
      "loss": 11.7297,
      "step": 27044
    },
    {
      "epoch": 1.4727083375875518,
      "grad_norm": 0.5470163791515646,
      "learning_rate": 3.429307052239536e-05,
      "loss": 11.8786,
      "step": 27045
    },
    {
      "epoch": 1.4727627915841348,
      "grad_norm": 0.5429542506097446,
      "learning_rate": 3.4286423308729524e-05,
      "loss": 11.7596,
      "step": 27046
    },
    {
      "epoch": 1.4728172455807178,
      "grad_norm": 0.5193121313872467,
      "learning_rate": 3.4279776606063496e-05,
      "loss": 11.8058,
      "step": 27047
    },
    {
      "epoch": 1.4728716995773008,
      "grad_norm": 0.569015986442028,
      "learning_rate": 3.427313041444896e-05,
      "loss": 11.921,
      "step": 27048
    },
    {
      "epoch": 1.4729261535738838,
      "grad_norm": 0.5396491219347833,
      "learning_rate": 3.4266484733937546e-05,
      "loss": 11.9579,
      "step": 27049
    },
    {
      "epoch": 1.472980607570467,
      "grad_norm": 0.5072043815273053,
      "learning_rate": 3.425983956458101e-05,
      "loss": 11.8264,
      "step": 27050
    },
    {
      "epoch": 1.47303506156705,
      "grad_norm": 0.5150497274906632,
      "learning_rate": 3.425319490643094e-05,
      "loss": 11.8747,
      "step": 27051
    },
    {
      "epoch": 1.473089515563633,
      "grad_norm": 0.5666219037141724,
      "learning_rate": 3.42465507595391e-05,
      "loss": 11.9068,
      "step": 27052
    },
    {
      "epoch": 1.473143969560216,
      "grad_norm": 0.4999371458346644,
      "learning_rate": 3.423990712395707e-05,
      "loss": 11.8077,
      "step": 27053
    },
    {
      "epoch": 1.473198423556799,
      "grad_norm": 0.5375262988369534,
      "learning_rate": 3.4233263999736543e-05,
      "loss": 11.8238,
      "step": 27054
    },
    {
      "epoch": 1.473252877553382,
      "grad_norm": 0.5524602798318212,
      "learning_rate": 3.4226621386929224e-05,
      "loss": 11.9338,
      "step": 27055
    },
    {
      "epoch": 1.473307331549965,
      "grad_norm": 0.5441144221728975,
      "learning_rate": 3.421997928558669e-05,
      "loss": 11.8395,
      "step": 27056
    },
    {
      "epoch": 1.473361785546548,
      "grad_norm": 0.5611645080034972,
      "learning_rate": 3.421333769576067e-05,
      "loss": 11.9071,
      "step": 27057
    },
    {
      "epoch": 1.473416239543131,
      "grad_norm": 0.5502104309668763,
      "learning_rate": 3.420669661750277e-05,
      "loss": 11.7065,
      "step": 27058
    },
    {
      "epoch": 1.473470693539714,
      "grad_norm": 0.5715937041666886,
      "learning_rate": 3.4200056050864584e-05,
      "loss": 11.7687,
      "step": 27059
    },
    {
      "epoch": 1.473525147536297,
      "grad_norm": 0.6181960696255637,
      "learning_rate": 3.419341599589784e-05,
      "loss": 11.8218,
      "step": 27060
    },
    {
      "epoch": 1.47357960153288,
      "grad_norm": 0.5294926288327422,
      "learning_rate": 3.418677645265409e-05,
      "loss": 11.7405,
      "step": 27061
    },
    {
      "epoch": 1.473634055529463,
      "grad_norm": 0.8029805969217824,
      "learning_rate": 3.4180137421185035e-05,
      "loss": 11.9257,
      "step": 27062
    },
    {
      "epoch": 1.473688509526046,
      "grad_norm": 0.5349932841043552,
      "learning_rate": 3.417349890154224e-05,
      "loss": 11.8885,
      "step": 27063
    },
    {
      "epoch": 1.473742963522629,
      "grad_norm": 0.5586481148900625,
      "learning_rate": 3.416686089377735e-05,
      "loss": 11.8633,
      "step": 27064
    },
    {
      "epoch": 1.4737974175192121,
      "grad_norm": 0.5273718362697052,
      "learning_rate": 3.4160223397942037e-05,
      "loss": 11.8968,
      "step": 27065
    },
    {
      "epoch": 1.473851871515795,
      "grad_norm": 0.5669173366538417,
      "learning_rate": 3.4153586414087824e-05,
      "loss": 11.9404,
      "step": 27066
    },
    {
      "epoch": 1.473906325512378,
      "grad_norm": 0.6056961832563036,
      "learning_rate": 3.414694994226644e-05,
      "loss": 11.8837,
      "step": 27067
    },
    {
      "epoch": 1.473960779508961,
      "grad_norm": 0.5432162802028999,
      "learning_rate": 3.414031398252935e-05,
      "loss": 11.9145,
      "step": 27068
    },
    {
      "epoch": 1.474015233505544,
      "grad_norm": 0.5311333608968709,
      "learning_rate": 3.4133678534928216e-05,
      "loss": 11.8465,
      "step": 27069
    },
    {
      "epoch": 1.474069687502127,
      "grad_norm": 0.535893314131085,
      "learning_rate": 3.412704359951467e-05,
      "loss": 11.8586,
      "step": 27070
    },
    {
      "epoch": 1.47412414149871,
      "grad_norm": 0.524757750295078,
      "learning_rate": 3.412040917634026e-05,
      "loss": 11.8801,
      "step": 27071
    },
    {
      "epoch": 1.474178595495293,
      "grad_norm": 0.5756576209868174,
      "learning_rate": 3.4113775265456626e-05,
      "loss": 11.9087,
      "step": 27072
    },
    {
      "epoch": 1.4742330494918763,
      "grad_norm": 0.6782042329146925,
      "learning_rate": 3.4107141866915294e-05,
      "loss": 11.9043,
      "step": 27073
    },
    {
      "epoch": 1.4742875034884593,
      "grad_norm": 0.5520135168377776,
      "learning_rate": 3.410050898076791e-05,
      "loss": 11.9487,
      "step": 27074
    },
    {
      "epoch": 1.4743419574850423,
      "grad_norm": 0.4930499779384862,
      "learning_rate": 3.4093876607065976e-05,
      "loss": 11.8126,
      "step": 27075
    },
    {
      "epoch": 1.4743964114816253,
      "grad_norm": 0.5479393717310947,
      "learning_rate": 3.4087244745861125e-05,
      "loss": 11.8866,
      "step": 27076
    },
    {
      "epoch": 1.4744508654782083,
      "grad_norm": 0.5562895428702546,
      "learning_rate": 3.4080613397204977e-05,
      "loss": 11.906,
      "step": 27077
    },
    {
      "epoch": 1.4745053194747912,
      "grad_norm": 0.5444885985476847,
      "learning_rate": 3.407398256114896e-05,
      "loss": 11.6886,
      "step": 27078
    },
    {
      "epoch": 1.4745597734713742,
      "grad_norm": 0.5548738264640611,
      "learning_rate": 3.406735223774471e-05,
      "loss": 11.836,
      "step": 27079
    },
    {
      "epoch": 1.4746142274679572,
      "grad_norm": 0.5348139660551197,
      "learning_rate": 3.406072242704382e-05,
      "loss": 11.8202,
      "step": 27080
    },
    {
      "epoch": 1.4746686814645402,
      "grad_norm": 0.511501103488106,
      "learning_rate": 3.405409312909777e-05,
      "loss": 11.6727,
      "step": 27081
    },
    {
      "epoch": 1.4747231354611232,
      "grad_norm": 0.5598396385414746,
      "learning_rate": 3.40474643439582e-05,
      "loss": 11.8583,
      "step": 27082
    },
    {
      "epoch": 1.4747775894577062,
      "grad_norm": 0.5076146583988852,
      "learning_rate": 3.404083607167657e-05,
      "loss": 11.9152,
      "step": 27083
    },
    {
      "epoch": 1.4748320434542892,
      "grad_norm": 0.5537755440569728,
      "learning_rate": 3.403420831230449e-05,
      "loss": 11.8594,
      "step": 27084
    },
    {
      "epoch": 1.4748864974508722,
      "grad_norm": 0.5293285022532415,
      "learning_rate": 3.402758106589343e-05,
      "loss": 11.9193,
      "step": 27085
    },
    {
      "epoch": 1.4749409514474552,
      "grad_norm": 0.5408443603695325,
      "learning_rate": 3.402095433249502e-05,
      "loss": 11.7753,
      "step": 27086
    },
    {
      "epoch": 1.4749954054440382,
      "grad_norm": 0.5749917278803268,
      "learning_rate": 3.401432811216071e-05,
      "loss": 11.9071,
      "step": 27087
    },
    {
      "epoch": 1.4750498594406214,
      "grad_norm": 0.5156306979181503,
      "learning_rate": 3.400770240494202e-05,
      "loss": 11.749,
      "step": 27088
    },
    {
      "epoch": 1.4751043134372044,
      "grad_norm": 0.559651083921084,
      "learning_rate": 3.400107721089054e-05,
      "loss": 11.765,
      "step": 27089
    },
    {
      "epoch": 1.4751587674337874,
      "grad_norm": 0.612724934333478,
      "learning_rate": 3.399445253005772e-05,
      "loss": 11.8683,
      "step": 27090
    },
    {
      "epoch": 1.4752132214303704,
      "grad_norm": 0.59789341134019,
      "learning_rate": 3.39878283624951e-05,
      "loss": 11.946,
      "step": 27091
    },
    {
      "epoch": 1.4752676754269534,
      "grad_norm": 0.5544059016115417,
      "learning_rate": 3.398120470825423e-05,
      "loss": 11.8487,
      "step": 27092
    },
    {
      "epoch": 1.4753221294235364,
      "grad_norm": 0.5617610908798364,
      "learning_rate": 3.3974581567386564e-05,
      "loss": 11.8222,
      "step": 27093
    },
    {
      "epoch": 1.4753765834201193,
      "grad_norm": 0.5125756027948952,
      "learning_rate": 3.396795893994365e-05,
      "loss": 11.8395,
      "step": 27094
    },
    {
      "epoch": 1.4754310374167023,
      "grad_norm": 0.5559725961950754,
      "learning_rate": 3.396133682597692e-05,
      "loss": 11.9392,
      "step": 27095
    },
    {
      "epoch": 1.4754854914132853,
      "grad_norm": 0.5762186452667207,
      "learning_rate": 3.395471522553795e-05,
      "loss": 11.7305,
      "step": 27096
    },
    {
      "epoch": 1.4755399454098685,
      "grad_norm": 0.531698633923273,
      "learning_rate": 3.394809413867819e-05,
      "loss": 11.8277,
      "step": 27097
    },
    {
      "epoch": 1.4755943994064515,
      "grad_norm": 0.5482631295246632,
      "learning_rate": 3.394147356544909e-05,
      "loss": 11.8116,
      "step": 27098
    },
    {
      "epoch": 1.4756488534030345,
      "grad_norm": 0.6278240149208363,
      "learning_rate": 3.393485350590221e-05,
      "loss": 11.8782,
      "step": 27099
    },
    {
      "epoch": 1.4757033073996175,
      "grad_norm": 0.5759202327727695,
      "learning_rate": 3.392823396008895e-05,
      "loss": 11.7914,
      "step": 27100
    },
    {
      "epoch": 1.4757577613962005,
      "grad_norm": 0.5773957919324828,
      "learning_rate": 3.392161492806085e-05,
      "loss": 11.8509,
      "step": 27101
    },
    {
      "epoch": 1.4758122153927835,
      "grad_norm": 0.5272776305146423,
      "learning_rate": 3.3914996409869335e-05,
      "loss": 11.8905,
      "step": 27102
    },
    {
      "epoch": 1.4758666693893665,
      "grad_norm": 0.6199382628951214,
      "learning_rate": 3.3908378405565876e-05,
      "loss": 11.9084,
      "step": 27103
    },
    {
      "epoch": 1.4759211233859495,
      "grad_norm": 0.5278872701414912,
      "learning_rate": 3.3901760915201995e-05,
      "loss": 11.8079,
      "step": 27104
    },
    {
      "epoch": 1.4759755773825325,
      "grad_norm": 0.542572280098002,
      "learning_rate": 3.389514393882906e-05,
      "loss": 11.9388,
      "step": 27105
    },
    {
      "epoch": 1.4760300313791155,
      "grad_norm": 0.5820755273708499,
      "learning_rate": 3.388852747649862e-05,
      "loss": 11.899,
      "step": 27106
    },
    {
      "epoch": 1.4760844853756985,
      "grad_norm": 0.5562626129763535,
      "learning_rate": 3.388191152826207e-05,
      "loss": 11.7676,
      "step": 27107
    },
    {
      "epoch": 1.4761389393722815,
      "grad_norm": 0.5507773169644343,
      "learning_rate": 3.3875296094170826e-05,
      "loss": 11.6364,
      "step": 27108
    },
    {
      "epoch": 1.4761933933688645,
      "grad_norm": 0.5315939561427894,
      "learning_rate": 3.38686811742764e-05,
      "loss": 11.7449,
      "step": 27109
    },
    {
      "epoch": 1.4762478473654475,
      "grad_norm": 0.5459717502917776,
      "learning_rate": 3.3862066768630184e-05,
      "loss": 11.7552,
      "step": 27110
    },
    {
      "epoch": 1.4763023013620307,
      "grad_norm": 0.5277619271971716,
      "learning_rate": 3.385545287728364e-05,
      "loss": 11.8546,
      "step": 27111
    },
    {
      "epoch": 1.4763567553586137,
      "grad_norm": 0.5954012509501209,
      "learning_rate": 3.3848839500288166e-05,
      "loss": 11.8795,
      "step": 27112
    },
    {
      "epoch": 1.4764112093551967,
      "grad_norm": 0.5124358491319496,
      "learning_rate": 3.384222663769521e-05,
      "loss": 11.8748,
      "step": 27113
    },
    {
      "epoch": 1.4764656633517796,
      "grad_norm": 0.6025619047189447,
      "learning_rate": 3.383561428955623e-05,
      "loss": 11.9018,
      "step": 27114
    },
    {
      "epoch": 1.4765201173483626,
      "grad_norm": 0.5699948163670611,
      "learning_rate": 3.382900245592257e-05,
      "loss": 11.9429,
      "step": 27115
    },
    {
      "epoch": 1.4765745713449456,
      "grad_norm": 0.517490357130833,
      "learning_rate": 3.382239113684571e-05,
      "loss": 11.8018,
      "step": 27116
    },
    {
      "epoch": 1.4766290253415286,
      "grad_norm": 0.5466848815284313,
      "learning_rate": 3.3815780332377044e-05,
      "loss": 11.7851,
      "step": 27117
    },
    {
      "epoch": 1.4766834793381116,
      "grad_norm": 0.5606494793356981,
      "learning_rate": 3.3809170042567925e-05,
      "loss": 11.9099,
      "step": 27118
    },
    {
      "epoch": 1.4767379333346946,
      "grad_norm": 0.5845511369310461,
      "learning_rate": 3.3802560267469855e-05,
      "loss": 11.9356,
      "step": 27119
    },
    {
      "epoch": 1.4767923873312778,
      "grad_norm": 0.5967304989228428,
      "learning_rate": 3.379595100713413e-05,
      "loss": 11.924,
      "step": 27120
    },
    {
      "epoch": 1.4768468413278608,
      "grad_norm": 0.513447914758633,
      "learning_rate": 3.3789342261612224e-05,
      "loss": 11.9511,
      "step": 27121
    },
    {
      "epoch": 1.4769012953244438,
      "grad_norm": 0.565957331964041,
      "learning_rate": 3.378273403095547e-05,
      "loss": 11.6514,
      "step": 27122
    },
    {
      "epoch": 1.4769557493210268,
      "grad_norm": 0.5842253427170512,
      "learning_rate": 3.37761263152153e-05,
      "loss": 11.962,
      "step": 27123
    },
    {
      "epoch": 1.4770102033176098,
      "grad_norm": 0.5845901353598117,
      "learning_rate": 3.376951911444306e-05,
      "loss": 11.9118,
      "step": 27124
    },
    {
      "epoch": 1.4770646573141928,
      "grad_norm": 0.5141469175908666,
      "learning_rate": 3.376291242869014e-05,
      "loss": 11.8073,
      "step": 27125
    },
    {
      "epoch": 1.4771191113107758,
      "grad_norm": 0.5259914534495058,
      "learning_rate": 3.3756306258007954e-05,
      "loss": 11.8461,
      "step": 27126
    },
    {
      "epoch": 1.4771735653073588,
      "grad_norm": 0.5459010431752896,
      "learning_rate": 3.374970060244784e-05,
      "loss": 11.8832,
      "step": 27127
    },
    {
      "epoch": 1.4772280193039418,
      "grad_norm": 0.5528517960244562,
      "learning_rate": 3.374309546206113e-05,
      "loss": 11.823,
      "step": 27128
    },
    {
      "epoch": 1.4772824733005248,
      "grad_norm": 0.5271159165239279,
      "learning_rate": 3.373649083689926e-05,
      "loss": 11.904,
      "step": 27129
    },
    {
      "epoch": 1.4773369272971077,
      "grad_norm": 0.5966994098474412,
      "learning_rate": 3.372988672701351e-05,
      "loss": 11.9556,
      "step": 27130
    },
    {
      "epoch": 1.4773913812936907,
      "grad_norm": 0.5744693204930843,
      "learning_rate": 3.372328313245531e-05,
      "loss": 11.8177,
      "step": 27131
    },
    {
      "epoch": 1.4774458352902737,
      "grad_norm": 0.5765051641289127,
      "learning_rate": 3.371668005327594e-05,
      "loss": 11.8644,
      "step": 27132
    },
    {
      "epoch": 1.4775002892868567,
      "grad_norm": 0.5233858411786155,
      "learning_rate": 3.371007748952681e-05,
      "loss": 11.8243,
      "step": 27133
    },
    {
      "epoch": 1.47755474328344,
      "grad_norm": 0.6552093420357177,
      "learning_rate": 3.370347544125921e-05,
      "loss": 11.9979,
      "step": 27134
    },
    {
      "epoch": 1.477609197280023,
      "grad_norm": 0.559480773537894,
      "learning_rate": 3.369687390852454e-05,
      "loss": 11.865,
      "step": 27135
    },
    {
      "epoch": 1.477663651276606,
      "grad_norm": 0.5642823448196956,
      "learning_rate": 3.369027289137405e-05,
      "loss": 11.9018,
      "step": 27136
    },
    {
      "epoch": 1.477718105273189,
      "grad_norm": 0.5322223232586138,
      "learning_rate": 3.3683672389859166e-05,
      "loss": 11.8025,
      "step": 27137
    },
    {
      "epoch": 1.477772559269772,
      "grad_norm": 0.587293504402518,
      "learning_rate": 3.367707240403115e-05,
      "loss": 11.8876,
      "step": 27138
    },
    {
      "epoch": 1.477827013266355,
      "grad_norm": 0.5464093118301003,
      "learning_rate": 3.367047293394132e-05,
      "loss": 11.9181,
      "step": 27139
    },
    {
      "epoch": 1.477881467262938,
      "grad_norm": 0.5373653748455144,
      "learning_rate": 3.366387397964101e-05,
      "loss": 11.8001,
      "step": 27140
    },
    {
      "epoch": 1.477935921259521,
      "grad_norm": 0.5878366181510546,
      "learning_rate": 3.365727554118159e-05,
      "loss": 11.9296,
      "step": 27141
    },
    {
      "epoch": 1.4779903752561039,
      "grad_norm": 0.5284923666358852,
      "learning_rate": 3.365067761861427e-05,
      "loss": 11.8898,
      "step": 27142
    },
    {
      "epoch": 1.478044829252687,
      "grad_norm": 0.5664966967686911,
      "learning_rate": 3.364408021199045e-05,
      "loss": 11.8015,
      "step": 27143
    },
    {
      "epoch": 1.47809928324927,
      "grad_norm": 0.566726081494145,
      "learning_rate": 3.363748332136135e-05,
      "loss": 11.79,
      "step": 27144
    },
    {
      "epoch": 1.478153737245853,
      "grad_norm": 0.508897171974987,
      "learning_rate": 3.363088694677834e-05,
      "loss": 11.7891,
      "step": 27145
    },
    {
      "epoch": 1.478208191242436,
      "grad_norm": 0.5855396399672527,
      "learning_rate": 3.362429108829266e-05,
      "loss": 12.0,
      "step": 27146
    },
    {
      "epoch": 1.478262645239019,
      "grad_norm": 0.5839440654246064,
      "learning_rate": 3.3617695745955646e-05,
      "loss": 11.7482,
      "step": 27147
    },
    {
      "epoch": 1.478317099235602,
      "grad_norm": 0.5294692175472873,
      "learning_rate": 3.361110091981857e-05,
      "loss": 11.9796,
      "step": 27148
    },
    {
      "epoch": 1.478371553232185,
      "grad_norm": 0.5490297186905433,
      "learning_rate": 3.3604506609932674e-05,
      "loss": 11.6764,
      "step": 27149
    },
    {
      "epoch": 1.478426007228768,
      "grad_norm": 0.5190701710804367,
      "learning_rate": 3.35979128163493e-05,
      "loss": 12.0312,
      "step": 27150
    },
    {
      "epoch": 1.478480461225351,
      "grad_norm": 0.5108576585974903,
      "learning_rate": 3.3591319539119656e-05,
      "loss": 11.8501,
      "step": 27151
    },
    {
      "epoch": 1.478534915221934,
      "grad_norm": 0.5622735264811781,
      "learning_rate": 3.358472677829505e-05,
      "loss": 11.9249,
      "step": 27152
    },
    {
      "epoch": 1.478589369218517,
      "grad_norm": 0.5439652232085131,
      "learning_rate": 3.357813453392679e-05,
      "loss": 11.8032,
      "step": 27153
    },
    {
      "epoch": 1.4786438232151,
      "grad_norm": 0.5089868565051143,
      "learning_rate": 3.3571542806066047e-05,
      "loss": 11.8561,
      "step": 27154
    },
    {
      "epoch": 1.478698277211683,
      "grad_norm": 0.5547826449591965,
      "learning_rate": 3.356495159476416e-05,
      "loss": 11.774,
      "step": 27155
    },
    {
      "epoch": 1.478752731208266,
      "grad_norm": 0.5473036932778695,
      "learning_rate": 3.3558360900072325e-05,
      "loss": 11.8986,
      "step": 27156
    },
    {
      "epoch": 1.478807185204849,
      "grad_norm": 0.4951983523661329,
      "learning_rate": 3.355177072204184e-05,
      "loss": 11.88,
      "step": 27157
    },
    {
      "epoch": 1.4788616392014322,
      "grad_norm": 0.5196955795091953,
      "learning_rate": 3.354518106072394e-05,
      "loss": 11.845,
      "step": 27158
    },
    {
      "epoch": 1.4789160931980152,
      "grad_norm": 0.5520316211913328,
      "learning_rate": 3.353859191616982e-05,
      "loss": 11.8677,
      "step": 27159
    },
    {
      "epoch": 1.4789705471945982,
      "grad_norm": 0.6203878430456891,
      "learning_rate": 3.353200328843079e-05,
      "loss": 11.92,
      "step": 27160
    },
    {
      "epoch": 1.4790250011911812,
      "grad_norm": 0.5786738160243766,
      "learning_rate": 3.352541517755802e-05,
      "loss": 11.9021,
      "step": 27161
    },
    {
      "epoch": 1.4790794551877642,
      "grad_norm": 0.5143903061802476,
      "learning_rate": 3.351882758360276e-05,
      "loss": 11.8808,
      "step": 27162
    },
    {
      "epoch": 1.4791339091843472,
      "grad_norm": 0.5973229100998572,
      "learning_rate": 3.351224050661629e-05,
      "loss": 11.9367,
      "step": 27163
    },
    {
      "epoch": 1.4791883631809302,
      "grad_norm": 0.528959301008005,
      "learning_rate": 3.350565394664974e-05,
      "loss": 11.8738,
      "step": 27164
    },
    {
      "epoch": 1.4792428171775132,
      "grad_norm": 0.5771028782922764,
      "learning_rate": 3.349906790375442e-05,
      "loss": 11.9089,
      "step": 27165
    },
    {
      "epoch": 1.4792972711740962,
      "grad_norm": 0.591567422479549,
      "learning_rate": 3.349248237798146e-05,
      "loss": 11.9304,
      "step": 27166
    },
    {
      "epoch": 1.4793517251706794,
      "grad_norm": 0.48787454573548533,
      "learning_rate": 3.3485897369382146e-05,
      "loss": 11.781,
      "step": 27167
    },
    {
      "epoch": 1.4794061791672624,
      "grad_norm": 0.5086313821470303,
      "learning_rate": 3.347931287800765e-05,
      "loss": 11.7327,
      "step": 27168
    },
    {
      "epoch": 1.4794606331638454,
      "grad_norm": 0.5750569091876963,
      "learning_rate": 3.347272890390915e-05,
      "loss": 11.9718,
      "step": 27169
    },
    {
      "epoch": 1.4795150871604283,
      "grad_norm": 0.5910052756781702,
      "learning_rate": 3.346614544713789e-05,
      "loss": 11.7934,
      "step": 27170
    },
    {
      "epoch": 1.4795695411570113,
      "grad_norm": 0.7537333335813179,
      "learning_rate": 3.3459562507745e-05,
      "loss": 11.8531,
      "step": 27171
    },
    {
      "epoch": 1.4796239951535943,
      "grad_norm": 0.5711350890046221,
      "learning_rate": 3.345298008578175e-05,
      "loss": 11.8902,
      "step": 27172
    },
    {
      "epoch": 1.4796784491501773,
      "grad_norm": 0.5146417451766456,
      "learning_rate": 3.344639818129926e-05,
      "loss": 11.8732,
      "step": 27173
    },
    {
      "epoch": 1.4797329031467603,
      "grad_norm": 0.5226190157213103,
      "learning_rate": 3.343981679434873e-05,
      "loss": 11.7984,
      "step": 27174
    },
    {
      "epoch": 1.4797873571433433,
      "grad_norm": 0.5976252647813438,
      "learning_rate": 3.343323592498138e-05,
      "loss": 11.995,
      "step": 27175
    },
    {
      "epoch": 1.4798418111399263,
      "grad_norm": 0.6078734907301634,
      "learning_rate": 3.3426655573248324e-05,
      "loss": 11.7342,
      "step": 27176
    },
    {
      "epoch": 1.4798962651365093,
      "grad_norm": 0.5852577048129345,
      "learning_rate": 3.342007573920078e-05,
      "loss": 11.8116,
      "step": 27177
    },
    {
      "epoch": 1.4799507191330923,
      "grad_norm": 0.5894873191799361,
      "learning_rate": 3.34134964228899e-05,
      "loss": 11.8434,
      "step": 27178
    },
    {
      "epoch": 1.4800051731296753,
      "grad_norm": 0.5910698681153836,
      "learning_rate": 3.340691762436681e-05,
      "loss": 11.688,
      "step": 27179
    },
    {
      "epoch": 1.4800596271262583,
      "grad_norm": 0.6702628794636158,
      "learning_rate": 3.340033934368273e-05,
      "loss": 11.8892,
      "step": 27180
    },
    {
      "epoch": 1.4801140811228415,
      "grad_norm": 0.5375409497292924,
      "learning_rate": 3.3393761580888736e-05,
      "loss": 11.7689,
      "step": 27181
    },
    {
      "epoch": 1.4801685351194245,
      "grad_norm": 0.5686594922761874,
      "learning_rate": 3.338718433603606e-05,
      "loss": 11.7577,
      "step": 27182
    },
    {
      "epoch": 1.4802229891160075,
      "grad_norm": 0.5169845193658228,
      "learning_rate": 3.3380607609175785e-05,
      "loss": 11.824,
      "step": 27183
    },
    {
      "epoch": 1.4802774431125905,
      "grad_norm": 0.5274731427167206,
      "learning_rate": 3.3374031400359095e-05,
      "loss": 11.8404,
      "step": 27184
    },
    {
      "epoch": 1.4803318971091735,
      "grad_norm": 0.567724516103824,
      "learning_rate": 3.336745570963709e-05,
      "loss": 11.7453,
      "step": 27185
    },
    {
      "epoch": 1.4803863511057564,
      "grad_norm": 0.5501182615912221,
      "learning_rate": 3.336088053706092e-05,
      "loss": 11.8781,
      "step": 27186
    },
    {
      "epoch": 1.4804408051023394,
      "grad_norm": 0.5725508482541988,
      "learning_rate": 3.3354305882681746e-05,
      "loss": 11.9257,
      "step": 27187
    },
    {
      "epoch": 1.4804952590989224,
      "grad_norm": 0.6068025079210324,
      "learning_rate": 3.334773174655067e-05,
      "loss": 11.9728,
      "step": 27188
    },
    {
      "epoch": 1.4805497130955054,
      "grad_norm": 0.5027437770027873,
      "learning_rate": 3.3341158128718776e-05,
      "loss": 11.7719,
      "step": 27189
    },
    {
      "epoch": 1.4806041670920886,
      "grad_norm": 0.5177253109463887,
      "learning_rate": 3.333458502923725e-05,
      "loss": 11.7254,
      "step": 27190
    },
    {
      "epoch": 1.4806586210886716,
      "grad_norm": 0.5774954096402204,
      "learning_rate": 3.332801244815714e-05,
      "loss": 11.8728,
      "step": 27191
    },
    {
      "epoch": 1.4807130750852546,
      "grad_norm": 0.6294710587232561,
      "learning_rate": 3.332144038552961e-05,
      "loss": 12.0012,
      "step": 27192
    },
    {
      "epoch": 1.4807675290818376,
      "grad_norm": 0.5595296786294132,
      "learning_rate": 3.331486884140572e-05,
      "loss": 11.8264,
      "step": 27193
    },
    {
      "epoch": 1.4808219830784206,
      "grad_norm": 0.527091612607428,
      "learning_rate": 3.330829781583662e-05,
      "loss": 11.8461,
      "step": 27194
    },
    {
      "epoch": 1.4808764370750036,
      "grad_norm": 0.5274658985174875,
      "learning_rate": 3.330172730887334e-05,
      "loss": 11.7673,
      "step": 27195
    },
    {
      "epoch": 1.4809308910715866,
      "grad_norm": 0.5809138311464556,
      "learning_rate": 3.329515732056703e-05,
      "loss": 11.9266,
      "step": 27196
    },
    {
      "epoch": 1.4809853450681696,
      "grad_norm": 0.5390868402140865,
      "learning_rate": 3.328858785096878e-05,
      "loss": 11.8711,
      "step": 27197
    },
    {
      "epoch": 1.4810397990647526,
      "grad_norm": 0.4756542962327035,
      "learning_rate": 3.328201890012966e-05,
      "loss": 11.8575,
      "step": 27198
    },
    {
      "epoch": 1.4810942530613356,
      "grad_norm": 0.5462471967775758,
      "learning_rate": 3.3275450468100764e-05,
      "loss": 11.8843,
      "step": 27199
    },
    {
      "epoch": 1.4811487070579186,
      "grad_norm": 0.5040527822746074,
      "learning_rate": 3.3268882554933114e-05,
      "loss": 11.8946,
      "step": 27200
    },
    {
      "epoch": 1.4812031610545016,
      "grad_norm": 0.5543580122913088,
      "learning_rate": 3.326231516067782e-05,
      "loss": 11.8207,
      "step": 27201
    },
    {
      "epoch": 1.4812576150510846,
      "grad_norm": 0.6029877438599717,
      "learning_rate": 3.325574828538599e-05,
      "loss": 11.9576,
      "step": 27202
    },
    {
      "epoch": 1.4813120690476675,
      "grad_norm": 0.5570732431349827,
      "learning_rate": 3.324918192910863e-05,
      "loss": 11.6534,
      "step": 27203
    },
    {
      "epoch": 1.4813665230442508,
      "grad_norm": 0.5318652584892374,
      "learning_rate": 3.324261609189685e-05,
      "loss": 11.9389,
      "step": 27204
    },
    {
      "epoch": 1.4814209770408338,
      "grad_norm": 0.5413052857900214,
      "learning_rate": 3.323605077380164e-05,
      "loss": 11.7172,
      "step": 27205
    },
    {
      "epoch": 1.4814754310374167,
      "grad_norm": 0.5925252601545924,
      "learning_rate": 3.322948597487414e-05,
      "loss": 11.8808,
      "step": 27206
    },
    {
      "epoch": 1.4815298850339997,
      "grad_norm": 0.5347771775967283,
      "learning_rate": 3.322292169516532e-05,
      "loss": 11.9355,
      "step": 27207
    },
    {
      "epoch": 1.4815843390305827,
      "grad_norm": 0.5609134224122345,
      "learning_rate": 3.3216357934726294e-05,
      "loss": 11.8862,
      "step": 27208
    },
    {
      "epoch": 1.4816387930271657,
      "grad_norm": 0.5312910095736958,
      "learning_rate": 3.320979469360808e-05,
      "loss": 11.8217,
      "step": 27209
    },
    {
      "epoch": 1.4816932470237487,
      "grad_norm": 0.5650646943245703,
      "learning_rate": 3.320323197186165e-05,
      "loss": 11.8769,
      "step": 27210
    },
    {
      "epoch": 1.4817477010203317,
      "grad_norm": 0.5883610555504796,
      "learning_rate": 3.31966697695381e-05,
      "loss": 11.8561,
      "step": 27211
    },
    {
      "epoch": 1.4818021550169147,
      "grad_norm": 0.5382984898612555,
      "learning_rate": 3.319010808668848e-05,
      "loss": 11.7359,
      "step": 27212
    },
    {
      "epoch": 1.481856609013498,
      "grad_norm": 0.6328079032372491,
      "learning_rate": 3.318354692336375e-05,
      "loss": 11.9215,
      "step": 27213
    },
    {
      "epoch": 1.481911063010081,
      "grad_norm": 0.5521219231718056,
      "learning_rate": 3.3176986279614994e-05,
      "loss": 11.7316,
      "step": 27214
    },
    {
      "epoch": 1.481965517006664,
      "grad_norm": 0.5810596107266891,
      "learning_rate": 3.317042615549317e-05,
      "loss": 12.0288,
      "step": 27215
    },
    {
      "epoch": 1.482019971003247,
      "grad_norm": 0.5162672593751939,
      "learning_rate": 3.316386655104935e-05,
      "loss": 11.8178,
      "step": 27216
    },
    {
      "epoch": 1.4820744249998299,
      "grad_norm": 0.5837226705935764,
      "learning_rate": 3.315730746633448e-05,
      "loss": 11.9309,
      "step": 27217
    },
    {
      "epoch": 1.4821288789964129,
      "grad_norm": 0.5838394899791232,
      "learning_rate": 3.3150748901399634e-05,
      "loss": 11.8939,
      "step": 27218
    },
    {
      "epoch": 1.4821833329929959,
      "grad_norm": 0.49102244179581195,
      "learning_rate": 3.314419085629577e-05,
      "loss": 11.893,
      "step": 27219
    },
    {
      "epoch": 1.4822377869895789,
      "grad_norm": 0.5228855868017309,
      "learning_rate": 3.3137633331073855e-05,
      "loss": 11.7481,
      "step": 27220
    },
    {
      "epoch": 1.4822922409861619,
      "grad_norm": 0.5828162795906983,
      "learning_rate": 3.313107632578495e-05,
      "loss": 11.9261,
      "step": 27221
    },
    {
      "epoch": 1.4823466949827449,
      "grad_norm": 0.5306854490185376,
      "learning_rate": 3.3124519840479985e-05,
      "loss": 11.8847,
      "step": 27222
    },
    {
      "epoch": 1.4824011489793278,
      "grad_norm": 0.49920585092632636,
      "learning_rate": 3.3117963875209955e-05,
      "loss": 11.9139,
      "step": 27223
    },
    {
      "epoch": 1.4824556029759108,
      "grad_norm": 0.5191636265092698,
      "learning_rate": 3.3111408430025906e-05,
      "loss": 11.8197,
      "step": 27224
    },
    {
      "epoch": 1.4825100569724938,
      "grad_norm": 0.5373763862972796,
      "learning_rate": 3.310485350497871e-05,
      "loss": 11.9339,
      "step": 27225
    },
    {
      "epoch": 1.4825645109690768,
      "grad_norm": 0.5363460680014575,
      "learning_rate": 3.3098299100119435e-05,
      "loss": 11.8584,
      "step": 27226
    },
    {
      "epoch": 1.4826189649656598,
      "grad_norm": 0.602641280895511,
      "learning_rate": 3.309174521549897e-05,
      "loss": 11.8895,
      "step": 27227
    },
    {
      "epoch": 1.482673418962243,
      "grad_norm": 0.4825496469900355,
      "learning_rate": 3.308519185116834e-05,
      "loss": 11.8072,
      "step": 27228
    },
    {
      "epoch": 1.482727872958826,
      "grad_norm": 0.5247850017060505,
      "learning_rate": 3.3078639007178495e-05,
      "loss": 11.8737,
      "step": 27229
    },
    {
      "epoch": 1.482782326955409,
      "grad_norm": 0.5279798067277318,
      "learning_rate": 3.307208668358033e-05,
      "loss": 11.8489,
      "step": 27230
    },
    {
      "epoch": 1.482836780951992,
      "grad_norm": 0.5190133333592362,
      "learning_rate": 3.306553488042489e-05,
      "loss": 11.8621,
      "step": 27231
    },
    {
      "epoch": 1.482891234948575,
      "grad_norm": 0.5516550293797355,
      "learning_rate": 3.3058983597763026e-05,
      "loss": 11.9184,
      "step": 27232
    },
    {
      "epoch": 1.482945688945158,
      "grad_norm": 0.4975954920367842,
      "learning_rate": 3.3052432835645744e-05,
      "loss": 11.8549,
      "step": 27233
    },
    {
      "epoch": 1.483000142941741,
      "grad_norm": 0.5627263185035444,
      "learning_rate": 3.304588259412399e-05,
      "loss": 11.8656,
      "step": 27234
    },
    {
      "epoch": 1.483054596938324,
      "grad_norm": 0.5895512964169496,
      "learning_rate": 3.3039332873248666e-05,
      "loss": 12.0113,
      "step": 27235
    },
    {
      "epoch": 1.4831090509349072,
      "grad_norm": 0.54498354106306,
      "learning_rate": 3.303278367307073e-05,
      "loss": 11.7448,
      "step": 27236
    },
    {
      "epoch": 1.4831635049314902,
      "grad_norm": 0.5789294053716956,
      "learning_rate": 3.3026234993641095e-05,
      "loss": 11.9806,
      "step": 27237
    },
    {
      "epoch": 1.4832179589280732,
      "grad_norm": 0.5721138297062113,
      "learning_rate": 3.30196868350107e-05,
      "loss": 11.8589,
      "step": 27238
    },
    {
      "epoch": 1.4832724129246562,
      "grad_norm": 0.5906386088344636,
      "learning_rate": 3.301313919723046e-05,
      "loss": 11.9199,
      "step": 27239
    },
    {
      "epoch": 1.4833268669212392,
      "grad_norm": 0.5347278439720046,
      "learning_rate": 3.300659208035124e-05,
      "loss": 11.8485,
      "step": 27240
    },
    {
      "epoch": 1.4833813209178222,
      "grad_norm": 0.5051752633238162,
      "learning_rate": 3.300004548442404e-05,
      "loss": 11.815,
      "step": 27241
    },
    {
      "epoch": 1.4834357749144051,
      "grad_norm": 0.6329251790011808,
      "learning_rate": 3.299349940949968e-05,
      "loss": 11.9331,
      "step": 27242
    },
    {
      "epoch": 1.4834902289109881,
      "grad_norm": 0.5489740006734455,
      "learning_rate": 3.2986953855629145e-05,
      "loss": 11.7848,
      "step": 27243
    },
    {
      "epoch": 1.4835446829075711,
      "grad_norm": 0.5848667235338048,
      "learning_rate": 3.2980408822863264e-05,
      "loss": 11.8227,
      "step": 27244
    },
    {
      "epoch": 1.4835991369041541,
      "grad_norm": 0.5789435272763168,
      "learning_rate": 3.297386431125296e-05,
      "loss": 11.8886,
      "step": 27245
    },
    {
      "epoch": 1.4836535909007371,
      "grad_norm": 0.529242405696322,
      "learning_rate": 3.296732032084916e-05,
      "loss": 11.724,
      "step": 27246
    },
    {
      "epoch": 1.48370804489732,
      "grad_norm": 0.5779624953404515,
      "learning_rate": 3.296077685170269e-05,
      "loss": 11.8703,
      "step": 27247
    },
    {
      "epoch": 1.483762498893903,
      "grad_norm": 0.5223023362809981,
      "learning_rate": 3.2954233903864515e-05,
      "loss": 11.7858,
      "step": 27248
    },
    {
      "epoch": 1.483816952890486,
      "grad_norm": 0.5962154089421761,
      "learning_rate": 3.294769147738541e-05,
      "loss": 11.888,
      "step": 27249
    },
    {
      "epoch": 1.483871406887069,
      "grad_norm": 0.5160660724809407,
      "learning_rate": 3.2941149572316286e-05,
      "loss": 11.8307,
      "step": 27250
    },
    {
      "epoch": 1.4839258608836523,
      "grad_norm": 0.5809806957806917,
      "learning_rate": 3.2934608188708075e-05,
      "loss": 11.8838,
      "step": 27251
    },
    {
      "epoch": 1.4839803148802353,
      "grad_norm": 0.48361595968755866,
      "learning_rate": 3.292806732661156e-05,
      "loss": 11.8405,
      "step": 27252
    },
    {
      "epoch": 1.4840347688768183,
      "grad_norm": 0.4974798421988643,
      "learning_rate": 3.292152698607768e-05,
      "loss": 11.8206,
      "step": 27253
    },
    {
      "epoch": 1.4840892228734013,
      "grad_norm": 0.5968044592205556,
      "learning_rate": 3.2914987167157206e-05,
      "loss": 11.8325,
      "step": 27254
    },
    {
      "epoch": 1.4841436768699843,
      "grad_norm": 0.5455372703310573,
      "learning_rate": 3.2908447869901095e-05,
      "loss": 11.8612,
      "step": 27255
    },
    {
      "epoch": 1.4841981308665673,
      "grad_norm": 0.616961306700275,
      "learning_rate": 3.2901909094360096e-05,
      "loss": 12.0473,
      "step": 27256
    },
    {
      "epoch": 1.4842525848631503,
      "grad_norm": 0.5819114537843179,
      "learning_rate": 3.2895370840585104e-05,
      "loss": 11.7873,
      "step": 27257
    },
    {
      "epoch": 1.4843070388597333,
      "grad_norm": 0.5389869348792732,
      "learning_rate": 3.2888833108627035e-05,
      "loss": 11.851,
      "step": 27258
    },
    {
      "epoch": 1.4843614928563162,
      "grad_norm": 0.5731211277536921,
      "learning_rate": 3.2882295898536584e-05,
      "loss": 11.8152,
      "step": 27259
    },
    {
      "epoch": 1.4844159468528995,
      "grad_norm": 0.5025569124630986,
      "learning_rate": 3.287575921036467e-05,
      "loss": 11.8047,
      "step": 27260
    },
    {
      "epoch": 1.4844704008494825,
      "grad_norm": 0.5881762750936129,
      "learning_rate": 3.2869223044162126e-05,
      "loss": 11.8476,
      "step": 27261
    },
    {
      "epoch": 1.4845248548460654,
      "grad_norm": 0.5608032378207175,
      "learning_rate": 3.286268739997973e-05,
      "loss": 11.863,
      "step": 27262
    },
    {
      "epoch": 1.4845793088426484,
      "grad_norm": 0.5110285164030312,
      "learning_rate": 3.285615227786838e-05,
      "loss": 11.8661,
      "step": 27263
    },
    {
      "epoch": 1.4846337628392314,
      "grad_norm": 0.704162504125869,
      "learning_rate": 3.2849617677878805e-05,
      "loss": 11.7386,
      "step": 27264
    },
    {
      "epoch": 1.4846882168358144,
      "grad_norm": 0.5746161075372516,
      "learning_rate": 3.2843083600061895e-05,
      "loss": 11.6459,
      "step": 27265
    },
    {
      "epoch": 1.4847426708323974,
      "grad_norm": 0.5199599932562631,
      "learning_rate": 3.28365500444684e-05,
      "loss": 11.73,
      "step": 27266
    },
    {
      "epoch": 1.4847971248289804,
      "grad_norm": 0.5906330491008318,
      "learning_rate": 3.283001701114916e-05,
      "loss": 12.0179,
      "step": 27267
    },
    {
      "epoch": 1.4848515788255634,
      "grad_norm": 0.5395303467151444,
      "learning_rate": 3.282348450015503e-05,
      "loss": 12.0826,
      "step": 27268
    },
    {
      "epoch": 1.4849060328221464,
      "grad_norm": 0.51679973935423,
      "learning_rate": 3.2816952511536705e-05,
      "loss": 11.7768,
      "step": 27269
    },
    {
      "epoch": 1.4849604868187294,
      "grad_norm": 0.5479399524253692,
      "learning_rate": 3.281042104534503e-05,
      "loss": 11.8409,
      "step": 27270
    },
    {
      "epoch": 1.4850149408153124,
      "grad_norm": 0.5411507669843482,
      "learning_rate": 3.280389010163077e-05,
      "loss": 11.9591,
      "step": 27271
    },
    {
      "epoch": 1.4850693948118954,
      "grad_norm": 0.5923250906597739,
      "learning_rate": 3.279735968044473e-05,
      "loss": 11.976,
      "step": 27272
    },
    {
      "epoch": 1.4851238488084784,
      "grad_norm": 0.4882082373529916,
      "learning_rate": 3.279082978183772e-05,
      "loss": 11.8757,
      "step": 27273
    },
    {
      "epoch": 1.4851783028050616,
      "grad_norm": 0.5303569538341536,
      "learning_rate": 3.2784300405860455e-05,
      "loss": 11.9316,
      "step": 27274
    },
    {
      "epoch": 1.4852327568016446,
      "grad_norm": 0.5347246877158316,
      "learning_rate": 3.277777155256378e-05,
      "loss": 11.7729,
      "step": 27275
    },
    {
      "epoch": 1.4852872107982276,
      "grad_norm": 0.5595970988754599,
      "learning_rate": 3.2771243221998384e-05,
      "loss": 11.869,
      "step": 27276
    },
    {
      "epoch": 1.4853416647948106,
      "grad_norm": 0.5650652753851264,
      "learning_rate": 3.2764715414215106e-05,
      "loss": 11.8208,
      "step": 27277
    },
    {
      "epoch": 1.4853961187913935,
      "grad_norm": 0.5484065079949246,
      "learning_rate": 3.275818812926469e-05,
      "loss": 11.8264,
      "step": 27278
    },
    {
      "epoch": 1.4854505727879765,
      "grad_norm": 0.5189689296845389,
      "learning_rate": 3.2751661367197836e-05,
      "loss": 11.667,
      "step": 27279
    },
    {
      "epoch": 1.4855050267845595,
      "grad_norm": 0.5436340697901008,
      "learning_rate": 3.2745135128065386e-05,
      "loss": 11.914,
      "step": 27280
    },
    {
      "epoch": 1.4855594807811425,
      "grad_norm": 0.5124001145040258,
      "learning_rate": 3.2738609411918e-05,
      "loss": 11.951,
      "step": 27281
    },
    {
      "epoch": 1.4856139347777255,
      "grad_norm": 0.5277418345053874,
      "learning_rate": 3.2732084218806467e-05,
      "loss": 11.8618,
      "step": 27282
    },
    {
      "epoch": 1.4856683887743087,
      "grad_norm": 0.6326896215702091,
      "learning_rate": 3.272555954878157e-05,
      "loss": 11.7973,
      "step": 27283
    },
    {
      "epoch": 1.4857228427708917,
      "grad_norm": 0.5674423458562502,
      "learning_rate": 3.271903540189396e-05,
      "loss": 11.9219,
      "step": 27284
    },
    {
      "epoch": 1.4857772967674747,
      "grad_norm": 0.5087084102268827,
      "learning_rate": 3.271251177819446e-05,
      "loss": 11.9181,
      "step": 27285
    },
    {
      "epoch": 1.4858317507640577,
      "grad_norm": 0.8901096032229812,
      "learning_rate": 3.27059886777337e-05,
      "loss": 11.8686,
      "step": 27286
    },
    {
      "epoch": 1.4858862047606407,
      "grad_norm": 0.5073872482333592,
      "learning_rate": 3.2699466100562504e-05,
      "loss": 11.7898,
      "step": 27287
    },
    {
      "epoch": 1.4859406587572237,
      "grad_norm": 0.5211698973456306,
      "learning_rate": 3.269294404673155e-05,
      "loss": 11.8486,
      "step": 27288
    },
    {
      "epoch": 1.4859951127538067,
      "grad_norm": 0.5368291141901312,
      "learning_rate": 3.2686422516291504e-05,
      "loss": 11.8866,
      "step": 27289
    },
    {
      "epoch": 1.4860495667503897,
      "grad_norm": 0.5693300384485326,
      "learning_rate": 3.2679901509293164e-05,
      "loss": 11.8772,
      "step": 27290
    },
    {
      "epoch": 1.4861040207469727,
      "grad_norm": 0.5215937928340507,
      "learning_rate": 3.2673381025787165e-05,
      "loss": 11.869,
      "step": 27291
    },
    {
      "epoch": 1.4861584747435557,
      "grad_norm": 0.5708249218961806,
      "learning_rate": 3.266686106582428e-05,
      "loss": 11.8806,
      "step": 27292
    },
    {
      "epoch": 1.4862129287401387,
      "grad_norm": 0.63677097635389,
      "learning_rate": 3.266034162945515e-05,
      "loss": 11.8537,
      "step": 27293
    },
    {
      "epoch": 1.4862673827367217,
      "grad_norm": 0.5857846513150303,
      "learning_rate": 3.265382271673048e-05,
      "loss": 11.8732,
      "step": 27294
    },
    {
      "epoch": 1.4863218367333046,
      "grad_norm": 0.520796712326528,
      "learning_rate": 3.2647304327701034e-05,
      "loss": 11.7786,
      "step": 27295
    },
    {
      "epoch": 1.4863762907298876,
      "grad_norm": 0.5983854752980385,
      "learning_rate": 3.264078646241739e-05,
      "loss": 11.9918,
      "step": 27296
    },
    {
      "epoch": 1.4864307447264706,
      "grad_norm": 0.5101282818263799,
      "learning_rate": 3.263426912093033e-05,
      "loss": 11.6416,
      "step": 27297
    },
    {
      "epoch": 1.4864851987230538,
      "grad_norm": 0.5038209482467594,
      "learning_rate": 3.2627752303290496e-05,
      "loss": 11.7531,
      "step": 27298
    },
    {
      "epoch": 1.4865396527196368,
      "grad_norm": 0.5320127817485272,
      "learning_rate": 3.262123600954852e-05,
      "loss": 11.7726,
      "step": 27299
    },
    {
      "epoch": 1.4865941067162198,
      "grad_norm": 0.5398572104230178,
      "learning_rate": 3.2614720239755135e-05,
      "loss": 11.7259,
      "step": 27300
    },
    {
      "epoch": 1.4866485607128028,
      "grad_norm": 0.5973062249520685,
      "learning_rate": 3.2608204993960964e-05,
      "loss": 11.9197,
      "step": 27301
    },
    {
      "epoch": 1.4867030147093858,
      "grad_norm": 0.5978273206228052,
      "learning_rate": 3.260169027221672e-05,
      "loss": 11.8774,
      "step": 27302
    },
    {
      "epoch": 1.4867574687059688,
      "grad_norm": 0.6872509022041908,
      "learning_rate": 3.259517607457301e-05,
      "loss": 11.906,
      "step": 27303
    },
    {
      "epoch": 1.4868119227025518,
      "grad_norm": 0.5169416606520894,
      "learning_rate": 3.258866240108054e-05,
      "loss": 11.8721,
      "step": 27304
    },
    {
      "epoch": 1.4868663766991348,
      "grad_norm": 0.5240611112350078,
      "learning_rate": 3.25821492517899e-05,
      "loss": 11.7707,
      "step": 27305
    },
    {
      "epoch": 1.486920830695718,
      "grad_norm": 0.5654095582004188,
      "learning_rate": 3.257563662675178e-05,
      "loss": 11.8115,
      "step": 27306
    },
    {
      "epoch": 1.486975284692301,
      "grad_norm": 0.5085421091455207,
      "learning_rate": 3.256912452601685e-05,
      "loss": 11.8982,
      "step": 27307
    },
    {
      "epoch": 1.487029738688884,
      "grad_norm": 0.5707968444123652,
      "learning_rate": 3.256261294963572e-05,
      "loss": 11.7993,
      "step": 27308
    },
    {
      "epoch": 1.487084192685467,
      "grad_norm": 0.50828426846241,
      "learning_rate": 3.2556101897658976e-05,
      "loss": 11.9359,
      "step": 27309
    },
    {
      "epoch": 1.48713864668205,
      "grad_norm": 0.5131446602261969,
      "learning_rate": 3.254959137013733e-05,
      "loss": 11.7652,
      "step": 27310
    },
    {
      "epoch": 1.487193100678633,
      "grad_norm": 0.5417230565673917,
      "learning_rate": 3.254308136712135e-05,
      "loss": 12.0064,
      "step": 27311
    },
    {
      "epoch": 1.487247554675216,
      "grad_norm": 0.557403104913808,
      "learning_rate": 3.2536571888661706e-05,
      "loss": 11.9038,
      "step": 27312
    },
    {
      "epoch": 1.487302008671799,
      "grad_norm": 0.5115471585697475,
      "learning_rate": 3.253006293480897e-05,
      "loss": 11.8616,
      "step": 27313
    },
    {
      "epoch": 1.487356462668382,
      "grad_norm": 0.5656769013152173,
      "learning_rate": 3.2523554505613796e-05,
      "loss": 11.8502,
      "step": 27314
    },
    {
      "epoch": 1.487410916664965,
      "grad_norm": 0.4902086284984136,
      "learning_rate": 3.251704660112675e-05,
      "loss": 11.773,
      "step": 27315
    },
    {
      "epoch": 1.487465370661548,
      "grad_norm": 0.5067827185346057,
      "learning_rate": 3.2510539221398475e-05,
      "loss": 11.8763,
      "step": 27316
    },
    {
      "epoch": 1.487519824658131,
      "grad_norm": 0.5197292409505939,
      "learning_rate": 3.25040323664796e-05,
      "loss": 11.8191,
      "step": 27317
    },
    {
      "epoch": 1.487574278654714,
      "grad_norm": 0.5697866881209924,
      "learning_rate": 3.2497526036420644e-05,
      "loss": 11.8394,
      "step": 27318
    },
    {
      "epoch": 1.487628732651297,
      "grad_norm": 0.5313578729803738,
      "learning_rate": 3.2491020231272317e-05,
      "loss": 11.7565,
      "step": 27319
    },
    {
      "epoch": 1.48768318664788,
      "grad_norm": 0.5369869662143938,
      "learning_rate": 3.2484514951085065e-05,
      "loss": 11.9415,
      "step": 27320
    },
    {
      "epoch": 1.4877376406444631,
      "grad_norm": 0.5238407392481245,
      "learning_rate": 3.2478010195909556e-05,
      "loss": 11.8026,
      "step": 27321
    },
    {
      "epoch": 1.4877920946410461,
      "grad_norm": 0.5856891411990525,
      "learning_rate": 3.247150596579639e-05,
      "loss": 11.9195,
      "step": 27322
    },
    {
      "epoch": 1.487846548637629,
      "grad_norm": 0.5187091122853142,
      "learning_rate": 3.24650022607961e-05,
      "loss": 11.9496,
      "step": 27323
    },
    {
      "epoch": 1.487901002634212,
      "grad_norm": 0.5612849698542438,
      "learning_rate": 3.245849908095929e-05,
      "loss": 11.8396,
      "step": 27324
    },
    {
      "epoch": 1.487955456630795,
      "grad_norm": 0.5008938401427246,
      "learning_rate": 3.245199642633649e-05,
      "loss": 11.7777,
      "step": 27325
    },
    {
      "epoch": 1.488009910627378,
      "grad_norm": 0.5242050525716808,
      "learning_rate": 3.244549429697834e-05,
      "loss": 11.962,
      "step": 27326
    },
    {
      "epoch": 1.488064364623961,
      "grad_norm": 0.5871022906923379,
      "learning_rate": 3.243899269293531e-05,
      "loss": 11.8592,
      "step": 27327
    },
    {
      "epoch": 1.488118818620544,
      "grad_norm": 0.5388255460765787,
      "learning_rate": 3.243249161425801e-05,
      "loss": 11.6366,
      "step": 27328
    },
    {
      "epoch": 1.488173272617127,
      "grad_norm": 0.5691555834204939,
      "learning_rate": 3.242599106099704e-05,
      "loss": 11.8457,
      "step": 27329
    },
    {
      "epoch": 1.4882277266137103,
      "grad_norm": 0.53414967703767,
      "learning_rate": 3.2419491033202843e-05,
      "loss": 11.8391,
      "step": 27330
    },
    {
      "epoch": 1.4882821806102933,
      "grad_norm": 0.5408069509464369,
      "learning_rate": 3.241299153092601e-05,
      "loss": 11.8947,
      "step": 27331
    },
    {
      "epoch": 1.4883366346068763,
      "grad_norm": 0.5642851220105962,
      "learning_rate": 3.240649255421713e-05,
      "loss": 11.8473,
      "step": 27332
    },
    {
      "epoch": 1.4883910886034593,
      "grad_norm": 0.5484846794469815,
      "learning_rate": 3.239999410312665e-05,
      "loss": 11.9208,
      "step": 27333
    },
    {
      "epoch": 1.4884455426000422,
      "grad_norm": 0.47128622223821753,
      "learning_rate": 3.23934961777052e-05,
      "loss": 11.8681,
      "step": 27334
    },
    {
      "epoch": 1.4884999965966252,
      "grad_norm": 0.5297904201959726,
      "learning_rate": 3.238699877800322e-05,
      "loss": 11.9234,
      "step": 27335
    },
    {
      "epoch": 1.4885544505932082,
      "grad_norm": 0.4994056879425562,
      "learning_rate": 3.2380501904071315e-05,
      "loss": 11.8726,
      "step": 27336
    },
    {
      "epoch": 1.4886089045897912,
      "grad_norm": 0.6012553319637874,
      "learning_rate": 3.237400555595993e-05,
      "loss": 11.8352,
      "step": 27337
    },
    {
      "epoch": 1.4886633585863742,
      "grad_norm": 0.5717342952793103,
      "learning_rate": 3.236750973371966e-05,
      "loss": 11.7764,
      "step": 27338
    },
    {
      "epoch": 1.4887178125829572,
      "grad_norm": 0.5721044207906831,
      "learning_rate": 3.236101443740097e-05,
      "loss": 11.9964,
      "step": 27339
    },
    {
      "epoch": 1.4887722665795402,
      "grad_norm": 0.543351701834864,
      "learning_rate": 3.235451966705434e-05,
      "loss": 11.6909,
      "step": 27340
    },
    {
      "epoch": 1.4888267205761232,
      "grad_norm": 0.5593434669192275,
      "learning_rate": 3.234802542273034e-05,
      "loss": 11.8715,
      "step": 27341
    },
    {
      "epoch": 1.4888811745727062,
      "grad_norm": 0.5956820363149719,
      "learning_rate": 3.234153170447941e-05,
      "loss": 11.8743,
      "step": 27342
    },
    {
      "epoch": 1.4889356285692892,
      "grad_norm": 0.5586765541934234,
      "learning_rate": 3.233503851235208e-05,
      "loss": 11.86,
      "step": 27343
    },
    {
      "epoch": 1.4889900825658724,
      "grad_norm": 0.5375570701571059,
      "learning_rate": 3.232854584639887e-05,
      "loss": 11.8627,
      "step": 27344
    },
    {
      "epoch": 1.4890445365624554,
      "grad_norm": 0.5788137075225459,
      "learning_rate": 3.2322053706670194e-05,
      "loss": 11.8417,
      "step": 27345
    },
    {
      "epoch": 1.4890989905590384,
      "grad_norm": 0.5739870435953575,
      "learning_rate": 3.231556209321662e-05,
      "loss": 11.8072,
      "step": 27346
    },
    {
      "epoch": 1.4891534445556214,
      "grad_norm": 0.5200634454133051,
      "learning_rate": 3.2309071006088554e-05,
      "loss": 11.8978,
      "step": 27347
    },
    {
      "epoch": 1.4892078985522044,
      "grad_norm": 0.5267569385414985,
      "learning_rate": 3.230258044533653e-05,
      "loss": 11.791,
      "step": 27348
    },
    {
      "epoch": 1.4892623525487874,
      "grad_norm": 0.5257919766051812,
      "learning_rate": 3.229609041101099e-05,
      "loss": 11.7765,
      "step": 27349
    },
    {
      "epoch": 1.4893168065453704,
      "grad_norm": 0.5379618945486823,
      "learning_rate": 3.228960090316239e-05,
      "loss": 11.7958,
      "step": 27350
    },
    {
      "epoch": 1.4893712605419533,
      "grad_norm": 0.5464085762061488,
      "learning_rate": 3.228311192184122e-05,
      "loss": 11.9128,
      "step": 27351
    },
    {
      "epoch": 1.4894257145385363,
      "grad_norm": 0.5120607901818375,
      "learning_rate": 3.227662346709791e-05,
      "loss": 11.8575,
      "step": 27352
    },
    {
      "epoch": 1.4894801685351196,
      "grad_norm": 0.555661724649654,
      "learning_rate": 3.227013553898296e-05,
      "loss": 11.8716,
      "step": 27353
    },
    {
      "epoch": 1.4895346225317025,
      "grad_norm": 0.5199119991369252,
      "learning_rate": 3.2263648137546765e-05,
      "loss": 11.8153,
      "step": 27354
    },
    {
      "epoch": 1.4895890765282855,
      "grad_norm": 0.514488265265781,
      "learning_rate": 3.22571612628398e-05,
      "loss": 11.7517,
      "step": 27355
    },
    {
      "epoch": 1.4896435305248685,
      "grad_norm": 0.5500673005441195,
      "learning_rate": 3.2250674914912536e-05,
      "loss": 11.918,
      "step": 27356
    },
    {
      "epoch": 1.4896979845214515,
      "grad_norm": 0.6316095639101168,
      "learning_rate": 3.224418909381536e-05,
      "loss": 11.861,
      "step": 27357
    },
    {
      "epoch": 1.4897524385180345,
      "grad_norm": 0.5401492324875024,
      "learning_rate": 3.223770379959876e-05,
      "loss": 11.827,
      "step": 27358
    },
    {
      "epoch": 1.4898068925146175,
      "grad_norm": 0.5351032328964,
      "learning_rate": 3.2231219032313144e-05,
      "loss": 11.917,
      "step": 27359
    },
    {
      "epoch": 1.4898613465112005,
      "grad_norm": 0.5419580382352344,
      "learning_rate": 3.222473479200889e-05,
      "loss": 11.8503,
      "step": 27360
    },
    {
      "epoch": 1.4899158005077835,
      "grad_norm": 0.5611074554911892,
      "learning_rate": 3.2218251078736505e-05,
      "loss": 11.9644,
      "step": 27361
    },
    {
      "epoch": 1.4899702545043665,
      "grad_norm": 0.5479997393111928,
      "learning_rate": 3.2211767892546344e-05,
      "loss": 11.8396,
      "step": 27362
    },
    {
      "epoch": 1.4900247085009495,
      "grad_norm": 0.566210287760261,
      "learning_rate": 3.220528523348886e-05,
      "loss": 11.7726,
      "step": 27363
    },
    {
      "epoch": 1.4900791624975325,
      "grad_norm": 0.5335510246762156,
      "learning_rate": 3.2198803101614425e-05,
      "loss": 11.9138,
      "step": 27364
    },
    {
      "epoch": 1.4901336164941155,
      "grad_norm": 0.5369260579828239,
      "learning_rate": 3.2192321496973465e-05,
      "loss": 11.6365,
      "step": 27365
    },
    {
      "epoch": 1.4901880704906985,
      "grad_norm": 0.5763778858222389,
      "learning_rate": 3.218584041961641e-05,
      "loss": 11.9058,
      "step": 27366
    },
    {
      "epoch": 1.4902425244872815,
      "grad_norm": 0.5695051398147223,
      "learning_rate": 3.217935986959361e-05,
      "loss": 11.9074,
      "step": 27367
    },
    {
      "epoch": 1.4902969784838647,
      "grad_norm": 0.5087613081774589,
      "learning_rate": 3.217287984695551e-05,
      "loss": 11.7241,
      "step": 27368
    },
    {
      "epoch": 1.4903514324804477,
      "grad_norm": 0.5395579202495738,
      "learning_rate": 3.2166400351752465e-05,
      "loss": 11.7826,
      "step": 27369
    },
    {
      "epoch": 1.4904058864770307,
      "grad_norm": 0.5027300400031914,
      "learning_rate": 3.215992138403484e-05,
      "loss": 11.7519,
      "step": 27370
    },
    {
      "epoch": 1.4904603404736136,
      "grad_norm": 0.5456917563778277,
      "learning_rate": 3.215344294385307e-05,
      "loss": 11.8791,
      "step": 27371
    },
    {
      "epoch": 1.4905147944701966,
      "grad_norm": 0.5098160299614349,
      "learning_rate": 3.214696503125748e-05,
      "loss": 11.7655,
      "step": 27372
    },
    {
      "epoch": 1.4905692484667796,
      "grad_norm": 0.5985508874443944,
      "learning_rate": 3.21404876462985e-05,
      "loss": 11.8313,
      "step": 27373
    },
    {
      "epoch": 1.4906237024633626,
      "grad_norm": 0.5881527697539021,
      "learning_rate": 3.213401078902644e-05,
      "loss": 11.9214,
      "step": 27374
    },
    {
      "epoch": 1.4906781564599456,
      "grad_norm": 0.5772487234516996,
      "learning_rate": 3.212753445949172e-05,
      "loss": 11.7601,
      "step": 27375
    },
    {
      "epoch": 1.4907326104565288,
      "grad_norm": 0.5953524314552884,
      "learning_rate": 3.212105865774464e-05,
      "loss": 11.8854,
      "step": 27376
    },
    {
      "epoch": 1.4907870644531118,
      "grad_norm": 0.5253379304485232,
      "learning_rate": 3.2114583383835595e-05,
      "loss": 11.7607,
      "step": 27377
    },
    {
      "epoch": 1.4908415184496948,
      "grad_norm": 0.5686002338526486,
      "learning_rate": 3.210810863781496e-05,
      "loss": 11.7952,
      "step": 27378
    },
    {
      "epoch": 1.4908959724462778,
      "grad_norm": 0.5781451795322011,
      "learning_rate": 3.210163441973306e-05,
      "loss": 11.8081,
      "step": 27379
    },
    {
      "epoch": 1.4909504264428608,
      "grad_norm": 0.550138091581712,
      "learning_rate": 3.209516072964022e-05,
      "loss": 11.8952,
      "step": 27380
    },
    {
      "epoch": 1.4910048804394438,
      "grad_norm": 0.5790573792479207,
      "learning_rate": 3.2088687567586804e-05,
      "loss": 11.7505,
      "step": 27381
    },
    {
      "epoch": 1.4910593344360268,
      "grad_norm": 0.5289096816977832,
      "learning_rate": 3.2082214933623134e-05,
      "loss": 11.7907,
      "step": 27382
    },
    {
      "epoch": 1.4911137884326098,
      "grad_norm": 0.46653037365099215,
      "learning_rate": 3.207574282779957e-05,
      "loss": 11.8671,
      "step": 27383
    },
    {
      "epoch": 1.4911682424291928,
      "grad_norm": 0.5352656293213426,
      "learning_rate": 3.2069271250166385e-05,
      "loss": 11.8939,
      "step": 27384
    },
    {
      "epoch": 1.4912226964257758,
      "grad_norm": 0.4975280705881872,
      "learning_rate": 3.206280020077398e-05,
      "loss": 11.9769,
      "step": 27385
    },
    {
      "epoch": 1.4912771504223588,
      "grad_norm": 0.5429788044732909,
      "learning_rate": 3.2056329679672606e-05,
      "loss": 11.8679,
      "step": 27386
    },
    {
      "epoch": 1.4913316044189417,
      "grad_norm": 0.551843620543023,
      "learning_rate": 3.2049859686912584e-05,
      "loss": 11.7358,
      "step": 27387
    },
    {
      "epoch": 1.4913860584155247,
      "grad_norm": 0.52908555993108,
      "learning_rate": 3.20433902225443e-05,
      "loss": 11.7486,
      "step": 27388
    },
    {
      "epoch": 1.4914405124121077,
      "grad_norm": 0.8751017592020649,
      "learning_rate": 3.2036921286618e-05,
      "loss": 11.9371,
      "step": 27389
    },
    {
      "epoch": 1.4914949664086907,
      "grad_norm": 0.6231913141871868,
      "learning_rate": 3.2030452879184014e-05,
      "loss": 11.8704,
      "step": 27390
    },
    {
      "epoch": 1.491549420405274,
      "grad_norm": 0.5603397525587256,
      "learning_rate": 3.2023985000292575e-05,
      "loss": 11.8782,
      "step": 27391
    },
    {
      "epoch": 1.491603874401857,
      "grad_norm": 0.6754076910478701,
      "learning_rate": 3.201751764999403e-05,
      "loss": 11.9525,
      "step": 27392
    },
    {
      "epoch": 1.49165832839844,
      "grad_norm": 0.5935902729391448,
      "learning_rate": 3.201105082833872e-05,
      "loss": 11.8441,
      "step": 27393
    },
    {
      "epoch": 1.491712782395023,
      "grad_norm": 0.5449317540943305,
      "learning_rate": 3.200458453537682e-05,
      "loss": 11.8019,
      "step": 27394
    },
    {
      "epoch": 1.491767236391606,
      "grad_norm": 0.5499611987491785,
      "learning_rate": 3.199811877115873e-05,
      "loss": 11.8566,
      "step": 27395
    },
    {
      "epoch": 1.491821690388189,
      "grad_norm": 0.5725463749876839,
      "learning_rate": 3.199165353573462e-05,
      "loss": 11.7469,
      "step": 27396
    },
    {
      "epoch": 1.491876144384772,
      "grad_norm": 0.5345352879477128,
      "learning_rate": 3.198518882915487e-05,
      "loss": 11.8076,
      "step": 27397
    },
    {
      "epoch": 1.4919305983813549,
      "grad_norm": 0.5150328479226377,
      "learning_rate": 3.1978724651469647e-05,
      "loss": 11.8037,
      "step": 27398
    },
    {
      "epoch": 1.4919850523779379,
      "grad_norm": 0.5672482133986002,
      "learning_rate": 3.197226100272931e-05,
      "loss": 11.9698,
      "step": 27399
    },
    {
      "epoch": 1.492039506374521,
      "grad_norm": 0.540766156130379,
      "learning_rate": 3.196579788298407e-05,
      "loss": 11.7337,
      "step": 27400
    },
    {
      "epoch": 1.492093960371104,
      "grad_norm": 0.5686484632273681,
      "learning_rate": 3.1959335292284175e-05,
      "loss": 11.9007,
      "step": 27401
    },
    {
      "epoch": 1.492148414367687,
      "grad_norm": 0.5120570343678219,
      "learning_rate": 3.1952873230679926e-05,
      "loss": 11.7939,
      "step": 27402
    },
    {
      "epoch": 1.49220286836427,
      "grad_norm": 0.5517134705371104,
      "learning_rate": 3.194641169822152e-05,
      "loss": 11.6883,
      "step": 27403
    },
    {
      "epoch": 1.492257322360853,
      "grad_norm": 0.5242483670753166,
      "learning_rate": 3.193995069495922e-05,
      "loss": 11.7722,
      "step": 27404
    },
    {
      "epoch": 1.492311776357436,
      "grad_norm": 0.6385521546705989,
      "learning_rate": 3.193349022094332e-05,
      "loss": 11.9355,
      "step": 27405
    },
    {
      "epoch": 1.492366230354019,
      "grad_norm": 0.5338069368985067,
      "learning_rate": 3.1927030276223966e-05,
      "loss": 11.7156,
      "step": 27406
    },
    {
      "epoch": 1.492420684350602,
      "grad_norm": 0.5373913974375919,
      "learning_rate": 3.192057086085148e-05,
      "loss": 11.8471,
      "step": 27407
    },
    {
      "epoch": 1.492475138347185,
      "grad_norm": 0.5928493974126836,
      "learning_rate": 3.1914111974876026e-05,
      "loss": 11.7267,
      "step": 27408
    },
    {
      "epoch": 1.492529592343768,
      "grad_norm": 0.5168927346027926,
      "learning_rate": 3.1907653618347886e-05,
      "loss": 11.8028,
      "step": 27409
    },
    {
      "epoch": 1.492584046340351,
      "grad_norm": 0.5313684037282164,
      "learning_rate": 3.190119579131725e-05,
      "loss": 11.8494,
      "step": 27410
    },
    {
      "epoch": 1.492638500336934,
      "grad_norm": 0.5136660413035197,
      "learning_rate": 3.18947384938343e-05,
      "loss": 11.7845,
      "step": 27411
    },
    {
      "epoch": 1.492692954333517,
      "grad_norm": 0.5357556278925202,
      "learning_rate": 3.188828172594932e-05,
      "loss": 11.8182,
      "step": 27412
    },
    {
      "epoch": 1.4927474083301,
      "grad_norm": 0.5275353878504169,
      "learning_rate": 3.188182548771245e-05,
      "loss": 11.7841,
      "step": 27413
    },
    {
      "epoch": 1.4928018623266832,
      "grad_norm": 0.5064172315746848,
      "learning_rate": 3.1875369779173924e-05,
      "loss": 11.7459,
      "step": 27414
    },
    {
      "epoch": 1.4928563163232662,
      "grad_norm": 0.5470100177461124,
      "learning_rate": 3.1868914600383994e-05,
      "loss": 11.9278,
      "step": 27415
    },
    {
      "epoch": 1.4929107703198492,
      "grad_norm": 0.4988042980869149,
      "learning_rate": 3.186245995139276e-05,
      "loss": 11.8884,
      "step": 27416
    },
    {
      "epoch": 1.4929652243164322,
      "grad_norm": 0.542600917398713,
      "learning_rate": 3.18560058322505e-05,
      "loss": 11.7948,
      "step": 27417
    },
    {
      "epoch": 1.4930196783130152,
      "grad_norm": 0.5337346171684353,
      "learning_rate": 3.184955224300734e-05,
      "loss": 11.8023,
      "step": 27418
    },
    {
      "epoch": 1.4930741323095982,
      "grad_norm": 0.5068117506775958,
      "learning_rate": 3.184309918371352e-05,
      "loss": 11.751,
      "step": 27419
    },
    {
      "epoch": 1.4931285863061812,
      "grad_norm": 0.5335797519704031,
      "learning_rate": 3.18366466544192e-05,
      "loss": 11.76,
      "step": 27420
    },
    {
      "epoch": 1.4931830403027642,
      "grad_norm": 0.5469394051136713,
      "learning_rate": 3.1830194655174505e-05,
      "loss": 11.7379,
      "step": 27421
    },
    {
      "epoch": 1.4932374942993472,
      "grad_norm": 0.5380651376103969,
      "learning_rate": 3.1823743186029675e-05,
      "loss": 11.9291,
      "step": 27422
    },
    {
      "epoch": 1.4932919482959304,
      "grad_norm": 0.50377001974569,
      "learning_rate": 3.1817292247034824e-05,
      "loss": 11.8882,
      "step": 27423
    },
    {
      "epoch": 1.4933464022925134,
      "grad_norm": 0.5398332871826095,
      "learning_rate": 3.181084183824018e-05,
      "loss": 11.8276,
      "step": 27424
    },
    {
      "epoch": 1.4934008562890964,
      "grad_norm": 0.5402220267565918,
      "learning_rate": 3.180439195969582e-05,
      "loss": 11.8085,
      "step": 27425
    },
    {
      "epoch": 1.4934553102856793,
      "grad_norm": 0.5355265862902128,
      "learning_rate": 3.179794261145196e-05,
      "loss": 11.8738,
      "step": 27426
    },
    {
      "epoch": 1.4935097642822623,
      "grad_norm": 0.55237386828305,
      "learning_rate": 3.179149379355877e-05,
      "loss": 11.8119,
      "step": 27427
    },
    {
      "epoch": 1.4935642182788453,
      "grad_norm": 0.497504691561615,
      "learning_rate": 3.178504550606631e-05,
      "loss": 11.9204,
      "step": 27428
    },
    {
      "epoch": 1.4936186722754283,
      "grad_norm": 0.5127452387862489,
      "learning_rate": 3.177859774902483e-05,
      "loss": 11.8487,
      "step": 27429
    },
    {
      "epoch": 1.4936731262720113,
      "grad_norm": 0.5758347834761272,
      "learning_rate": 3.17721505224844e-05,
      "loss": 11.9149,
      "step": 27430
    },
    {
      "epoch": 1.4937275802685943,
      "grad_norm": 0.5547243846772557,
      "learning_rate": 3.1765703826495144e-05,
      "loss": 11.7884,
      "step": 27431
    },
    {
      "epoch": 1.4937820342651773,
      "grad_norm": 0.5571913268408929,
      "learning_rate": 3.1759257661107245e-05,
      "loss": 11.715,
      "step": 27432
    },
    {
      "epoch": 1.4938364882617603,
      "grad_norm": 0.5442979747736589,
      "learning_rate": 3.175281202637077e-05,
      "loss": 11.8839,
      "step": 27433
    },
    {
      "epoch": 1.4938909422583433,
      "grad_norm": 0.5704327202416556,
      "learning_rate": 3.174636692233591e-05,
      "loss": 11.8487,
      "step": 27434
    },
    {
      "epoch": 1.4939453962549263,
      "grad_norm": 0.5372537584832451,
      "learning_rate": 3.173992234905272e-05,
      "loss": 11.8977,
      "step": 27435
    },
    {
      "epoch": 1.4939998502515093,
      "grad_norm": 0.575209458369802,
      "learning_rate": 3.173347830657133e-05,
      "loss": 11.9214,
      "step": 27436
    },
    {
      "epoch": 1.4940543042480925,
      "grad_norm": 0.5796686629508384,
      "learning_rate": 3.17270347949419e-05,
      "loss": 11.8689,
      "step": 27437
    },
    {
      "epoch": 1.4941087582446755,
      "grad_norm": 0.5433666967461873,
      "learning_rate": 3.1720591814214464e-05,
      "loss": 11.9079,
      "step": 27438
    },
    {
      "epoch": 1.4941632122412585,
      "grad_norm": 0.5236638767969538,
      "learning_rate": 3.1714149364439215e-05,
      "loss": 11.852,
      "step": 27439
    },
    {
      "epoch": 1.4942176662378415,
      "grad_norm": 0.5577906353201473,
      "learning_rate": 3.1707707445666135e-05,
      "loss": 11.8241,
      "step": 27440
    },
    {
      "epoch": 1.4942721202344245,
      "grad_norm": 0.49045688724140346,
      "learning_rate": 3.1701266057945376e-05,
      "loss": 11.8646,
      "step": 27441
    },
    {
      "epoch": 1.4943265742310075,
      "grad_norm": 0.5170163146468425,
      "learning_rate": 3.169482520132705e-05,
      "loss": 11.8163,
      "step": 27442
    },
    {
      "epoch": 1.4943810282275904,
      "grad_norm": 0.5074604276050942,
      "learning_rate": 3.168838487586119e-05,
      "loss": 11.8348,
      "step": 27443
    },
    {
      "epoch": 1.4944354822241734,
      "grad_norm": 0.4994581872740865,
      "learning_rate": 3.168194508159794e-05,
      "loss": 11.8019,
      "step": 27444
    },
    {
      "epoch": 1.4944899362207564,
      "grad_norm": 0.5539482960695543,
      "learning_rate": 3.16755058185873e-05,
      "loss": 11.9257,
      "step": 27445
    },
    {
      "epoch": 1.4945443902173396,
      "grad_norm": 0.5394138641393187,
      "learning_rate": 3.166906708687943e-05,
      "loss": 11.8341,
      "step": 27446
    },
    {
      "epoch": 1.4945988442139226,
      "grad_norm": 0.5958720947355558,
      "learning_rate": 3.1662628886524314e-05,
      "loss": 11.7851,
      "step": 27447
    },
    {
      "epoch": 1.4946532982105056,
      "grad_norm": 0.5055716382702036,
      "learning_rate": 3.165619121757206e-05,
      "loss": 11.9507,
      "step": 27448
    },
    {
      "epoch": 1.4947077522070886,
      "grad_norm": 0.5850583357709767,
      "learning_rate": 3.164975408007279e-05,
      "loss": 11.8939,
      "step": 27449
    },
    {
      "epoch": 1.4947622062036716,
      "grad_norm": 0.5091230101697962,
      "learning_rate": 3.1643317474076415e-05,
      "loss": 11.8763,
      "step": 27450
    },
    {
      "epoch": 1.4948166602002546,
      "grad_norm": 0.5378135812842038,
      "learning_rate": 3.163688139963311e-05,
      "loss": 11.8288,
      "step": 27451
    },
    {
      "epoch": 1.4948711141968376,
      "grad_norm": 0.550914214934827,
      "learning_rate": 3.163044585679286e-05,
      "loss": 11.957,
      "step": 27452
    },
    {
      "epoch": 1.4949255681934206,
      "grad_norm": 0.5402679003309571,
      "learning_rate": 3.162401084560571e-05,
      "loss": 11.8529,
      "step": 27453
    },
    {
      "epoch": 1.4949800221900036,
      "grad_norm": 0.5627741211877516,
      "learning_rate": 3.161757636612176e-05,
      "loss": 11.8402,
      "step": 27454
    },
    {
      "epoch": 1.4950344761865866,
      "grad_norm": 0.5389489879191961,
      "learning_rate": 3.161114241839096e-05,
      "loss": 11.7944,
      "step": 27455
    },
    {
      "epoch": 1.4950889301831696,
      "grad_norm": 0.5577135803377378,
      "learning_rate": 3.1604709002463426e-05,
      "loss": 11.8685,
      "step": 27456
    },
    {
      "epoch": 1.4951433841797526,
      "grad_norm": 0.5315667289223625,
      "learning_rate": 3.159827611838912e-05,
      "loss": 11.853,
      "step": 27457
    },
    {
      "epoch": 1.4951978381763356,
      "grad_norm": 0.5531299342302193,
      "learning_rate": 3.159184376621811e-05,
      "loss": 11.8088,
      "step": 27458
    },
    {
      "epoch": 1.4952522921729186,
      "grad_norm": 0.5639067210014592,
      "learning_rate": 3.15854119460004e-05,
      "loss": 11.87,
      "step": 27459
    },
    {
      "epoch": 1.4953067461695015,
      "grad_norm": 0.5060470268627262,
      "learning_rate": 3.1578980657785975e-05,
      "loss": 11.9204,
      "step": 27460
    },
    {
      "epoch": 1.4953612001660848,
      "grad_norm": 0.7839953306743349,
      "learning_rate": 3.1572549901624895e-05,
      "loss": 11.9339,
      "step": 27461
    },
    {
      "epoch": 1.4954156541626678,
      "grad_norm": 0.5774357181060241,
      "learning_rate": 3.156611967756711e-05,
      "loss": 11.6995,
      "step": 27462
    },
    {
      "epoch": 1.4954701081592507,
      "grad_norm": 0.5582189399702306,
      "learning_rate": 3.155968998566264e-05,
      "loss": 11.823,
      "step": 27463
    },
    {
      "epoch": 1.4955245621558337,
      "grad_norm": 0.53568887973156,
      "learning_rate": 3.155326082596155e-05,
      "loss": 11.8139,
      "step": 27464
    },
    {
      "epoch": 1.4955790161524167,
      "grad_norm": 0.5031229417344054,
      "learning_rate": 3.1546832198513754e-05,
      "loss": 11.9377,
      "step": 27465
    },
    {
      "epoch": 1.4956334701489997,
      "grad_norm": 0.5423070313971351,
      "learning_rate": 3.154040410336929e-05,
      "loss": 11.8203,
      "step": 27466
    },
    {
      "epoch": 1.4956879241455827,
      "grad_norm": 0.5539134542696412,
      "learning_rate": 3.1533976540578095e-05,
      "loss": 11.8151,
      "step": 27467
    },
    {
      "epoch": 1.4957423781421657,
      "grad_norm": 0.6590430201518052,
      "learning_rate": 3.152754951019021e-05,
      "loss": 12.0469,
      "step": 27468
    },
    {
      "epoch": 1.495796832138749,
      "grad_norm": 0.5380775390880129,
      "learning_rate": 3.152112301225556e-05,
      "loss": 11.8812,
      "step": 27469
    },
    {
      "epoch": 1.495851286135332,
      "grad_norm": 0.5424455925133175,
      "learning_rate": 3.151469704682416e-05,
      "loss": 11.8462,
      "step": 27470
    },
    {
      "epoch": 1.495905740131915,
      "grad_norm": 0.5286895412193657,
      "learning_rate": 3.150827161394597e-05,
      "loss": 11.8277,
      "step": 27471
    },
    {
      "epoch": 1.495960194128498,
      "grad_norm": 0.5286892757526992,
      "learning_rate": 3.150184671367091e-05,
      "loss": 11.8305,
      "step": 27472
    },
    {
      "epoch": 1.496014648125081,
      "grad_norm": 0.5745537850910586,
      "learning_rate": 3.149542234604902e-05,
      "loss": 11.9197,
      "step": 27473
    },
    {
      "epoch": 1.4960691021216639,
      "grad_norm": 0.48156888306081325,
      "learning_rate": 3.148899851113018e-05,
      "loss": 11.7803,
      "step": 27474
    },
    {
      "epoch": 1.4961235561182469,
      "grad_norm": 0.5226758497171498,
      "learning_rate": 3.148257520896436e-05,
      "loss": 11.8624,
      "step": 27475
    },
    {
      "epoch": 1.4961780101148299,
      "grad_norm": 0.6284035486208138,
      "learning_rate": 3.147615243960157e-05,
      "loss": 11.9189,
      "step": 27476
    },
    {
      "epoch": 1.4962324641114129,
      "grad_norm": 0.5525820937708962,
      "learning_rate": 3.146973020309168e-05,
      "loss": 11.8693,
      "step": 27477
    },
    {
      "epoch": 1.4962869181079959,
      "grad_norm": 0.5280852682684858,
      "learning_rate": 3.146330849948468e-05,
      "loss": 11.8642,
      "step": 27478
    },
    {
      "epoch": 1.4963413721045788,
      "grad_norm": 0.5388026049611822,
      "learning_rate": 3.145688732883047e-05,
      "loss": 11.8511,
      "step": 27479
    },
    {
      "epoch": 1.4963958261011618,
      "grad_norm": 0.5225958946227364,
      "learning_rate": 3.1450466691179014e-05,
      "loss": 11.8485,
      "step": 27480
    },
    {
      "epoch": 1.4964502800977448,
      "grad_norm": 0.5239720890394016,
      "learning_rate": 3.144404658658024e-05,
      "loss": 11.9028,
      "step": 27481
    },
    {
      "epoch": 1.4965047340943278,
      "grad_norm": 0.5550563471128637,
      "learning_rate": 3.1437627015084016e-05,
      "loss": 11.8473,
      "step": 27482
    },
    {
      "epoch": 1.4965591880909108,
      "grad_norm": 0.5317330658116989,
      "learning_rate": 3.143120797674034e-05,
      "loss": 11.8971,
      "step": 27483
    },
    {
      "epoch": 1.496613642087494,
      "grad_norm": 0.5109791195007246,
      "learning_rate": 3.142478947159906e-05,
      "loss": 12.0197,
      "step": 27484
    },
    {
      "epoch": 1.496668096084077,
      "grad_norm": 0.5453036897113615,
      "learning_rate": 3.141837149971011e-05,
      "loss": 11.6837,
      "step": 27485
    },
    {
      "epoch": 1.49672255008066,
      "grad_norm": 0.5486934779186424,
      "learning_rate": 3.141195406112344e-05,
      "loss": 11.8049,
      "step": 27486
    },
    {
      "epoch": 1.496777004077243,
      "grad_norm": 0.5505034490465567,
      "learning_rate": 3.1405537155888876e-05,
      "loss": 11.9074,
      "step": 27487
    },
    {
      "epoch": 1.496831458073826,
      "grad_norm": 0.4991785681869255,
      "learning_rate": 3.1399120784056404e-05,
      "loss": 11.8099,
      "step": 27488
    },
    {
      "epoch": 1.496885912070409,
      "grad_norm": 0.483806822887898,
      "learning_rate": 3.1392704945675835e-05,
      "loss": 11.6764,
      "step": 27489
    },
    {
      "epoch": 1.496940366066992,
      "grad_norm": 0.5662450510412216,
      "learning_rate": 3.1386289640797126e-05,
      "loss": 11.9674,
      "step": 27490
    },
    {
      "epoch": 1.496994820063575,
      "grad_norm": 0.4846114197861738,
      "learning_rate": 3.137987486947015e-05,
      "loss": 11.7917,
      "step": 27491
    },
    {
      "epoch": 1.497049274060158,
      "grad_norm": 0.5326804141127223,
      "learning_rate": 3.137346063174472e-05,
      "loss": 11.8509,
      "step": 27492
    },
    {
      "epoch": 1.4971037280567412,
      "grad_norm": 0.54381282457603,
      "learning_rate": 3.1367046927670815e-05,
      "loss": 11.7573,
      "step": 27493
    },
    {
      "epoch": 1.4971581820533242,
      "grad_norm": 0.6531117726889754,
      "learning_rate": 3.136063375729823e-05,
      "loss": 12.0029,
      "step": 27494
    },
    {
      "epoch": 1.4972126360499072,
      "grad_norm": 0.7901619653358373,
      "learning_rate": 3.135422112067691e-05,
      "loss": 11.8016,
      "step": 27495
    },
    {
      "epoch": 1.4972670900464902,
      "grad_norm": 0.547921772757661,
      "learning_rate": 3.134780901785663e-05,
      "loss": 11.8898,
      "step": 27496
    },
    {
      "epoch": 1.4973215440430732,
      "grad_norm": 0.581070980445588,
      "learning_rate": 3.13413974488873e-05,
      "loss": 11.8743,
      "step": 27497
    },
    {
      "epoch": 1.4973759980396562,
      "grad_norm": 0.5626893852553062,
      "learning_rate": 3.1334986413818826e-05,
      "loss": 11.9637,
      "step": 27498
    },
    {
      "epoch": 1.4974304520362391,
      "grad_norm": 0.56260109220869,
      "learning_rate": 3.132857591270096e-05,
      "loss": 11.9424,
      "step": 27499
    },
    {
      "epoch": 1.4974849060328221,
      "grad_norm": 0.5682717579984866,
      "learning_rate": 3.132216594558368e-05,
      "loss": 11.9458,
      "step": 27500
    },
    {
      "epoch": 1.4975393600294051,
      "grad_norm": 0.5087272968067512,
      "learning_rate": 3.1315756512516694e-05,
      "loss": 11.8205,
      "step": 27501
    },
    {
      "epoch": 1.4975938140259881,
      "grad_norm": 0.5251530675231519,
      "learning_rate": 3.130934761354989e-05,
      "loss": 11.8179,
      "step": 27502
    },
    {
      "epoch": 1.4976482680225711,
      "grad_norm": 0.6019276926576036,
      "learning_rate": 3.1302939248733164e-05,
      "loss": 11.9282,
      "step": 27503
    },
    {
      "epoch": 1.497702722019154,
      "grad_norm": 0.5623108223998298,
      "learning_rate": 3.129653141811626e-05,
      "loss": 11.8528,
      "step": 27504
    },
    {
      "epoch": 1.497757176015737,
      "grad_norm": 0.5165853157628106,
      "learning_rate": 3.1290124121749087e-05,
      "loss": 11.8269,
      "step": 27505
    },
    {
      "epoch": 1.49781163001232,
      "grad_norm": 0.5454551760674319,
      "learning_rate": 3.1283717359681394e-05,
      "loss": 11.8371,
      "step": 27506
    },
    {
      "epoch": 1.4978660840089033,
      "grad_norm": 0.513751383959271,
      "learning_rate": 3.127731113196308e-05,
      "loss": 11.8529,
      "step": 27507
    },
    {
      "epoch": 1.4979205380054863,
      "grad_norm": 0.7296877830900464,
      "learning_rate": 3.1270905438643885e-05,
      "loss": 11.7038,
      "step": 27508
    },
    {
      "epoch": 1.4979749920020693,
      "grad_norm": 0.5451329658973117,
      "learning_rate": 3.126450027977366e-05,
      "loss": 11.8812,
      "step": 27509
    },
    {
      "epoch": 1.4980294459986523,
      "grad_norm": 0.5304672373545064,
      "learning_rate": 3.125809565540225e-05,
      "loss": 11.9342,
      "step": 27510
    },
    {
      "epoch": 1.4980838999952353,
      "grad_norm": 0.5229017499272024,
      "learning_rate": 3.1251691565579376e-05,
      "loss": 11.8463,
      "step": 27511
    },
    {
      "epoch": 1.4981383539918183,
      "grad_norm": 0.5460362783778945,
      "learning_rate": 3.124528801035487e-05,
      "loss": 11.8818,
      "step": 27512
    },
    {
      "epoch": 1.4981928079884013,
      "grad_norm": 0.5116544705483668,
      "learning_rate": 3.123888498977856e-05,
      "loss": 11.9706,
      "step": 27513
    },
    {
      "epoch": 1.4982472619849843,
      "grad_norm": 0.5778308106786579,
      "learning_rate": 3.1232482503900185e-05,
      "loss": 11.8801,
      "step": 27514
    },
    {
      "epoch": 1.4983017159815673,
      "grad_norm": 0.6156861772561578,
      "learning_rate": 3.122608055276959e-05,
      "loss": 11.9433,
      "step": 27515
    },
    {
      "epoch": 1.4983561699781505,
      "grad_norm": 0.6010348328230253,
      "learning_rate": 3.1219679136436494e-05,
      "loss": 11.933,
      "step": 27516
    },
    {
      "epoch": 1.4984106239747335,
      "grad_norm": 0.6208290159700537,
      "learning_rate": 3.121327825495074e-05,
      "loss": 11.9482,
      "step": 27517
    },
    {
      "epoch": 1.4984650779713165,
      "grad_norm": 0.5385137364443083,
      "learning_rate": 3.120687790836204e-05,
      "loss": 11.7787,
      "step": 27518
    },
    {
      "epoch": 1.4985195319678994,
      "grad_norm": 0.5725944104336594,
      "learning_rate": 3.1200478096720185e-05,
      "loss": 11.8878,
      "step": 27519
    },
    {
      "epoch": 1.4985739859644824,
      "grad_norm": 0.6063329120900592,
      "learning_rate": 3.1194078820075026e-05,
      "loss": 11.7532,
      "step": 27520
    },
    {
      "epoch": 1.4986284399610654,
      "grad_norm": 0.5704832095542197,
      "learning_rate": 3.118768007847618e-05,
      "loss": 11.8878,
      "step": 27521
    },
    {
      "epoch": 1.4986828939576484,
      "grad_norm": 0.5636863502859587,
      "learning_rate": 3.1181281871973514e-05,
      "loss": 11.9443,
      "step": 27522
    },
    {
      "epoch": 1.4987373479542314,
      "grad_norm": 0.5147934802112546,
      "learning_rate": 3.117488420061669e-05,
      "loss": 11.9173,
      "step": 27523
    },
    {
      "epoch": 1.4987918019508144,
      "grad_norm": 0.51464273422368,
      "learning_rate": 3.1168487064455524e-05,
      "loss": 11.9362,
      "step": 27524
    },
    {
      "epoch": 1.4988462559473974,
      "grad_norm": 0.4749948706665351,
      "learning_rate": 3.1162090463539773e-05,
      "loss": 11.7872,
      "step": 27525
    },
    {
      "epoch": 1.4989007099439804,
      "grad_norm": 0.5811691982443821,
      "learning_rate": 3.115569439791911e-05,
      "loss": 11.7787,
      "step": 27526
    },
    {
      "epoch": 1.4989551639405634,
      "grad_norm": 0.5207677503485039,
      "learning_rate": 3.114929886764335e-05,
      "loss": 11.8194,
      "step": 27527
    },
    {
      "epoch": 1.4990096179371464,
      "grad_norm": 0.5618206163224754,
      "learning_rate": 3.114290387276216e-05,
      "loss": 11.8607,
      "step": 27528
    },
    {
      "epoch": 1.4990640719337294,
      "grad_norm": 0.6222593230654404,
      "learning_rate": 3.113650941332533e-05,
      "loss": 11.8066,
      "step": 27529
    },
    {
      "epoch": 1.4991185259303124,
      "grad_norm": 0.5485997733806728,
      "learning_rate": 3.113011548938255e-05,
      "loss": 11.7379,
      "step": 27530
    },
    {
      "epoch": 1.4991729799268956,
      "grad_norm": 0.5699558382653481,
      "learning_rate": 3.112372210098351e-05,
      "loss": 11.9523,
      "step": 27531
    },
    {
      "epoch": 1.4992274339234786,
      "grad_norm": 0.5336572321578744,
      "learning_rate": 3.1117329248177984e-05,
      "loss": 11.8289,
      "step": 27532
    },
    {
      "epoch": 1.4992818879200616,
      "grad_norm": 0.5966800422283287,
      "learning_rate": 3.111093693101563e-05,
      "loss": 11.9526,
      "step": 27533
    },
    {
      "epoch": 1.4993363419166446,
      "grad_norm": 0.5315714128725967,
      "learning_rate": 3.1104545149546184e-05,
      "loss": 11.8079,
      "step": 27534
    },
    {
      "epoch": 1.4993907959132275,
      "grad_norm": 0.5280188799287588,
      "learning_rate": 3.109815390381938e-05,
      "loss": 11.9168,
      "step": 27535
    },
    {
      "epoch": 1.4994452499098105,
      "grad_norm": 0.4887410793402791,
      "learning_rate": 3.109176319388485e-05,
      "loss": 11.8738,
      "step": 27536
    },
    {
      "epoch": 1.4994997039063935,
      "grad_norm": 0.606420131416671,
      "learning_rate": 3.1085373019792366e-05,
      "loss": 11.8706,
      "step": 27537
    },
    {
      "epoch": 1.4995541579029765,
      "grad_norm": 0.5491822896002987,
      "learning_rate": 3.107898338159153e-05,
      "loss": 11.773,
      "step": 27538
    },
    {
      "epoch": 1.4996086118995597,
      "grad_norm": 0.5137174202577975,
      "learning_rate": 3.107259427933212e-05,
      "loss": 11.7138,
      "step": 27539
    },
    {
      "epoch": 1.4996630658961427,
      "grad_norm": 0.5568255752004765,
      "learning_rate": 3.106620571306378e-05,
      "loss": 11.78,
      "step": 27540
    },
    {
      "epoch": 1.4997175198927257,
      "grad_norm": 0.5316097749800234,
      "learning_rate": 3.105981768283614e-05,
      "loss": 11.8818,
      "step": 27541
    },
    {
      "epoch": 1.4997719738893087,
      "grad_norm": 0.5207641553822407,
      "learning_rate": 3.1053430188698976e-05,
      "loss": 11.7427,
      "step": 27542
    },
    {
      "epoch": 1.4998264278858917,
      "grad_norm": 0.5309694357002798,
      "learning_rate": 3.1047043230701844e-05,
      "loss": 11.8229,
      "step": 27543
    },
    {
      "epoch": 1.4998808818824747,
      "grad_norm": 0.5583661611807208,
      "learning_rate": 3.1040656808894505e-05,
      "loss": 11.8332,
      "step": 27544
    },
    {
      "epoch": 1.4999353358790577,
      "grad_norm": 0.5743922278384787,
      "learning_rate": 3.103427092332656e-05,
      "loss": 11.8431,
      "step": 27545
    },
    {
      "epoch": 1.4999897898756407,
      "grad_norm": 0.5261854774839165,
      "learning_rate": 3.1027885574047687e-05,
      "loss": 11.8617,
      "step": 27546
    },
    {
      "epoch": 1.5000442438722237,
      "grad_norm": 0.5468380951743005,
      "learning_rate": 3.102150076110757e-05,
      "loss": 11.7561,
      "step": 27547
    },
    {
      "epoch": 1.5000986978688067,
      "grad_norm": 0.5654126809901402,
      "learning_rate": 3.101511648455579e-05,
      "loss": 11.8551,
      "step": 27548
    },
    {
      "epoch": 1.5001531518653897,
      "grad_norm": 0.5495863535567939,
      "learning_rate": 3.100873274444208e-05,
      "loss": 11.809,
      "step": 27549
    },
    {
      "epoch": 1.5002076058619727,
      "grad_norm": 0.5647235853031947,
      "learning_rate": 3.1002349540816036e-05,
      "loss": 11.8476,
      "step": 27550
    },
    {
      "epoch": 1.5002620598585557,
      "grad_norm": 0.5435244911535195,
      "learning_rate": 3.0995966873727244e-05,
      "loss": 11.7316,
      "step": 27551
    },
    {
      "epoch": 1.5003165138551386,
      "grad_norm": 0.529410883315443,
      "learning_rate": 3.098958474322543e-05,
      "loss": 11.8691,
      "step": 27552
    },
    {
      "epoch": 1.5003709678517216,
      "grad_norm": 0.570198110631626,
      "learning_rate": 3.098320314936015e-05,
      "loss": 11.7387,
      "step": 27553
    },
    {
      "epoch": 1.5004254218483046,
      "grad_norm": 0.5399444238497815,
      "learning_rate": 3.0976822092181076e-05,
      "loss": 11.7691,
      "step": 27554
    },
    {
      "epoch": 1.5004798758448878,
      "grad_norm": 0.6302267951639604,
      "learning_rate": 3.097044157173778e-05,
      "loss": 11.8268,
      "step": 27555
    },
    {
      "epoch": 1.5005343298414708,
      "grad_norm": 0.5475409955087848,
      "learning_rate": 3.096406158807995e-05,
      "loss": 11.7685,
      "step": 27556
    },
    {
      "epoch": 1.5005887838380538,
      "grad_norm": 0.5299594529500552,
      "learning_rate": 3.0957682141257104e-05,
      "loss": 11.8292,
      "step": 27557
    },
    {
      "epoch": 1.5006432378346368,
      "grad_norm": 0.5538666764621011,
      "learning_rate": 3.0951303231318916e-05,
      "loss": 11.6462,
      "step": 27558
    },
    {
      "epoch": 1.5006976918312198,
      "grad_norm": 0.5317166062254615,
      "learning_rate": 3.0944924858314994e-05,
      "loss": 11.7942,
      "step": 27559
    },
    {
      "epoch": 1.5007521458278028,
      "grad_norm": 0.581170481397359,
      "learning_rate": 3.093854702229493e-05,
      "loss": 11.7875,
      "step": 27560
    },
    {
      "epoch": 1.500806599824386,
      "grad_norm": 0.5594560718388232,
      "learning_rate": 3.093216972330827e-05,
      "loss": 11.7466,
      "step": 27561
    },
    {
      "epoch": 1.500861053820969,
      "grad_norm": 0.5414119794473163,
      "learning_rate": 3.092579296140467e-05,
      "loss": 11.8654,
      "step": 27562
    },
    {
      "epoch": 1.500915507817552,
      "grad_norm": 0.5203958651338018,
      "learning_rate": 3.0919416736633646e-05,
      "loss": 11.9641,
      "step": 27563
    },
    {
      "epoch": 1.500969961814135,
      "grad_norm": 0.5535061061668877,
      "learning_rate": 3.091304104904487e-05,
      "loss": 11.8742,
      "step": 27564
    },
    {
      "epoch": 1.501024415810718,
      "grad_norm": 0.5362178767221011,
      "learning_rate": 3.0906665898687826e-05,
      "loss": 11.8637,
      "step": 27565
    },
    {
      "epoch": 1.501078869807301,
      "grad_norm": 0.5127350859931332,
      "learning_rate": 3.090029128561218e-05,
      "loss": 11.8934,
      "step": 27566
    },
    {
      "epoch": 1.501133323803884,
      "grad_norm": 0.5298054538718383,
      "learning_rate": 3.089391720986742e-05,
      "loss": 11.8764,
      "step": 27567
    },
    {
      "epoch": 1.501187777800467,
      "grad_norm": 0.6092771664732751,
      "learning_rate": 3.088754367150315e-05,
      "loss": 11.8243,
      "step": 27568
    },
    {
      "epoch": 1.50124223179705,
      "grad_norm": 0.5284515644674422,
      "learning_rate": 3.088117067056896e-05,
      "loss": 11.8535,
      "step": 27569
    },
    {
      "epoch": 1.501296685793633,
      "grad_norm": 0.571331530496941,
      "learning_rate": 3.0874798207114374e-05,
      "loss": 11.9048,
      "step": 27570
    },
    {
      "epoch": 1.501351139790216,
      "grad_norm": 0.6116606438283967,
      "learning_rate": 3.0868426281188953e-05,
      "loss": 11.9523,
      "step": 27571
    },
    {
      "epoch": 1.501405593786799,
      "grad_norm": 0.5388547942213677,
      "learning_rate": 3.0862054892842215e-05,
      "loss": 11.8108,
      "step": 27572
    },
    {
      "epoch": 1.501460047783382,
      "grad_norm": 0.5460617994491654,
      "learning_rate": 3.0855684042123734e-05,
      "loss": 11.8827,
      "step": 27573
    },
    {
      "epoch": 1.501514501779965,
      "grad_norm": 0.5466345431925912,
      "learning_rate": 3.084931372908307e-05,
      "loss": 11.662,
      "step": 27574
    },
    {
      "epoch": 1.501568955776548,
      "grad_norm": 0.5919496457101667,
      "learning_rate": 3.0842943953769724e-05,
      "loss": 11.9597,
      "step": 27575
    },
    {
      "epoch": 1.501623409773131,
      "grad_norm": 0.5290296969291574,
      "learning_rate": 3.083657471623326e-05,
      "loss": 11.8332,
      "step": 27576
    },
    {
      "epoch": 1.501677863769714,
      "grad_norm": 0.5476704734827672,
      "learning_rate": 3.0830206016523165e-05,
      "loss": 11.913,
      "step": 27577
    },
    {
      "epoch": 1.501732317766297,
      "grad_norm": 0.5052560576496058,
      "learning_rate": 3.0823837854689016e-05,
      "loss": 11.9057,
      "step": 27578
    },
    {
      "epoch": 1.5017867717628801,
      "grad_norm": 0.5616683807127564,
      "learning_rate": 3.081747023078028e-05,
      "loss": 11.8198,
      "step": 27579
    },
    {
      "epoch": 1.501841225759463,
      "grad_norm": 0.5784261645398846,
      "learning_rate": 3.081110314484652e-05,
      "loss": 11.697,
      "step": 27580
    },
    {
      "epoch": 1.501895679756046,
      "grad_norm": 0.4839773202934603,
      "learning_rate": 3.0804736596937225e-05,
      "loss": 11.8623,
      "step": 27581
    },
    {
      "epoch": 1.501950133752629,
      "grad_norm": 0.5926065659373663,
      "learning_rate": 3.079837058710188e-05,
      "loss": 11.8601,
      "step": 27582
    },
    {
      "epoch": 1.502004587749212,
      "grad_norm": 0.5667121528056517,
      "learning_rate": 3.079200511539e-05,
      "loss": 11.8723,
      "step": 27583
    },
    {
      "epoch": 1.502059041745795,
      "grad_norm": 0.5492222967762245,
      "learning_rate": 3.0785640181851125e-05,
      "loss": 11.8033,
      "step": 27584
    },
    {
      "epoch": 1.5021134957423783,
      "grad_norm": 0.5894426805620854,
      "learning_rate": 3.07792757865347e-05,
      "loss": 11.8386,
      "step": 27585
    },
    {
      "epoch": 1.5021679497389613,
      "grad_norm": 0.5891161954158775,
      "learning_rate": 3.0772911929490265e-05,
      "loss": 11.8987,
      "step": 27586
    },
    {
      "epoch": 1.5022224037355443,
      "grad_norm": 0.5071649840710323,
      "learning_rate": 3.076654861076723e-05,
      "loss": 11.8914,
      "step": 27587
    },
    {
      "epoch": 1.5022768577321273,
      "grad_norm": 0.5372359479170464,
      "learning_rate": 3.076018583041517e-05,
      "loss": 11.8784,
      "step": 27588
    },
    {
      "epoch": 1.5023313117287103,
      "grad_norm": 0.53300440436641,
      "learning_rate": 3.075382358848348e-05,
      "loss": 11.8932,
      "step": 27589
    },
    {
      "epoch": 1.5023857657252933,
      "grad_norm": 0.5549183348199073,
      "learning_rate": 3.07474618850217e-05,
      "loss": 11.8055,
      "step": 27590
    },
    {
      "epoch": 1.5024402197218762,
      "grad_norm": 0.5364623809143403,
      "learning_rate": 3.074110072007927e-05,
      "loss": 11.8346,
      "step": 27591
    },
    {
      "epoch": 1.5024946737184592,
      "grad_norm": 0.5980968037882382,
      "learning_rate": 3.073474009370563e-05,
      "loss": 11.8335,
      "step": 27592
    },
    {
      "epoch": 1.5025491277150422,
      "grad_norm": 0.5319294149873626,
      "learning_rate": 3.07283800059503e-05,
      "loss": 11.7779,
      "step": 27593
    },
    {
      "epoch": 1.5026035817116252,
      "grad_norm": 0.5135358376415784,
      "learning_rate": 3.072202045686265e-05,
      "loss": 11.7976,
      "step": 27594
    },
    {
      "epoch": 1.5026580357082082,
      "grad_norm": 0.5577114057239492,
      "learning_rate": 3.0715661446492217e-05,
      "loss": 11.8907,
      "step": 27595
    },
    {
      "epoch": 1.5027124897047912,
      "grad_norm": 0.5472486261440314,
      "learning_rate": 3.070930297488843e-05,
      "loss": 11.9129,
      "step": 27596
    },
    {
      "epoch": 1.5027669437013742,
      "grad_norm": 0.5574978807079234,
      "learning_rate": 3.0702945042100706e-05,
      "loss": 11.8204,
      "step": 27597
    },
    {
      "epoch": 1.5028213976979572,
      "grad_norm": 0.5722485819548241,
      "learning_rate": 3.0696587648178523e-05,
      "loss": 11.8558,
      "step": 27598
    },
    {
      "epoch": 1.5028758516945402,
      "grad_norm": 0.5395526032028541,
      "learning_rate": 3.069023079317127e-05,
      "loss": 11.8176,
      "step": 27599
    },
    {
      "epoch": 1.5029303056911232,
      "grad_norm": 0.5277757089922277,
      "learning_rate": 3.0683874477128436e-05,
      "loss": 11.8558,
      "step": 27600
    },
    {
      "epoch": 1.5029847596877062,
      "grad_norm": 0.5240624033785303,
      "learning_rate": 3.067751870009942e-05,
      "loss": 11.9188,
      "step": 27601
    },
    {
      "epoch": 1.5030392136842894,
      "grad_norm": 0.5084139043867187,
      "learning_rate": 3.067116346213361e-05,
      "loss": 11.8472,
      "step": 27602
    },
    {
      "epoch": 1.5030936676808724,
      "grad_norm": 0.5409512735160412,
      "learning_rate": 3.06648087632805e-05,
      "loss": 11.8957,
      "step": 27603
    },
    {
      "epoch": 1.5031481216774554,
      "grad_norm": 0.5711421532119695,
      "learning_rate": 3.0658454603589416e-05,
      "loss": 11.9514,
      "step": 27604
    },
    {
      "epoch": 1.5032025756740384,
      "grad_norm": 0.5798785861361018,
      "learning_rate": 3.065210098310985e-05,
      "loss": 11.9032,
      "step": 27605
    },
    {
      "epoch": 1.5032570296706214,
      "grad_norm": 0.5517773569951979,
      "learning_rate": 3.0645747901891164e-05,
      "loss": 11.7489,
      "step": 27606
    },
    {
      "epoch": 1.5033114836672044,
      "grad_norm": 0.60294794538028,
      "learning_rate": 3.063939535998276e-05,
      "loss": 11.8538,
      "step": 27607
    },
    {
      "epoch": 1.5033659376637876,
      "grad_norm": 0.5379968417713321,
      "learning_rate": 3.0633043357434074e-05,
      "loss": 11.7519,
      "step": 27608
    },
    {
      "epoch": 1.5034203916603706,
      "grad_norm": 0.5558431679438995,
      "learning_rate": 3.0626691894294456e-05,
      "loss": 11.8692,
      "step": 27609
    },
    {
      "epoch": 1.5034748456569536,
      "grad_norm": 0.56468562645044,
      "learning_rate": 3.0620340970613345e-05,
      "loss": 11.7864,
      "step": 27610
    },
    {
      "epoch": 1.5035292996535365,
      "grad_norm": 0.5267122393320834,
      "learning_rate": 3.06139905864401e-05,
      "loss": 11.7804,
      "step": 27611
    },
    {
      "epoch": 1.5035837536501195,
      "grad_norm": 0.5499475642516489,
      "learning_rate": 3.060764074182406e-05,
      "loss": 11.775,
      "step": 27612
    },
    {
      "epoch": 1.5036382076467025,
      "grad_norm": 0.5143490316338917,
      "learning_rate": 3.0601291436814684e-05,
      "loss": 11.8559,
      "step": 27613
    },
    {
      "epoch": 1.5036926616432855,
      "grad_norm": 0.5706754059374372,
      "learning_rate": 3.059494267146127e-05,
      "loss": 11.8854,
      "step": 27614
    },
    {
      "epoch": 1.5037471156398685,
      "grad_norm": 0.523767343686082,
      "learning_rate": 3.058859444581326e-05,
      "loss": 11.885,
      "step": 27615
    },
    {
      "epoch": 1.5038015696364515,
      "grad_norm": 0.5624296865878132,
      "learning_rate": 3.058224675991993e-05,
      "loss": 11.6968,
      "step": 27616
    },
    {
      "epoch": 1.5038560236330345,
      "grad_norm": 0.49422260520353717,
      "learning_rate": 3.0575899613830706e-05,
      "loss": 11.7223,
      "step": 27617
    },
    {
      "epoch": 1.5039104776296175,
      "grad_norm": 0.5247207589597501,
      "learning_rate": 3.0569553007594956e-05,
      "loss": 11.8117,
      "step": 27618
    },
    {
      "epoch": 1.5039649316262005,
      "grad_norm": 0.5646454568774119,
      "learning_rate": 3.056320694126197e-05,
      "loss": 11.8483,
      "step": 27619
    },
    {
      "epoch": 1.5040193856227835,
      "grad_norm": 0.5403338629392758,
      "learning_rate": 3.0556861414881154e-05,
      "loss": 11.8416,
      "step": 27620
    },
    {
      "epoch": 1.5040738396193665,
      "grad_norm": 0.5505254544530112,
      "learning_rate": 3.0550516428501854e-05,
      "loss": 11.8749,
      "step": 27621
    },
    {
      "epoch": 1.5041282936159495,
      "grad_norm": 0.4985326885128025,
      "learning_rate": 3.054417198217333e-05,
      "loss": 11.7613,
      "step": 27622
    },
    {
      "epoch": 1.5041827476125325,
      "grad_norm": 0.5408151241099561,
      "learning_rate": 3.0537828075945016e-05,
      "loss": 11.9199,
      "step": 27623
    },
    {
      "epoch": 1.5042372016091154,
      "grad_norm": 0.5483937991568044,
      "learning_rate": 3.053148470986617e-05,
      "loss": 11.9476,
      "step": 27624
    },
    {
      "epoch": 1.5042916556056987,
      "grad_norm": 0.6080094206391734,
      "learning_rate": 3.052514188398617e-05,
      "loss": 11.7556,
      "step": 27625
    },
    {
      "epoch": 1.5043461096022817,
      "grad_norm": 0.5883997187373496,
      "learning_rate": 3.05187995983543e-05,
      "loss": 11.8255,
      "step": 27626
    },
    {
      "epoch": 1.5044005635988646,
      "grad_norm": 0.563012064872864,
      "learning_rate": 3.0512457853019917e-05,
      "loss": 11.9002,
      "step": 27627
    },
    {
      "epoch": 1.5044550175954476,
      "grad_norm": 0.566417441781685,
      "learning_rate": 3.050611664803228e-05,
      "loss": 11.7473,
      "step": 27628
    },
    {
      "epoch": 1.5045094715920306,
      "grad_norm": 0.5262720486523642,
      "learning_rate": 3.049977598344075e-05,
      "loss": 11.9355,
      "step": 27629
    },
    {
      "epoch": 1.5045639255886136,
      "grad_norm": 0.5155669848516421,
      "learning_rate": 3.0493435859294628e-05,
      "loss": 11.7297,
      "step": 27630
    },
    {
      "epoch": 1.5046183795851968,
      "grad_norm": 0.5817023527518846,
      "learning_rate": 3.0487096275643224e-05,
      "loss": 11.8661,
      "step": 27631
    },
    {
      "epoch": 1.5046728335817798,
      "grad_norm": 0.5217576368753092,
      "learning_rate": 3.0480757232535772e-05,
      "loss": 11.8647,
      "step": 27632
    },
    {
      "epoch": 1.5047272875783628,
      "grad_norm": 0.5230937332016243,
      "learning_rate": 3.0474418730021648e-05,
      "loss": 11.8024,
      "step": 27633
    },
    {
      "epoch": 1.5047817415749458,
      "grad_norm": 0.5179976205184285,
      "learning_rate": 3.0468080768150076e-05,
      "loss": 11.7769,
      "step": 27634
    },
    {
      "epoch": 1.5048361955715288,
      "grad_norm": 0.5224160824425776,
      "learning_rate": 3.0461743346970395e-05,
      "loss": 11.9731,
      "step": 27635
    },
    {
      "epoch": 1.5048906495681118,
      "grad_norm": 0.5364814966305287,
      "learning_rate": 3.045540646653182e-05,
      "loss": 11.7945,
      "step": 27636
    },
    {
      "epoch": 1.5049451035646948,
      "grad_norm": 0.5789394424236494,
      "learning_rate": 3.0449070126883707e-05,
      "loss": 11.8448,
      "step": 27637
    },
    {
      "epoch": 1.5049995575612778,
      "grad_norm": 0.5730519306389764,
      "learning_rate": 3.0442734328075263e-05,
      "loss": 11.8303,
      "step": 27638
    },
    {
      "epoch": 1.5050540115578608,
      "grad_norm": 0.5712768170623889,
      "learning_rate": 3.0436399070155774e-05,
      "loss": 11.9416,
      "step": 27639
    },
    {
      "epoch": 1.5051084655544438,
      "grad_norm": 0.5352127555076341,
      "learning_rate": 3.0430064353174538e-05,
      "loss": 11.8445,
      "step": 27640
    },
    {
      "epoch": 1.5051629195510268,
      "grad_norm": 0.5426642678899108,
      "learning_rate": 3.0423730177180797e-05,
      "loss": 11.8103,
      "step": 27641
    },
    {
      "epoch": 1.5052173735476098,
      "grad_norm": 0.5461720622822323,
      "learning_rate": 3.0417396542223798e-05,
      "loss": 11.9051,
      "step": 27642
    },
    {
      "epoch": 1.5052718275441928,
      "grad_norm": 0.5544016741604803,
      "learning_rate": 3.041106344835275e-05,
      "loss": 11.8508,
      "step": 27643
    },
    {
      "epoch": 1.5053262815407757,
      "grad_norm": 0.5802545894786407,
      "learning_rate": 3.040473089561695e-05,
      "loss": 11.7987,
      "step": 27644
    },
    {
      "epoch": 1.5053807355373587,
      "grad_norm": 0.5435056876544804,
      "learning_rate": 3.039839888406567e-05,
      "loss": 11.7582,
      "step": 27645
    },
    {
      "epoch": 1.5054351895339417,
      "grad_norm": 0.5498441096741119,
      "learning_rate": 3.0392067413748083e-05,
      "loss": 11.8857,
      "step": 27646
    },
    {
      "epoch": 1.5054896435305247,
      "grad_norm": 0.5868732495443124,
      "learning_rate": 3.0385736484713477e-05,
      "loss": 11.7373,
      "step": 27647
    },
    {
      "epoch": 1.5055440975271077,
      "grad_norm": 0.5631848433435084,
      "learning_rate": 3.037940609701102e-05,
      "loss": 11.7864,
      "step": 27648
    },
    {
      "epoch": 1.505598551523691,
      "grad_norm": 0.5860801684071715,
      "learning_rate": 3.0373076250690026e-05,
      "loss": 11.7146,
      "step": 27649
    },
    {
      "epoch": 1.505653005520274,
      "grad_norm": 0.5585283707571668,
      "learning_rate": 3.036674694579962e-05,
      "loss": 11.8924,
      "step": 27650
    },
    {
      "epoch": 1.505707459516857,
      "grad_norm": 0.5920091119800553,
      "learning_rate": 3.0360418182389105e-05,
      "loss": 11.9214,
      "step": 27651
    },
    {
      "epoch": 1.50576191351344,
      "grad_norm": 0.5741591184526227,
      "learning_rate": 3.035408996050766e-05,
      "loss": 11.8055,
      "step": 27652
    },
    {
      "epoch": 1.505816367510023,
      "grad_norm": 0.4973276984223243,
      "learning_rate": 3.0347762280204462e-05,
      "loss": 11.7137,
      "step": 27653
    },
    {
      "epoch": 1.5058708215066061,
      "grad_norm": 0.5516625562789796,
      "learning_rate": 3.0341435141528763e-05,
      "loss": 11.7877,
      "step": 27654
    },
    {
      "epoch": 1.505925275503189,
      "grad_norm": 0.5708887913135768,
      "learning_rate": 3.033510854452972e-05,
      "loss": 11.9488,
      "step": 27655
    },
    {
      "epoch": 1.505979729499772,
      "grad_norm": 0.5741516223263851,
      "learning_rate": 3.032878248925657e-05,
      "loss": 11.9446,
      "step": 27656
    },
    {
      "epoch": 1.506034183496355,
      "grad_norm": 0.5531204656209718,
      "learning_rate": 3.0322456975758505e-05,
      "loss": 11.8402,
      "step": 27657
    },
    {
      "epoch": 1.506088637492938,
      "grad_norm": 0.6218445852612702,
      "learning_rate": 3.0316132004084674e-05,
      "loss": 11.9344,
      "step": 27658
    },
    {
      "epoch": 1.506143091489521,
      "grad_norm": 0.5769309911162305,
      "learning_rate": 3.030980757428432e-05,
      "loss": 11.9579,
      "step": 27659
    },
    {
      "epoch": 1.506197545486104,
      "grad_norm": 0.5402758990034201,
      "learning_rate": 3.0303483686406555e-05,
      "loss": 11.889,
      "step": 27660
    },
    {
      "epoch": 1.506251999482687,
      "grad_norm": 0.6158061534664355,
      "learning_rate": 3.029716034050062e-05,
      "loss": 11.792,
      "step": 27661
    },
    {
      "epoch": 1.50630645347927,
      "grad_norm": 0.5213652094806597,
      "learning_rate": 3.0290837536615656e-05,
      "loss": 11.7095,
      "step": 27662
    },
    {
      "epoch": 1.506360907475853,
      "grad_norm": 0.5524929572467947,
      "learning_rate": 3.0284515274800807e-05,
      "loss": 11.8645,
      "step": 27663
    },
    {
      "epoch": 1.506415361472436,
      "grad_norm": 0.5287964101011609,
      "learning_rate": 3.0278193555105283e-05,
      "loss": 11.9561,
      "step": 27664
    },
    {
      "epoch": 1.506469815469019,
      "grad_norm": 0.5456827261862065,
      "learning_rate": 3.027187237757818e-05,
      "loss": 11.7935,
      "step": 27665
    },
    {
      "epoch": 1.506524269465602,
      "grad_norm": 0.5560675880179424,
      "learning_rate": 3.0265551742268693e-05,
      "loss": 11.8534,
      "step": 27666
    },
    {
      "epoch": 1.506578723462185,
      "grad_norm": 0.4839734380980397,
      "learning_rate": 3.0259231649226015e-05,
      "loss": 11.7421,
      "step": 27667
    },
    {
      "epoch": 1.506633177458768,
      "grad_norm": 0.5314535741315,
      "learning_rate": 3.0252912098499207e-05,
      "loss": 11.7904,
      "step": 27668
    },
    {
      "epoch": 1.506687631455351,
      "grad_norm": 0.5362603181572028,
      "learning_rate": 3.0246593090137476e-05,
      "loss": 11.803,
      "step": 27669
    },
    {
      "epoch": 1.506742085451934,
      "grad_norm": 0.5808088826584501,
      "learning_rate": 3.0240274624189913e-05,
      "loss": 11.9047,
      "step": 27670
    },
    {
      "epoch": 1.506796539448517,
      "grad_norm": 0.578753416107352,
      "learning_rate": 3.02339567007057e-05,
      "loss": 11.8853,
      "step": 27671
    },
    {
      "epoch": 1.5068509934451002,
      "grad_norm": 0.508534977626483,
      "learning_rate": 3.0227639319733936e-05,
      "loss": 11.8025,
      "step": 27672
    },
    {
      "epoch": 1.5069054474416832,
      "grad_norm": 0.5839536048259578,
      "learning_rate": 3.0221322481323723e-05,
      "loss": 11.8539,
      "step": 27673
    },
    {
      "epoch": 1.5069599014382662,
      "grad_norm": 0.5056608957084763,
      "learning_rate": 3.021500618552424e-05,
      "loss": 11.8683,
      "step": 27674
    },
    {
      "epoch": 1.5070143554348492,
      "grad_norm": 0.48905232113237707,
      "learning_rate": 3.0208690432384546e-05,
      "loss": 11.7323,
      "step": 27675
    },
    {
      "epoch": 1.5070688094314322,
      "grad_norm": 0.5823855342076538,
      "learning_rate": 3.0202375221953805e-05,
      "loss": 11.8104,
      "step": 27676
    },
    {
      "epoch": 1.5071232634280152,
      "grad_norm": 0.545191595494048,
      "learning_rate": 3.019606055428106e-05,
      "loss": 11.9079,
      "step": 27677
    },
    {
      "epoch": 1.5071777174245984,
      "grad_norm": 0.5396027113593479,
      "learning_rate": 3.0189746429415468e-05,
      "loss": 11.7344,
      "step": 27678
    },
    {
      "epoch": 1.5072321714211814,
      "grad_norm": 0.5411755554388232,
      "learning_rate": 3.0183432847406134e-05,
      "loss": 12.0117,
      "step": 27679
    },
    {
      "epoch": 1.5072866254177644,
      "grad_norm": 0.5251088975317232,
      "learning_rate": 3.0177119808302113e-05,
      "loss": 11.8839,
      "step": 27680
    },
    {
      "epoch": 1.5073410794143474,
      "grad_norm": 0.5613355898052612,
      "learning_rate": 3.0170807312152537e-05,
      "loss": 11.7858,
      "step": 27681
    },
    {
      "epoch": 1.5073955334109304,
      "grad_norm": 0.5848409840479958,
      "learning_rate": 3.0164495359006484e-05,
      "loss": 11.9158,
      "step": 27682
    },
    {
      "epoch": 1.5074499874075133,
      "grad_norm": 0.5313528696351432,
      "learning_rate": 3.0158183948912988e-05,
      "loss": 11.8243,
      "step": 27683
    },
    {
      "epoch": 1.5075044414040963,
      "grad_norm": 0.5753534768744759,
      "learning_rate": 3.0151873081921213e-05,
      "loss": 11.905,
      "step": 27684
    },
    {
      "epoch": 1.5075588954006793,
      "grad_norm": 0.5783521566718891,
      "learning_rate": 3.0145562758080137e-05,
      "loss": 11.8615,
      "step": 27685
    },
    {
      "epoch": 1.5076133493972623,
      "grad_norm": 0.5092573533801059,
      "learning_rate": 3.0139252977438914e-05,
      "loss": 11.8508,
      "step": 27686
    },
    {
      "epoch": 1.5076678033938453,
      "grad_norm": 0.5199298621276031,
      "learning_rate": 3.013294374004655e-05,
      "loss": 11.6786,
      "step": 27687
    },
    {
      "epoch": 1.5077222573904283,
      "grad_norm": 0.562591696967236,
      "learning_rate": 3.0126635045952133e-05,
      "loss": 11.7552,
      "step": 27688
    },
    {
      "epoch": 1.5077767113870113,
      "grad_norm": 0.567123205368442,
      "learning_rate": 3.0120326895204753e-05,
      "loss": 11.9026,
      "step": 27689
    },
    {
      "epoch": 1.5078311653835943,
      "grad_norm": 0.5685875255988978,
      "learning_rate": 3.011401928785339e-05,
      "loss": 11.7773,
      "step": 27690
    },
    {
      "epoch": 1.5078856193801773,
      "grad_norm": 0.5327316258202117,
      "learning_rate": 3.0107712223947203e-05,
      "loss": 11.7722,
      "step": 27691
    },
    {
      "epoch": 1.5079400733767603,
      "grad_norm": 0.5250633768345059,
      "learning_rate": 3.0101405703535103e-05,
      "loss": 11.8296,
      "step": 27692
    },
    {
      "epoch": 1.5079945273733433,
      "grad_norm": 0.5514445851102612,
      "learning_rate": 3.0095099726666187e-05,
      "loss": 11.8859,
      "step": 27693
    },
    {
      "epoch": 1.5080489813699263,
      "grad_norm": 0.6038059950739845,
      "learning_rate": 3.0088794293389532e-05,
      "loss": 11.8462,
      "step": 27694
    },
    {
      "epoch": 1.5081034353665095,
      "grad_norm": 0.5387197878737624,
      "learning_rate": 3.008248940375411e-05,
      "loss": 11.9464,
      "step": 27695
    },
    {
      "epoch": 1.5081578893630925,
      "grad_norm": 0.5045327878282647,
      "learning_rate": 3.0076185057809003e-05,
      "loss": 11.8414,
      "step": 27696
    },
    {
      "epoch": 1.5082123433596755,
      "grad_norm": 0.543435502192343,
      "learning_rate": 3.0069881255603182e-05,
      "loss": 11.7743,
      "step": 27697
    },
    {
      "epoch": 1.5082667973562585,
      "grad_norm": 0.49555653997269605,
      "learning_rate": 3.006357799718572e-05,
      "loss": 11.8638,
      "step": 27698
    },
    {
      "epoch": 1.5083212513528415,
      "grad_norm": 0.6831213653704303,
      "learning_rate": 3.005727528260557e-05,
      "loss": 11.8205,
      "step": 27699
    },
    {
      "epoch": 1.5083757053494244,
      "grad_norm": 0.5093461458203183,
      "learning_rate": 3.0050973111911772e-05,
      "loss": 11.8062,
      "step": 27700
    },
    {
      "epoch": 1.5084301593460077,
      "grad_norm": 0.5545139771784615,
      "learning_rate": 3.004467148515341e-05,
      "loss": 11.8289,
      "step": 27701
    },
    {
      "epoch": 1.5084846133425907,
      "grad_norm": 0.6564560130369709,
      "learning_rate": 3.0038370402379344e-05,
      "loss": 11.7958,
      "step": 27702
    },
    {
      "epoch": 1.5085390673391736,
      "grad_norm": 0.5399395695077479,
      "learning_rate": 3.0032069863638678e-05,
      "loss": 11.8155,
      "step": 27703
    },
    {
      "epoch": 1.5085935213357566,
      "grad_norm": 0.5239996833966908,
      "learning_rate": 3.0025769868980335e-05,
      "loss": 11.8158,
      "step": 27704
    },
    {
      "epoch": 1.5086479753323396,
      "grad_norm": 0.527652555018816,
      "learning_rate": 3.0019470418453345e-05,
      "loss": 11.877,
      "step": 27705
    },
    {
      "epoch": 1.5087024293289226,
      "grad_norm": 0.5482037668581788,
      "learning_rate": 3.0013171512106718e-05,
      "loss": 11.9077,
      "step": 27706
    },
    {
      "epoch": 1.5087568833255056,
      "grad_norm": 0.5593091392044128,
      "learning_rate": 3.0006873149989377e-05,
      "loss": 11.8471,
      "step": 27707
    },
    {
      "epoch": 1.5088113373220886,
      "grad_norm": 0.5489678883026025,
      "learning_rate": 3.000057533215036e-05,
      "loss": 11.9846,
      "step": 27708
    },
    {
      "epoch": 1.5088657913186716,
      "grad_norm": 0.5048261198947238,
      "learning_rate": 2.999427805863858e-05,
      "loss": 11.767,
      "step": 27709
    },
    {
      "epoch": 1.5089202453152546,
      "grad_norm": 0.5385155346180013,
      "learning_rate": 2.9987981329503056e-05,
      "loss": 11.802,
      "step": 27710
    },
    {
      "epoch": 1.5089746993118376,
      "grad_norm": 0.5682527811224917,
      "learning_rate": 2.9981685144792737e-05,
      "loss": 11.7803,
      "step": 27711
    },
    {
      "epoch": 1.5090291533084206,
      "grad_norm": 0.5867884195522941,
      "learning_rate": 2.9975389504556538e-05,
      "loss": 11.7611,
      "step": 27712
    },
    {
      "epoch": 1.5090836073050036,
      "grad_norm": 0.5388582827725946,
      "learning_rate": 2.996909440884349e-05,
      "loss": 11.8301,
      "step": 27713
    },
    {
      "epoch": 1.5091380613015866,
      "grad_norm": 0.5467698982091564,
      "learning_rate": 2.9962799857702474e-05,
      "loss": 11.7841,
      "step": 27714
    },
    {
      "epoch": 1.5091925152981696,
      "grad_norm": 0.6212780696901338,
      "learning_rate": 2.995650585118247e-05,
      "loss": 11.985,
      "step": 27715
    },
    {
      "epoch": 1.5092469692947525,
      "grad_norm": 0.5645021872395943,
      "learning_rate": 2.9950212389332466e-05,
      "loss": 11.8428,
      "step": 27716
    },
    {
      "epoch": 1.5093014232913355,
      "grad_norm": 0.5347293351712962,
      "learning_rate": 2.994391947220131e-05,
      "loss": 11.7633,
      "step": 27717
    },
    {
      "epoch": 1.5093558772879185,
      "grad_norm": 0.6065627903828362,
      "learning_rate": 2.993762709983803e-05,
      "loss": 11.8202,
      "step": 27718
    },
    {
      "epoch": 1.5094103312845017,
      "grad_norm": 0.5478554910871952,
      "learning_rate": 2.9931335272291472e-05,
      "loss": 11.8343,
      "step": 27719
    },
    {
      "epoch": 1.5094647852810847,
      "grad_norm": 0.5887533294741611,
      "learning_rate": 2.9925043989610635e-05,
      "loss": 11.9709,
      "step": 27720
    },
    {
      "epoch": 1.5095192392776677,
      "grad_norm": 0.5367691135942377,
      "learning_rate": 2.99187532518444e-05,
      "loss": 11.7524,
      "step": 27721
    },
    {
      "epoch": 1.5095736932742507,
      "grad_norm": 0.5920294223461078,
      "learning_rate": 2.9912463059041673e-05,
      "loss": 11.8158,
      "step": 27722
    },
    {
      "epoch": 1.5096281472708337,
      "grad_norm": 0.5799886961127216,
      "learning_rate": 2.9906173411251414e-05,
      "loss": 11.9282,
      "step": 27723
    },
    {
      "epoch": 1.509682601267417,
      "grad_norm": 0.5420188804440307,
      "learning_rate": 2.9899884308522475e-05,
      "loss": 11.7917,
      "step": 27724
    },
    {
      "epoch": 1.509737055264,
      "grad_norm": 0.6077614752035385,
      "learning_rate": 2.9893595750903813e-05,
      "loss": 11.8512,
      "step": 27725
    },
    {
      "epoch": 1.509791509260583,
      "grad_norm": 0.500352912740398,
      "learning_rate": 2.9887307738444293e-05,
      "loss": 11.8185,
      "step": 27726
    },
    {
      "epoch": 1.509845963257166,
      "grad_norm": 0.6414151378311607,
      "learning_rate": 2.9881020271192806e-05,
      "loss": 11.8646,
      "step": 27727
    },
    {
      "epoch": 1.509900417253749,
      "grad_norm": 0.562359974934004,
      "learning_rate": 2.9874733349198315e-05,
      "loss": 11.8485,
      "step": 27728
    },
    {
      "epoch": 1.509954871250332,
      "grad_norm": 0.5324743453276087,
      "learning_rate": 2.9868446972509612e-05,
      "loss": 11.9092,
      "step": 27729
    },
    {
      "epoch": 1.510009325246915,
      "grad_norm": 0.6062978713611572,
      "learning_rate": 2.986216114117566e-05,
      "loss": 11.7869,
      "step": 27730
    },
    {
      "epoch": 1.5100637792434979,
      "grad_norm": 0.5741654731406654,
      "learning_rate": 2.985587585524532e-05,
      "loss": 11.872,
      "step": 27731
    },
    {
      "epoch": 1.5101182332400809,
      "grad_norm": 0.5104704082013496,
      "learning_rate": 2.9849591114767406e-05,
      "loss": 11.7474,
      "step": 27732
    },
    {
      "epoch": 1.5101726872366639,
      "grad_norm": 0.674178818188482,
      "learning_rate": 2.984330691979087e-05,
      "loss": 11.9635,
      "step": 27733
    },
    {
      "epoch": 1.5102271412332469,
      "grad_norm": 0.5652969158957791,
      "learning_rate": 2.9837023270364506e-05,
      "loss": 11.9251,
      "step": 27734
    },
    {
      "epoch": 1.5102815952298299,
      "grad_norm": 0.5609797645699056,
      "learning_rate": 2.9830740166537264e-05,
      "loss": 11.8837,
      "step": 27735
    },
    {
      "epoch": 1.5103360492264128,
      "grad_norm": 0.5159221518667632,
      "learning_rate": 2.9824457608357902e-05,
      "loss": 11.8842,
      "step": 27736
    },
    {
      "epoch": 1.5103905032229958,
      "grad_norm": 0.5129693179919498,
      "learning_rate": 2.9818175595875342e-05,
      "loss": 11.6384,
      "step": 27737
    },
    {
      "epoch": 1.5104449572195788,
      "grad_norm": 0.606667141464947,
      "learning_rate": 2.9811894129138452e-05,
      "loss": 11.9409,
      "step": 27738
    },
    {
      "epoch": 1.5104994112161618,
      "grad_norm": 0.5693205394135664,
      "learning_rate": 2.9805613208196003e-05,
      "loss": 11.8966,
      "step": 27739
    },
    {
      "epoch": 1.5105538652127448,
      "grad_norm": 0.5250548845680687,
      "learning_rate": 2.9799332833096906e-05,
      "loss": 11.711,
      "step": 27740
    },
    {
      "epoch": 1.5106083192093278,
      "grad_norm": 0.5524343060557678,
      "learning_rate": 2.979305300388997e-05,
      "loss": 11.8614,
      "step": 27741
    },
    {
      "epoch": 1.510662773205911,
      "grad_norm": 0.5485382355577155,
      "learning_rate": 2.9786773720624007e-05,
      "loss": 11.8623,
      "step": 27742
    },
    {
      "epoch": 1.510717227202494,
      "grad_norm": 0.5195175777305084,
      "learning_rate": 2.9780494983347885e-05,
      "loss": 11.8537,
      "step": 27743
    },
    {
      "epoch": 1.510771681199077,
      "grad_norm": 0.5646632657230068,
      "learning_rate": 2.9774216792110386e-05,
      "loss": 11.9142,
      "step": 27744
    },
    {
      "epoch": 1.51082613519566,
      "grad_norm": 0.5326355239330797,
      "learning_rate": 2.976793914696039e-05,
      "loss": 11.7431,
      "step": 27745
    },
    {
      "epoch": 1.510880589192243,
      "grad_norm": 0.5640738171778821,
      "learning_rate": 2.9761662047946638e-05,
      "loss": 11.8633,
      "step": 27746
    },
    {
      "epoch": 1.510935043188826,
      "grad_norm": 0.6548038237412614,
      "learning_rate": 2.9755385495118014e-05,
      "loss": 11.8872,
      "step": 27747
    },
    {
      "epoch": 1.5109894971854092,
      "grad_norm": 0.6111004721881695,
      "learning_rate": 2.9749109488523265e-05,
      "loss": 11.8724,
      "step": 27748
    },
    {
      "epoch": 1.5110439511819922,
      "grad_norm": 0.5840157481701392,
      "learning_rate": 2.9742834028211207e-05,
      "loss": 11.888,
      "step": 27749
    },
    {
      "epoch": 1.5110984051785752,
      "grad_norm": 0.512270983141964,
      "learning_rate": 2.97365591142307e-05,
      "loss": 11.8493,
      "step": 27750
    },
    {
      "epoch": 1.5111528591751582,
      "grad_norm": 0.5047354628998809,
      "learning_rate": 2.9730284746630454e-05,
      "loss": 11.7814,
      "step": 27751
    },
    {
      "epoch": 1.5112073131717412,
      "grad_norm": 0.5192742936666448,
      "learning_rate": 2.9724010925459368e-05,
      "loss": 11.9035,
      "step": 27752
    },
    {
      "epoch": 1.5112617671683242,
      "grad_norm": 0.5700787671591359,
      "learning_rate": 2.9717737650766085e-05,
      "loss": 11.7511,
      "step": 27753
    },
    {
      "epoch": 1.5113162211649072,
      "grad_norm": 0.4802551680400022,
      "learning_rate": 2.9711464922599474e-05,
      "loss": 11.6913,
      "step": 27754
    },
    {
      "epoch": 1.5113706751614902,
      "grad_norm": 0.5204942213256752,
      "learning_rate": 2.9705192741008325e-05,
      "loss": 11.7288,
      "step": 27755
    },
    {
      "epoch": 1.5114251291580731,
      "grad_norm": 0.5405846651405644,
      "learning_rate": 2.9698921106041354e-05,
      "loss": 11.797,
      "step": 27756
    },
    {
      "epoch": 1.5114795831546561,
      "grad_norm": 0.5781230033029797,
      "learning_rate": 2.969265001774739e-05,
      "loss": 11.7932,
      "step": 27757
    },
    {
      "epoch": 1.5115340371512391,
      "grad_norm": 0.5185954624122529,
      "learning_rate": 2.968637947617514e-05,
      "loss": 11.756,
      "step": 27758
    },
    {
      "epoch": 1.5115884911478221,
      "grad_norm": 0.5698397862436131,
      "learning_rate": 2.968010948137343e-05,
      "loss": 11.8523,
      "step": 27759
    },
    {
      "epoch": 1.5116429451444051,
      "grad_norm": 0.5756019090718169,
      "learning_rate": 2.9673840033390943e-05,
      "loss": 11.8631,
      "step": 27760
    },
    {
      "epoch": 1.511697399140988,
      "grad_norm": 0.5585769282172337,
      "learning_rate": 2.9667571132276474e-05,
      "loss": 11.8178,
      "step": 27761
    },
    {
      "epoch": 1.511751853137571,
      "grad_norm": 0.5598159438504424,
      "learning_rate": 2.9661302778078826e-05,
      "loss": 11.8843,
      "step": 27762
    },
    {
      "epoch": 1.511806307134154,
      "grad_norm": 0.5718899983595637,
      "learning_rate": 2.965503497084663e-05,
      "loss": 11.9957,
      "step": 27763
    },
    {
      "epoch": 1.511860761130737,
      "grad_norm": 0.5435589226540075,
      "learning_rate": 2.9648767710628665e-05,
      "loss": 11.9282,
      "step": 27764
    },
    {
      "epoch": 1.5119152151273203,
      "grad_norm": 0.5583589531844237,
      "learning_rate": 2.964250099747372e-05,
      "loss": 11.8914,
      "step": 27765
    },
    {
      "epoch": 1.5119696691239033,
      "grad_norm": 0.6002200385594159,
      "learning_rate": 2.9636234831430454e-05,
      "loss": 11.7539,
      "step": 27766
    },
    {
      "epoch": 1.5120241231204863,
      "grad_norm": 0.5466406512985008,
      "learning_rate": 2.962996921254766e-05,
      "loss": 11.797,
      "step": 27767
    },
    {
      "epoch": 1.5120785771170693,
      "grad_norm": 0.49651751791199267,
      "learning_rate": 2.9623704140873986e-05,
      "loss": 11.8796,
      "step": 27768
    },
    {
      "epoch": 1.5121330311136523,
      "grad_norm": 0.5640010234271187,
      "learning_rate": 2.961743961645823e-05,
      "loss": 11.9479,
      "step": 27769
    },
    {
      "epoch": 1.5121874851102353,
      "grad_norm": 0.4978289056008298,
      "learning_rate": 2.961117563934903e-05,
      "loss": 11.758,
      "step": 27770
    },
    {
      "epoch": 1.5122419391068185,
      "grad_norm": 0.5258343399477016,
      "learning_rate": 2.9604912209595136e-05,
      "loss": 11.7486,
      "step": 27771
    },
    {
      "epoch": 1.5122963931034015,
      "grad_norm": 0.5353552731019385,
      "learning_rate": 2.959864932724532e-05,
      "loss": 11.8828,
      "step": 27772
    },
    {
      "epoch": 1.5123508470999845,
      "grad_norm": 0.5380838597696318,
      "learning_rate": 2.959238699234814e-05,
      "loss": 11.9841,
      "step": 27773
    },
    {
      "epoch": 1.5124053010965675,
      "grad_norm": 0.5472535488504849,
      "learning_rate": 2.9586125204952398e-05,
      "loss": 11.7373,
      "step": 27774
    },
    {
      "epoch": 1.5124597550931504,
      "grad_norm": 0.49168558506863086,
      "learning_rate": 2.9579863965106724e-05,
      "loss": 11.677,
      "step": 27775
    },
    {
      "epoch": 1.5125142090897334,
      "grad_norm": 0.6033380433241394,
      "learning_rate": 2.957360327285984e-05,
      "loss": 11.9156,
      "step": 27776
    },
    {
      "epoch": 1.5125686630863164,
      "grad_norm": 0.5972369825776153,
      "learning_rate": 2.956734312826046e-05,
      "loss": 11.9468,
      "step": 27777
    },
    {
      "epoch": 1.5126231170828994,
      "grad_norm": 0.5765631999622778,
      "learning_rate": 2.9561083531357203e-05,
      "loss": 11.8256,
      "step": 27778
    },
    {
      "epoch": 1.5126775710794824,
      "grad_norm": 0.5718339356550207,
      "learning_rate": 2.9554824482198794e-05,
      "loss": 11.9699,
      "step": 27779
    },
    {
      "epoch": 1.5127320250760654,
      "grad_norm": 0.5322174718543619,
      "learning_rate": 2.9548565980833854e-05,
      "loss": 11.8512,
      "step": 27780
    },
    {
      "epoch": 1.5127864790726484,
      "grad_norm": 0.5482797675858028,
      "learning_rate": 2.9542308027311116e-05,
      "loss": 11.9528,
      "step": 27781
    },
    {
      "epoch": 1.5128409330692314,
      "grad_norm": 0.5400244281915337,
      "learning_rate": 2.953605062167921e-05,
      "loss": 11.7887,
      "step": 27782
    },
    {
      "epoch": 1.5128953870658144,
      "grad_norm": 0.6011909227453015,
      "learning_rate": 2.9529793763986758e-05,
      "loss": 11.8717,
      "step": 27783
    },
    {
      "epoch": 1.5129498410623974,
      "grad_norm": 0.5506693073369622,
      "learning_rate": 2.952353745428247e-05,
      "loss": 11.7844,
      "step": 27784
    },
    {
      "epoch": 1.5130042950589804,
      "grad_norm": 0.5740683367698717,
      "learning_rate": 2.9517281692614952e-05,
      "loss": 11.9149,
      "step": 27785
    },
    {
      "epoch": 1.5130587490555634,
      "grad_norm": 0.5624485248051482,
      "learning_rate": 2.9511026479032867e-05,
      "loss": 11.9163,
      "step": 27786
    },
    {
      "epoch": 1.5131132030521464,
      "grad_norm": 0.54204018303972,
      "learning_rate": 2.9504771813584887e-05,
      "loss": 11.9727,
      "step": 27787
    },
    {
      "epoch": 1.5131676570487296,
      "grad_norm": 0.5594505465196601,
      "learning_rate": 2.9498517696319605e-05,
      "loss": 11.9219,
      "step": 27788
    },
    {
      "epoch": 1.5132221110453126,
      "grad_norm": 0.6689923677952158,
      "learning_rate": 2.9492264127285695e-05,
      "loss": 12.0173,
      "step": 27789
    },
    {
      "epoch": 1.5132765650418956,
      "grad_norm": 0.5000079040973773,
      "learning_rate": 2.948601110653173e-05,
      "loss": 11.8615,
      "step": 27790
    },
    {
      "epoch": 1.5133310190384786,
      "grad_norm": 0.5409981726138181,
      "learning_rate": 2.94797586341064e-05,
      "loss": 11.7796,
      "step": 27791
    },
    {
      "epoch": 1.5133854730350615,
      "grad_norm": 0.5024907674996787,
      "learning_rate": 2.947350671005831e-05,
      "loss": 11.8399,
      "step": 27792
    },
    {
      "epoch": 1.5134399270316445,
      "grad_norm": 0.5657093388356965,
      "learning_rate": 2.9467255334436006e-05,
      "loss": 11.8278,
      "step": 27793
    },
    {
      "epoch": 1.5134943810282278,
      "grad_norm": 0.512819115467977,
      "learning_rate": 2.9461004507288194e-05,
      "loss": 11.8364,
      "step": 27794
    },
    {
      "epoch": 1.5135488350248107,
      "grad_norm": 0.5853750031511613,
      "learning_rate": 2.9454754228663407e-05,
      "loss": 11.7854,
      "step": 27795
    },
    {
      "epoch": 1.5136032890213937,
      "grad_norm": 0.5658294543457406,
      "learning_rate": 2.9448504498610307e-05,
      "loss": 11.9533,
      "step": 27796
    },
    {
      "epoch": 1.5136577430179767,
      "grad_norm": 0.516569400584985,
      "learning_rate": 2.9442255317177448e-05,
      "loss": 11.7878,
      "step": 27797
    },
    {
      "epoch": 1.5137121970145597,
      "grad_norm": 0.5623119523680566,
      "learning_rate": 2.9436006684413444e-05,
      "loss": 11.8565,
      "step": 27798
    },
    {
      "epoch": 1.5137666510111427,
      "grad_norm": 0.5305491088570766,
      "learning_rate": 2.9429758600366906e-05,
      "loss": 11.7577,
      "step": 27799
    },
    {
      "epoch": 1.5138211050077257,
      "grad_norm": 0.4661915877437278,
      "learning_rate": 2.9423511065086375e-05,
      "loss": 11.6817,
      "step": 27800
    },
    {
      "epoch": 1.5138755590043087,
      "grad_norm": 0.6026615705906947,
      "learning_rate": 2.941726407862049e-05,
      "loss": 11.9328,
      "step": 27801
    },
    {
      "epoch": 1.5139300130008917,
      "grad_norm": 0.5188263326926184,
      "learning_rate": 2.941101764101779e-05,
      "loss": 11.8143,
      "step": 27802
    },
    {
      "epoch": 1.5139844669974747,
      "grad_norm": 0.5143942788249989,
      "learning_rate": 2.940477175232683e-05,
      "loss": 11.8826,
      "step": 27803
    },
    {
      "epoch": 1.5140389209940577,
      "grad_norm": 0.5289844667408242,
      "learning_rate": 2.9398526412596228e-05,
      "loss": 11.6707,
      "step": 27804
    },
    {
      "epoch": 1.5140933749906407,
      "grad_norm": 0.5190835605898211,
      "learning_rate": 2.9392281621874495e-05,
      "loss": 11.8505,
      "step": 27805
    },
    {
      "epoch": 1.5141478289872237,
      "grad_norm": 0.6307976520850446,
      "learning_rate": 2.938603738021026e-05,
      "loss": 11.8924,
      "step": 27806
    },
    {
      "epoch": 1.5142022829838067,
      "grad_norm": 0.5463165588990535,
      "learning_rate": 2.9379793687652003e-05,
      "loss": 11.8728,
      "step": 27807
    },
    {
      "epoch": 1.5142567369803897,
      "grad_norm": 0.5091881255534446,
      "learning_rate": 2.937355054424835e-05,
      "loss": 11.7606,
      "step": 27808
    },
    {
      "epoch": 1.5143111909769726,
      "grad_norm": 0.5107749718863238,
      "learning_rate": 2.9367307950047775e-05,
      "loss": 11.6613,
      "step": 27809
    },
    {
      "epoch": 1.5143656449735556,
      "grad_norm": 0.605009771388253,
      "learning_rate": 2.9361065905098862e-05,
      "loss": 11.8747,
      "step": 27810
    },
    {
      "epoch": 1.5144200989701386,
      "grad_norm": 0.6343331401690552,
      "learning_rate": 2.9354824409450165e-05,
      "loss": 11.9454,
      "step": 27811
    },
    {
      "epoch": 1.5144745529667218,
      "grad_norm": 0.5961261935633988,
      "learning_rate": 2.9348583463150215e-05,
      "loss": 11.9897,
      "step": 27812
    },
    {
      "epoch": 1.5145290069633048,
      "grad_norm": 0.5260408723621902,
      "learning_rate": 2.934234306624749e-05,
      "loss": 11.8594,
      "step": 27813
    },
    {
      "epoch": 1.5145834609598878,
      "grad_norm": 0.5676865393284748,
      "learning_rate": 2.933610321879059e-05,
      "loss": 11.6873,
      "step": 27814
    },
    {
      "epoch": 1.5146379149564708,
      "grad_norm": 0.5789886000789263,
      "learning_rate": 2.9329863920827973e-05,
      "loss": 12.0022,
      "step": 27815
    },
    {
      "epoch": 1.5146923689530538,
      "grad_norm": 0.5660964656367419,
      "learning_rate": 2.932362517240822e-05,
      "loss": 11.9032,
      "step": 27816
    },
    {
      "epoch": 1.5147468229496368,
      "grad_norm": 0.5911587393227951,
      "learning_rate": 2.9317386973579764e-05,
      "loss": 11.9297,
      "step": 27817
    },
    {
      "epoch": 1.51480127694622,
      "grad_norm": 0.5413495498279537,
      "learning_rate": 2.931114932439121e-05,
      "loss": 11.8286,
      "step": 27818
    },
    {
      "epoch": 1.514855730942803,
      "grad_norm": 0.5625383941350024,
      "learning_rate": 2.930491222489097e-05,
      "loss": 11.887,
      "step": 27819
    },
    {
      "epoch": 1.514910184939386,
      "grad_norm": 0.5705753135611374,
      "learning_rate": 2.9298675675127586e-05,
      "loss": 11.9516,
      "step": 27820
    },
    {
      "epoch": 1.514964638935969,
      "grad_norm": 0.5495970961301904,
      "learning_rate": 2.9292439675149587e-05,
      "loss": 11.8987,
      "step": 27821
    },
    {
      "epoch": 1.515019092932552,
      "grad_norm": 0.5799716159335289,
      "learning_rate": 2.928620422500544e-05,
      "loss": 11.807,
      "step": 27822
    },
    {
      "epoch": 1.515073546929135,
      "grad_norm": 0.5299167729543424,
      "learning_rate": 2.927996932474363e-05,
      "loss": 11.8671,
      "step": 27823
    },
    {
      "epoch": 1.515128000925718,
      "grad_norm": 0.549514754283632,
      "learning_rate": 2.9273734974412605e-05,
      "loss": 11.9484,
      "step": 27824
    },
    {
      "epoch": 1.515182454922301,
      "grad_norm": 0.6565246978050049,
      "learning_rate": 2.9267501174060873e-05,
      "loss": 11.8285,
      "step": 27825
    },
    {
      "epoch": 1.515236908918884,
      "grad_norm": 0.5094424060972945,
      "learning_rate": 2.9261267923736958e-05,
      "loss": 11.8731,
      "step": 27826
    },
    {
      "epoch": 1.515291362915467,
      "grad_norm": 0.5401169161084873,
      "learning_rate": 2.9255035223489236e-05,
      "loss": 11.8489,
      "step": 27827
    },
    {
      "epoch": 1.51534581691205,
      "grad_norm": 0.534441901673353,
      "learning_rate": 2.9248803073366272e-05,
      "loss": 11.8566,
      "step": 27828
    },
    {
      "epoch": 1.515400270908633,
      "grad_norm": 0.5261033805036803,
      "learning_rate": 2.9242571473416436e-05,
      "loss": 11.8294,
      "step": 27829
    },
    {
      "epoch": 1.515454724905216,
      "grad_norm": 0.6081218612704858,
      "learning_rate": 2.9236340423688268e-05,
      "loss": 11.9544,
      "step": 27830
    },
    {
      "epoch": 1.515509178901799,
      "grad_norm": 0.5769409886346919,
      "learning_rate": 2.9230109924230152e-05,
      "loss": 11.925,
      "step": 27831
    },
    {
      "epoch": 1.515563632898382,
      "grad_norm": 0.5710254215367669,
      "learning_rate": 2.9223879975090606e-05,
      "loss": 11.9083,
      "step": 27832
    },
    {
      "epoch": 1.515618086894965,
      "grad_norm": 0.5514124110800704,
      "learning_rate": 2.9217650576318036e-05,
      "loss": 11.8565,
      "step": 27833
    },
    {
      "epoch": 1.515672540891548,
      "grad_norm": 0.5206379089906886,
      "learning_rate": 2.9211421727960854e-05,
      "loss": 11.7541,
      "step": 27834
    },
    {
      "epoch": 1.5157269948881311,
      "grad_norm": 0.537492180588417,
      "learning_rate": 2.9205193430067525e-05,
      "loss": 11.8885,
      "step": 27835
    },
    {
      "epoch": 1.515781448884714,
      "grad_norm": 0.5459418827556682,
      "learning_rate": 2.919896568268652e-05,
      "loss": 11.741,
      "step": 27836
    },
    {
      "epoch": 1.515835902881297,
      "grad_norm": 0.5690263196747039,
      "learning_rate": 2.919273848586619e-05,
      "loss": 11.7847,
      "step": 27837
    },
    {
      "epoch": 1.51589035687788,
      "grad_norm": 0.518363252654648,
      "learning_rate": 2.918651183965504e-05,
      "loss": 11.7857,
      "step": 27838
    },
    {
      "epoch": 1.515944810874463,
      "grad_norm": 0.5698791668369638,
      "learning_rate": 2.918028574410141e-05,
      "loss": 11.724,
      "step": 27839
    },
    {
      "epoch": 1.515999264871046,
      "grad_norm": 0.6821101984848678,
      "learning_rate": 2.9174060199253794e-05,
      "loss": 11.8832,
      "step": 27840
    },
    {
      "epoch": 1.5160537188676293,
      "grad_norm": 0.5393663450828089,
      "learning_rate": 2.9167835205160532e-05,
      "loss": 11.7887,
      "step": 27841
    },
    {
      "epoch": 1.5161081728642123,
      "grad_norm": 0.6874066226660146,
      "learning_rate": 2.9161610761870082e-05,
      "loss": 11.7755,
      "step": 27842
    },
    {
      "epoch": 1.5161626268607953,
      "grad_norm": 0.5503605609436028,
      "learning_rate": 2.9155386869430847e-05,
      "loss": 11.8203,
      "step": 27843
    },
    {
      "epoch": 1.5162170808573783,
      "grad_norm": 0.5233963511568601,
      "learning_rate": 2.9149163527891154e-05,
      "loss": 11.8154,
      "step": 27844
    },
    {
      "epoch": 1.5162715348539613,
      "grad_norm": 0.635581410885624,
      "learning_rate": 2.9142940737299485e-05,
      "loss": 11.8079,
      "step": 27845
    },
    {
      "epoch": 1.5163259888505443,
      "grad_norm": 0.6003910004603072,
      "learning_rate": 2.9136718497704164e-05,
      "loss": 11.8677,
      "step": 27846
    },
    {
      "epoch": 1.5163804428471273,
      "grad_norm": 0.55421737256066,
      "learning_rate": 2.9130496809153597e-05,
      "loss": 11.8329,
      "step": 27847
    },
    {
      "epoch": 1.5164348968437102,
      "grad_norm": 0.5580708026299369,
      "learning_rate": 2.9124275671696212e-05,
      "loss": 11.9821,
      "step": 27848
    },
    {
      "epoch": 1.5164893508402932,
      "grad_norm": 0.521589600014856,
      "learning_rate": 2.9118055085380303e-05,
      "loss": 11.8927,
      "step": 27849
    },
    {
      "epoch": 1.5165438048368762,
      "grad_norm": 0.5953309181436236,
      "learning_rate": 2.9111835050254323e-05,
      "loss": 11.9729,
      "step": 27850
    },
    {
      "epoch": 1.5165982588334592,
      "grad_norm": 0.5634892078543181,
      "learning_rate": 2.9105615566366563e-05,
      "loss": 11.8954,
      "step": 27851
    },
    {
      "epoch": 1.5166527128300422,
      "grad_norm": 0.5375423234571668,
      "learning_rate": 2.9099396633765464e-05,
      "loss": 11.7917,
      "step": 27852
    },
    {
      "epoch": 1.5167071668266252,
      "grad_norm": 0.7058152558283332,
      "learning_rate": 2.9093178252499344e-05,
      "loss": 12.0143,
      "step": 27853
    },
    {
      "epoch": 1.5167616208232082,
      "grad_norm": 0.627399877302454,
      "learning_rate": 2.9086960422616527e-05,
      "loss": 11.8652,
      "step": 27854
    },
    {
      "epoch": 1.5168160748197912,
      "grad_norm": 0.5432207752183109,
      "learning_rate": 2.9080743144165433e-05,
      "loss": 11.9182,
      "step": 27855
    },
    {
      "epoch": 1.5168705288163742,
      "grad_norm": 0.5795550600974179,
      "learning_rate": 2.9074526417194347e-05,
      "loss": 11.918,
      "step": 27856
    },
    {
      "epoch": 1.5169249828129572,
      "grad_norm": 0.6469021761797537,
      "learning_rate": 2.906831024175166e-05,
      "loss": 11.8331,
      "step": 27857
    },
    {
      "epoch": 1.5169794368095404,
      "grad_norm": 0.5351030400617772,
      "learning_rate": 2.906209461788566e-05,
      "loss": 11.8457,
      "step": 27858
    },
    {
      "epoch": 1.5170338908061234,
      "grad_norm": 0.5242390532435152,
      "learning_rate": 2.9055879545644716e-05,
      "loss": 11.8352,
      "step": 27859
    },
    {
      "epoch": 1.5170883448027064,
      "grad_norm": 0.536360864304305,
      "learning_rate": 2.9049665025077178e-05,
      "loss": 11.7729,
      "step": 27860
    },
    {
      "epoch": 1.5171427987992894,
      "grad_norm": 0.5182475054363096,
      "learning_rate": 2.9043451056231295e-05,
      "loss": 11.8093,
      "step": 27861
    },
    {
      "epoch": 1.5171972527958724,
      "grad_norm": 0.5674892112315953,
      "learning_rate": 2.9037237639155492e-05,
      "loss": 11.9089,
      "step": 27862
    },
    {
      "epoch": 1.5172517067924554,
      "grad_norm": 0.48246487052542764,
      "learning_rate": 2.9031024773898018e-05,
      "loss": 11.7977,
      "step": 27863
    },
    {
      "epoch": 1.5173061607890386,
      "grad_norm": 0.6185070494429271,
      "learning_rate": 2.9024812460507154e-05,
      "loss": 11.7855,
      "step": 27864
    },
    {
      "epoch": 1.5173606147856216,
      "grad_norm": 0.5718141434432262,
      "learning_rate": 2.9018600699031294e-05,
      "loss": 11.9102,
      "step": 27865
    },
    {
      "epoch": 1.5174150687822046,
      "grad_norm": 0.5511896418875497,
      "learning_rate": 2.9012389489518677e-05,
      "loss": 11.823,
      "step": 27866
    },
    {
      "epoch": 1.5174695227787875,
      "grad_norm": 0.5754939719354365,
      "learning_rate": 2.9006178832017638e-05,
      "loss": 11.9012,
      "step": 27867
    },
    {
      "epoch": 1.5175239767753705,
      "grad_norm": 0.5322506960199332,
      "learning_rate": 2.8999968726576442e-05,
      "loss": 11.8449,
      "step": 27868
    },
    {
      "epoch": 1.5175784307719535,
      "grad_norm": 0.5396397627818631,
      "learning_rate": 2.8993759173243386e-05,
      "loss": 11.8189,
      "step": 27869
    },
    {
      "epoch": 1.5176328847685365,
      "grad_norm": 0.5185878582681657,
      "learning_rate": 2.89875501720668e-05,
      "loss": 11.895,
      "step": 27870
    },
    {
      "epoch": 1.5176873387651195,
      "grad_norm": 0.5792286258753058,
      "learning_rate": 2.8981341723094902e-05,
      "loss": 11.8972,
      "step": 27871
    },
    {
      "epoch": 1.5177417927617025,
      "grad_norm": 0.5322116546301725,
      "learning_rate": 2.8975133826376057e-05,
      "loss": 11.9257,
      "step": 27872
    },
    {
      "epoch": 1.5177962467582855,
      "grad_norm": 0.5851757597525883,
      "learning_rate": 2.8968926481958424e-05,
      "loss": 11.8221,
      "step": 27873
    },
    {
      "epoch": 1.5178507007548685,
      "grad_norm": 0.5829732504888988,
      "learning_rate": 2.896271968989034e-05,
      "loss": 11.9159,
      "step": 27874
    },
    {
      "epoch": 1.5179051547514515,
      "grad_norm": 0.5144869012805328,
      "learning_rate": 2.895651345022008e-05,
      "loss": 11.6957,
      "step": 27875
    },
    {
      "epoch": 1.5179596087480345,
      "grad_norm": 0.5280357587995841,
      "learning_rate": 2.8950307762995853e-05,
      "loss": 11.8242,
      "step": 27876
    },
    {
      "epoch": 1.5180140627446175,
      "grad_norm": 0.5394983149057343,
      "learning_rate": 2.894410262826599e-05,
      "loss": 11.8521,
      "step": 27877
    },
    {
      "epoch": 1.5180685167412005,
      "grad_norm": 0.5396967352632152,
      "learning_rate": 2.893789804607866e-05,
      "loss": 11.8221,
      "step": 27878
    },
    {
      "epoch": 1.5181229707377835,
      "grad_norm": 0.5640942151315489,
      "learning_rate": 2.893169401648218e-05,
      "loss": 11.9148,
      "step": 27879
    },
    {
      "epoch": 1.5181774247343665,
      "grad_norm": 0.5478598688414913,
      "learning_rate": 2.8925490539524746e-05,
      "loss": 11.8535,
      "step": 27880
    },
    {
      "epoch": 1.5182318787309494,
      "grad_norm": 0.5769870862339115,
      "learning_rate": 2.891928761525461e-05,
      "loss": 12.0046,
      "step": 27881
    },
    {
      "epoch": 1.5182863327275327,
      "grad_norm": 0.5464094118829763,
      "learning_rate": 2.8913085243720085e-05,
      "loss": 11.9371,
      "step": 27882
    },
    {
      "epoch": 1.5183407867241157,
      "grad_norm": 0.5639966166131771,
      "learning_rate": 2.8906883424969257e-05,
      "loss": 11.9645,
      "step": 27883
    },
    {
      "epoch": 1.5183952407206986,
      "grad_norm": 0.561523889345088,
      "learning_rate": 2.890068215905043e-05,
      "loss": 11.8737,
      "step": 27884
    },
    {
      "epoch": 1.5184496947172816,
      "grad_norm": 0.5584318488250063,
      "learning_rate": 2.8894481446011847e-05,
      "loss": 11.7879,
      "step": 27885
    },
    {
      "epoch": 1.5185041487138646,
      "grad_norm": 0.5195656150142299,
      "learning_rate": 2.8888281285901674e-05,
      "loss": 11.8961,
      "step": 27886
    },
    {
      "epoch": 1.5185586027104476,
      "grad_norm": 0.5258584797947724,
      "learning_rate": 2.8882081678768193e-05,
      "loss": 11.8562,
      "step": 27887
    },
    {
      "epoch": 1.5186130567070308,
      "grad_norm": 0.5705927095621556,
      "learning_rate": 2.8875882624659524e-05,
      "loss": 11.9243,
      "step": 27888
    },
    {
      "epoch": 1.5186675107036138,
      "grad_norm": 0.5482577633922507,
      "learning_rate": 2.8869684123623963e-05,
      "loss": 11.8742,
      "step": 27889
    },
    {
      "epoch": 1.5187219647001968,
      "grad_norm": 0.5771281166583392,
      "learning_rate": 2.886348617570963e-05,
      "loss": 11.8703,
      "step": 27890
    },
    {
      "epoch": 1.5187764186967798,
      "grad_norm": 0.5262282254611751,
      "learning_rate": 2.8857288780964753e-05,
      "loss": 11.8149,
      "step": 27891
    },
    {
      "epoch": 1.5188308726933628,
      "grad_norm": 0.5398936561700859,
      "learning_rate": 2.8851091939437602e-05,
      "loss": 11.7759,
      "step": 27892
    },
    {
      "epoch": 1.5188853266899458,
      "grad_norm": 0.5436349613015419,
      "learning_rate": 2.8844895651176218e-05,
      "loss": 11.9125,
      "step": 27893
    },
    {
      "epoch": 1.5189397806865288,
      "grad_norm": 0.5211540971772278,
      "learning_rate": 2.8838699916228894e-05,
      "loss": 11.8642,
      "step": 27894
    },
    {
      "epoch": 1.5189942346831118,
      "grad_norm": 0.5274338892558537,
      "learning_rate": 2.883250473464374e-05,
      "loss": 11.8229,
      "step": 27895
    },
    {
      "epoch": 1.5190486886796948,
      "grad_norm": 0.5211417927003539,
      "learning_rate": 2.8826310106468968e-05,
      "loss": 11.7099,
      "step": 27896
    },
    {
      "epoch": 1.5191031426762778,
      "grad_norm": 0.5494026998814043,
      "learning_rate": 2.8820116031752774e-05,
      "loss": 11.807,
      "step": 27897
    },
    {
      "epoch": 1.5191575966728608,
      "grad_norm": 0.6055377194136393,
      "learning_rate": 2.8813922510543267e-05,
      "loss": 11.8004,
      "step": 27898
    },
    {
      "epoch": 1.5192120506694438,
      "grad_norm": 0.5225621783353847,
      "learning_rate": 2.8807729542888662e-05,
      "loss": 11.7967,
      "step": 27899
    },
    {
      "epoch": 1.5192665046660268,
      "grad_norm": 0.5567172433033473,
      "learning_rate": 2.8801537128837065e-05,
      "loss": 11.7802,
      "step": 27900
    },
    {
      "epoch": 1.5193209586626097,
      "grad_norm": 0.5196774715295803,
      "learning_rate": 2.879534526843668e-05,
      "loss": 11.8116,
      "step": 27901
    },
    {
      "epoch": 1.5193754126591927,
      "grad_norm": 0.5388871073186456,
      "learning_rate": 2.8789153961735605e-05,
      "loss": 11.8597,
      "step": 27902
    },
    {
      "epoch": 1.5194298666557757,
      "grad_norm": 0.5610179797228153,
      "learning_rate": 2.8782963208782042e-05,
      "loss": 11.847,
      "step": 27903
    },
    {
      "epoch": 1.5194843206523587,
      "grad_norm": 0.5846937180348672,
      "learning_rate": 2.87767730096241e-05,
      "loss": 11.8652,
      "step": 27904
    },
    {
      "epoch": 1.519538774648942,
      "grad_norm": 0.5537353097572987,
      "learning_rate": 2.877058336430989e-05,
      "loss": 11.8942,
      "step": 27905
    },
    {
      "epoch": 1.519593228645525,
      "grad_norm": 0.5957336635846588,
      "learning_rate": 2.8764394272887584e-05,
      "loss": 11.8114,
      "step": 27906
    },
    {
      "epoch": 1.519647682642108,
      "grad_norm": 0.48896804956338424,
      "learning_rate": 2.8758205735405276e-05,
      "loss": 11.8408,
      "step": 27907
    },
    {
      "epoch": 1.519702136638691,
      "grad_norm": 0.5693470935341843,
      "learning_rate": 2.875201775191111e-05,
      "loss": 11.6823,
      "step": 27908
    },
    {
      "epoch": 1.519756590635274,
      "grad_norm": 0.5409164269378668,
      "learning_rate": 2.8745830322453226e-05,
      "loss": 11.8604,
      "step": 27909
    },
    {
      "epoch": 1.519811044631857,
      "grad_norm": 0.5218839293379983,
      "learning_rate": 2.873964344707968e-05,
      "loss": 11.8008,
      "step": 27910
    },
    {
      "epoch": 1.5198654986284401,
      "grad_norm": 0.5377627599129235,
      "learning_rate": 2.8733457125838658e-05,
      "loss": 11.7381,
      "step": 27911
    },
    {
      "epoch": 1.519919952625023,
      "grad_norm": 0.5765601938153482,
      "learning_rate": 2.8727271358778185e-05,
      "loss": 11.8799,
      "step": 27912
    },
    {
      "epoch": 1.519974406621606,
      "grad_norm": 0.5392376194512571,
      "learning_rate": 2.872108614594644e-05,
      "loss": 11.766,
      "step": 27913
    },
    {
      "epoch": 1.520028860618189,
      "grad_norm": 0.5889251853275724,
      "learning_rate": 2.8714901487391477e-05,
      "loss": 11.8027,
      "step": 27914
    },
    {
      "epoch": 1.520083314614772,
      "grad_norm": 0.5743403154080624,
      "learning_rate": 2.8708717383161366e-05,
      "loss": 11.8232,
      "step": 27915
    },
    {
      "epoch": 1.520137768611355,
      "grad_norm": 0.6280411365971715,
      "learning_rate": 2.8702533833304256e-05,
      "loss": 11.9877,
      "step": 27916
    },
    {
      "epoch": 1.520192222607938,
      "grad_norm": 0.5101934537252294,
      "learning_rate": 2.8696350837868168e-05,
      "loss": 11.8981,
      "step": 27917
    },
    {
      "epoch": 1.520246676604521,
      "grad_norm": 0.5393295124478232,
      "learning_rate": 2.869016839690122e-05,
      "loss": 11.7954,
      "step": 27918
    },
    {
      "epoch": 1.520301130601104,
      "grad_norm": 0.5665424341421378,
      "learning_rate": 2.86839865104515e-05,
      "loss": 11.9649,
      "step": 27919
    },
    {
      "epoch": 1.520355584597687,
      "grad_norm": 0.5372327587059348,
      "learning_rate": 2.867780517856703e-05,
      "loss": 11.8867,
      "step": 27920
    },
    {
      "epoch": 1.52041003859427,
      "grad_norm": 0.49457536122256185,
      "learning_rate": 2.8671624401295947e-05,
      "loss": 11.7953,
      "step": 27921
    },
    {
      "epoch": 1.520464492590853,
      "grad_norm": 0.5682484408441838,
      "learning_rate": 2.866544417868624e-05,
      "loss": 11.865,
      "step": 27922
    },
    {
      "epoch": 1.520518946587436,
      "grad_norm": 0.5497355483267916,
      "learning_rate": 2.865926451078603e-05,
      "loss": 11.7453,
      "step": 27923
    },
    {
      "epoch": 1.520573400584019,
      "grad_norm": 0.5312350452988377,
      "learning_rate": 2.8653085397643355e-05,
      "loss": 11.7752,
      "step": 27924
    },
    {
      "epoch": 1.520627854580602,
      "grad_norm": 0.5337459565912904,
      "learning_rate": 2.864690683930621e-05,
      "loss": 11.7479,
      "step": 27925
    },
    {
      "epoch": 1.520682308577185,
      "grad_norm": 0.5492033095210955,
      "learning_rate": 2.8640728835822715e-05,
      "loss": 11.8602,
      "step": 27926
    },
    {
      "epoch": 1.520736762573768,
      "grad_norm": 0.5882903796976917,
      "learning_rate": 2.863455138724085e-05,
      "loss": 11.9274,
      "step": 27927
    },
    {
      "epoch": 1.5207912165703512,
      "grad_norm": 0.5464474040295534,
      "learning_rate": 2.862837449360871e-05,
      "loss": 11.8038,
      "step": 27928
    },
    {
      "epoch": 1.5208456705669342,
      "grad_norm": 0.5259964522455184,
      "learning_rate": 2.862219815497428e-05,
      "loss": 11.8672,
      "step": 27929
    },
    {
      "epoch": 1.5209001245635172,
      "grad_norm": 0.5164772302638427,
      "learning_rate": 2.86160223713856e-05,
      "loss": 11.8341,
      "step": 27930
    },
    {
      "epoch": 1.5209545785601002,
      "grad_norm": 0.5307716165980209,
      "learning_rate": 2.860984714289072e-05,
      "loss": 11.7679,
      "step": 27931
    },
    {
      "epoch": 1.5210090325566832,
      "grad_norm": 0.6043738213104092,
      "learning_rate": 2.8603672469537622e-05,
      "loss": 11.8279,
      "step": 27932
    },
    {
      "epoch": 1.5210634865532662,
      "grad_norm": 0.5445614445443454,
      "learning_rate": 2.8597498351374376e-05,
      "loss": 11.7401,
      "step": 27933
    },
    {
      "epoch": 1.5211179405498494,
      "grad_norm": 0.5289380992588216,
      "learning_rate": 2.8591324788448948e-05,
      "loss": 11.8751,
      "step": 27934
    },
    {
      "epoch": 1.5211723945464324,
      "grad_norm": 0.533654251638464,
      "learning_rate": 2.8585151780809328e-05,
      "loss": 11.8401,
      "step": 27935
    },
    {
      "epoch": 1.5212268485430154,
      "grad_norm": 0.5220964872865659,
      "learning_rate": 2.8578979328503576e-05,
      "loss": 11.8242,
      "step": 27936
    },
    {
      "epoch": 1.5212813025395984,
      "grad_norm": 0.5114364612323417,
      "learning_rate": 2.857280743157962e-05,
      "loss": 11.8001,
      "step": 27937
    },
    {
      "epoch": 1.5213357565361814,
      "grad_norm": 0.5523660842971195,
      "learning_rate": 2.856663609008553e-05,
      "loss": 11.8084,
      "step": 27938
    },
    {
      "epoch": 1.5213902105327644,
      "grad_norm": 0.52333885065501,
      "learning_rate": 2.856046530406922e-05,
      "loss": 11.8418,
      "step": 27939
    },
    {
      "epoch": 1.5214446645293473,
      "grad_norm": 0.5295371740160958,
      "learning_rate": 2.8554295073578708e-05,
      "loss": 11.8021,
      "step": 27940
    },
    {
      "epoch": 1.5214991185259303,
      "grad_norm": 0.4960778756772864,
      "learning_rate": 2.8548125398662017e-05,
      "loss": 11.8328,
      "step": 27941
    },
    {
      "epoch": 1.5215535725225133,
      "grad_norm": 0.46425175646156275,
      "learning_rate": 2.8541956279367053e-05,
      "loss": 11.729,
      "step": 27942
    },
    {
      "epoch": 1.5216080265190963,
      "grad_norm": 0.5012255929276607,
      "learning_rate": 2.853578771574189e-05,
      "loss": 11.8035,
      "step": 27943
    },
    {
      "epoch": 1.5216624805156793,
      "grad_norm": 0.5888817324911219,
      "learning_rate": 2.8529619707834355e-05,
      "loss": 11.8283,
      "step": 27944
    },
    {
      "epoch": 1.5217169345122623,
      "grad_norm": 0.6008255980546373,
      "learning_rate": 2.8523452255692485e-05,
      "loss": 11.9394,
      "step": 27945
    },
    {
      "epoch": 1.5217713885088453,
      "grad_norm": 0.6249937246002554,
      "learning_rate": 2.8517285359364266e-05,
      "loss": 11.92,
      "step": 27946
    },
    {
      "epoch": 1.5218258425054283,
      "grad_norm": 0.5441509701530162,
      "learning_rate": 2.851111901889759e-05,
      "loss": 11.9523,
      "step": 27947
    },
    {
      "epoch": 1.5218802965020113,
      "grad_norm": 0.5510325862380181,
      "learning_rate": 2.850495323434048e-05,
      "loss": 11.8965,
      "step": 27948
    },
    {
      "epoch": 1.5219347504985943,
      "grad_norm": 0.5338604703249146,
      "learning_rate": 2.8498788005740816e-05,
      "loss": 11.863,
      "step": 27949
    },
    {
      "epoch": 1.5219892044951773,
      "grad_norm": 0.6182806408308447,
      "learning_rate": 2.8492623333146585e-05,
      "loss": 12.011,
      "step": 27950
    },
    {
      "epoch": 1.5220436584917603,
      "grad_norm": 0.5586257192807768,
      "learning_rate": 2.8486459216605688e-05,
      "loss": 12.0284,
      "step": 27951
    },
    {
      "epoch": 1.5220981124883435,
      "grad_norm": 0.5386383629515035,
      "learning_rate": 2.848029565616607e-05,
      "loss": 11.6983,
      "step": 27952
    },
    {
      "epoch": 1.5221525664849265,
      "grad_norm": 0.6378948206143068,
      "learning_rate": 2.847413265187573e-05,
      "loss": 12.0341,
      "step": 27953
    },
    {
      "epoch": 1.5222070204815095,
      "grad_norm": 0.6468948500218227,
      "learning_rate": 2.8467970203782467e-05,
      "loss": 11.8029,
      "step": 27954
    },
    {
      "epoch": 1.5222614744780925,
      "grad_norm": 0.5759773580548446,
      "learning_rate": 2.846180831193429e-05,
      "loss": 11.8949,
      "step": 27955
    },
    {
      "epoch": 1.5223159284746755,
      "grad_norm": 0.6236305897816501,
      "learning_rate": 2.8455646976379068e-05,
      "loss": 11.9684,
      "step": 27956
    },
    {
      "epoch": 1.5223703824712587,
      "grad_norm": 0.6303340620397581,
      "learning_rate": 2.844948619716473e-05,
      "loss": 11.6356,
      "step": 27957
    },
    {
      "epoch": 1.5224248364678417,
      "grad_norm": 0.5853771207077727,
      "learning_rate": 2.844332597433921e-05,
      "loss": 11.7633,
      "step": 27958
    },
    {
      "epoch": 1.5224792904644247,
      "grad_norm": 0.5212751154680605,
      "learning_rate": 2.8437166307950368e-05,
      "loss": 11.9005,
      "step": 27959
    },
    {
      "epoch": 1.5225337444610076,
      "grad_norm": 0.5753545971375285,
      "learning_rate": 2.8431007198046144e-05,
      "loss": 11.848,
      "step": 27960
    },
    {
      "epoch": 1.5225881984575906,
      "grad_norm": 0.5314260131438653,
      "learning_rate": 2.8424848644674385e-05,
      "loss": 11.6512,
      "step": 27961
    },
    {
      "epoch": 1.5226426524541736,
      "grad_norm": 0.5560664066073572,
      "learning_rate": 2.841869064788303e-05,
      "loss": 11.8517,
      "step": 27962
    },
    {
      "epoch": 1.5226971064507566,
      "grad_norm": 0.5235610838337695,
      "learning_rate": 2.841253320771994e-05,
      "loss": 11.8405,
      "step": 27963
    },
    {
      "epoch": 1.5227515604473396,
      "grad_norm": 0.6135400382910773,
      "learning_rate": 2.8406376324232974e-05,
      "loss": 11.6442,
      "step": 27964
    },
    {
      "epoch": 1.5228060144439226,
      "grad_norm": 0.6182186923824203,
      "learning_rate": 2.8400219997470056e-05,
      "loss": 11.7927,
      "step": 27965
    },
    {
      "epoch": 1.5228604684405056,
      "grad_norm": 0.5503423157192251,
      "learning_rate": 2.839406422747901e-05,
      "loss": 11.8811,
      "step": 27966
    },
    {
      "epoch": 1.5229149224370886,
      "grad_norm": 0.5613776072349622,
      "learning_rate": 2.8387909014307722e-05,
      "loss": 11.8915,
      "step": 27967
    },
    {
      "epoch": 1.5229693764336716,
      "grad_norm": 0.5330578499553932,
      "learning_rate": 2.8381754358004097e-05,
      "loss": 11.7234,
      "step": 27968
    },
    {
      "epoch": 1.5230238304302546,
      "grad_norm": 0.556692852189812,
      "learning_rate": 2.837560025861593e-05,
      "loss": 11.6645,
      "step": 27969
    },
    {
      "epoch": 1.5230782844268376,
      "grad_norm": 0.529759914988266,
      "learning_rate": 2.836944671619114e-05,
      "loss": 11.8636,
      "step": 27970
    },
    {
      "epoch": 1.5231327384234206,
      "grad_norm": 0.5558955504395425,
      "learning_rate": 2.8363293730777517e-05,
      "loss": 11.9139,
      "step": 27971
    },
    {
      "epoch": 1.5231871924200036,
      "grad_norm": 0.5073574032386341,
      "learning_rate": 2.8357141302422962e-05,
      "loss": 11.8774,
      "step": 27972
    },
    {
      "epoch": 1.5232416464165865,
      "grad_norm": 0.5408827877794644,
      "learning_rate": 2.835098943117529e-05,
      "loss": 11.8919,
      "step": 27973
    },
    {
      "epoch": 1.5232961004131695,
      "grad_norm": 0.5485433703136475,
      "learning_rate": 2.8344838117082306e-05,
      "loss": 11.6821,
      "step": 27974
    },
    {
      "epoch": 1.5233505544097528,
      "grad_norm": 0.5524202991264017,
      "learning_rate": 2.8338687360191906e-05,
      "loss": 11.7799,
      "step": 27975
    },
    {
      "epoch": 1.5234050084063357,
      "grad_norm": 0.5407395598972906,
      "learning_rate": 2.8332537160551864e-05,
      "loss": 11.7938,
      "step": 27976
    },
    {
      "epoch": 1.5234594624029187,
      "grad_norm": 0.5985515785085527,
      "learning_rate": 2.8326387518210063e-05,
      "loss": 11.9475,
      "step": 27977
    },
    {
      "epoch": 1.5235139163995017,
      "grad_norm": 0.5743204461344149,
      "learning_rate": 2.8320238433214263e-05,
      "loss": 11.9857,
      "step": 27978
    },
    {
      "epoch": 1.5235683703960847,
      "grad_norm": 0.5341807793487325,
      "learning_rate": 2.8314089905612306e-05,
      "loss": 11.8742,
      "step": 27979
    },
    {
      "epoch": 1.5236228243926677,
      "grad_norm": 0.5275568888986025,
      "learning_rate": 2.8307941935452043e-05,
      "loss": 11.8501,
      "step": 27980
    },
    {
      "epoch": 1.523677278389251,
      "grad_norm": 0.5393791917241146,
      "learning_rate": 2.830179452278121e-05,
      "loss": 11.766,
      "step": 27981
    },
    {
      "epoch": 1.523731732385834,
      "grad_norm": 0.5768720464720104,
      "learning_rate": 2.8295647667647685e-05,
      "loss": 11.9164,
      "step": 27982
    },
    {
      "epoch": 1.523786186382417,
      "grad_norm": 0.5896321753178678,
      "learning_rate": 2.8289501370099225e-05,
      "loss": 11.7657,
      "step": 27983
    },
    {
      "epoch": 1.523840640379,
      "grad_norm": 0.5534454910447092,
      "learning_rate": 2.8283355630183593e-05,
      "loss": 11.8392,
      "step": 27984
    },
    {
      "epoch": 1.523895094375583,
      "grad_norm": 0.5516639208822186,
      "learning_rate": 2.8277210447948653e-05,
      "loss": 11.8244,
      "step": 27985
    },
    {
      "epoch": 1.523949548372166,
      "grad_norm": 0.686931899725569,
      "learning_rate": 2.8271065823442123e-05,
      "loss": 11.8538,
      "step": 27986
    },
    {
      "epoch": 1.524004002368749,
      "grad_norm": 0.5315660720617317,
      "learning_rate": 2.8264921756711837e-05,
      "loss": 11.9026,
      "step": 27987
    },
    {
      "epoch": 1.5240584563653319,
      "grad_norm": 0.6007169368610384,
      "learning_rate": 2.8258778247805517e-05,
      "loss": 11.8662,
      "step": 27988
    },
    {
      "epoch": 1.5241129103619149,
      "grad_norm": 0.5597611080799025,
      "learning_rate": 2.825263529677097e-05,
      "loss": 11.8968,
      "step": 27989
    },
    {
      "epoch": 1.5241673643584979,
      "grad_norm": 0.5920000808107831,
      "learning_rate": 2.8246492903655998e-05,
      "loss": 11.8771,
      "step": 27990
    },
    {
      "epoch": 1.5242218183550809,
      "grad_norm": 0.5544028723552854,
      "learning_rate": 2.8240351068508297e-05,
      "loss": 11.8372,
      "step": 27991
    },
    {
      "epoch": 1.5242762723516639,
      "grad_norm": 0.5826707811122956,
      "learning_rate": 2.82342097913757e-05,
      "loss": 11.8462,
      "step": 27992
    },
    {
      "epoch": 1.5243307263482468,
      "grad_norm": 0.5593061795049994,
      "learning_rate": 2.8228069072305907e-05,
      "loss": 11.7339,
      "step": 27993
    },
    {
      "epoch": 1.5243851803448298,
      "grad_norm": 0.555009710492283,
      "learning_rate": 2.8221928911346663e-05,
      "loss": 11.7679,
      "step": 27994
    },
    {
      "epoch": 1.5244396343414128,
      "grad_norm": 0.588777087305147,
      "learning_rate": 2.821578930854577e-05,
      "loss": 11.8769,
      "step": 27995
    },
    {
      "epoch": 1.5244940883379958,
      "grad_norm": 0.5612662130441314,
      "learning_rate": 2.8209650263950905e-05,
      "loss": 11.6895,
      "step": 27996
    },
    {
      "epoch": 1.5245485423345788,
      "grad_norm": 0.5517453495317268,
      "learning_rate": 2.8203511777609858e-05,
      "loss": 11.7565,
      "step": 27997
    },
    {
      "epoch": 1.524602996331162,
      "grad_norm": 0.5360699191588418,
      "learning_rate": 2.819737384957033e-05,
      "loss": 11.8972,
      "step": 27998
    },
    {
      "epoch": 1.524657450327745,
      "grad_norm": 0.5216156089956738,
      "learning_rate": 2.819123647988009e-05,
      "loss": 11.7221,
      "step": 27999
    },
    {
      "epoch": 1.524711904324328,
      "grad_norm": 0.5835597939173611,
      "learning_rate": 2.81850996685868e-05,
      "loss": 11.8809,
      "step": 28000
    },
    {
      "epoch": 1.524766358320911,
      "grad_norm": 0.5840532266995864,
      "learning_rate": 2.8178963415738225e-05,
      "loss": 11.8049,
      "step": 28001
    },
    {
      "epoch": 1.524820812317494,
      "grad_norm": 0.6183406611645063,
      "learning_rate": 2.81728277213821e-05,
      "loss": 11.8851,
      "step": 28002
    },
    {
      "epoch": 1.524875266314077,
      "grad_norm": 0.5232583144111126,
      "learning_rate": 2.816669258556611e-05,
      "loss": 11.8841,
      "step": 28003
    },
    {
      "epoch": 1.5249297203106602,
      "grad_norm": 0.5773888316303224,
      "learning_rate": 2.8160558008337977e-05,
      "loss": 11.8857,
      "step": 28004
    },
    {
      "epoch": 1.5249841743072432,
      "grad_norm": 0.576444798531799,
      "learning_rate": 2.815442398974536e-05,
      "loss": 11.8206,
      "step": 28005
    },
    {
      "epoch": 1.5250386283038262,
      "grad_norm": 0.475299752941559,
      "learning_rate": 2.8148290529835987e-05,
      "loss": 11.8208,
      "step": 28006
    },
    {
      "epoch": 1.5250930823004092,
      "grad_norm": 0.5790512256596804,
      "learning_rate": 2.8142157628657583e-05,
      "loss": 11.8499,
      "step": 28007
    },
    {
      "epoch": 1.5251475362969922,
      "grad_norm": 0.5234532016890456,
      "learning_rate": 2.8136025286257784e-05,
      "loss": 11.8159,
      "step": 28008
    },
    {
      "epoch": 1.5252019902935752,
      "grad_norm": 0.5630963999270417,
      "learning_rate": 2.8129893502684334e-05,
      "loss": 11.8672,
      "step": 28009
    },
    {
      "epoch": 1.5252564442901582,
      "grad_norm": 0.5455318970182873,
      "learning_rate": 2.8123762277984856e-05,
      "loss": 11.8057,
      "step": 28010
    },
    {
      "epoch": 1.5253108982867412,
      "grad_norm": 0.5171199728328628,
      "learning_rate": 2.8117631612207084e-05,
      "loss": 11.721,
      "step": 28011
    },
    {
      "epoch": 1.5253653522833241,
      "grad_norm": 0.5221933409063921,
      "learning_rate": 2.811150150539863e-05,
      "loss": 11.8984,
      "step": 28012
    },
    {
      "epoch": 1.5254198062799071,
      "grad_norm": 0.5187042053234086,
      "learning_rate": 2.8105371957607228e-05,
      "loss": 11.8053,
      "step": 28013
    },
    {
      "epoch": 1.5254742602764901,
      "grad_norm": 0.5429878060486782,
      "learning_rate": 2.8099242968880512e-05,
      "loss": 11.8103,
      "step": 28014
    },
    {
      "epoch": 1.5255287142730731,
      "grad_norm": 0.6503001961805629,
      "learning_rate": 2.809311453926612e-05,
      "loss": 11.9886,
      "step": 28015
    },
    {
      "epoch": 1.5255831682696561,
      "grad_norm": 0.5230869261194098,
      "learning_rate": 2.808698666881171e-05,
      "loss": 11.8418,
      "step": 28016
    },
    {
      "epoch": 1.5256376222662391,
      "grad_norm": 0.5723371956038334,
      "learning_rate": 2.8080859357564993e-05,
      "loss": 11.9124,
      "step": 28017
    },
    {
      "epoch": 1.525692076262822,
      "grad_norm": 0.5966982453619117,
      "learning_rate": 2.807473260557355e-05,
      "loss": 11.9901,
      "step": 28018
    },
    {
      "epoch": 1.525746530259405,
      "grad_norm": 0.49105321129198276,
      "learning_rate": 2.806860641288507e-05,
      "loss": 11.849,
      "step": 28019
    },
    {
      "epoch": 1.525800984255988,
      "grad_norm": 0.5817851378221217,
      "learning_rate": 2.806248077954714e-05,
      "loss": 11.7378,
      "step": 28020
    },
    {
      "epoch": 1.525855438252571,
      "grad_norm": 0.5241202288140655,
      "learning_rate": 2.8056355705607452e-05,
      "loss": 11.7037,
      "step": 28021
    },
    {
      "epoch": 1.5259098922491543,
      "grad_norm": 0.49166570019066175,
      "learning_rate": 2.8050231191113573e-05,
      "loss": 11.795,
      "step": 28022
    },
    {
      "epoch": 1.5259643462457373,
      "grad_norm": 0.5136362026482226,
      "learning_rate": 2.8044107236113203e-05,
      "loss": 11.8897,
      "step": 28023
    },
    {
      "epoch": 1.5260188002423203,
      "grad_norm": 0.5261733293137251,
      "learning_rate": 2.8037983840653913e-05,
      "loss": 11.8093,
      "step": 28024
    },
    {
      "epoch": 1.5260732542389033,
      "grad_norm": 0.4760619588143287,
      "learning_rate": 2.80318610047833e-05,
      "loss": 11.7524,
      "step": 28025
    },
    {
      "epoch": 1.5261277082354863,
      "grad_norm": 0.5216907813608113,
      "learning_rate": 2.8025738728549046e-05,
      "loss": 11.8057,
      "step": 28026
    },
    {
      "epoch": 1.5261821622320695,
      "grad_norm": 0.4826666149391475,
      "learning_rate": 2.8019617011998677e-05,
      "loss": 11.759,
      "step": 28027
    },
    {
      "epoch": 1.5262366162286525,
      "grad_norm": 0.5486646520149847,
      "learning_rate": 2.8013495855179837e-05,
      "loss": 11.8722,
      "step": 28028
    },
    {
      "epoch": 1.5262910702252355,
      "grad_norm": 0.5339442745290974,
      "learning_rate": 2.8007375258140156e-05,
      "loss": 11.8051,
      "step": 28029
    },
    {
      "epoch": 1.5263455242218185,
      "grad_norm": 0.5702150244759141,
      "learning_rate": 2.8001255220927168e-05,
      "loss": 11.8569,
      "step": 28030
    },
    {
      "epoch": 1.5263999782184015,
      "grad_norm": 0.5393840986818391,
      "learning_rate": 2.7995135743588528e-05,
      "loss": 11.8107,
      "step": 28031
    },
    {
      "epoch": 1.5264544322149844,
      "grad_norm": 0.5263479220858477,
      "learning_rate": 2.7989016826171755e-05,
      "loss": 11.9613,
      "step": 28032
    },
    {
      "epoch": 1.5265088862115674,
      "grad_norm": 0.5609368998722691,
      "learning_rate": 2.7982898468724493e-05,
      "loss": 11.9058,
      "step": 28033
    },
    {
      "epoch": 1.5265633402081504,
      "grad_norm": 0.5145333261289492,
      "learning_rate": 2.797678067129429e-05,
      "loss": 11.8689,
      "step": 28034
    },
    {
      "epoch": 1.5266177942047334,
      "grad_norm": 0.5112366708808339,
      "learning_rate": 2.797066343392869e-05,
      "loss": 11.7795,
      "step": 28035
    },
    {
      "epoch": 1.5266722482013164,
      "grad_norm": 0.5292153657436052,
      "learning_rate": 2.7964546756675313e-05,
      "loss": 11.8977,
      "step": 28036
    },
    {
      "epoch": 1.5267267021978994,
      "grad_norm": 0.5187700953438499,
      "learning_rate": 2.795843063958168e-05,
      "loss": 11.885,
      "step": 28037
    },
    {
      "epoch": 1.5267811561944824,
      "grad_norm": 0.5260785610507742,
      "learning_rate": 2.7952315082695367e-05,
      "loss": 11.7061,
      "step": 28038
    },
    {
      "epoch": 1.5268356101910654,
      "grad_norm": 0.5926034175885343,
      "learning_rate": 2.794620008606397e-05,
      "loss": 11.8106,
      "step": 28039
    },
    {
      "epoch": 1.5268900641876484,
      "grad_norm": 0.5698490004315689,
      "learning_rate": 2.794008564973497e-05,
      "loss": 11.8,
      "step": 28040
    },
    {
      "epoch": 1.5269445181842314,
      "grad_norm": 0.6678278354852758,
      "learning_rate": 2.7933971773755986e-05,
      "loss": 11.8865,
      "step": 28041
    },
    {
      "epoch": 1.5269989721808144,
      "grad_norm": 0.5213235120952687,
      "learning_rate": 2.7927858458174483e-05,
      "loss": 11.752,
      "step": 28042
    },
    {
      "epoch": 1.5270534261773974,
      "grad_norm": 0.5463359333407151,
      "learning_rate": 2.7921745703038072e-05,
      "loss": 11.6669,
      "step": 28043
    },
    {
      "epoch": 1.5271078801739804,
      "grad_norm": 0.5344727131368201,
      "learning_rate": 2.7915633508394258e-05,
      "loss": 11.8769,
      "step": 28044
    },
    {
      "epoch": 1.5271623341705636,
      "grad_norm": 0.5402139975432181,
      "learning_rate": 2.7909521874290524e-05,
      "loss": 11.9086,
      "step": 28045
    },
    {
      "epoch": 1.5272167881671466,
      "grad_norm": 0.6095364033149826,
      "learning_rate": 2.7903410800774478e-05,
      "loss": 11.901,
      "step": 28046
    },
    {
      "epoch": 1.5272712421637296,
      "grad_norm": 0.5795695853077685,
      "learning_rate": 2.7897300287893568e-05,
      "loss": 11.9074,
      "step": 28047
    },
    {
      "epoch": 1.5273256961603126,
      "grad_norm": 0.541051857417764,
      "learning_rate": 2.7891190335695373e-05,
      "loss": 11.8282,
      "step": 28048
    },
    {
      "epoch": 1.5273801501568955,
      "grad_norm": 0.6023298925941373,
      "learning_rate": 2.788508094422735e-05,
      "loss": 11.9771,
      "step": 28049
    },
    {
      "epoch": 1.5274346041534785,
      "grad_norm": 0.590164425957543,
      "learning_rate": 2.7878972113537017e-05,
      "loss": 11.8233,
      "step": 28050
    },
    {
      "epoch": 1.5274890581500618,
      "grad_norm": 0.5883901291502481,
      "learning_rate": 2.787286384367194e-05,
      "loss": 11.859,
      "step": 28051
    },
    {
      "epoch": 1.5275435121466447,
      "grad_norm": 0.5598461273887382,
      "learning_rate": 2.7866756134679528e-05,
      "loss": 11.8201,
      "step": 28052
    },
    {
      "epoch": 1.5275979661432277,
      "grad_norm": 0.5510855141903137,
      "learning_rate": 2.786064898660734e-05,
      "loss": 11.9274,
      "step": 28053
    },
    {
      "epoch": 1.5276524201398107,
      "grad_norm": 0.6171418019889008,
      "learning_rate": 2.785454239950286e-05,
      "loss": 11.9351,
      "step": 28054
    },
    {
      "epoch": 1.5277068741363937,
      "grad_norm": 0.5441721929752072,
      "learning_rate": 2.784843637341351e-05,
      "loss": 11.8461,
      "step": 28055
    },
    {
      "epoch": 1.5277613281329767,
      "grad_norm": 0.550427509481184,
      "learning_rate": 2.784233090838686e-05,
      "loss": 11.7901,
      "step": 28056
    },
    {
      "epoch": 1.5278157821295597,
      "grad_norm": 0.5388240833165424,
      "learning_rate": 2.7836226004470322e-05,
      "loss": 11.9641,
      "step": 28057
    },
    {
      "epoch": 1.5278702361261427,
      "grad_norm": 0.554171740403026,
      "learning_rate": 2.7830121661711405e-05,
      "loss": 11.837,
      "step": 28058
    },
    {
      "epoch": 1.5279246901227257,
      "grad_norm": 0.5383805673000276,
      "learning_rate": 2.782401788015755e-05,
      "loss": 11.8651,
      "step": 28059
    },
    {
      "epoch": 1.5279791441193087,
      "grad_norm": 0.5657370833017105,
      "learning_rate": 2.781791465985626e-05,
      "loss": 11.8957,
      "step": 28060
    },
    {
      "epoch": 1.5280335981158917,
      "grad_norm": 0.5216301810143582,
      "learning_rate": 2.781181200085494e-05,
      "loss": 11.7694,
      "step": 28061
    },
    {
      "epoch": 1.5280880521124747,
      "grad_norm": 0.4991395494787023,
      "learning_rate": 2.7805709903201073e-05,
      "loss": 11.8382,
      "step": 28062
    },
    {
      "epoch": 1.5281425061090577,
      "grad_norm": 0.5198154381445942,
      "learning_rate": 2.7799608366942143e-05,
      "loss": 11.8512,
      "step": 28063
    },
    {
      "epoch": 1.5281969601056407,
      "grad_norm": 0.5222690177492786,
      "learning_rate": 2.779350739212557e-05,
      "loss": 11.8377,
      "step": 28064
    },
    {
      "epoch": 1.5282514141022236,
      "grad_norm": 0.4910046177255441,
      "learning_rate": 2.778740697879877e-05,
      "loss": 11.7966,
      "step": 28065
    },
    {
      "epoch": 1.5283058680988066,
      "grad_norm": 0.5890142611054503,
      "learning_rate": 2.7781307127009226e-05,
      "loss": 11.8705,
      "step": 28066
    },
    {
      "epoch": 1.5283603220953896,
      "grad_norm": 0.5150666867395313,
      "learning_rate": 2.7775207836804317e-05,
      "loss": 11.635,
      "step": 28067
    },
    {
      "epoch": 1.5284147760919728,
      "grad_norm": 0.5877921587720709,
      "learning_rate": 2.7769109108231528e-05,
      "loss": 11.8718,
      "step": 28068
    },
    {
      "epoch": 1.5284692300885558,
      "grad_norm": 0.5905701907420984,
      "learning_rate": 2.776301094133824e-05,
      "loss": 11.8247,
      "step": 28069
    },
    {
      "epoch": 1.5285236840851388,
      "grad_norm": 0.5319275146743632,
      "learning_rate": 2.7756913336171917e-05,
      "loss": 11.885,
      "step": 28070
    },
    {
      "epoch": 1.5285781380817218,
      "grad_norm": 0.5710362753324053,
      "learning_rate": 2.7750816292779926e-05,
      "loss": 11.9469,
      "step": 28071
    },
    {
      "epoch": 1.5286325920783048,
      "grad_norm": 0.5839829945359638,
      "learning_rate": 2.7744719811209695e-05,
      "loss": 11.8371,
      "step": 28072
    },
    {
      "epoch": 1.5286870460748878,
      "grad_norm": 0.5087264130086645,
      "learning_rate": 2.7738623891508688e-05,
      "loss": 11.7981,
      "step": 28073
    },
    {
      "epoch": 1.528741500071471,
      "grad_norm": 0.6097435097493412,
      "learning_rate": 2.7732528533724245e-05,
      "loss": 12.018,
      "step": 28074
    },
    {
      "epoch": 1.528795954068054,
      "grad_norm": 0.581255067665826,
      "learning_rate": 2.772643373790378e-05,
      "loss": 11.9142,
      "step": 28075
    },
    {
      "epoch": 1.528850408064637,
      "grad_norm": 0.5529909667015764,
      "learning_rate": 2.772033950409466e-05,
      "loss": 11.6942,
      "step": 28076
    },
    {
      "epoch": 1.52890486206122,
      "grad_norm": 0.5507846355961354,
      "learning_rate": 2.7714245832344298e-05,
      "loss": 11.9129,
      "step": 28077
    },
    {
      "epoch": 1.528959316057803,
      "grad_norm": 0.5263795313275622,
      "learning_rate": 2.770815272270012e-05,
      "loss": 11.8114,
      "step": 28078
    },
    {
      "epoch": 1.529013770054386,
      "grad_norm": 0.5915882574630816,
      "learning_rate": 2.7702060175209433e-05,
      "loss": 11.947,
      "step": 28079
    },
    {
      "epoch": 1.529068224050969,
      "grad_norm": 0.5420336359228751,
      "learning_rate": 2.7695968189919684e-05,
      "loss": 11.6884,
      "step": 28080
    },
    {
      "epoch": 1.529122678047552,
      "grad_norm": 0.5761177601026898,
      "learning_rate": 2.7689876766878177e-05,
      "loss": 11.9456,
      "step": 28081
    },
    {
      "epoch": 1.529177132044135,
      "grad_norm": 0.6179788515082738,
      "learning_rate": 2.7683785906132353e-05,
      "loss": 11.9586,
      "step": 28082
    },
    {
      "epoch": 1.529231586040718,
      "grad_norm": 0.5376348725469609,
      "learning_rate": 2.7677695607729505e-05,
      "loss": 11.7374,
      "step": 28083
    },
    {
      "epoch": 1.529286040037301,
      "grad_norm": 0.5875898380240233,
      "learning_rate": 2.7671605871717044e-05,
      "loss": 11.8818,
      "step": 28084
    },
    {
      "epoch": 1.529340494033884,
      "grad_norm": 0.5319121946915005,
      "learning_rate": 2.7665516698142314e-05,
      "loss": 11.812,
      "step": 28085
    },
    {
      "epoch": 1.529394948030467,
      "grad_norm": 0.5232715675086153,
      "learning_rate": 2.765942808705263e-05,
      "loss": 11.6767,
      "step": 28086
    },
    {
      "epoch": 1.52944940202705,
      "grad_norm": 0.587033249086594,
      "learning_rate": 2.7653340038495358e-05,
      "loss": 11.9199,
      "step": 28087
    },
    {
      "epoch": 1.529503856023633,
      "grad_norm": 0.5670701469594571,
      "learning_rate": 2.7647252552517878e-05,
      "loss": 11.8395,
      "step": 28088
    },
    {
      "epoch": 1.529558310020216,
      "grad_norm": 0.5456216942507206,
      "learning_rate": 2.7641165629167464e-05,
      "loss": 11.8476,
      "step": 28089
    },
    {
      "epoch": 1.529612764016799,
      "grad_norm": 0.6133476880784436,
      "learning_rate": 2.7635079268491516e-05,
      "loss": 11.7352,
      "step": 28090
    },
    {
      "epoch": 1.5296672180133821,
      "grad_norm": 0.5689587520370178,
      "learning_rate": 2.762899347053729e-05,
      "loss": 11.9563,
      "step": 28091
    },
    {
      "epoch": 1.5297216720099651,
      "grad_norm": 0.5608064813813687,
      "learning_rate": 2.7622908235352185e-05,
      "loss": 11.9223,
      "step": 28092
    },
    {
      "epoch": 1.529776126006548,
      "grad_norm": 0.5265785301100568,
      "learning_rate": 2.7616823562983453e-05,
      "loss": 11.8348,
      "step": 28093
    },
    {
      "epoch": 1.529830580003131,
      "grad_norm": 0.5261958437030664,
      "learning_rate": 2.761073945347846e-05,
      "loss": 11.901,
      "step": 28094
    },
    {
      "epoch": 1.529885033999714,
      "grad_norm": 0.5544943092947802,
      "learning_rate": 2.7604655906884502e-05,
      "loss": 11.8705,
      "step": 28095
    },
    {
      "epoch": 1.529939487996297,
      "grad_norm": 0.5932595399334087,
      "learning_rate": 2.7598572923248857e-05,
      "loss": 11.9126,
      "step": 28096
    },
    {
      "epoch": 1.5299939419928803,
      "grad_norm": 0.5241433903541007,
      "learning_rate": 2.759249050261887e-05,
      "loss": 11.8082,
      "step": 28097
    },
    {
      "epoch": 1.5300483959894633,
      "grad_norm": 0.508394728574014,
      "learning_rate": 2.7586408645041796e-05,
      "loss": 11.8364,
      "step": 28098
    },
    {
      "epoch": 1.5301028499860463,
      "grad_norm": 0.5660475859610954,
      "learning_rate": 2.758032735056495e-05,
      "loss": 11.9086,
      "step": 28099
    },
    {
      "epoch": 1.5301573039826293,
      "grad_norm": 0.5854390292446331,
      "learning_rate": 2.7574246619235654e-05,
      "loss": 11.892,
      "step": 28100
    },
    {
      "epoch": 1.5302117579792123,
      "grad_norm": 0.6135300347394426,
      "learning_rate": 2.756816645110113e-05,
      "loss": 11.8013,
      "step": 28101
    },
    {
      "epoch": 1.5302662119757953,
      "grad_norm": 0.47022733445491777,
      "learning_rate": 2.7562086846208723e-05,
      "loss": 11.7453,
      "step": 28102
    },
    {
      "epoch": 1.5303206659723783,
      "grad_norm": 0.5604935484080049,
      "learning_rate": 2.7556007804605654e-05,
      "loss": 11.8919,
      "step": 28103
    },
    {
      "epoch": 1.5303751199689613,
      "grad_norm": 0.5560539287446726,
      "learning_rate": 2.754992932633924e-05,
      "loss": 11.9655,
      "step": 28104
    },
    {
      "epoch": 1.5304295739655442,
      "grad_norm": 0.5035252030556944,
      "learning_rate": 2.7543851411456722e-05,
      "loss": 11.845,
      "step": 28105
    },
    {
      "epoch": 1.5304840279621272,
      "grad_norm": 0.5998247291009688,
      "learning_rate": 2.753777406000534e-05,
      "loss": 11.8718,
      "step": 28106
    },
    {
      "epoch": 1.5305384819587102,
      "grad_norm": 0.6277000221854228,
      "learning_rate": 2.7531697272032408e-05,
      "loss": 11.8319,
      "step": 28107
    },
    {
      "epoch": 1.5305929359552932,
      "grad_norm": 0.5311575974019177,
      "learning_rate": 2.7525621047585115e-05,
      "loss": 11.862,
      "step": 28108
    },
    {
      "epoch": 1.5306473899518762,
      "grad_norm": 0.5240917378904438,
      "learning_rate": 2.7519545386710787e-05,
      "loss": 11.6807,
      "step": 28109
    },
    {
      "epoch": 1.5307018439484592,
      "grad_norm": 0.5482481864827582,
      "learning_rate": 2.75134702894566e-05,
      "loss": 11.7755,
      "step": 28110
    },
    {
      "epoch": 1.5307562979450422,
      "grad_norm": 0.5199580600118068,
      "learning_rate": 2.7507395755869813e-05,
      "loss": 11.7715,
      "step": 28111
    },
    {
      "epoch": 1.5308107519416252,
      "grad_norm": 0.46853550384032233,
      "learning_rate": 2.7501321785997713e-05,
      "loss": 11.8175,
      "step": 28112
    },
    {
      "epoch": 1.5308652059382082,
      "grad_norm": 0.5378687057721647,
      "learning_rate": 2.7495248379887462e-05,
      "loss": 11.794,
      "step": 28113
    },
    {
      "epoch": 1.5309196599347912,
      "grad_norm": 0.6076040383716326,
      "learning_rate": 2.7489175537586342e-05,
      "loss": 11.9345,
      "step": 28114
    },
    {
      "epoch": 1.5309741139313744,
      "grad_norm": 0.5302827915958461,
      "learning_rate": 2.748310325914155e-05,
      "loss": 11.78,
      "step": 28115
    },
    {
      "epoch": 1.5310285679279574,
      "grad_norm": 0.5206254842623453,
      "learning_rate": 2.7477031544600297e-05,
      "loss": 11.844,
      "step": 28116
    },
    {
      "epoch": 1.5310830219245404,
      "grad_norm": 0.5782492612681784,
      "learning_rate": 2.7470960394009826e-05,
      "loss": 11.6907,
      "step": 28117
    },
    {
      "epoch": 1.5311374759211234,
      "grad_norm": 0.5816187740888729,
      "learning_rate": 2.7464889807417293e-05,
      "loss": 11.7165,
      "step": 28118
    },
    {
      "epoch": 1.5311919299177064,
      "grad_norm": 0.5147939602641817,
      "learning_rate": 2.7458819784869983e-05,
      "loss": 11.7735,
      "step": 28119
    },
    {
      "epoch": 1.5312463839142894,
      "grad_norm": 0.5354121280135299,
      "learning_rate": 2.745275032641502e-05,
      "loss": 11.6952,
      "step": 28120
    },
    {
      "epoch": 1.5313008379108726,
      "grad_norm": 0.5344169272867186,
      "learning_rate": 2.7446681432099642e-05,
      "loss": 11.8144,
      "step": 28121
    },
    {
      "epoch": 1.5313552919074556,
      "grad_norm": 0.5251465455208807,
      "learning_rate": 2.744061310197106e-05,
      "loss": 11.766,
      "step": 28122
    },
    {
      "epoch": 1.5314097459040386,
      "grad_norm": 0.5721180654907436,
      "learning_rate": 2.743454533607641e-05,
      "loss": 11.814,
      "step": 28123
    },
    {
      "epoch": 1.5314641999006215,
      "grad_norm": 0.5198986708643252,
      "learning_rate": 2.7428478134462963e-05,
      "loss": 11.801,
      "step": 28124
    },
    {
      "epoch": 1.5315186538972045,
      "grad_norm": 0.6349634894030276,
      "learning_rate": 2.742241149717779e-05,
      "loss": 11.8605,
      "step": 28125
    },
    {
      "epoch": 1.5315731078937875,
      "grad_norm": 0.5476776064162383,
      "learning_rate": 2.7416345424268107e-05,
      "loss": 11.8861,
      "step": 28126
    },
    {
      "epoch": 1.5316275618903705,
      "grad_norm": 0.5389532605450704,
      "learning_rate": 2.7410279915781123e-05,
      "loss": 11.9331,
      "step": 28127
    },
    {
      "epoch": 1.5316820158869535,
      "grad_norm": 0.7137800699458078,
      "learning_rate": 2.7404214971763952e-05,
      "loss": 11.8757,
      "step": 28128
    },
    {
      "epoch": 1.5317364698835365,
      "grad_norm": 0.6023696650918151,
      "learning_rate": 2.7398150592263805e-05,
      "loss": 11.7711,
      "step": 28129
    },
    {
      "epoch": 1.5317909238801195,
      "grad_norm": 0.5275100704514966,
      "learning_rate": 2.7392086777327787e-05,
      "loss": 11.9261,
      "step": 28130
    },
    {
      "epoch": 1.5318453778767025,
      "grad_norm": 0.5377998200518751,
      "learning_rate": 2.738602352700311e-05,
      "loss": 11.8072,
      "step": 28131
    },
    {
      "epoch": 1.5318998318732855,
      "grad_norm": 0.5857096227630232,
      "learning_rate": 2.737996084133686e-05,
      "loss": 11.9087,
      "step": 28132
    },
    {
      "epoch": 1.5319542858698685,
      "grad_norm": 0.608249892848749,
      "learning_rate": 2.7373898720376212e-05,
      "loss": 11.8523,
      "step": 28133
    },
    {
      "epoch": 1.5320087398664515,
      "grad_norm": 0.5743496564238021,
      "learning_rate": 2.7367837164168375e-05,
      "loss": 11.7277,
      "step": 28134
    },
    {
      "epoch": 1.5320631938630345,
      "grad_norm": 0.5774906890183263,
      "learning_rate": 2.7361776172760346e-05,
      "loss": 11.8523,
      "step": 28135
    },
    {
      "epoch": 1.5321176478596175,
      "grad_norm": 0.5693847254039437,
      "learning_rate": 2.7355715746199327e-05,
      "loss": 11.8029,
      "step": 28136
    },
    {
      "epoch": 1.5321721018562005,
      "grad_norm": 0.48642514926955,
      "learning_rate": 2.734965588453248e-05,
      "loss": 11.7881,
      "step": 28137
    },
    {
      "epoch": 1.5322265558527837,
      "grad_norm": 0.5208125835512221,
      "learning_rate": 2.7343596587806865e-05,
      "loss": 11.769,
      "step": 28138
    },
    {
      "epoch": 1.5322810098493667,
      "grad_norm": 0.5756246884970783,
      "learning_rate": 2.7337537856069652e-05,
      "loss": 11.8422,
      "step": 28139
    },
    {
      "epoch": 1.5323354638459497,
      "grad_norm": 0.5510921702394357,
      "learning_rate": 2.7331479689367888e-05,
      "loss": 11.8868,
      "step": 28140
    },
    {
      "epoch": 1.5323899178425326,
      "grad_norm": 0.6054078622464308,
      "learning_rate": 2.7325422087748776e-05,
      "loss": 11.8649,
      "step": 28141
    },
    {
      "epoch": 1.5324443718391156,
      "grad_norm": 0.5215026850061641,
      "learning_rate": 2.7319365051259326e-05,
      "loss": 11.848,
      "step": 28142
    },
    {
      "epoch": 1.5324988258356986,
      "grad_norm": 0.5448259488081963,
      "learning_rate": 2.7313308579946684e-05,
      "loss": 11.8352,
      "step": 28143
    },
    {
      "epoch": 1.5325532798322818,
      "grad_norm": 0.536583365184915,
      "learning_rate": 2.7307252673858007e-05,
      "loss": 11.7611,
      "step": 28144
    },
    {
      "epoch": 1.5326077338288648,
      "grad_norm": 0.5328057211891392,
      "learning_rate": 2.7301197333040264e-05,
      "loss": 11.8555,
      "step": 28145
    },
    {
      "epoch": 1.5326621878254478,
      "grad_norm": 0.5382358859349715,
      "learning_rate": 2.729514255754063e-05,
      "loss": 11.8034,
      "step": 28146
    },
    {
      "epoch": 1.5327166418220308,
      "grad_norm": 0.54372606929296,
      "learning_rate": 2.7289088347406135e-05,
      "loss": 11.8172,
      "step": 28147
    },
    {
      "epoch": 1.5327710958186138,
      "grad_norm": 0.5071314621892234,
      "learning_rate": 2.7283034702683885e-05,
      "loss": 11.771,
      "step": 28148
    },
    {
      "epoch": 1.5328255498151968,
      "grad_norm": 0.47760232581095086,
      "learning_rate": 2.7276981623420974e-05,
      "loss": 11.7506,
      "step": 28149
    },
    {
      "epoch": 1.5328800038117798,
      "grad_norm": 0.5779126067885992,
      "learning_rate": 2.7270929109664423e-05,
      "loss": 11.776,
      "step": 28150
    },
    {
      "epoch": 1.5329344578083628,
      "grad_norm": 0.532756986419023,
      "learning_rate": 2.7264877161461343e-05,
      "loss": 11.8969,
      "step": 28151
    },
    {
      "epoch": 1.5329889118049458,
      "grad_norm": 0.5301360810526304,
      "learning_rate": 2.7258825778858764e-05,
      "loss": 11.909,
      "step": 28152
    },
    {
      "epoch": 1.5330433658015288,
      "grad_norm": 0.5409486462875552,
      "learning_rate": 2.725277496190377e-05,
      "loss": 11.855,
      "step": 28153
    },
    {
      "epoch": 1.5330978197981118,
      "grad_norm": 0.5483761385219517,
      "learning_rate": 2.724672471064341e-05,
      "loss": 11.8397,
      "step": 28154
    },
    {
      "epoch": 1.5331522737946948,
      "grad_norm": 0.5196613633243852,
      "learning_rate": 2.7240675025124684e-05,
      "loss": 11.8376,
      "step": 28155
    },
    {
      "epoch": 1.5332067277912778,
      "grad_norm": 0.5921240861695857,
      "learning_rate": 2.7234625905394697e-05,
      "loss": 11.9667,
      "step": 28156
    },
    {
      "epoch": 1.5332611817878607,
      "grad_norm": 0.5912985763166101,
      "learning_rate": 2.7228577351500427e-05,
      "loss": 11.991,
      "step": 28157
    },
    {
      "epoch": 1.5333156357844437,
      "grad_norm": 0.54196239765597,
      "learning_rate": 2.7222529363488992e-05,
      "loss": 11.7573,
      "step": 28158
    },
    {
      "epoch": 1.5333700897810267,
      "grad_norm": 0.5850025102559534,
      "learning_rate": 2.7216481941407323e-05,
      "loss": 11.9364,
      "step": 28159
    },
    {
      "epoch": 1.5334245437776097,
      "grad_norm": 0.5977439025733949,
      "learning_rate": 2.721043508530251e-05,
      "loss": 11.9675,
      "step": 28160
    },
    {
      "epoch": 1.533478997774193,
      "grad_norm": 0.5388939351892333,
      "learning_rate": 2.720438879522158e-05,
      "loss": 11.8457,
      "step": 28161
    },
    {
      "epoch": 1.533533451770776,
      "grad_norm": 0.5213198958516342,
      "learning_rate": 2.7198343071211508e-05,
      "loss": 11.789,
      "step": 28162
    },
    {
      "epoch": 1.533587905767359,
      "grad_norm": 0.5201233933753517,
      "learning_rate": 2.7192297913319363e-05,
      "loss": 11.7326,
      "step": 28163
    },
    {
      "epoch": 1.533642359763942,
      "grad_norm": 0.5696029144591914,
      "learning_rate": 2.7186253321592114e-05,
      "loss": 11.8743,
      "step": 28164
    },
    {
      "epoch": 1.533696813760525,
      "grad_norm": 0.5661831990271644,
      "learning_rate": 2.718020929607673e-05,
      "loss": 11.5473,
      "step": 28165
    },
    {
      "epoch": 1.533751267757108,
      "grad_norm": 0.5733737118565293,
      "learning_rate": 2.7174165836820297e-05,
      "loss": 11.897,
      "step": 28166
    },
    {
      "epoch": 1.5338057217536911,
      "grad_norm": 0.5742035351847482,
      "learning_rate": 2.7168122943869723e-05,
      "loss": 11.8856,
      "step": 28167
    },
    {
      "epoch": 1.5338601757502741,
      "grad_norm": 0.5999607268231191,
      "learning_rate": 2.716208061727208e-05,
      "loss": 11.8658,
      "step": 28168
    },
    {
      "epoch": 1.533914629746857,
      "grad_norm": 0.5218650452531289,
      "learning_rate": 2.715603885707427e-05,
      "loss": 11.7596,
      "step": 28169
    },
    {
      "epoch": 1.53396908374344,
      "grad_norm": 0.5579045882793019,
      "learning_rate": 2.7149997663323323e-05,
      "loss": 11.8881,
      "step": 28170
    },
    {
      "epoch": 1.534023537740023,
      "grad_norm": 0.5119751726399356,
      "learning_rate": 2.7143957036066238e-05,
      "loss": 11.8144,
      "step": 28171
    },
    {
      "epoch": 1.534077991736606,
      "grad_norm": 0.528054243795056,
      "learning_rate": 2.713791697534994e-05,
      "loss": 11.7638,
      "step": 28172
    },
    {
      "epoch": 1.534132445733189,
      "grad_norm": 0.6120665347017282,
      "learning_rate": 2.7131877481221448e-05,
      "loss": 11.6907,
      "step": 28173
    },
    {
      "epoch": 1.534186899729772,
      "grad_norm": 0.5020852473806008,
      "learning_rate": 2.712583855372769e-05,
      "loss": 11.791,
      "step": 28174
    },
    {
      "epoch": 1.534241353726355,
      "grad_norm": 0.5512264992726531,
      "learning_rate": 2.71198001929156e-05,
      "loss": 11.7513,
      "step": 28175
    },
    {
      "epoch": 1.534295807722938,
      "grad_norm": 0.5514363641125494,
      "learning_rate": 2.71137623988322e-05,
      "loss": 11.8073,
      "step": 28176
    },
    {
      "epoch": 1.534350261719521,
      "grad_norm": 0.5312078707535559,
      "learning_rate": 2.7107725171524378e-05,
      "loss": 11.8327,
      "step": 28177
    },
    {
      "epoch": 1.534404715716104,
      "grad_norm": 0.5247512696484431,
      "learning_rate": 2.7101688511039137e-05,
      "loss": 11.7713,
      "step": 28178
    },
    {
      "epoch": 1.534459169712687,
      "grad_norm": 0.5459385675423508,
      "learning_rate": 2.7095652417423357e-05,
      "loss": 11.8542,
      "step": 28179
    },
    {
      "epoch": 1.53451362370927,
      "grad_norm": 0.5749601248779559,
      "learning_rate": 2.7089616890724044e-05,
      "loss": 11.8236,
      "step": 28180
    },
    {
      "epoch": 1.534568077705853,
      "grad_norm": 0.5664622210684159,
      "learning_rate": 2.7083581930988065e-05,
      "loss": 11.8597,
      "step": 28181
    },
    {
      "epoch": 1.534622531702436,
      "grad_norm": 0.5690130145344745,
      "learning_rate": 2.7077547538262392e-05,
      "loss": 11.6244,
      "step": 28182
    },
    {
      "epoch": 1.534676985699019,
      "grad_norm": 0.6098614195320473,
      "learning_rate": 2.7071513712593954e-05,
      "loss": 11.8586,
      "step": 28183
    },
    {
      "epoch": 1.534731439695602,
      "grad_norm": 0.5345772468011127,
      "learning_rate": 2.7065480454029635e-05,
      "loss": 11.8058,
      "step": 28184
    },
    {
      "epoch": 1.5347858936921852,
      "grad_norm": 0.5004064107412757,
      "learning_rate": 2.7059447762616396e-05,
      "loss": 11.692,
      "step": 28185
    },
    {
      "epoch": 1.5348403476887682,
      "grad_norm": 0.5509148116583732,
      "learning_rate": 2.7053415638401125e-05,
      "loss": 11.8781,
      "step": 28186
    },
    {
      "epoch": 1.5348948016853512,
      "grad_norm": 0.5388835013163247,
      "learning_rate": 2.7047384081430706e-05,
      "loss": 11.9216,
      "step": 28187
    },
    {
      "epoch": 1.5349492556819342,
      "grad_norm": 0.5831123752442173,
      "learning_rate": 2.7041353091752076e-05,
      "loss": 11.878,
      "step": 28188
    },
    {
      "epoch": 1.5350037096785172,
      "grad_norm": 0.5942715795712129,
      "learning_rate": 2.70353226694121e-05,
      "loss": 11.8724,
      "step": 28189
    },
    {
      "epoch": 1.5350581636751004,
      "grad_norm": 0.576071406714683,
      "learning_rate": 2.7029292814457717e-05,
      "loss": 11.8425,
      "step": 28190
    },
    {
      "epoch": 1.5351126176716834,
      "grad_norm": 0.5763627257223947,
      "learning_rate": 2.702326352693576e-05,
      "loss": 11.7898,
      "step": 28191
    },
    {
      "epoch": 1.5351670716682664,
      "grad_norm": 0.543371020897107,
      "learning_rate": 2.7017234806893143e-05,
      "loss": 11.8641,
      "step": 28192
    },
    {
      "epoch": 1.5352215256648494,
      "grad_norm": 0.6112162290672735,
      "learning_rate": 2.7011206654376787e-05,
      "loss": 11.8811,
      "step": 28193
    },
    {
      "epoch": 1.5352759796614324,
      "grad_norm": 0.5228178862268024,
      "learning_rate": 2.7005179069433482e-05,
      "loss": 11.7781,
      "step": 28194
    },
    {
      "epoch": 1.5353304336580154,
      "grad_norm": 0.5842052961124035,
      "learning_rate": 2.6999152052110222e-05,
      "loss": 11.9354,
      "step": 28195
    },
    {
      "epoch": 1.5353848876545984,
      "grad_norm": 0.5523614778190836,
      "learning_rate": 2.6993125602453728e-05,
      "loss": 11.8795,
      "step": 28196
    },
    {
      "epoch": 1.5354393416511813,
      "grad_norm": 0.561699747227993,
      "learning_rate": 2.6987099720510933e-05,
      "loss": 11.7798,
      "step": 28197
    },
    {
      "epoch": 1.5354937956477643,
      "grad_norm": 0.5862358470888648,
      "learning_rate": 2.6981074406328732e-05,
      "loss": 11.894,
      "step": 28198
    },
    {
      "epoch": 1.5355482496443473,
      "grad_norm": 0.5816568902861724,
      "learning_rate": 2.697504965995391e-05,
      "loss": 11.6072,
      "step": 28199
    },
    {
      "epoch": 1.5356027036409303,
      "grad_norm": 0.5415710531098747,
      "learning_rate": 2.6969025481433375e-05,
      "loss": 11.8295,
      "step": 28200
    },
    {
      "epoch": 1.5356571576375133,
      "grad_norm": 0.5528077817902902,
      "learning_rate": 2.6963001870813907e-05,
      "loss": 11.9264,
      "step": 28201
    },
    {
      "epoch": 1.5357116116340963,
      "grad_norm": 0.5498728942280875,
      "learning_rate": 2.695697882814243e-05,
      "loss": 11.808,
      "step": 28202
    },
    {
      "epoch": 1.5357660656306793,
      "grad_norm": 0.5274560242694658,
      "learning_rate": 2.6950956353465697e-05,
      "loss": 11.847,
      "step": 28203
    },
    {
      "epoch": 1.5358205196272623,
      "grad_norm": 0.7194185033400121,
      "learning_rate": 2.6944934446830584e-05,
      "loss": 11.9278,
      "step": 28204
    },
    {
      "epoch": 1.5358749736238453,
      "grad_norm": 0.5294529582944018,
      "learning_rate": 2.693891310828396e-05,
      "loss": 11.8302,
      "step": 28205
    },
    {
      "epoch": 1.5359294276204283,
      "grad_norm": 0.5764854018767049,
      "learning_rate": 2.693289233787255e-05,
      "loss": 11.8552,
      "step": 28206
    },
    {
      "epoch": 1.5359838816170113,
      "grad_norm": 0.5182438175989933,
      "learning_rate": 2.6926872135643244e-05,
      "loss": 11.7975,
      "step": 28207
    },
    {
      "epoch": 1.5360383356135945,
      "grad_norm": 0.5586890704804036,
      "learning_rate": 2.6920852501642812e-05,
      "loss": 11.9141,
      "step": 28208
    },
    {
      "epoch": 1.5360927896101775,
      "grad_norm": 0.5515382589733375,
      "learning_rate": 2.6914833435918085e-05,
      "loss": 11.6309,
      "step": 28209
    },
    {
      "epoch": 1.5361472436067605,
      "grad_norm": 0.5639640398732918,
      "learning_rate": 2.6908814938515904e-05,
      "loss": 11.7213,
      "step": 28210
    },
    {
      "epoch": 1.5362016976033435,
      "grad_norm": 0.5368623491940676,
      "learning_rate": 2.6902797009482992e-05,
      "loss": 11.7836,
      "step": 28211
    },
    {
      "epoch": 1.5362561515999265,
      "grad_norm": 0.5347723030275935,
      "learning_rate": 2.6896779648866234e-05,
      "loss": 11.7712,
      "step": 28212
    },
    {
      "epoch": 1.5363106055965094,
      "grad_norm": 0.561264192490462,
      "learning_rate": 2.6890762856712338e-05,
      "loss": 11.9036,
      "step": 28213
    },
    {
      "epoch": 1.5363650595930927,
      "grad_norm": 0.5754915513990667,
      "learning_rate": 2.688474663306816e-05,
      "loss": 11.867,
      "step": 28214
    },
    {
      "epoch": 1.5364195135896757,
      "grad_norm": 0.5205203897511063,
      "learning_rate": 2.687873097798046e-05,
      "loss": 11.8342,
      "step": 28215
    },
    {
      "epoch": 1.5364739675862586,
      "grad_norm": 0.5193764182969883,
      "learning_rate": 2.6872715891495982e-05,
      "loss": 11.7938,
      "step": 28216
    },
    {
      "epoch": 1.5365284215828416,
      "grad_norm": 0.593055727163197,
      "learning_rate": 2.6866701373661552e-05,
      "loss": 11.9238,
      "step": 28217
    },
    {
      "epoch": 1.5365828755794246,
      "grad_norm": 0.5334897617058839,
      "learning_rate": 2.686068742452389e-05,
      "loss": 11.8103,
      "step": 28218
    },
    {
      "epoch": 1.5366373295760076,
      "grad_norm": 0.5530211029638681,
      "learning_rate": 2.685467404412978e-05,
      "loss": 11.9189,
      "step": 28219
    },
    {
      "epoch": 1.5366917835725906,
      "grad_norm": 0.5918491619100299,
      "learning_rate": 2.684866123252603e-05,
      "loss": 11.916,
      "step": 28220
    },
    {
      "epoch": 1.5367462375691736,
      "grad_norm": 0.6564766542400814,
      "learning_rate": 2.684264898975932e-05,
      "loss": 11.9384,
      "step": 28221
    },
    {
      "epoch": 1.5368006915657566,
      "grad_norm": 0.5300217143991779,
      "learning_rate": 2.683663731587648e-05,
      "loss": 11.8974,
      "step": 28222
    },
    {
      "epoch": 1.5368551455623396,
      "grad_norm": 0.6238055965656362,
      "learning_rate": 2.6830626210924182e-05,
      "loss": 11.8434,
      "step": 28223
    },
    {
      "epoch": 1.5369095995589226,
      "grad_norm": 0.4747300939338794,
      "learning_rate": 2.6824615674949227e-05,
      "loss": 11.876,
      "step": 28224
    },
    {
      "epoch": 1.5369640535555056,
      "grad_norm": 0.5202331247292236,
      "learning_rate": 2.6818605707998345e-05,
      "loss": 11.7362,
      "step": 28225
    },
    {
      "epoch": 1.5370185075520886,
      "grad_norm": 0.6266174323981987,
      "learning_rate": 2.681259631011822e-05,
      "loss": 11.8971,
      "step": 28226
    },
    {
      "epoch": 1.5370729615486716,
      "grad_norm": 0.5401195218607323,
      "learning_rate": 2.680658748135566e-05,
      "loss": 11.8222,
      "step": 28227
    },
    {
      "epoch": 1.5371274155452546,
      "grad_norm": 0.5287654703216333,
      "learning_rate": 2.6800579221757306e-05,
      "loss": 11.8862,
      "step": 28228
    },
    {
      "epoch": 1.5371818695418376,
      "grad_norm": 0.5382665462414589,
      "learning_rate": 2.679457153136996e-05,
      "loss": 11.838,
      "step": 28229
    },
    {
      "epoch": 1.5372363235384205,
      "grad_norm": 0.5244955209680553,
      "learning_rate": 2.678856441024028e-05,
      "loss": 11.7849,
      "step": 28230
    },
    {
      "epoch": 1.5372907775350038,
      "grad_norm": 0.5285389177552661,
      "learning_rate": 2.6782557858414992e-05,
      "loss": 11.8859,
      "step": 28231
    },
    {
      "epoch": 1.5373452315315868,
      "grad_norm": 0.5646328371339733,
      "learning_rate": 2.6776551875940847e-05,
      "loss": 11.8143,
      "step": 28232
    },
    {
      "epoch": 1.5373996855281697,
      "grad_norm": 0.5363893936188058,
      "learning_rate": 2.6770546462864478e-05,
      "loss": 11.8303,
      "step": 28233
    },
    {
      "epoch": 1.5374541395247527,
      "grad_norm": 0.5184131565912657,
      "learning_rate": 2.676454161923265e-05,
      "loss": 11.9607,
      "step": 28234
    },
    {
      "epoch": 1.5375085935213357,
      "grad_norm": 0.5467243664910497,
      "learning_rate": 2.6758537345092037e-05,
      "loss": 11.7396,
      "step": 28235
    },
    {
      "epoch": 1.5375630475179187,
      "grad_norm": 0.5002391440336198,
      "learning_rate": 2.6752533640489284e-05,
      "loss": 11.6786,
      "step": 28236
    },
    {
      "epoch": 1.537617501514502,
      "grad_norm": 0.5546305245961396,
      "learning_rate": 2.6746530505471147e-05,
      "loss": 11.8863,
      "step": 28237
    },
    {
      "epoch": 1.537671955511085,
      "grad_norm": 0.5463276192766239,
      "learning_rate": 2.6740527940084236e-05,
      "loss": 11.7478,
      "step": 28238
    },
    {
      "epoch": 1.537726409507668,
      "grad_norm": 0.6154320058044087,
      "learning_rate": 2.67345259443753e-05,
      "loss": 11.8903,
      "step": 28239
    },
    {
      "epoch": 1.537780863504251,
      "grad_norm": 0.5863681216047066,
      "learning_rate": 2.6728524518390952e-05,
      "loss": 11.8566,
      "step": 28240
    },
    {
      "epoch": 1.537835317500834,
      "grad_norm": 0.5603478206952835,
      "learning_rate": 2.672252366217788e-05,
      "loss": 11.8774,
      "step": 28241
    },
    {
      "epoch": 1.537889771497417,
      "grad_norm": 0.5593682753283273,
      "learning_rate": 2.6716523375782776e-05,
      "loss": 11.7486,
      "step": 28242
    },
    {
      "epoch": 1.537944225494,
      "grad_norm": 0.5579982858200164,
      "learning_rate": 2.6710523659252252e-05,
      "loss": 11.6957,
      "step": 28243
    },
    {
      "epoch": 1.5379986794905829,
      "grad_norm": 0.5395499117612986,
      "learning_rate": 2.670452451263301e-05,
      "loss": 11.9338,
      "step": 28244
    },
    {
      "epoch": 1.5380531334871659,
      "grad_norm": 0.5455376572211799,
      "learning_rate": 2.6698525935971685e-05,
      "loss": 11.7827,
      "step": 28245
    },
    {
      "epoch": 1.5381075874837489,
      "grad_norm": 0.5247406524040752,
      "learning_rate": 2.6692527929314893e-05,
      "loss": 11.8625,
      "step": 28246
    },
    {
      "epoch": 1.5381620414803319,
      "grad_norm": 0.5160767431977465,
      "learning_rate": 2.6686530492709316e-05,
      "loss": 11.7607,
      "step": 28247
    },
    {
      "epoch": 1.5382164954769149,
      "grad_norm": 0.6140882322238727,
      "learning_rate": 2.6680533626201544e-05,
      "loss": 11.9011,
      "step": 28248
    },
    {
      "epoch": 1.5382709494734979,
      "grad_norm": 0.5190865977042493,
      "learning_rate": 2.6674537329838267e-05,
      "loss": 11.8215,
      "step": 28249
    },
    {
      "epoch": 1.5383254034700808,
      "grad_norm": 0.5606977311297707,
      "learning_rate": 2.6668541603666052e-05,
      "loss": 11.7624,
      "step": 28250
    },
    {
      "epoch": 1.5383798574666638,
      "grad_norm": 0.5544608244481404,
      "learning_rate": 2.6662546447731594e-05,
      "loss": 11.866,
      "step": 28251
    },
    {
      "epoch": 1.5384343114632468,
      "grad_norm": 0.5405060388636937,
      "learning_rate": 2.6656551862081437e-05,
      "loss": 11.8598,
      "step": 28252
    },
    {
      "epoch": 1.5384887654598298,
      "grad_norm": 0.5814147566330362,
      "learning_rate": 2.6650557846762237e-05,
      "loss": 11.7744,
      "step": 28253
    },
    {
      "epoch": 1.5385432194564128,
      "grad_norm": 0.4946928011055275,
      "learning_rate": 2.6644564401820626e-05,
      "loss": 11.7776,
      "step": 28254
    },
    {
      "epoch": 1.538597673452996,
      "grad_norm": 0.5513304184837484,
      "learning_rate": 2.663857152730319e-05,
      "loss": 11.8948,
      "step": 28255
    },
    {
      "epoch": 1.538652127449579,
      "grad_norm": 0.5325891969231507,
      "learning_rate": 2.6632579223256515e-05,
      "loss": 11.8826,
      "step": 28256
    },
    {
      "epoch": 1.538706581446162,
      "grad_norm": 0.5272251203217627,
      "learning_rate": 2.6626587489727184e-05,
      "loss": 11.6687,
      "step": 28257
    },
    {
      "epoch": 1.538761035442745,
      "grad_norm": 0.5104341806424597,
      "learning_rate": 2.6620596326761814e-05,
      "loss": 11.7976,
      "step": 28258
    },
    {
      "epoch": 1.538815489439328,
      "grad_norm": 0.6044950523686626,
      "learning_rate": 2.6614605734407016e-05,
      "loss": 12.0231,
      "step": 28259
    },
    {
      "epoch": 1.5388699434359112,
      "grad_norm": 0.509313261244652,
      "learning_rate": 2.660861571270933e-05,
      "loss": 11.8104,
      "step": 28260
    },
    {
      "epoch": 1.5389243974324942,
      "grad_norm": 0.5832287795387571,
      "learning_rate": 2.660262626171538e-05,
      "loss": 11.9543,
      "step": 28261
    },
    {
      "epoch": 1.5389788514290772,
      "grad_norm": 0.5313190113670253,
      "learning_rate": 2.659663738147169e-05,
      "loss": 11.9242,
      "step": 28262
    },
    {
      "epoch": 1.5390333054256602,
      "grad_norm": 0.5440514986939275,
      "learning_rate": 2.659064907202489e-05,
      "loss": 11.9818,
      "step": 28263
    },
    {
      "epoch": 1.5390877594222432,
      "grad_norm": 0.5605878741267645,
      "learning_rate": 2.658466133342147e-05,
      "loss": 11.9153,
      "step": 28264
    },
    {
      "epoch": 1.5391422134188262,
      "grad_norm": 0.5532632315372058,
      "learning_rate": 2.6578674165708073e-05,
      "loss": 11.8579,
      "step": 28265
    },
    {
      "epoch": 1.5391966674154092,
      "grad_norm": 0.5740420566631218,
      "learning_rate": 2.657268756893122e-05,
      "loss": 11.8121,
      "step": 28266
    },
    {
      "epoch": 1.5392511214119922,
      "grad_norm": 0.5150297688552529,
      "learning_rate": 2.656670154313744e-05,
      "loss": 11.7826,
      "step": 28267
    },
    {
      "epoch": 1.5393055754085752,
      "grad_norm": 0.5225642522836905,
      "learning_rate": 2.6560716088373294e-05,
      "loss": 11.7241,
      "step": 28268
    },
    {
      "epoch": 1.5393600294051581,
      "grad_norm": 0.5335641716258503,
      "learning_rate": 2.655473120468537e-05,
      "loss": 11.8361,
      "step": 28269
    },
    {
      "epoch": 1.5394144834017411,
      "grad_norm": 0.5868537446099409,
      "learning_rate": 2.6548746892120136e-05,
      "loss": 11.8187,
      "step": 28270
    },
    {
      "epoch": 1.5394689373983241,
      "grad_norm": 0.5275408793199444,
      "learning_rate": 2.6542763150724192e-05,
      "loss": 11.8821,
      "step": 28271
    },
    {
      "epoch": 1.5395233913949071,
      "grad_norm": 0.5738982718035953,
      "learning_rate": 2.6536779980544024e-05,
      "loss": 11.7754,
      "step": 28272
    },
    {
      "epoch": 1.5395778453914901,
      "grad_norm": 0.550355726212609,
      "learning_rate": 2.6530797381626194e-05,
      "loss": 11.8614,
      "step": 28273
    },
    {
      "epoch": 1.5396322993880731,
      "grad_norm": 0.5776006089965645,
      "learning_rate": 2.6524815354017175e-05,
      "loss": 11.8015,
      "step": 28274
    },
    {
      "epoch": 1.539686753384656,
      "grad_norm": 0.5433736193633831,
      "learning_rate": 2.651883389776355e-05,
      "loss": 11.8397,
      "step": 28275
    },
    {
      "epoch": 1.539741207381239,
      "grad_norm": 0.5378596487178605,
      "learning_rate": 2.6512853012911787e-05,
      "loss": 11.7706,
      "step": 28276
    },
    {
      "epoch": 1.539795661377822,
      "grad_norm": 0.5774975567190883,
      "learning_rate": 2.650687269950838e-05,
      "loss": 11.8928,
      "step": 28277
    },
    {
      "epoch": 1.5398501153744053,
      "grad_norm": 0.670412656161244,
      "learning_rate": 2.650089295759989e-05,
      "loss": 11.7932,
      "step": 28278
    },
    {
      "epoch": 1.5399045693709883,
      "grad_norm": 0.6209052285760718,
      "learning_rate": 2.6494913787232735e-05,
      "loss": 12.094,
      "step": 28279
    },
    {
      "epoch": 1.5399590233675713,
      "grad_norm": 0.5434472006128124,
      "learning_rate": 2.648893518845347e-05,
      "loss": 11.7558,
      "step": 28280
    },
    {
      "epoch": 1.5400134773641543,
      "grad_norm": 0.507490407696909,
      "learning_rate": 2.6482957161308598e-05,
      "loss": 11.7851,
      "step": 28281
    },
    {
      "epoch": 1.5400679313607373,
      "grad_norm": 0.5185829514338437,
      "learning_rate": 2.6476979705844552e-05,
      "loss": 11.8925,
      "step": 28282
    },
    {
      "epoch": 1.5401223853573203,
      "grad_norm": 0.5818697985336496,
      "learning_rate": 2.647100282210787e-05,
      "loss": 11.7104,
      "step": 28283
    },
    {
      "epoch": 1.5401768393539035,
      "grad_norm": 0.5285137941617885,
      "learning_rate": 2.6465026510144975e-05,
      "loss": 11.8305,
      "step": 28284
    },
    {
      "epoch": 1.5402312933504865,
      "grad_norm": 0.5613587002440537,
      "learning_rate": 2.645905077000239e-05,
      "loss": 11.8033,
      "step": 28285
    },
    {
      "epoch": 1.5402857473470695,
      "grad_norm": 0.5563490881689548,
      "learning_rate": 2.6453075601726562e-05,
      "loss": 11.981,
      "step": 28286
    },
    {
      "epoch": 1.5403402013436525,
      "grad_norm": 0.5663931025263615,
      "learning_rate": 2.6447101005363927e-05,
      "loss": 11.826,
      "step": 28287
    },
    {
      "epoch": 1.5403946553402355,
      "grad_norm": 0.5531980679284013,
      "learning_rate": 2.6441126980960994e-05,
      "loss": 11.8848,
      "step": 28288
    },
    {
      "epoch": 1.5404491093368184,
      "grad_norm": 0.5424048762741972,
      "learning_rate": 2.643515352856416e-05,
      "loss": 11.7974,
      "step": 28289
    },
    {
      "epoch": 1.5405035633334014,
      "grad_norm": 0.533448142546899,
      "learning_rate": 2.642918064821992e-05,
      "loss": 11.8026,
      "step": 28290
    },
    {
      "epoch": 1.5405580173299844,
      "grad_norm": 0.5165036958476561,
      "learning_rate": 2.6423208339974725e-05,
      "loss": 11.9402,
      "step": 28291
    },
    {
      "epoch": 1.5406124713265674,
      "grad_norm": 0.5354052136056343,
      "learning_rate": 2.6417236603874982e-05,
      "loss": 11.8617,
      "step": 28292
    },
    {
      "epoch": 1.5406669253231504,
      "grad_norm": 0.592659981834093,
      "learning_rate": 2.641126543996717e-05,
      "loss": 11.8951,
      "step": 28293
    },
    {
      "epoch": 1.5407213793197334,
      "grad_norm": 0.5192023515452038,
      "learning_rate": 2.640529484829768e-05,
      "loss": 11.7969,
      "step": 28294
    },
    {
      "epoch": 1.5407758333163164,
      "grad_norm": 0.5714755675377791,
      "learning_rate": 2.6399324828912973e-05,
      "loss": 11.9247,
      "step": 28295
    },
    {
      "epoch": 1.5408302873128994,
      "grad_norm": 0.5223828682124572,
      "learning_rate": 2.6393355381859475e-05,
      "loss": 11.7491,
      "step": 28296
    },
    {
      "epoch": 1.5408847413094824,
      "grad_norm": 0.6046746840026958,
      "learning_rate": 2.638738650718354e-05,
      "loss": 11.8696,
      "step": 28297
    },
    {
      "epoch": 1.5409391953060654,
      "grad_norm": 0.5080778644015773,
      "learning_rate": 2.6381418204931684e-05,
      "loss": 11.8089,
      "step": 28298
    },
    {
      "epoch": 1.5409936493026484,
      "grad_norm": 0.5254248962893969,
      "learning_rate": 2.637545047515022e-05,
      "loss": 11.7532,
      "step": 28299
    },
    {
      "epoch": 1.5410481032992314,
      "grad_norm": 0.5641932917252481,
      "learning_rate": 2.6369483317885637e-05,
      "loss": 11.8597,
      "step": 28300
    },
    {
      "epoch": 1.5411025572958146,
      "grad_norm": 0.5175369706337605,
      "learning_rate": 2.636351673318427e-05,
      "loss": 11.7573,
      "step": 28301
    },
    {
      "epoch": 1.5411570112923976,
      "grad_norm": 0.6062429143215914,
      "learning_rate": 2.6357550721092538e-05,
      "loss": 11.9393,
      "step": 28302
    },
    {
      "epoch": 1.5412114652889806,
      "grad_norm": 0.5595891504648797,
      "learning_rate": 2.6351585281656875e-05,
      "loss": 11.7512,
      "step": 28303
    },
    {
      "epoch": 1.5412659192855636,
      "grad_norm": 0.5788012801230924,
      "learning_rate": 2.634562041492361e-05,
      "loss": 11.9586,
      "step": 28304
    },
    {
      "epoch": 1.5413203732821465,
      "grad_norm": 0.5805037700823928,
      "learning_rate": 2.6339656120939182e-05,
      "loss": 11.9215,
      "step": 28305
    },
    {
      "epoch": 1.5413748272787295,
      "grad_norm": 0.5127059839878486,
      "learning_rate": 2.6333692399749942e-05,
      "loss": 11.8413,
      "step": 28306
    },
    {
      "epoch": 1.5414292812753128,
      "grad_norm": 0.5395360288185435,
      "learning_rate": 2.6327729251402233e-05,
      "loss": 11.7857,
      "step": 28307
    },
    {
      "epoch": 1.5414837352718957,
      "grad_norm": 0.5152761726323076,
      "learning_rate": 2.6321766675942483e-05,
      "loss": 11.8383,
      "step": 28308
    },
    {
      "epoch": 1.5415381892684787,
      "grad_norm": 0.5600835552397703,
      "learning_rate": 2.6315804673416998e-05,
      "loss": 11.805,
      "step": 28309
    },
    {
      "epoch": 1.5415926432650617,
      "grad_norm": 0.5454841709735719,
      "learning_rate": 2.630984324387221e-05,
      "loss": 11.8448,
      "step": 28310
    },
    {
      "epoch": 1.5416470972616447,
      "grad_norm": 0.6253522379181685,
      "learning_rate": 2.6303882387354406e-05,
      "loss": 11.9069,
      "step": 28311
    },
    {
      "epoch": 1.5417015512582277,
      "grad_norm": 0.6219127188901813,
      "learning_rate": 2.629792210391e-05,
      "loss": 11.8694,
      "step": 28312
    },
    {
      "epoch": 1.5417560052548107,
      "grad_norm": 0.5797193422957406,
      "learning_rate": 2.6291962393585278e-05,
      "loss": 11.9399,
      "step": 28313
    },
    {
      "epoch": 1.5418104592513937,
      "grad_norm": 0.5359935669224456,
      "learning_rate": 2.6286003256426628e-05,
      "loss": 11.7806,
      "step": 28314
    },
    {
      "epoch": 1.5418649132479767,
      "grad_norm": 0.5482595131675084,
      "learning_rate": 2.6280044692480432e-05,
      "loss": 11.9322,
      "step": 28315
    },
    {
      "epoch": 1.5419193672445597,
      "grad_norm": 0.5412211758477348,
      "learning_rate": 2.627408670179291e-05,
      "loss": 11.8118,
      "step": 28316
    },
    {
      "epoch": 1.5419738212411427,
      "grad_norm": 0.5031157443018751,
      "learning_rate": 2.6268129284410458e-05,
      "loss": 11.724,
      "step": 28317
    },
    {
      "epoch": 1.5420282752377257,
      "grad_norm": 0.48082930373461896,
      "learning_rate": 2.6262172440379416e-05,
      "loss": 11.9277,
      "step": 28318
    },
    {
      "epoch": 1.5420827292343087,
      "grad_norm": 0.5756715142540825,
      "learning_rate": 2.625621616974607e-05,
      "loss": 11.8244,
      "step": 28319
    },
    {
      "epoch": 1.5421371832308917,
      "grad_norm": 0.5387402752687224,
      "learning_rate": 2.6250260472556777e-05,
      "loss": 11.8202,
      "step": 28320
    },
    {
      "epoch": 1.5421916372274747,
      "grad_norm": 0.5338966248872453,
      "learning_rate": 2.62443053488578e-05,
      "loss": 11.9323,
      "step": 28321
    },
    {
      "epoch": 1.5422460912240576,
      "grad_norm": 0.49394165854848227,
      "learning_rate": 2.623835079869551e-05,
      "loss": 11.8239,
      "step": 28322
    },
    {
      "epoch": 1.5423005452206406,
      "grad_norm": 0.5510301241991037,
      "learning_rate": 2.6232396822116134e-05,
      "loss": 11.7766,
      "step": 28323
    },
    {
      "epoch": 1.5423549992172239,
      "grad_norm": 0.5920767057702554,
      "learning_rate": 2.6226443419166012e-05,
      "loss": 11.9232,
      "step": 28324
    },
    {
      "epoch": 1.5424094532138068,
      "grad_norm": 0.5743929819507603,
      "learning_rate": 2.62204905898915e-05,
      "loss": 11.8195,
      "step": 28325
    },
    {
      "epoch": 1.5424639072103898,
      "grad_norm": 0.5439752153702302,
      "learning_rate": 2.6214538334338766e-05,
      "loss": 11.9344,
      "step": 28326
    },
    {
      "epoch": 1.5425183612069728,
      "grad_norm": 0.5646101969548867,
      "learning_rate": 2.620858665255418e-05,
      "loss": 11.8105,
      "step": 28327
    },
    {
      "epoch": 1.5425728152035558,
      "grad_norm": 0.5761307663888735,
      "learning_rate": 2.6202635544583975e-05,
      "loss": 11.7824,
      "step": 28328
    },
    {
      "epoch": 1.5426272692001388,
      "grad_norm": 0.532423942765136,
      "learning_rate": 2.6196685010474452e-05,
      "loss": 11.9044,
      "step": 28329
    },
    {
      "epoch": 1.542681723196722,
      "grad_norm": 0.5435841278020633,
      "learning_rate": 2.619073505027192e-05,
      "loss": 11.9201,
      "step": 28330
    },
    {
      "epoch": 1.542736177193305,
      "grad_norm": 0.5538508421217291,
      "learning_rate": 2.618478566402257e-05,
      "loss": 11.9541,
      "step": 28331
    },
    {
      "epoch": 1.542790631189888,
      "grad_norm": 0.6162874260229597,
      "learning_rate": 2.6178836851772736e-05,
      "loss": 11.9137,
      "step": 28332
    },
    {
      "epoch": 1.542845085186471,
      "grad_norm": 0.5373943280467628,
      "learning_rate": 2.6172888613568612e-05,
      "loss": 11.8199,
      "step": 28333
    },
    {
      "epoch": 1.542899539183054,
      "grad_norm": 0.5179828636771145,
      "learning_rate": 2.616694094945653e-05,
      "loss": 11.8554,
      "step": 28334
    },
    {
      "epoch": 1.542953993179637,
      "grad_norm": 0.48028407935061235,
      "learning_rate": 2.6160993859482665e-05,
      "loss": 11.9126,
      "step": 28335
    },
    {
      "epoch": 1.54300844717622,
      "grad_norm": 0.5225454847238858,
      "learning_rate": 2.6155047343693317e-05,
      "loss": 11.8516,
      "step": 28336
    },
    {
      "epoch": 1.543062901172803,
      "grad_norm": 0.563803946682016,
      "learning_rate": 2.6149101402134702e-05,
      "loss": 11.8689,
      "step": 28337
    },
    {
      "epoch": 1.543117355169386,
      "grad_norm": 0.5273704956752464,
      "learning_rate": 2.6143156034853034e-05,
      "loss": 11.8562,
      "step": 28338
    },
    {
      "epoch": 1.543171809165969,
      "grad_norm": 0.599175341602558,
      "learning_rate": 2.6137211241894576e-05,
      "loss": 11.8287,
      "step": 28339
    },
    {
      "epoch": 1.543226263162552,
      "grad_norm": 0.5615828578947035,
      "learning_rate": 2.613126702330557e-05,
      "loss": 11.7832,
      "step": 28340
    },
    {
      "epoch": 1.543280717159135,
      "grad_norm": 0.5481125681678248,
      "learning_rate": 2.6125323379132193e-05,
      "loss": 11.6501,
      "step": 28341
    },
    {
      "epoch": 1.543335171155718,
      "grad_norm": 0.5813305801399807,
      "learning_rate": 2.6119380309420727e-05,
      "loss": 11.8198,
      "step": 28342
    },
    {
      "epoch": 1.543389625152301,
      "grad_norm": 0.5535303105664144,
      "learning_rate": 2.6113437814217313e-05,
      "loss": 11.9125,
      "step": 28343
    },
    {
      "epoch": 1.543444079148884,
      "grad_norm": 0.5820910869739198,
      "learning_rate": 2.6107495893568236e-05,
      "loss": 11.7817,
      "step": 28344
    },
    {
      "epoch": 1.543498533145467,
      "grad_norm": 0.614254078685859,
      "learning_rate": 2.6101554547519626e-05,
      "loss": 11.8894,
      "step": 28345
    },
    {
      "epoch": 1.54355298714205,
      "grad_norm": 0.5264325906590919,
      "learning_rate": 2.6095613776117743e-05,
      "loss": 11.8236,
      "step": 28346
    },
    {
      "epoch": 1.543607441138633,
      "grad_norm": 0.5658556435382978,
      "learning_rate": 2.608967357940877e-05,
      "loss": 11.8537,
      "step": 28347
    },
    {
      "epoch": 1.5436618951352161,
      "grad_norm": 0.5796891032107981,
      "learning_rate": 2.608373395743886e-05,
      "loss": 11.9766,
      "step": 28348
    },
    {
      "epoch": 1.5437163491317991,
      "grad_norm": 0.5228768863041379,
      "learning_rate": 2.607779491025426e-05,
      "loss": 11.8632,
      "step": 28349
    },
    {
      "epoch": 1.543770803128382,
      "grad_norm": 0.5562725029840014,
      "learning_rate": 2.607185643790109e-05,
      "loss": 11.7465,
      "step": 28350
    },
    {
      "epoch": 1.543825257124965,
      "grad_norm": 0.524910704869663,
      "learning_rate": 2.606591854042556e-05,
      "loss": 11.8588,
      "step": 28351
    },
    {
      "epoch": 1.543879711121548,
      "grad_norm": 0.5637855734704784,
      "learning_rate": 2.6059981217873885e-05,
      "loss": 11.9517,
      "step": 28352
    },
    {
      "epoch": 1.543934165118131,
      "grad_norm": 0.6208971999782115,
      "learning_rate": 2.605404447029216e-05,
      "loss": 11.8617,
      "step": 28353
    },
    {
      "epoch": 1.5439886191147143,
      "grad_norm": 0.5726753186084881,
      "learning_rate": 2.6048108297726613e-05,
      "loss": 11.9586,
      "step": 28354
    },
    {
      "epoch": 1.5440430731112973,
      "grad_norm": 0.5358515479739011,
      "learning_rate": 2.6042172700223357e-05,
      "loss": 11.8994,
      "step": 28355
    },
    {
      "epoch": 1.5440975271078803,
      "grad_norm": 0.548695997930056,
      "learning_rate": 2.603623767782859e-05,
      "loss": 11.9534,
      "step": 28356
    },
    {
      "epoch": 1.5441519811044633,
      "grad_norm": 0.5525914459172369,
      "learning_rate": 2.603030323058845e-05,
      "loss": 11.8632,
      "step": 28357
    },
    {
      "epoch": 1.5442064351010463,
      "grad_norm": 0.5346785184587497,
      "learning_rate": 2.602436935854904e-05,
      "loss": 11.7681,
      "step": 28358
    },
    {
      "epoch": 1.5442608890976293,
      "grad_norm": 0.5612780373793211,
      "learning_rate": 2.601843606175658e-05,
      "loss": 11.9691,
      "step": 28359
    },
    {
      "epoch": 1.5443153430942123,
      "grad_norm": 0.5176343562940072,
      "learning_rate": 2.6012503340257123e-05,
      "loss": 11.6743,
      "step": 28360
    },
    {
      "epoch": 1.5443697970907952,
      "grad_norm": 0.5351190897668885,
      "learning_rate": 2.6006571194096885e-05,
      "loss": 11.8922,
      "step": 28361
    },
    {
      "epoch": 1.5444242510873782,
      "grad_norm": 0.522983848147743,
      "learning_rate": 2.6000639623321933e-05,
      "loss": 11.5706,
      "step": 28362
    },
    {
      "epoch": 1.5444787050839612,
      "grad_norm": 0.5916990401567385,
      "learning_rate": 2.5994708627978415e-05,
      "loss": 11.8936,
      "step": 28363
    },
    {
      "epoch": 1.5445331590805442,
      "grad_norm": 0.5807024279553158,
      "learning_rate": 2.5988778208112476e-05,
      "loss": 11.8119,
      "step": 28364
    },
    {
      "epoch": 1.5445876130771272,
      "grad_norm": 0.5416564826302289,
      "learning_rate": 2.5982848363770184e-05,
      "loss": 11.7792,
      "step": 28365
    },
    {
      "epoch": 1.5446420670737102,
      "grad_norm": 0.545381216888862,
      "learning_rate": 2.597691909499771e-05,
      "loss": 11.7271,
      "step": 28366
    },
    {
      "epoch": 1.5446965210702932,
      "grad_norm": 0.5237571080501712,
      "learning_rate": 2.5970990401841123e-05,
      "loss": 11.8412,
      "step": 28367
    },
    {
      "epoch": 1.5447509750668762,
      "grad_norm": 0.5145817796750763,
      "learning_rate": 2.5965062284346496e-05,
      "loss": 11.6807,
      "step": 28368
    },
    {
      "epoch": 1.5448054290634592,
      "grad_norm": 0.5881393218023219,
      "learning_rate": 2.5959134742559998e-05,
      "loss": 11.7546,
      "step": 28369
    },
    {
      "epoch": 1.5448598830600422,
      "grad_norm": 0.5752119891535337,
      "learning_rate": 2.5953207776527642e-05,
      "loss": 11.8851,
      "step": 28370
    },
    {
      "epoch": 1.5449143370566254,
      "grad_norm": 0.6303330141537662,
      "learning_rate": 2.594728138629561e-05,
      "loss": 11.8358,
      "step": 28371
    },
    {
      "epoch": 1.5449687910532084,
      "grad_norm": 0.5962712900911351,
      "learning_rate": 2.5941355571909886e-05,
      "loss": 11.7304,
      "step": 28372
    },
    {
      "epoch": 1.5450232450497914,
      "grad_norm": 0.5654189325468316,
      "learning_rate": 2.593543033341661e-05,
      "loss": 11.8266,
      "step": 28373
    },
    {
      "epoch": 1.5450776990463744,
      "grad_norm": 0.5788037406236803,
      "learning_rate": 2.5929505670861875e-05,
      "loss": 11.7691,
      "step": 28374
    },
    {
      "epoch": 1.5451321530429574,
      "grad_norm": 0.5022287286365017,
      "learning_rate": 2.59235815842917e-05,
      "loss": 11.7054,
      "step": 28375
    },
    {
      "epoch": 1.5451866070395404,
      "grad_norm": 0.5976819508719686,
      "learning_rate": 2.591765807375224e-05,
      "loss": 11.8247,
      "step": 28376
    },
    {
      "epoch": 1.5452410610361236,
      "grad_norm": 0.5356904181928719,
      "learning_rate": 2.591173513928943e-05,
      "loss": 11.7389,
      "step": 28377
    },
    {
      "epoch": 1.5452955150327066,
      "grad_norm": 0.5956826321487316,
      "learning_rate": 2.590581278094939e-05,
      "loss": 11.8591,
      "step": 28378
    },
    {
      "epoch": 1.5453499690292896,
      "grad_norm": 0.5405582684955066,
      "learning_rate": 2.589989099877821e-05,
      "loss": 11.7199,
      "step": 28379
    },
    {
      "epoch": 1.5454044230258726,
      "grad_norm": 0.58390632451909,
      "learning_rate": 2.589396979282187e-05,
      "loss": 11.9566,
      "step": 28380
    },
    {
      "epoch": 1.5454588770224555,
      "grad_norm": 0.5505994202984245,
      "learning_rate": 2.588804916312648e-05,
      "loss": 11.766,
      "step": 28381
    },
    {
      "epoch": 1.5455133310190385,
      "grad_norm": 0.6379255049520043,
      "learning_rate": 2.5882129109738018e-05,
      "loss": 11.895,
      "step": 28382
    },
    {
      "epoch": 1.5455677850156215,
      "grad_norm": 0.5558429276491551,
      "learning_rate": 2.5876209632702586e-05,
      "loss": 11.8519,
      "step": 28383
    },
    {
      "epoch": 1.5456222390122045,
      "grad_norm": 0.5415143297445657,
      "learning_rate": 2.587029073206614e-05,
      "loss": 11.8384,
      "step": 28384
    },
    {
      "epoch": 1.5456766930087875,
      "grad_norm": 0.5644130794259087,
      "learning_rate": 2.586437240787475e-05,
      "loss": 11.8423,
      "step": 28385
    },
    {
      "epoch": 1.5457311470053705,
      "grad_norm": 0.5344614842194936,
      "learning_rate": 2.585845466017448e-05,
      "loss": 11.8722,
      "step": 28386
    },
    {
      "epoch": 1.5457856010019535,
      "grad_norm": 0.5564494369049631,
      "learning_rate": 2.585253748901125e-05,
      "loss": 11.8582,
      "step": 28387
    },
    {
      "epoch": 1.5458400549985365,
      "grad_norm": 0.6049125728363719,
      "learning_rate": 2.5846620894431128e-05,
      "loss": 11.7804,
      "step": 28388
    },
    {
      "epoch": 1.5458945089951195,
      "grad_norm": 0.5961697799108411,
      "learning_rate": 2.584070487648015e-05,
      "loss": 11.9494,
      "step": 28389
    },
    {
      "epoch": 1.5459489629917025,
      "grad_norm": 0.5841081208904166,
      "learning_rate": 2.5834789435204243e-05,
      "loss": 11.8902,
      "step": 28390
    },
    {
      "epoch": 1.5460034169882855,
      "grad_norm": 0.5050965529600596,
      "learning_rate": 2.5828874570649486e-05,
      "loss": 11.9289,
      "step": 28391
    },
    {
      "epoch": 1.5460578709848685,
      "grad_norm": 0.5294184189231204,
      "learning_rate": 2.582296028286182e-05,
      "loss": 11.7554,
      "step": 28392
    },
    {
      "epoch": 1.5461123249814515,
      "grad_norm": 0.5135657146092494,
      "learning_rate": 2.581704657188727e-05,
      "loss": 11.7776,
      "step": 28393
    },
    {
      "epoch": 1.5461667789780347,
      "grad_norm": 0.5651805158509523,
      "learning_rate": 2.5811133437771785e-05,
      "loss": 11.8092,
      "step": 28394
    },
    {
      "epoch": 1.5462212329746177,
      "grad_norm": 0.5587154402497274,
      "learning_rate": 2.5805220880561364e-05,
      "loss": 11.8334,
      "step": 28395
    },
    {
      "epoch": 1.5462756869712007,
      "grad_norm": 0.5270724338500411,
      "learning_rate": 2.5799308900302044e-05,
      "loss": 11.7817,
      "step": 28396
    },
    {
      "epoch": 1.5463301409677837,
      "grad_norm": 0.5263739900704169,
      "learning_rate": 2.5793397497039683e-05,
      "loss": 11.7137,
      "step": 28397
    },
    {
      "epoch": 1.5463845949643666,
      "grad_norm": 0.5602872287487679,
      "learning_rate": 2.5787486670820348e-05,
      "loss": 11.9086,
      "step": 28398
    },
    {
      "epoch": 1.5464390489609496,
      "grad_norm": 0.517047938860014,
      "learning_rate": 2.5781576421689923e-05,
      "loss": 11.7993,
      "step": 28399
    },
    {
      "epoch": 1.5464935029575329,
      "grad_norm": 0.5108722857351483,
      "learning_rate": 2.5775666749694406e-05,
      "loss": 11.8566,
      "step": 28400
    },
    {
      "epoch": 1.5465479569541158,
      "grad_norm": 0.6174935293929322,
      "learning_rate": 2.5769757654879777e-05,
      "loss": 11.7961,
      "step": 28401
    },
    {
      "epoch": 1.5466024109506988,
      "grad_norm": 0.5302922625882244,
      "learning_rate": 2.576384913729194e-05,
      "loss": 11.9029,
      "step": 28402
    },
    {
      "epoch": 1.5466568649472818,
      "grad_norm": 0.5783597376976017,
      "learning_rate": 2.5757941196976887e-05,
      "loss": 11.8606,
      "step": 28403
    },
    {
      "epoch": 1.5467113189438648,
      "grad_norm": 0.5795724963024235,
      "learning_rate": 2.5752033833980515e-05,
      "loss": 11.9066,
      "step": 28404
    },
    {
      "epoch": 1.5467657729404478,
      "grad_norm": 0.5695245194364605,
      "learning_rate": 2.57461270483488e-05,
      "loss": 11.9406,
      "step": 28405
    },
    {
      "epoch": 1.5468202269370308,
      "grad_norm": 0.5543568927912289,
      "learning_rate": 2.574022084012765e-05,
      "loss": 11.8522,
      "step": 28406
    },
    {
      "epoch": 1.5468746809336138,
      "grad_norm": 0.5389756714339937,
      "learning_rate": 2.573431520936297e-05,
      "loss": 11.7696,
      "step": 28407
    },
    {
      "epoch": 1.5469291349301968,
      "grad_norm": 0.5439451675427848,
      "learning_rate": 2.5728410156100735e-05,
      "loss": 11.7786,
      "step": 28408
    },
    {
      "epoch": 1.5469835889267798,
      "grad_norm": 0.5777718417749071,
      "learning_rate": 2.572250568038681e-05,
      "loss": 11.8458,
      "step": 28409
    },
    {
      "epoch": 1.5470380429233628,
      "grad_norm": 0.5440717683471554,
      "learning_rate": 2.5716601782267168e-05,
      "loss": 11.7257,
      "step": 28410
    },
    {
      "epoch": 1.5470924969199458,
      "grad_norm": 0.5503235949718415,
      "learning_rate": 2.5710698461787664e-05,
      "loss": 11.875,
      "step": 28411
    },
    {
      "epoch": 1.5471469509165288,
      "grad_norm": 0.5311155260575879,
      "learning_rate": 2.5704795718994225e-05,
      "loss": 11.854,
      "step": 28412
    },
    {
      "epoch": 1.5472014049131118,
      "grad_norm": 0.5580733513631506,
      "learning_rate": 2.5698893553932778e-05,
      "loss": 11.8475,
      "step": 28413
    },
    {
      "epoch": 1.5472558589096947,
      "grad_norm": 0.5629690841127116,
      "learning_rate": 2.569299196664916e-05,
      "loss": 11.8143,
      "step": 28414
    },
    {
      "epoch": 1.5473103129062777,
      "grad_norm": 0.5358089708920208,
      "learning_rate": 2.568709095718934e-05,
      "loss": 11.8512,
      "step": 28415
    },
    {
      "epoch": 1.5473647669028607,
      "grad_norm": 0.5409977297627018,
      "learning_rate": 2.5681190525599163e-05,
      "loss": 11.9244,
      "step": 28416
    },
    {
      "epoch": 1.5474192208994437,
      "grad_norm": 0.638433176437136,
      "learning_rate": 2.5675290671924468e-05,
      "loss": 11.9127,
      "step": 28417
    },
    {
      "epoch": 1.547473674896027,
      "grad_norm": 0.5756050355483072,
      "learning_rate": 2.566939139621122e-05,
      "loss": 11.9142,
      "step": 28418
    },
    {
      "epoch": 1.54752812889261,
      "grad_norm": 0.5431040821555536,
      "learning_rate": 2.566349269850522e-05,
      "loss": 11.7574,
      "step": 28419
    },
    {
      "epoch": 1.547582582889193,
      "grad_norm": 0.5511736086108341,
      "learning_rate": 2.5657594578852386e-05,
      "loss": 11.8797,
      "step": 28420
    },
    {
      "epoch": 1.547637036885776,
      "grad_norm": 0.5323173722084245,
      "learning_rate": 2.565169703729854e-05,
      "loss": 11.8559,
      "step": 28421
    },
    {
      "epoch": 1.547691490882359,
      "grad_norm": 0.5596644594630502,
      "learning_rate": 2.5645800073889568e-05,
      "loss": 11.823,
      "step": 28422
    },
    {
      "epoch": 1.547745944878942,
      "grad_norm": 0.549082073241267,
      "learning_rate": 2.5639903688671352e-05,
      "loss": 11.9428,
      "step": 28423
    },
    {
      "epoch": 1.5478003988755251,
      "grad_norm": 0.5475631173156968,
      "learning_rate": 2.563400788168969e-05,
      "loss": 11.6797,
      "step": 28424
    },
    {
      "epoch": 1.5478548528721081,
      "grad_norm": 0.5149174796601396,
      "learning_rate": 2.5628112652990478e-05,
      "loss": 11.7562,
      "step": 28425
    },
    {
      "epoch": 1.547909306868691,
      "grad_norm": 0.550137163536973,
      "learning_rate": 2.5622218002619537e-05,
      "loss": 11.9227,
      "step": 28426
    },
    {
      "epoch": 1.547963760865274,
      "grad_norm": 0.6155296306761008,
      "learning_rate": 2.5616323930622665e-05,
      "loss": 12.0026,
      "step": 28427
    },
    {
      "epoch": 1.548018214861857,
      "grad_norm": 0.5078000253024684,
      "learning_rate": 2.5610430437045775e-05,
      "loss": 11.7096,
      "step": 28428
    },
    {
      "epoch": 1.54807266885844,
      "grad_norm": 0.5655854998697462,
      "learning_rate": 2.5604537521934614e-05,
      "loss": 11.7819,
      "step": 28429
    },
    {
      "epoch": 1.548127122855023,
      "grad_norm": 0.5253510272861863,
      "learning_rate": 2.559864518533508e-05,
      "loss": 11.7842,
      "step": 28430
    },
    {
      "epoch": 1.548181576851606,
      "grad_norm": 0.5411136871300276,
      "learning_rate": 2.5592753427292927e-05,
      "loss": 11.932,
      "step": 28431
    },
    {
      "epoch": 1.548236030848189,
      "grad_norm": 0.6140542561957026,
      "learning_rate": 2.5586862247854027e-05,
      "loss": 11.8188,
      "step": 28432
    },
    {
      "epoch": 1.548290484844772,
      "grad_norm": 0.5187459708397918,
      "learning_rate": 2.5580971647064134e-05,
      "loss": 11.7711,
      "step": 28433
    },
    {
      "epoch": 1.548344938841355,
      "grad_norm": 0.5237170621281736,
      "learning_rate": 2.5575081624969087e-05,
      "loss": 11.9784,
      "step": 28434
    },
    {
      "epoch": 1.548399392837938,
      "grad_norm": 0.5275190190654206,
      "learning_rate": 2.556919218161472e-05,
      "loss": 11.7082,
      "step": 28435
    },
    {
      "epoch": 1.548453846834521,
      "grad_norm": 0.5589132428169242,
      "learning_rate": 2.5563303317046795e-05,
      "loss": 11.8246,
      "step": 28436
    },
    {
      "epoch": 1.548508300831104,
      "grad_norm": 0.5084913066271427,
      "learning_rate": 2.5557415031311082e-05,
      "loss": 11.862,
      "step": 28437
    },
    {
      "epoch": 1.548562754827687,
      "grad_norm": 0.522068060313236,
      "learning_rate": 2.5551527324453417e-05,
      "loss": 11.7411,
      "step": 28438
    },
    {
      "epoch": 1.54861720882427,
      "grad_norm": 0.5739219829998441,
      "learning_rate": 2.554564019651954e-05,
      "loss": 11.8917,
      "step": 28439
    },
    {
      "epoch": 1.548671662820853,
      "grad_norm": 0.5725865084331427,
      "learning_rate": 2.5539753647555266e-05,
      "loss": 11.9694,
      "step": 28440
    },
    {
      "epoch": 1.5487261168174362,
      "grad_norm": 0.5222121633979407,
      "learning_rate": 2.5533867677606337e-05,
      "loss": 11.7907,
      "step": 28441
    },
    {
      "epoch": 1.5487805708140192,
      "grad_norm": 0.5343862732679546,
      "learning_rate": 2.5527982286718567e-05,
      "loss": 11.9875,
      "step": 28442
    },
    {
      "epoch": 1.5488350248106022,
      "grad_norm": 0.5173530999527631,
      "learning_rate": 2.5522097474937658e-05,
      "loss": 11.8422,
      "step": 28443
    },
    {
      "epoch": 1.5488894788071852,
      "grad_norm": 0.5505206385587831,
      "learning_rate": 2.5516213242309418e-05,
      "loss": 11.8241,
      "step": 28444
    },
    {
      "epoch": 1.5489439328037682,
      "grad_norm": 0.5601652186740096,
      "learning_rate": 2.5510329588879624e-05,
      "loss": 11.8184,
      "step": 28445
    },
    {
      "epoch": 1.5489983868003512,
      "grad_norm": 0.5260787115235537,
      "learning_rate": 2.5504446514693992e-05,
      "loss": 11.9064,
      "step": 28446
    },
    {
      "epoch": 1.5490528407969344,
      "grad_norm": 0.5527946293696209,
      "learning_rate": 2.549856401979829e-05,
      "loss": 11.8156,
      "step": 28447
    },
    {
      "epoch": 1.5491072947935174,
      "grad_norm": 0.5783068020689694,
      "learning_rate": 2.5492682104238207e-05,
      "loss": 11.8586,
      "step": 28448
    },
    {
      "epoch": 1.5491617487901004,
      "grad_norm": 0.5013085575154154,
      "learning_rate": 2.5486800768059527e-05,
      "loss": 11.7434,
      "step": 28449
    },
    {
      "epoch": 1.5492162027866834,
      "grad_norm": 0.4907623554578605,
      "learning_rate": 2.548092001130801e-05,
      "loss": 11.788,
      "step": 28450
    },
    {
      "epoch": 1.5492706567832664,
      "grad_norm": 0.60379728605464,
      "learning_rate": 2.5475039834029324e-05,
      "loss": 11.7984,
      "step": 28451
    },
    {
      "epoch": 1.5493251107798494,
      "grad_norm": 0.5735537390763572,
      "learning_rate": 2.546916023626925e-05,
      "loss": 11.832,
      "step": 28452
    },
    {
      "epoch": 1.5493795647764323,
      "grad_norm": 0.5880665310677062,
      "learning_rate": 2.5463281218073466e-05,
      "loss": 11.8112,
      "step": 28453
    },
    {
      "epoch": 1.5494340187730153,
      "grad_norm": 0.5662966118919047,
      "learning_rate": 2.5457402779487728e-05,
      "loss": 12.0215,
      "step": 28454
    },
    {
      "epoch": 1.5494884727695983,
      "grad_norm": 0.5487430492166419,
      "learning_rate": 2.54515249205577e-05,
      "loss": 11.8194,
      "step": 28455
    },
    {
      "epoch": 1.5495429267661813,
      "grad_norm": 0.6073259437457524,
      "learning_rate": 2.544564764132915e-05,
      "loss": 11.9199,
      "step": 28456
    },
    {
      "epoch": 1.5495973807627643,
      "grad_norm": 0.5110416125123185,
      "learning_rate": 2.543977094184775e-05,
      "loss": 11.9007,
      "step": 28457
    },
    {
      "epoch": 1.5496518347593473,
      "grad_norm": 0.6010475075449443,
      "learning_rate": 2.5433894822159155e-05,
      "loss": 11.7367,
      "step": 28458
    },
    {
      "epoch": 1.5497062887559303,
      "grad_norm": 0.5799295765125994,
      "learning_rate": 2.54280192823091e-05,
      "loss": 11.7556,
      "step": 28459
    },
    {
      "epoch": 1.5497607427525133,
      "grad_norm": 0.5737195536592768,
      "learning_rate": 2.542214432234331e-05,
      "loss": 11.8661,
      "step": 28460
    },
    {
      "epoch": 1.5498151967490963,
      "grad_norm": 0.5060511930665154,
      "learning_rate": 2.5416269942307404e-05,
      "loss": 11.8723,
      "step": 28461
    },
    {
      "epoch": 1.5498696507456793,
      "grad_norm": 0.4857564571140649,
      "learning_rate": 2.5410396142247107e-05,
      "loss": 11.8094,
      "step": 28462
    },
    {
      "epoch": 1.5499241047422623,
      "grad_norm": 0.5395258894578068,
      "learning_rate": 2.5404522922208064e-05,
      "loss": 11.8783,
      "step": 28463
    },
    {
      "epoch": 1.5499785587388455,
      "grad_norm": 0.5520990718572701,
      "learning_rate": 2.539865028223598e-05,
      "loss": 11.8394,
      "step": 28464
    },
    {
      "epoch": 1.5500330127354285,
      "grad_norm": 0.5408865396162762,
      "learning_rate": 2.5392778222376467e-05,
      "loss": 11.9677,
      "step": 28465
    },
    {
      "epoch": 1.5500874667320115,
      "grad_norm": 0.5381429081715795,
      "learning_rate": 2.5386906742675265e-05,
      "loss": 11.8065,
      "step": 28466
    },
    {
      "epoch": 1.5501419207285945,
      "grad_norm": 0.5616159830219871,
      "learning_rate": 2.538103584317798e-05,
      "loss": 11.9628,
      "step": 28467
    },
    {
      "epoch": 1.5501963747251775,
      "grad_norm": 0.5400137642162649,
      "learning_rate": 2.5375165523930256e-05,
      "loss": 11.7801,
      "step": 28468
    },
    {
      "epoch": 1.5502508287217605,
      "grad_norm": 0.5158123918662787,
      "learning_rate": 2.5369295784977776e-05,
      "loss": 11.8542,
      "step": 28469
    },
    {
      "epoch": 1.5503052827183437,
      "grad_norm": 0.5271437661328544,
      "learning_rate": 2.536342662636614e-05,
      "loss": 11.7967,
      "step": 28470
    },
    {
      "epoch": 1.5503597367149267,
      "grad_norm": 0.5534959222742973,
      "learning_rate": 2.5357558048141017e-05,
      "loss": 11.8952,
      "step": 28471
    },
    {
      "epoch": 1.5504141907115097,
      "grad_norm": 0.5482585284101907,
      "learning_rate": 2.5351690050348075e-05,
      "loss": 11.9437,
      "step": 28472
    },
    {
      "epoch": 1.5504686447080926,
      "grad_norm": 0.5579770962441959,
      "learning_rate": 2.534582263303288e-05,
      "loss": 11.8728,
      "step": 28473
    },
    {
      "epoch": 1.5505230987046756,
      "grad_norm": 0.8902138915915079,
      "learning_rate": 2.533995579624111e-05,
      "loss": 11.7841,
      "step": 28474
    },
    {
      "epoch": 1.5505775527012586,
      "grad_norm": 0.5406013223693736,
      "learning_rate": 2.5334089540018336e-05,
      "loss": 11.6726,
      "step": 28475
    },
    {
      "epoch": 1.5506320066978416,
      "grad_norm": 0.6255594525097159,
      "learning_rate": 2.5328223864410227e-05,
      "loss": 11.8823,
      "step": 28476
    },
    {
      "epoch": 1.5506864606944246,
      "grad_norm": 0.5325474589054756,
      "learning_rate": 2.5322358769462383e-05,
      "loss": 11.8107,
      "step": 28477
    },
    {
      "epoch": 1.5507409146910076,
      "grad_norm": 0.5368601359246209,
      "learning_rate": 2.531649425522037e-05,
      "loss": 11.68,
      "step": 28478
    },
    {
      "epoch": 1.5507953686875906,
      "grad_norm": 0.5529466446931488,
      "learning_rate": 2.5310630321729846e-05,
      "loss": 11.8102,
      "step": 28479
    },
    {
      "epoch": 1.5508498226841736,
      "grad_norm": 0.6059228591987602,
      "learning_rate": 2.5304766969036363e-05,
      "loss": 11.9629,
      "step": 28480
    },
    {
      "epoch": 1.5509042766807566,
      "grad_norm": 0.5819827990530834,
      "learning_rate": 2.5298904197185557e-05,
      "loss": 11.8608,
      "step": 28481
    },
    {
      "epoch": 1.5509587306773396,
      "grad_norm": 0.5142102163683073,
      "learning_rate": 2.5293042006222976e-05,
      "loss": 11.8563,
      "step": 28482
    },
    {
      "epoch": 1.5510131846739226,
      "grad_norm": 0.5319599480372926,
      "learning_rate": 2.5287180396194233e-05,
      "loss": 11.6907,
      "step": 28483
    },
    {
      "epoch": 1.5510676386705056,
      "grad_norm": 0.5819049921814107,
      "learning_rate": 2.528131936714493e-05,
      "loss": 11.9163,
      "step": 28484
    },
    {
      "epoch": 1.5511220926670886,
      "grad_norm": 0.5380853151840385,
      "learning_rate": 2.5275458919120598e-05,
      "loss": 11.7327,
      "step": 28485
    },
    {
      "epoch": 1.5511765466636716,
      "grad_norm": 0.5537818658038224,
      "learning_rate": 2.526959905216685e-05,
      "loss": 11.7454,
      "step": 28486
    },
    {
      "epoch": 1.5512310006602545,
      "grad_norm": 0.6021263448844547,
      "learning_rate": 2.526373976632923e-05,
      "loss": 11.9231,
      "step": 28487
    },
    {
      "epoch": 1.5512854546568378,
      "grad_norm": 0.6145604956693037,
      "learning_rate": 2.5257881061653277e-05,
      "loss": 11.9234,
      "step": 28488
    },
    {
      "epoch": 1.5513399086534208,
      "grad_norm": 0.5299598670829708,
      "learning_rate": 2.5252022938184594e-05,
      "loss": 11.8107,
      "step": 28489
    },
    {
      "epoch": 1.5513943626500037,
      "grad_norm": 0.5900073754156276,
      "learning_rate": 2.5246165395968702e-05,
      "loss": 11.843,
      "step": 28490
    },
    {
      "epoch": 1.5514488166465867,
      "grad_norm": 0.587429098817428,
      "learning_rate": 2.5240308435051187e-05,
      "loss": 11.9314,
      "step": 28491
    },
    {
      "epoch": 1.5515032706431697,
      "grad_norm": 0.5925412366155998,
      "learning_rate": 2.523445205547753e-05,
      "loss": 11.9291,
      "step": 28492
    },
    {
      "epoch": 1.551557724639753,
      "grad_norm": 0.4975812315264232,
      "learning_rate": 2.5228596257293325e-05,
      "loss": 11.6459,
      "step": 28493
    },
    {
      "epoch": 1.551612178636336,
      "grad_norm": 0.5454202643079502,
      "learning_rate": 2.522274104054413e-05,
      "loss": 11.6712,
      "step": 28494
    },
    {
      "epoch": 1.551666632632919,
      "grad_norm": 0.6599507266548589,
      "learning_rate": 2.5216886405275396e-05,
      "loss": 11.8713,
      "step": 28495
    },
    {
      "epoch": 1.551721086629502,
      "grad_norm": 0.5091180273869543,
      "learning_rate": 2.5211032351532738e-05,
      "loss": 11.6595,
      "step": 28496
    },
    {
      "epoch": 1.551775540626085,
      "grad_norm": 0.5563702209320073,
      "learning_rate": 2.5205178879361624e-05,
      "loss": 11.8779,
      "step": 28497
    },
    {
      "epoch": 1.551829994622668,
      "grad_norm": 0.5664771627422197,
      "learning_rate": 2.5199325988807553e-05,
      "loss": 11.8834,
      "step": 28498
    },
    {
      "epoch": 1.551884448619251,
      "grad_norm": 0.6076489917821639,
      "learning_rate": 2.5193473679916103e-05,
      "loss": 11.87,
      "step": 28499
    },
    {
      "epoch": 1.551938902615834,
      "grad_norm": 0.5669662502693399,
      "learning_rate": 2.518762195273272e-05,
      "loss": 11.8221,
      "step": 28500
    },
    {
      "epoch": 1.5519933566124169,
      "grad_norm": 0.5707755767582692,
      "learning_rate": 2.5181770807302953e-05,
      "loss": 11.7684,
      "step": 28501
    },
    {
      "epoch": 1.5520478106089999,
      "grad_norm": 0.5518388417723734,
      "learning_rate": 2.517592024367226e-05,
      "loss": 11.8613,
      "step": 28502
    },
    {
      "epoch": 1.5521022646055829,
      "grad_norm": 0.5008864884307273,
      "learning_rate": 2.517007026188619e-05,
      "loss": 11.7114,
      "step": 28503
    },
    {
      "epoch": 1.5521567186021659,
      "grad_norm": 0.4934029122753554,
      "learning_rate": 2.5164220861990174e-05,
      "loss": 11.7556,
      "step": 28504
    },
    {
      "epoch": 1.5522111725987489,
      "grad_norm": 0.5584005164458216,
      "learning_rate": 2.515837204402972e-05,
      "loss": 11.8764,
      "step": 28505
    },
    {
      "epoch": 1.5522656265953318,
      "grad_norm": 0.5847427836113182,
      "learning_rate": 2.5152523808050353e-05,
      "loss": 11.7411,
      "step": 28506
    },
    {
      "epoch": 1.5523200805919148,
      "grad_norm": 0.5587139265149721,
      "learning_rate": 2.5146676154097504e-05,
      "loss": 11.7873,
      "step": 28507
    },
    {
      "epoch": 1.5523745345884978,
      "grad_norm": 0.5743506563430758,
      "learning_rate": 2.5140829082216634e-05,
      "loss": 11.9409,
      "step": 28508
    },
    {
      "epoch": 1.5524289885850808,
      "grad_norm": 0.6032488970331529,
      "learning_rate": 2.5134982592453248e-05,
      "loss": 11.9309,
      "step": 28509
    },
    {
      "epoch": 1.5524834425816638,
      "grad_norm": 0.5852034334633738,
      "learning_rate": 2.5129136684852773e-05,
      "loss": 11.8758,
      "step": 28510
    },
    {
      "epoch": 1.552537896578247,
      "grad_norm": 0.5936254615136347,
      "learning_rate": 2.512329135946072e-05,
      "loss": 11.7461,
      "step": 28511
    },
    {
      "epoch": 1.55259235057483,
      "grad_norm": 0.6038112322972872,
      "learning_rate": 2.511744661632247e-05,
      "loss": 11.9481,
      "step": 28512
    },
    {
      "epoch": 1.552646804571413,
      "grad_norm": 0.57374905226439,
      "learning_rate": 2.5111602455483552e-05,
      "loss": 11.8354,
      "step": 28513
    },
    {
      "epoch": 1.552701258567996,
      "grad_norm": 0.6271858490672713,
      "learning_rate": 2.510575887698934e-05,
      "loss": 11.9047,
      "step": 28514
    },
    {
      "epoch": 1.552755712564579,
      "grad_norm": 0.5534109474902617,
      "learning_rate": 2.5099915880885328e-05,
      "loss": 11.8675,
      "step": 28515
    },
    {
      "epoch": 1.552810166561162,
      "grad_norm": 0.6694132274300042,
      "learning_rate": 2.5094073467216906e-05,
      "loss": 11.8989,
      "step": 28516
    },
    {
      "epoch": 1.5528646205577452,
      "grad_norm": 0.571900046650996,
      "learning_rate": 2.508823163602957e-05,
      "loss": 11.9213,
      "step": 28517
    },
    {
      "epoch": 1.5529190745543282,
      "grad_norm": 0.526527084733199,
      "learning_rate": 2.5082390387368694e-05,
      "loss": 11.8168,
      "step": 28518
    },
    {
      "epoch": 1.5529735285509112,
      "grad_norm": 0.5603474863790836,
      "learning_rate": 2.5076549721279685e-05,
      "loss": 11.9179,
      "step": 28519
    },
    {
      "epoch": 1.5530279825474942,
      "grad_norm": 0.5400150236607307,
      "learning_rate": 2.5070709637807998e-05,
      "loss": 11.7396,
      "step": 28520
    },
    {
      "epoch": 1.5530824365440772,
      "grad_norm": 0.5325796354521942,
      "learning_rate": 2.5064870136999064e-05,
      "loss": 11.785,
      "step": 28521
    },
    {
      "epoch": 1.5531368905406602,
      "grad_norm": 0.5709758062936658,
      "learning_rate": 2.5059031218898243e-05,
      "loss": 11.92,
      "step": 28522
    },
    {
      "epoch": 1.5531913445372432,
      "grad_norm": 0.5419729540075258,
      "learning_rate": 2.505319288355098e-05,
      "loss": 11.8677,
      "step": 28523
    },
    {
      "epoch": 1.5532457985338262,
      "grad_norm": 0.5044308542791872,
      "learning_rate": 2.5047355131002638e-05,
      "loss": 11.9088,
      "step": 28524
    },
    {
      "epoch": 1.5533002525304092,
      "grad_norm": 0.5875763905242495,
      "learning_rate": 2.5041517961298655e-05,
      "loss": 11.8522,
      "step": 28525
    },
    {
      "epoch": 1.5533547065269921,
      "grad_norm": 0.5436673170751153,
      "learning_rate": 2.5035681374484375e-05,
      "loss": 11.6491,
      "step": 28526
    },
    {
      "epoch": 1.5534091605235751,
      "grad_norm": 0.5018654095517917,
      "learning_rate": 2.502984537060524e-05,
      "loss": 11.8711,
      "step": 28527
    },
    {
      "epoch": 1.5534636145201581,
      "grad_norm": 0.5694929635891186,
      "learning_rate": 2.5024009949706596e-05,
      "loss": 11.8014,
      "step": 28528
    },
    {
      "epoch": 1.5535180685167411,
      "grad_norm": 0.555056874138216,
      "learning_rate": 2.5018175111833796e-05,
      "loss": 11.6892,
      "step": 28529
    },
    {
      "epoch": 1.5535725225133241,
      "grad_norm": 0.54490926612928,
      "learning_rate": 2.5012340857032267e-05,
      "loss": 11.819,
      "step": 28530
    },
    {
      "epoch": 1.553626976509907,
      "grad_norm": 0.5463372306875046,
      "learning_rate": 2.5006507185347317e-05,
      "loss": 11.8369,
      "step": 28531
    },
    {
      "epoch": 1.55368143050649,
      "grad_norm": 0.5448397787529085,
      "learning_rate": 2.500067409682435e-05,
      "loss": 11.8249,
      "step": 28532
    },
    {
      "epoch": 1.553735884503073,
      "grad_norm": 0.5675516686988226,
      "learning_rate": 2.4994841591508754e-05,
      "loss": 11.9393,
      "step": 28533
    },
    {
      "epoch": 1.5537903384996563,
      "grad_norm": 0.5267371689428354,
      "learning_rate": 2.4989009669445807e-05,
      "loss": 11.5129,
      "step": 28534
    },
    {
      "epoch": 1.5538447924962393,
      "grad_norm": 0.5425475610079252,
      "learning_rate": 2.4983178330680936e-05,
      "loss": 11.8204,
      "step": 28535
    },
    {
      "epoch": 1.5538992464928223,
      "grad_norm": 0.549845211647986,
      "learning_rate": 2.4977347575259425e-05,
      "loss": 11.8349,
      "step": 28536
    },
    {
      "epoch": 1.5539537004894053,
      "grad_norm": 0.4935824476753059,
      "learning_rate": 2.4971517403226663e-05,
      "loss": 11.867,
      "step": 28537
    },
    {
      "epoch": 1.5540081544859883,
      "grad_norm": 0.6923932588157574,
      "learning_rate": 2.496568781462797e-05,
      "loss": 11.9196,
      "step": 28538
    },
    {
      "epoch": 1.5540626084825713,
      "grad_norm": 0.5260881989908586,
      "learning_rate": 2.495985880950863e-05,
      "loss": 11.773,
      "step": 28539
    },
    {
      "epoch": 1.5541170624791545,
      "grad_norm": 0.5364431472614594,
      "learning_rate": 2.495403038791405e-05,
      "loss": 11.7791,
      "step": 28540
    },
    {
      "epoch": 1.5541715164757375,
      "grad_norm": 0.5812548888784491,
      "learning_rate": 2.4948202549889488e-05,
      "loss": 11.7457,
      "step": 28541
    },
    {
      "epoch": 1.5542259704723205,
      "grad_norm": 0.5919997638028225,
      "learning_rate": 2.494237529548029e-05,
      "loss": 11.9624,
      "step": 28542
    },
    {
      "epoch": 1.5542804244689035,
      "grad_norm": 0.5281734432144186,
      "learning_rate": 2.4936548624731792e-05,
      "loss": 11.9087,
      "step": 28543
    },
    {
      "epoch": 1.5543348784654865,
      "grad_norm": 0.5807438940712529,
      "learning_rate": 2.4930722537689256e-05,
      "loss": 11.803,
      "step": 28544
    },
    {
      "epoch": 1.5543893324620695,
      "grad_norm": 0.5800437925530684,
      "learning_rate": 2.4924897034398033e-05,
      "loss": 11.782,
      "step": 28545
    },
    {
      "epoch": 1.5544437864586524,
      "grad_norm": 0.5268069571755901,
      "learning_rate": 2.4919072114903374e-05,
      "loss": 11.8909,
      "step": 28546
    },
    {
      "epoch": 1.5544982404552354,
      "grad_norm": 0.545833204562664,
      "learning_rate": 2.4913247779250636e-05,
      "loss": 11.8384,
      "step": 28547
    },
    {
      "epoch": 1.5545526944518184,
      "grad_norm": 0.6160011127069638,
      "learning_rate": 2.490742402748507e-05,
      "loss": 11.7852,
      "step": 28548
    },
    {
      "epoch": 1.5546071484484014,
      "grad_norm": 0.5290535096343847,
      "learning_rate": 2.490160085965194e-05,
      "loss": 11.9017,
      "step": 28549
    },
    {
      "epoch": 1.5546616024449844,
      "grad_norm": 0.571680133415217,
      "learning_rate": 2.4895778275796587e-05,
      "loss": 11.8059,
      "step": 28550
    },
    {
      "epoch": 1.5547160564415674,
      "grad_norm": 0.6052844268196166,
      "learning_rate": 2.4889956275964232e-05,
      "loss": 11.8639,
      "step": 28551
    },
    {
      "epoch": 1.5547705104381504,
      "grad_norm": 0.6723872804078066,
      "learning_rate": 2.4884134860200194e-05,
      "loss": 11.8993,
      "step": 28552
    },
    {
      "epoch": 1.5548249644347334,
      "grad_norm": 0.5220939478677491,
      "learning_rate": 2.4878314028549686e-05,
      "loss": 11.6449,
      "step": 28553
    },
    {
      "epoch": 1.5548794184313164,
      "grad_norm": 0.5959196608776058,
      "learning_rate": 2.4872493781058015e-05,
      "loss": 11.8946,
      "step": 28554
    },
    {
      "epoch": 1.5549338724278994,
      "grad_norm": 0.5283774698462669,
      "learning_rate": 2.486667411777046e-05,
      "loss": 11.7955,
      "step": 28555
    },
    {
      "epoch": 1.5549883264244824,
      "grad_norm": 0.5808798509977745,
      "learning_rate": 2.4860855038732223e-05,
      "loss": 11.8956,
      "step": 28556
    },
    {
      "epoch": 1.5550427804210654,
      "grad_norm": 0.5713598747774712,
      "learning_rate": 2.4855036543988598e-05,
      "loss": 11.9146,
      "step": 28557
    },
    {
      "epoch": 1.5550972344176486,
      "grad_norm": 0.5475176573672785,
      "learning_rate": 2.4849218633584814e-05,
      "loss": 11.8442,
      "step": 28558
    },
    {
      "epoch": 1.5551516884142316,
      "grad_norm": 0.5350549055637662,
      "learning_rate": 2.4843401307566082e-05,
      "loss": 11.8207,
      "step": 28559
    },
    {
      "epoch": 1.5552061424108146,
      "grad_norm": 0.704467150637401,
      "learning_rate": 2.483758456597769e-05,
      "loss": 11.8277,
      "step": 28560
    },
    {
      "epoch": 1.5552605964073976,
      "grad_norm": 0.5585654171172186,
      "learning_rate": 2.483176840886482e-05,
      "loss": 11.8686,
      "step": 28561
    },
    {
      "epoch": 1.5553150504039805,
      "grad_norm": 0.6053319547903152,
      "learning_rate": 2.4825952836272747e-05,
      "loss": 11.8209,
      "step": 28562
    },
    {
      "epoch": 1.5553695044005638,
      "grad_norm": 0.58797267331446,
      "learning_rate": 2.4820137848246648e-05,
      "loss": 11.8193,
      "step": 28563
    },
    {
      "epoch": 1.5554239583971468,
      "grad_norm": 0.5585981773514379,
      "learning_rate": 2.4814323444831787e-05,
      "loss": 11.7418,
      "step": 28564
    },
    {
      "epoch": 1.5554784123937297,
      "grad_norm": 0.632299090688014,
      "learning_rate": 2.480850962607334e-05,
      "loss": 11.8569,
      "step": 28565
    },
    {
      "epoch": 1.5555328663903127,
      "grad_norm": 0.5404204742042362,
      "learning_rate": 2.4802696392016522e-05,
      "loss": 11.7614,
      "step": 28566
    },
    {
      "epoch": 1.5555873203868957,
      "grad_norm": 0.5307115052866692,
      "learning_rate": 2.4796883742706613e-05,
      "loss": 11.8431,
      "step": 28567
    },
    {
      "epoch": 1.5556417743834787,
      "grad_norm": 0.5748895070945195,
      "learning_rate": 2.479107167818868e-05,
      "loss": 11.8516,
      "step": 28568
    },
    {
      "epoch": 1.5556962283800617,
      "grad_norm": 0.5185364565502456,
      "learning_rate": 2.4785260198507987e-05,
      "loss": 11.7532,
      "step": 28569
    },
    {
      "epoch": 1.5557506823766447,
      "grad_norm": 0.5824939631145019,
      "learning_rate": 2.4779449303709758e-05,
      "loss": 11.8828,
      "step": 28570
    },
    {
      "epoch": 1.5558051363732277,
      "grad_norm": 0.5823898208825926,
      "learning_rate": 2.477363899383911e-05,
      "loss": 11.8927,
      "step": 28571
    },
    {
      "epoch": 1.5558595903698107,
      "grad_norm": 0.498832121471464,
      "learning_rate": 2.4767829268941302e-05,
      "loss": 11.8058,
      "step": 28572
    },
    {
      "epoch": 1.5559140443663937,
      "grad_norm": 0.5967633998287475,
      "learning_rate": 2.4762020129061427e-05,
      "loss": 11.8112,
      "step": 28573
    },
    {
      "epoch": 1.5559684983629767,
      "grad_norm": 0.5864411539195628,
      "learning_rate": 2.4756211574244735e-05,
      "loss": 11.9025,
      "step": 28574
    },
    {
      "epoch": 1.5560229523595597,
      "grad_norm": 0.5456037698325555,
      "learning_rate": 2.4750403604536333e-05,
      "loss": 11.8419,
      "step": 28575
    },
    {
      "epoch": 1.5560774063561427,
      "grad_norm": 0.5512963607380309,
      "learning_rate": 2.4744596219981418e-05,
      "loss": 11.7866,
      "step": 28576
    },
    {
      "epoch": 1.5561318603527257,
      "grad_norm": 0.5508067364195802,
      "learning_rate": 2.4738789420625195e-05,
      "loss": 11.7468,
      "step": 28577
    },
    {
      "epoch": 1.5561863143493087,
      "grad_norm": 0.6184534697758933,
      "learning_rate": 2.4732983206512706e-05,
      "loss": 11.9362,
      "step": 28578
    },
    {
      "epoch": 1.5562407683458916,
      "grad_norm": 0.5479859137355779,
      "learning_rate": 2.47271775776892e-05,
      "loss": 11.8337,
      "step": 28579
    },
    {
      "epoch": 1.5562952223424746,
      "grad_norm": 0.5082027032151528,
      "learning_rate": 2.472137253419974e-05,
      "loss": 11.7432,
      "step": 28580
    },
    {
      "epoch": 1.5563496763390579,
      "grad_norm": 0.49428153199139857,
      "learning_rate": 2.4715568076089522e-05,
      "loss": 11.7991,
      "step": 28581
    },
    {
      "epoch": 1.5564041303356408,
      "grad_norm": 0.5589649387211867,
      "learning_rate": 2.4709764203403697e-05,
      "loss": 11.8683,
      "step": 28582
    },
    {
      "epoch": 1.5564585843322238,
      "grad_norm": 0.53255396563382,
      "learning_rate": 2.470396091618734e-05,
      "loss": 11.7046,
      "step": 28583
    },
    {
      "epoch": 1.5565130383288068,
      "grad_norm": 0.4977437695350165,
      "learning_rate": 2.4698158214485644e-05,
      "loss": 11.8066,
      "step": 28584
    },
    {
      "epoch": 1.5565674923253898,
      "grad_norm": 0.5626315967222058,
      "learning_rate": 2.4692356098343673e-05,
      "loss": 11.8381,
      "step": 28585
    },
    {
      "epoch": 1.5566219463219728,
      "grad_norm": 0.5286919067641952,
      "learning_rate": 2.4686554567806597e-05,
      "loss": 11.8094,
      "step": 28586
    },
    {
      "epoch": 1.556676400318556,
      "grad_norm": 0.553498917671147,
      "learning_rate": 2.4680753622919496e-05,
      "loss": 11.8322,
      "step": 28587
    },
    {
      "epoch": 1.556730854315139,
      "grad_norm": 0.5602149080639953,
      "learning_rate": 2.4674953263727464e-05,
      "loss": 11.9096,
      "step": 28588
    },
    {
      "epoch": 1.556785308311722,
      "grad_norm": 0.5276623364621802,
      "learning_rate": 2.4669153490275652e-05,
      "loss": 11.7724,
      "step": 28589
    },
    {
      "epoch": 1.556839762308305,
      "grad_norm": 0.49386661278834587,
      "learning_rate": 2.466335430260911e-05,
      "loss": 11.7903,
      "step": 28590
    },
    {
      "epoch": 1.556894216304888,
      "grad_norm": 0.5721983479039223,
      "learning_rate": 2.4657555700772962e-05,
      "loss": 11.7868,
      "step": 28591
    },
    {
      "epoch": 1.556948670301471,
      "grad_norm": 0.5269906038780431,
      "learning_rate": 2.4651757684812317e-05,
      "loss": 11.9258,
      "step": 28592
    },
    {
      "epoch": 1.557003124298054,
      "grad_norm": 0.6267277822180448,
      "learning_rate": 2.4645960254772215e-05,
      "loss": 11.8316,
      "step": 28593
    },
    {
      "epoch": 1.557057578294637,
      "grad_norm": 0.5675082189744377,
      "learning_rate": 2.4640163410697793e-05,
      "loss": 11.654,
      "step": 28594
    },
    {
      "epoch": 1.55711203229122,
      "grad_norm": 0.5715847666698621,
      "learning_rate": 2.4634367152634062e-05,
      "loss": 11.7683,
      "step": 28595
    },
    {
      "epoch": 1.557166486287803,
      "grad_norm": 0.5335686527000096,
      "learning_rate": 2.4628571480626163e-05,
      "loss": 11.8012,
      "step": 28596
    },
    {
      "epoch": 1.557220940284386,
      "grad_norm": 0.5619368982704767,
      "learning_rate": 2.4622776394719137e-05,
      "loss": 11.7206,
      "step": 28597
    },
    {
      "epoch": 1.557275394280969,
      "grad_norm": 0.5735706142849618,
      "learning_rate": 2.4616981894958013e-05,
      "loss": 11.8803,
      "step": 28598
    },
    {
      "epoch": 1.557329848277552,
      "grad_norm": 0.5575511692095365,
      "learning_rate": 2.4611187981387907e-05,
      "loss": 11.717,
      "step": 28599
    },
    {
      "epoch": 1.557384302274135,
      "grad_norm": 0.5332991999274577,
      "learning_rate": 2.4605394654053814e-05,
      "loss": 11.6635,
      "step": 28600
    },
    {
      "epoch": 1.557438756270718,
      "grad_norm": 0.5375788513377285,
      "learning_rate": 2.4599601913000846e-05,
      "loss": 11.8682,
      "step": 28601
    },
    {
      "epoch": 1.557493210267301,
      "grad_norm": 0.5642020985469542,
      "learning_rate": 2.459380975827399e-05,
      "loss": 11.8607,
      "step": 28602
    },
    {
      "epoch": 1.557547664263884,
      "grad_norm": 0.5423063000863227,
      "learning_rate": 2.4588018189918317e-05,
      "loss": 11.8366,
      "step": 28603
    },
    {
      "epoch": 1.5576021182604671,
      "grad_norm": 0.5343092928001854,
      "learning_rate": 2.458222720797888e-05,
      "loss": 11.944,
      "step": 28604
    },
    {
      "epoch": 1.5576565722570501,
      "grad_norm": 0.4843757052977693,
      "learning_rate": 2.457643681250068e-05,
      "loss": 11.9183,
      "step": 28605
    },
    {
      "epoch": 1.5577110262536331,
      "grad_norm": 0.548516745397688,
      "learning_rate": 2.457064700352877e-05,
      "loss": 11.8403,
      "step": 28606
    },
    {
      "epoch": 1.557765480250216,
      "grad_norm": 0.5216081875371917,
      "learning_rate": 2.4564857781108165e-05,
      "loss": 11.7289,
      "step": 28607
    },
    {
      "epoch": 1.557819934246799,
      "grad_norm": 0.508128508668198,
      "learning_rate": 2.4559069145283852e-05,
      "loss": 11.7655,
      "step": 28608
    },
    {
      "epoch": 1.557874388243382,
      "grad_norm": 0.5711385745905865,
      "learning_rate": 2.4553281096100888e-05,
      "loss": 11.7131,
      "step": 28609
    },
    {
      "epoch": 1.5579288422399653,
      "grad_norm": 0.6333255154336631,
      "learning_rate": 2.4547493633604235e-05,
      "loss": 11.7489,
      "step": 28610
    },
    {
      "epoch": 1.5579832962365483,
      "grad_norm": 0.5312599057644889,
      "learning_rate": 2.4541706757838957e-05,
      "loss": 11.7314,
      "step": 28611
    },
    {
      "epoch": 1.5580377502331313,
      "grad_norm": 0.5378413349580727,
      "learning_rate": 2.4535920468849993e-05,
      "loss": 11.8509,
      "step": 28612
    },
    {
      "epoch": 1.5580922042297143,
      "grad_norm": 0.5637246262646303,
      "learning_rate": 2.45301347666824e-05,
      "loss": 11.8517,
      "step": 28613
    },
    {
      "epoch": 1.5581466582262973,
      "grad_norm": 0.5419334294028164,
      "learning_rate": 2.4524349651381095e-05,
      "loss": 11.6795,
      "step": 28614
    },
    {
      "epoch": 1.5582011122228803,
      "grad_norm": 0.6493976749345264,
      "learning_rate": 2.4518565122991112e-05,
      "loss": 11.8643,
      "step": 28615
    },
    {
      "epoch": 1.5582555662194633,
      "grad_norm": 0.5473265317011902,
      "learning_rate": 2.4512781181557455e-05,
      "loss": 11.6634,
      "step": 28616
    },
    {
      "epoch": 1.5583100202160463,
      "grad_norm": 0.5643660364536964,
      "learning_rate": 2.4506997827125067e-05,
      "loss": 11.879,
      "step": 28617
    },
    {
      "epoch": 1.5583644742126292,
      "grad_norm": 0.5333052829511333,
      "learning_rate": 2.4501215059738902e-05,
      "loss": 11.8666,
      "step": 28618
    },
    {
      "epoch": 1.5584189282092122,
      "grad_norm": 0.533843454397603,
      "learning_rate": 2.449543287944397e-05,
      "loss": 11.7767,
      "step": 28619
    },
    {
      "epoch": 1.5584733822057952,
      "grad_norm": 0.5117656452129317,
      "learning_rate": 2.4489651286285186e-05,
      "loss": 11.8108,
      "step": 28620
    },
    {
      "epoch": 1.5585278362023782,
      "grad_norm": 0.5018256139928625,
      "learning_rate": 2.4483870280307576e-05,
      "loss": 11.7305,
      "step": 28621
    },
    {
      "epoch": 1.5585822901989612,
      "grad_norm": 0.5154778160752138,
      "learning_rate": 2.4478089861556007e-05,
      "loss": 11.7754,
      "step": 28622
    },
    {
      "epoch": 1.5586367441955442,
      "grad_norm": 0.5008525101542722,
      "learning_rate": 2.4472310030075518e-05,
      "loss": 11.9024,
      "step": 28623
    },
    {
      "epoch": 1.5586911981921272,
      "grad_norm": 0.5203044454204292,
      "learning_rate": 2.446653078591099e-05,
      "loss": 11.8236,
      "step": 28624
    },
    {
      "epoch": 1.5587456521887102,
      "grad_norm": 0.5297525290030621,
      "learning_rate": 2.4460752129107377e-05,
      "loss": 11.865,
      "step": 28625
    },
    {
      "epoch": 1.5588001061852932,
      "grad_norm": 0.5086130964742646,
      "learning_rate": 2.4454974059709644e-05,
      "loss": 11.7829,
      "step": 28626
    },
    {
      "epoch": 1.5588545601818764,
      "grad_norm": 0.5712787289226434,
      "learning_rate": 2.444919657776269e-05,
      "loss": 11.8218,
      "step": 28627
    },
    {
      "epoch": 1.5589090141784594,
      "grad_norm": 0.5381708369746944,
      "learning_rate": 2.4443419683311496e-05,
      "loss": 11.9287,
      "step": 28628
    },
    {
      "epoch": 1.5589634681750424,
      "grad_norm": 0.5295786602324816,
      "learning_rate": 2.4437643376400888e-05,
      "loss": 11.8455,
      "step": 28629
    },
    {
      "epoch": 1.5590179221716254,
      "grad_norm": 0.5203389601693851,
      "learning_rate": 2.4431867657075846e-05,
      "loss": 11.8296,
      "step": 28630
    },
    {
      "epoch": 1.5590723761682084,
      "grad_norm": 0.6216417078768269,
      "learning_rate": 2.4426092525381294e-05,
      "loss": 11.8956,
      "step": 28631
    },
    {
      "epoch": 1.5591268301647914,
      "grad_norm": 0.5659958596684848,
      "learning_rate": 2.4420317981362094e-05,
      "loss": 11.8099,
      "step": 28632
    },
    {
      "epoch": 1.5591812841613746,
      "grad_norm": 0.5445559846984509,
      "learning_rate": 2.4414544025063202e-05,
      "loss": 11.8889,
      "step": 28633
    },
    {
      "epoch": 1.5592357381579576,
      "grad_norm": 0.5817520990401447,
      "learning_rate": 2.4408770656529468e-05,
      "loss": 11.9012,
      "step": 28634
    },
    {
      "epoch": 1.5592901921545406,
      "grad_norm": 0.5697807847156862,
      "learning_rate": 2.440299787580583e-05,
      "loss": 11.825,
      "step": 28635
    },
    {
      "epoch": 1.5593446461511236,
      "grad_norm": 0.524748583111397,
      "learning_rate": 2.439722568293713e-05,
      "loss": 11.9108,
      "step": 28636
    },
    {
      "epoch": 1.5593991001477066,
      "grad_norm": 0.5784674837697873,
      "learning_rate": 2.439145407796829e-05,
      "loss": 11.7483,
      "step": 28637
    },
    {
      "epoch": 1.5594535541442895,
      "grad_norm": 0.5499970080038136,
      "learning_rate": 2.4385683060944244e-05,
      "loss": 11.6448,
      "step": 28638
    },
    {
      "epoch": 1.5595080081408725,
      "grad_norm": 0.5523570417335587,
      "learning_rate": 2.437991263190974e-05,
      "loss": 11.8614,
      "step": 28639
    },
    {
      "epoch": 1.5595624621374555,
      "grad_norm": 0.5371483363138093,
      "learning_rate": 2.4374142790909716e-05,
      "loss": 11.6866,
      "step": 28640
    },
    {
      "epoch": 1.5596169161340385,
      "grad_norm": 0.5225912993809458,
      "learning_rate": 2.4368373537989065e-05,
      "loss": 11.7559,
      "step": 28641
    },
    {
      "epoch": 1.5596713701306215,
      "grad_norm": 0.5885400975090556,
      "learning_rate": 2.4362604873192608e-05,
      "loss": 11.9619,
      "step": 28642
    },
    {
      "epoch": 1.5597258241272045,
      "grad_norm": 0.6045698927878795,
      "learning_rate": 2.4356836796565242e-05,
      "loss": 11.9076,
      "step": 28643
    },
    {
      "epoch": 1.5597802781237875,
      "grad_norm": 0.5977187037410151,
      "learning_rate": 2.4351069308151775e-05,
      "loss": 11.8175,
      "step": 28644
    },
    {
      "epoch": 1.5598347321203705,
      "grad_norm": 0.629120694430693,
      "learning_rate": 2.4345302407997106e-05,
      "loss": 11.7188,
      "step": 28645
    },
    {
      "epoch": 1.5598891861169535,
      "grad_norm": 0.5351879694628767,
      "learning_rate": 2.433953609614602e-05,
      "loss": 11.8421,
      "step": 28646
    },
    {
      "epoch": 1.5599436401135365,
      "grad_norm": 0.526824232359522,
      "learning_rate": 2.4333770372643382e-05,
      "loss": 11.8083,
      "step": 28647
    },
    {
      "epoch": 1.5599980941101195,
      "grad_norm": 0.5323823091066532,
      "learning_rate": 2.4328005237534112e-05,
      "loss": 11.7918,
      "step": 28648
    },
    {
      "epoch": 1.5600525481067025,
      "grad_norm": 0.5136817842203153,
      "learning_rate": 2.4322240690862896e-05,
      "loss": 11.8293,
      "step": 28649
    },
    {
      "epoch": 1.5601070021032855,
      "grad_norm": 0.6024753448170723,
      "learning_rate": 2.4316476732674654e-05,
      "loss": 11.8917,
      "step": 28650
    },
    {
      "epoch": 1.5601614560998687,
      "grad_norm": 0.6061125754433051,
      "learning_rate": 2.4310713363014147e-05,
      "loss": 11.8552,
      "step": 28651
    },
    {
      "epoch": 1.5602159100964517,
      "grad_norm": 0.5350605569422646,
      "learning_rate": 2.4304950581926246e-05,
      "loss": 11.7366,
      "step": 28652
    },
    {
      "epoch": 1.5602703640930347,
      "grad_norm": 0.5338812927349402,
      "learning_rate": 2.429918838945575e-05,
      "loss": 11.6912,
      "step": 28653
    },
    {
      "epoch": 1.5603248180896176,
      "grad_norm": 0.5487878983649543,
      "learning_rate": 2.4293426785647444e-05,
      "loss": 11.8467,
      "step": 28654
    },
    {
      "epoch": 1.5603792720862006,
      "grad_norm": 0.6095213129678665,
      "learning_rate": 2.4287665770546176e-05,
      "loss": 11.8893,
      "step": 28655
    },
    {
      "epoch": 1.5604337260827836,
      "grad_norm": 0.5485124508683139,
      "learning_rate": 2.4281905344196686e-05,
      "loss": 11.8722,
      "step": 28656
    },
    {
      "epoch": 1.5604881800793668,
      "grad_norm": 0.5454587961981922,
      "learning_rate": 2.4276145506643822e-05,
      "loss": 11.8044,
      "step": 28657
    },
    {
      "epoch": 1.5605426340759498,
      "grad_norm": 0.5391338900803166,
      "learning_rate": 2.4270386257932353e-05,
      "loss": 11.9079,
      "step": 28658
    },
    {
      "epoch": 1.5605970880725328,
      "grad_norm": 0.5377075082659929,
      "learning_rate": 2.4264627598107025e-05,
      "loss": 11.8558,
      "step": 28659
    },
    {
      "epoch": 1.5606515420691158,
      "grad_norm": 0.5359101307237248,
      "learning_rate": 2.425886952721269e-05,
      "loss": 11.8104,
      "step": 28660
    },
    {
      "epoch": 1.5607059960656988,
      "grad_norm": 0.5198996882303529,
      "learning_rate": 2.4253112045294056e-05,
      "loss": 11.7534,
      "step": 28661
    },
    {
      "epoch": 1.5607604500622818,
      "grad_norm": 0.6005535960545674,
      "learning_rate": 2.4247355152395933e-05,
      "loss": 11.753,
      "step": 28662
    },
    {
      "epoch": 1.5608149040588648,
      "grad_norm": 0.5374859344460433,
      "learning_rate": 2.4241598848563095e-05,
      "loss": 11.8552,
      "step": 28663
    },
    {
      "epoch": 1.5608693580554478,
      "grad_norm": 0.5435563746272594,
      "learning_rate": 2.4235843133840276e-05,
      "loss": 11.8205,
      "step": 28664
    },
    {
      "epoch": 1.5609238120520308,
      "grad_norm": 0.514249889739879,
      "learning_rate": 2.4230088008272267e-05,
      "loss": 11.8019,
      "step": 28665
    },
    {
      "epoch": 1.5609782660486138,
      "grad_norm": 0.5519709215934762,
      "learning_rate": 2.422433347190377e-05,
      "loss": 11.8763,
      "step": 28666
    },
    {
      "epoch": 1.5610327200451968,
      "grad_norm": 0.6500519150146741,
      "learning_rate": 2.4218579524779596e-05,
      "loss": 11.8816,
      "step": 28667
    },
    {
      "epoch": 1.5610871740417798,
      "grad_norm": 0.540962738663549,
      "learning_rate": 2.4212826166944445e-05,
      "loss": 11.7951,
      "step": 28668
    },
    {
      "epoch": 1.5611416280383628,
      "grad_norm": 0.5331272631238472,
      "learning_rate": 2.420707339844305e-05,
      "loss": 11.7081,
      "step": 28669
    },
    {
      "epoch": 1.5611960820349458,
      "grad_norm": 0.5311968185544254,
      "learning_rate": 2.4201321219320194e-05,
      "loss": 11.8584,
      "step": 28670
    },
    {
      "epoch": 1.5612505360315287,
      "grad_norm": 0.5329710549646063,
      "learning_rate": 2.419556962962054e-05,
      "loss": 11.7512,
      "step": 28671
    },
    {
      "epoch": 1.5613049900281117,
      "grad_norm": 0.5494203517761478,
      "learning_rate": 2.4189818629388884e-05,
      "loss": 11.9823,
      "step": 28672
    },
    {
      "epoch": 1.5613594440246947,
      "grad_norm": 0.5331030500575514,
      "learning_rate": 2.4184068218669877e-05,
      "loss": 11.7479,
      "step": 28673
    },
    {
      "epoch": 1.561413898021278,
      "grad_norm": 0.565972514856663,
      "learning_rate": 2.417831839750827e-05,
      "loss": 11.8592,
      "step": 28674
    },
    {
      "epoch": 1.561468352017861,
      "grad_norm": 0.5680900215380795,
      "learning_rate": 2.41725691659488e-05,
      "loss": 12.0057,
      "step": 28675
    },
    {
      "epoch": 1.561522806014444,
      "grad_norm": 0.5255724644831455,
      "learning_rate": 2.416682052403613e-05,
      "loss": 11.8153,
      "step": 28676
    },
    {
      "epoch": 1.561577260011027,
      "grad_norm": 0.5279055876700216,
      "learning_rate": 2.4161072471814995e-05,
      "loss": 11.8486,
      "step": 28677
    },
    {
      "epoch": 1.56163171400761,
      "grad_norm": 0.595591677256108,
      "learning_rate": 2.415532500933009e-05,
      "loss": 11.8528,
      "step": 28678
    },
    {
      "epoch": 1.561686168004193,
      "grad_norm": 0.5061505186200804,
      "learning_rate": 2.4149578136626062e-05,
      "loss": 11.7846,
      "step": 28679
    },
    {
      "epoch": 1.5617406220007761,
      "grad_norm": 0.5060773506766636,
      "learning_rate": 2.4143831853747668e-05,
      "loss": 11.8524,
      "step": 28680
    },
    {
      "epoch": 1.5617950759973591,
      "grad_norm": 0.584715774655512,
      "learning_rate": 2.4138086160739525e-05,
      "loss": 11.7046,
      "step": 28681
    },
    {
      "epoch": 1.561849529993942,
      "grad_norm": 0.6205679596133787,
      "learning_rate": 2.413234105764638e-05,
      "loss": 11.8545,
      "step": 28682
    },
    {
      "epoch": 1.561903983990525,
      "grad_norm": 0.558499851873605,
      "learning_rate": 2.4126596544512846e-05,
      "loss": 11.7731,
      "step": 28683
    },
    {
      "epoch": 1.561958437987108,
      "grad_norm": 0.49533448777764894,
      "learning_rate": 2.4120852621383648e-05,
      "loss": 11.9185,
      "step": 28684
    },
    {
      "epoch": 1.562012891983691,
      "grad_norm": 0.5125488555076525,
      "learning_rate": 2.4115109288303395e-05,
      "loss": 11.7386,
      "step": 28685
    },
    {
      "epoch": 1.562067345980274,
      "grad_norm": 0.5737346535227084,
      "learning_rate": 2.410936654531678e-05,
      "loss": 11.8959,
      "step": 28686
    },
    {
      "epoch": 1.562121799976857,
      "grad_norm": 0.5614848846577719,
      "learning_rate": 2.410362439246848e-05,
      "loss": 11.6778,
      "step": 28687
    },
    {
      "epoch": 1.56217625397344,
      "grad_norm": 0.5149510034252667,
      "learning_rate": 2.409788282980313e-05,
      "loss": 11.7197,
      "step": 28688
    },
    {
      "epoch": 1.562230707970023,
      "grad_norm": 0.5971872967791887,
      "learning_rate": 2.4092141857365348e-05,
      "loss": 11.8822,
      "step": 28689
    },
    {
      "epoch": 1.562285161966606,
      "grad_norm": 0.5987782035748999,
      "learning_rate": 2.408640147519983e-05,
      "loss": 11.716,
      "step": 28690
    },
    {
      "epoch": 1.562339615963189,
      "grad_norm": 0.5242083412362638,
      "learning_rate": 2.4080661683351147e-05,
      "loss": 11.7833,
      "step": 28691
    },
    {
      "epoch": 1.562394069959772,
      "grad_norm": 0.6180979334920832,
      "learning_rate": 2.4074922481864005e-05,
      "loss": 11.8354,
      "step": 28692
    },
    {
      "epoch": 1.562448523956355,
      "grad_norm": 0.5467262921957935,
      "learning_rate": 2.4069183870782962e-05,
      "loss": 11.8422,
      "step": 28693
    },
    {
      "epoch": 1.562502977952938,
      "grad_norm": 0.5297791975345009,
      "learning_rate": 2.406344585015271e-05,
      "loss": 11.7195,
      "step": 28694
    },
    {
      "epoch": 1.562557431949521,
      "grad_norm": 0.5166941071554032,
      "learning_rate": 2.405770842001781e-05,
      "loss": 11.7458,
      "step": 28695
    },
    {
      "epoch": 1.562611885946104,
      "grad_norm": 0.5626793895017317,
      "learning_rate": 2.4051971580422917e-05,
      "loss": 11.898,
      "step": 28696
    },
    {
      "epoch": 1.5626663399426872,
      "grad_norm": 0.5460906787316868,
      "learning_rate": 2.4046235331412648e-05,
      "loss": 11.8376,
      "step": 28697
    },
    {
      "epoch": 1.5627207939392702,
      "grad_norm": 0.5257022270095185,
      "learning_rate": 2.404049967303158e-05,
      "loss": 11.8347,
      "step": 28698
    },
    {
      "epoch": 1.5627752479358532,
      "grad_norm": 0.5000313753235595,
      "learning_rate": 2.4034764605324344e-05,
      "loss": 11.8122,
      "step": 28699
    },
    {
      "epoch": 1.5628297019324362,
      "grad_norm": 0.5569986287852403,
      "learning_rate": 2.402903012833547e-05,
      "loss": 11.793,
      "step": 28700
    },
    {
      "epoch": 1.5628841559290192,
      "grad_norm": 0.5336683536513435,
      "learning_rate": 2.4023296242109616e-05,
      "loss": 11.8115,
      "step": 28701
    },
    {
      "epoch": 1.5629386099256022,
      "grad_norm": 0.5475684687673211,
      "learning_rate": 2.4017562946691376e-05,
      "loss": 11.9115,
      "step": 28702
    },
    {
      "epoch": 1.5629930639221854,
      "grad_norm": 0.5337789947071677,
      "learning_rate": 2.4011830242125277e-05,
      "loss": 11.8374,
      "step": 28703
    },
    {
      "epoch": 1.5630475179187684,
      "grad_norm": 0.5904045267095109,
      "learning_rate": 2.4006098128455957e-05,
      "loss": 11.8028,
      "step": 28704
    },
    {
      "epoch": 1.5631019719153514,
      "grad_norm": 0.5083725581979951,
      "learning_rate": 2.400036660572793e-05,
      "loss": 11.8354,
      "step": 28705
    },
    {
      "epoch": 1.5631564259119344,
      "grad_norm": 0.5801147322330663,
      "learning_rate": 2.3994635673985832e-05,
      "loss": 11.8596,
      "step": 28706
    },
    {
      "epoch": 1.5632108799085174,
      "grad_norm": 0.6197128490651064,
      "learning_rate": 2.398890533327416e-05,
      "loss": 11.724,
      "step": 28707
    },
    {
      "epoch": 1.5632653339051004,
      "grad_norm": 0.509327581039951,
      "learning_rate": 2.398317558363754e-05,
      "loss": 11.7433,
      "step": 28708
    },
    {
      "epoch": 1.5633197879016834,
      "grad_norm": 0.5578360607636319,
      "learning_rate": 2.3977446425120498e-05,
      "loss": 11.8905,
      "step": 28709
    },
    {
      "epoch": 1.5633742418982663,
      "grad_norm": 0.568355764591363,
      "learning_rate": 2.397171785776755e-05,
      "loss": 11.8355,
      "step": 28710
    },
    {
      "epoch": 1.5634286958948493,
      "grad_norm": 0.4980275299516433,
      "learning_rate": 2.3965989881623263e-05,
      "loss": 11.7844,
      "step": 28711
    },
    {
      "epoch": 1.5634831498914323,
      "grad_norm": 0.541978878698937,
      "learning_rate": 2.3960262496732234e-05,
      "loss": 11.8814,
      "step": 28712
    },
    {
      "epoch": 1.5635376038880153,
      "grad_norm": 0.501969711216694,
      "learning_rate": 2.3954535703138914e-05,
      "loss": 11.9145,
      "step": 28713
    },
    {
      "epoch": 1.5635920578845983,
      "grad_norm": 0.5280477172839124,
      "learning_rate": 2.3948809500887914e-05,
      "loss": 11.7605,
      "step": 28714
    },
    {
      "epoch": 1.5636465118811813,
      "grad_norm": 0.5200954490724421,
      "learning_rate": 2.39430838900237e-05,
      "loss": 11.7397,
      "step": 28715
    },
    {
      "epoch": 1.5637009658777643,
      "grad_norm": 0.5243192315248342,
      "learning_rate": 2.3937358870590842e-05,
      "loss": 11.9174,
      "step": 28716
    },
    {
      "epoch": 1.5637554198743473,
      "grad_norm": 0.5686220440936848,
      "learning_rate": 2.393163444263381e-05,
      "loss": 11.9139,
      "step": 28717
    },
    {
      "epoch": 1.5638098738709303,
      "grad_norm": 0.5883168712787715,
      "learning_rate": 2.392591060619718e-05,
      "loss": 11.861,
      "step": 28718
    },
    {
      "epoch": 1.5638643278675133,
      "grad_norm": 0.5133332182405562,
      "learning_rate": 2.392018736132542e-05,
      "loss": 11.7445,
      "step": 28719
    },
    {
      "epoch": 1.5639187818640963,
      "grad_norm": 0.5413224039592875,
      "learning_rate": 2.391446470806301e-05,
      "loss": 11.9039,
      "step": 28720
    },
    {
      "epoch": 1.5639732358606795,
      "grad_norm": 0.5726298650546662,
      "learning_rate": 2.3908742646454517e-05,
      "loss": 11.8146,
      "step": 28721
    },
    {
      "epoch": 1.5640276898572625,
      "grad_norm": 0.5707113450541269,
      "learning_rate": 2.3903021176544372e-05,
      "loss": 11.8959,
      "step": 28722
    },
    {
      "epoch": 1.5640821438538455,
      "grad_norm": 0.5102116110964123,
      "learning_rate": 2.389730029837709e-05,
      "loss": 11.8204,
      "step": 28723
    },
    {
      "epoch": 1.5641365978504285,
      "grad_norm": 0.5654603646911585,
      "learning_rate": 2.3891580011997194e-05,
      "loss": 11.7731,
      "step": 28724
    },
    {
      "epoch": 1.5641910518470115,
      "grad_norm": 0.5402573401043785,
      "learning_rate": 2.3885860317449116e-05,
      "loss": 11.7931,
      "step": 28725
    },
    {
      "epoch": 1.5642455058435947,
      "grad_norm": 0.5231421169027981,
      "learning_rate": 2.3880141214777375e-05,
      "loss": 11.8217,
      "step": 28726
    },
    {
      "epoch": 1.5642999598401777,
      "grad_norm": 0.5214264989303993,
      "learning_rate": 2.387442270402639e-05,
      "loss": 11.8492,
      "step": 28727
    },
    {
      "epoch": 1.5643544138367607,
      "grad_norm": 0.5118071365436173,
      "learning_rate": 2.3868704785240693e-05,
      "loss": 11.8548,
      "step": 28728
    },
    {
      "epoch": 1.5644088678333437,
      "grad_norm": 0.5094263838292173,
      "learning_rate": 2.3862987458464704e-05,
      "loss": 11.8949,
      "step": 28729
    },
    {
      "epoch": 1.5644633218299266,
      "grad_norm": 0.5738672324059615,
      "learning_rate": 2.3857270723742874e-05,
      "loss": 11.8904,
      "step": 28730
    },
    {
      "epoch": 1.5645177758265096,
      "grad_norm": 0.5148932033644436,
      "learning_rate": 2.3851554581119708e-05,
      "loss": 11.8232,
      "step": 28731
    },
    {
      "epoch": 1.5645722298230926,
      "grad_norm": 0.5352406901516237,
      "learning_rate": 2.3845839030639582e-05,
      "loss": 11.822,
      "step": 28732
    },
    {
      "epoch": 1.5646266838196756,
      "grad_norm": 0.49418822148510005,
      "learning_rate": 2.3840124072347027e-05,
      "loss": 11.8528,
      "step": 28733
    },
    {
      "epoch": 1.5646811378162586,
      "grad_norm": 0.5375250399413598,
      "learning_rate": 2.3834409706286397e-05,
      "loss": 11.7332,
      "step": 28734
    },
    {
      "epoch": 1.5647355918128416,
      "grad_norm": 0.58188569843298,
      "learning_rate": 2.3828695932502177e-05,
      "loss": 11.8946,
      "step": 28735
    },
    {
      "epoch": 1.5647900458094246,
      "grad_norm": 0.5830529094803893,
      "learning_rate": 2.3822982751038825e-05,
      "loss": 11.894,
      "step": 28736
    },
    {
      "epoch": 1.5648444998060076,
      "grad_norm": 0.5446066079976716,
      "learning_rate": 2.3817270161940696e-05,
      "loss": 11.9443,
      "step": 28737
    },
    {
      "epoch": 1.5648989538025906,
      "grad_norm": 0.5943059947098568,
      "learning_rate": 2.381155816525228e-05,
      "loss": 11.8442,
      "step": 28738
    },
    {
      "epoch": 1.5649534077991736,
      "grad_norm": 0.5221723910675296,
      "learning_rate": 2.380584676101797e-05,
      "loss": 11.7763,
      "step": 28739
    },
    {
      "epoch": 1.5650078617957566,
      "grad_norm": 0.5839712836393786,
      "learning_rate": 2.380013594928214e-05,
      "loss": 11.8905,
      "step": 28740
    },
    {
      "epoch": 1.5650623157923396,
      "grad_norm": 0.5390609129932482,
      "learning_rate": 2.3794425730089263e-05,
      "loss": 11.8606,
      "step": 28741
    },
    {
      "epoch": 1.5651167697889226,
      "grad_norm": 0.507204504875137,
      "learning_rate": 2.3788716103483677e-05,
      "loss": 11.7389,
      "step": 28742
    },
    {
      "epoch": 1.5651712237855055,
      "grad_norm": 0.5677803797688926,
      "learning_rate": 2.3783007069509856e-05,
      "loss": 11.9405,
      "step": 28743
    },
    {
      "epoch": 1.5652256777820888,
      "grad_norm": 0.48342845713468824,
      "learning_rate": 2.377729862821212e-05,
      "loss": 11.7874,
      "step": 28744
    },
    {
      "epoch": 1.5652801317786718,
      "grad_norm": 0.5616391977923335,
      "learning_rate": 2.377159077963489e-05,
      "loss": 11.9251,
      "step": 28745
    },
    {
      "epoch": 1.5653345857752547,
      "grad_norm": 0.5282233862160536,
      "learning_rate": 2.3765883523822575e-05,
      "loss": 11.9397,
      "step": 28746
    },
    {
      "epoch": 1.5653890397718377,
      "grad_norm": 0.5284359597713528,
      "learning_rate": 2.3760176860819517e-05,
      "loss": 11.7287,
      "step": 28747
    },
    {
      "epoch": 1.5654434937684207,
      "grad_norm": 0.5204217061484021,
      "learning_rate": 2.3754470790670158e-05,
      "loss": 11.9262,
      "step": 28748
    },
    {
      "epoch": 1.5654979477650037,
      "grad_norm": 0.5371181748420035,
      "learning_rate": 2.3748765313418765e-05,
      "loss": 11.8768,
      "step": 28749
    },
    {
      "epoch": 1.565552401761587,
      "grad_norm": 0.5137552902370959,
      "learning_rate": 2.3743060429109765e-05,
      "loss": 11.727,
      "step": 28750
    },
    {
      "epoch": 1.56560685575817,
      "grad_norm": 0.5572033761507299,
      "learning_rate": 2.3737356137787548e-05,
      "loss": 11.8423,
      "step": 28751
    },
    {
      "epoch": 1.565661309754753,
      "grad_norm": 0.5129204105216301,
      "learning_rate": 2.3731652439496398e-05,
      "loss": 11.7398,
      "step": 28752
    },
    {
      "epoch": 1.565715763751336,
      "grad_norm": 0.6097920625736223,
      "learning_rate": 2.372594933428074e-05,
      "loss": 11.8881,
      "step": 28753
    },
    {
      "epoch": 1.565770217747919,
      "grad_norm": 0.5645443131079321,
      "learning_rate": 2.3720246822184866e-05,
      "loss": 11.824,
      "step": 28754
    },
    {
      "epoch": 1.565824671744502,
      "grad_norm": 0.5537955200202854,
      "learning_rate": 2.371454490325318e-05,
      "loss": 11.908,
      "step": 28755
    },
    {
      "epoch": 1.565879125741085,
      "grad_norm": 0.5893560843152212,
      "learning_rate": 2.370884357752995e-05,
      "loss": 11.9121,
      "step": 28756
    },
    {
      "epoch": 1.565933579737668,
      "grad_norm": 0.5274576306479433,
      "learning_rate": 2.3703142845059545e-05,
      "loss": 11.8951,
      "step": 28757
    },
    {
      "epoch": 1.5659880337342509,
      "grad_norm": 0.540132917346285,
      "learning_rate": 2.369744270588635e-05,
      "loss": 11.791,
      "step": 28758
    },
    {
      "epoch": 1.5660424877308339,
      "grad_norm": 0.5511282991532422,
      "learning_rate": 2.369174316005459e-05,
      "loss": 11.7187,
      "step": 28759
    },
    {
      "epoch": 1.5660969417274169,
      "grad_norm": 0.5231365561063569,
      "learning_rate": 2.3686044207608626e-05,
      "loss": 11.8112,
      "step": 28760
    },
    {
      "epoch": 1.5661513957239999,
      "grad_norm": 0.48191017846474765,
      "learning_rate": 2.368034584859281e-05,
      "loss": 11.8063,
      "step": 28761
    },
    {
      "epoch": 1.5662058497205829,
      "grad_norm": 0.5128220263647797,
      "learning_rate": 2.3674648083051387e-05,
      "loss": 11.7613,
      "step": 28762
    },
    {
      "epoch": 1.5662603037171658,
      "grad_norm": 0.5198616359674489,
      "learning_rate": 2.3668950911028743e-05,
      "loss": 11.8377,
      "step": 28763
    },
    {
      "epoch": 1.5663147577137488,
      "grad_norm": 0.5331805507713756,
      "learning_rate": 2.3663254332569094e-05,
      "loss": 11.8622,
      "step": 28764
    },
    {
      "epoch": 1.5663692117103318,
      "grad_norm": 0.5312240906252309,
      "learning_rate": 2.3657558347716824e-05,
      "loss": 11.7363,
      "step": 28765
    },
    {
      "epoch": 1.5664236657069148,
      "grad_norm": 0.5669592320093093,
      "learning_rate": 2.3651862956516146e-05,
      "loss": 11.8808,
      "step": 28766
    },
    {
      "epoch": 1.566478119703498,
      "grad_norm": 0.5556726740596933,
      "learning_rate": 2.3646168159011418e-05,
      "loss": 11.6917,
      "step": 28767
    },
    {
      "epoch": 1.566532573700081,
      "grad_norm": 0.6241357849272708,
      "learning_rate": 2.3640473955246856e-05,
      "loss": 11.6563,
      "step": 28768
    },
    {
      "epoch": 1.566587027696664,
      "grad_norm": 0.5384150007817814,
      "learning_rate": 2.3634780345266806e-05,
      "loss": 11.8801,
      "step": 28769
    },
    {
      "epoch": 1.566641481693247,
      "grad_norm": 0.6633948331576193,
      "learning_rate": 2.3629087329115517e-05,
      "loss": 11.907,
      "step": 28770
    },
    {
      "epoch": 1.56669593568983,
      "grad_norm": 0.5392172734467195,
      "learning_rate": 2.3623394906837216e-05,
      "loss": 11.8092,
      "step": 28771
    },
    {
      "epoch": 1.566750389686413,
      "grad_norm": 0.5751667334222652,
      "learning_rate": 2.3617703078476204e-05,
      "loss": 11.7699,
      "step": 28772
    },
    {
      "epoch": 1.5668048436829962,
      "grad_norm": 0.5564432575087046,
      "learning_rate": 2.3612011844076776e-05,
      "loss": 11.7863,
      "step": 28773
    },
    {
      "epoch": 1.5668592976795792,
      "grad_norm": 0.555151398696221,
      "learning_rate": 2.3606321203683125e-05,
      "loss": 11.758,
      "step": 28774
    },
    {
      "epoch": 1.5669137516761622,
      "grad_norm": 0.5823342213344037,
      "learning_rate": 2.360063115733956e-05,
      "loss": 11.786,
      "step": 28775
    },
    {
      "epoch": 1.5669682056727452,
      "grad_norm": 0.5084271925018297,
      "learning_rate": 2.359494170509027e-05,
      "loss": 11.8565,
      "step": 28776
    },
    {
      "epoch": 1.5670226596693282,
      "grad_norm": 0.5624785039688724,
      "learning_rate": 2.3589252846979547e-05,
      "loss": 11.8468,
      "step": 28777
    },
    {
      "epoch": 1.5670771136659112,
      "grad_norm": 0.5853940280131675,
      "learning_rate": 2.3583564583051586e-05,
      "loss": 11.8713,
      "step": 28778
    },
    {
      "epoch": 1.5671315676624942,
      "grad_norm": 0.5133279565725016,
      "learning_rate": 2.357787691335067e-05,
      "loss": 11.7816,
      "step": 28779
    },
    {
      "epoch": 1.5671860216590772,
      "grad_norm": 0.6258315850356592,
      "learning_rate": 2.3572189837920998e-05,
      "loss": 11.7263,
      "step": 28780
    },
    {
      "epoch": 1.5672404756556602,
      "grad_norm": 0.5319551282542647,
      "learning_rate": 2.3566503356806768e-05,
      "loss": 11.7758,
      "step": 28781
    },
    {
      "epoch": 1.5672949296522432,
      "grad_norm": 0.5462647364412913,
      "learning_rate": 2.3560817470052253e-05,
      "loss": 11.8098,
      "step": 28782
    },
    {
      "epoch": 1.5673493836488261,
      "grad_norm": 0.5138441339635628,
      "learning_rate": 2.3555132177701612e-05,
      "loss": 11.8368,
      "step": 28783
    },
    {
      "epoch": 1.5674038376454091,
      "grad_norm": 0.5074516251173836,
      "learning_rate": 2.3549447479799093e-05,
      "loss": 11.8303,
      "step": 28784
    },
    {
      "epoch": 1.5674582916419921,
      "grad_norm": 0.5123121406713322,
      "learning_rate": 2.3543763376388904e-05,
      "loss": 11.8151,
      "step": 28785
    },
    {
      "epoch": 1.5675127456385751,
      "grad_norm": 0.5743749465006065,
      "learning_rate": 2.3538079867515217e-05,
      "loss": 11.8573,
      "step": 28786
    },
    {
      "epoch": 1.5675671996351581,
      "grad_norm": 0.5419491477759231,
      "learning_rate": 2.3532396953222257e-05,
      "loss": 11.8738,
      "step": 28787
    },
    {
      "epoch": 1.567621653631741,
      "grad_norm": 0.5100219492146478,
      "learning_rate": 2.352671463355418e-05,
      "loss": 11.8024,
      "step": 28788
    },
    {
      "epoch": 1.567676107628324,
      "grad_norm": 0.6024295216855875,
      "learning_rate": 2.352103290855523e-05,
      "loss": 11.8999,
      "step": 28789
    },
    {
      "epoch": 1.567730561624907,
      "grad_norm": 0.5244918516681831,
      "learning_rate": 2.3515351778269547e-05,
      "loss": 11.8695,
      "step": 28790
    },
    {
      "epoch": 1.5677850156214903,
      "grad_norm": 0.5251132217855241,
      "learning_rate": 2.3509671242741293e-05,
      "loss": 11.8705,
      "step": 28791
    },
    {
      "epoch": 1.5678394696180733,
      "grad_norm": 0.5617429267671153,
      "learning_rate": 2.3503991302014682e-05,
      "loss": 11.7922,
      "step": 28792
    },
    {
      "epoch": 1.5678939236146563,
      "grad_norm": 0.5013374606752228,
      "learning_rate": 2.3498311956133845e-05,
      "loss": 11.8182,
      "step": 28793
    },
    {
      "epoch": 1.5679483776112393,
      "grad_norm": 0.529311636265472,
      "learning_rate": 2.3492633205142957e-05,
      "loss": 11.8551,
      "step": 28794
    },
    {
      "epoch": 1.5680028316078223,
      "grad_norm": 0.5522262472677143,
      "learning_rate": 2.348695504908621e-05,
      "loss": 11.8119,
      "step": 28795
    },
    {
      "epoch": 1.5680572856044055,
      "grad_norm": 0.5637816975019991,
      "learning_rate": 2.34812774880077e-05,
      "loss": 11.9208,
      "step": 28796
    },
    {
      "epoch": 1.5681117396009885,
      "grad_norm": 0.5124110850407693,
      "learning_rate": 2.3475600521951646e-05,
      "loss": 11.8226,
      "step": 28797
    },
    {
      "epoch": 1.5681661935975715,
      "grad_norm": 0.5061190377644172,
      "learning_rate": 2.346992415096213e-05,
      "loss": 11.7736,
      "step": 28798
    },
    {
      "epoch": 1.5682206475941545,
      "grad_norm": 0.5952185542731735,
      "learning_rate": 2.3464248375083343e-05,
      "loss": 11.9282,
      "step": 28799
    },
    {
      "epoch": 1.5682751015907375,
      "grad_norm": 0.5735068389300485,
      "learning_rate": 2.345857319435939e-05,
      "loss": 11.9166,
      "step": 28800
    },
    {
      "epoch": 1.5683295555873205,
      "grad_norm": 0.5137494152933586,
      "learning_rate": 2.3452898608834385e-05,
      "loss": 11.8012,
      "step": 28801
    },
    {
      "epoch": 1.5683840095839034,
      "grad_norm": 0.5625967930993502,
      "learning_rate": 2.3447224618552498e-05,
      "loss": 11.8578,
      "step": 28802
    },
    {
      "epoch": 1.5684384635804864,
      "grad_norm": 0.5146389222455794,
      "learning_rate": 2.344155122355781e-05,
      "loss": 11.8517,
      "step": 28803
    },
    {
      "epoch": 1.5684929175770694,
      "grad_norm": 0.5264964698827507,
      "learning_rate": 2.3435878423894488e-05,
      "loss": 11.7719,
      "step": 28804
    },
    {
      "epoch": 1.5685473715736524,
      "grad_norm": 0.5464145866372988,
      "learning_rate": 2.3430206219606578e-05,
      "loss": 11.7619,
      "step": 28805
    },
    {
      "epoch": 1.5686018255702354,
      "grad_norm": 0.5585292284540881,
      "learning_rate": 2.3424534610738235e-05,
      "loss": 11.8803,
      "step": 28806
    },
    {
      "epoch": 1.5686562795668184,
      "grad_norm": 0.5819457914709306,
      "learning_rate": 2.3418863597333573e-05,
      "loss": 11.8771,
      "step": 28807
    },
    {
      "epoch": 1.5687107335634014,
      "grad_norm": 0.5685507530431541,
      "learning_rate": 2.3413193179436644e-05,
      "loss": 11.8529,
      "step": 28808
    },
    {
      "epoch": 1.5687651875599844,
      "grad_norm": 0.6031635409779674,
      "learning_rate": 2.34075233570916e-05,
      "loss": 11.8167,
      "step": 28809
    },
    {
      "epoch": 1.5688196415565674,
      "grad_norm": 0.5577260190987443,
      "learning_rate": 2.340185413034249e-05,
      "loss": 11.9123,
      "step": 28810
    },
    {
      "epoch": 1.5688740955531504,
      "grad_norm": 0.5429463809666294,
      "learning_rate": 2.3396185499233393e-05,
      "loss": 11.7829,
      "step": 28811
    },
    {
      "epoch": 1.5689285495497334,
      "grad_norm": 0.5210444587902235,
      "learning_rate": 2.3390517463808427e-05,
      "loss": 11.7514,
      "step": 28812
    },
    {
      "epoch": 1.5689830035463164,
      "grad_norm": 0.562737098318813,
      "learning_rate": 2.3384850024111616e-05,
      "loss": 11.8266,
      "step": 28813
    },
    {
      "epoch": 1.5690374575428996,
      "grad_norm": 0.6035665160900674,
      "learning_rate": 2.337918318018708e-05,
      "loss": 11.8448,
      "step": 28814
    },
    {
      "epoch": 1.5690919115394826,
      "grad_norm": 0.6524794324622821,
      "learning_rate": 2.3373516932078843e-05,
      "loss": 11.9193,
      "step": 28815
    },
    {
      "epoch": 1.5691463655360656,
      "grad_norm": 0.5289140152754468,
      "learning_rate": 2.3367851279830988e-05,
      "loss": 11.5499,
      "step": 28816
    },
    {
      "epoch": 1.5692008195326486,
      "grad_norm": 0.615430679097053,
      "learning_rate": 2.3362186223487603e-05,
      "loss": 11.8119,
      "step": 28817
    },
    {
      "epoch": 1.5692552735292316,
      "grad_norm": 0.5461840473533941,
      "learning_rate": 2.3356521763092687e-05,
      "loss": 11.7733,
      "step": 28818
    },
    {
      "epoch": 1.5693097275258145,
      "grad_norm": 0.5904324219559542,
      "learning_rate": 2.335085789869036e-05,
      "loss": 11.8883,
      "step": 28819
    },
    {
      "epoch": 1.5693641815223978,
      "grad_norm": 0.5416018914944514,
      "learning_rate": 2.3345194630324563e-05,
      "loss": 11.7112,
      "step": 28820
    },
    {
      "epoch": 1.5694186355189808,
      "grad_norm": 0.5592029661413549,
      "learning_rate": 2.3339531958039384e-05,
      "loss": 11.8843,
      "step": 28821
    },
    {
      "epoch": 1.5694730895155637,
      "grad_norm": 0.6197204075331203,
      "learning_rate": 2.333386988187889e-05,
      "loss": 11.918,
      "step": 28822
    },
    {
      "epoch": 1.5695275435121467,
      "grad_norm": 0.5527706793530026,
      "learning_rate": 2.3328208401887054e-05,
      "loss": 11.8334,
      "step": 28823
    },
    {
      "epoch": 1.5695819975087297,
      "grad_norm": 0.5986971658936749,
      "learning_rate": 2.332254751810795e-05,
      "loss": 11.815,
      "step": 28824
    },
    {
      "epoch": 1.5696364515053127,
      "grad_norm": 0.54743412591157,
      "learning_rate": 2.3316887230585548e-05,
      "loss": 11.7309,
      "step": 28825
    },
    {
      "epoch": 1.5696909055018957,
      "grad_norm": 0.5232788361556617,
      "learning_rate": 2.331122753936392e-05,
      "loss": 11.884,
      "step": 28826
    },
    {
      "epoch": 1.5697453594984787,
      "grad_norm": 0.5132268042937693,
      "learning_rate": 2.3305568444487026e-05,
      "loss": 11.7641,
      "step": 28827
    },
    {
      "epoch": 1.5697998134950617,
      "grad_norm": 0.5243162193204141,
      "learning_rate": 2.329990994599889e-05,
      "loss": 11.8505,
      "step": 28828
    },
    {
      "epoch": 1.5698542674916447,
      "grad_norm": 0.5632512571249487,
      "learning_rate": 2.3294252043943565e-05,
      "loss": 11.7132,
      "step": 28829
    },
    {
      "epoch": 1.5699087214882277,
      "grad_norm": 0.558958465573504,
      "learning_rate": 2.3288594738364955e-05,
      "loss": 11.7075,
      "step": 28830
    },
    {
      "epoch": 1.5699631754848107,
      "grad_norm": 0.5638982990361233,
      "learning_rate": 2.328293802930712e-05,
      "loss": 11.916,
      "step": 28831
    },
    {
      "epoch": 1.5700176294813937,
      "grad_norm": 0.5584317871891189,
      "learning_rate": 2.3277281916813998e-05,
      "loss": 11.8094,
      "step": 28832
    },
    {
      "epoch": 1.5700720834779767,
      "grad_norm": 0.534225741168446,
      "learning_rate": 2.3271626400929592e-05,
      "loss": 11.7414,
      "step": 28833
    },
    {
      "epoch": 1.5701265374745597,
      "grad_norm": 0.5439174665096898,
      "learning_rate": 2.3265971481697925e-05,
      "loss": 11.8063,
      "step": 28834
    },
    {
      "epoch": 1.5701809914711427,
      "grad_norm": 0.4844349434575177,
      "learning_rate": 2.3260317159162893e-05,
      "loss": 11.8095,
      "step": 28835
    },
    {
      "epoch": 1.5702354454677256,
      "grad_norm": 0.5011890767572778,
      "learning_rate": 2.325466343336854e-05,
      "loss": 11.7949,
      "step": 28836
    },
    {
      "epoch": 1.5702898994643089,
      "grad_norm": 0.5429282307084502,
      "learning_rate": 2.3249010304358767e-05,
      "loss": 11.8371,
      "step": 28837
    },
    {
      "epoch": 1.5703443534608919,
      "grad_norm": 0.5885668486888262,
      "learning_rate": 2.3243357772177587e-05,
      "loss": 11.8381,
      "step": 28838
    },
    {
      "epoch": 1.5703988074574748,
      "grad_norm": 0.516917885066894,
      "learning_rate": 2.3237705836868928e-05,
      "loss": 11.6619,
      "step": 28839
    },
    {
      "epoch": 1.5704532614540578,
      "grad_norm": 0.5392968565847012,
      "learning_rate": 2.323205449847671e-05,
      "loss": 11.8768,
      "step": 28840
    },
    {
      "epoch": 1.5705077154506408,
      "grad_norm": 0.5249524459624826,
      "learning_rate": 2.3226403757044936e-05,
      "loss": 11.8003,
      "step": 28841
    },
    {
      "epoch": 1.5705621694472238,
      "grad_norm": 0.5771607004204238,
      "learning_rate": 2.3220753612617495e-05,
      "loss": 11.9423,
      "step": 28842
    },
    {
      "epoch": 1.570616623443807,
      "grad_norm": 0.538561955421746,
      "learning_rate": 2.3215104065238357e-05,
      "loss": 11.784,
      "step": 28843
    },
    {
      "epoch": 1.57067107744039,
      "grad_norm": 0.5394213655913459,
      "learning_rate": 2.3209455114951474e-05,
      "loss": 11.8195,
      "step": 28844
    },
    {
      "epoch": 1.570725531436973,
      "grad_norm": 0.5664551279352269,
      "learning_rate": 2.3203806761800717e-05,
      "loss": 11.8103,
      "step": 28845
    },
    {
      "epoch": 1.570779985433556,
      "grad_norm": 0.5625213965904582,
      "learning_rate": 2.3198159005830055e-05,
      "loss": 11.8841,
      "step": 28846
    },
    {
      "epoch": 1.570834439430139,
      "grad_norm": 0.565816915360413,
      "learning_rate": 2.3192511847083375e-05,
      "loss": 11.7987,
      "step": 28847
    },
    {
      "epoch": 1.570888893426722,
      "grad_norm": 0.5448686503596304,
      "learning_rate": 2.318686528560462e-05,
      "loss": 11.6822,
      "step": 28848
    },
    {
      "epoch": 1.570943347423305,
      "grad_norm": 0.5553673942431313,
      "learning_rate": 2.3181219321437687e-05,
      "loss": 11.9259,
      "step": 28849
    },
    {
      "epoch": 1.570997801419888,
      "grad_norm": 0.5950616031474728,
      "learning_rate": 2.3175573954626452e-05,
      "loss": 11.8688,
      "step": 28850
    },
    {
      "epoch": 1.571052255416471,
      "grad_norm": 0.4919833826596297,
      "learning_rate": 2.3169929185214856e-05,
      "loss": 11.7691,
      "step": 28851
    },
    {
      "epoch": 1.571106709413054,
      "grad_norm": 0.5778741528414014,
      "learning_rate": 2.3164285013246755e-05,
      "loss": 11.8501,
      "step": 28852
    },
    {
      "epoch": 1.571161163409637,
      "grad_norm": 0.5534783101305234,
      "learning_rate": 2.3158641438766083e-05,
      "loss": 11.796,
      "step": 28853
    },
    {
      "epoch": 1.57121561740622,
      "grad_norm": 0.569115913555047,
      "learning_rate": 2.315299846181668e-05,
      "loss": 11.9566,
      "step": 28854
    },
    {
      "epoch": 1.571270071402803,
      "grad_norm": 0.5468155802080271,
      "learning_rate": 2.314735608244244e-05,
      "loss": 11.8586,
      "step": 28855
    },
    {
      "epoch": 1.571324525399386,
      "grad_norm": 0.5401501043180141,
      "learning_rate": 2.3141714300687278e-05,
      "loss": 11.8065,
      "step": 28856
    },
    {
      "epoch": 1.571378979395969,
      "grad_norm": 0.4796767885691581,
      "learning_rate": 2.3136073116595003e-05,
      "loss": 11.7506,
      "step": 28857
    },
    {
      "epoch": 1.571433433392552,
      "grad_norm": 0.5759294664512107,
      "learning_rate": 2.3130432530209545e-05,
      "loss": 11.776,
      "step": 28858
    },
    {
      "epoch": 1.571487887389135,
      "grad_norm": 0.5283198939236153,
      "learning_rate": 2.3124792541574737e-05,
      "loss": 11.7619,
      "step": 28859
    },
    {
      "epoch": 1.5715423413857181,
      "grad_norm": 0.5819955804357554,
      "learning_rate": 2.3119153150734397e-05,
      "loss": 11.9188,
      "step": 28860
    },
    {
      "epoch": 1.5715967953823011,
      "grad_norm": 0.5253213142452875,
      "learning_rate": 2.311351435773246e-05,
      "loss": 11.766,
      "step": 28861
    },
    {
      "epoch": 1.5716512493788841,
      "grad_norm": 0.5531674322889195,
      "learning_rate": 2.3107876162612684e-05,
      "loss": 11.8874,
      "step": 28862
    },
    {
      "epoch": 1.5717057033754671,
      "grad_norm": 0.5201859640964867,
      "learning_rate": 2.3102238565419e-05,
      "loss": 11.7146,
      "step": 28863
    },
    {
      "epoch": 1.57176015737205,
      "grad_norm": 0.5856481414476795,
      "learning_rate": 2.3096601566195163e-05,
      "loss": 11.9887,
      "step": 28864
    },
    {
      "epoch": 1.571814611368633,
      "grad_norm": 0.5451818598652349,
      "learning_rate": 2.3090965164985058e-05,
      "loss": 11.7176,
      "step": 28865
    },
    {
      "epoch": 1.5718690653652163,
      "grad_norm": 0.5155272465145542,
      "learning_rate": 2.308532936183253e-05,
      "loss": 11.7139,
      "step": 28866
    },
    {
      "epoch": 1.5719235193617993,
      "grad_norm": 0.5971233770583798,
      "learning_rate": 2.3079694156781352e-05,
      "loss": 11.9034,
      "step": 28867
    },
    {
      "epoch": 1.5719779733583823,
      "grad_norm": 0.5359770605652687,
      "learning_rate": 2.3074059549875393e-05,
      "loss": 11.9564,
      "step": 28868
    },
    {
      "epoch": 1.5720324273549653,
      "grad_norm": 0.6001300294057664,
      "learning_rate": 2.3068425541158456e-05,
      "loss": 11.7861,
      "step": 28869
    },
    {
      "epoch": 1.5720868813515483,
      "grad_norm": 0.49531358904600387,
      "learning_rate": 2.3062792130674314e-05,
      "loss": 11.7003,
      "step": 28870
    },
    {
      "epoch": 1.5721413353481313,
      "grad_norm": 0.588119631962856,
      "learning_rate": 2.3057159318466836e-05,
      "loss": 11.9108,
      "step": 28871
    },
    {
      "epoch": 1.5721957893447143,
      "grad_norm": 0.5981029271313729,
      "learning_rate": 2.305152710457975e-05,
      "loss": 11.7787,
      "step": 28872
    },
    {
      "epoch": 1.5722502433412973,
      "grad_norm": 0.5604200596126128,
      "learning_rate": 2.3045895489056934e-05,
      "loss": 11.8497,
      "step": 28873
    },
    {
      "epoch": 1.5723046973378803,
      "grad_norm": 0.5760121192842886,
      "learning_rate": 2.3040264471942108e-05,
      "loss": 11.7729,
      "step": 28874
    },
    {
      "epoch": 1.5723591513344632,
      "grad_norm": 0.5054891954410827,
      "learning_rate": 2.3034634053279115e-05,
      "loss": 11.691,
      "step": 28875
    },
    {
      "epoch": 1.5724136053310462,
      "grad_norm": 0.6305847167877786,
      "learning_rate": 2.3029004233111694e-05,
      "loss": 11.8834,
      "step": 28876
    },
    {
      "epoch": 1.5724680593276292,
      "grad_norm": 0.5231289623932586,
      "learning_rate": 2.3023375011483638e-05,
      "loss": 11.8407,
      "step": 28877
    },
    {
      "epoch": 1.5725225133242122,
      "grad_norm": 0.5354428442379003,
      "learning_rate": 2.3017746388438753e-05,
      "loss": 11.9022,
      "step": 28878
    },
    {
      "epoch": 1.5725769673207952,
      "grad_norm": 0.529924526298394,
      "learning_rate": 2.3012118364020787e-05,
      "loss": 11.8539,
      "step": 28879
    },
    {
      "epoch": 1.5726314213173782,
      "grad_norm": 0.49937226386890826,
      "learning_rate": 2.3006490938273507e-05,
      "loss": 11.773,
      "step": 28880
    },
    {
      "epoch": 1.5726858753139612,
      "grad_norm": 0.600246857127341,
      "learning_rate": 2.3000864111240627e-05,
      "loss": 11.9778,
      "step": 28881
    },
    {
      "epoch": 1.5727403293105442,
      "grad_norm": 0.5150061147575434,
      "learning_rate": 2.2995237882965936e-05,
      "loss": 11.7229,
      "step": 28882
    },
    {
      "epoch": 1.5727947833071272,
      "grad_norm": 0.5389269690688803,
      "learning_rate": 2.298961225349322e-05,
      "loss": 11.7591,
      "step": 28883
    },
    {
      "epoch": 1.5728492373037104,
      "grad_norm": 0.5496834218271008,
      "learning_rate": 2.2983987222866176e-05,
      "loss": 11.8051,
      "step": 28884
    },
    {
      "epoch": 1.5729036913002934,
      "grad_norm": 0.5408526547759474,
      "learning_rate": 2.2978362791128582e-05,
      "loss": 11.8004,
      "step": 28885
    },
    {
      "epoch": 1.5729581452968764,
      "grad_norm": 0.48644510699505733,
      "learning_rate": 2.2972738958324123e-05,
      "loss": 11.7983,
      "step": 28886
    },
    {
      "epoch": 1.5730125992934594,
      "grad_norm": 0.5125963878475169,
      "learning_rate": 2.29671157244966e-05,
      "loss": 11.7825,
      "step": 28887
    },
    {
      "epoch": 1.5730670532900424,
      "grad_norm": 0.5456381862501376,
      "learning_rate": 2.2961493089689677e-05,
      "loss": 11.8283,
      "step": 28888
    },
    {
      "epoch": 1.5731215072866254,
      "grad_norm": 0.5832132101324694,
      "learning_rate": 2.295587105394713e-05,
      "loss": 11.9061,
      "step": 28889
    },
    {
      "epoch": 1.5731759612832086,
      "grad_norm": 0.5117851731701568,
      "learning_rate": 2.2950249617312648e-05,
      "loss": 11.8274,
      "step": 28890
    },
    {
      "epoch": 1.5732304152797916,
      "grad_norm": 0.7988447022716587,
      "learning_rate": 2.2944628779829913e-05,
      "loss": 12.0,
      "step": 28891
    },
    {
      "epoch": 1.5732848692763746,
      "grad_norm": 0.5515940665282327,
      "learning_rate": 2.293900854154267e-05,
      "loss": 11.8128,
      "step": 28892
    },
    {
      "epoch": 1.5733393232729576,
      "grad_norm": 0.5458483583398712,
      "learning_rate": 2.2933388902494646e-05,
      "loss": 11.9411,
      "step": 28893
    },
    {
      "epoch": 1.5733937772695405,
      "grad_norm": 0.5000229674779554,
      "learning_rate": 2.2927769862729497e-05,
      "loss": 11.8226,
      "step": 28894
    },
    {
      "epoch": 1.5734482312661235,
      "grad_norm": 0.5421644200612199,
      "learning_rate": 2.292215142229095e-05,
      "loss": 11.9527,
      "step": 28895
    },
    {
      "epoch": 1.5735026852627065,
      "grad_norm": 0.48402343152433586,
      "learning_rate": 2.2916533581222655e-05,
      "loss": 11.7499,
      "step": 28896
    },
    {
      "epoch": 1.5735571392592895,
      "grad_norm": 0.5165782833986247,
      "learning_rate": 2.2910916339568354e-05,
      "loss": 11.8198,
      "step": 28897
    },
    {
      "epoch": 1.5736115932558725,
      "grad_norm": 0.5627621105234702,
      "learning_rate": 2.290529969737166e-05,
      "loss": 11.7158,
      "step": 28898
    },
    {
      "epoch": 1.5736660472524555,
      "grad_norm": 0.5331697384083023,
      "learning_rate": 2.289968365467632e-05,
      "loss": 11.7072,
      "step": 28899
    },
    {
      "epoch": 1.5737205012490385,
      "grad_norm": 0.5808663226201757,
      "learning_rate": 2.2894068211525964e-05,
      "loss": 11.8476,
      "step": 28900
    },
    {
      "epoch": 1.5737749552456215,
      "grad_norm": 0.5411353924438677,
      "learning_rate": 2.2888453367964235e-05,
      "loss": 11.8446,
      "step": 28901
    },
    {
      "epoch": 1.5738294092422045,
      "grad_norm": 0.5439941125721316,
      "learning_rate": 2.288283912403486e-05,
      "loss": 11.8777,
      "step": 28902
    },
    {
      "epoch": 1.5738838632387875,
      "grad_norm": 0.5354142726054871,
      "learning_rate": 2.2877225479781427e-05,
      "loss": 11.8446,
      "step": 28903
    },
    {
      "epoch": 1.5739383172353705,
      "grad_norm": 0.5238275096977838,
      "learning_rate": 2.2871612435247625e-05,
      "loss": 11.8189,
      "step": 28904
    },
    {
      "epoch": 1.5739927712319535,
      "grad_norm": 0.5314271423713879,
      "learning_rate": 2.2865999990477128e-05,
      "loss": 11.8378,
      "step": 28905
    },
    {
      "epoch": 1.5740472252285365,
      "grad_norm": 0.5406769231803211,
      "learning_rate": 2.286038814551351e-05,
      "loss": 11.7432,
      "step": 28906
    },
    {
      "epoch": 1.5741016792251197,
      "grad_norm": 0.5294794764373592,
      "learning_rate": 2.28547769004005e-05,
      "loss": 11.6909,
      "step": 28907
    },
    {
      "epoch": 1.5741561332217027,
      "grad_norm": 0.5633527580065322,
      "learning_rate": 2.284916625518163e-05,
      "loss": 11.7931,
      "step": 28908
    },
    {
      "epoch": 1.5742105872182857,
      "grad_norm": 0.5253651236233929,
      "learning_rate": 2.2843556209900628e-05,
      "loss": 11.8092,
      "step": 28909
    },
    {
      "epoch": 1.5742650412148687,
      "grad_norm": 0.537188452610232,
      "learning_rate": 2.283794676460107e-05,
      "loss": 11.9092,
      "step": 28910
    },
    {
      "epoch": 1.5743194952114516,
      "grad_norm": 0.5943887836728728,
      "learning_rate": 2.283233791932654e-05,
      "loss": 11.9433,
      "step": 28911
    },
    {
      "epoch": 1.5743739492080346,
      "grad_norm": 0.5744023067687126,
      "learning_rate": 2.2826729674120728e-05,
      "loss": 11.8958,
      "step": 28912
    },
    {
      "epoch": 1.5744284032046179,
      "grad_norm": 0.5141809582026872,
      "learning_rate": 2.2821122029027176e-05,
      "loss": 11.8696,
      "step": 28913
    },
    {
      "epoch": 1.5744828572012008,
      "grad_norm": 0.5748075044435534,
      "learning_rate": 2.2815514984089526e-05,
      "loss": 11.858,
      "step": 28914
    },
    {
      "epoch": 1.5745373111977838,
      "grad_norm": 0.560832134890702,
      "learning_rate": 2.28099085393514e-05,
      "loss": 11.8423,
      "step": 28915
    },
    {
      "epoch": 1.5745917651943668,
      "grad_norm": 0.5371437284242473,
      "learning_rate": 2.280430269485634e-05,
      "loss": 11.9207,
      "step": 28916
    },
    {
      "epoch": 1.5746462191909498,
      "grad_norm": 0.53380184172574,
      "learning_rate": 2.2798697450648e-05,
      "loss": 11.8518,
      "step": 28917
    },
    {
      "epoch": 1.5747006731875328,
      "grad_norm": 0.540136552633758,
      "learning_rate": 2.27930928067699e-05,
      "loss": 11.8117,
      "step": 28918
    },
    {
      "epoch": 1.5747551271841158,
      "grad_norm": 0.5279692252142438,
      "learning_rate": 2.2787488763265697e-05,
      "loss": 11.828,
      "step": 28919
    },
    {
      "epoch": 1.5748095811806988,
      "grad_norm": 0.5667267210614443,
      "learning_rate": 2.278188532017892e-05,
      "loss": 11.7413,
      "step": 28920
    },
    {
      "epoch": 1.5748640351772818,
      "grad_norm": 0.5641548462221883,
      "learning_rate": 2.2776282477553125e-05,
      "loss": 11.8776,
      "step": 28921
    },
    {
      "epoch": 1.5749184891738648,
      "grad_norm": 0.5165989644735156,
      "learning_rate": 2.277068023543194e-05,
      "loss": 11.7896,
      "step": 28922
    },
    {
      "epoch": 1.5749729431704478,
      "grad_norm": 0.5499267654843936,
      "learning_rate": 2.2765078593858868e-05,
      "loss": 11.8233,
      "step": 28923
    },
    {
      "epoch": 1.5750273971670308,
      "grad_norm": 0.5689880773549624,
      "learning_rate": 2.275947755287753e-05,
      "loss": 11.8807,
      "step": 28924
    },
    {
      "epoch": 1.5750818511636138,
      "grad_norm": 0.5704698987243558,
      "learning_rate": 2.275387711253142e-05,
      "loss": 11.9132,
      "step": 28925
    },
    {
      "epoch": 1.5751363051601968,
      "grad_norm": 0.5903502315610777,
      "learning_rate": 2.2748277272864106e-05,
      "loss": 11.8719,
      "step": 28926
    },
    {
      "epoch": 1.5751907591567798,
      "grad_norm": 0.5582185909323738,
      "learning_rate": 2.2742678033919175e-05,
      "loss": 11.83,
      "step": 28927
    },
    {
      "epoch": 1.5752452131533627,
      "grad_norm": 0.5236327125435509,
      "learning_rate": 2.2737079395740114e-05,
      "loss": 11.8483,
      "step": 28928
    },
    {
      "epoch": 1.5752996671499457,
      "grad_norm": 0.517888638304181,
      "learning_rate": 2.2731481358370498e-05,
      "loss": 11.7218,
      "step": 28929
    },
    {
      "epoch": 1.575354121146529,
      "grad_norm": 0.5806616262289762,
      "learning_rate": 2.272588392185384e-05,
      "loss": 11.8007,
      "step": 28930
    },
    {
      "epoch": 1.575408575143112,
      "grad_norm": 0.5388379469177362,
      "learning_rate": 2.272028708623365e-05,
      "loss": 11.7921,
      "step": 28931
    },
    {
      "epoch": 1.575463029139695,
      "grad_norm": 0.5361134179981693,
      "learning_rate": 2.2714690851553488e-05,
      "loss": 11.7761,
      "step": 28932
    },
    {
      "epoch": 1.575517483136278,
      "grad_norm": 0.5423441461487326,
      "learning_rate": 2.2709095217856836e-05,
      "loss": 11.8017,
      "step": 28933
    },
    {
      "epoch": 1.575571937132861,
      "grad_norm": 0.5245884311961633,
      "learning_rate": 2.270350018518723e-05,
      "loss": 11.7721,
      "step": 28934
    },
    {
      "epoch": 1.575626391129444,
      "grad_norm": 0.5602690809245169,
      "learning_rate": 2.2697905753588156e-05,
      "loss": 11.8095,
      "step": 28935
    },
    {
      "epoch": 1.5756808451260271,
      "grad_norm": 0.6439708276703576,
      "learning_rate": 2.269231192310315e-05,
      "loss": 11.8879,
      "step": 28936
    },
    {
      "epoch": 1.5757352991226101,
      "grad_norm": 0.6756873643787613,
      "learning_rate": 2.2686718693775664e-05,
      "loss": 11.8543,
      "step": 28937
    },
    {
      "epoch": 1.5757897531191931,
      "grad_norm": 0.5729403690540208,
      "learning_rate": 2.2681126065649227e-05,
      "loss": 11.9723,
      "step": 28938
    },
    {
      "epoch": 1.575844207115776,
      "grad_norm": 0.5200836804350855,
      "learning_rate": 2.2675534038767342e-05,
      "loss": 11.8063,
      "step": 28939
    },
    {
      "epoch": 1.575898661112359,
      "grad_norm": 0.5864393205334614,
      "learning_rate": 2.2669942613173465e-05,
      "loss": 11.9725,
      "step": 28940
    },
    {
      "epoch": 1.575953115108942,
      "grad_norm": 0.5406517164507786,
      "learning_rate": 2.2664351788911063e-05,
      "loss": 11.8057,
      "step": 28941
    },
    {
      "epoch": 1.576007569105525,
      "grad_norm": 0.5329268958336405,
      "learning_rate": 2.2658761566023655e-05,
      "loss": 11.7887,
      "step": 28942
    },
    {
      "epoch": 1.576062023102108,
      "grad_norm": 0.5332785095692129,
      "learning_rate": 2.2653171944554662e-05,
      "loss": 11.8428,
      "step": 28943
    },
    {
      "epoch": 1.576116477098691,
      "grad_norm": 0.5469093885972705,
      "learning_rate": 2.2647582924547606e-05,
      "loss": 11.8983,
      "step": 28944
    },
    {
      "epoch": 1.576170931095274,
      "grad_norm": 0.5199717403612628,
      "learning_rate": 2.2641994506045883e-05,
      "loss": 11.8231,
      "step": 28945
    },
    {
      "epoch": 1.576225385091857,
      "grad_norm": 0.5879884774777993,
      "learning_rate": 2.2636406689093014e-05,
      "loss": 11.8622,
      "step": 28946
    },
    {
      "epoch": 1.57627983908844,
      "grad_norm": 0.5686514620226485,
      "learning_rate": 2.263081947373239e-05,
      "loss": 11.649,
      "step": 28947
    },
    {
      "epoch": 1.576334293085023,
      "grad_norm": 0.507483113874304,
      "learning_rate": 2.2625232860007495e-05,
      "loss": 11.8155,
      "step": 28948
    },
    {
      "epoch": 1.576388747081606,
      "grad_norm": 0.6117196971041503,
      "learning_rate": 2.2619646847961785e-05,
      "loss": 11.8837,
      "step": 28949
    },
    {
      "epoch": 1.576443201078189,
      "grad_norm": 0.5720458429327793,
      "learning_rate": 2.261406143763868e-05,
      "loss": 11.8197,
      "step": 28950
    },
    {
      "epoch": 1.576497655074772,
      "grad_norm": 0.5534445260877011,
      "learning_rate": 2.260847662908161e-05,
      "loss": 11.8195,
      "step": 28951
    },
    {
      "epoch": 1.576552109071355,
      "grad_norm": 0.5583281817128195,
      "learning_rate": 2.2602892422333976e-05,
      "loss": 11.6677,
      "step": 28952
    },
    {
      "epoch": 1.576606563067938,
      "grad_norm": 0.531915341939756,
      "learning_rate": 2.2597308817439232e-05,
      "loss": 11.7879,
      "step": 28953
    },
    {
      "epoch": 1.5766610170645212,
      "grad_norm": 0.5098542220047537,
      "learning_rate": 2.2591725814440835e-05,
      "loss": 11.7829,
      "step": 28954
    },
    {
      "epoch": 1.5767154710611042,
      "grad_norm": 0.4992514116282411,
      "learning_rate": 2.258614341338211e-05,
      "loss": 11.7571,
      "step": 28955
    },
    {
      "epoch": 1.5767699250576872,
      "grad_norm": 0.5449980587058868,
      "learning_rate": 2.258056161430656e-05,
      "loss": 11.9823,
      "step": 28956
    },
    {
      "epoch": 1.5768243790542702,
      "grad_norm": 0.6281234219372464,
      "learning_rate": 2.257498041725751e-05,
      "loss": 11.8211,
      "step": 28957
    },
    {
      "epoch": 1.5768788330508532,
      "grad_norm": 0.5436626228492178,
      "learning_rate": 2.2569399822278425e-05,
      "loss": 11.8588,
      "step": 28958
    },
    {
      "epoch": 1.5769332870474362,
      "grad_norm": 0.631549265387338,
      "learning_rate": 2.2563819829412647e-05,
      "loss": 11.8126,
      "step": 28959
    },
    {
      "epoch": 1.5769877410440194,
      "grad_norm": 0.5676873706665952,
      "learning_rate": 2.2558240438703625e-05,
      "loss": 11.7445,
      "step": 28960
    },
    {
      "epoch": 1.5770421950406024,
      "grad_norm": 0.5769554586685568,
      "learning_rate": 2.2552661650194706e-05,
      "loss": 11.7879,
      "step": 28961
    },
    {
      "epoch": 1.5770966490371854,
      "grad_norm": 0.5362000100321781,
      "learning_rate": 2.2547083463929242e-05,
      "loss": 11.8459,
      "step": 28962
    },
    {
      "epoch": 1.5771511030337684,
      "grad_norm": 0.5134673533249084,
      "learning_rate": 2.2541505879950665e-05,
      "loss": 11.7598,
      "step": 28963
    },
    {
      "epoch": 1.5772055570303514,
      "grad_norm": 0.5357515394780799,
      "learning_rate": 2.253592889830234e-05,
      "loss": 11.9005,
      "step": 28964
    },
    {
      "epoch": 1.5772600110269344,
      "grad_norm": 0.524983362575471,
      "learning_rate": 2.2530352519027596e-05,
      "loss": 11.8162,
      "step": 28965
    },
    {
      "epoch": 1.5773144650235174,
      "grad_norm": 0.6154698033924738,
      "learning_rate": 2.252477674216986e-05,
      "loss": 11.8156,
      "step": 28966
    },
    {
      "epoch": 1.5773689190201003,
      "grad_norm": 0.5504062250391509,
      "learning_rate": 2.2519201567772418e-05,
      "loss": 11.6462,
      "step": 28967
    },
    {
      "epoch": 1.5774233730166833,
      "grad_norm": 0.5430983524423318,
      "learning_rate": 2.2513626995878688e-05,
      "loss": 11.7889,
      "step": 28968
    },
    {
      "epoch": 1.5774778270132663,
      "grad_norm": 0.5188498487490496,
      "learning_rate": 2.2508053026531962e-05,
      "loss": 11.7778,
      "step": 28969
    },
    {
      "epoch": 1.5775322810098493,
      "grad_norm": 0.5121725059802743,
      "learning_rate": 2.2502479659775632e-05,
      "loss": 11.866,
      "step": 28970
    },
    {
      "epoch": 1.5775867350064323,
      "grad_norm": 0.5290015315785176,
      "learning_rate": 2.2496906895653026e-05,
      "loss": 11.7747,
      "step": 28971
    },
    {
      "epoch": 1.5776411890030153,
      "grad_norm": 0.6569043931659422,
      "learning_rate": 2.2491334734207436e-05,
      "loss": 11.9028,
      "step": 28972
    },
    {
      "epoch": 1.5776956429995983,
      "grad_norm": 0.5543234105763295,
      "learning_rate": 2.2485763175482255e-05,
      "loss": 11.7604,
      "step": 28973
    },
    {
      "epoch": 1.5777500969961813,
      "grad_norm": 0.49838727870874877,
      "learning_rate": 2.2480192219520745e-05,
      "loss": 11.7692,
      "step": 28974
    },
    {
      "epoch": 1.5778045509927643,
      "grad_norm": 0.5135857097967986,
      "learning_rate": 2.2474621866366265e-05,
      "loss": 11.746,
      "step": 28975
    },
    {
      "epoch": 1.5778590049893473,
      "grad_norm": 0.5333201044484813,
      "learning_rate": 2.2469052116062162e-05,
      "loss": 11.7192,
      "step": 28976
    },
    {
      "epoch": 1.5779134589859305,
      "grad_norm": 0.529647847874188,
      "learning_rate": 2.2463482968651673e-05,
      "loss": 11.693,
      "step": 28977
    },
    {
      "epoch": 1.5779679129825135,
      "grad_norm": 0.5440209011817407,
      "learning_rate": 2.245791442417817e-05,
      "loss": 11.7874,
      "step": 28978
    },
    {
      "epoch": 1.5780223669790965,
      "grad_norm": 0.6376308003114978,
      "learning_rate": 2.2452346482684904e-05,
      "loss": 11.7973,
      "step": 28979
    },
    {
      "epoch": 1.5780768209756795,
      "grad_norm": 0.5519943422912124,
      "learning_rate": 2.2446779144215226e-05,
      "loss": 11.8261,
      "step": 28980
    },
    {
      "epoch": 1.5781312749722625,
      "grad_norm": 0.5295756531059973,
      "learning_rate": 2.2441212408812406e-05,
      "loss": 11.8154,
      "step": 28981
    },
    {
      "epoch": 1.5781857289688455,
      "grad_norm": 0.6778275703681259,
      "learning_rate": 2.2435646276519684e-05,
      "loss": 11.8203,
      "step": 28982
    },
    {
      "epoch": 1.5782401829654287,
      "grad_norm": 0.6029728680115989,
      "learning_rate": 2.243008074738042e-05,
      "loss": 11.6193,
      "step": 28983
    },
    {
      "epoch": 1.5782946369620117,
      "grad_norm": 0.5648586442144191,
      "learning_rate": 2.242451582143782e-05,
      "loss": 11.8152,
      "step": 28984
    },
    {
      "epoch": 1.5783490909585947,
      "grad_norm": 0.5750749028367127,
      "learning_rate": 2.241895149873523e-05,
      "loss": 11.8914,
      "step": 28985
    },
    {
      "epoch": 1.5784035449551777,
      "grad_norm": 0.5815848033836704,
      "learning_rate": 2.2413387779315854e-05,
      "loss": 11.8193,
      "step": 28986
    },
    {
      "epoch": 1.5784579989517606,
      "grad_norm": 0.5911850440102399,
      "learning_rate": 2.2407824663222988e-05,
      "loss": 11.9349,
      "step": 28987
    },
    {
      "epoch": 1.5785124529483436,
      "grad_norm": 0.5667578798371958,
      "learning_rate": 2.240226215049992e-05,
      "loss": 11.8732,
      "step": 28988
    },
    {
      "epoch": 1.5785669069449266,
      "grad_norm": 0.5261334268084475,
      "learning_rate": 2.239670024118984e-05,
      "loss": 11.7511,
      "step": 28989
    },
    {
      "epoch": 1.5786213609415096,
      "grad_norm": 0.5301617304382541,
      "learning_rate": 2.2391138935336065e-05,
      "loss": 11.9436,
      "step": 28990
    },
    {
      "epoch": 1.5786758149380926,
      "grad_norm": 0.6135552732462849,
      "learning_rate": 2.2385578232981807e-05,
      "loss": 12.0065,
      "step": 28991
    },
    {
      "epoch": 1.5787302689346756,
      "grad_norm": 0.5642411080239568,
      "learning_rate": 2.2380018134170278e-05,
      "loss": 11.8878,
      "step": 28992
    },
    {
      "epoch": 1.5787847229312586,
      "grad_norm": 0.5857130422423739,
      "learning_rate": 2.2374458638944775e-05,
      "loss": 11.8529,
      "step": 28993
    },
    {
      "epoch": 1.5788391769278416,
      "grad_norm": 0.6379855297922655,
      "learning_rate": 2.236889974734847e-05,
      "loss": 11.9326,
      "step": 28994
    },
    {
      "epoch": 1.5788936309244246,
      "grad_norm": 0.5219813063400816,
      "learning_rate": 2.236334145942465e-05,
      "loss": 11.7913,
      "step": 28995
    },
    {
      "epoch": 1.5789480849210076,
      "grad_norm": 0.5256263252262056,
      "learning_rate": 2.2357783775216478e-05,
      "loss": 11.7567,
      "step": 28996
    },
    {
      "epoch": 1.5790025389175906,
      "grad_norm": 0.6304368816838619,
      "learning_rate": 2.2352226694767198e-05,
      "loss": 11.8917,
      "step": 28997
    },
    {
      "epoch": 1.5790569929141736,
      "grad_norm": 0.5439879866079371,
      "learning_rate": 2.234667021812006e-05,
      "loss": 11.8921,
      "step": 28998
    },
    {
      "epoch": 1.5791114469107566,
      "grad_norm": 0.5229767730065559,
      "learning_rate": 2.2341114345318192e-05,
      "loss": 11.7656,
      "step": 28999
    },
    {
      "epoch": 1.5791659009073398,
      "grad_norm": 0.5446884088217604,
      "learning_rate": 2.233555907640491e-05,
      "loss": 11.8311,
      "step": 29000
    },
    {
      "epoch": 1.5792203549039228,
      "grad_norm": 0.506451489020867,
      "learning_rate": 2.2330004411423288e-05,
      "loss": 11.6914,
      "step": 29001
    },
    {
      "epoch": 1.5792748089005058,
      "grad_norm": 0.5656469671271962,
      "learning_rate": 2.232445035041658e-05,
      "loss": 11.9168,
      "step": 29002
    },
    {
      "epoch": 1.5793292628970887,
      "grad_norm": 0.6068004920605006,
      "learning_rate": 2.2318896893427998e-05,
      "loss": 11.7953,
      "step": 29003
    },
    {
      "epoch": 1.5793837168936717,
      "grad_norm": 0.6278159447890614,
      "learning_rate": 2.2313344040500673e-05,
      "loss": 11.9229,
      "step": 29004
    },
    {
      "epoch": 1.5794381708902547,
      "grad_norm": 0.5440617499970428,
      "learning_rate": 2.2307791791677845e-05,
      "loss": 11.7915,
      "step": 29005
    },
    {
      "epoch": 1.579492624886838,
      "grad_norm": 0.781416627615001,
      "learning_rate": 2.2302240147002616e-05,
      "loss": 11.7311,
      "step": 29006
    },
    {
      "epoch": 1.579547078883421,
      "grad_norm": 0.5246497992509671,
      "learning_rate": 2.229668910651824e-05,
      "loss": 11.6722,
      "step": 29007
    },
    {
      "epoch": 1.579601532880004,
      "grad_norm": 0.5731466136407526,
      "learning_rate": 2.2291138670267808e-05,
      "loss": 11.9134,
      "step": 29008
    },
    {
      "epoch": 1.579655986876587,
      "grad_norm": 0.5130475726398709,
      "learning_rate": 2.2285588838294514e-05,
      "loss": 11.8347,
      "step": 29009
    },
    {
      "epoch": 1.57971044087317,
      "grad_norm": 0.5376548533099087,
      "learning_rate": 2.2280039610641556e-05,
      "loss": 11.7917,
      "step": 29010
    },
    {
      "epoch": 1.579764894869753,
      "grad_norm": 0.6086318740399723,
      "learning_rate": 2.2274490987352003e-05,
      "loss": 11.901,
      "step": 29011
    },
    {
      "epoch": 1.579819348866336,
      "grad_norm": 0.6488987762219234,
      "learning_rate": 2.2268942968469043e-05,
      "loss": 11.8067,
      "step": 29012
    },
    {
      "epoch": 1.579873802862919,
      "grad_norm": 0.549697596344946,
      "learning_rate": 2.226339555403584e-05,
      "loss": 11.9265,
      "step": 29013
    },
    {
      "epoch": 1.579928256859502,
      "grad_norm": 0.4918710492757763,
      "learning_rate": 2.225784874409549e-05,
      "loss": 11.8429,
      "step": 29014
    },
    {
      "epoch": 1.5799827108560849,
      "grad_norm": 0.5244451401497985,
      "learning_rate": 2.225230253869116e-05,
      "loss": 11.7575,
      "step": 29015
    },
    {
      "epoch": 1.5800371648526679,
      "grad_norm": 0.5493826641505074,
      "learning_rate": 2.2246756937865943e-05,
      "loss": 11.7747,
      "step": 29016
    },
    {
      "epoch": 1.5800916188492509,
      "grad_norm": 0.5741686517478704,
      "learning_rate": 2.224121194166301e-05,
      "loss": 11.9342,
      "step": 29017
    },
    {
      "epoch": 1.5801460728458339,
      "grad_norm": 0.5597580542195663,
      "learning_rate": 2.2235667550125427e-05,
      "loss": 11.8798,
      "step": 29018
    },
    {
      "epoch": 1.5802005268424169,
      "grad_norm": 0.5389378037250077,
      "learning_rate": 2.223012376329633e-05,
      "loss": 11.8639,
      "step": 29019
    },
    {
      "epoch": 1.5802549808389998,
      "grad_norm": 0.5279581187588729,
      "learning_rate": 2.222458058121889e-05,
      "loss": 11.712,
      "step": 29020
    },
    {
      "epoch": 1.5803094348355828,
      "grad_norm": 0.5616077014528822,
      "learning_rate": 2.221903800393611e-05,
      "loss": 11.9719,
      "step": 29021
    },
    {
      "epoch": 1.5803638888321658,
      "grad_norm": 0.5172516190327713,
      "learning_rate": 2.2213496031491142e-05,
      "loss": 11.7983,
      "step": 29022
    },
    {
      "epoch": 1.5804183428287488,
      "grad_norm": 0.5774367667137277,
      "learning_rate": 2.2207954663927066e-05,
      "loss": 11.8706,
      "step": 29023
    },
    {
      "epoch": 1.580472796825332,
      "grad_norm": 0.5558358176704967,
      "learning_rate": 2.2202413901286968e-05,
      "loss": 11.8833,
      "step": 29024
    },
    {
      "epoch": 1.580527250821915,
      "grad_norm": 0.5207789713979787,
      "learning_rate": 2.219687374361398e-05,
      "loss": 11.7573,
      "step": 29025
    },
    {
      "epoch": 1.580581704818498,
      "grad_norm": 0.5306067291703788,
      "learning_rate": 2.2191334190951118e-05,
      "loss": 11.8661,
      "step": 29026
    },
    {
      "epoch": 1.580636158815081,
      "grad_norm": 0.5335870808746337,
      "learning_rate": 2.218579524334151e-05,
      "loss": 11.8255,
      "step": 29027
    },
    {
      "epoch": 1.580690612811664,
      "grad_norm": 0.518464436909379,
      "learning_rate": 2.218025690082819e-05,
      "loss": 11.7333,
      "step": 29028
    },
    {
      "epoch": 1.5807450668082472,
      "grad_norm": 0.5065462281628947,
      "learning_rate": 2.217471916345427e-05,
      "loss": 11.8099,
      "step": 29029
    },
    {
      "epoch": 1.5807995208048302,
      "grad_norm": 0.5028103002169194,
      "learning_rate": 2.2169182031262782e-05,
      "loss": 11.7827,
      "step": 29030
    },
    {
      "epoch": 1.5808539748014132,
      "grad_norm": 0.4958842347440862,
      "learning_rate": 2.216364550429676e-05,
      "loss": 11.8149,
      "step": 29031
    },
    {
      "epoch": 1.5809084287979962,
      "grad_norm": 0.6093113753707403,
      "learning_rate": 2.2158109582599305e-05,
      "loss": 11.7194,
      "step": 29032
    },
    {
      "epoch": 1.5809628827945792,
      "grad_norm": 0.5681429641056872,
      "learning_rate": 2.2152574266213434e-05,
      "loss": 11.9397,
      "step": 29033
    },
    {
      "epoch": 1.5810173367911622,
      "grad_norm": 0.6107264754564226,
      "learning_rate": 2.2147039555182216e-05,
      "loss": 11.5987,
      "step": 29034
    },
    {
      "epoch": 1.5810717907877452,
      "grad_norm": 0.5332209531813399,
      "learning_rate": 2.214150544954865e-05,
      "loss": 11.8597,
      "step": 29035
    },
    {
      "epoch": 1.5811262447843282,
      "grad_norm": 0.5354953413126309,
      "learning_rate": 2.213597194935578e-05,
      "loss": 11.8647,
      "step": 29036
    },
    {
      "epoch": 1.5811806987809112,
      "grad_norm": 0.5808049495861983,
      "learning_rate": 2.21304390546467e-05,
      "loss": 11.9228,
      "step": 29037
    },
    {
      "epoch": 1.5812351527774942,
      "grad_norm": 0.5742406354612188,
      "learning_rate": 2.2124906765464347e-05,
      "loss": 11.8291,
      "step": 29038
    },
    {
      "epoch": 1.5812896067740771,
      "grad_norm": 0.5044437219628712,
      "learning_rate": 2.21193750818518e-05,
      "loss": 11.8074,
      "step": 29039
    },
    {
      "epoch": 1.5813440607706601,
      "grad_norm": 0.4933525140210394,
      "learning_rate": 2.2113844003852057e-05,
      "loss": 11.8077,
      "step": 29040
    },
    {
      "epoch": 1.5813985147672431,
      "grad_norm": 0.5170722586985054,
      "learning_rate": 2.2108313531508108e-05,
      "loss": 11.7912,
      "step": 29041
    },
    {
      "epoch": 1.5814529687638261,
      "grad_norm": 0.54294628014179,
      "learning_rate": 2.2102783664862992e-05,
      "loss": 11.812,
      "step": 29042
    },
    {
      "epoch": 1.5815074227604091,
      "grad_norm": 0.5359145055436966,
      "learning_rate": 2.2097254403959666e-05,
      "loss": 11.7214,
      "step": 29043
    },
    {
      "epoch": 1.5815618767569921,
      "grad_norm": 0.5144911463018499,
      "learning_rate": 2.2091725748841187e-05,
      "loss": 11.6378,
      "step": 29044
    },
    {
      "epoch": 1.581616330753575,
      "grad_norm": 0.5511888396833232,
      "learning_rate": 2.208619769955048e-05,
      "loss": 11.8841,
      "step": 29045
    },
    {
      "epoch": 1.581670784750158,
      "grad_norm": 0.5605769639697662,
      "learning_rate": 2.2080670256130564e-05,
      "loss": 11.7991,
      "step": 29046
    },
    {
      "epoch": 1.5817252387467413,
      "grad_norm": 0.5613098356939531,
      "learning_rate": 2.2075143418624454e-05,
      "loss": 11.6674,
      "step": 29047
    },
    {
      "epoch": 1.5817796927433243,
      "grad_norm": 0.5292845915904641,
      "learning_rate": 2.2069617187075076e-05,
      "loss": 11.6973,
      "step": 29048
    },
    {
      "epoch": 1.5818341467399073,
      "grad_norm": 0.5340355912272015,
      "learning_rate": 2.2064091561525445e-05,
      "loss": 11.7223,
      "step": 29049
    },
    {
      "epoch": 1.5818886007364903,
      "grad_norm": 0.5426519904518304,
      "learning_rate": 2.205856654201851e-05,
      "loss": 11.8386,
      "step": 29050
    },
    {
      "epoch": 1.5819430547330733,
      "grad_norm": 0.5843536996058114,
      "learning_rate": 2.20530421285972e-05,
      "loss": 11.7543,
      "step": 29051
    },
    {
      "epoch": 1.5819975087296563,
      "grad_norm": 0.5281374525024105,
      "learning_rate": 2.2047518321304538e-05,
      "loss": 11.8029,
      "step": 29052
    },
    {
      "epoch": 1.5820519627262395,
      "grad_norm": 0.5468818828836998,
      "learning_rate": 2.2041995120183424e-05,
      "loss": 11.8517,
      "step": 29053
    },
    {
      "epoch": 1.5821064167228225,
      "grad_norm": 0.5696897498571843,
      "learning_rate": 2.2036472525276852e-05,
      "loss": 11.7462,
      "step": 29054
    },
    {
      "epoch": 1.5821608707194055,
      "grad_norm": 0.548390241746292,
      "learning_rate": 2.2030950536627715e-05,
      "loss": 11.9156,
      "step": 29055
    },
    {
      "epoch": 1.5822153247159885,
      "grad_norm": 0.5841769023178146,
      "learning_rate": 2.2025429154279008e-05,
      "loss": 11.7717,
      "step": 29056
    },
    {
      "epoch": 1.5822697787125715,
      "grad_norm": 0.536544838418208,
      "learning_rate": 2.2019908378273612e-05,
      "loss": 11.8411,
      "step": 29057
    },
    {
      "epoch": 1.5823242327091545,
      "grad_norm": 0.546668705865174,
      "learning_rate": 2.2014388208654492e-05,
      "loss": 11.7562,
      "step": 29058
    },
    {
      "epoch": 1.5823786867057374,
      "grad_norm": 0.5637228763955602,
      "learning_rate": 2.2008868645464586e-05,
      "loss": 11.7531,
      "step": 29059
    },
    {
      "epoch": 1.5824331407023204,
      "grad_norm": 0.5349452483227524,
      "learning_rate": 2.2003349688746776e-05,
      "loss": 11.8587,
      "step": 29060
    },
    {
      "epoch": 1.5824875946989034,
      "grad_norm": 0.5177356234154459,
      "learning_rate": 2.199783133854403e-05,
      "loss": 11.7372,
      "step": 29061
    },
    {
      "epoch": 1.5825420486954864,
      "grad_norm": 0.5816474828550473,
      "learning_rate": 2.1992313594899216e-05,
      "loss": 11.8017,
      "step": 29062
    },
    {
      "epoch": 1.5825965026920694,
      "grad_norm": 0.5381057753374475,
      "learning_rate": 2.198679645785524e-05,
      "loss": 11.8486,
      "step": 29063
    },
    {
      "epoch": 1.5826509566886524,
      "grad_norm": 0.548465577977049,
      "learning_rate": 2.1981279927455034e-05,
      "loss": 11.8801,
      "step": 29064
    },
    {
      "epoch": 1.5827054106852354,
      "grad_norm": 0.5347820313151733,
      "learning_rate": 2.1975764003741462e-05,
      "loss": 11.8356,
      "step": 29065
    },
    {
      "epoch": 1.5827598646818184,
      "grad_norm": 0.5559862770546784,
      "learning_rate": 2.1970248686757454e-05,
      "loss": 11.8921,
      "step": 29066
    },
    {
      "epoch": 1.5828143186784014,
      "grad_norm": 0.6062493227172571,
      "learning_rate": 2.1964733976545847e-05,
      "loss": 11.7938,
      "step": 29067
    },
    {
      "epoch": 1.5828687726749844,
      "grad_norm": 0.5570617421374567,
      "learning_rate": 2.195921987314956e-05,
      "loss": 11.8624,
      "step": 29068
    },
    {
      "epoch": 1.5829232266715674,
      "grad_norm": 0.5493201048040416,
      "learning_rate": 2.1953706376611495e-05,
      "loss": 11.853,
      "step": 29069
    },
    {
      "epoch": 1.5829776806681506,
      "grad_norm": 0.6152634032043247,
      "learning_rate": 2.1948193486974466e-05,
      "loss": 11.793,
      "step": 29070
    },
    {
      "epoch": 1.5830321346647336,
      "grad_norm": 0.5396840445424315,
      "learning_rate": 2.1942681204281433e-05,
      "loss": 11.9135,
      "step": 29071
    },
    {
      "epoch": 1.5830865886613166,
      "grad_norm": 0.5363726241797955,
      "learning_rate": 2.193716952857515e-05,
      "loss": 11.81,
      "step": 29072
    },
    {
      "epoch": 1.5831410426578996,
      "grad_norm": 0.5262139466133963,
      "learning_rate": 2.1931658459898518e-05,
      "loss": 11.844,
      "step": 29073
    },
    {
      "epoch": 1.5831954966544826,
      "grad_norm": 0.5416015297460568,
      "learning_rate": 2.1926147998294433e-05,
      "loss": 11.7217,
      "step": 29074
    },
    {
      "epoch": 1.5832499506510656,
      "grad_norm": 0.6406901360328437,
      "learning_rate": 2.1920638143805695e-05,
      "loss": 11.9512,
      "step": 29075
    },
    {
      "epoch": 1.5833044046476488,
      "grad_norm": 0.58056937267003,
      "learning_rate": 2.1915128896475188e-05,
      "loss": 11.7572,
      "step": 29076
    },
    {
      "epoch": 1.5833588586442318,
      "grad_norm": 0.5641658998258171,
      "learning_rate": 2.1909620256345708e-05,
      "loss": 11.7556,
      "step": 29077
    },
    {
      "epoch": 1.5834133126408148,
      "grad_norm": 0.47841146649147465,
      "learning_rate": 2.1904112223460138e-05,
      "loss": 11.7228,
      "step": 29078
    },
    {
      "epoch": 1.5834677666373977,
      "grad_norm": 0.5374175110604116,
      "learning_rate": 2.1898604797861267e-05,
      "loss": 11.7036,
      "step": 29079
    },
    {
      "epoch": 1.5835222206339807,
      "grad_norm": 0.5960610157409361,
      "learning_rate": 2.1893097979591937e-05,
      "loss": 11.7403,
      "step": 29080
    },
    {
      "epoch": 1.5835766746305637,
      "grad_norm": 0.5297496147641881,
      "learning_rate": 2.1887591768695036e-05,
      "loss": 11.7162,
      "step": 29081
    },
    {
      "epoch": 1.5836311286271467,
      "grad_norm": 0.5736564245092398,
      "learning_rate": 2.1882086165213268e-05,
      "loss": 11.8737,
      "step": 29082
    },
    {
      "epoch": 1.5836855826237297,
      "grad_norm": 0.5862986527536662,
      "learning_rate": 2.1876581169189527e-05,
      "loss": 11.7519,
      "step": 29083
    },
    {
      "epoch": 1.5837400366203127,
      "grad_norm": 0.5418749097848605,
      "learning_rate": 2.1871076780666556e-05,
      "loss": 11.8964,
      "step": 29084
    },
    {
      "epoch": 1.5837944906168957,
      "grad_norm": 0.4936289500691851,
      "learning_rate": 2.18655729996872e-05,
      "loss": 11.8897,
      "step": 29085
    },
    {
      "epoch": 1.5838489446134787,
      "grad_norm": 0.5013235215036588,
      "learning_rate": 2.186006982629427e-05,
      "loss": 11.8063,
      "step": 29086
    },
    {
      "epoch": 1.5839033986100617,
      "grad_norm": 0.5099163128588902,
      "learning_rate": 2.185456726053052e-05,
      "loss": 11.8442,
      "step": 29087
    },
    {
      "epoch": 1.5839578526066447,
      "grad_norm": 0.5917029892828737,
      "learning_rate": 2.1849065302438797e-05,
      "loss": 11.8341,
      "step": 29088
    },
    {
      "epoch": 1.5840123066032277,
      "grad_norm": 0.5021925925260688,
      "learning_rate": 2.1843563952061808e-05,
      "loss": 11.8131,
      "step": 29089
    },
    {
      "epoch": 1.5840667605998107,
      "grad_norm": 0.497188094143562,
      "learning_rate": 2.1838063209442407e-05,
      "loss": 11.8229,
      "step": 29090
    },
    {
      "epoch": 1.5841212145963937,
      "grad_norm": 0.626381462049018,
      "learning_rate": 2.1832563074623335e-05,
      "loss": 11.8201,
      "step": 29091
    },
    {
      "epoch": 1.5841756685929766,
      "grad_norm": 0.5784104874652187,
      "learning_rate": 2.182706354764733e-05,
      "loss": 11.9355,
      "step": 29092
    },
    {
      "epoch": 1.5842301225895596,
      "grad_norm": 0.5532987423111362,
      "learning_rate": 2.182156462855721e-05,
      "loss": 11.7749,
      "step": 29093
    },
    {
      "epoch": 1.5842845765861429,
      "grad_norm": 0.5022861028858535,
      "learning_rate": 2.18160663173957e-05,
      "loss": 11.7734,
      "step": 29094
    },
    {
      "epoch": 1.5843390305827258,
      "grad_norm": 0.5617752136216879,
      "learning_rate": 2.1810568614205562e-05,
      "loss": 11.8503,
      "step": 29095
    },
    {
      "epoch": 1.5843934845793088,
      "grad_norm": 0.5429460549899313,
      "learning_rate": 2.1805071519029586e-05,
      "loss": 11.8458,
      "step": 29096
    },
    {
      "epoch": 1.5844479385758918,
      "grad_norm": 0.5316821255156369,
      "learning_rate": 2.1799575031910447e-05,
      "loss": 11.8789,
      "step": 29097
    },
    {
      "epoch": 1.5845023925724748,
      "grad_norm": 0.5462723679800302,
      "learning_rate": 2.1794079152890966e-05,
      "loss": 11.8165,
      "step": 29098
    },
    {
      "epoch": 1.584556846569058,
      "grad_norm": 0.5281100326559773,
      "learning_rate": 2.1788583882013812e-05,
      "loss": 11.7381,
      "step": 29099
    },
    {
      "epoch": 1.584611300565641,
      "grad_norm": 0.5652817255406261,
      "learning_rate": 2.178308921932177e-05,
      "loss": 11.8184,
      "step": 29100
    },
    {
      "epoch": 1.584665754562224,
      "grad_norm": 0.559364291169176,
      "learning_rate": 2.1777595164857544e-05,
      "loss": 11.77,
      "step": 29101
    },
    {
      "epoch": 1.584720208558807,
      "grad_norm": 0.5517743434517466,
      "learning_rate": 2.1772101718663827e-05,
      "loss": 11.7682,
      "step": 29102
    },
    {
      "epoch": 1.58477466255539,
      "grad_norm": 0.6718982106305822,
      "learning_rate": 2.17666088807834e-05,
      "loss": 11.8081,
      "step": 29103
    },
    {
      "epoch": 1.584829116551973,
      "grad_norm": 0.5600334935057784,
      "learning_rate": 2.1761116651258918e-05,
      "loss": 11.8599,
      "step": 29104
    },
    {
      "epoch": 1.584883570548556,
      "grad_norm": 0.5398688980994854,
      "learning_rate": 2.175562503013313e-05,
      "loss": 11.8,
      "step": 29105
    },
    {
      "epoch": 1.584938024545139,
      "grad_norm": 0.520637674409149,
      "learning_rate": 2.17501340174487e-05,
      "loss": 11.7779,
      "step": 29106
    },
    {
      "epoch": 1.584992478541722,
      "grad_norm": 0.6543683148255448,
      "learning_rate": 2.174464361324835e-05,
      "loss": 11.6661,
      "step": 29107
    },
    {
      "epoch": 1.585046932538305,
      "grad_norm": 0.4959228769340991,
      "learning_rate": 2.1739153817574796e-05,
      "loss": 11.7957,
      "step": 29108
    },
    {
      "epoch": 1.585101386534888,
      "grad_norm": 0.5685651267143339,
      "learning_rate": 2.1733664630470685e-05,
      "loss": 11.8713,
      "step": 29109
    },
    {
      "epoch": 1.585155840531471,
      "grad_norm": 0.5651378296195563,
      "learning_rate": 2.1728176051978754e-05,
      "loss": 11.9213,
      "step": 29110
    },
    {
      "epoch": 1.585210294528054,
      "grad_norm": 0.5153589105640635,
      "learning_rate": 2.1722688082141652e-05,
      "loss": 11.795,
      "step": 29111
    },
    {
      "epoch": 1.585264748524637,
      "grad_norm": 0.6036755918870512,
      "learning_rate": 2.1717200721002017e-05,
      "loss": 11.8628,
      "step": 29112
    },
    {
      "epoch": 1.58531920252122,
      "grad_norm": 0.6678899420844301,
      "learning_rate": 2.17117139686026e-05,
      "loss": 11.8299,
      "step": 29113
    },
    {
      "epoch": 1.585373656517803,
      "grad_norm": 0.5068025624269648,
      "learning_rate": 2.170622782498598e-05,
      "loss": 11.8039,
      "step": 29114
    },
    {
      "epoch": 1.585428110514386,
      "grad_norm": 0.5893143485657308,
      "learning_rate": 2.170074229019491e-05,
      "loss": 11.7748,
      "step": 29115
    },
    {
      "epoch": 1.585482564510969,
      "grad_norm": 0.5289392164569375,
      "learning_rate": 2.1695257364271948e-05,
      "loss": 11.7358,
      "step": 29116
    },
    {
      "epoch": 1.5855370185075521,
      "grad_norm": 0.5227695138565449,
      "learning_rate": 2.168977304725981e-05,
      "loss": 11.8593,
      "step": 29117
    },
    {
      "epoch": 1.5855914725041351,
      "grad_norm": 0.5659794584806885,
      "learning_rate": 2.168428933920116e-05,
      "loss": 11.9605,
      "step": 29118
    },
    {
      "epoch": 1.5856459265007181,
      "grad_norm": 0.5498119223535392,
      "learning_rate": 2.167880624013857e-05,
      "loss": 11.7594,
      "step": 29119
    },
    {
      "epoch": 1.585700380497301,
      "grad_norm": 0.5337225061114226,
      "learning_rate": 2.167332375011476e-05,
      "loss": 11.8674,
      "step": 29120
    },
    {
      "epoch": 1.585754834493884,
      "grad_norm": 0.5580952724767342,
      "learning_rate": 2.1667841869172313e-05,
      "loss": 11.7955,
      "step": 29121
    },
    {
      "epoch": 1.585809288490467,
      "grad_norm": 0.5462561153115113,
      "learning_rate": 2.1662360597353826e-05,
      "loss": 11.8734,
      "step": 29122
    },
    {
      "epoch": 1.5858637424870503,
      "grad_norm": 0.5461428899994698,
      "learning_rate": 2.1656879934702e-05,
      "loss": 11.8885,
      "step": 29123
    },
    {
      "epoch": 1.5859181964836333,
      "grad_norm": 0.5583712414235588,
      "learning_rate": 2.165139988125938e-05,
      "loss": 11.8426,
      "step": 29124
    },
    {
      "epoch": 1.5859726504802163,
      "grad_norm": 0.6166205933702942,
      "learning_rate": 2.1645920437068645e-05,
      "loss": 11.6789,
      "step": 29125
    },
    {
      "epoch": 1.5860271044767993,
      "grad_norm": 0.5771649355488141,
      "learning_rate": 2.1640441602172347e-05,
      "loss": 11.8718,
      "step": 29126
    },
    {
      "epoch": 1.5860815584733823,
      "grad_norm": 0.5711758625196081,
      "learning_rate": 2.1634963376613136e-05,
      "loss": 11.7602,
      "step": 29127
    },
    {
      "epoch": 1.5861360124699653,
      "grad_norm": 0.5728149068220495,
      "learning_rate": 2.1629485760433575e-05,
      "loss": 11.7312,
      "step": 29128
    },
    {
      "epoch": 1.5861904664665483,
      "grad_norm": 0.6268439389283404,
      "learning_rate": 2.1624008753676262e-05,
      "loss": 11.8792,
      "step": 29129
    },
    {
      "epoch": 1.5862449204631313,
      "grad_norm": 0.6069611203707442,
      "learning_rate": 2.1618532356383835e-05,
      "loss": 11.831,
      "step": 29130
    },
    {
      "epoch": 1.5862993744597143,
      "grad_norm": 0.5386232493514849,
      "learning_rate": 2.1613056568598843e-05,
      "loss": 11.8296,
      "step": 29131
    },
    {
      "epoch": 1.5863538284562972,
      "grad_norm": 0.5265429702693016,
      "learning_rate": 2.1607581390363873e-05,
      "loss": 11.6696,
      "step": 29132
    },
    {
      "epoch": 1.5864082824528802,
      "grad_norm": 0.5530855200515045,
      "learning_rate": 2.160210682172147e-05,
      "loss": 11.7573,
      "step": 29133
    },
    {
      "epoch": 1.5864627364494632,
      "grad_norm": 0.5583801460465538,
      "learning_rate": 2.1596632862714228e-05,
      "loss": 11.9413,
      "step": 29134
    },
    {
      "epoch": 1.5865171904460462,
      "grad_norm": 0.565629203525206,
      "learning_rate": 2.159115951338475e-05,
      "loss": 11.8124,
      "step": 29135
    },
    {
      "epoch": 1.5865716444426292,
      "grad_norm": 0.6122724835057962,
      "learning_rate": 2.1585686773775525e-05,
      "loss": 11.949,
      "step": 29136
    },
    {
      "epoch": 1.5866260984392122,
      "grad_norm": 0.5596665136297907,
      "learning_rate": 2.1580214643929187e-05,
      "loss": 11.8034,
      "step": 29137
    },
    {
      "epoch": 1.5866805524357952,
      "grad_norm": 0.5271993444831127,
      "learning_rate": 2.1574743123888218e-05,
      "loss": 11.7126,
      "step": 29138
    },
    {
      "epoch": 1.5867350064323782,
      "grad_norm": 0.5547945712060374,
      "learning_rate": 2.1569272213695236e-05,
      "loss": 11.8732,
      "step": 29139
    },
    {
      "epoch": 1.5867894604289614,
      "grad_norm": 0.5334795982773989,
      "learning_rate": 2.156380191339271e-05,
      "loss": 11.7308,
      "step": 29140
    },
    {
      "epoch": 1.5868439144255444,
      "grad_norm": 0.5445794762733351,
      "learning_rate": 2.1558332223023247e-05,
      "loss": 11.6873,
      "step": 29141
    },
    {
      "epoch": 1.5868983684221274,
      "grad_norm": 0.5507707318790223,
      "learning_rate": 2.1552863142629344e-05,
      "loss": 11.7586,
      "step": 29142
    },
    {
      "epoch": 1.5869528224187104,
      "grad_norm": 0.5147203627377872,
      "learning_rate": 2.1547394672253496e-05,
      "loss": 11.765,
      "step": 29143
    },
    {
      "epoch": 1.5870072764152934,
      "grad_norm": 0.7584100790072255,
      "learning_rate": 2.1541926811938274e-05,
      "loss": 11.6893,
      "step": 29144
    },
    {
      "epoch": 1.5870617304118764,
      "grad_norm": 0.5401167545297461,
      "learning_rate": 2.153645956172622e-05,
      "loss": 11.7704,
      "step": 29145
    },
    {
      "epoch": 1.5871161844084596,
      "grad_norm": 0.558777060546608,
      "learning_rate": 2.1530992921659775e-05,
      "loss": 11.7655,
      "step": 29146
    },
    {
      "epoch": 1.5871706384050426,
      "grad_norm": 0.49229068922422564,
      "learning_rate": 2.1525526891781522e-05,
      "loss": 11.7972,
      "step": 29147
    },
    {
      "epoch": 1.5872250924016256,
      "grad_norm": 0.5533981064329909,
      "learning_rate": 2.1520061472133902e-05,
      "loss": 11.8416,
      "step": 29148
    },
    {
      "epoch": 1.5872795463982086,
      "grad_norm": 0.5013605542073992,
      "learning_rate": 2.1514596662759467e-05,
      "loss": 11.8212,
      "step": 29149
    },
    {
      "epoch": 1.5873340003947916,
      "grad_norm": 0.5900255638176347,
      "learning_rate": 2.1509132463700677e-05,
      "loss": 11.6981,
      "step": 29150
    },
    {
      "epoch": 1.5873884543913745,
      "grad_norm": 0.5330474399920727,
      "learning_rate": 2.150366887500005e-05,
      "loss": 11.8208,
      "step": 29151
    },
    {
      "epoch": 1.5874429083879575,
      "grad_norm": 0.5134170634657004,
      "learning_rate": 2.1498205896700063e-05,
      "loss": 11.6273,
      "step": 29152
    },
    {
      "epoch": 1.5874973623845405,
      "grad_norm": 0.5683029043787393,
      "learning_rate": 2.1492743528843173e-05,
      "loss": 11.8019,
      "step": 29153
    },
    {
      "epoch": 1.5875518163811235,
      "grad_norm": 0.5519986747922824,
      "learning_rate": 2.148728177147189e-05,
      "loss": 11.6967,
      "step": 29154
    },
    {
      "epoch": 1.5876062703777065,
      "grad_norm": 0.5351536096407199,
      "learning_rate": 2.1481820624628644e-05,
      "loss": 11.789,
      "step": 29155
    },
    {
      "epoch": 1.5876607243742895,
      "grad_norm": 0.514282314891165,
      "learning_rate": 2.1476360088355928e-05,
      "loss": 11.8068,
      "step": 29156
    },
    {
      "epoch": 1.5877151783708725,
      "grad_norm": 0.5509461107819521,
      "learning_rate": 2.147090016269624e-05,
      "loss": 11.8408,
      "step": 29157
    },
    {
      "epoch": 1.5877696323674555,
      "grad_norm": 0.5855389748059536,
      "learning_rate": 2.1465440847691975e-05,
      "loss": 11.8164,
      "step": 29158
    },
    {
      "epoch": 1.5878240863640385,
      "grad_norm": 0.5449832269316832,
      "learning_rate": 2.1459982143385627e-05,
      "loss": 11.8003,
      "step": 29159
    },
    {
      "epoch": 1.5878785403606215,
      "grad_norm": 0.5963238684244109,
      "learning_rate": 2.1454524049819613e-05,
      "loss": 11.9584,
      "step": 29160
    },
    {
      "epoch": 1.5879329943572045,
      "grad_norm": 0.5565491697938304,
      "learning_rate": 2.1449066567036413e-05,
      "loss": 11.7727,
      "step": 29161
    },
    {
      "epoch": 1.5879874483537875,
      "grad_norm": 0.5219938308345949,
      "learning_rate": 2.144360969507845e-05,
      "loss": 11.875,
      "step": 29162
    },
    {
      "epoch": 1.5880419023503707,
      "grad_norm": 0.5408579161222582,
      "learning_rate": 2.1438153433988117e-05,
      "loss": 11.8056,
      "step": 29163
    },
    {
      "epoch": 1.5880963563469537,
      "grad_norm": 0.4886017432595483,
      "learning_rate": 2.143269778380791e-05,
      "loss": 11.8131,
      "step": 29164
    },
    {
      "epoch": 1.5881508103435367,
      "grad_norm": 0.5771394631036637,
      "learning_rate": 2.14272427445802e-05,
      "loss": 11.8153,
      "step": 29165
    },
    {
      "epoch": 1.5882052643401197,
      "grad_norm": 0.5217238601603127,
      "learning_rate": 2.1421788316347415e-05,
      "loss": 11.819,
      "step": 29166
    },
    {
      "epoch": 1.5882597183367027,
      "grad_norm": 0.5630323168814597,
      "learning_rate": 2.1416334499152013e-05,
      "loss": 11.842,
      "step": 29167
    },
    {
      "epoch": 1.5883141723332856,
      "grad_norm": 0.5376945597360916,
      "learning_rate": 2.1410881293036344e-05,
      "loss": 11.7791,
      "step": 29168
    },
    {
      "epoch": 1.5883686263298689,
      "grad_norm": 0.5248237679813952,
      "learning_rate": 2.1405428698042874e-05,
      "loss": 11.7955,
      "step": 29169
    },
    {
      "epoch": 1.5884230803264519,
      "grad_norm": 0.5436701563811874,
      "learning_rate": 2.1399976714213942e-05,
      "loss": 11.8776,
      "step": 29170
    },
    {
      "epoch": 1.5884775343230348,
      "grad_norm": 0.5128821896212716,
      "learning_rate": 2.1394525341591997e-05,
      "loss": 11.8182,
      "step": 29171
    },
    {
      "epoch": 1.5885319883196178,
      "grad_norm": 0.4992722527555482,
      "learning_rate": 2.1389074580219402e-05,
      "loss": 11.7359,
      "step": 29172
    },
    {
      "epoch": 1.5885864423162008,
      "grad_norm": 0.5467252671302966,
      "learning_rate": 2.138362443013853e-05,
      "loss": 11.8906,
      "step": 29173
    },
    {
      "epoch": 1.5886408963127838,
      "grad_norm": 0.5843163581667822,
      "learning_rate": 2.137817489139179e-05,
      "loss": 11.8127,
      "step": 29174
    },
    {
      "epoch": 1.5886953503093668,
      "grad_norm": 0.629587246009668,
      "learning_rate": 2.1372725964021534e-05,
      "loss": 11.9161,
      "step": 29175
    },
    {
      "epoch": 1.5887498043059498,
      "grad_norm": 0.5397705353121288,
      "learning_rate": 2.136727764807016e-05,
      "loss": 11.8118,
      "step": 29176
    },
    {
      "epoch": 1.5888042583025328,
      "grad_norm": 0.540204443399126,
      "learning_rate": 2.1361829943580004e-05,
      "loss": 11.8334,
      "step": 29177
    },
    {
      "epoch": 1.5888587122991158,
      "grad_norm": 0.6136391249747147,
      "learning_rate": 2.135638285059345e-05,
      "loss": 11.7844,
      "step": 29178
    },
    {
      "epoch": 1.5889131662956988,
      "grad_norm": 0.5118044243007966,
      "learning_rate": 2.1350936369152873e-05,
      "loss": 11.9372,
      "step": 29179
    },
    {
      "epoch": 1.5889676202922818,
      "grad_norm": 0.5571087194325227,
      "learning_rate": 2.134549049930058e-05,
      "loss": 11.773,
      "step": 29180
    },
    {
      "epoch": 1.5890220742888648,
      "grad_norm": 0.5520172025077966,
      "learning_rate": 2.134004524107899e-05,
      "loss": 11.8087,
      "step": 29181
    },
    {
      "epoch": 1.5890765282854478,
      "grad_norm": 0.5189616400894338,
      "learning_rate": 2.1334600594530353e-05,
      "loss": 11.8258,
      "step": 29182
    },
    {
      "epoch": 1.5891309822820308,
      "grad_norm": 0.5060035468310032,
      "learning_rate": 2.132915655969705e-05,
      "loss": 11.8142,
      "step": 29183
    },
    {
      "epoch": 1.5891854362786138,
      "grad_norm": 0.5444357482948762,
      "learning_rate": 2.1323713136621447e-05,
      "loss": 11.9714,
      "step": 29184
    },
    {
      "epoch": 1.5892398902751967,
      "grad_norm": 0.5036453223751051,
      "learning_rate": 2.1318270325345813e-05,
      "loss": 11.7883,
      "step": 29185
    },
    {
      "epoch": 1.5892943442717797,
      "grad_norm": 0.5272278688885228,
      "learning_rate": 2.1312828125912542e-05,
      "loss": 11.7632,
      "step": 29186
    },
    {
      "epoch": 1.589348798268363,
      "grad_norm": 0.5516678752768197,
      "learning_rate": 2.1307386538363872e-05,
      "loss": 11.9416,
      "step": 29187
    },
    {
      "epoch": 1.589403252264946,
      "grad_norm": 0.5390981805405269,
      "learning_rate": 2.1301945562742198e-05,
      "loss": 11.8479,
      "step": 29188
    },
    {
      "epoch": 1.589457706261529,
      "grad_norm": 0.5455874018990825,
      "learning_rate": 2.1296505199089767e-05,
      "loss": 11.745,
      "step": 29189
    },
    {
      "epoch": 1.589512160258112,
      "grad_norm": 0.5358756695688616,
      "learning_rate": 2.129106544744891e-05,
      "loss": 11.9746,
      "step": 29190
    },
    {
      "epoch": 1.589566614254695,
      "grad_norm": 0.5572858884820283,
      "learning_rate": 2.1285626307861985e-05,
      "loss": 11.684,
      "step": 29191
    },
    {
      "epoch": 1.589621068251278,
      "grad_norm": 0.5749974252294444,
      "learning_rate": 2.1280187780371164e-05,
      "loss": 11.7451,
      "step": 29192
    },
    {
      "epoch": 1.5896755222478611,
      "grad_norm": 0.6094768452017864,
      "learning_rate": 2.1274749865018817e-05,
      "loss": 11.784,
      "step": 29193
    },
    {
      "epoch": 1.5897299762444441,
      "grad_norm": 0.5155799518670798,
      "learning_rate": 2.1269312561847243e-05,
      "loss": 11.8573,
      "step": 29194
    },
    {
      "epoch": 1.5897844302410271,
      "grad_norm": 0.554418934830859,
      "learning_rate": 2.1263875870898663e-05,
      "loss": 11.8116,
      "step": 29195
    },
    {
      "epoch": 1.58983888423761,
      "grad_norm": 0.5677755816917791,
      "learning_rate": 2.1258439792215424e-05,
      "loss": 11.9431,
      "step": 29196
    },
    {
      "epoch": 1.589893338234193,
      "grad_norm": 0.5491318303962497,
      "learning_rate": 2.125300432583972e-05,
      "loss": 11.8706,
      "step": 29197
    },
    {
      "epoch": 1.589947792230776,
      "grad_norm": 0.5536616969800391,
      "learning_rate": 2.12475694718139e-05,
      "loss": 11.7946,
      "step": 29198
    },
    {
      "epoch": 1.590002246227359,
      "grad_norm": 0.5421698118841684,
      "learning_rate": 2.124213523018016e-05,
      "loss": 11.7416,
      "step": 29199
    },
    {
      "epoch": 1.590056700223942,
      "grad_norm": 0.5689022927176461,
      "learning_rate": 2.1236701600980778e-05,
      "loss": 11.9269,
      "step": 29200
    },
    {
      "epoch": 1.590111154220525,
      "grad_norm": 0.6153413866925239,
      "learning_rate": 2.1231268584258045e-05,
      "loss": 11.8882,
      "step": 29201
    },
    {
      "epoch": 1.590165608217108,
      "grad_norm": 0.5124053512110116,
      "learning_rate": 2.122583618005417e-05,
      "loss": 11.8088,
      "step": 29202
    },
    {
      "epoch": 1.590220062213691,
      "grad_norm": 0.580719462557604,
      "learning_rate": 2.122040438841141e-05,
      "loss": 11.7163,
      "step": 29203
    },
    {
      "epoch": 1.590274516210274,
      "grad_norm": 0.5151472244442663,
      "learning_rate": 2.1214973209371968e-05,
      "loss": 11.7885,
      "step": 29204
    },
    {
      "epoch": 1.590328970206857,
      "grad_norm": 0.5076383471266717,
      "learning_rate": 2.1209542642978108e-05,
      "loss": 11.7529,
      "step": 29205
    },
    {
      "epoch": 1.59038342420344,
      "grad_norm": 0.5193509259548776,
      "learning_rate": 2.1204112689272084e-05,
      "loss": 11.7794,
      "step": 29206
    },
    {
      "epoch": 1.590437878200023,
      "grad_norm": 0.5828889025431375,
      "learning_rate": 2.1198683348296066e-05,
      "loss": 11.7939,
      "step": 29207
    },
    {
      "epoch": 1.590492332196606,
      "grad_norm": 0.5650755250368186,
      "learning_rate": 2.119325462009233e-05,
      "loss": 11.8566,
      "step": 29208
    },
    {
      "epoch": 1.590546786193189,
      "grad_norm": 0.5297085796436609,
      "learning_rate": 2.1187826504703035e-05,
      "loss": 11.887,
      "step": 29209
    },
    {
      "epoch": 1.5906012401897722,
      "grad_norm": 0.5064611510433685,
      "learning_rate": 2.118239900217044e-05,
      "loss": 11.7344,
      "step": 29210
    },
    {
      "epoch": 1.5906556941863552,
      "grad_norm": 0.5342162214270872,
      "learning_rate": 2.1176972112536707e-05,
      "loss": 11.8243,
      "step": 29211
    },
    {
      "epoch": 1.5907101481829382,
      "grad_norm": 0.5392048206008597,
      "learning_rate": 2.1171545835844074e-05,
      "loss": 11.7277,
      "step": 29212
    },
    {
      "epoch": 1.5907646021795212,
      "grad_norm": 0.5489343232386106,
      "learning_rate": 2.116612017213473e-05,
      "loss": 11.7642,
      "step": 29213
    },
    {
      "epoch": 1.5908190561761042,
      "grad_norm": 0.5556368544405371,
      "learning_rate": 2.1160695121450835e-05,
      "loss": 11.9001,
      "step": 29214
    },
    {
      "epoch": 1.5908735101726872,
      "grad_norm": 0.5637806875218498,
      "learning_rate": 2.1155270683834583e-05,
      "loss": 11.846,
      "step": 29215
    },
    {
      "epoch": 1.5909279641692704,
      "grad_norm": 0.5675488791609382,
      "learning_rate": 2.1149846859328204e-05,
      "loss": 11.8661,
      "step": 29216
    },
    {
      "epoch": 1.5909824181658534,
      "grad_norm": 0.5517703220420297,
      "learning_rate": 2.1144423647973798e-05,
      "loss": 11.8339,
      "step": 29217
    },
    {
      "epoch": 1.5910368721624364,
      "grad_norm": 0.5623212189048679,
      "learning_rate": 2.1139001049813623e-05,
      "loss": 11.7671,
      "step": 29218
    },
    {
      "epoch": 1.5910913261590194,
      "grad_norm": 0.49704509406640446,
      "learning_rate": 2.1133579064889764e-05,
      "loss": 11.8135,
      "step": 29219
    },
    {
      "epoch": 1.5911457801556024,
      "grad_norm": 0.5886064269462271,
      "learning_rate": 2.1128157693244454e-05,
      "loss": 11.8631,
      "step": 29220
    },
    {
      "epoch": 1.5912002341521854,
      "grad_norm": 0.5392727739873534,
      "learning_rate": 2.1122736934919783e-05,
      "loss": 11.7098,
      "step": 29221
    },
    {
      "epoch": 1.5912546881487684,
      "grad_norm": 0.565132586191543,
      "learning_rate": 2.1117316789957962e-05,
      "loss": 11.8217,
      "step": 29222
    },
    {
      "epoch": 1.5913091421453514,
      "grad_norm": 0.542112994531766,
      "learning_rate": 2.1111897258401125e-05,
      "loss": 11.795,
      "step": 29223
    },
    {
      "epoch": 1.5913635961419343,
      "grad_norm": 0.5161550032683274,
      "learning_rate": 2.110647834029137e-05,
      "loss": 11.8477,
      "step": 29224
    },
    {
      "epoch": 1.5914180501385173,
      "grad_norm": 0.5026830349343652,
      "learning_rate": 2.1101060035670893e-05,
      "loss": 11.8038,
      "step": 29225
    },
    {
      "epoch": 1.5914725041351003,
      "grad_norm": 0.5379469062411203,
      "learning_rate": 2.1095642344581778e-05,
      "loss": 11.8555,
      "step": 29226
    },
    {
      "epoch": 1.5915269581316833,
      "grad_norm": 0.5024891120704371,
      "learning_rate": 2.1090225267066187e-05,
      "loss": 11.6885,
      "step": 29227
    },
    {
      "epoch": 1.5915814121282663,
      "grad_norm": 0.56469021355504,
      "learning_rate": 2.1084808803166267e-05,
      "loss": 11.9066,
      "step": 29228
    },
    {
      "epoch": 1.5916358661248493,
      "grad_norm": 0.5676993194955461,
      "learning_rate": 2.107939295292407e-05,
      "loss": 11.7328,
      "step": 29229
    },
    {
      "epoch": 1.5916903201214323,
      "grad_norm": 0.5052910014520902,
      "learning_rate": 2.1073977716381787e-05,
      "loss": 11.7752,
      "step": 29230
    },
    {
      "epoch": 1.5917447741180153,
      "grad_norm": 0.5207248151621681,
      "learning_rate": 2.106856309358145e-05,
      "loss": 11.8814,
      "step": 29231
    },
    {
      "epoch": 1.5917992281145983,
      "grad_norm": 0.5204707893107874,
      "learning_rate": 2.1063149084565238e-05,
      "loss": 11.8628,
      "step": 29232
    },
    {
      "epoch": 1.5918536821111815,
      "grad_norm": 0.604036827527067,
      "learning_rate": 2.1057735689375212e-05,
      "loss": 11.9725,
      "step": 29233
    },
    {
      "epoch": 1.5919081361077645,
      "grad_norm": 0.5089674645710892,
      "learning_rate": 2.1052322908053457e-05,
      "loss": 11.7684,
      "step": 29234
    },
    {
      "epoch": 1.5919625901043475,
      "grad_norm": 0.5745763659233781,
      "learning_rate": 2.104691074064209e-05,
      "loss": 11.8978,
      "step": 29235
    },
    {
      "epoch": 1.5920170441009305,
      "grad_norm": 0.591309404312403,
      "learning_rate": 2.1041499187183167e-05,
      "loss": 11.8323,
      "step": 29236
    },
    {
      "epoch": 1.5920714980975135,
      "grad_norm": 0.5545029480301127,
      "learning_rate": 2.103608824771881e-05,
      "loss": 11.796,
      "step": 29237
    },
    {
      "epoch": 1.5921259520940965,
      "grad_norm": 0.5184792550490716,
      "learning_rate": 2.1030677922291054e-05,
      "loss": 11.79,
      "step": 29238
    },
    {
      "epoch": 1.5921804060906797,
      "grad_norm": 0.5539966035999682,
      "learning_rate": 2.1025268210941984e-05,
      "loss": 11.9322,
      "step": 29239
    },
    {
      "epoch": 1.5922348600872627,
      "grad_norm": 0.576787261578154,
      "learning_rate": 2.1019859113713702e-05,
      "loss": 11.8311,
      "step": 29240
    },
    {
      "epoch": 1.5922893140838457,
      "grad_norm": 0.5353127817297225,
      "learning_rate": 2.101445063064821e-05,
      "loss": 11.8213,
      "step": 29241
    },
    {
      "epoch": 1.5923437680804287,
      "grad_norm": 0.5028518180042723,
      "learning_rate": 2.1009042761787622e-05,
      "loss": 11.6888,
      "step": 29242
    },
    {
      "epoch": 1.5923982220770116,
      "grad_norm": 0.5227168638947924,
      "learning_rate": 2.100363550717397e-05,
      "loss": 11.7396,
      "step": 29243
    },
    {
      "epoch": 1.5924526760735946,
      "grad_norm": 0.5207687491171965,
      "learning_rate": 2.0998228866849258e-05,
      "loss": 11.8756,
      "step": 29244
    },
    {
      "epoch": 1.5925071300701776,
      "grad_norm": 0.5689433497115669,
      "learning_rate": 2.0992822840855607e-05,
      "loss": 11.9099,
      "step": 29245
    },
    {
      "epoch": 1.5925615840667606,
      "grad_norm": 0.5170718065392603,
      "learning_rate": 2.0987417429234978e-05,
      "loss": 11.6968,
      "step": 29246
    },
    {
      "epoch": 1.5926160380633436,
      "grad_norm": 0.7161047318327063,
      "learning_rate": 2.098201263202948e-05,
      "loss": 11.9096,
      "step": 29247
    },
    {
      "epoch": 1.5926704920599266,
      "grad_norm": 0.5449667988099567,
      "learning_rate": 2.0976608449281065e-05,
      "loss": 11.7962,
      "step": 29248
    },
    {
      "epoch": 1.5927249460565096,
      "grad_norm": 0.5422448402646335,
      "learning_rate": 2.0971204881031804e-05,
      "loss": 11.7914,
      "step": 29249
    },
    {
      "epoch": 1.5927794000530926,
      "grad_norm": 0.5204730984461148,
      "learning_rate": 2.0965801927323723e-05,
      "loss": 11.8461,
      "step": 29250
    },
    {
      "epoch": 1.5928338540496756,
      "grad_norm": 0.5911032488345888,
      "learning_rate": 2.09603995881988e-05,
      "loss": 11.9048,
      "step": 29251
    },
    {
      "epoch": 1.5928883080462586,
      "grad_norm": 0.5380247928248235,
      "learning_rate": 2.095499786369912e-05,
      "loss": 11.8832,
      "step": 29252
    },
    {
      "epoch": 1.5929427620428416,
      "grad_norm": 0.5702427190662462,
      "learning_rate": 2.0949596753866573e-05,
      "loss": 11.7932,
      "step": 29253
    },
    {
      "epoch": 1.5929972160394246,
      "grad_norm": 0.5117243886033823,
      "learning_rate": 2.094419625874322e-05,
      "loss": 11.8682,
      "step": 29254
    },
    {
      "epoch": 1.5930516700360076,
      "grad_norm": 0.562316585517797,
      "learning_rate": 2.0938796378371084e-05,
      "loss": 11.7828,
      "step": 29255
    },
    {
      "epoch": 1.5931061240325906,
      "grad_norm": 0.564703827891049,
      "learning_rate": 2.0933397112792097e-05,
      "loss": 11.8802,
      "step": 29256
    },
    {
      "epoch": 1.5931605780291738,
      "grad_norm": 0.5840082363291428,
      "learning_rate": 2.0927998462048305e-05,
      "loss": 11.8813,
      "step": 29257
    },
    {
      "epoch": 1.5932150320257568,
      "grad_norm": 0.5169333620750146,
      "learning_rate": 2.0922600426181627e-05,
      "loss": 11.8397,
      "step": 29258
    },
    {
      "epoch": 1.5932694860223398,
      "grad_norm": 0.5618592799405008,
      "learning_rate": 2.09172030052341e-05,
      "loss": 11.742,
      "step": 29259
    },
    {
      "epoch": 1.5933239400189227,
      "grad_norm": 0.6014043592532105,
      "learning_rate": 2.091180619924763e-05,
      "loss": 11.7218,
      "step": 29260
    },
    {
      "epoch": 1.5933783940155057,
      "grad_norm": 0.5450336197543947,
      "learning_rate": 2.090641000826422e-05,
      "loss": 11.8609,
      "step": 29261
    },
    {
      "epoch": 1.593432848012089,
      "grad_norm": 0.5122395469798758,
      "learning_rate": 2.0901014432325894e-05,
      "loss": 11.758,
      "step": 29262
    },
    {
      "epoch": 1.593487302008672,
      "grad_norm": 0.5099102735598482,
      "learning_rate": 2.0895619471474482e-05,
      "loss": 11.8422,
      "step": 29263
    },
    {
      "epoch": 1.593541756005255,
      "grad_norm": 0.6136624409372344,
      "learning_rate": 2.0890225125751996e-05,
      "loss": 11.7182,
      "step": 29264
    },
    {
      "epoch": 1.593596210001838,
      "grad_norm": 0.6155525934292949,
      "learning_rate": 2.088483139520042e-05,
      "loss": 11.9642,
      "step": 29265
    },
    {
      "epoch": 1.593650663998421,
      "grad_norm": 0.5131323241283007,
      "learning_rate": 2.087943827986163e-05,
      "loss": 11.7876,
      "step": 29266
    },
    {
      "epoch": 1.593705117995004,
      "grad_norm": 0.592275942631102,
      "learning_rate": 2.087404577977763e-05,
      "loss": 11.7975,
      "step": 29267
    },
    {
      "epoch": 1.593759571991587,
      "grad_norm": 0.5590714359972297,
      "learning_rate": 2.0868653894990286e-05,
      "loss": 11.7961,
      "step": 29268
    },
    {
      "epoch": 1.59381402598817,
      "grad_norm": 0.5687129917076879,
      "learning_rate": 2.086326262554159e-05,
      "loss": 11.8583,
      "step": 29269
    },
    {
      "epoch": 1.593868479984753,
      "grad_norm": 0.5440672765535223,
      "learning_rate": 2.0857871971473396e-05,
      "loss": 11.7144,
      "step": 29270
    },
    {
      "epoch": 1.5939229339813359,
      "grad_norm": 0.5096900234979592,
      "learning_rate": 2.0852481932827683e-05,
      "loss": 11.8566,
      "step": 29271
    },
    {
      "epoch": 1.5939773879779189,
      "grad_norm": 0.6039441842701397,
      "learning_rate": 2.0847092509646382e-05,
      "loss": 11.8094,
      "step": 29272
    },
    {
      "epoch": 1.5940318419745019,
      "grad_norm": 0.5224844081263593,
      "learning_rate": 2.0841703701971317e-05,
      "loss": 11.8094,
      "step": 29273
    },
    {
      "epoch": 1.5940862959710849,
      "grad_norm": 0.5925389058157018,
      "learning_rate": 2.0836315509844462e-05,
      "loss": 11.8046,
      "step": 29274
    },
    {
      "epoch": 1.5941407499676679,
      "grad_norm": 0.5479508959563254,
      "learning_rate": 2.0830927933307666e-05,
      "loss": 11.7138,
      "step": 29275
    },
    {
      "epoch": 1.5941952039642509,
      "grad_norm": 0.637556042757401,
      "learning_rate": 2.0825540972402858e-05,
      "loss": 11.8174,
      "step": 29276
    },
    {
      "epoch": 1.5942496579608338,
      "grad_norm": 0.5866111568680433,
      "learning_rate": 2.082015462717194e-05,
      "loss": 11.844,
      "step": 29277
    },
    {
      "epoch": 1.5943041119574168,
      "grad_norm": 0.5123714898958694,
      "learning_rate": 2.0814768897656754e-05,
      "loss": 11.7639,
      "step": 29278
    },
    {
      "epoch": 1.5943585659539998,
      "grad_norm": 0.6224809269400543,
      "learning_rate": 2.080938378389923e-05,
      "loss": 11.9527,
      "step": 29279
    },
    {
      "epoch": 1.594413019950583,
      "grad_norm": 0.5313001344689005,
      "learning_rate": 2.080399928594119e-05,
      "loss": 11.8202,
      "step": 29280
    },
    {
      "epoch": 1.594467473947166,
      "grad_norm": 0.592526489518852,
      "learning_rate": 2.0798615403824562e-05,
      "loss": 11.8038,
      "step": 29281
    },
    {
      "epoch": 1.594521927943749,
      "grad_norm": 0.5396856684300272,
      "learning_rate": 2.079323213759119e-05,
      "loss": 11.8334,
      "step": 29282
    },
    {
      "epoch": 1.594576381940332,
      "grad_norm": 0.5204311666070865,
      "learning_rate": 2.0787849487282894e-05,
      "loss": 11.8591,
      "step": 29283
    },
    {
      "epoch": 1.594630835936915,
      "grad_norm": 0.5625748473460092,
      "learning_rate": 2.078246745294159e-05,
      "loss": 11.8796,
      "step": 29284
    },
    {
      "epoch": 1.594685289933498,
      "grad_norm": 0.5504402028504651,
      "learning_rate": 2.077708603460907e-05,
      "loss": 11.8352,
      "step": 29285
    },
    {
      "epoch": 1.5947397439300812,
      "grad_norm": 0.5938363247497671,
      "learning_rate": 2.0771705232327253e-05,
      "loss": 11.8516,
      "step": 29286
    },
    {
      "epoch": 1.5947941979266642,
      "grad_norm": 0.5375808369991738,
      "learning_rate": 2.0766325046137915e-05,
      "loss": 11.7668,
      "step": 29287
    },
    {
      "epoch": 1.5948486519232472,
      "grad_norm": 0.6255520458851344,
      "learning_rate": 2.0760945476082914e-05,
      "loss": 11.9198,
      "step": 29288
    },
    {
      "epoch": 1.5949031059198302,
      "grad_norm": 0.6078822174774229,
      "learning_rate": 2.075556652220413e-05,
      "loss": 11.861,
      "step": 29289
    },
    {
      "epoch": 1.5949575599164132,
      "grad_norm": 0.5727629181299997,
      "learning_rate": 2.0750188184543306e-05,
      "loss": 11.8628,
      "step": 29290
    },
    {
      "epoch": 1.5950120139129962,
      "grad_norm": 0.567216252456352,
      "learning_rate": 2.074481046314235e-05,
      "loss": 11.7328,
      "step": 29291
    },
    {
      "epoch": 1.5950664679095792,
      "grad_norm": 0.6022782443729904,
      "learning_rate": 2.0739433358043026e-05,
      "loss": 11.7829,
      "step": 29292
    },
    {
      "epoch": 1.5951209219061622,
      "grad_norm": 0.5616427075100456,
      "learning_rate": 2.0734056869287144e-05,
      "loss": 11.8856,
      "step": 29293
    },
    {
      "epoch": 1.5951753759027452,
      "grad_norm": 0.5327986980656418,
      "learning_rate": 2.072868099691655e-05,
      "loss": 11.9241,
      "step": 29294
    },
    {
      "epoch": 1.5952298298993282,
      "grad_norm": 0.5593695576862636,
      "learning_rate": 2.0723305740972996e-05,
      "loss": 11.8909,
      "step": 29295
    },
    {
      "epoch": 1.5952842838959111,
      "grad_norm": 0.5311426509712349,
      "learning_rate": 2.0717931101498344e-05,
      "loss": 11.9296,
      "step": 29296
    },
    {
      "epoch": 1.5953387378924941,
      "grad_norm": 0.5176813535764708,
      "learning_rate": 2.0712557078534335e-05,
      "loss": 11.8577,
      "step": 29297
    },
    {
      "epoch": 1.5953931918890771,
      "grad_norm": 0.5496162841096986,
      "learning_rate": 2.0707183672122765e-05,
      "loss": 11.8236,
      "step": 29298
    },
    {
      "epoch": 1.5954476458856601,
      "grad_norm": 0.504755071941615,
      "learning_rate": 2.0701810882305462e-05,
      "loss": 11.7105,
      "step": 29299
    },
    {
      "epoch": 1.5955020998822431,
      "grad_norm": 0.5039080260857132,
      "learning_rate": 2.0696438709124157e-05,
      "loss": 11.7858,
      "step": 29300
    },
    {
      "epoch": 1.5955565538788261,
      "grad_norm": 0.6188281055076668,
      "learning_rate": 2.069106715262067e-05,
      "loss": 11.8878,
      "step": 29301
    },
    {
      "epoch": 1.595611007875409,
      "grad_norm": 0.5333251973907444,
      "learning_rate": 2.0685696212836737e-05,
      "loss": 11.8109,
      "step": 29302
    },
    {
      "epoch": 1.5956654618719923,
      "grad_norm": 0.5721325884762738,
      "learning_rate": 2.0680325889814112e-05,
      "loss": 11.85,
      "step": 29303
    },
    {
      "epoch": 1.5957199158685753,
      "grad_norm": 0.5291263883005053,
      "learning_rate": 2.0674956183594595e-05,
      "loss": 11.8698,
      "step": 29304
    },
    {
      "epoch": 1.5957743698651583,
      "grad_norm": 0.5646301205861213,
      "learning_rate": 2.06695870942199e-05,
      "loss": 11.8938,
      "step": 29305
    },
    {
      "epoch": 1.5958288238617413,
      "grad_norm": 0.5432610891425843,
      "learning_rate": 2.0664218621731823e-05,
      "loss": 11.7214,
      "step": 29306
    },
    {
      "epoch": 1.5958832778583243,
      "grad_norm": 0.539769048527323,
      "learning_rate": 2.0658850766172054e-05,
      "loss": 11.9207,
      "step": 29307
    },
    {
      "epoch": 1.5959377318549073,
      "grad_norm": 0.588714758758604,
      "learning_rate": 2.0653483527582408e-05,
      "loss": 11.8795,
      "step": 29308
    },
    {
      "epoch": 1.5959921858514905,
      "grad_norm": 0.5916651483477288,
      "learning_rate": 2.0648116906004543e-05,
      "loss": 11.7702,
      "step": 29309
    },
    {
      "epoch": 1.5960466398480735,
      "grad_norm": 0.5108358492591021,
      "learning_rate": 2.0642750901480233e-05,
      "loss": 11.8893,
      "step": 29310
    },
    {
      "epoch": 1.5961010938446565,
      "grad_norm": 0.49544440087693337,
      "learning_rate": 2.0637385514051223e-05,
      "loss": 11.7864,
      "step": 29311
    },
    {
      "epoch": 1.5961555478412395,
      "grad_norm": 0.5423380161079329,
      "learning_rate": 2.0632020743759217e-05,
      "loss": 11.691,
      "step": 29312
    },
    {
      "epoch": 1.5962100018378225,
      "grad_norm": 0.5046285314022753,
      "learning_rate": 2.06266565906459e-05,
      "loss": 11.9099,
      "step": 29313
    },
    {
      "epoch": 1.5962644558344055,
      "grad_norm": 0.5434175528898965,
      "learning_rate": 2.0621293054753032e-05,
      "loss": 11.9292,
      "step": 29314
    },
    {
      "epoch": 1.5963189098309885,
      "grad_norm": 0.513969180975846,
      "learning_rate": 2.0615930136122286e-05,
      "loss": 11.8333,
      "step": 29315
    },
    {
      "epoch": 1.5963733638275714,
      "grad_norm": 0.5537772534660337,
      "learning_rate": 2.061056783479539e-05,
      "loss": 11.7639,
      "step": 29316
    },
    {
      "epoch": 1.5964278178241544,
      "grad_norm": 0.5658981839435101,
      "learning_rate": 2.0605206150814017e-05,
      "loss": 11.9302,
      "step": 29317
    },
    {
      "epoch": 1.5964822718207374,
      "grad_norm": 0.5156821553687051,
      "learning_rate": 2.0599845084219905e-05,
      "loss": 11.8463,
      "step": 29318
    },
    {
      "epoch": 1.5965367258173204,
      "grad_norm": 0.6033784564546595,
      "learning_rate": 2.059448463505468e-05,
      "loss": 11.8912,
      "step": 29319
    },
    {
      "epoch": 1.5965911798139034,
      "grad_norm": 0.5023804642949706,
      "learning_rate": 2.058912480336006e-05,
      "loss": 11.5914,
      "step": 29320
    },
    {
      "epoch": 1.5966456338104864,
      "grad_norm": 0.565739164961108,
      "learning_rate": 2.058376558917775e-05,
      "loss": 11.7791,
      "step": 29321
    },
    {
      "epoch": 1.5967000878070694,
      "grad_norm": 0.5280704510056182,
      "learning_rate": 2.0578406992549405e-05,
      "loss": 11.6932,
      "step": 29322
    },
    {
      "epoch": 1.5967545418036524,
      "grad_norm": 0.5126183917450234,
      "learning_rate": 2.0573049013516676e-05,
      "loss": 11.8379,
      "step": 29323
    },
    {
      "epoch": 1.5968089958002354,
      "grad_norm": 0.5440726386513677,
      "learning_rate": 2.056769165212121e-05,
      "loss": 11.8529,
      "step": 29324
    },
    {
      "epoch": 1.5968634497968184,
      "grad_norm": 0.5233461159863342,
      "learning_rate": 2.05623349084047e-05,
      "loss": 11.727,
      "step": 29325
    },
    {
      "epoch": 1.5969179037934014,
      "grad_norm": 0.5363070990828511,
      "learning_rate": 2.055697878240882e-05,
      "loss": 11.839,
      "step": 29326
    },
    {
      "epoch": 1.5969723577899846,
      "grad_norm": 0.614792689281094,
      "learning_rate": 2.0551623274175167e-05,
      "loss": 11.7179,
      "step": 29327
    },
    {
      "epoch": 1.5970268117865676,
      "grad_norm": 0.6002726534508523,
      "learning_rate": 2.0546268383745447e-05,
      "loss": 11.8518,
      "step": 29328
    },
    {
      "epoch": 1.5970812657831506,
      "grad_norm": 0.6787932544133423,
      "learning_rate": 2.0540914111161246e-05,
      "loss": 11.7234,
      "step": 29329
    },
    {
      "epoch": 1.5971357197797336,
      "grad_norm": 0.53390654356914,
      "learning_rate": 2.0535560456464244e-05,
      "loss": 11.7811,
      "step": 29330
    },
    {
      "epoch": 1.5971901737763166,
      "grad_norm": 0.5723515378116586,
      "learning_rate": 2.0530207419696014e-05,
      "loss": 11.8642,
      "step": 29331
    },
    {
      "epoch": 1.5972446277728998,
      "grad_norm": 0.5323370881988478,
      "learning_rate": 2.052485500089826e-05,
      "loss": 11.8854,
      "step": 29332
    },
    {
      "epoch": 1.5972990817694828,
      "grad_norm": 0.5355162973079043,
      "learning_rate": 2.0519503200112544e-05,
      "loss": 11.8937,
      "step": 29333
    },
    {
      "epoch": 1.5973535357660658,
      "grad_norm": 0.6096867267092838,
      "learning_rate": 2.0514152017380482e-05,
      "loss": 11.7952,
      "step": 29334
    },
    {
      "epoch": 1.5974079897626488,
      "grad_norm": 0.6575767372477747,
      "learning_rate": 2.050880145274373e-05,
      "loss": 11.9411,
      "step": 29335
    },
    {
      "epoch": 1.5974624437592317,
      "grad_norm": 0.5766491995228399,
      "learning_rate": 2.0503451506243844e-05,
      "loss": 11.8742,
      "step": 29336
    },
    {
      "epoch": 1.5975168977558147,
      "grad_norm": 0.5564938888422738,
      "learning_rate": 2.049810217792245e-05,
      "loss": 11.6938,
      "step": 29337
    },
    {
      "epoch": 1.5975713517523977,
      "grad_norm": 0.5633894880921775,
      "learning_rate": 2.049275346782118e-05,
      "loss": 11.8614,
      "step": 29338
    },
    {
      "epoch": 1.5976258057489807,
      "grad_norm": 0.6026291967793118,
      "learning_rate": 2.048740537598155e-05,
      "loss": 11.9225,
      "step": 29339
    },
    {
      "epoch": 1.5976802597455637,
      "grad_norm": 0.5454465730805981,
      "learning_rate": 2.048205790244523e-05,
      "loss": 11.806,
      "step": 29340
    },
    {
      "epoch": 1.5977347137421467,
      "grad_norm": 0.5324393254097145,
      "learning_rate": 2.047671104725373e-05,
      "loss": 11.7547,
      "step": 29341
    },
    {
      "epoch": 1.5977891677387297,
      "grad_norm": 0.539211253220576,
      "learning_rate": 2.047136481044869e-05,
      "loss": 11.8763,
      "step": 29342
    },
    {
      "epoch": 1.5978436217353127,
      "grad_norm": 0.4994208877902171,
      "learning_rate": 2.0466019192071652e-05,
      "loss": 11.8807,
      "step": 29343
    },
    {
      "epoch": 1.5978980757318957,
      "grad_norm": 0.5234971724937825,
      "learning_rate": 2.0460674192164163e-05,
      "loss": 11.8054,
      "step": 29344
    },
    {
      "epoch": 1.5979525297284787,
      "grad_norm": 0.6142188536115024,
      "learning_rate": 2.045532981076783e-05,
      "loss": 11.7835,
      "step": 29345
    },
    {
      "epoch": 1.5980069837250617,
      "grad_norm": 0.5165604775158151,
      "learning_rate": 2.0449986047924173e-05,
      "loss": 11.7835,
      "step": 29346
    },
    {
      "epoch": 1.5980614377216447,
      "grad_norm": 0.5374935990097058,
      "learning_rate": 2.044464290367476e-05,
      "loss": 11.6989,
      "step": 29347
    },
    {
      "epoch": 1.5981158917182277,
      "grad_norm": 0.5124067275860661,
      "learning_rate": 2.0439300378061178e-05,
      "loss": 11.7579,
      "step": 29348
    },
    {
      "epoch": 1.5981703457148106,
      "grad_norm": 0.5376089506350612,
      "learning_rate": 2.0433958471124902e-05,
      "loss": 11.77,
      "step": 29349
    },
    {
      "epoch": 1.5982247997113939,
      "grad_norm": 0.5845308154121257,
      "learning_rate": 2.042861718290754e-05,
      "loss": 11.831,
      "step": 29350
    },
    {
      "epoch": 1.5982792537079769,
      "grad_norm": 0.5422491709840996,
      "learning_rate": 2.0423276513450572e-05,
      "loss": 11.8305,
      "step": 29351
    },
    {
      "epoch": 1.5983337077045598,
      "grad_norm": 0.5882532223644485,
      "learning_rate": 2.041793646279557e-05,
      "loss": 11.8764,
      "step": 29352
    },
    {
      "epoch": 1.5983881617011428,
      "grad_norm": 0.5068251702623046,
      "learning_rate": 2.041259703098405e-05,
      "loss": 11.8794,
      "step": 29353
    },
    {
      "epoch": 1.5984426156977258,
      "grad_norm": 0.5423090314044817,
      "learning_rate": 2.040725821805749e-05,
      "loss": 11.6394,
      "step": 29354
    },
    {
      "epoch": 1.5984970696943088,
      "grad_norm": 0.5244575659651973,
      "learning_rate": 2.0401920024057464e-05,
      "loss": 11.8177,
      "step": 29355
    },
    {
      "epoch": 1.598551523690892,
      "grad_norm": 0.534760458178448,
      "learning_rate": 2.0396582449025438e-05,
      "loss": 11.8421,
      "step": 29356
    },
    {
      "epoch": 1.598605977687475,
      "grad_norm": 0.595690836982202,
      "learning_rate": 2.0391245493002952e-05,
      "loss": 11.9205,
      "step": 29357
    },
    {
      "epoch": 1.598660431684058,
      "grad_norm": 0.5297888333733959,
      "learning_rate": 2.0385909156031467e-05,
      "loss": 11.8155,
      "step": 29358
    },
    {
      "epoch": 1.598714885680641,
      "grad_norm": 0.5589288586725722,
      "learning_rate": 2.0380573438152507e-05,
      "loss": 11.6377,
      "step": 29359
    },
    {
      "epoch": 1.598769339677224,
      "grad_norm": 0.5116527023082919,
      "learning_rate": 2.0375238339407577e-05,
      "loss": 11.7126,
      "step": 29360
    },
    {
      "epoch": 1.598823793673807,
      "grad_norm": 0.5354004677928198,
      "learning_rate": 2.0369903859838135e-05,
      "loss": 11.7904,
      "step": 29361
    },
    {
      "epoch": 1.59887824767039,
      "grad_norm": 0.5822814331580956,
      "learning_rate": 2.0364569999485695e-05,
      "loss": 11.7983,
      "step": 29362
    },
    {
      "epoch": 1.598932701666973,
      "grad_norm": 0.4890187835699303,
      "learning_rate": 2.035923675839171e-05,
      "loss": 11.8121,
      "step": 29363
    },
    {
      "epoch": 1.598987155663556,
      "grad_norm": 0.5322577516379914,
      "learning_rate": 2.0353904136597635e-05,
      "loss": 11.7383,
      "step": 29364
    },
    {
      "epoch": 1.599041609660139,
      "grad_norm": 0.5383542847029062,
      "learning_rate": 2.034857213414497e-05,
      "loss": 11.6958,
      "step": 29365
    },
    {
      "epoch": 1.599096063656722,
      "grad_norm": 0.5210816974437572,
      "learning_rate": 2.0343240751075154e-05,
      "loss": 11.5566,
      "step": 29366
    },
    {
      "epoch": 1.599150517653305,
      "grad_norm": 0.5457318379706793,
      "learning_rate": 2.0337909987429683e-05,
      "loss": 11.6954,
      "step": 29367
    },
    {
      "epoch": 1.599204971649888,
      "grad_norm": 0.5168958739928255,
      "learning_rate": 2.0332579843249954e-05,
      "loss": 11.6211,
      "step": 29368
    },
    {
      "epoch": 1.599259425646471,
      "grad_norm": 0.5850823126483844,
      "learning_rate": 2.032725031857744e-05,
      "loss": 11.737,
      "step": 29369
    },
    {
      "epoch": 1.599313879643054,
      "grad_norm": 0.5253489496144531,
      "learning_rate": 2.0321921413453627e-05,
      "loss": 11.8412,
      "step": 29370
    },
    {
      "epoch": 1.599368333639637,
      "grad_norm": 0.5565454633267117,
      "learning_rate": 2.031659312791987e-05,
      "loss": 11.8185,
      "step": 29371
    },
    {
      "epoch": 1.59942278763622,
      "grad_norm": 0.5149642405061943,
      "learning_rate": 2.0311265462017693e-05,
      "loss": 11.8535,
      "step": 29372
    },
    {
      "epoch": 1.5994772416328031,
      "grad_norm": 0.5864200899043692,
      "learning_rate": 2.0305938415788472e-05,
      "loss": 11.7961,
      "step": 29373
    },
    {
      "epoch": 1.5995316956293861,
      "grad_norm": 0.5950075065925858,
      "learning_rate": 2.030061198927361e-05,
      "loss": 11.8858,
      "step": 29374
    },
    {
      "epoch": 1.5995861496259691,
      "grad_norm": 0.6078560209968946,
      "learning_rate": 2.0295286182514584e-05,
      "loss": 11.9283,
      "step": 29375
    },
    {
      "epoch": 1.5996406036225521,
      "grad_norm": 0.5447446069690238,
      "learning_rate": 2.028996099555275e-05,
      "loss": 11.7079,
      "step": 29376
    },
    {
      "epoch": 1.599695057619135,
      "grad_norm": 0.5171646235522743,
      "learning_rate": 2.028463642842957e-05,
      "loss": 11.7078,
      "step": 29377
    },
    {
      "epoch": 1.599749511615718,
      "grad_norm": 0.5406602315200316,
      "learning_rate": 2.0279312481186407e-05,
      "loss": 11.8501,
      "step": 29378
    },
    {
      "epoch": 1.5998039656123013,
      "grad_norm": 0.554756835062345,
      "learning_rate": 2.0273989153864703e-05,
      "loss": 11.9458,
      "step": 29379
    },
    {
      "epoch": 1.5998584196088843,
      "grad_norm": 0.6053443037557681,
      "learning_rate": 2.0268666446505803e-05,
      "loss": 11.8822,
      "step": 29380
    },
    {
      "epoch": 1.5999128736054673,
      "grad_norm": 0.5052093775294951,
      "learning_rate": 2.0263344359151114e-05,
      "loss": 11.8362,
      "step": 29381
    },
    {
      "epoch": 1.5999673276020503,
      "grad_norm": 0.5497470843840734,
      "learning_rate": 2.0258022891842066e-05,
      "loss": 11.7865,
      "step": 29382
    },
    {
      "epoch": 1.6000217815986333,
      "grad_norm": 0.5696991347372337,
      "learning_rate": 2.0252702044620007e-05,
      "loss": 11.7343,
      "step": 29383
    },
    {
      "epoch": 1.6000762355952163,
      "grad_norm": 0.5391041147615704,
      "learning_rate": 2.024738181752631e-05,
      "loss": 11.8043,
      "step": 29384
    },
    {
      "epoch": 1.6001306895917993,
      "grad_norm": 0.5445163514808162,
      "learning_rate": 2.0242062210602318e-05,
      "loss": 11.8279,
      "step": 29385
    },
    {
      "epoch": 1.6001851435883823,
      "grad_norm": 0.4901077404302885,
      "learning_rate": 2.0236743223889422e-05,
      "loss": 11.7292,
      "step": 29386
    },
    {
      "epoch": 1.6002395975849653,
      "grad_norm": 0.5709519578301949,
      "learning_rate": 2.0231424857429026e-05,
      "loss": 11.8579,
      "step": 29387
    },
    {
      "epoch": 1.6002940515815482,
      "grad_norm": 0.5521014017851856,
      "learning_rate": 2.0226107111262404e-05,
      "loss": 11.7892,
      "step": 29388
    },
    {
      "epoch": 1.6003485055781312,
      "grad_norm": 0.5981844161381498,
      "learning_rate": 2.0220789985430986e-05,
      "loss": 11.8641,
      "step": 29389
    },
    {
      "epoch": 1.6004029595747142,
      "grad_norm": 0.6060426741470232,
      "learning_rate": 2.0215473479976067e-05,
      "loss": 11.8387,
      "step": 29390
    },
    {
      "epoch": 1.6004574135712972,
      "grad_norm": 0.5382350872141481,
      "learning_rate": 2.0210157594939016e-05,
      "loss": 11.7363,
      "step": 29391
    },
    {
      "epoch": 1.6005118675678802,
      "grad_norm": 0.5575719557209478,
      "learning_rate": 2.020484233036114e-05,
      "loss": 11.9765,
      "step": 29392
    },
    {
      "epoch": 1.6005663215644632,
      "grad_norm": 0.5301358191553304,
      "learning_rate": 2.0199527686283827e-05,
      "loss": 11.673,
      "step": 29393
    },
    {
      "epoch": 1.6006207755610462,
      "grad_norm": 0.5860651116209377,
      "learning_rate": 2.0194213662748362e-05,
      "loss": 11.7878,
      "step": 29394
    },
    {
      "epoch": 1.6006752295576292,
      "grad_norm": 0.5871124440341244,
      "learning_rate": 2.018890025979604e-05,
      "loss": 11.7139,
      "step": 29395
    },
    {
      "epoch": 1.6007296835542122,
      "grad_norm": 0.5616677883384459,
      "learning_rate": 2.0183587477468226e-05,
      "loss": 11.8742,
      "step": 29396
    },
    {
      "epoch": 1.6007841375507954,
      "grad_norm": 0.5191381149894279,
      "learning_rate": 2.0178275315806237e-05,
      "loss": 11.7855,
      "step": 29397
    },
    {
      "epoch": 1.6008385915473784,
      "grad_norm": 0.5091419315561139,
      "learning_rate": 2.017296377485135e-05,
      "loss": 11.8683,
      "step": 29398
    },
    {
      "epoch": 1.6008930455439614,
      "grad_norm": 0.5352089856035932,
      "learning_rate": 2.0167652854644902e-05,
      "loss": 11.8727,
      "step": 29399
    },
    {
      "epoch": 1.6009474995405444,
      "grad_norm": 0.5537442666076666,
      "learning_rate": 2.0162342555228152e-05,
      "loss": 11.8444,
      "step": 29400
    },
    {
      "epoch": 1.6010019535371274,
      "grad_norm": 0.5936712654447079,
      "learning_rate": 2.0157032876642445e-05,
      "loss": 11.7059,
      "step": 29401
    },
    {
      "epoch": 1.6010564075337106,
      "grad_norm": 0.5149627350136136,
      "learning_rate": 2.0151723818929002e-05,
      "loss": 11.7851,
      "step": 29402
    },
    {
      "epoch": 1.6011108615302936,
      "grad_norm": 0.5368922638871556,
      "learning_rate": 2.014641538212918e-05,
      "loss": 11.763,
      "step": 29403
    },
    {
      "epoch": 1.6011653155268766,
      "grad_norm": 0.5987494576207734,
      "learning_rate": 2.0141107566284224e-05,
      "loss": 11.7101,
      "step": 29404
    },
    {
      "epoch": 1.6012197695234596,
      "grad_norm": 0.5622285933443621,
      "learning_rate": 2.0135800371435377e-05,
      "loss": 11.7634,
      "step": 29405
    },
    {
      "epoch": 1.6012742235200426,
      "grad_norm": 0.587119010358452,
      "learning_rate": 2.013049379762397e-05,
      "loss": 11.8866,
      "step": 29406
    },
    {
      "epoch": 1.6013286775166256,
      "grad_norm": 0.5303848846970265,
      "learning_rate": 2.0125187844891214e-05,
      "loss": 11.8284,
      "step": 29407
    },
    {
      "epoch": 1.6013831315132085,
      "grad_norm": 0.5169890060139251,
      "learning_rate": 2.0119882513278376e-05,
      "loss": 11.7693,
      "step": 29408
    },
    {
      "epoch": 1.6014375855097915,
      "grad_norm": 0.4841309745431037,
      "learning_rate": 2.011457780282677e-05,
      "loss": 11.6597,
      "step": 29409
    },
    {
      "epoch": 1.6014920395063745,
      "grad_norm": 0.5296250126575548,
      "learning_rate": 2.010927371357757e-05,
      "loss": 11.8657,
      "step": 29410
    },
    {
      "epoch": 1.6015464935029575,
      "grad_norm": 0.5264449967244791,
      "learning_rate": 2.010397024557209e-05,
      "loss": 11.902,
      "step": 29411
    },
    {
      "epoch": 1.6016009474995405,
      "grad_norm": 0.5242921438198742,
      "learning_rate": 2.0098667398851502e-05,
      "loss": 11.761,
      "step": 29412
    },
    {
      "epoch": 1.6016554014961235,
      "grad_norm": 0.5567536072830178,
      "learning_rate": 2.0093365173457102e-05,
      "loss": 11.9027,
      "step": 29413
    },
    {
      "epoch": 1.6017098554927065,
      "grad_norm": 0.5864266023755338,
      "learning_rate": 2.0088063569430093e-05,
      "loss": 11.9558,
      "step": 29414
    },
    {
      "epoch": 1.6017643094892895,
      "grad_norm": 0.5302354177762099,
      "learning_rate": 2.0082762586811686e-05,
      "loss": 11.8522,
      "step": 29415
    },
    {
      "epoch": 1.6018187634858725,
      "grad_norm": 0.5340893847277921,
      "learning_rate": 2.007746222564314e-05,
      "loss": 11.8762,
      "step": 29416
    },
    {
      "epoch": 1.6018732174824555,
      "grad_norm": 0.5841347553224232,
      "learning_rate": 2.0072162485965618e-05,
      "loss": 11.5254,
      "step": 29417
    },
    {
      "epoch": 1.6019276714790385,
      "grad_norm": 0.5277587417691794,
      "learning_rate": 2.0066863367820376e-05,
      "loss": 11.8683,
      "step": 29418
    },
    {
      "epoch": 1.6019821254756215,
      "grad_norm": 0.5513675047556913,
      "learning_rate": 2.006156487124863e-05,
      "loss": 11.8087,
      "step": 29419
    },
    {
      "epoch": 1.6020365794722047,
      "grad_norm": 0.523210258162122,
      "learning_rate": 2.005626699629153e-05,
      "loss": 11.8505,
      "step": 29420
    },
    {
      "epoch": 1.6020910334687877,
      "grad_norm": 0.5376344267347557,
      "learning_rate": 2.0050969742990346e-05,
      "loss": 11.7899,
      "step": 29421
    },
    {
      "epoch": 1.6021454874653707,
      "grad_norm": 0.535397752464809,
      "learning_rate": 2.0045673111386187e-05,
      "loss": 11.7908,
      "step": 29422
    },
    {
      "epoch": 1.6021999414619537,
      "grad_norm": 0.5702444878502699,
      "learning_rate": 2.0040377101520313e-05,
      "loss": 11.7889,
      "step": 29423
    },
    {
      "epoch": 1.6022543954585367,
      "grad_norm": 0.6231034304487845,
      "learning_rate": 2.0035081713433878e-05,
      "loss": 11.8418,
      "step": 29424
    },
    {
      "epoch": 1.6023088494551196,
      "grad_norm": 0.49722995197810943,
      "learning_rate": 2.0029786947168028e-05,
      "loss": 11.8859,
      "step": 29425
    },
    {
      "epoch": 1.6023633034517029,
      "grad_norm": 0.5580957798454991,
      "learning_rate": 2.0024492802763996e-05,
      "loss": 11.746,
      "step": 29426
    },
    {
      "epoch": 1.6024177574482859,
      "grad_norm": 0.5889126367332558,
      "learning_rate": 2.001919928026289e-05,
      "loss": 11.8205,
      "step": 29427
    },
    {
      "epoch": 1.6024722114448688,
      "grad_norm": 0.5209238836756184,
      "learning_rate": 2.0013906379705925e-05,
      "loss": 11.9455,
      "step": 29428
    },
    {
      "epoch": 1.6025266654414518,
      "grad_norm": 0.5677628338776259,
      "learning_rate": 2.0008614101134228e-05,
      "loss": 11.8863,
      "step": 29429
    },
    {
      "epoch": 1.6025811194380348,
      "grad_norm": 0.6074607586573398,
      "learning_rate": 2.0003322444588945e-05,
      "loss": 11.7899,
      "step": 29430
    },
    {
      "epoch": 1.6026355734346178,
      "grad_norm": 0.5618640876632571,
      "learning_rate": 1.9998031410111275e-05,
      "loss": 11.7676,
      "step": 29431
    },
    {
      "epoch": 1.6026900274312008,
      "grad_norm": 0.5084939224750649,
      "learning_rate": 1.999274099774231e-05,
      "loss": 11.7653,
      "step": 29432
    },
    {
      "epoch": 1.6027444814277838,
      "grad_norm": 0.5814571428367679,
      "learning_rate": 1.9987451207523245e-05,
      "loss": 11.8396,
      "step": 29433
    },
    {
      "epoch": 1.6027989354243668,
      "grad_norm": 0.5685096359853066,
      "learning_rate": 1.9982162039495133e-05,
      "loss": 11.818,
      "step": 29434
    },
    {
      "epoch": 1.6028533894209498,
      "grad_norm": 0.5745809363328269,
      "learning_rate": 1.9976873493699155e-05,
      "loss": 11.6109,
      "step": 29435
    },
    {
      "epoch": 1.6029078434175328,
      "grad_norm": 0.5289668498802075,
      "learning_rate": 1.9971585570176443e-05,
      "loss": 11.705,
      "step": 29436
    },
    {
      "epoch": 1.6029622974141158,
      "grad_norm": 0.6143617425459307,
      "learning_rate": 1.9966298268968086e-05,
      "loss": 11.8005,
      "step": 29437
    },
    {
      "epoch": 1.6030167514106988,
      "grad_norm": 0.49538255166652645,
      "learning_rate": 1.996101159011524e-05,
      "loss": 11.7992,
      "step": 29438
    },
    {
      "epoch": 1.6030712054072818,
      "grad_norm": 0.5543726132688722,
      "learning_rate": 1.9955725533658954e-05,
      "loss": 11.7016,
      "step": 29439
    },
    {
      "epoch": 1.6031256594038648,
      "grad_norm": 0.6354922580447147,
      "learning_rate": 1.995044009964041e-05,
      "loss": 11.8876,
      "step": 29440
    },
    {
      "epoch": 1.6031801134004477,
      "grad_norm": 0.5291584829323384,
      "learning_rate": 1.9945155288100636e-05,
      "loss": 11.8345,
      "step": 29441
    },
    {
      "epoch": 1.6032345673970307,
      "grad_norm": 0.5579322998389947,
      "learning_rate": 1.9939871099080753e-05,
      "loss": 11.8444,
      "step": 29442
    },
    {
      "epoch": 1.603289021393614,
      "grad_norm": 0.5522996991215278,
      "learning_rate": 1.993458753262192e-05,
      "loss": 11.9134,
      "step": 29443
    },
    {
      "epoch": 1.603343475390197,
      "grad_norm": 0.5340596804135831,
      "learning_rate": 1.9929304588765107e-05,
      "loss": 11.8661,
      "step": 29444
    },
    {
      "epoch": 1.60339792938678,
      "grad_norm": 0.5900912133232146,
      "learning_rate": 1.9924022267551444e-05,
      "loss": 11.9336,
      "step": 29445
    },
    {
      "epoch": 1.603452383383363,
      "grad_norm": 0.6204722787699938,
      "learning_rate": 1.9918740569022042e-05,
      "loss": 11.954,
      "step": 29446
    },
    {
      "epoch": 1.603506837379946,
      "grad_norm": 0.5473150052355028,
      "learning_rate": 1.9913459493217923e-05,
      "loss": 11.7859,
      "step": 29447
    },
    {
      "epoch": 1.603561291376529,
      "grad_norm": 0.6360911442389331,
      "learning_rate": 1.9908179040180197e-05,
      "loss": 11.8471,
      "step": 29448
    },
    {
      "epoch": 1.6036157453731121,
      "grad_norm": 0.5718820184914463,
      "learning_rate": 1.9902899209949865e-05,
      "loss": 11.8596,
      "step": 29449
    },
    {
      "epoch": 1.6036701993696951,
      "grad_norm": 0.5702905523345672,
      "learning_rate": 1.9897620002568064e-05,
      "loss": 11.9062,
      "step": 29450
    },
    {
      "epoch": 1.6037246533662781,
      "grad_norm": 0.674056427309686,
      "learning_rate": 1.9892341418075766e-05,
      "loss": 12.0222,
      "step": 29451
    },
    {
      "epoch": 1.6037791073628611,
      "grad_norm": 0.5430100968272906,
      "learning_rate": 1.9887063456514055e-05,
      "loss": 11.6578,
      "step": 29452
    },
    {
      "epoch": 1.603833561359444,
      "grad_norm": 0.5437180308953273,
      "learning_rate": 1.9881786117924027e-05,
      "loss": 11.7666,
      "step": 29453
    },
    {
      "epoch": 1.603888015356027,
      "grad_norm": 0.5978516870845708,
      "learning_rate": 1.9876509402346623e-05,
      "loss": 11.7347,
      "step": 29454
    },
    {
      "epoch": 1.60394246935261,
      "grad_norm": 0.5734227760594006,
      "learning_rate": 1.9871233309822935e-05,
      "loss": 11.9542,
      "step": 29455
    },
    {
      "epoch": 1.603996923349193,
      "grad_norm": 0.5099606994702084,
      "learning_rate": 1.9865957840393945e-05,
      "loss": 11.858,
      "step": 29456
    },
    {
      "epoch": 1.604051377345776,
      "grad_norm": 0.5513875227445373,
      "learning_rate": 1.986068299410071e-05,
      "loss": 11.7734,
      "step": 29457
    },
    {
      "epoch": 1.604105831342359,
      "grad_norm": 0.571443913331018,
      "learning_rate": 1.9855408770984274e-05,
      "loss": 11.7681,
      "step": 29458
    },
    {
      "epoch": 1.604160285338942,
      "grad_norm": 0.6271376497212182,
      "learning_rate": 1.9850135171085582e-05,
      "loss": 11.7576,
      "step": 29459
    },
    {
      "epoch": 1.604214739335525,
      "grad_norm": 0.5429512264294071,
      "learning_rate": 1.9844862194445713e-05,
      "loss": 11.8022,
      "step": 29460
    },
    {
      "epoch": 1.604269193332108,
      "grad_norm": 0.5231861899643807,
      "learning_rate": 1.9839589841105623e-05,
      "loss": 11.8079,
      "step": 29461
    },
    {
      "epoch": 1.604323647328691,
      "grad_norm": 0.530296903557611,
      "learning_rate": 1.983431811110633e-05,
      "loss": 11.8469,
      "step": 29462
    },
    {
      "epoch": 1.604378101325274,
      "grad_norm": 0.5534393628654909,
      "learning_rate": 1.9829047004488832e-05,
      "loss": 11.7574,
      "step": 29463
    },
    {
      "epoch": 1.604432555321857,
      "grad_norm": 0.5181474191870714,
      "learning_rate": 1.982377652129409e-05,
      "loss": 11.7387,
      "step": 29464
    },
    {
      "epoch": 1.60448700931844,
      "grad_norm": 0.5370931947527635,
      "learning_rate": 1.981850666156313e-05,
      "loss": 11.7648,
      "step": 29465
    },
    {
      "epoch": 1.6045414633150232,
      "grad_norm": 0.5592638879431615,
      "learning_rate": 1.9813237425336883e-05,
      "loss": 11.8377,
      "step": 29466
    },
    {
      "epoch": 1.6045959173116062,
      "grad_norm": 0.6084962679463427,
      "learning_rate": 1.9807968812656342e-05,
      "loss": 11.7357,
      "step": 29467
    },
    {
      "epoch": 1.6046503713081892,
      "grad_norm": 0.5625876362901717,
      "learning_rate": 1.9802700823562514e-05,
      "loss": 11.8689,
      "step": 29468
    },
    {
      "epoch": 1.6047048253047722,
      "grad_norm": 0.5597426028816667,
      "learning_rate": 1.9797433458096304e-05,
      "loss": 11.825,
      "step": 29469
    },
    {
      "epoch": 1.6047592793013552,
      "grad_norm": 0.5074386905928788,
      "learning_rate": 1.979216671629873e-05,
      "loss": 11.7852,
      "step": 29470
    },
    {
      "epoch": 1.6048137332979382,
      "grad_norm": 0.5922805815533806,
      "learning_rate": 1.97869005982107e-05,
      "loss": 11.8745,
      "step": 29471
    },
    {
      "epoch": 1.6048681872945214,
      "grad_norm": 0.5292795437817838,
      "learning_rate": 1.978163510387321e-05,
      "loss": 11.7987,
      "step": 29472
    },
    {
      "epoch": 1.6049226412911044,
      "grad_norm": 0.5729941697010434,
      "learning_rate": 1.977637023332717e-05,
      "loss": 11.8809,
      "step": 29473
    },
    {
      "epoch": 1.6049770952876874,
      "grad_norm": 0.6235051491056085,
      "learning_rate": 1.9771105986613492e-05,
      "loss": 11.8302,
      "step": 29474
    },
    {
      "epoch": 1.6050315492842704,
      "grad_norm": 0.5094087735677347,
      "learning_rate": 1.976584236377319e-05,
      "loss": 11.7308,
      "step": 29475
    },
    {
      "epoch": 1.6050860032808534,
      "grad_norm": 0.5222837277982497,
      "learning_rate": 1.976057936484712e-05,
      "loss": 11.7958,
      "step": 29476
    },
    {
      "epoch": 1.6051404572774364,
      "grad_norm": 0.5484101553209316,
      "learning_rate": 1.9755316989876273e-05,
      "loss": 11.7901,
      "step": 29477
    },
    {
      "epoch": 1.6051949112740194,
      "grad_norm": 0.5685191554377844,
      "learning_rate": 1.975005523890151e-05,
      "loss": 11.7697,
      "step": 29478
    },
    {
      "epoch": 1.6052493652706024,
      "grad_norm": 0.538431149989638,
      "learning_rate": 1.974479411196377e-05,
      "loss": 11.7302,
      "step": 29479
    },
    {
      "epoch": 1.6053038192671854,
      "grad_norm": 0.5511395883280289,
      "learning_rate": 1.9739533609104e-05,
      "loss": 11.8739,
      "step": 29480
    },
    {
      "epoch": 1.6053582732637683,
      "grad_norm": 0.5394145093542255,
      "learning_rate": 1.973427373036305e-05,
      "loss": 11.7719,
      "step": 29481
    },
    {
      "epoch": 1.6054127272603513,
      "grad_norm": 0.5821856532083591,
      "learning_rate": 1.972901447578188e-05,
      "loss": 11.6762,
      "step": 29482
    },
    {
      "epoch": 1.6054671812569343,
      "grad_norm": 0.554589386980011,
      "learning_rate": 1.9723755845401348e-05,
      "loss": 11.6822,
      "step": 29483
    },
    {
      "epoch": 1.6055216352535173,
      "grad_norm": 0.6017702453644517,
      "learning_rate": 1.9718497839262327e-05,
      "loss": 11.8896,
      "step": 29484
    },
    {
      "epoch": 1.6055760892501003,
      "grad_norm": 0.5876700246333384,
      "learning_rate": 1.9713240457405756e-05,
      "loss": 11.8857,
      "step": 29485
    },
    {
      "epoch": 1.6056305432466833,
      "grad_norm": 0.5051138645146366,
      "learning_rate": 1.9707983699872467e-05,
      "loss": 11.7911,
      "step": 29486
    },
    {
      "epoch": 1.6056849972432663,
      "grad_norm": 0.5804991215949988,
      "learning_rate": 1.9702727566703383e-05,
      "loss": 11.8263,
      "step": 29487
    },
    {
      "epoch": 1.6057394512398493,
      "grad_norm": 0.5762805663177812,
      "learning_rate": 1.9697472057939336e-05,
      "loss": 11.7438,
      "step": 29488
    },
    {
      "epoch": 1.6057939052364323,
      "grad_norm": 0.5301925356628214,
      "learning_rate": 1.969221717362124e-05,
      "loss": 11.8418,
      "step": 29489
    },
    {
      "epoch": 1.6058483592330155,
      "grad_norm": 0.4838981274097415,
      "learning_rate": 1.9686962913789897e-05,
      "loss": 11.8709,
      "step": 29490
    },
    {
      "epoch": 1.6059028132295985,
      "grad_norm": 0.5256166653957722,
      "learning_rate": 1.9681709278486204e-05,
      "loss": 11.7395,
      "step": 29491
    },
    {
      "epoch": 1.6059572672261815,
      "grad_norm": 0.5874138183369193,
      "learning_rate": 1.967645626775103e-05,
      "loss": 11.6483,
      "step": 29492
    },
    {
      "epoch": 1.6060117212227645,
      "grad_norm": 0.619175014361116,
      "learning_rate": 1.9671203881625187e-05,
      "loss": 11.7709,
      "step": 29493
    },
    {
      "epoch": 1.6060661752193475,
      "grad_norm": 0.5397721233361413,
      "learning_rate": 1.9665952120149557e-05,
      "loss": 11.841,
      "step": 29494
    },
    {
      "epoch": 1.6061206292159305,
      "grad_norm": 0.5133565440623458,
      "learning_rate": 1.9660700983364943e-05,
      "loss": 11.635,
      "step": 29495
    },
    {
      "epoch": 1.6061750832125137,
      "grad_norm": 0.5781675428524408,
      "learning_rate": 1.9655450471312176e-05,
      "loss": 11.8864,
      "step": 29496
    },
    {
      "epoch": 1.6062295372090967,
      "grad_norm": 0.5758039603189183,
      "learning_rate": 1.9650200584032118e-05,
      "loss": 11.8107,
      "step": 29497
    },
    {
      "epoch": 1.6062839912056797,
      "grad_norm": 0.5710336443251128,
      "learning_rate": 1.9644951321565564e-05,
      "loss": 11.8368,
      "step": 29498
    },
    {
      "epoch": 1.6063384452022627,
      "grad_norm": 0.6379260319670289,
      "learning_rate": 1.9639702683953355e-05,
      "loss": 11.9761,
      "step": 29499
    },
    {
      "epoch": 1.6063928991988456,
      "grad_norm": 0.5398940733126778,
      "learning_rate": 1.963445467123628e-05,
      "loss": 11.8037,
      "step": 29500
    },
    {
      "epoch": 1.6064473531954286,
      "grad_norm": 0.560128835292989,
      "learning_rate": 1.9629207283455152e-05,
      "loss": 11.8461,
      "step": 29501
    },
    {
      "epoch": 1.6065018071920116,
      "grad_norm": 0.5908217534337917,
      "learning_rate": 1.9623960520650818e-05,
      "loss": 11.8833,
      "step": 29502
    },
    {
      "epoch": 1.6065562611885946,
      "grad_norm": 0.6062157160382261,
      "learning_rate": 1.9618714382864022e-05,
      "loss": 11.9305,
      "step": 29503
    },
    {
      "epoch": 1.6066107151851776,
      "grad_norm": 0.5354432956893059,
      "learning_rate": 1.9613468870135632e-05,
      "loss": 11.7287,
      "step": 29504
    },
    {
      "epoch": 1.6066651691817606,
      "grad_norm": 0.5175288607008623,
      "learning_rate": 1.9608223982506336e-05,
      "loss": 11.7869,
      "step": 29505
    },
    {
      "epoch": 1.6067196231783436,
      "grad_norm": 0.5830635355602947,
      "learning_rate": 1.9602979720016966e-05,
      "loss": 11.9349,
      "step": 29506
    },
    {
      "epoch": 1.6067740771749266,
      "grad_norm": 0.5880066495957835,
      "learning_rate": 1.959773608270835e-05,
      "loss": 11.6871,
      "step": 29507
    },
    {
      "epoch": 1.6068285311715096,
      "grad_norm": 0.5211816113989604,
      "learning_rate": 1.9592493070621186e-05,
      "loss": 11.8174,
      "step": 29508
    },
    {
      "epoch": 1.6068829851680926,
      "grad_norm": 0.5881013403450657,
      "learning_rate": 1.95872506837963e-05,
      "loss": 11.7841,
      "step": 29509
    },
    {
      "epoch": 1.6069374391646756,
      "grad_norm": 0.5116156019886727,
      "learning_rate": 1.9582008922274418e-05,
      "loss": 11.8707,
      "step": 29510
    },
    {
      "epoch": 1.6069918931612586,
      "grad_norm": 0.556690276586919,
      "learning_rate": 1.9576767786096352e-05,
      "loss": 11.7254,
      "step": 29511
    },
    {
      "epoch": 1.6070463471578416,
      "grad_norm": 0.6215590417171741,
      "learning_rate": 1.9571527275302792e-05,
      "loss": 11.905,
      "step": 29512
    },
    {
      "epoch": 1.6071008011544248,
      "grad_norm": 0.5608214710341549,
      "learning_rate": 1.9566287389934535e-05,
      "loss": 11.9164,
      "step": 29513
    },
    {
      "epoch": 1.6071552551510078,
      "grad_norm": 0.5599770801378372,
      "learning_rate": 1.9561048130032357e-05,
      "loss": 11.8274,
      "step": 29514
    },
    {
      "epoch": 1.6072097091475908,
      "grad_norm": 0.531641870382106,
      "learning_rate": 1.9555809495636912e-05,
      "loss": 11.6633,
      "step": 29515
    },
    {
      "epoch": 1.6072641631441738,
      "grad_norm": 0.5398785670105353,
      "learning_rate": 1.9550571486788984e-05,
      "loss": 11.829,
      "step": 29516
    },
    {
      "epoch": 1.6073186171407567,
      "grad_norm": 0.5244099261527798,
      "learning_rate": 1.9545334103529324e-05,
      "loss": 11.7315,
      "step": 29517
    },
    {
      "epoch": 1.6073730711373397,
      "grad_norm": 0.5923949215629317,
      "learning_rate": 1.954009734589861e-05,
      "loss": 11.8393,
      "step": 29518
    },
    {
      "epoch": 1.607427525133923,
      "grad_norm": 0.5406979365669363,
      "learning_rate": 1.9534861213937627e-05,
      "loss": 11.7673,
      "step": 29519
    },
    {
      "epoch": 1.607481979130506,
      "grad_norm": 0.5557786652710994,
      "learning_rate": 1.9529625707687027e-05,
      "loss": 11.8586,
      "step": 29520
    },
    {
      "epoch": 1.607536433127089,
      "grad_norm": 0.5494921665501551,
      "learning_rate": 1.9524390827187577e-05,
      "loss": 11.8828,
      "step": 29521
    },
    {
      "epoch": 1.607590887123672,
      "grad_norm": 0.5415854479649429,
      "learning_rate": 1.9519156572479934e-05,
      "loss": 11.8841,
      "step": 29522
    },
    {
      "epoch": 1.607645341120255,
      "grad_norm": 0.5048770497296734,
      "learning_rate": 1.9513922943604834e-05,
      "loss": 11.8813,
      "step": 29523
    },
    {
      "epoch": 1.607699795116838,
      "grad_norm": 0.4881420841267677,
      "learning_rate": 1.9508689940603008e-05,
      "loss": 11.9356,
      "step": 29524
    },
    {
      "epoch": 1.607754249113421,
      "grad_norm": 0.5381087130994184,
      "learning_rate": 1.950345756351506e-05,
      "loss": 11.8552,
      "step": 29525
    },
    {
      "epoch": 1.607808703110004,
      "grad_norm": 0.5556975514909981,
      "learning_rate": 1.9498225812381755e-05,
      "loss": 11.7983,
      "step": 29526
    },
    {
      "epoch": 1.607863157106587,
      "grad_norm": 0.5920559232220721,
      "learning_rate": 1.9492994687243714e-05,
      "loss": 11.876,
      "step": 29527
    },
    {
      "epoch": 1.6079176111031699,
      "grad_norm": 0.5737121249998529,
      "learning_rate": 1.9487764188141655e-05,
      "loss": 11.7988,
      "step": 29528
    },
    {
      "epoch": 1.6079720650997529,
      "grad_norm": 0.5837325675912988,
      "learning_rate": 1.948253431511626e-05,
      "loss": 11.8061,
      "step": 29529
    },
    {
      "epoch": 1.6080265190963359,
      "grad_norm": 0.5208751115569004,
      "learning_rate": 1.9477305068208163e-05,
      "loss": 11.8655,
      "step": 29530
    },
    {
      "epoch": 1.6080809730929189,
      "grad_norm": 0.5807413749379425,
      "learning_rate": 1.9472076447458064e-05,
      "loss": 11.8194,
      "step": 29531
    },
    {
      "epoch": 1.6081354270895019,
      "grad_norm": 0.5780867307509088,
      "learning_rate": 1.946684845290658e-05,
      "loss": 11.748,
      "step": 29532
    },
    {
      "epoch": 1.6081898810860848,
      "grad_norm": 0.5718270503096778,
      "learning_rate": 1.9461621084594418e-05,
      "loss": 11.6538,
      "step": 29533
    },
    {
      "epoch": 1.6082443350826678,
      "grad_norm": 0.5256356144512199,
      "learning_rate": 1.945639434256219e-05,
      "loss": 11.8155,
      "step": 29534
    },
    {
      "epoch": 1.6082987890792508,
      "grad_norm": 0.5758285504641771,
      "learning_rate": 1.9451168226850524e-05,
      "loss": 11.7021,
      "step": 29535
    },
    {
      "epoch": 1.608353243075834,
      "grad_norm": 0.5559875475881851,
      "learning_rate": 1.9445942737500113e-05,
      "loss": 11.8845,
      "step": 29536
    },
    {
      "epoch": 1.608407697072417,
      "grad_norm": 0.534015103734107,
      "learning_rate": 1.9440717874551528e-05,
      "loss": 11.883,
      "step": 29537
    },
    {
      "epoch": 1.608462151069,
      "grad_norm": 0.5331950711834742,
      "learning_rate": 1.9435493638045455e-05,
      "loss": 11.7914,
      "step": 29538
    },
    {
      "epoch": 1.608516605065583,
      "grad_norm": 0.5458694608076182,
      "learning_rate": 1.943027002802247e-05,
      "loss": 11.725,
      "step": 29539
    },
    {
      "epoch": 1.608571059062166,
      "grad_norm": 0.5828304155005862,
      "learning_rate": 1.9425047044523226e-05,
      "loss": 11.8955,
      "step": 29540
    },
    {
      "epoch": 1.608625513058749,
      "grad_norm": 0.5225958157818917,
      "learning_rate": 1.941982468758834e-05,
      "loss": 11.7342,
      "step": 29541
    },
    {
      "epoch": 1.6086799670553322,
      "grad_norm": 0.5496488900618468,
      "learning_rate": 1.941460295725839e-05,
      "loss": 11.8381,
      "step": 29542
    },
    {
      "epoch": 1.6087344210519152,
      "grad_norm": 0.5062961263757868,
      "learning_rate": 1.9409381853574026e-05,
      "loss": 11.8887,
      "step": 29543
    },
    {
      "epoch": 1.6087888750484982,
      "grad_norm": 0.5302396278303994,
      "learning_rate": 1.9404161376575835e-05,
      "loss": 11.813,
      "step": 29544
    },
    {
      "epoch": 1.6088433290450812,
      "grad_norm": 0.49799933580786143,
      "learning_rate": 1.939894152630436e-05,
      "loss": 11.8171,
      "step": 29545
    },
    {
      "epoch": 1.6088977830416642,
      "grad_norm": 0.5289288469149018,
      "learning_rate": 1.9393722302800266e-05,
      "loss": 11.7155,
      "step": 29546
    },
    {
      "epoch": 1.6089522370382472,
      "grad_norm": 0.6175431664817955,
      "learning_rate": 1.9388503706104077e-05,
      "loss": 11.8465,
      "step": 29547
    },
    {
      "epoch": 1.6090066910348302,
      "grad_norm": 0.5713663245887997,
      "learning_rate": 1.938328573625643e-05,
      "loss": 11.9048,
      "step": 29548
    },
    {
      "epoch": 1.6090611450314132,
      "grad_norm": 0.7576427256232343,
      "learning_rate": 1.9378068393297844e-05,
      "loss": 11.9706,
      "step": 29549
    },
    {
      "epoch": 1.6091155990279962,
      "grad_norm": 0.5443327726839043,
      "learning_rate": 1.937285167726892e-05,
      "loss": 11.9013,
      "step": 29550
    },
    {
      "epoch": 1.6091700530245792,
      "grad_norm": 0.5647505781259433,
      "learning_rate": 1.9367635588210253e-05,
      "loss": 11.9367,
      "step": 29551
    },
    {
      "epoch": 1.6092245070211622,
      "grad_norm": 0.5655720292030478,
      "learning_rate": 1.936242012616234e-05,
      "loss": 11.8224,
      "step": 29552
    },
    {
      "epoch": 1.6092789610177451,
      "grad_norm": 0.5978227737092479,
      "learning_rate": 1.9357205291165802e-05,
      "loss": 11.8153,
      "step": 29553
    },
    {
      "epoch": 1.6093334150143281,
      "grad_norm": 0.5091386406530556,
      "learning_rate": 1.9351991083261156e-05,
      "loss": 11.828,
      "step": 29554
    },
    {
      "epoch": 1.6093878690109111,
      "grad_norm": 0.5762205756489719,
      "learning_rate": 1.9346777502488923e-05,
      "loss": 11.8939,
      "step": 29555
    },
    {
      "epoch": 1.6094423230074941,
      "grad_norm": 0.5396369529761386,
      "learning_rate": 1.9341564548889713e-05,
      "loss": 11.5902,
      "step": 29556
    },
    {
      "epoch": 1.6094967770040771,
      "grad_norm": 0.5152461717621627,
      "learning_rate": 1.9336352222503984e-05,
      "loss": 11.713,
      "step": 29557
    },
    {
      "epoch": 1.60955123100066,
      "grad_norm": 0.5538421050177275,
      "learning_rate": 1.933114052337234e-05,
      "loss": 11.8857,
      "step": 29558
    },
    {
      "epoch": 1.609605684997243,
      "grad_norm": 0.5159800025316812,
      "learning_rate": 1.932592945153524e-05,
      "loss": 11.8314,
      "step": 29559
    },
    {
      "epoch": 1.6096601389938263,
      "grad_norm": 0.5413113498923491,
      "learning_rate": 1.9320719007033282e-05,
      "loss": 11.8667,
      "step": 29560
    },
    {
      "epoch": 1.6097145929904093,
      "grad_norm": 0.5531103888116603,
      "learning_rate": 1.931550918990691e-05,
      "loss": 11.7705,
      "step": 29561
    },
    {
      "epoch": 1.6097690469869923,
      "grad_norm": 0.5782706319008367,
      "learning_rate": 1.9310300000196668e-05,
      "loss": 11.926,
      "step": 29562
    },
    {
      "epoch": 1.6098235009835753,
      "grad_norm": 0.5584620417309196,
      "learning_rate": 1.930509143794309e-05,
      "loss": 11.8614,
      "step": 29563
    },
    {
      "epoch": 1.6098779549801583,
      "grad_norm": 0.5381583087166442,
      "learning_rate": 1.9299883503186656e-05,
      "loss": 11.7382,
      "step": 29564
    },
    {
      "epoch": 1.6099324089767415,
      "grad_norm": 0.5125804308240364,
      "learning_rate": 1.9294676195967833e-05,
      "loss": 11.8325,
      "step": 29565
    },
    {
      "epoch": 1.6099868629733245,
      "grad_norm": 0.5481157917892993,
      "learning_rate": 1.9289469516327175e-05,
      "loss": 11.838,
      "step": 29566
    },
    {
      "epoch": 1.6100413169699075,
      "grad_norm": 0.5936471282787849,
      "learning_rate": 1.9284263464305108e-05,
      "loss": 11.9087,
      "step": 29567
    },
    {
      "epoch": 1.6100957709664905,
      "grad_norm": 0.5251934267782864,
      "learning_rate": 1.9279058039942165e-05,
      "loss": 11.8108,
      "step": 29568
    },
    {
      "epoch": 1.6101502249630735,
      "grad_norm": 0.6441650840009321,
      "learning_rate": 1.9273853243278782e-05,
      "loss": 11.9299,
      "step": 29569
    },
    {
      "epoch": 1.6102046789596565,
      "grad_norm": 0.5996784450447684,
      "learning_rate": 1.9268649074355484e-05,
      "loss": 11.8155,
      "step": 29570
    },
    {
      "epoch": 1.6102591329562395,
      "grad_norm": 0.5240673711257662,
      "learning_rate": 1.9263445533212677e-05,
      "loss": 11.8043,
      "step": 29571
    },
    {
      "epoch": 1.6103135869528225,
      "grad_norm": 0.5922263540896808,
      "learning_rate": 1.9258242619890854e-05,
      "loss": 11.8052,
      "step": 29572
    },
    {
      "epoch": 1.6103680409494054,
      "grad_norm": 0.5400260135135027,
      "learning_rate": 1.9253040334430505e-05,
      "loss": 11.7609,
      "step": 29573
    },
    {
      "epoch": 1.6104224949459884,
      "grad_norm": 0.5622988521635481,
      "learning_rate": 1.9247838676872064e-05,
      "loss": 11.7748,
      "step": 29574
    },
    {
      "epoch": 1.6104769489425714,
      "grad_norm": 0.47277871171211544,
      "learning_rate": 1.9242637647255967e-05,
      "loss": 11.7018,
      "step": 29575
    },
    {
      "epoch": 1.6105314029391544,
      "grad_norm": 0.5598215265791844,
      "learning_rate": 1.9237437245622635e-05,
      "loss": 11.977,
      "step": 29576
    },
    {
      "epoch": 1.6105858569357374,
      "grad_norm": 0.5066079895315062,
      "learning_rate": 1.923223747201254e-05,
      "loss": 11.8769,
      "step": 29577
    },
    {
      "epoch": 1.6106403109323204,
      "grad_norm": 0.5737905323530964,
      "learning_rate": 1.922703832646613e-05,
      "loss": 11.8221,
      "step": 29578
    },
    {
      "epoch": 1.6106947649289034,
      "grad_norm": 0.5371741190676275,
      "learning_rate": 1.922183980902379e-05,
      "loss": 11.8263,
      "step": 29579
    },
    {
      "epoch": 1.6107492189254864,
      "grad_norm": 0.5470637761312328,
      "learning_rate": 1.9216641919725996e-05,
      "loss": 11.829,
      "step": 29580
    },
    {
      "epoch": 1.6108036729220694,
      "grad_norm": 0.5755290994175941,
      "learning_rate": 1.9211444658613122e-05,
      "loss": 11.8048,
      "step": 29581
    },
    {
      "epoch": 1.6108581269186524,
      "grad_norm": 0.6298942677106962,
      "learning_rate": 1.9206248025725614e-05,
      "loss": 11.9161,
      "step": 29582
    },
    {
      "epoch": 1.6109125809152356,
      "grad_norm": 0.5313163553259902,
      "learning_rate": 1.920105202110386e-05,
      "loss": 11.9056,
      "step": 29583
    },
    {
      "epoch": 1.6109670349118186,
      "grad_norm": 0.5472964463557821,
      "learning_rate": 1.91958566447883e-05,
      "loss": 11.905,
      "step": 29584
    },
    {
      "epoch": 1.6110214889084016,
      "grad_norm": 0.5912321516673279,
      "learning_rate": 1.9190661896819307e-05,
      "loss": 11.8834,
      "step": 29585
    },
    {
      "epoch": 1.6110759429049846,
      "grad_norm": 0.6536873624048212,
      "learning_rate": 1.9185467777237254e-05,
      "loss": 11.923,
      "step": 29586
    },
    {
      "epoch": 1.6111303969015676,
      "grad_norm": 0.5090252953567536,
      "learning_rate": 1.9180274286082578e-05,
      "loss": 11.833,
      "step": 29587
    },
    {
      "epoch": 1.6111848508981506,
      "grad_norm": 0.5001710337858857,
      "learning_rate": 1.9175081423395612e-05,
      "loss": 11.8627,
      "step": 29588
    },
    {
      "epoch": 1.6112393048947338,
      "grad_norm": 0.5291093489575781,
      "learning_rate": 1.9169889189216783e-05,
      "loss": 11.8392,
      "step": 29589
    },
    {
      "epoch": 1.6112937588913168,
      "grad_norm": 0.5202906696704613,
      "learning_rate": 1.9164697583586468e-05,
      "loss": 11.776,
      "step": 29590
    },
    {
      "epoch": 1.6113482128878998,
      "grad_norm": 0.5246389746290668,
      "learning_rate": 1.9159506606544984e-05,
      "loss": 11.8681,
      "step": 29591
    },
    {
      "epoch": 1.6114026668844827,
      "grad_norm": 0.4898275564259432,
      "learning_rate": 1.9154316258132777e-05,
      "loss": 11.6848,
      "step": 29592
    },
    {
      "epoch": 1.6114571208810657,
      "grad_norm": 0.5413377027556796,
      "learning_rate": 1.9149126538390127e-05,
      "loss": 11.8868,
      "step": 29593
    },
    {
      "epoch": 1.6115115748776487,
      "grad_norm": 0.5192466163693963,
      "learning_rate": 1.9143937447357462e-05,
      "loss": 11.761,
      "step": 29594
    },
    {
      "epoch": 1.6115660288742317,
      "grad_norm": 0.5587969031217869,
      "learning_rate": 1.9138748985075094e-05,
      "loss": 11.8258,
      "step": 29595
    },
    {
      "epoch": 1.6116204828708147,
      "grad_norm": 0.5353248257573864,
      "learning_rate": 1.9133561151583355e-05,
      "loss": 11.7902,
      "step": 29596
    },
    {
      "epoch": 1.6116749368673977,
      "grad_norm": 0.5227399056629031,
      "learning_rate": 1.912837394692262e-05,
      "loss": 11.7976,
      "step": 29597
    },
    {
      "epoch": 1.6117293908639807,
      "grad_norm": 0.5508571534806835,
      "learning_rate": 1.91231873711332e-05,
      "loss": 11.6627,
      "step": 29598
    },
    {
      "epoch": 1.6117838448605637,
      "grad_norm": 0.5102923687347977,
      "learning_rate": 1.9118001424255427e-05,
      "loss": 11.8478,
      "step": 29599
    },
    {
      "epoch": 1.6118382988571467,
      "grad_norm": 0.531953171103753,
      "learning_rate": 1.9112816106329667e-05,
      "loss": 11.7117,
      "step": 29600
    },
    {
      "epoch": 1.6118927528537297,
      "grad_norm": 0.5297788108626057,
      "learning_rate": 1.9107631417396188e-05,
      "loss": 11.8004,
      "step": 29601
    },
    {
      "epoch": 1.6119472068503127,
      "grad_norm": 0.7068541638325369,
      "learning_rate": 1.9102447357495355e-05,
      "loss": 11.7185,
      "step": 29602
    },
    {
      "epoch": 1.6120016608468957,
      "grad_norm": 0.49332287462322977,
      "learning_rate": 1.909726392666744e-05,
      "loss": 11.8006,
      "step": 29603
    },
    {
      "epoch": 1.6120561148434787,
      "grad_norm": 0.5468416721993087,
      "learning_rate": 1.9092081124952787e-05,
      "loss": 11.7957,
      "step": 29604
    },
    {
      "epoch": 1.6121105688400617,
      "grad_norm": 0.5464211206930017,
      "learning_rate": 1.9086898952391686e-05,
      "loss": 11.7933,
      "step": 29605
    },
    {
      "epoch": 1.6121650228366449,
      "grad_norm": 0.5639125354698105,
      "learning_rate": 1.90817174090244e-05,
      "loss": 11.8746,
      "step": 29606
    },
    {
      "epoch": 1.6122194768332279,
      "grad_norm": 0.5275710519658461,
      "learning_rate": 1.9076536494891272e-05,
      "loss": 11.8444,
      "step": 29607
    },
    {
      "epoch": 1.6122739308298109,
      "grad_norm": 0.5395387033152028,
      "learning_rate": 1.9071356210032544e-05,
      "loss": 11.7913,
      "step": 29608
    },
    {
      "epoch": 1.6123283848263938,
      "grad_norm": 0.627910051876697,
      "learning_rate": 1.9066176554488545e-05,
      "loss": 12.0437,
      "step": 29609
    },
    {
      "epoch": 1.6123828388229768,
      "grad_norm": 0.6087750569054463,
      "learning_rate": 1.9060997528299505e-05,
      "loss": 11.8598,
      "step": 29610
    },
    {
      "epoch": 1.6124372928195598,
      "grad_norm": 0.5898224923045239,
      "learning_rate": 1.9055819131505724e-05,
      "loss": 11.8917,
      "step": 29611
    },
    {
      "epoch": 1.612491746816143,
      "grad_norm": 0.5226180741584564,
      "learning_rate": 1.9050641364147493e-05,
      "loss": 11.7502,
      "step": 29612
    },
    {
      "epoch": 1.612546200812726,
      "grad_norm": 0.5586843605855791,
      "learning_rate": 1.904546422626502e-05,
      "loss": 11.7246,
      "step": 29613
    },
    {
      "epoch": 1.612600654809309,
      "grad_norm": 0.5259545980686726,
      "learning_rate": 1.904028771789863e-05,
      "loss": 11.823,
      "step": 29614
    },
    {
      "epoch": 1.612655108805892,
      "grad_norm": 0.5658945433406595,
      "learning_rate": 1.9035111839088527e-05,
      "loss": 11.8673,
      "step": 29615
    },
    {
      "epoch": 1.612709562802475,
      "grad_norm": 0.5527330441003985,
      "learning_rate": 1.902993658987495e-05,
      "loss": 11.7979,
      "step": 29616
    },
    {
      "epoch": 1.612764016799058,
      "grad_norm": 0.5096908852313172,
      "learning_rate": 1.9024761970298187e-05,
      "loss": 11.7417,
      "step": 29617
    },
    {
      "epoch": 1.612818470795641,
      "grad_norm": 0.5566952959549946,
      "learning_rate": 1.9019587980398445e-05,
      "loss": 11.9029,
      "step": 29618
    },
    {
      "epoch": 1.612872924792224,
      "grad_norm": 0.5164784893964677,
      "learning_rate": 1.901441462021598e-05,
      "loss": 11.7604,
      "step": 29619
    },
    {
      "epoch": 1.612927378788807,
      "grad_norm": 0.505202894194634,
      "learning_rate": 1.9009241889790984e-05,
      "loss": 11.8244,
      "step": 29620
    },
    {
      "epoch": 1.61298183278539,
      "grad_norm": 0.49931685006218834,
      "learning_rate": 1.9004069789163702e-05,
      "loss": 11.7195,
      "step": 29621
    },
    {
      "epoch": 1.613036286781973,
      "grad_norm": 0.5455425210517945,
      "learning_rate": 1.8998898318374382e-05,
      "loss": 11.7581,
      "step": 29622
    },
    {
      "epoch": 1.613090740778556,
      "grad_norm": 0.5432936625568962,
      "learning_rate": 1.8993727477463197e-05,
      "loss": 11.8066,
      "step": 29623
    },
    {
      "epoch": 1.613145194775139,
      "grad_norm": 0.5828583096469647,
      "learning_rate": 1.8988557266470408e-05,
      "loss": 11.8331,
      "step": 29624
    },
    {
      "epoch": 1.613199648771722,
      "grad_norm": 0.526464711439938,
      "learning_rate": 1.898338768543615e-05,
      "loss": 11.8147,
      "step": 29625
    },
    {
      "epoch": 1.613254102768305,
      "grad_norm": 0.51089263905331,
      "learning_rate": 1.8978218734400645e-05,
      "loss": 11.8426,
      "step": 29626
    },
    {
      "epoch": 1.613308556764888,
      "grad_norm": 0.5337080174071616,
      "learning_rate": 1.8973050413404126e-05,
      "loss": 11.8138,
      "step": 29627
    },
    {
      "epoch": 1.613363010761471,
      "grad_norm": 0.6126107910181418,
      "learning_rate": 1.896788272248673e-05,
      "loss": 11.8685,
      "step": 29628
    },
    {
      "epoch": 1.613417464758054,
      "grad_norm": 0.5444234316369566,
      "learning_rate": 1.896271566168869e-05,
      "loss": 11.7673,
      "step": 29629
    },
    {
      "epoch": 1.6134719187546371,
      "grad_norm": 0.6123301985222828,
      "learning_rate": 1.895754923105013e-05,
      "loss": 11.7898,
      "step": 29630
    },
    {
      "epoch": 1.6135263727512201,
      "grad_norm": 0.5254807159606458,
      "learning_rate": 1.8952383430611298e-05,
      "loss": 11.8347,
      "step": 29631
    },
    {
      "epoch": 1.6135808267478031,
      "grad_norm": 0.5005848331393221,
      "learning_rate": 1.894721826041229e-05,
      "loss": 11.8329,
      "step": 29632
    },
    {
      "epoch": 1.6136352807443861,
      "grad_norm": 0.5490648289902342,
      "learning_rate": 1.8942053720493314e-05,
      "loss": 11.8345,
      "step": 29633
    },
    {
      "epoch": 1.613689734740969,
      "grad_norm": 0.5419209812540147,
      "learning_rate": 1.8936889810894542e-05,
      "loss": 11.8052,
      "step": 29634
    },
    {
      "epoch": 1.6137441887375523,
      "grad_norm": 0.5832089670673982,
      "learning_rate": 1.89317265316561e-05,
      "loss": 11.6428,
      "step": 29635
    },
    {
      "epoch": 1.6137986427341353,
      "grad_norm": 0.553729642072451,
      "learning_rate": 1.892656388281816e-05,
      "loss": 11.9131,
      "step": 29636
    },
    {
      "epoch": 1.6138530967307183,
      "grad_norm": 0.5502991842329528,
      "learning_rate": 1.892140186442083e-05,
      "loss": 11.8427,
      "step": 29637
    },
    {
      "epoch": 1.6139075507273013,
      "grad_norm": 0.555463008611924,
      "learning_rate": 1.8916240476504276e-05,
      "loss": 11.9569,
      "step": 29638
    },
    {
      "epoch": 1.6139620047238843,
      "grad_norm": 0.6543402161789428,
      "learning_rate": 1.891107971910865e-05,
      "loss": 11.9743,
      "step": 29639
    },
    {
      "epoch": 1.6140164587204673,
      "grad_norm": 0.5126772533103994,
      "learning_rate": 1.8905919592274046e-05,
      "loss": 11.5509,
      "step": 29640
    },
    {
      "epoch": 1.6140709127170503,
      "grad_norm": 0.559791595103359,
      "learning_rate": 1.890076009604064e-05,
      "loss": 11.872,
      "step": 29641
    },
    {
      "epoch": 1.6141253667136333,
      "grad_norm": 0.6019338330864715,
      "learning_rate": 1.8895601230448502e-05,
      "loss": 11.8674,
      "step": 29642
    },
    {
      "epoch": 1.6141798207102163,
      "grad_norm": 0.5680816470425887,
      "learning_rate": 1.8890442995537782e-05,
      "loss": 11.8414,
      "step": 29643
    },
    {
      "epoch": 1.6142342747067993,
      "grad_norm": 0.5496959065861055,
      "learning_rate": 1.888528539134856e-05,
      "loss": 11.805,
      "step": 29644
    },
    {
      "epoch": 1.6142887287033822,
      "grad_norm": 0.5399562832626262,
      "learning_rate": 1.8880128417920993e-05,
      "loss": 11.8105,
      "step": 29645
    },
    {
      "epoch": 1.6143431826999652,
      "grad_norm": 0.5282321670642584,
      "learning_rate": 1.887497207529514e-05,
      "loss": 11.7994,
      "step": 29646
    },
    {
      "epoch": 1.6143976366965482,
      "grad_norm": 0.5414942430792258,
      "learning_rate": 1.8869816363511095e-05,
      "loss": 11.7747,
      "step": 29647
    },
    {
      "epoch": 1.6144520906931312,
      "grad_norm": 0.5670112969663105,
      "learning_rate": 1.8864661282608954e-05,
      "loss": 11.8084,
      "step": 29648
    },
    {
      "epoch": 1.6145065446897142,
      "grad_norm": 0.543724242751264,
      "learning_rate": 1.8859506832628848e-05,
      "loss": 11.6512,
      "step": 29649
    },
    {
      "epoch": 1.6145609986862972,
      "grad_norm": 0.6046962502639556,
      "learning_rate": 1.885435301361079e-05,
      "loss": 11.881,
      "step": 29650
    },
    {
      "epoch": 1.6146154526828802,
      "grad_norm": 0.5295983012501382,
      "learning_rate": 1.8849199825594922e-05,
      "loss": 11.6617,
      "step": 29651
    },
    {
      "epoch": 1.6146699066794632,
      "grad_norm": 0.5356367137513399,
      "learning_rate": 1.8844047268621256e-05,
      "loss": 11.8382,
      "step": 29652
    },
    {
      "epoch": 1.6147243606760464,
      "grad_norm": 0.5692716482021974,
      "learning_rate": 1.8838895342729913e-05,
      "loss": 11.7674,
      "step": 29653
    },
    {
      "epoch": 1.6147788146726294,
      "grad_norm": 0.5406897308559924,
      "learning_rate": 1.8833744047960912e-05,
      "loss": 11.832,
      "step": 29654
    },
    {
      "epoch": 1.6148332686692124,
      "grad_norm": 0.5131634859804876,
      "learning_rate": 1.882859338435434e-05,
      "loss": 11.7801,
      "step": 29655
    },
    {
      "epoch": 1.6148877226657954,
      "grad_norm": 0.5414352540346842,
      "learning_rate": 1.8823443351950254e-05,
      "loss": 11.8916,
      "step": 29656
    },
    {
      "epoch": 1.6149421766623784,
      "grad_norm": 0.6039948249554931,
      "learning_rate": 1.8818293950788656e-05,
      "loss": 11.8279,
      "step": 29657
    },
    {
      "epoch": 1.6149966306589614,
      "grad_norm": 0.5631411197719803,
      "learning_rate": 1.881314518090964e-05,
      "loss": 11.771,
      "step": 29658
    },
    {
      "epoch": 1.6150510846555446,
      "grad_norm": 0.5250399020585735,
      "learning_rate": 1.8807997042353198e-05,
      "loss": 11.7496,
      "step": 29659
    },
    {
      "epoch": 1.6151055386521276,
      "grad_norm": 0.559565163232491,
      "learning_rate": 1.8802849535159393e-05,
      "loss": 11.8464,
      "step": 29660
    },
    {
      "epoch": 1.6151599926487106,
      "grad_norm": 0.5732934200833762,
      "learning_rate": 1.8797702659368265e-05,
      "loss": 11.8802,
      "step": 29661
    },
    {
      "epoch": 1.6152144466452936,
      "grad_norm": 0.5437281490005131,
      "learning_rate": 1.879255641501979e-05,
      "loss": 11.8055,
      "step": 29662
    },
    {
      "epoch": 1.6152689006418766,
      "grad_norm": 0.5146443961078804,
      "learning_rate": 1.8787410802154048e-05,
      "loss": 11.7519,
      "step": 29663
    },
    {
      "epoch": 1.6153233546384596,
      "grad_norm": 0.5901833914398004,
      "learning_rate": 1.8782265820810995e-05,
      "loss": 11.824,
      "step": 29664
    },
    {
      "epoch": 1.6153778086350425,
      "grad_norm": 0.49409431228392275,
      "learning_rate": 1.877712147103069e-05,
      "loss": 11.8383,
      "step": 29665
    },
    {
      "epoch": 1.6154322626316255,
      "grad_norm": 0.5510657008505888,
      "learning_rate": 1.87719777528531e-05,
      "loss": 11.843,
      "step": 29666
    },
    {
      "epoch": 1.6154867166282085,
      "grad_norm": 0.5682934847729291,
      "learning_rate": 1.876683466631821e-05,
      "loss": 11.8976,
      "step": 29667
    },
    {
      "epoch": 1.6155411706247915,
      "grad_norm": 0.5586429230938673,
      "learning_rate": 1.8761692211466063e-05,
      "loss": 11.6376,
      "step": 29668
    },
    {
      "epoch": 1.6155956246213745,
      "grad_norm": 0.5575927369629163,
      "learning_rate": 1.8756550388336603e-05,
      "loss": 11.8644,
      "step": 29669
    },
    {
      "epoch": 1.6156500786179575,
      "grad_norm": 0.5522321182844298,
      "learning_rate": 1.8751409196969817e-05,
      "loss": 11.8512,
      "step": 29670
    },
    {
      "epoch": 1.6157045326145405,
      "grad_norm": 0.5691194886311696,
      "learning_rate": 1.874626863740574e-05,
      "loss": 11.7803,
      "step": 29671
    },
    {
      "epoch": 1.6157589866111235,
      "grad_norm": 0.5387213643850569,
      "learning_rate": 1.8741128709684264e-05,
      "loss": 11.8652,
      "step": 29672
    },
    {
      "epoch": 1.6158134406077065,
      "grad_norm": 0.5629550626198992,
      "learning_rate": 1.8735989413845432e-05,
      "loss": 11.8926,
      "step": 29673
    },
    {
      "epoch": 1.6158678946042895,
      "grad_norm": 0.5400894612339091,
      "learning_rate": 1.873085074992915e-05,
      "loss": 11.6906,
      "step": 29674
    },
    {
      "epoch": 1.6159223486008725,
      "grad_norm": 0.5797810667066575,
      "learning_rate": 1.8725712717975418e-05,
      "loss": 11.8288,
      "step": 29675
    },
    {
      "epoch": 1.6159768025974557,
      "grad_norm": 0.583512170594587,
      "learning_rate": 1.872057531802419e-05,
      "loss": 11.9047,
      "step": 29676
    },
    {
      "epoch": 1.6160312565940387,
      "grad_norm": 0.5796086398446647,
      "learning_rate": 1.871543855011536e-05,
      "loss": 11.8306,
      "step": 29677
    },
    {
      "epoch": 1.6160857105906217,
      "grad_norm": 0.6026315869755731,
      "learning_rate": 1.871030241428894e-05,
      "loss": 11.9264,
      "step": 29678
    },
    {
      "epoch": 1.6161401645872047,
      "grad_norm": 0.5549657449014572,
      "learning_rate": 1.8705166910584815e-05,
      "loss": 11.8103,
      "step": 29679
    },
    {
      "epoch": 1.6161946185837877,
      "grad_norm": 0.5021729527965968,
      "learning_rate": 1.870003203904297e-05,
      "loss": 11.7316,
      "step": 29680
    },
    {
      "epoch": 1.6162490725803706,
      "grad_norm": 0.5175615033962297,
      "learning_rate": 1.8694897799703282e-05,
      "loss": 11.9144,
      "step": 29681
    },
    {
      "epoch": 1.6163035265769539,
      "grad_norm": 0.5262945184296868,
      "learning_rate": 1.8689764192605707e-05,
      "loss": 11.8049,
      "step": 29682
    },
    {
      "epoch": 1.6163579805735369,
      "grad_norm": 0.4973390433921258,
      "learning_rate": 1.8684631217790194e-05,
      "loss": 11.8239,
      "step": 29683
    },
    {
      "epoch": 1.6164124345701198,
      "grad_norm": 0.5495698019580566,
      "learning_rate": 1.8679498875296588e-05,
      "loss": 11.808,
      "step": 29684
    },
    {
      "epoch": 1.6164668885667028,
      "grad_norm": 0.5129640458517766,
      "learning_rate": 1.8674367165164898e-05,
      "loss": 11.8014,
      "step": 29685
    },
    {
      "epoch": 1.6165213425632858,
      "grad_norm": 0.5501871845800342,
      "learning_rate": 1.8669236087434904e-05,
      "loss": 11.7824,
      "step": 29686
    },
    {
      "epoch": 1.6165757965598688,
      "grad_norm": 0.5263849480792924,
      "learning_rate": 1.866410564214658e-05,
      "loss": 11.651,
      "step": 29687
    },
    {
      "epoch": 1.6166302505564518,
      "grad_norm": 0.5654000764992241,
      "learning_rate": 1.8658975829339832e-05,
      "loss": 11.6673,
      "step": 29688
    },
    {
      "epoch": 1.6166847045530348,
      "grad_norm": 0.5329454968180634,
      "learning_rate": 1.8653846649054497e-05,
      "loss": 11.7625,
      "step": 29689
    },
    {
      "epoch": 1.6167391585496178,
      "grad_norm": 0.5874770734532969,
      "learning_rate": 1.864871810133053e-05,
      "loss": 11.8525,
      "step": 29690
    },
    {
      "epoch": 1.6167936125462008,
      "grad_norm": 0.5889592799692791,
      "learning_rate": 1.8643590186207737e-05,
      "loss": 11.8978,
      "step": 29691
    },
    {
      "epoch": 1.6168480665427838,
      "grad_norm": 0.5247054306891863,
      "learning_rate": 1.8638462903726062e-05,
      "loss": 11.7815,
      "step": 29692
    },
    {
      "epoch": 1.6169025205393668,
      "grad_norm": 0.5405937891158469,
      "learning_rate": 1.8633336253925314e-05,
      "loss": 11.8829,
      "step": 29693
    },
    {
      "epoch": 1.6169569745359498,
      "grad_norm": 0.5864781772542362,
      "learning_rate": 1.862821023684539e-05,
      "loss": 11.8152,
      "step": 29694
    },
    {
      "epoch": 1.6170114285325328,
      "grad_norm": 0.5940501313197909,
      "learning_rate": 1.8623084852526208e-05,
      "loss": 11.9119,
      "step": 29695
    },
    {
      "epoch": 1.6170658825291158,
      "grad_norm": 0.5226921153191229,
      "learning_rate": 1.8617960101007504e-05,
      "loss": 11.761,
      "step": 29696
    },
    {
      "epoch": 1.6171203365256988,
      "grad_norm": 0.5452004933781307,
      "learning_rate": 1.861283598232919e-05,
      "loss": 11.7848,
      "step": 29697
    },
    {
      "epoch": 1.6171747905222817,
      "grad_norm": 0.5318178540595372,
      "learning_rate": 1.8607712496531148e-05,
      "loss": 11.7976,
      "step": 29698
    },
    {
      "epoch": 1.617229244518865,
      "grad_norm": 0.5477669497775027,
      "learning_rate": 1.8602589643653144e-05,
      "loss": 11.8655,
      "step": 29699
    },
    {
      "epoch": 1.617283698515448,
      "grad_norm": 0.5344043123632418,
      "learning_rate": 1.8597467423735092e-05,
      "loss": 11.8005,
      "step": 29700
    },
    {
      "epoch": 1.617338152512031,
      "grad_norm": 0.5575845681308145,
      "learning_rate": 1.859234583681675e-05,
      "loss": 11.852,
      "step": 29701
    },
    {
      "epoch": 1.617392606508614,
      "grad_norm": 0.5262517809572043,
      "learning_rate": 1.8587224882938003e-05,
      "loss": 11.78,
      "step": 29702
    },
    {
      "epoch": 1.617447060505197,
      "grad_norm": 0.5882528780569757,
      "learning_rate": 1.8582104562138637e-05,
      "loss": 11.838,
      "step": 29703
    },
    {
      "epoch": 1.61750151450178,
      "grad_norm": 0.6035131653248957,
      "learning_rate": 1.857698487445846e-05,
      "loss": 11.8683,
      "step": 29704
    },
    {
      "epoch": 1.6175559684983631,
      "grad_norm": 0.5144802107133795,
      "learning_rate": 1.857186581993737e-05,
      "loss": 11.7094,
      "step": 29705
    },
    {
      "epoch": 1.6176104224949461,
      "grad_norm": 0.5667786980611123,
      "learning_rate": 1.8566747398615048e-05,
      "loss": 11.8018,
      "step": 29706
    },
    {
      "epoch": 1.6176648764915291,
      "grad_norm": 0.5400424489562431,
      "learning_rate": 1.8561629610531385e-05,
      "loss": 11.8653,
      "step": 29707
    },
    {
      "epoch": 1.6177193304881121,
      "grad_norm": 0.563215643402406,
      "learning_rate": 1.8556512455726115e-05,
      "loss": 11.9103,
      "step": 29708
    },
    {
      "epoch": 1.617773784484695,
      "grad_norm": 0.5516286598899117,
      "learning_rate": 1.8551395934239068e-05,
      "loss": 11.7368,
      "step": 29709
    },
    {
      "epoch": 1.617828238481278,
      "grad_norm": 0.5328982898114879,
      "learning_rate": 1.8546280046110042e-05,
      "loss": 11.7824,
      "step": 29710
    },
    {
      "epoch": 1.617882692477861,
      "grad_norm": 0.5870227609709985,
      "learning_rate": 1.8541164791378786e-05,
      "loss": 11.9064,
      "step": 29711
    },
    {
      "epoch": 1.617937146474444,
      "grad_norm": 0.5749743102814574,
      "learning_rate": 1.8536050170085117e-05,
      "loss": 11.9272,
      "step": 29712
    },
    {
      "epoch": 1.617991600471027,
      "grad_norm": 0.5111909105755295,
      "learning_rate": 1.8530936182268755e-05,
      "loss": 11.6833,
      "step": 29713
    },
    {
      "epoch": 1.61804605446761,
      "grad_norm": 0.5118690492104214,
      "learning_rate": 1.852582282796952e-05,
      "loss": 11.7455,
      "step": 29714
    },
    {
      "epoch": 1.618100508464193,
      "grad_norm": 0.5177070100658223,
      "learning_rate": 1.8520710107227156e-05,
      "loss": 11.8511,
      "step": 29715
    },
    {
      "epoch": 1.618154962460776,
      "grad_norm": 0.5429501437094439,
      "learning_rate": 1.8515598020081383e-05,
      "loss": 11.9224,
      "step": 29716
    },
    {
      "epoch": 1.618209416457359,
      "grad_norm": 0.5475021896792563,
      "learning_rate": 1.8510486566572017e-05,
      "loss": 11.6796,
      "step": 29717
    },
    {
      "epoch": 1.618263870453942,
      "grad_norm": 0.5924743382317234,
      "learning_rate": 1.850537574673874e-05,
      "loss": 11.8485,
      "step": 29718
    },
    {
      "epoch": 1.618318324450525,
      "grad_norm": 0.5013354626793133,
      "learning_rate": 1.8500265560621334e-05,
      "loss": 11.8067,
      "step": 29719
    },
    {
      "epoch": 1.618372778447108,
      "grad_norm": 0.5114146283955324,
      "learning_rate": 1.8495156008259552e-05,
      "loss": 11.6471,
      "step": 29720
    },
    {
      "epoch": 1.618427232443691,
      "grad_norm": 0.5786062154256025,
      "learning_rate": 1.8490047089693084e-05,
      "loss": 11.934,
      "step": 29721
    },
    {
      "epoch": 1.618481686440274,
      "grad_norm": 0.5626412667315792,
      "learning_rate": 1.8484938804961706e-05,
      "loss": 11.8384,
      "step": 29722
    },
    {
      "epoch": 1.6185361404368572,
      "grad_norm": 0.5544560528376635,
      "learning_rate": 1.8479831154105075e-05,
      "loss": 11.6767,
      "step": 29723
    },
    {
      "epoch": 1.6185905944334402,
      "grad_norm": 0.5589589299532259,
      "learning_rate": 1.847472413716298e-05,
      "loss": 11.8483,
      "step": 29724
    },
    {
      "epoch": 1.6186450484300232,
      "grad_norm": 0.5928141170937883,
      "learning_rate": 1.8469617754175107e-05,
      "loss": 11.7703,
      "step": 29725
    },
    {
      "epoch": 1.6186995024266062,
      "grad_norm": 0.5576237658610851,
      "learning_rate": 1.8464512005181133e-05,
      "loss": 11.8894,
      "step": 29726
    },
    {
      "epoch": 1.6187539564231892,
      "grad_norm": 0.5290010663145447,
      "learning_rate": 1.845940689022081e-05,
      "loss": 11.7963,
      "step": 29727
    },
    {
      "epoch": 1.6188084104197722,
      "grad_norm": 0.5617072598563205,
      "learning_rate": 1.845430240933379e-05,
      "loss": 11.7636,
      "step": 29728
    },
    {
      "epoch": 1.6188628644163554,
      "grad_norm": 0.5955959921049784,
      "learning_rate": 1.844919856255981e-05,
      "loss": 11.879,
      "step": 29729
    },
    {
      "epoch": 1.6189173184129384,
      "grad_norm": 0.5005111701661928,
      "learning_rate": 1.8444095349938518e-05,
      "loss": 11.8425,
      "step": 29730
    },
    {
      "epoch": 1.6189717724095214,
      "grad_norm": 0.5753867860748185,
      "learning_rate": 1.843899277150961e-05,
      "loss": 11.8455,
      "step": 29731
    },
    {
      "epoch": 1.6190262264061044,
      "grad_norm": 0.6022810675614024,
      "learning_rate": 1.8433890827312806e-05,
      "loss": 11.7727,
      "step": 29732
    },
    {
      "epoch": 1.6190806804026874,
      "grad_norm": 0.5737669885927515,
      "learning_rate": 1.8428789517387723e-05,
      "loss": 11.6892,
      "step": 29733
    },
    {
      "epoch": 1.6191351343992704,
      "grad_norm": 0.5383511411239242,
      "learning_rate": 1.8423688841774067e-05,
      "loss": 11.7437,
      "step": 29734
    },
    {
      "epoch": 1.6191895883958534,
      "grad_norm": 0.6560133762952048,
      "learning_rate": 1.8418588800511493e-05,
      "loss": 11.9499,
      "step": 29735
    },
    {
      "epoch": 1.6192440423924364,
      "grad_norm": 0.5467763424292883,
      "learning_rate": 1.841348939363963e-05,
      "loss": 11.8131,
      "step": 29736
    },
    {
      "epoch": 1.6192984963890193,
      "grad_norm": 0.5654688761193011,
      "learning_rate": 1.840839062119818e-05,
      "loss": 11.6356,
      "step": 29737
    },
    {
      "epoch": 1.6193529503856023,
      "grad_norm": 0.5736787912331952,
      "learning_rate": 1.840329248322673e-05,
      "loss": 11.7675,
      "step": 29738
    },
    {
      "epoch": 1.6194074043821853,
      "grad_norm": 0.7146954815898924,
      "learning_rate": 1.8398194979765005e-05,
      "loss": 11.8188,
      "step": 29739
    },
    {
      "epoch": 1.6194618583787683,
      "grad_norm": 0.5270226949178375,
      "learning_rate": 1.839309811085257e-05,
      "loss": 11.651,
      "step": 29740
    },
    {
      "epoch": 1.6195163123753513,
      "grad_norm": 0.5697433592279761,
      "learning_rate": 1.8388001876529105e-05,
      "loss": 11.8665,
      "step": 29741
    },
    {
      "epoch": 1.6195707663719343,
      "grad_norm": 0.5792422543746626,
      "learning_rate": 1.8382906276834212e-05,
      "loss": 11.8301,
      "step": 29742
    },
    {
      "epoch": 1.6196252203685173,
      "grad_norm": 0.5314354702809112,
      "learning_rate": 1.8377811311807514e-05,
      "loss": 11.7925,
      "step": 29743
    },
    {
      "epoch": 1.6196796743651003,
      "grad_norm": 0.5408031908187323,
      "learning_rate": 1.837271698148868e-05,
      "loss": 11.9217,
      "step": 29744
    },
    {
      "epoch": 1.6197341283616833,
      "grad_norm": 0.5682392557677713,
      "learning_rate": 1.836762328591728e-05,
      "loss": 11.7606,
      "step": 29745
    },
    {
      "epoch": 1.6197885823582665,
      "grad_norm": 0.47152443855339354,
      "learning_rate": 1.8362530225132902e-05,
      "loss": 11.817,
      "step": 29746
    },
    {
      "epoch": 1.6198430363548495,
      "grad_norm": 0.5890601620756605,
      "learning_rate": 1.8357437799175216e-05,
      "loss": 11.9369,
      "step": 29747
    },
    {
      "epoch": 1.6198974903514325,
      "grad_norm": 0.512614243080736,
      "learning_rate": 1.8352346008083753e-05,
      "loss": 11.7113,
      "step": 29748
    },
    {
      "epoch": 1.6199519443480155,
      "grad_norm": 0.5001521175061336,
      "learning_rate": 1.8347254851898166e-05,
      "loss": 11.7671,
      "step": 29749
    },
    {
      "epoch": 1.6200063983445985,
      "grad_norm": 0.5526447643640972,
      "learning_rate": 1.8342164330657997e-05,
      "loss": 11.6839,
      "step": 29750
    },
    {
      "epoch": 1.6200608523411815,
      "grad_norm": 0.5244986183923309,
      "learning_rate": 1.833707444440288e-05,
      "loss": 11.6831,
      "step": 29751
    },
    {
      "epoch": 1.6201153063377647,
      "grad_norm": 0.5890321091260146,
      "learning_rate": 1.833198519317233e-05,
      "loss": 11.8685,
      "step": 29752
    },
    {
      "epoch": 1.6201697603343477,
      "grad_norm": 0.5218310798077371,
      "learning_rate": 1.832689657700597e-05,
      "loss": 11.7616,
      "step": 29753
    },
    {
      "epoch": 1.6202242143309307,
      "grad_norm": 0.5239361598665088,
      "learning_rate": 1.832180859594338e-05,
      "loss": 11.7665,
      "step": 29754
    },
    {
      "epoch": 1.6202786683275137,
      "grad_norm": 0.5561360692793038,
      "learning_rate": 1.8316721250024093e-05,
      "loss": 11.8565,
      "step": 29755
    },
    {
      "epoch": 1.6203331223240967,
      "grad_norm": 0.58588166861285,
      "learning_rate": 1.8311634539287692e-05,
      "loss": 11.8331,
      "step": 29756
    },
    {
      "epoch": 1.6203875763206796,
      "grad_norm": 0.47225103007604813,
      "learning_rate": 1.83065484637737e-05,
      "loss": 11.7213,
      "step": 29757
    },
    {
      "epoch": 1.6204420303172626,
      "grad_norm": 0.5730737085838575,
      "learning_rate": 1.830146302352168e-05,
      "loss": 11.7649,
      "step": 29758
    },
    {
      "epoch": 1.6204964843138456,
      "grad_norm": 0.480864723884432,
      "learning_rate": 1.8296378218571207e-05,
      "loss": 11.7468,
      "step": 29759
    },
    {
      "epoch": 1.6205509383104286,
      "grad_norm": 0.6027636164304947,
      "learning_rate": 1.829129404896177e-05,
      "loss": 11.8949,
      "step": 29760
    },
    {
      "epoch": 1.6206053923070116,
      "grad_norm": 0.515908200547813,
      "learning_rate": 1.8286210514732973e-05,
      "loss": 11.6664,
      "step": 29761
    },
    {
      "epoch": 1.6206598463035946,
      "grad_norm": 0.5588522111663349,
      "learning_rate": 1.8281127615924264e-05,
      "loss": 11.7865,
      "step": 29762
    },
    {
      "epoch": 1.6207143003001776,
      "grad_norm": 0.6389388471645864,
      "learning_rate": 1.8276045352575244e-05,
      "loss": 11.8541,
      "step": 29763
    },
    {
      "epoch": 1.6207687542967606,
      "grad_norm": 0.5136481223944924,
      "learning_rate": 1.8270963724725375e-05,
      "loss": 11.8994,
      "step": 29764
    },
    {
      "epoch": 1.6208232082933436,
      "grad_norm": 0.5349080581632016,
      "learning_rate": 1.826588273241423e-05,
      "loss": 11.7461,
      "step": 29765
    },
    {
      "epoch": 1.6208776622899266,
      "grad_norm": 0.5986740150879933,
      "learning_rate": 1.8260802375681286e-05,
      "loss": 11.9312,
      "step": 29766
    },
    {
      "epoch": 1.6209321162865096,
      "grad_norm": 0.5560229739189517,
      "learning_rate": 1.825572265456602e-05,
      "loss": 11.8586,
      "step": 29767
    },
    {
      "epoch": 1.6209865702830926,
      "grad_norm": 0.582516417404331,
      "learning_rate": 1.825064356910796e-05,
      "loss": 11.8453,
      "step": 29768
    },
    {
      "epoch": 1.6210410242796758,
      "grad_norm": 0.6281607864308909,
      "learning_rate": 1.8245565119346632e-05,
      "loss": 11.795,
      "step": 29769
    },
    {
      "epoch": 1.6210954782762588,
      "grad_norm": 0.5695811605645611,
      "learning_rate": 1.8240487305321473e-05,
      "loss": 11.7029,
      "step": 29770
    },
    {
      "epoch": 1.6211499322728418,
      "grad_norm": 0.5451215618444492,
      "learning_rate": 1.8235410127072027e-05,
      "loss": 11.7372,
      "step": 29771
    },
    {
      "epoch": 1.6212043862694248,
      "grad_norm": 0.5705223579328994,
      "learning_rate": 1.8230333584637716e-05,
      "loss": 11.9475,
      "step": 29772
    },
    {
      "epoch": 1.6212588402660078,
      "grad_norm": 0.5490309796689397,
      "learning_rate": 1.8225257678058072e-05,
      "loss": 11.8289,
      "step": 29773
    },
    {
      "epoch": 1.6213132942625907,
      "grad_norm": 0.5834407945646138,
      "learning_rate": 1.822018240737251e-05,
      "loss": 11.9484,
      "step": 29774
    },
    {
      "epoch": 1.621367748259174,
      "grad_norm": 0.5774574060540624,
      "learning_rate": 1.8215107772620554e-05,
      "loss": 11.9275,
      "step": 29775
    },
    {
      "epoch": 1.621422202255757,
      "grad_norm": 0.5749672081949425,
      "learning_rate": 1.8210033773841627e-05,
      "loss": 11.8054,
      "step": 29776
    },
    {
      "epoch": 1.62147665625234,
      "grad_norm": 0.5108241365433569,
      "learning_rate": 1.820496041107518e-05,
      "loss": 11.7704,
      "step": 29777
    },
    {
      "epoch": 1.621531110248923,
      "grad_norm": 0.5355894214491904,
      "learning_rate": 1.8199887684360696e-05,
      "loss": 11.7478,
      "step": 29778
    },
    {
      "epoch": 1.621585564245506,
      "grad_norm": 0.5344683838749086,
      "learning_rate": 1.819481559373757e-05,
      "loss": 11.721,
      "step": 29779
    },
    {
      "epoch": 1.621640018242089,
      "grad_norm": 0.5484251651408244,
      "learning_rate": 1.818974413924528e-05,
      "loss": 11.7141,
      "step": 29780
    },
    {
      "epoch": 1.621694472238672,
      "grad_norm": 0.5330372194509918,
      "learning_rate": 1.8184673320923296e-05,
      "loss": 11.847,
      "step": 29781
    },
    {
      "epoch": 1.621748926235255,
      "grad_norm": 0.54969047449176,
      "learning_rate": 1.8179603138810973e-05,
      "loss": 11.9113,
      "step": 29782
    },
    {
      "epoch": 1.621803380231838,
      "grad_norm": 0.5540264519426792,
      "learning_rate": 1.8174533592947817e-05,
      "loss": 11.8589,
      "step": 29783
    },
    {
      "epoch": 1.621857834228421,
      "grad_norm": 0.5027733970076889,
      "learning_rate": 1.8169464683373172e-05,
      "loss": 11.8258,
      "step": 29784
    },
    {
      "epoch": 1.6219122882250039,
      "grad_norm": 0.5338662523738564,
      "learning_rate": 1.8164396410126527e-05,
      "loss": 11.8409,
      "step": 29785
    },
    {
      "epoch": 1.6219667422215869,
      "grad_norm": 0.5355778796105455,
      "learning_rate": 1.8159328773247254e-05,
      "loss": 11.6876,
      "step": 29786
    },
    {
      "epoch": 1.6220211962181699,
      "grad_norm": 0.5506582771023127,
      "learning_rate": 1.8154261772774738e-05,
      "loss": 11.7884,
      "step": 29787
    },
    {
      "epoch": 1.6220756502147529,
      "grad_norm": 0.5706165051098933,
      "learning_rate": 1.8149195408748443e-05,
      "loss": 11.8531,
      "step": 29788
    },
    {
      "epoch": 1.6221301042113359,
      "grad_norm": 0.5373574049088574,
      "learning_rate": 1.8144129681207688e-05,
      "loss": 11.7053,
      "step": 29789
    },
    {
      "epoch": 1.6221845582079188,
      "grad_norm": 0.5458616252224401,
      "learning_rate": 1.8139064590191946e-05,
      "loss": 11.7965,
      "step": 29790
    },
    {
      "epoch": 1.6222390122045018,
      "grad_norm": 0.5840944206279342,
      "learning_rate": 1.8134000135740536e-05,
      "loss": 11.8291,
      "step": 29791
    },
    {
      "epoch": 1.6222934662010848,
      "grad_norm": 0.5795284200751835,
      "learning_rate": 1.8128936317892875e-05,
      "loss": 11.7689,
      "step": 29792
    },
    {
      "epoch": 1.622347920197668,
      "grad_norm": 0.5292672808196972,
      "learning_rate": 1.8123873136688364e-05,
      "loss": 11.774,
      "step": 29793
    },
    {
      "epoch": 1.622402374194251,
      "grad_norm": 0.505826063359344,
      "learning_rate": 1.8118810592166314e-05,
      "loss": 11.8555,
      "step": 29794
    },
    {
      "epoch": 1.622456828190834,
      "grad_norm": 0.5177697343648606,
      "learning_rate": 1.811374868436615e-05,
      "loss": 11.7862,
      "step": 29795
    },
    {
      "epoch": 1.622511282187417,
      "grad_norm": 0.6289377222011117,
      "learning_rate": 1.810868741332721e-05,
      "loss": 11.8154,
      "step": 29796
    },
    {
      "epoch": 1.622565736184,
      "grad_norm": 0.5185660813581372,
      "learning_rate": 1.810362677908882e-05,
      "loss": 11.7476,
      "step": 29797
    },
    {
      "epoch": 1.622620190180583,
      "grad_norm": 0.8677192061925832,
      "learning_rate": 1.8098566781690397e-05,
      "loss": 11.6404,
      "step": 29798
    },
    {
      "epoch": 1.6226746441771662,
      "grad_norm": 0.5562605901423329,
      "learning_rate": 1.8093507421171217e-05,
      "loss": 11.8005,
      "step": 29799
    },
    {
      "epoch": 1.6227290981737492,
      "grad_norm": 0.5416724699517385,
      "learning_rate": 1.8088448697570693e-05,
      "loss": 11.8058,
      "step": 29800
    },
    {
      "epoch": 1.6227835521703322,
      "grad_norm": 0.5413622332538577,
      "learning_rate": 1.8083390610928097e-05,
      "loss": 11.8775,
      "step": 29801
    },
    {
      "epoch": 1.6228380061669152,
      "grad_norm": 0.5086184504807458,
      "learning_rate": 1.807833316128279e-05,
      "loss": 11.8459,
      "step": 29802
    },
    {
      "epoch": 1.6228924601634982,
      "grad_norm": 0.5248310507755106,
      "learning_rate": 1.807327634867414e-05,
      "loss": 11.7233,
      "step": 29803
    },
    {
      "epoch": 1.6229469141600812,
      "grad_norm": 0.5570920371789084,
      "learning_rate": 1.8068220173141394e-05,
      "loss": 11.8058,
      "step": 29804
    },
    {
      "epoch": 1.6230013681566642,
      "grad_norm": 0.504956431398303,
      "learning_rate": 1.806316463472394e-05,
      "loss": 11.7385,
      "step": 29805
    },
    {
      "epoch": 1.6230558221532472,
      "grad_norm": 0.5343066618336901,
      "learning_rate": 1.805810973346106e-05,
      "loss": 11.8999,
      "step": 29806
    },
    {
      "epoch": 1.6231102761498302,
      "grad_norm": 0.5745609893400554,
      "learning_rate": 1.8053055469392034e-05,
      "loss": 11.9275,
      "step": 29807
    },
    {
      "epoch": 1.6231647301464132,
      "grad_norm": 0.5751834413970852,
      "learning_rate": 1.8048001842556216e-05,
      "loss": 11.8573,
      "step": 29808
    },
    {
      "epoch": 1.6232191841429962,
      "grad_norm": 0.5415861978949785,
      "learning_rate": 1.804294885299286e-05,
      "loss": 11.792,
      "step": 29809
    },
    {
      "epoch": 1.6232736381395791,
      "grad_norm": 0.5979319586553563,
      "learning_rate": 1.8037896500741292e-05,
      "loss": 11.8604,
      "step": 29810
    },
    {
      "epoch": 1.6233280921361621,
      "grad_norm": 0.5332030842511283,
      "learning_rate": 1.8032844785840765e-05,
      "loss": 11.825,
      "step": 29811
    },
    {
      "epoch": 1.6233825461327451,
      "grad_norm": 0.5672826300173477,
      "learning_rate": 1.802779370833061e-05,
      "loss": 11.7771,
      "step": 29812
    },
    {
      "epoch": 1.6234370001293281,
      "grad_norm": 0.5456811727174755,
      "learning_rate": 1.8022743268250053e-05,
      "loss": 11.7762,
      "step": 29813
    },
    {
      "epoch": 1.6234914541259111,
      "grad_norm": 0.5188778840167558,
      "learning_rate": 1.8017693465638385e-05,
      "loss": 11.8128,
      "step": 29814
    },
    {
      "epoch": 1.623545908122494,
      "grad_norm": 0.6029305953430277,
      "learning_rate": 1.8012644300534908e-05,
      "loss": 11.9242,
      "step": 29815
    },
    {
      "epoch": 1.6236003621190773,
      "grad_norm": 0.576940166603129,
      "learning_rate": 1.800759577297887e-05,
      "loss": 11.8113,
      "step": 29816
    },
    {
      "epoch": 1.6236548161156603,
      "grad_norm": 0.5358719089609392,
      "learning_rate": 1.800254788300948e-05,
      "loss": 11.8687,
      "step": 29817
    },
    {
      "epoch": 1.6237092701122433,
      "grad_norm": 0.5202429387029124,
      "learning_rate": 1.7997500630666053e-05,
      "loss": 11.6968,
      "step": 29818
    },
    {
      "epoch": 1.6237637241088263,
      "grad_norm": 0.5431275583205528,
      "learning_rate": 1.7992454015987793e-05,
      "loss": 11.7837,
      "step": 29819
    },
    {
      "epoch": 1.6238181781054093,
      "grad_norm": 0.5793584771153201,
      "learning_rate": 1.7987408039013987e-05,
      "loss": 11.7659,
      "step": 29820
    },
    {
      "epoch": 1.6238726321019923,
      "grad_norm": 0.5579550905480155,
      "learning_rate": 1.7982362699783827e-05,
      "loss": 11.6604,
      "step": 29821
    },
    {
      "epoch": 1.6239270860985755,
      "grad_norm": 0.5744584766801781,
      "learning_rate": 1.7977317998336597e-05,
      "loss": 12.0137,
      "step": 29822
    },
    {
      "epoch": 1.6239815400951585,
      "grad_norm": 0.6315104965522051,
      "learning_rate": 1.797227393471147e-05,
      "loss": 11.8804,
      "step": 29823
    },
    {
      "epoch": 1.6240359940917415,
      "grad_norm": 0.562141376184874,
      "learning_rate": 1.7967230508947687e-05,
      "loss": 11.8801,
      "step": 29824
    },
    {
      "epoch": 1.6240904480883245,
      "grad_norm": 0.538474797622595,
      "learning_rate": 1.796218772108451e-05,
      "loss": 11.7563,
      "step": 29825
    },
    {
      "epoch": 1.6241449020849075,
      "grad_norm": 0.5280226524983459,
      "learning_rate": 1.7957145571161117e-05,
      "loss": 11.8238,
      "step": 29826
    },
    {
      "epoch": 1.6241993560814905,
      "grad_norm": 0.5367644806566167,
      "learning_rate": 1.7952104059216722e-05,
      "loss": 11.8238,
      "step": 29827
    },
    {
      "epoch": 1.6242538100780735,
      "grad_norm": 0.536063174405373,
      "learning_rate": 1.7947063185290502e-05,
      "loss": 11.8473,
      "step": 29828
    },
    {
      "epoch": 1.6243082640746564,
      "grad_norm": 0.5080321495151783,
      "learning_rate": 1.7942022949421665e-05,
      "loss": 11.8336,
      "step": 29829
    },
    {
      "epoch": 1.6243627180712394,
      "grad_norm": 0.5543109604420468,
      "learning_rate": 1.793698335164946e-05,
      "loss": 11.8157,
      "step": 29830
    },
    {
      "epoch": 1.6244171720678224,
      "grad_norm": 0.5446841505257469,
      "learning_rate": 1.7931944392013e-05,
      "loss": 11.7777,
      "step": 29831
    },
    {
      "epoch": 1.6244716260644054,
      "grad_norm": 0.5390297565914725,
      "learning_rate": 1.792690607055153e-05,
      "loss": 11.6937,
      "step": 29832
    },
    {
      "epoch": 1.6245260800609884,
      "grad_norm": 0.6024428999512876,
      "learning_rate": 1.7921868387304187e-05,
      "loss": 11.715,
      "step": 29833
    },
    {
      "epoch": 1.6245805340575714,
      "grad_norm": 0.4685025058250226,
      "learning_rate": 1.791683134231017e-05,
      "loss": 11.8521,
      "step": 29834
    },
    {
      "epoch": 1.6246349880541544,
      "grad_norm": 0.537252562932653,
      "learning_rate": 1.7911794935608616e-05,
      "loss": 11.8172,
      "step": 29835
    },
    {
      "epoch": 1.6246894420507374,
      "grad_norm": 0.5740644328894018,
      "learning_rate": 1.790675916723874e-05,
      "loss": 11.7572,
      "step": 29836
    },
    {
      "epoch": 1.6247438960473204,
      "grad_norm": 0.49532925050483767,
      "learning_rate": 1.7901724037239675e-05,
      "loss": 11.7963,
      "step": 29837
    },
    {
      "epoch": 1.6247983500439034,
      "grad_norm": 0.4812212648939318,
      "learning_rate": 1.7896689545650548e-05,
      "loss": 11.6105,
      "step": 29838
    },
    {
      "epoch": 1.6248528040404866,
      "grad_norm": 0.5808507902498412,
      "learning_rate": 1.7891655692510547e-05,
      "loss": 11.7792,
      "step": 29839
    },
    {
      "epoch": 1.6249072580370696,
      "grad_norm": 0.5716854153082225,
      "learning_rate": 1.7886622477858783e-05,
      "loss": 11.8356,
      "step": 29840
    },
    {
      "epoch": 1.6249617120336526,
      "grad_norm": 0.5438784820571364,
      "learning_rate": 1.7881589901734408e-05,
      "loss": 11.7404,
      "step": 29841
    },
    {
      "epoch": 1.6250161660302356,
      "grad_norm": 0.5522321049595004,
      "learning_rate": 1.7876557964176587e-05,
      "loss": 11.8787,
      "step": 29842
    },
    {
      "epoch": 1.6250706200268186,
      "grad_norm": 0.5385128981210782,
      "learning_rate": 1.7871526665224404e-05,
      "loss": 11.9742,
      "step": 29843
    },
    {
      "epoch": 1.6251250740234016,
      "grad_norm": 0.5642394369165362,
      "learning_rate": 1.786649600491702e-05,
      "loss": 11.8231,
      "step": 29844
    },
    {
      "epoch": 1.6251795280199848,
      "grad_norm": 0.5282025949759572,
      "learning_rate": 1.7861465983293514e-05,
      "loss": 11.8951,
      "step": 29845
    },
    {
      "epoch": 1.6252339820165678,
      "grad_norm": 0.5067521434221105,
      "learning_rate": 1.7856436600393046e-05,
      "loss": 11.8502,
      "step": 29846
    },
    {
      "epoch": 1.6252884360131508,
      "grad_norm": 0.5880429059922319,
      "learning_rate": 1.785140785625471e-05,
      "loss": 11.8383,
      "step": 29847
    },
    {
      "epoch": 1.6253428900097338,
      "grad_norm": 0.532984901509907,
      "learning_rate": 1.7846379750917576e-05,
      "loss": 11.8345,
      "step": 29848
    },
    {
      "epoch": 1.6253973440063167,
      "grad_norm": 0.5544920082952747,
      "learning_rate": 1.7841352284420786e-05,
      "loss": 11.7894,
      "step": 29849
    },
    {
      "epoch": 1.6254517980028997,
      "grad_norm": 0.5217674602501226,
      "learning_rate": 1.7836325456803404e-05,
      "loss": 11.7569,
      "step": 29850
    },
    {
      "epoch": 1.6255062519994827,
      "grad_norm": 0.5172372220391482,
      "learning_rate": 1.783129926810453e-05,
      "loss": 11.7676,
      "step": 29851
    },
    {
      "epoch": 1.6255607059960657,
      "grad_norm": 0.5156677079660713,
      "learning_rate": 1.7826273718363274e-05,
      "loss": 11.6364,
      "step": 29852
    },
    {
      "epoch": 1.6256151599926487,
      "grad_norm": 0.5500854937460841,
      "learning_rate": 1.7821248807618674e-05,
      "loss": 11.7937,
      "step": 29853
    },
    {
      "epoch": 1.6256696139892317,
      "grad_norm": 0.6009562225192525,
      "learning_rate": 1.7816224535909854e-05,
      "loss": 11.8653,
      "step": 29854
    },
    {
      "epoch": 1.6257240679858147,
      "grad_norm": 0.5897278157514485,
      "learning_rate": 1.7811200903275826e-05,
      "loss": 11.9097,
      "step": 29855
    },
    {
      "epoch": 1.6257785219823977,
      "grad_norm": 0.5196931484303855,
      "learning_rate": 1.7806177909755706e-05,
      "loss": 11.7866,
      "step": 29856
    },
    {
      "epoch": 1.6258329759789807,
      "grad_norm": 0.5259285433385095,
      "learning_rate": 1.7801155555388528e-05,
      "loss": 11.9078,
      "step": 29857
    },
    {
      "epoch": 1.6258874299755637,
      "grad_norm": 0.5379667007032992,
      "learning_rate": 1.7796133840213323e-05,
      "loss": 11.8614,
      "step": 29858
    },
    {
      "epoch": 1.6259418839721467,
      "grad_norm": 0.5596329701440825,
      "learning_rate": 1.77911127642692e-05,
      "loss": 11.7767,
      "step": 29859
    },
    {
      "epoch": 1.6259963379687297,
      "grad_norm": 0.5160637975508303,
      "learning_rate": 1.778609232759514e-05,
      "loss": 11.8189,
      "step": 29860
    },
    {
      "epoch": 1.6260507919653127,
      "grad_norm": 0.565438200037967,
      "learning_rate": 1.7781072530230236e-05,
      "loss": 11.7822,
      "step": 29861
    },
    {
      "epoch": 1.6261052459618957,
      "grad_norm": 0.58634501271915,
      "learning_rate": 1.7776053372213486e-05,
      "loss": 11.8256,
      "step": 29862
    },
    {
      "epoch": 1.6261596999584789,
      "grad_norm": 0.557731556141301,
      "learning_rate": 1.7771034853583923e-05,
      "loss": 11.7614,
      "step": 29863
    },
    {
      "epoch": 1.6262141539550619,
      "grad_norm": 0.5498987337364991,
      "learning_rate": 1.7766016974380606e-05,
      "loss": 11.8434,
      "step": 29864
    },
    {
      "epoch": 1.6262686079516449,
      "grad_norm": 0.5814591000650591,
      "learning_rate": 1.7760999734642513e-05,
      "loss": 11.8587,
      "step": 29865
    },
    {
      "epoch": 1.6263230619482278,
      "grad_norm": 0.5252663649406915,
      "learning_rate": 1.7755983134408703e-05,
      "loss": 11.7993,
      "step": 29866
    },
    {
      "epoch": 1.6263775159448108,
      "grad_norm": 0.5483473578770199,
      "learning_rate": 1.775096717371816e-05,
      "loss": 11.8305,
      "step": 29867
    },
    {
      "epoch": 1.626431969941394,
      "grad_norm": 0.5179828168760365,
      "learning_rate": 1.7745951852609865e-05,
      "loss": 11.8467,
      "step": 29868
    },
    {
      "epoch": 1.626486423937977,
      "grad_norm": 0.6214975217796868,
      "learning_rate": 1.7740937171122864e-05,
      "loss": 11.7297,
      "step": 29869
    },
    {
      "epoch": 1.62654087793456,
      "grad_norm": 0.5395443085330129,
      "learning_rate": 1.7735923129296107e-05,
      "loss": 11.805,
      "step": 29870
    },
    {
      "epoch": 1.626595331931143,
      "grad_norm": 0.6039017919913058,
      "learning_rate": 1.7730909727168632e-05,
      "loss": 11.8849,
      "step": 29871
    },
    {
      "epoch": 1.626649785927726,
      "grad_norm": 0.5699838813347412,
      "learning_rate": 1.7725896964779375e-05,
      "loss": 11.8384,
      "step": 29872
    },
    {
      "epoch": 1.626704239924309,
      "grad_norm": 0.5390644551119237,
      "learning_rate": 1.772088484216734e-05,
      "loss": 11.8132,
      "step": 29873
    },
    {
      "epoch": 1.626758693920892,
      "grad_norm": 0.581202814320493,
      "learning_rate": 1.7715873359371526e-05,
      "loss": 11.885,
      "step": 29874
    },
    {
      "epoch": 1.626813147917475,
      "grad_norm": 0.4968017133864991,
      "learning_rate": 1.7710862516430858e-05,
      "loss": 11.6651,
      "step": 29875
    },
    {
      "epoch": 1.626867601914058,
      "grad_norm": 0.5312199139071078,
      "learning_rate": 1.7705852313384374e-05,
      "loss": 11.8351,
      "step": 29876
    },
    {
      "epoch": 1.626922055910641,
      "grad_norm": 0.5009578686172519,
      "learning_rate": 1.770084275027093e-05,
      "loss": 11.773,
      "step": 29877
    },
    {
      "epoch": 1.626976509907224,
      "grad_norm": 0.5437621951448142,
      "learning_rate": 1.7695833827129538e-05,
      "loss": 11.7318,
      "step": 29878
    },
    {
      "epoch": 1.627030963903807,
      "grad_norm": 0.5263257599612827,
      "learning_rate": 1.7690825543999167e-05,
      "loss": 11.86,
      "step": 29879
    },
    {
      "epoch": 1.62708541790039,
      "grad_norm": 0.529554063485207,
      "learning_rate": 1.7685817900918723e-05,
      "loss": 11.7946,
      "step": 29880
    },
    {
      "epoch": 1.627139871896973,
      "grad_norm": 0.5981416094577404,
      "learning_rate": 1.768081089792717e-05,
      "loss": 11.9005,
      "step": 29881
    },
    {
      "epoch": 1.627194325893556,
      "grad_norm": 0.589911703001778,
      "learning_rate": 1.7675804535063424e-05,
      "loss": 11.8847,
      "step": 29882
    },
    {
      "epoch": 1.627248779890139,
      "grad_norm": 0.5577245965792366,
      "learning_rate": 1.767079881236645e-05,
      "loss": 11.843,
      "step": 29883
    },
    {
      "epoch": 1.627303233886722,
      "grad_norm": 0.5370008557756406,
      "learning_rate": 1.7665793729875125e-05,
      "loss": 11.7251,
      "step": 29884
    },
    {
      "epoch": 1.627357687883305,
      "grad_norm": 0.55426542388383,
      "learning_rate": 1.76607892876284e-05,
      "loss": 11.8755,
      "step": 29885
    },
    {
      "epoch": 1.6274121418798881,
      "grad_norm": 0.5663218347031582,
      "learning_rate": 1.7655785485665233e-05,
      "loss": 11.7826,
      "step": 29886
    },
    {
      "epoch": 1.6274665958764711,
      "grad_norm": 0.5194411723218578,
      "learning_rate": 1.765078232402444e-05,
      "loss": 11.7238,
      "step": 29887
    },
    {
      "epoch": 1.6275210498730541,
      "grad_norm": 0.5452031387886116,
      "learning_rate": 1.7645779802744965e-05,
      "loss": 11.8034,
      "step": 29888
    },
    {
      "epoch": 1.6275755038696371,
      "grad_norm": 0.625163580505798,
      "learning_rate": 1.7640777921865746e-05,
      "loss": 11.7914,
      "step": 29889
    },
    {
      "epoch": 1.6276299578662201,
      "grad_norm": 0.562152490599785,
      "learning_rate": 1.7635776681425632e-05,
      "loss": 11.8573,
      "step": 29890
    },
    {
      "epoch": 1.627684411862803,
      "grad_norm": 0.5631411373166973,
      "learning_rate": 1.763077608146355e-05,
      "loss": 11.8009,
      "step": 29891
    },
    {
      "epoch": 1.6277388658593863,
      "grad_norm": 0.5004948928729843,
      "learning_rate": 1.7625776122018333e-05,
      "loss": 11.824,
      "step": 29892
    },
    {
      "epoch": 1.6277933198559693,
      "grad_norm": 0.5780593933229375,
      "learning_rate": 1.7620776803128936e-05,
      "loss": 11.9005,
      "step": 29893
    },
    {
      "epoch": 1.6278477738525523,
      "grad_norm": 0.543401389222193,
      "learning_rate": 1.761577812483416e-05,
      "loss": 11.7474,
      "step": 29894
    },
    {
      "epoch": 1.6279022278491353,
      "grad_norm": 0.5595408395569539,
      "learning_rate": 1.761078008717293e-05,
      "loss": 11.8585,
      "step": 29895
    },
    {
      "epoch": 1.6279566818457183,
      "grad_norm": 0.5635121012342469,
      "learning_rate": 1.7605782690184103e-05,
      "loss": 11.9676,
      "step": 29896
    },
    {
      "epoch": 1.6280111358423013,
      "grad_norm": 0.5949180173017401,
      "learning_rate": 1.7600785933906494e-05,
      "loss": 11.9263,
      "step": 29897
    },
    {
      "epoch": 1.6280655898388843,
      "grad_norm": 0.6192375097089666,
      "learning_rate": 1.7595789818379017e-05,
      "loss": 11.8609,
      "step": 29898
    },
    {
      "epoch": 1.6281200438354673,
      "grad_norm": 0.5377687262894562,
      "learning_rate": 1.7590794343640472e-05,
      "loss": 11.7447,
      "step": 29899
    },
    {
      "epoch": 1.6281744978320503,
      "grad_norm": 0.5328658151317262,
      "learning_rate": 1.7585799509729727e-05,
      "loss": 11.8867,
      "step": 29900
    },
    {
      "epoch": 1.6282289518286333,
      "grad_norm": 0.6000130189421203,
      "learning_rate": 1.7580805316685666e-05,
      "loss": 11.7832,
      "step": 29901
    },
    {
      "epoch": 1.6282834058252162,
      "grad_norm": 0.6347525803748942,
      "learning_rate": 1.7575811764547045e-05,
      "loss": 11.9967,
      "step": 29902
    },
    {
      "epoch": 1.6283378598217992,
      "grad_norm": 0.507984416260509,
      "learning_rate": 1.7570818853352766e-05,
      "loss": 11.6879,
      "step": 29903
    },
    {
      "epoch": 1.6283923138183822,
      "grad_norm": 0.5272719736698143,
      "learning_rate": 1.75658265831416e-05,
      "loss": 11.7738,
      "step": 29904
    },
    {
      "epoch": 1.6284467678149652,
      "grad_norm": 0.5171362303261697,
      "learning_rate": 1.756083495395241e-05,
      "loss": 11.7618,
      "step": 29905
    },
    {
      "epoch": 1.6285012218115482,
      "grad_norm": 0.573823690687866,
      "learning_rate": 1.7555843965823992e-05,
      "loss": 11.8225,
      "step": 29906
    },
    {
      "epoch": 1.6285556758081312,
      "grad_norm": 0.6025332639191686,
      "learning_rate": 1.7550853618795125e-05,
      "loss": 11.8546,
      "step": 29907
    },
    {
      "epoch": 1.6286101298047142,
      "grad_norm": 0.5876088671905414,
      "learning_rate": 1.7545863912904693e-05,
      "loss": 11.9995,
      "step": 29908
    },
    {
      "epoch": 1.6286645838012974,
      "grad_norm": 0.6659103166319886,
      "learning_rate": 1.7540874848191413e-05,
      "loss": 11.7968,
      "step": 29909
    },
    {
      "epoch": 1.6287190377978804,
      "grad_norm": 0.5783262871274817,
      "learning_rate": 1.753588642469414e-05,
      "loss": 11.7612,
      "step": 29910
    },
    {
      "epoch": 1.6287734917944634,
      "grad_norm": 0.49121689900863397,
      "learning_rate": 1.753089864245162e-05,
      "loss": 11.7456,
      "step": 29911
    },
    {
      "epoch": 1.6288279457910464,
      "grad_norm": 0.554789154861225,
      "learning_rate": 1.7525911501502666e-05,
      "loss": 11.6595,
      "step": 29912
    },
    {
      "epoch": 1.6288823997876294,
      "grad_norm": 0.4976445127465602,
      "learning_rate": 1.752092500188608e-05,
      "loss": 11.8202,
      "step": 29913
    },
    {
      "epoch": 1.6289368537842124,
      "grad_norm": 0.574762160355171,
      "learning_rate": 1.7515939143640592e-05,
      "loss": 11.7653,
      "step": 29914
    },
    {
      "epoch": 1.6289913077807956,
      "grad_norm": 0.54272193204427,
      "learning_rate": 1.7510953926805006e-05,
      "loss": 11.8093,
      "step": 29915
    },
    {
      "epoch": 1.6290457617773786,
      "grad_norm": 0.5370533238410519,
      "learning_rate": 1.7505969351418083e-05,
      "loss": 11.7391,
      "step": 29916
    },
    {
      "epoch": 1.6291002157739616,
      "grad_norm": 0.5176173028365498,
      "learning_rate": 1.7500985417518555e-05,
      "loss": 11.7328,
      "step": 29917
    },
    {
      "epoch": 1.6291546697705446,
      "grad_norm": 0.5533274271616844,
      "learning_rate": 1.7496002125145217e-05,
      "loss": 11.7789,
      "step": 29918
    },
    {
      "epoch": 1.6292091237671276,
      "grad_norm": 0.5672350726348384,
      "learning_rate": 1.7491019474336788e-05,
      "loss": 11.8602,
      "step": 29919
    },
    {
      "epoch": 1.6292635777637106,
      "grad_norm": 0.5632535034997334,
      "learning_rate": 1.7486037465132042e-05,
      "loss": 11.8678,
      "step": 29920
    },
    {
      "epoch": 1.6293180317602936,
      "grad_norm": 0.5189339183346685,
      "learning_rate": 1.7481056097569692e-05,
      "loss": 11.8659,
      "step": 29921
    },
    {
      "epoch": 1.6293724857568765,
      "grad_norm": 0.5896665295666436,
      "learning_rate": 1.7476075371688484e-05,
      "loss": 11.8734,
      "step": 29922
    },
    {
      "epoch": 1.6294269397534595,
      "grad_norm": 0.5639517663215442,
      "learning_rate": 1.7471095287527173e-05,
      "loss": 11.8643,
      "step": 29923
    },
    {
      "epoch": 1.6294813937500425,
      "grad_norm": 0.532895021995113,
      "learning_rate": 1.7466115845124443e-05,
      "loss": 11.783,
      "step": 29924
    },
    {
      "epoch": 1.6295358477466255,
      "grad_norm": 0.5485912463971948,
      "learning_rate": 1.746113704451906e-05,
      "loss": 11.7881,
      "step": 29925
    },
    {
      "epoch": 1.6295903017432085,
      "grad_norm": 0.5370351342773633,
      "learning_rate": 1.7456158885749698e-05,
      "loss": 11.8659,
      "step": 29926
    },
    {
      "epoch": 1.6296447557397915,
      "grad_norm": 0.5385903916694121,
      "learning_rate": 1.7451181368855108e-05,
      "loss": 11.8075,
      "step": 29927
    },
    {
      "epoch": 1.6296992097363745,
      "grad_norm": 0.5417105813799838,
      "learning_rate": 1.7446204493873985e-05,
      "loss": 11.8404,
      "step": 29928
    },
    {
      "epoch": 1.6297536637329575,
      "grad_norm": 0.5365161137756895,
      "learning_rate": 1.7441228260844977e-05,
      "loss": 11.8451,
      "step": 29929
    },
    {
      "epoch": 1.6298081177295405,
      "grad_norm": 0.5404649118432637,
      "learning_rate": 1.7436252669806864e-05,
      "loss": 11.6875,
      "step": 29930
    },
    {
      "epoch": 1.6298625717261235,
      "grad_norm": 0.5416142284054279,
      "learning_rate": 1.7431277720798266e-05,
      "loss": 11.7902,
      "step": 29931
    },
    {
      "epoch": 1.6299170257227065,
      "grad_norm": 0.5846514279150229,
      "learning_rate": 1.7426303413857924e-05,
      "loss": 11.7458,
      "step": 29932
    },
    {
      "epoch": 1.6299714797192897,
      "grad_norm": 0.526056418719975,
      "learning_rate": 1.742132974902446e-05,
      "loss": 11.6389,
      "step": 29933
    },
    {
      "epoch": 1.6300259337158727,
      "grad_norm": 0.5459289100184062,
      "learning_rate": 1.74163567263366e-05,
      "loss": 11.7018,
      "step": 29934
    },
    {
      "epoch": 1.6300803877124557,
      "grad_norm": 0.4947918508224737,
      "learning_rate": 1.7411384345833016e-05,
      "loss": 11.775,
      "step": 29935
    },
    {
      "epoch": 1.6301348417090387,
      "grad_norm": 0.5410534306996849,
      "learning_rate": 1.7406412607552335e-05,
      "loss": 11.7166,
      "step": 29936
    },
    {
      "epoch": 1.6301892957056217,
      "grad_norm": 0.6530609830229392,
      "learning_rate": 1.7401441511533268e-05,
      "loss": 11.9182,
      "step": 29937
    },
    {
      "epoch": 1.6302437497022049,
      "grad_norm": 0.5322161878776143,
      "learning_rate": 1.7396471057814433e-05,
      "loss": 11.8298,
      "step": 29938
    },
    {
      "epoch": 1.6302982036987879,
      "grad_norm": 0.5067145283136489,
      "learning_rate": 1.7391501246434472e-05,
      "loss": 11.8654,
      "step": 29939
    },
    {
      "epoch": 1.6303526576953709,
      "grad_norm": 0.5367518130966582,
      "learning_rate": 1.7386532077432084e-05,
      "loss": 11.7867,
      "step": 29940
    },
    {
      "epoch": 1.6304071116919538,
      "grad_norm": 0.5940582748469746,
      "learning_rate": 1.7381563550845836e-05,
      "loss": 11.8191,
      "step": 29941
    },
    {
      "epoch": 1.6304615656885368,
      "grad_norm": 0.5854313763757858,
      "learning_rate": 1.7376595666714444e-05,
      "loss": 11.7924,
      "step": 29942
    },
    {
      "epoch": 1.6305160196851198,
      "grad_norm": 0.586362723890138,
      "learning_rate": 1.737162842507647e-05,
      "loss": 11.6086,
      "step": 29943
    },
    {
      "epoch": 1.6305704736817028,
      "grad_norm": 0.513156294977651,
      "learning_rate": 1.73666618259706e-05,
      "loss": 11.7184,
      "step": 29944
    },
    {
      "epoch": 1.6306249276782858,
      "grad_norm": 0.553937100181963,
      "learning_rate": 1.736169586943539e-05,
      "loss": 12.0196,
      "step": 29945
    },
    {
      "epoch": 1.6306793816748688,
      "grad_norm": 0.5563188406615613,
      "learning_rate": 1.7356730555509494e-05,
      "loss": 11.8782,
      "step": 29946
    },
    {
      "epoch": 1.6307338356714518,
      "grad_norm": 0.5886290555930446,
      "learning_rate": 1.7351765884231575e-05,
      "loss": 11.9051,
      "step": 29947
    },
    {
      "epoch": 1.6307882896680348,
      "grad_norm": 0.5713424153146086,
      "learning_rate": 1.7346801855640138e-05,
      "loss": 11.9171,
      "step": 29948
    },
    {
      "epoch": 1.6308427436646178,
      "grad_norm": 0.5629399749142001,
      "learning_rate": 1.7341838469773818e-05,
      "loss": 11.7363,
      "step": 29949
    },
    {
      "epoch": 1.6308971976612008,
      "grad_norm": 0.5192354403945891,
      "learning_rate": 1.7336875726671265e-05,
      "loss": 11.9123,
      "step": 29950
    },
    {
      "epoch": 1.6309516516577838,
      "grad_norm": 0.4996096798522264,
      "learning_rate": 1.733191362637099e-05,
      "loss": 11.8283,
      "step": 29951
    },
    {
      "epoch": 1.6310061056543668,
      "grad_norm": 0.5517685381132719,
      "learning_rate": 1.732695216891165e-05,
      "loss": 11.7819,
      "step": 29952
    },
    {
      "epoch": 1.6310605596509498,
      "grad_norm": 0.5743685710000227,
      "learning_rate": 1.7321991354331768e-05,
      "loss": 11.8194,
      "step": 29953
    },
    {
      "epoch": 1.6311150136475328,
      "grad_norm": 0.502849034050244,
      "learning_rate": 1.7317031182669975e-05,
      "loss": 11.7695,
      "step": 29954
    },
    {
      "epoch": 1.6311694676441157,
      "grad_norm": 0.557595428652575,
      "learning_rate": 1.7312071653964778e-05,
      "loss": 11.9792,
      "step": 29955
    },
    {
      "epoch": 1.631223921640699,
      "grad_norm": 0.5255916282851473,
      "learning_rate": 1.7307112768254785e-05,
      "loss": 11.7699,
      "step": 29956
    },
    {
      "epoch": 1.631278375637282,
      "grad_norm": 0.5822218988876974,
      "learning_rate": 1.7302154525578607e-05,
      "loss": 11.6688,
      "step": 29957
    },
    {
      "epoch": 1.631332829633865,
      "grad_norm": 0.5970869466234149,
      "learning_rate": 1.729719692597468e-05,
      "loss": 11.9408,
      "step": 29958
    },
    {
      "epoch": 1.631387283630448,
      "grad_norm": 0.5221473922026875,
      "learning_rate": 1.729223996948166e-05,
      "loss": 11.7715,
      "step": 29959
    },
    {
      "epoch": 1.631441737627031,
      "grad_norm": 0.5946245539765873,
      "learning_rate": 1.728728365613801e-05,
      "loss": 11.7498,
      "step": 29960
    },
    {
      "epoch": 1.631496191623614,
      "grad_norm": 0.5729044952909914,
      "learning_rate": 1.728232798598233e-05,
      "loss": 11.9128,
      "step": 29961
    },
    {
      "epoch": 1.6315506456201971,
      "grad_norm": 0.6188856348056977,
      "learning_rate": 1.727737295905315e-05,
      "loss": 11.8017,
      "step": 29962
    },
    {
      "epoch": 1.6316050996167801,
      "grad_norm": 0.5219248367702782,
      "learning_rate": 1.727241857538897e-05,
      "loss": 11.7027,
      "step": 29963
    },
    {
      "epoch": 1.6316595536133631,
      "grad_norm": 0.5771117140186215,
      "learning_rate": 1.7267464835028368e-05,
      "loss": 11.757,
      "step": 29964
    },
    {
      "epoch": 1.6317140076099461,
      "grad_norm": 0.534418423431781,
      "learning_rate": 1.72625117380098e-05,
      "loss": 11.8091,
      "step": 29965
    },
    {
      "epoch": 1.631768461606529,
      "grad_norm": 0.5382609954575446,
      "learning_rate": 1.725755928437185e-05,
      "loss": 11.6708,
      "step": 29966
    },
    {
      "epoch": 1.631822915603112,
      "grad_norm": 0.573478177342807,
      "learning_rate": 1.725260747415299e-05,
      "loss": 11.787,
      "step": 29967
    },
    {
      "epoch": 1.631877369599695,
      "grad_norm": 0.5501560654196728,
      "learning_rate": 1.7247656307391703e-05,
      "loss": 11.8144,
      "step": 29968
    },
    {
      "epoch": 1.631931823596278,
      "grad_norm": 0.5329772046645241,
      "learning_rate": 1.7242705784126545e-05,
      "loss": 11.7624,
      "step": 29969
    },
    {
      "epoch": 1.631986277592861,
      "grad_norm": 0.5403872408862898,
      "learning_rate": 1.723775590439596e-05,
      "loss": 11.843,
      "step": 29970
    },
    {
      "epoch": 1.632040731589444,
      "grad_norm": 0.6013170532934077,
      "learning_rate": 1.7232806668238466e-05,
      "loss": 11.8641,
      "step": 29971
    },
    {
      "epoch": 1.632095185586027,
      "grad_norm": 0.5806003289194821,
      "learning_rate": 1.7227858075692572e-05,
      "loss": 11.8855,
      "step": 29972
    },
    {
      "epoch": 1.63214963958261,
      "grad_norm": 0.5536369154366305,
      "learning_rate": 1.722291012679671e-05,
      "loss": 11.6345,
      "step": 29973
    },
    {
      "epoch": 1.632204093579193,
      "grad_norm": 0.5051355613369085,
      "learning_rate": 1.7217962821589405e-05,
      "loss": 11.8649,
      "step": 29974
    },
    {
      "epoch": 1.632258547575776,
      "grad_norm": 0.5162418393512304,
      "learning_rate": 1.7213016160109075e-05,
      "loss": 11.8208,
      "step": 29975
    },
    {
      "epoch": 1.632313001572359,
      "grad_norm": 0.5359214026692442,
      "learning_rate": 1.720807014239425e-05,
      "loss": 11.8198,
      "step": 29976
    },
    {
      "epoch": 1.632367455568942,
      "grad_norm": 0.5661563304303698,
      "learning_rate": 1.7203124768483347e-05,
      "loss": 11.8668,
      "step": 29977
    },
    {
      "epoch": 1.632421909565525,
      "grad_norm": 0.5385703886541486,
      "learning_rate": 1.7198180038414814e-05,
      "loss": 11.81,
      "step": 29978
    },
    {
      "epoch": 1.6324763635621082,
      "grad_norm": 0.5266639322681721,
      "learning_rate": 1.7193235952227127e-05,
      "loss": 11.8471,
      "step": 29979
    },
    {
      "epoch": 1.6325308175586912,
      "grad_norm": 0.5411447754752525,
      "learning_rate": 1.7188292509958714e-05,
      "loss": 11.7058,
      "step": 29980
    },
    {
      "epoch": 1.6325852715552742,
      "grad_norm": 0.5698511022891626,
      "learning_rate": 1.7183349711648045e-05,
      "loss": 11.795,
      "step": 29981
    },
    {
      "epoch": 1.6326397255518572,
      "grad_norm": 0.5468815026330077,
      "learning_rate": 1.717840755733352e-05,
      "loss": 11.7333,
      "step": 29982
    },
    {
      "epoch": 1.6326941795484402,
      "grad_norm": 0.5504093319599296,
      "learning_rate": 1.7173466047053576e-05,
      "loss": 11.7767,
      "step": 29983
    },
    {
      "epoch": 1.6327486335450232,
      "grad_norm": 0.6155818544855903,
      "learning_rate": 1.7168525180846683e-05,
      "loss": 11.9369,
      "step": 29984
    },
    {
      "epoch": 1.6328030875416064,
      "grad_norm": 0.5670589884714078,
      "learning_rate": 1.71635849587512e-05,
      "loss": 11.8164,
      "step": 29985
    },
    {
      "epoch": 1.6328575415381894,
      "grad_norm": 0.607814367470063,
      "learning_rate": 1.7158645380805595e-05,
      "loss": 11.7662,
      "step": 29986
    },
    {
      "epoch": 1.6329119955347724,
      "grad_norm": 0.5492429874772824,
      "learning_rate": 1.7153706447048256e-05,
      "loss": 11.8007,
      "step": 29987
    },
    {
      "epoch": 1.6329664495313554,
      "grad_norm": 0.5837794851697979,
      "learning_rate": 1.714876815751757e-05,
      "loss": 11.8114,
      "step": 29988
    },
    {
      "epoch": 1.6330209035279384,
      "grad_norm": 0.6224607450061648,
      "learning_rate": 1.714383051225198e-05,
      "loss": 11.7728,
      "step": 29989
    },
    {
      "epoch": 1.6330753575245214,
      "grad_norm": 0.5772532331114856,
      "learning_rate": 1.7138893511289833e-05,
      "loss": 11.8376,
      "step": 29990
    },
    {
      "epoch": 1.6331298115211044,
      "grad_norm": 0.5929361195225834,
      "learning_rate": 1.7133957154669566e-05,
      "loss": 11.7147,
      "step": 29991
    },
    {
      "epoch": 1.6331842655176874,
      "grad_norm": 0.5619559270786373,
      "learning_rate": 1.7129021442429526e-05,
      "loss": 11.8793,
      "step": 29992
    },
    {
      "epoch": 1.6332387195142704,
      "grad_norm": 0.5084116920288405,
      "learning_rate": 1.712408637460814e-05,
      "loss": 11.6987,
      "step": 29993
    },
    {
      "epoch": 1.6332931735108533,
      "grad_norm": 0.6164379057478121,
      "learning_rate": 1.711915195124373e-05,
      "loss": 11.8974,
      "step": 29994
    },
    {
      "epoch": 1.6333476275074363,
      "grad_norm": 0.8771008723158041,
      "learning_rate": 1.7114218172374687e-05,
      "loss": 11.771,
      "step": 29995
    },
    {
      "epoch": 1.6334020815040193,
      "grad_norm": 0.6433309937496873,
      "learning_rate": 1.7109285038039414e-05,
      "loss": 11.9548,
      "step": 29996
    },
    {
      "epoch": 1.6334565355006023,
      "grad_norm": 0.5610217464026294,
      "learning_rate": 1.7104352548276238e-05,
      "loss": 11.8948,
      "step": 29997
    },
    {
      "epoch": 1.6335109894971853,
      "grad_norm": 0.5950032725760158,
      "learning_rate": 1.7099420703123492e-05,
      "loss": 11.9114,
      "step": 29998
    },
    {
      "epoch": 1.6335654434937683,
      "grad_norm": 0.5461232231821463,
      "learning_rate": 1.7094489502619583e-05,
      "loss": 11.7725,
      "step": 29999
    },
    {
      "epoch": 1.6336198974903513,
      "grad_norm": 0.6059168275598296,
      "learning_rate": 1.7089558946802808e-05,
      "loss": 11.9028,
      "step": 30000
    },
    {
      "epoch": 1.6336743514869343,
      "grad_norm": 0.6105096510741903,
      "learning_rate": 1.708462903571154e-05,
      "loss": 11.7295,
      "step": 30001
    },
    {
      "epoch": 1.6337288054835175,
      "grad_norm": 0.5402355451457493,
      "learning_rate": 1.7079699769384084e-05,
      "loss": 11.8044,
      "step": 30002
    },
    {
      "epoch": 1.6337832594801005,
      "grad_norm": 0.5329796961915664,
      "learning_rate": 1.7074771147858805e-05,
      "loss": 11.8456,
      "step": 30003
    },
    {
      "epoch": 1.6338377134766835,
      "grad_norm": 0.6321166473939128,
      "learning_rate": 1.7069843171173993e-05,
      "loss": 11.8749,
      "step": 30004
    },
    {
      "epoch": 1.6338921674732665,
      "grad_norm": 0.5538225118784644,
      "learning_rate": 1.706491583936799e-05,
      "loss": 11.7861,
      "step": 30005
    },
    {
      "epoch": 1.6339466214698495,
      "grad_norm": 0.5914003584658366,
      "learning_rate": 1.7059989152479126e-05,
      "loss": 11.7819,
      "step": 30006
    },
    {
      "epoch": 1.6340010754664325,
      "grad_norm": 0.5376738789432051,
      "learning_rate": 1.7055063110545698e-05,
      "loss": 11.7034,
      "step": 30007
    },
    {
      "epoch": 1.6340555294630157,
      "grad_norm": 0.5155795542157379,
      "learning_rate": 1.705013771360602e-05,
      "loss": 11.8084,
      "step": 30008
    },
    {
      "epoch": 1.6341099834595987,
      "grad_norm": 0.5703404727879696,
      "learning_rate": 1.7045212961698342e-05,
      "loss": 11.8154,
      "step": 30009
    },
    {
      "epoch": 1.6341644374561817,
      "grad_norm": 0.5368328682726928,
      "learning_rate": 1.704028885486101e-05,
      "loss": 11.849,
      "step": 30010
    },
    {
      "epoch": 1.6342188914527647,
      "grad_norm": 0.5565666632153429,
      "learning_rate": 1.7035365393132328e-05,
      "loss": 11.8997,
      "step": 30011
    },
    {
      "epoch": 1.6342733454493477,
      "grad_norm": 0.5732944831513871,
      "learning_rate": 1.7030442576550533e-05,
      "loss": 11.8533,
      "step": 30012
    },
    {
      "epoch": 1.6343277994459307,
      "grad_norm": 0.563069851826406,
      "learning_rate": 1.702552040515395e-05,
      "loss": 11.7898,
      "step": 30013
    },
    {
      "epoch": 1.6343822534425136,
      "grad_norm": 0.6865294138891473,
      "learning_rate": 1.702059887898081e-05,
      "loss": 11.859,
      "step": 30014
    },
    {
      "epoch": 1.6344367074390966,
      "grad_norm": 0.6114203703375761,
      "learning_rate": 1.7015677998069436e-05,
      "loss": 11.898,
      "step": 30015
    },
    {
      "epoch": 1.6344911614356796,
      "grad_norm": 0.5619269326249757,
      "learning_rate": 1.701075776245804e-05,
      "loss": 11.8784,
      "step": 30016
    },
    {
      "epoch": 1.6345456154322626,
      "grad_norm": 0.5496292266512631,
      "learning_rate": 1.7005838172184942e-05,
      "loss": 11.6915,
      "step": 30017
    },
    {
      "epoch": 1.6346000694288456,
      "grad_norm": 0.61453076425597,
      "learning_rate": 1.700091922728836e-05,
      "loss": 11.9333,
      "step": 30018
    },
    {
      "epoch": 1.6346545234254286,
      "grad_norm": 0.5557187034857363,
      "learning_rate": 1.6996000927806522e-05,
      "loss": 11.7656,
      "step": 30019
    },
    {
      "epoch": 1.6347089774220116,
      "grad_norm": 0.544668931532488,
      "learning_rate": 1.69910832737777e-05,
      "loss": 11.7086,
      "step": 30020
    },
    {
      "epoch": 1.6347634314185946,
      "grad_norm": 0.5320903825442154,
      "learning_rate": 1.6986166265240165e-05,
      "loss": 11.9186,
      "step": 30021
    },
    {
      "epoch": 1.6348178854151776,
      "grad_norm": 0.5975450327333364,
      "learning_rate": 1.698124990223209e-05,
      "loss": 11.8909,
      "step": 30022
    },
    {
      "epoch": 1.6348723394117606,
      "grad_norm": 0.6641660955700938,
      "learning_rate": 1.6976334184791766e-05,
      "loss": 11.7259,
      "step": 30023
    },
    {
      "epoch": 1.6349267934083436,
      "grad_norm": 0.565918647424538,
      "learning_rate": 1.6971419112957366e-05,
      "loss": 11.833,
      "step": 30024
    },
    {
      "epoch": 1.6349812474049266,
      "grad_norm": 0.5664943585391604,
      "learning_rate": 1.696650468676716e-05,
      "loss": 11.8858,
      "step": 30025
    },
    {
      "epoch": 1.6350357014015098,
      "grad_norm": 0.5758799834049017,
      "learning_rate": 1.6961590906259317e-05,
      "loss": 11.7644,
      "step": 30026
    },
    {
      "epoch": 1.6350901553980928,
      "grad_norm": 0.5871389275039496,
      "learning_rate": 1.6956677771472086e-05,
      "loss": 11.6974,
      "step": 30027
    },
    {
      "epoch": 1.6351446093946758,
      "grad_norm": 0.5864890693843933,
      "learning_rate": 1.6951765282443665e-05,
      "loss": 11.8469,
      "step": 30028
    },
    {
      "epoch": 1.6351990633912588,
      "grad_norm": 0.5496732357600271,
      "learning_rate": 1.6946853439212208e-05,
      "loss": 11.7803,
      "step": 30029
    },
    {
      "epoch": 1.6352535173878417,
      "grad_norm": 0.54321256875059,
      "learning_rate": 1.6941942241815976e-05,
      "loss": 11.8153,
      "step": 30030
    },
    {
      "epoch": 1.6353079713844247,
      "grad_norm": 0.49234013000421545,
      "learning_rate": 1.6937031690293114e-05,
      "loss": 11.8611,
      "step": 30031
    },
    {
      "epoch": 1.635362425381008,
      "grad_norm": 0.5436932201135376,
      "learning_rate": 1.693212178468181e-05,
      "loss": 11.7982,
      "step": 30032
    },
    {
      "epoch": 1.635416879377591,
      "grad_norm": 0.6021818624904028,
      "learning_rate": 1.6927212525020286e-05,
      "loss": 11.8615,
      "step": 30033
    },
    {
      "epoch": 1.635471333374174,
      "grad_norm": 0.5271136833375871,
      "learning_rate": 1.6922303911346672e-05,
      "loss": 11.6763,
      "step": 30034
    },
    {
      "epoch": 1.635525787370757,
      "grad_norm": 0.5546249424141428,
      "learning_rate": 1.6917395943699167e-05,
      "loss": 11.7943,
      "step": 30035
    },
    {
      "epoch": 1.63558024136734,
      "grad_norm": 0.5544407333531863,
      "learning_rate": 1.6912488622115906e-05,
      "loss": 11.798,
      "step": 30036
    },
    {
      "epoch": 1.635634695363923,
      "grad_norm": 0.55837627104082,
      "learning_rate": 1.69075819466351e-05,
      "loss": 11.7274,
      "step": 30037
    },
    {
      "epoch": 1.635689149360506,
      "grad_norm": 0.4966642026542986,
      "learning_rate": 1.6902675917294864e-05,
      "loss": 11.7882,
      "step": 30038
    },
    {
      "epoch": 1.635743603357089,
      "grad_norm": 0.5733304750743955,
      "learning_rate": 1.689777053413333e-05,
      "loss": 11.9074,
      "step": 30039
    },
    {
      "epoch": 1.635798057353672,
      "grad_norm": 0.5606136053844597,
      "learning_rate": 1.6892865797188683e-05,
      "loss": 11.854,
      "step": 30040
    },
    {
      "epoch": 1.635852511350255,
      "grad_norm": 0.5457936238075886,
      "learning_rate": 1.6887961706499033e-05,
      "loss": 11.8125,
      "step": 30041
    },
    {
      "epoch": 1.6359069653468379,
      "grad_norm": 0.5306080361527131,
      "learning_rate": 1.6883058262102558e-05,
      "loss": 11.8243,
      "step": 30042
    },
    {
      "epoch": 1.6359614193434209,
      "grad_norm": 0.5193432870429455,
      "learning_rate": 1.6878155464037338e-05,
      "loss": 11.7919,
      "step": 30043
    },
    {
      "epoch": 1.6360158733400039,
      "grad_norm": 0.6253338911875089,
      "learning_rate": 1.687325331234152e-05,
      "loss": 11.9306,
      "step": 30044
    },
    {
      "epoch": 1.6360703273365869,
      "grad_norm": 0.5701814352057775,
      "learning_rate": 1.686835180705324e-05,
      "loss": 11.8268,
      "step": 30045
    },
    {
      "epoch": 1.6361247813331699,
      "grad_norm": 0.5394099316510441,
      "learning_rate": 1.686345094821058e-05,
      "loss": 11.8104,
      "step": 30046
    },
    {
      "epoch": 1.6361792353297528,
      "grad_norm": 0.5045219247526627,
      "learning_rate": 1.6858550735851695e-05,
      "loss": 11.6289,
      "step": 30047
    },
    {
      "epoch": 1.6362336893263358,
      "grad_norm": 0.5671328828710982,
      "learning_rate": 1.685365117001466e-05,
      "loss": 11.6058,
      "step": 30048
    },
    {
      "epoch": 1.636288143322919,
      "grad_norm": 0.538110442919234,
      "learning_rate": 1.6848752250737554e-05,
      "loss": 11.8118,
      "step": 30049
    },
    {
      "epoch": 1.636342597319502,
      "grad_norm": 0.522312921593881,
      "learning_rate": 1.684385397805851e-05,
      "loss": 11.7837,
      "step": 30050
    },
    {
      "epoch": 1.636397051316085,
      "grad_norm": 0.5498876554399803,
      "learning_rate": 1.683895635201559e-05,
      "loss": 11.7524,
      "step": 30051
    },
    {
      "epoch": 1.636451505312668,
      "grad_norm": 0.5556707509641179,
      "learning_rate": 1.6834059372646904e-05,
      "loss": 11.8493,
      "step": 30052
    },
    {
      "epoch": 1.636505959309251,
      "grad_norm": 0.5582342262947769,
      "learning_rate": 1.6829163039990503e-05,
      "loss": 11.7411,
      "step": 30053
    },
    {
      "epoch": 1.636560413305834,
      "grad_norm": 0.5381200686560274,
      "learning_rate": 1.682426735408448e-05,
      "loss": 11.8162,
      "step": 30054
    },
    {
      "epoch": 1.6366148673024172,
      "grad_norm": 0.5421399558650414,
      "learning_rate": 1.6819372314966932e-05,
      "loss": 11.8233,
      "step": 30055
    },
    {
      "epoch": 1.6366693212990002,
      "grad_norm": 0.5242904898004394,
      "learning_rate": 1.6814477922675853e-05,
      "loss": 11.7721,
      "step": 30056
    },
    {
      "epoch": 1.6367237752955832,
      "grad_norm": 0.5682402609569833,
      "learning_rate": 1.6809584177249405e-05,
      "loss": 11.8661,
      "step": 30057
    },
    {
      "epoch": 1.6367782292921662,
      "grad_norm": 0.5518682391718505,
      "learning_rate": 1.6804691078725533e-05,
      "loss": 11.7571,
      "step": 30058
    },
    {
      "epoch": 1.6368326832887492,
      "grad_norm": 0.5388755418216244,
      "learning_rate": 1.6799798627142326e-05,
      "loss": 11.8295,
      "step": 30059
    },
    {
      "epoch": 1.6368871372853322,
      "grad_norm": 0.5676185457478856,
      "learning_rate": 1.679490682253787e-05,
      "loss": 11.8375,
      "step": 30060
    },
    {
      "epoch": 1.6369415912819152,
      "grad_norm": 0.5274055204742559,
      "learning_rate": 1.679001566495014e-05,
      "loss": 11.8399,
      "step": 30061
    },
    {
      "epoch": 1.6369960452784982,
      "grad_norm": 0.5763032762948113,
      "learning_rate": 1.6785125154417235e-05,
      "loss": 11.8082,
      "step": 30062
    },
    {
      "epoch": 1.6370504992750812,
      "grad_norm": 0.5071738550733418,
      "learning_rate": 1.678023529097712e-05,
      "loss": 11.8153,
      "step": 30063
    },
    {
      "epoch": 1.6371049532716642,
      "grad_norm": 0.5430073858298432,
      "learning_rate": 1.6775346074667887e-05,
      "loss": 11.736,
      "step": 30064
    },
    {
      "epoch": 1.6371594072682472,
      "grad_norm": 0.5404312233014523,
      "learning_rate": 1.677045750552748e-05,
      "loss": 11.8809,
      "step": 30065
    },
    {
      "epoch": 1.6372138612648302,
      "grad_norm": 0.5368471376442372,
      "learning_rate": 1.6765569583593964e-05,
      "loss": 11.7555,
      "step": 30066
    },
    {
      "epoch": 1.6372683152614131,
      "grad_norm": 0.537025994929866,
      "learning_rate": 1.676068230890536e-05,
      "loss": 11.6391,
      "step": 30067
    },
    {
      "epoch": 1.6373227692579961,
      "grad_norm": 0.5311610970612696,
      "learning_rate": 1.6755795681499653e-05,
      "loss": 11.7425,
      "step": 30068
    },
    {
      "epoch": 1.6373772232545791,
      "grad_norm": 0.5370816819715026,
      "learning_rate": 1.6750909701414808e-05,
      "loss": 11.7935,
      "step": 30069
    },
    {
      "epoch": 1.6374316772511621,
      "grad_norm": 0.5189362084418445,
      "learning_rate": 1.6746024368688872e-05,
      "loss": 11.8864,
      "step": 30070
    },
    {
      "epoch": 1.6374861312477451,
      "grad_norm": 0.5576943627089798,
      "learning_rate": 1.6741139683359796e-05,
      "loss": 11.616,
      "step": 30071
    },
    {
      "epoch": 1.6375405852443283,
      "grad_norm": 0.5894375453579125,
      "learning_rate": 1.6736255645465602e-05,
      "loss": 11.7119,
      "step": 30072
    },
    {
      "epoch": 1.6375950392409113,
      "grad_norm": 0.5471688132992194,
      "learning_rate": 1.673137225504422e-05,
      "loss": 11.6751,
      "step": 30073
    },
    {
      "epoch": 1.6376494932374943,
      "grad_norm": 0.5673981566929037,
      "learning_rate": 1.672648951213368e-05,
      "loss": 11.8766,
      "step": 30074
    },
    {
      "epoch": 1.6377039472340773,
      "grad_norm": 0.5408324528312617,
      "learning_rate": 1.672160741677189e-05,
      "loss": 11.7351,
      "step": 30075
    },
    {
      "epoch": 1.6377584012306603,
      "grad_norm": 0.5820788572291198,
      "learning_rate": 1.671672596899686e-05,
      "loss": 11.8543,
      "step": 30076
    },
    {
      "epoch": 1.6378128552272433,
      "grad_norm": 0.557561109908252,
      "learning_rate": 1.6711845168846553e-05,
      "loss": 11.6013,
      "step": 30077
    },
    {
      "epoch": 1.6378673092238265,
      "grad_norm": 0.5336726556037551,
      "learning_rate": 1.67069650163589e-05,
      "loss": 11.845,
      "step": 30078
    },
    {
      "epoch": 1.6379217632204095,
      "grad_norm": 0.6094915021122238,
      "learning_rate": 1.6702085511571863e-05,
      "loss": 11.8543,
      "step": 30079
    },
    {
      "epoch": 1.6379762172169925,
      "grad_norm": 0.5317362788737234,
      "learning_rate": 1.6697206654523344e-05,
      "loss": 11.8369,
      "step": 30080
    },
    {
      "epoch": 1.6380306712135755,
      "grad_norm": 0.5327956438040591,
      "learning_rate": 1.6692328445251325e-05,
      "loss": 11.7825,
      "step": 30081
    },
    {
      "epoch": 1.6380851252101585,
      "grad_norm": 0.5461396379068768,
      "learning_rate": 1.6687450883793753e-05,
      "loss": 11.8955,
      "step": 30082
    },
    {
      "epoch": 1.6381395792067415,
      "grad_norm": 0.5050090392465167,
      "learning_rate": 1.668257397018851e-05,
      "loss": 11.8116,
      "step": 30083
    },
    {
      "epoch": 1.6381940332033245,
      "grad_norm": 0.5690084558942501,
      "learning_rate": 1.667769770447357e-05,
      "loss": 11.6996,
      "step": 30084
    },
    {
      "epoch": 1.6382484871999075,
      "grad_norm": 0.5827578713703137,
      "learning_rate": 1.6672822086686803e-05,
      "loss": 11.8359,
      "step": 30085
    },
    {
      "epoch": 1.6383029411964904,
      "grad_norm": 0.542803027702863,
      "learning_rate": 1.6667947116866168e-05,
      "loss": 11.7655,
      "step": 30086
    },
    {
      "epoch": 1.6383573951930734,
      "grad_norm": 0.5515404128187872,
      "learning_rate": 1.6663072795049515e-05,
      "loss": 11.8565,
      "step": 30087
    },
    {
      "epoch": 1.6384118491896564,
      "grad_norm": 0.5164036976239055,
      "learning_rate": 1.665819912127482e-05,
      "loss": 11.8333,
      "step": 30088
    },
    {
      "epoch": 1.6384663031862394,
      "grad_norm": 0.592115041557363,
      "learning_rate": 1.6653326095579946e-05,
      "loss": 11.6284,
      "step": 30089
    },
    {
      "epoch": 1.6385207571828224,
      "grad_norm": 0.5585561045423707,
      "learning_rate": 1.6648453718002765e-05,
      "loss": 11.8202,
      "step": 30090
    },
    {
      "epoch": 1.6385752111794054,
      "grad_norm": 0.5429523758392679,
      "learning_rate": 1.664358198858117e-05,
      "loss": 11.7837,
      "step": 30091
    },
    {
      "epoch": 1.6386296651759884,
      "grad_norm": 0.5787009703870182,
      "learning_rate": 1.6638710907353095e-05,
      "loss": 11.9051,
      "step": 30092
    },
    {
      "epoch": 1.6386841191725714,
      "grad_norm": 0.5372462408997021,
      "learning_rate": 1.6633840474356366e-05,
      "loss": 11.5681,
      "step": 30093
    },
    {
      "epoch": 1.6387385731691544,
      "grad_norm": 0.5849190833703195,
      "learning_rate": 1.6628970689628887e-05,
      "loss": 11.9292,
      "step": 30094
    },
    {
      "epoch": 1.6387930271657374,
      "grad_norm": 0.5209925881249817,
      "learning_rate": 1.6624101553208492e-05,
      "loss": 11.882,
      "step": 30095
    },
    {
      "epoch": 1.6388474811623206,
      "grad_norm": 0.5124654393783703,
      "learning_rate": 1.661923306513309e-05,
      "loss": 11.7205,
      "step": 30096
    },
    {
      "epoch": 1.6389019351589036,
      "grad_norm": 0.560753906045654,
      "learning_rate": 1.6614365225440488e-05,
      "loss": 11.805,
      "step": 30097
    },
    {
      "epoch": 1.6389563891554866,
      "grad_norm": 0.5489227700795934,
      "learning_rate": 1.6609498034168603e-05,
      "loss": 11.9082,
      "step": 30098
    },
    {
      "epoch": 1.6390108431520696,
      "grad_norm": 0.5765246165371151,
      "learning_rate": 1.6604631491355226e-05,
      "loss": 11.8043,
      "step": 30099
    },
    {
      "epoch": 1.6390652971486526,
      "grad_norm": 0.5548932072939367,
      "learning_rate": 1.6599765597038207e-05,
      "loss": 11.8307,
      "step": 30100
    },
    {
      "epoch": 1.6391197511452358,
      "grad_norm": 0.530195112560014,
      "learning_rate": 1.6594900351255428e-05,
      "loss": 11.841,
      "step": 30101
    },
    {
      "epoch": 1.6391742051418188,
      "grad_norm": 0.5312002150356473,
      "learning_rate": 1.659003575404465e-05,
      "loss": 11.6799,
      "step": 30102
    },
    {
      "epoch": 1.6392286591384018,
      "grad_norm": 0.5707074796569253,
      "learning_rate": 1.6585171805443754e-05,
      "loss": 11.8772,
      "step": 30103
    },
    {
      "epoch": 1.6392831131349848,
      "grad_norm": 0.6623149603267436,
      "learning_rate": 1.658030850549057e-05,
      "loss": 11.9273,
      "step": 30104
    },
    {
      "epoch": 1.6393375671315678,
      "grad_norm": 0.5247996329746817,
      "learning_rate": 1.657544585422287e-05,
      "loss": 11.7142,
      "step": 30105
    },
    {
      "epoch": 1.6393920211281507,
      "grad_norm": 0.5231783955010734,
      "learning_rate": 1.6570583851678524e-05,
      "loss": 11.8849,
      "step": 30106
    },
    {
      "epoch": 1.6394464751247337,
      "grad_norm": 0.5614133123226214,
      "learning_rate": 1.6565722497895287e-05,
      "loss": 11.7412,
      "step": 30107
    },
    {
      "epoch": 1.6395009291213167,
      "grad_norm": 0.575766303853106,
      "learning_rate": 1.6560861792910997e-05,
      "loss": 11.9719,
      "step": 30108
    },
    {
      "epoch": 1.6395553831178997,
      "grad_norm": 0.5500778176237447,
      "learning_rate": 1.655600173676345e-05,
      "loss": 11.7736,
      "step": 30109
    },
    {
      "epoch": 1.6396098371144827,
      "grad_norm": 0.5503625238016333,
      "learning_rate": 1.655114232949039e-05,
      "loss": 11.8682,
      "step": 30110
    },
    {
      "epoch": 1.6396642911110657,
      "grad_norm": 0.5653805119479136,
      "learning_rate": 1.6546283571129672e-05,
      "loss": 11.9648,
      "step": 30111
    },
    {
      "epoch": 1.6397187451076487,
      "grad_norm": 0.5157511129896286,
      "learning_rate": 1.6541425461719027e-05,
      "loss": 11.7993,
      "step": 30112
    },
    {
      "epoch": 1.6397731991042317,
      "grad_norm": 0.5705645957327735,
      "learning_rate": 1.653656800129628e-05,
      "loss": 11.7648,
      "step": 30113
    },
    {
      "epoch": 1.6398276531008147,
      "grad_norm": 0.5515964624708107,
      "learning_rate": 1.6531711189899145e-05,
      "loss": 11.7286,
      "step": 30114
    },
    {
      "epoch": 1.6398821070973977,
      "grad_norm": 0.5069048453469294,
      "learning_rate": 1.6526855027565423e-05,
      "loss": 11.8409,
      "step": 30115
    },
    {
      "epoch": 1.6399365610939807,
      "grad_norm": 0.5606669819797543,
      "learning_rate": 1.652199951433291e-05,
      "loss": 11.8605,
      "step": 30116
    },
    {
      "epoch": 1.6399910150905637,
      "grad_norm": 0.6099907291433857,
      "learning_rate": 1.651714465023929e-05,
      "loss": 11.9511,
      "step": 30117
    },
    {
      "epoch": 1.6400454690871467,
      "grad_norm": 0.5727123118523072,
      "learning_rate": 1.6512290435322385e-05,
      "loss": 11.6288,
      "step": 30118
    },
    {
      "epoch": 1.6400999230837299,
      "grad_norm": 0.4939985069499722,
      "learning_rate": 1.6507436869619907e-05,
      "loss": 11.7311,
      "step": 30119
    },
    {
      "epoch": 1.6401543770803129,
      "grad_norm": 0.5129205285412832,
      "learning_rate": 1.650258395316957e-05,
      "loss": 11.6636,
      "step": 30120
    },
    {
      "epoch": 1.6402088310768959,
      "grad_norm": 0.5085701211172086,
      "learning_rate": 1.6497731686009176e-05,
      "loss": 11.7575,
      "step": 30121
    },
    {
      "epoch": 1.6402632850734788,
      "grad_norm": 0.5101237714626635,
      "learning_rate": 1.6492880068176396e-05,
      "loss": 11.7947,
      "step": 30122
    },
    {
      "epoch": 1.6403177390700618,
      "grad_norm": 0.5512918945382997,
      "learning_rate": 1.6488029099709023e-05,
      "loss": 11.8164,
      "step": 30123
    },
    {
      "epoch": 1.6403721930666448,
      "grad_norm": 0.5242432731708799,
      "learning_rate": 1.6483178780644702e-05,
      "loss": 11.7004,
      "step": 30124
    },
    {
      "epoch": 1.640426647063228,
      "grad_norm": 0.568338929432154,
      "learning_rate": 1.6478329111021185e-05,
      "loss": 11.9587,
      "step": 30125
    },
    {
      "epoch": 1.640481101059811,
      "grad_norm": 0.507992246282294,
      "learning_rate": 1.6473480090876226e-05,
      "loss": 11.7301,
      "step": 30126
    },
    {
      "epoch": 1.640535555056394,
      "grad_norm": 0.5983717861836699,
      "learning_rate": 1.646863172024746e-05,
      "loss": 11.8204,
      "step": 30127
    },
    {
      "epoch": 1.640590009052977,
      "grad_norm": 0.5772742570237682,
      "learning_rate": 1.6463783999172665e-05,
      "loss": 11.7378,
      "step": 30128
    },
    {
      "epoch": 1.64064446304956,
      "grad_norm": 0.5491561143340316,
      "learning_rate": 1.6458936927689462e-05,
      "loss": 11.8633,
      "step": 30129
    },
    {
      "epoch": 1.640698917046143,
      "grad_norm": 0.5221765408554284,
      "learning_rate": 1.6454090505835562e-05,
      "loss": 11.8537,
      "step": 30130
    },
    {
      "epoch": 1.640753371042726,
      "grad_norm": 0.5578128299948817,
      "learning_rate": 1.6449244733648695e-05,
      "loss": 11.8589,
      "step": 30131
    },
    {
      "epoch": 1.640807825039309,
      "grad_norm": 0.6028340515247499,
      "learning_rate": 1.6444399611166484e-05,
      "loss": 11.8208,
      "step": 30132
    },
    {
      "epoch": 1.640862279035892,
      "grad_norm": 0.5399518459484299,
      "learning_rate": 1.6439555138426666e-05,
      "loss": 11.7228,
      "step": 30133
    },
    {
      "epoch": 1.640916733032475,
      "grad_norm": 0.5902670897304149,
      "learning_rate": 1.6434711315466844e-05,
      "loss": 11.8582,
      "step": 30134
    },
    {
      "epoch": 1.640971187029058,
      "grad_norm": 0.5743644022290435,
      "learning_rate": 1.6429868142324757e-05,
      "loss": 11.8942,
      "step": 30135
    },
    {
      "epoch": 1.641025641025641,
      "grad_norm": 0.5386308223798555,
      "learning_rate": 1.642502561903799e-05,
      "loss": 11.7774,
      "step": 30136
    },
    {
      "epoch": 1.641080095022224,
      "grad_norm": 0.5331663092826117,
      "learning_rate": 1.6420183745644258e-05,
      "loss": 11.8669,
      "step": 30137
    },
    {
      "epoch": 1.641134549018807,
      "grad_norm": 0.5882156028799918,
      "learning_rate": 1.6415342522181233e-05,
      "loss": 11.8053,
      "step": 30138
    },
    {
      "epoch": 1.64118900301539,
      "grad_norm": 0.5164712670058067,
      "learning_rate": 1.6410501948686473e-05,
      "loss": 11.8903,
      "step": 30139
    },
    {
      "epoch": 1.641243457011973,
      "grad_norm": 0.596815269015819,
      "learning_rate": 1.6405662025197665e-05,
      "loss": 11.8447,
      "step": 30140
    },
    {
      "epoch": 1.641297911008556,
      "grad_norm": 0.5299777840801392,
      "learning_rate": 1.6400822751752475e-05,
      "loss": 11.7045,
      "step": 30141
    },
    {
      "epoch": 1.6413523650051391,
      "grad_norm": 0.5109550589165297,
      "learning_rate": 1.6395984128388487e-05,
      "loss": 11.8194,
      "step": 30142
    },
    {
      "epoch": 1.6414068190017221,
      "grad_norm": 0.600931446336842,
      "learning_rate": 1.6391146155143367e-05,
      "loss": 11.857,
      "step": 30143
    },
    {
      "epoch": 1.6414612729983051,
      "grad_norm": 0.5144424977095644,
      "learning_rate": 1.63863088320547e-05,
      "loss": 11.6053,
      "step": 30144
    },
    {
      "epoch": 1.6415157269948881,
      "grad_norm": 0.5553563956708736,
      "learning_rate": 1.6381472159160126e-05,
      "loss": 11.8207,
      "step": 30145
    },
    {
      "epoch": 1.6415701809914711,
      "grad_norm": 0.5406254046813125,
      "learning_rate": 1.6376636136497235e-05,
      "loss": 11.8224,
      "step": 30146
    },
    {
      "epoch": 1.641624634988054,
      "grad_norm": 0.6141331659409237,
      "learning_rate": 1.6371800764103663e-05,
      "loss": 11.8021,
      "step": 30147
    },
    {
      "epoch": 1.6416790889846373,
      "grad_norm": 0.5601358481483191,
      "learning_rate": 1.6366966042016996e-05,
      "loss": 11.6448,
      "step": 30148
    },
    {
      "epoch": 1.6417335429812203,
      "grad_norm": 0.484738020824408,
      "learning_rate": 1.6362131970274796e-05,
      "loss": 11.8006,
      "step": 30149
    },
    {
      "epoch": 1.6417879969778033,
      "grad_norm": 0.5427513378769239,
      "learning_rate": 1.6357298548914722e-05,
      "loss": 11.8511,
      "step": 30150
    },
    {
      "epoch": 1.6418424509743863,
      "grad_norm": 0.5756763379337209,
      "learning_rate": 1.635246577797428e-05,
      "loss": 11.8875,
      "step": 30151
    },
    {
      "epoch": 1.6418969049709693,
      "grad_norm": 0.5385630663061246,
      "learning_rate": 1.63476336574911e-05,
      "loss": 11.7996,
      "step": 30152
    },
    {
      "epoch": 1.6419513589675523,
      "grad_norm": 0.6060538780518313,
      "learning_rate": 1.6342802187502782e-05,
      "loss": 11.7774,
      "step": 30153
    },
    {
      "epoch": 1.6420058129641353,
      "grad_norm": 0.5927889484959951,
      "learning_rate": 1.6337971368046823e-05,
      "loss": 11.7934,
      "step": 30154
    },
    {
      "epoch": 1.6420602669607183,
      "grad_norm": 0.5672123417583669,
      "learning_rate": 1.633314119916086e-05,
      "loss": 11.8023,
      "step": 30155
    },
    {
      "epoch": 1.6421147209573013,
      "grad_norm": 0.5047756720524929,
      "learning_rate": 1.63283116808824e-05,
      "loss": 11.8242,
      "step": 30156
    },
    {
      "epoch": 1.6421691749538843,
      "grad_norm": 0.5481889386605263,
      "learning_rate": 1.6323482813249046e-05,
      "loss": 11.8706,
      "step": 30157
    },
    {
      "epoch": 1.6422236289504673,
      "grad_norm": 0.5739071727362628,
      "learning_rate": 1.631865459629832e-05,
      "loss": 11.9472,
      "step": 30158
    },
    {
      "epoch": 1.6422780829470502,
      "grad_norm": 0.5230641715863068,
      "learning_rate": 1.6313827030067742e-05,
      "loss": 11.7824,
      "step": 30159
    },
    {
      "epoch": 1.6423325369436332,
      "grad_norm": 0.533055946320717,
      "learning_rate": 1.6309000114594907e-05,
      "loss": 11.7953,
      "step": 30160
    },
    {
      "epoch": 1.6423869909402162,
      "grad_norm": 0.5262768491491955,
      "learning_rate": 1.6304173849917293e-05,
      "loss": 11.7402,
      "step": 30161
    },
    {
      "epoch": 1.6424414449367992,
      "grad_norm": 0.5180209889053046,
      "learning_rate": 1.6299348236072487e-05,
      "loss": 11.7873,
      "step": 30162
    },
    {
      "epoch": 1.6424958989333822,
      "grad_norm": 0.5472664669211041,
      "learning_rate": 1.629452327309796e-05,
      "loss": 11.7478,
      "step": 30163
    },
    {
      "epoch": 1.6425503529299652,
      "grad_norm": 0.5622909712539453,
      "learning_rate": 1.6289698961031265e-05,
      "loss": 11.917,
      "step": 30164
    },
    {
      "epoch": 1.6426048069265482,
      "grad_norm": 0.5305997335205761,
      "learning_rate": 1.628487529990992e-05,
      "loss": 11.813,
      "step": 30165
    },
    {
      "epoch": 1.6426592609231314,
      "grad_norm": 0.49457213538677725,
      "learning_rate": 1.6280052289771407e-05,
      "loss": 11.7294,
      "step": 30166
    },
    {
      "epoch": 1.6427137149197144,
      "grad_norm": 0.5504786587383643,
      "learning_rate": 1.627522993065327e-05,
      "loss": 11.8793,
      "step": 30167
    },
    {
      "epoch": 1.6427681689162974,
      "grad_norm": 0.5563024258222868,
      "learning_rate": 1.6270408222592992e-05,
      "loss": 11.7451,
      "step": 30168
    },
    {
      "epoch": 1.6428226229128804,
      "grad_norm": 0.5538065961453419,
      "learning_rate": 1.6265587165628026e-05,
      "loss": 11.8441,
      "step": 30169
    },
    {
      "epoch": 1.6428770769094634,
      "grad_norm": 0.5133949483620408,
      "learning_rate": 1.6260766759795932e-05,
      "loss": 11.8452,
      "step": 30170
    },
    {
      "epoch": 1.6429315309060466,
      "grad_norm": 0.537967420213983,
      "learning_rate": 1.625594700513413e-05,
      "loss": 11.7883,
      "step": 30171
    },
    {
      "epoch": 1.6429859849026296,
      "grad_norm": 0.5286085004048998,
      "learning_rate": 1.6251127901680154e-05,
      "loss": 11.8057,
      "step": 30172
    },
    {
      "epoch": 1.6430404388992126,
      "grad_norm": 0.5052098441762602,
      "learning_rate": 1.624630944947142e-05,
      "loss": 11.702,
      "step": 30173
    },
    {
      "epoch": 1.6430948928957956,
      "grad_norm": 0.6221990772921728,
      "learning_rate": 1.6241491648545425e-05,
      "loss": 12.0073,
      "step": 30174
    },
    {
      "epoch": 1.6431493468923786,
      "grad_norm": 0.5572868259678778,
      "learning_rate": 1.6236674498939674e-05,
      "loss": 11.7741,
      "step": 30175
    },
    {
      "epoch": 1.6432038008889616,
      "grad_norm": 0.6730592903415611,
      "learning_rate": 1.623185800069157e-05,
      "loss": 11.8505,
      "step": 30176
    },
    {
      "epoch": 1.6432582548855446,
      "grad_norm": 0.5398495166799504,
      "learning_rate": 1.6227042153838602e-05,
      "loss": 11.8462,
      "step": 30177
    },
    {
      "epoch": 1.6433127088821275,
      "grad_norm": 0.5434122483300948,
      "learning_rate": 1.6222226958418196e-05,
      "loss": 11.8473,
      "step": 30178
    },
    {
      "epoch": 1.6433671628787105,
      "grad_norm": 0.5141177947597703,
      "learning_rate": 1.621741241446778e-05,
      "loss": 11.7393,
      "step": 30179
    },
    {
      "epoch": 1.6434216168752935,
      "grad_norm": 0.5144025252010122,
      "learning_rate": 1.6212598522024846e-05,
      "loss": 11.7931,
      "step": 30180
    },
    {
      "epoch": 1.6434760708718765,
      "grad_norm": 0.5644670811107217,
      "learning_rate": 1.6207785281126776e-05,
      "loss": 11.7436,
      "step": 30181
    },
    {
      "epoch": 1.6435305248684595,
      "grad_norm": 0.5343302417104526,
      "learning_rate": 1.6202972691811036e-05,
      "loss": 11.8889,
      "step": 30182
    },
    {
      "epoch": 1.6435849788650425,
      "grad_norm": 0.5538776457357333,
      "learning_rate": 1.619816075411501e-05,
      "loss": 11.7634,
      "step": 30183
    },
    {
      "epoch": 1.6436394328616255,
      "grad_norm": 0.5331346245091763,
      "learning_rate": 1.619334946807617e-05,
      "loss": 11.6585,
      "step": 30184
    },
    {
      "epoch": 1.6436938868582085,
      "grad_norm": 0.5369501897949668,
      "learning_rate": 1.618853883373187e-05,
      "loss": 11.5974,
      "step": 30185
    },
    {
      "epoch": 1.6437483408547915,
      "grad_norm": 0.549389119772309,
      "learning_rate": 1.618372885111955e-05,
      "loss": 11.8103,
      "step": 30186
    },
    {
      "epoch": 1.6438027948513745,
      "grad_norm": 0.5392353962985108,
      "learning_rate": 1.6178919520276636e-05,
      "loss": 11.8543,
      "step": 30187
    },
    {
      "epoch": 1.6438572488479575,
      "grad_norm": 0.5745473003725531,
      "learning_rate": 1.6174110841240498e-05,
      "loss": 11.8401,
      "step": 30188
    },
    {
      "epoch": 1.6439117028445407,
      "grad_norm": 0.5425281327084583,
      "learning_rate": 1.6169302814048505e-05,
      "loss": 11.756,
      "step": 30189
    },
    {
      "epoch": 1.6439661568411237,
      "grad_norm": 0.5404762589688521,
      "learning_rate": 1.6164495438738093e-05,
      "loss": 11.7027,
      "step": 30190
    },
    {
      "epoch": 1.6440206108377067,
      "grad_norm": 0.5224215118945175,
      "learning_rate": 1.615968871534661e-05,
      "loss": 11.7933,
      "step": 30191
    },
    {
      "epoch": 1.6440750648342897,
      "grad_norm": 0.5558364736720943,
      "learning_rate": 1.6154882643911462e-05,
      "loss": 11.8388,
      "step": 30192
    },
    {
      "epoch": 1.6441295188308727,
      "grad_norm": 0.5201611773068436,
      "learning_rate": 1.6150077224469982e-05,
      "loss": 11.7992,
      "step": 30193
    },
    {
      "epoch": 1.6441839728274557,
      "grad_norm": 0.6008427608907387,
      "learning_rate": 1.61452724570596e-05,
      "loss": 11.9217,
      "step": 30194
    },
    {
      "epoch": 1.6442384268240389,
      "grad_norm": 0.5242592097140267,
      "learning_rate": 1.6140468341717606e-05,
      "loss": 11.8247,
      "step": 30195
    },
    {
      "epoch": 1.6442928808206219,
      "grad_norm": 0.5179011124989028,
      "learning_rate": 1.6135664878481427e-05,
      "loss": 11.7884,
      "step": 30196
    },
    {
      "epoch": 1.6443473348172049,
      "grad_norm": 0.5513964123639697,
      "learning_rate": 1.6130862067388353e-05,
      "loss": 11.8346,
      "step": 30197
    },
    {
      "epoch": 1.6444017888137878,
      "grad_norm": 0.5532758824760037,
      "learning_rate": 1.6126059908475778e-05,
      "loss": 11.8882,
      "step": 30198
    },
    {
      "epoch": 1.6444562428103708,
      "grad_norm": 0.5063132610517528,
      "learning_rate": 1.6121258401781035e-05,
      "loss": 11.7788,
      "step": 30199
    },
    {
      "epoch": 1.6445106968069538,
      "grad_norm": 0.5780013179652417,
      "learning_rate": 1.6116457547341425e-05,
      "loss": 11.7088,
      "step": 30200
    },
    {
      "epoch": 1.6445651508035368,
      "grad_norm": 0.5981226697750344,
      "learning_rate": 1.6111657345194308e-05,
      "loss": 11.7495,
      "step": 30201
    },
    {
      "epoch": 1.6446196048001198,
      "grad_norm": 0.5244753579200823,
      "learning_rate": 1.6106857795377038e-05,
      "loss": 11.8968,
      "step": 30202
    },
    {
      "epoch": 1.6446740587967028,
      "grad_norm": 0.5259705428364313,
      "learning_rate": 1.6102058897926886e-05,
      "loss": 11.747,
      "step": 30203
    },
    {
      "epoch": 1.6447285127932858,
      "grad_norm": 0.6379847630238716,
      "learning_rate": 1.609726065288122e-05,
      "loss": 11.8394,
      "step": 30204
    },
    {
      "epoch": 1.6447829667898688,
      "grad_norm": 0.5361327029031906,
      "learning_rate": 1.609246306027731e-05,
      "loss": 11.8516,
      "step": 30205
    },
    {
      "epoch": 1.6448374207864518,
      "grad_norm": 0.46549869099427416,
      "learning_rate": 1.608766612015249e-05,
      "loss": 11.7019,
      "step": 30206
    },
    {
      "epoch": 1.6448918747830348,
      "grad_norm": 0.5261882995306035,
      "learning_rate": 1.6082869832544022e-05,
      "loss": 11.8061,
      "step": 30207
    },
    {
      "epoch": 1.6449463287796178,
      "grad_norm": 0.49370947612749594,
      "learning_rate": 1.607807419748927e-05,
      "loss": 11.7223,
      "step": 30208
    },
    {
      "epoch": 1.6450007827762008,
      "grad_norm": 0.5000206853764118,
      "learning_rate": 1.6073279215025473e-05,
      "loss": 11.816,
      "step": 30209
    },
    {
      "epoch": 1.6450552367727838,
      "grad_norm": 0.5787497720942635,
      "learning_rate": 1.6068484885189915e-05,
      "loss": 11.9479,
      "step": 30210
    },
    {
      "epoch": 1.6451096907693668,
      "grad_norm": 0.5372897855231669,
      "learning_rate": 1.6063691208019917e-05,
      "loss": 11.8536,
      "step": 30211
    },
    {
      "epoch": 1.64516414476595,
      "grad_norm": 0.5002106263799222,
      "learning_rate": 1.6058898183552708e-05,
      "loss": 11.661,
      "step": 30212
    },
    {
      "epoch": 1.645218598762533,
      "grad_norm": 0.5660293224008779,
      "learning_rate": 1.6054105811825582e-05,
      "loss": 11.8339,
      "step": 30213
    },
    {
      "epoch": 1.645273052759116,
      "grad_norm": 0.5275471143776981,
      "learning_rate": 1.6049314092875832e-05,
      "loss": 11.8119,
      "step": 30214
    },
    {
      "epoch": 1.645327506755699,
      "grad_norm": 0.5427198343184654,
      "learning_rate": 1.6044523026740666e-05,
      "loss": 11.8751,
      "step": 30215
    },
    {
      "epoch": 1.645381960752282,
      "grad_norm": 0.5249795762334607,
      "learning_rate": 1.603973261345739e-05,
      "loss": 11.7967,
      "step": 30216
    },
    {
      "epoch": 1.645436414748865,
      "grad_norm": 0.5720662556319112,
      "learning_rate": 1.6034942853063218e-05,
      "loss": 11.844,
      "step": 30217
    },
    {
      "epoch": 1.6454908687454481,
      "grad_norm": 0.6051020981225518,
      "learning_rate": 1.603015374559543e-05,
      "loss": 11.9617,
      "step": 30218
    },
    {
      "epoch": 1.6455453227420311,
      "grad_norm": 0.5839193438142962,
      "learning_rate": 1.6025365291091253e-05,
      "loss": 11.8775,
      "step": 30219
    },
    {
      "epoch": 1.6455997767386141,
      "grad_norm": 0.5290094140171435,
      "learning_rate": 1.6020577489587897e-05,
      "loss": 11.8442,
      "step": 30220
    },
    {
      "epoch": 1.6456542307351971,
      "grad_norm": 0.5457381959843188,
      "learning_rate": 1.6015790341122637e-05,
      "loss": 11.8472,
      "step": 30221
    },
    {
      "epoch": 1.6457086847317801,
      "grad_norm": 0.5302315325245227,
      "learning_rate": 1.601100384573264e-05,
      "loss": 11.7743,
      "step": 30222
    },
    {
      "epoch": 1.645763138728363,
      "grad_norm": 0.5443525039553707,
      "learning_rate": 1.6006218003455175e-05,
      "loss": 11.8231,
      "step": 30223
    },
    {
      "epoch": 1.645817592724946,
      "grad_norm": 0.5151473320863345,
      "learning_rate": 1.6001432814327467e-05,
      "loss": 11.7664,
      "step": 30224
    },
    {
      "epoch": 1.645872046721529,
      "grad_norm": 0.5958881746221502,
      "learning_rate": 1.599664827838667e-05,
      "loss": 11.846,
      "step": 30225
    },
    {
      "epoch": 1.645926500718112,
      "grad_norm": 0.5797504547148997,
      "learning_rate": 1.5991864395670052e-05,
      "loss": 11.8341,
      "step": 30226
    },
    {
      "epoch": 1.645980954714695,
      "grad_norm": 0.5883214560695113,
      "learning_rate": 1.598708116621477e-05,
      "loss": 11.8504,
      "step": 30227
    },
    {
      "epoch": 1.646035408711278,
      "grad_norm": 0.6053640051559289,
      "learning_rate": 1.598229859005804e-05,
      "loss": 11.8396,
      "step": 30228
    },
    {
      "epoch": 1.646089862707861,
      "grad_norm": 0.5922548361657342,
      "learning_rate": 1.597751666723706e-05,
      "loss": 11.8569,
      "step": 30229
    },
    {
      "epoch": 1.646144316704444,
      "grad_norm": 0.5600807084813513,
      "learning_rate": 1.5972735397788972e-05,
      "loss": 11.7831,
      "step": 30230
    },
    {
      "epoch": 1.646198770701027,
      "grad_norm": 0.509136169809699,
      "learning_rate": 1.5967954781751004e-05,
      "loss": 11.7968,
      "step": 30231
    },
    {
      "epoch": 1.64625322469761,
      "grad_norm": 0.5757226218075859,
      "learning_rate": 1.5963174819160297e-05,
      "loss": 11.8568,
      "step": 30232
    },
    {
      "epoch": 1.646307678694193,
      "grad_norm": 0.5645557112183426,
      "learning_rate": 1.5958395510054058e-05,
      "loss": 11.786,
      "step": 30233
    },
    {
      "epoch": 1.646362132690776,
      "grad_norm": 0.5352800378522243,
      "learning_rate": 1.59536168544694e-05,
      "loss": 11.66,
      "step": 30234
    },
    {
      "epoch": 1.6464165866873592,
      "grad_norm": 0.520343803438192,
      "learning_rate": 1.5948838852443515e-05,
      "loss": 11.8258,
      "step": 30235
    },
    {
      "epoch": 1.6464710406839422,
      "grad_norm": 0.5734099053021384,
      "learning_rate": 1.5944061504013584e-05,
      "loss": 11.9015,
      "step": 30236
    },
    {
      "epoch": 1.6465254946805252,
      "grad_norm": 0.5804734427098937,
      "learning_rate": 1.593928480921669e-05,
      "loss": 11.8885,
      "step": 30237
    },
    {
      "epoch": 1.6465799486771082,
      "grad_norm": 0.5559074652720587,
      "learning_rate": 1.593450876809005e-05,
      "loss": 11.8093,
      "step": 30238
    },
    {
      "epoch": 1.6466344026736912,
      "grad_norm": 0.5548681081568707,
      "learning_rate": 1.5929733380670763e-05,
      "loss": 11.7741,
      "step": 30239
    },
    {
      "epoch": 1.6466888566702742,
      "grad_norm": 0.5482775809889404,
      "learning_rate": 1.5924958646995948e-05,
      "loss": 11.8452,
      "step": 30240
    },
    {
      "epoch": 1.6467433106668574,
      "grad_norm": 0.515091895675061,
      "learning_rate": 1.5920184567102768e-05,
      "loss": 11.8557,
      "step": 30241
    },
    {
      "epoch": 1.6467977646634404,
      "grad_norm": 0.6422393359974997,
      "learning_rate": 1.5915411141028326e-05,
      "loss": 11.6629,
      "step": 30242
    },
    {
      "epoch": 1.6468522186600234,
      "grad_norm": 0.5747118067216739,
      "learning_rate": 1.591063836880976e-05,
      "loss": 11.7587,
      "step": 30243
    },
    {
      "epoch": 1.6469066726566064,
      "grad_norm": 0.5208001241281782,
      "learning_rate": 1.5905866250484146e-05,
      "loss": 11.764,
      "step": 30244
    },
    {
      "epoch": 1.6469611266531894,
      "grad_norm": 0.5943357567202925,
      "learning_rate": 1.590109478608862e-05,
      "loss": 11.6988,
      "step": 30245
    },
    {
      "epoch": 1.6470155806497724,
      "grad_norm": 0.6174542726047334,
      "learning_rate": 1.589632397566031e-05,
      "loss": 11.8192,
      "step": 30246
    },
    {
      "epoch": 1.6470700346463554,
      "grad_norm": 0.5589766098234975,
      "learning_rate": 1.589155381923627e-05,
      "loss": 11.8471,
      "step": 30247
    },
    {
      "epoch": 1.6471244886429384,
      "grad_norm": 0.526663364675429,
      "learning_rate": 1.588678431685363e-05,
      "loss": 11.833,
      "step": 30248
    },
    {
      "epoch": 1.6471789426395214,
      "grad_norm": 0.5420646816181991,
      "learning_rate": 1.588201546854946e-05,
      "loss": 11.7753,
      "step": 30249
    },
    {
      "epoch": 1.6472333966361044,
      "grad_norm": 0.5585573894236098,
      "learning_rate": 1.587724727436082e-05,
      "loss": 11.8415,
      "step": 30250
    },
    {
      "epoch": 1.6472878506326873,
      "grad_norm": 0.5718008066308626,
      "learning_rate": 1.5872479734324843e-05,
      "loss": 11.7366,
      "step": 30251
    },
    {
      "epoch": 1.6473423046292703,
      "grad_norm": 0.4943534299757422,
      "learning_rate": 1.5867712848478545e-05,
      "loss": 11.7126,
      "step": 30252
    },
    {
      "epoch": 1.6473967586258533,
      "grad_norm": 0.5023601942903341,
      "learning_rate": 1.586294661685904e-05,
      "loss": 11.7995,
      "step": 30253
    },
    {
      "epoch": 1.6474512126224363,
      "grad_norm": 0.5113537105801732,
      "learning_rate": 1.585818103950335e-05,
      "loss": 11.8232,
      "step": 30254
    },
    {
      "epoch": 1.6475056666190193,
      "grad_norm": 0.5339525225899784,
      "learning_rate": 1.5853416116448582e-05,
      "loss": 11.7476,
      "step": 30255
    },
    {
      "epoch": 1.6475601206156023,
      "grad_norm": 0.5436259771386474,
      "learning_rate": 1.5848651847731745e-05,
      "loss": 11.7871,
      "step": 30256
    },
    {
      "epoch": 1.6476145746121853,
      "grad_norm": 0.6170914582958583,
      "learning_rate": 1.5843888233389893e-05,
      "loss": 11.8158,
      "step": 30257
    },
    {
      "epoch": 1.6476690286087683,
      "grad_norm": 0.6716163108483342,
      "learning_rate": 1.58391252734601e-05,
      "loss": 11.836,
      "step": 30258
    },
    {
      "epoch": 1.6477234826053515,
      "grad_norm": 0.5310794370224393,
      "learning_rate": 1.5834362967979387e-05,
      "loss": 11.829,
      "step": 30259
    },
    {
      "epoch": 1.6477779366019345,
      "grad_norm": 0.481759676350806,
      "learning_rate": 1.582960131698479e-05,
      "loss": 11.8247,
      "step": 30260
    },
    {
      "epoch": 1.6478323905985175,
      "grad_norm": 0.5315280858473589,
      "learning_rate": 1.582484032051329e-05,
      "loss": 11.8731,
      "step": 30261
    },
    {
      "epoch": 1.6478868445951005,
      "grad_norm": 0.5378813463110909,
      "learning_rate": 1.5820079978601955e-05,
      "loss": 11.8076,
      "step": 30262
    },
    {
      "epoch": 1.6479412985916835,
      "grad_norm": 0.55903266225185,
      "learning_rate": 1.5815320291287816e-05,
      "loss": 11.7443,
      "step": 30263
    },
    {
      "epoch": 1.6479957525882665,
      "grad_norm": 0.6012034498040094,
      "learning_rate": 1.5810561258607847e-05,
      "loss": 11.7531,
      "step": 30264
    },
    {
      "epoch": 1.6480502065848497,
      "grad_norm": 0.5144155695918641,
      "learning_rate": 1.5805802880599097e-05,
      "loss": 11.9225,
      "step": 30265
    },
    {
      "epoch": 1.6481046605814327,
      "grad_norm": 0.5474350811292334,
      "learning_rate": 1.580104515729851e-05,
      "loss": 11.8487,
      "step": 30266
    },
    {
      "epoch": 1.6481591145780157,
      "grad_norm": 0.6174978287790941,
      "learning_rate": 1.579628808874315e-05,
      "loss": 11.8679,
      "step": 30267
    },
    {
      "epoch": 1.6482135685745987,
      "grad_norm": 0.5420979287960093,
      "learning_rate": 1.5791531674969938e-05,
      "loss": 11.8612,
      "step": 30268
    },
    {
      "epoch": 1.6482680225711817,
      "grad_norm": 0.5702792554346019,
      "learning_rate": 1.5786775916015938e-05,
      "loss": 11.8418,
      "step": 30269
    },
    {
      "epoch": 1.6483224765677646,
      "grad_norm": 0.5200834366442612,
      "learning_rate": 1.5782020811918075e-05,
      "loss": 11.8059,
      "step": 30270
    },
    {
      "epoch": 1.6483769305643476,
      "grad_norm": 0.5350795284900471,
      "learning_rate": 1.5777266362713327e-05,
      "loss": 11.8158,
      "step": 30271
    },
    {
      "epoch": 1.6484313845609306,
      "grad_norm": 0.5144431219346606,
      "learning_rate": 1.577251256843868e-05,
      "loss": 11.8311,
      "step": 30272
    },
    {
      "epoch": 1.6484858385575136,
      "grad_norm": 0.5051398509971821,
      "learning_rate": 1.5767759429131123e-05,
      "loss": 11.805,
      "step": 30273
    },
    {
      "epoch": 1.6485402925540966,
      "grad_norm": 0.5139305998432078,
      "learning_rate": 1.5763006944827564e-05,
      "loss": 11.8075,
      "step": 30274
    },
    {
      "epoch": 1.6485947465506796,
      "grad_norm": 0.5260812256509733,
      "learning_rate": 1.5758255115565023e-05,
      "loss": 11.7722,
      "step": 30275
    },
    {
      "epoch": 1.6486492005472626,
      "grad_norm": 0.5193301853136946,
      "learning_rate": 1.575350394138039e-05,
      "loss": 11.7299,
      "step": 30276
    },
    {
      "epoch": 1.6487036545438456,
      "grad_norm": 0.551471942030761,
      "learning_rate": 1.5748753422310657e-05,
      "loss": 11.7086,
      "step": 30277
    },
    {
      "epoch": 1.6487581085404286,
      "grad_norm": 0.5041331920105072,
      "learning_rate": 1.5744003558392727e-05,
      "loss": 11.7779,
      "step": 30278
    },
    {
      "epoch": 1.6488125625370116,
      "grad_norm": 0.5717935631152075,
      "learning_rate": 1.573925434966358e-05,
      "loss": 11.8936,
      "step": 30279
    },
    {
      "epoch": 1.6488670165335946,
      "grad_norm": 0.5568158823108074,
      "learning_rate": 1.5734505796160125e-05,
      "loss": 11.9016,
      "step": 30280
    },
    {
      "epoch": 1.6489214705301776,
      "grad_norm": 0.5757758848790586,
      "learning_rate": 1.572975789791925e-05,
      "loss": 11.8203,
      "step": 30281
    },
    {
      "epoch": 1.6489759245267608,
      "grad_norm": 0.5473651209505923,
      "learning_rate": 1.5725010654977944e-05,
      "loss": 11.7335,
      "step": 30282
    },
    {
      "epoch": 1.6490303785233438,
      "grad_norm": 0.5518298560932837,
      "learning_rate": 1.572026406737306e-05,
      "loss": 11.7585,
      "step": 30283
    },
    {
      "epoch": 1.6490848325199268,
      "grad_norm": 0.5395694848191006,
      "learning_rate": 1.5715518135141537e-05,
      "loss": 11.8149,
      "step": 30284
    },
    {
      "epoch": 1.6491392865165098,
      "grad_norm": 0.5357026261927198,
      "learning_rate": 1.57107728583203e-05,
      "loss": 11.8469,
      "step": 30285
    },
    {
      "epoch": 1.6491937405130928,
      "grad_norm": 0.5788451144378695,
      "learning_rate": 1.5706028236946213e-05,
      "loss": 11.9288,
      "step": 30286
    },
    {
      "epoch": 1.6492481945096757,
      "grad_norm": 0.49511969403709766,
      "learning_rate": 1.5701284271056206e-05,
      "loss": 11.7889,
      "step": 30287
    },
    {
      "epoch": 1.649302648506259,
      "grad_norm": 0.5333781080811473,
      "learning_rate": 1.5696540960687124e-05,
      "loss": 11.8837,
      "step": 30288
    },
    {
      "epoch": 1.649357102502842,
      "grad_norm": 0.485940058173729,
      "learning_rate": 1.5691798305875892e-05,
      "loss": 11.7135,
      "step": 30289
    },
    {
      "epoch": 1.649411556499425,
      "grad_norm": 0.5588762528727661,
      "learning_rate": 1.5687056306659385e-05,
      "loss": 11.7117,
      "step": 30290
    },
    {
      "epoch": 1.649466010496008,
      "grad_norm": 0.4969574205986947,
      "learning_rate": 1.568231496307445e-05,
      "loss": 11.8301,
      "step": 30291
    },
    {
      "epoch": 1.649520464492591,
      "grad_norm": 0.5388298627014794,
      "learning_rate": 1.567757427515799e-05,
      "loss": 11.7702,
      "step": 30292
    },
    {
      "epoch": 1.649574918489174,
      "grad_norm": 0.559164915027884,
      "learning_rate": 1.5672834242946833e-05,
      "loss": 11.8038,
      "step": 30293
    },
    {
      "epoch": 1.649629372485757,
      "grad_norm": 0.5254196074212372,
      "learning_rate": 1.5668094866477846e-05,
      "loss": 11.7543,
      "step": 30294
    },
    {
      "epoch": 1.64968382648234,
      "grad_norm": 0.5450844572147387,
      "learning_rate": 1.5663356145787932e-05,
      "loss": 11.8074,
      "step": 30295
    },
    {
      "epoch": 1.649738280478923,
      "grad_norm": 0.5800732777198301,
      "learning_rate": 1.5658618080913878e-05,
      "loss": 11.9646,
      "step": 30296
    },
    {
      "epoch": 1.649792734475506,
      "grad_norm": 0.5408316042486881,
      "learning_rate": 1.5653880671892573e-05,
      "loss": 11.7845,
      "step": 30297
    },
    {
      "epoch": 1.6498471884720889,
      "grad_norm": 0.5049158407211579,
      "learning_rate": 1.5649143918760822e-05,
      "loss": 11.7873,
      "step": 30298
    },
    {
      "epoch": 1.6499016424686719,
      "grad_norm": 0.5507430811172999,
      "learning_rate": 1.5644407821555495e-05,
      "loss": 11.9438,
      "step": 30299
    },
    {
      "epoch": 1.6499560964652549,
      "grad_norm": 0.537260845691589,
      "learning_rate": 1.5639672380313398e-05,
      "loss": 11.714,
      "step": 30300
    },
    {
      "epoch": 1.6500105504618379,
      "grad_norm": 0.5595521725432183,
      "learning_rate": 1.563493759507133e-05,
      "loss": 11.8271,
      "step": 30301
    },
    {
      "epoch": 1.6500650044584209,
      "grad_norm": 0.5486761625586757,
      "learning_rate": 1.5630203465866156e-05,
      "loss": 11.8217,
      "step": 30302
    },
    {
      "epoch": 1.6501194584550039,
      "grad_norm": 0.5580839278671575,
      "learning_rate": 1.5625469992734642e-05,
      "loss": 11.7532,
      "step": 30303
    },
    {
      "epoch": 1.6501739124515868,
      "grad_norm": 0.5809952505539281,
      "learning_rate": 1.5620737175713663e-05,
      "loss": 11.8901,
      "step": 30304
    },
    {
      "epoch": 1.65022836644817,
      "grad_norm": 0.543482058822921,
      "learning_rate": 1.561600501483994e-05,
      "loss": 11.7574,
      "step": 30305
    },
    {
      "epoch": 1.650282820444753,
      "grad_norm": 0.5544393298231484,
      "learning_rate": 1.5611273510150326e-05,
      "loss": 11.7676,
      "step": 30306
    },
    {
      "epoch": 1.650337274441336,
      "grad_norm": 0.5144533714967504,
      "learning_rate": 1.5606542661681622e-05,
      "loss": 11.79,
      "step": 30307
    },
    {
      "epoch": 1.650391728437919,
      "grad_norm": 0.5170192773414732,
      "learning_rate": 1.5601812469470565e-05,
      "loss": 11.7147,
      "step": 30308
    },
    {
      "epoch": 1.650446182434502,
      "grad_norm": 0.545316082602362,
      "learning_rate": 1.559708293355402e-05,
      "loss": 11.8639,
      "step": 30309
    },
    {
      "epoch": 1.650500636431085,
      "grad_norm": 0.5979883381590966,
      "learning_rate": 1.559235405396867e-05,
      "loss": 11.9237,
      "step": 30310
    },
    {
      "epoch": 1.6505550904276682,
      "grad_norm": 0.5953268062769637,
      "learning_rate": 1.558762583075133e-05,
      "loss": 11.7034,
      "step": 30311
    },
    {
      "epoch": 1.6506095444242512,
      "grad_norm": 0.6066605854287557,
      "learning_rate": 1.5582898263938784e-05,
      "loss": 11.858,
      "step": 30312
    },
    {
      "epoch": 1.6506639984208342,
      "grad_norm": 0.5447694093535638,
      "learning_rate": 1.557817135356775e-05,
      "loss": 11.8601,
      "step": 30313
    },
    {
      "epoch": 1.6507184524174172,
      "grad_norm": 0.6123454736841145,
      "learning_rate": 1.5573445099675045e-05,
      "loss": 11.8336,
      "step": 30314
    },
    {
      "epoch": 1.6507729064140002,
      "grad_norm": 0.5389667746472384,
      "learning_rate": 1.556871950229737e-05,
      "loss": 11.7927,
      "step": 30315
    },
    {
      "epoch": 1.6508273604105832,
      "grad_norm": 0.5308657024967885,
      "learning_rate": 1.5563994561471506e-05,
      "loss": 11.675,
      "step": 30316
    },
    {
      "epoch": 1.6508818144071662,
      "grad_norm": 0.525862668934587,
      "learning_rate": 1.555927027723416e-05,
      "loss": 11.7492,
      "step": 30317
    },
    {
      "epoch": 1.6509362684037492,
      "grad_norm": 0.5386362426469454,
      "learning_rate": 1.5554546649622094e-05,
      "loss": 11.6782,
      "step": 30318
    },
    {
      "epoch": 1.6509907224003322,
      "grad_norm": 0.5667891737255518,
      "learning_rate": 1.5549823678672072e-05,
      "loss": 11.9245,
      "step": 30319
    },
    {
      "epoch": 1.6510451763969152,
      "grad_norm": 0.5157830108174235,
      "learning_rate": 1.5545101364420744e-05,
      "loss": 11.5794,
      "step": 30320
    },
    {
      "epoch": 1.6510996303934982,
      "grad_norm": 0.6135712245199829,
      "learning_rate": 1.5540379706904862e-05,
      "loss": 11.9485,
      "step": 30321
    },
    {
      "epoch": 1.6511540843900812,
      "grad_norm": 0.5558481003182353,
      "learning_rate": 1.5535658706161183e-05,
      "loss": 11.8048,
      "step": 30322
    },
    {
      "epoch": 1.6512085383866641,
      "grad_norm": 0.5875586621045195,
      "learning_rate": 1.5530938362226353e-05,
      "loss": 11.8165,
      "step": 30323
    },
    {
      "epoch": 1.6512629923832471,
      "grad_norm": 0.537000074397403,
      "learning_rate": 1.552621867513715e-05,
      "loss": 11.8471,
      "step": 30324
    },
    {
      "epoch": 1.6513174463798301,
      "grad_norm": 0.557250138866359,
      "learning_rate": 1.5521499644930203e-05,
      "loss": 11.7105,
      "step": 30325
    },
    {
      "epoch": 1.6513719003764131,
      "grad_norm": 0.5779101757630384,
      "learning_rate": 1.5516781271642265e-05,
      "loss": 11.6837,
      "step": 30326
    },
    {
      "epoch": 1.6514263543729961,
      "grad_norm": 0.5137701020962699,
      "learning_rate": 1.5512063555309985e-05,
      "loss": 11.7143,
      "step": 30327
    },
    {
      "epoch": 1.6514808083695791,
      "grad_norm": 0.5416251137432319,
      "learning_rate": 1.5507346495970075e-05,
      "loss": 11.7964,
      "step": 30328
    },
    {
      "epoch": 1.6515352623661623,
      "grad_norm": 0.5257476400477481,
      "learning_rate": 1.5502630093659243e-05,
      "loss": 11.7226,
      "step": 30329
    },
    {
      "epoch": 1.6515897163627453,
      "grad_norm": 0.5530633292834322,
      "learning_rate": 1.549791434841409e-05,
      "loss": 11.7645,
      "step": 30330
    },
    {
      "epoch": 1.6516441703593283,
      "grad_norm": 0.4887682575319276,
      "learning_rate": 1.549319926027134e-05,
      "loss": 11.7486,
      "step": 30331
    },
    {
      "epoch": 1.6516986243559113,
      "grad_norm": 0.5035435737874265,
      "learning_rate": 1.5488484829267626e-05,
      "loss": 11.6502,
      "step": 30332
    },
    {
      "epoch": 1.6517530783524943,
      "grad_norm": 0.5236023357418601,
      "learning_rate": 1.5483771055439634e-05,
      "loss": 11.8346,
      "step": 30333
    },
    {
      "epoch": 1.6518075323490773,
      "grad_norm": 0.5720862268818151,
      "learning_rate": 1.5479057938824025e-05,
      "loss": 11.7759,
      "step": 30334
    },
    {
      "epoch": 1.6518619863456605,
      "grad_norm": 0.5189373404562976,
      "learning_rate": 1.5474345479457408e-05,
      "loss": 11.794,
      "step": 30335
    },
    {
      "epoch": 1.6519164403422435,
      "grad_norm": 0.596743891031663,
      "learning_rate": 1.5469633677376495e-05,
      "loss": 11.8006,
      "step": 30336
    },
    {
      "epoch": 1.6519708943388265,
      "grad_norm": 0.6333154097587338,
      "learning_rate": 1.5464922532617853e-05,
      "loss": 11.9669,
      "step": 30337
    },
    {
      "epoch": 1.6520253483354095,
      "grad_norm": 0.5352705912976505,
      "learning_rate": 1.546021204521817e-05,
      "loss": 11.8758,
      "step": 30338
    },
    {
      "epoch": 1.6520798023319925,
      "grad_norm": 0.5458485553451803,
      "learning_rate": 1.5455502215214057e-05,
      "loss": 11.8238,
      "step": 30339
    },
    {
      "epoch": 1.6521342563285755,
      "grad_norm": 0.5712397342395055,
      "learning_rate": 1.545079304264212e-05,
      "loss": 11.7973,
      "step": 30340
    },
    {
      "epoch": 1.6521887103251585,
      "grad_norm": 0.666926277135753,
      "learning_rate": 1.544608452753902e-05,
      "loss": 11.8262,
      "step": 30341
    },
    {
      "epoch": 1.6522431643217415,
      "grad_norm": 0.543368776603423,
      "learning_rate": 1.544137666994131e-05,
      "loss": 11.8499,
      "step": 30342
    },
    {
      "epoch": 1.6522976183183244,
      "grad_norm": 0.5593660809025408,
      "learning_rate": 1.5436669469885645e-05,
      "loss": 11.8503,
      "step": 30343
    },
    {
      "epoch": 1.6523520723149074,
      "grad_norm": 0.5784121439909142,
      "learning_rate": 1.543196292740864e-05,
      "loss": 11.7,
      "step": 30344
    },
    {
      "epoch": 1.6524065263114904,
      "grad_norm": 0.5811856404692756,
      "learning_rate": 1.542725704254685e-05,
      "loss": 11.8399,
      "step": 30345
    },
    {
      "epoch": 1.6524609803080734,
      "grad_norm": 0.5337992690973378,
      "learning_rate": 1.5422551815336916e-05,
      "loss": 11.7704,
      "step": 30346
    },
    {
      "epoch": 1.6525154343046564,
      "grad_norm": 0.5361280886355042,
      "learning_rate": 1.5417847245815387e-05,
      "loss": 11.6659,
      "step": 30347
    },
    {
      "epoch": 1.6525698883012394,
      "grad_norm": 0.5805437058398001,
      "learning_rate": 1.5413143334018875e-05,
      "loss": 11.8149,
      "step": 30348
    },
    {
      "epoch": 1.6526243422978224,
      "grad_norm": 0.5204293584801625,
      "learning_rate": 1.5408440079983945e-05,
      "loss": 11.7907,
      "step": 30349
    },
    {
      "epoch": 1.6526787962944054,
      "grad_norm": 0.5598694718774125,
      "learning_rate": 1.540373748374715e-05,
      "loss": 11.7298,
      "step": 30350
    },
    {
      "epoch": 1.6527332502909884,
      "grad_norm": 0.5730257959846164,
      "learning_rate": 1.539903554534511e-05,
      "loss": 11.7879,
      "step": 30351
    },
    {
      "epoch": 1.6527877042875716,
      "grad_norm": 0.5471553983808888,
      "learning_rate": 1.5394334264814326e-05,
      "loss": 11.796,
      "step": 30352
    },
    {
      "epoch": 1.6528421582841546,
      "grad_norm": 0.6051570299155735,
      "learning_rate": 1.5389633642191415e-05,
      "loss": 11.8685,
      "step": 30353
    },
    {
      "epoch": 1.6528966122807376,
      "grad_norm": 0.5979375283709545,
      "learning_rate": 1.5384933677512868e-05,
      "loss": 11.8454,
      "step": 30354
    },
    {
      "epoch": 1.6529510662773206,
      "grad_norm": 0.5348757715511375,
      "learning_rate": 1.538023437081528e-05,
      "loss": 11.883,
      "step": 30355
    },
    {
      "epoch": 1.6530055202739036,
      "grad_norm": 0.705201060188786,
      "learning_rate": 1.5375535722135203e-05,
      "loss": 11.8147,
      "step": 30356
    },
    {
      "epoch": 1.6530599742704866,
      "grad_norm": 0.581476274725346,
      "learning_rate": 1.5370837731509125e-05,
      "loss": 11.8792,
      "step": 30357
    },
    {
      "epoch": 1.6531144282670698,
      "grad_norm": 0.5791998452877999,
      "learning_rate": 1.5366140398973627e-05,
      "loss": 11.8992,
      "step": 30358
    },
    {
      "epoch": 1.6531688822636528,
      "grad_norm": 0.56410517208088,
      "learning_rate": 1.5361443724565183e-05,
      "loss": 11.8791,
      "step": 30359
    },
    {
      "epoch": 1.6532233362602358,
      "grad_norm": 0.6183107415967306,
      "learning_rate": 1.5356747708320386e-05,
      "loss": 11.7988,
      "step": 30360
    },
    {
      "epoch": 1.6532777902568188,
      "grad_norm": 0.6875640496560392,
      "learning_rate": 1.5352052350275703e-05,
      "loss": 11.8829,
      "step": 30361
    },
    {
      "epoch": 1.6533322442534018,
      "grad_norm": 0.5497935646659076,
      "learning_rate": 1.534735765046763e-05,
      "loss": 11.791,
      "step": 30362
    },
    {
      "epoch": 1.6533866982499847,
      "grad_norm": 0.5619957708161861,
      "learning_rate": 1.5342663608932738e-05,
      "loss": 11.8213,
      "step": 30363
    },
    {
      "epoch": 1.6534411522465677,
      "grad_norm": 0.5555121280641989,
      "learning_rate": 1.5337970225707456e-05,
      "loss": 11.882,
      "step": 30364
    },
    {
      "epoch": 1.6534956062431507,
      "grad_norm": 0.48188241317873043,
      "learning_rate": 1.5333277500828337e-05,
      "loss": 11.8077,
      "step": 30365
    },
    {
      "epoch": 1.6535500602397337,
      "grad_norm": 0.6001084996859429,
      "learning_rate": 1.5328585434331834e-05,
      "loss": 11.8115,
      "step": 30366
    },
    {
      "epoch": 1.6536045142363167,
      "grad_norm": 0.5428956276223575,
      "learning_rate": 1.5323894026254448e-05,
      "loss": 11.8563,
      "step": 30367
    },
    {
      "epoch": 1.6536589682328997,
      "grad_norm": 0.5685570349966749,
      "learning_rate": 1.531920327663269e-05,
      "loss": 11.7507,
      "step": 30368
    },
    {
      "epoch": 1.6537134222294827,
      "grad_norm": 0.5485690021047516,
      "learning_rate": 1.5314513185502976e-05,
      "loss": 11.8139,
      "step": 30369
    },
    {
      "epoch": 1.6537678762260657,
      "grad_norm": 0.5509489858868376,
      "learning_rate": 1.530982375290184e-05,
      "loss": 11.8292,
      "step": 30370
    },
    {
      "epoch": 1.6538223302226487,
      "grad_norm": 0.5656163722575241,
      "learning_rate": 1.530513497886571e-05,
      "loss": 11.7971,
      "step": 30371
    },
    {
      "epoch": 1.6538767842192317,
      "grad_norm": 0.5823733713054023,
      "learning_rate": 1.530044686343104e-05,
      "loss": 11.8057,
      "step": 30372
    },
    {
      "epoch": 1.6539312382158147,
      "grad_norm": 0.5984367437471964,
      "learning_rate": 1.5295759406634312e-05,
      "loss": 11.8007,
      "step": 30373
    },
    {
      "epoch": 1.6539856922123977,
      "grad_norm": 0.5833388668835244,
      "learning_rate": 1.5291072608511946e-05,
      "loss": 11.7687,
      "step": 30374
    },
    {
      "epoch": 1.6540401462089809,
      "grad_norm": 0.614076766546919,
      "learning_rate": 1.5286386469100434e-05,
      "loss": 11.9744,
      "step": 30375
    },
    {
      "epoch": 1.6540946002055639,
      "grad_norm": 0.5085048132639166,
      "learning_rate": 1.5281700988436153e-05,
      "loss": 11.6403,
      "step": 30376
    },
    {
      "epoch": 1.6541490542021469,
      "grad_norm": 0.5215632385558154,
      "learning_rate": 1.527701616655558e-05,
      "loss": 11.7214,
      "step": 30377
    },
    {
      "epoch": 1.6542035081987299,
      "grad_norm": 0.532165653013137,
      "learning_rate": 1.527233200349516e-05,
      "loss": 11.7481,
      "step": 30378
    },
    {
      "epoch": 1.6542579621953128,
      "grad_norm": 0.6044869715802822,
      "learning_rate": 1.5267648499291266e-05,
      "loss": 11.8685,
      "step": 30379
    },
    {
      "epoch": 1.6543124161918958,
      "grad_norm": 0.5182815276832565,
      "learning_rate": 1.52629656539804e-05,
      "loss": 11.8329,
      "step": 30380
    },
    {
      "epoch": 1.654366870188479,
      "grad_norm": 0.5775482579320368,
      "learning_rate": 1.5258283467598878e-05,
      "loss": 11.8831,
      "step": 30381
    },
    {
      "epoch": 1.654421324185062,
      "grad_norm": 0.5034491015701649,
      "learning_rate": 1.525360194018315e-05,
      "loss": 11.7911,
      "step": 30382
    },
    {
      "epoch": 1.654475778181645,
      "grad_norm": 0.52738638170991,
      "learning_rate": 1.524892107176964e-05,
      "loss": 11.6845,
      "step": 30383
    },
    {
      "epoch": 1.654530232178228,
      "grad_norm": 0.587513577590588,
      "learning_rate": 1.5244240862394721e-05,
      "loss": 11.806,
      "step": 30384
    },
    {
      "epoch": 1.654584686174811,
      "grad_norm": 0.6151047411750562,
      "learning_rate": 1.523956131209482e-05,
      "loss": 11.7422,
      "step": 30385
    },
    {
      "epoch": 1.654639140171394,
      "grad_norm": 0.4971705865165115,
      "learning_rate": 1.5234882420906282e-05,
      "loss": 11.7241,
      "step": 30386
    },
    {
      "epoch": 1.654693594167977,
      "grad_norm": 0.6124434570136407,
      "learning_rate": 1.5230204188865538e-05,
      "loss": 11.8893,
      "step": 30387
    },
    {
      "epoch": 1.65474804816456,
      "grad_norm": 0.5647274186823761,
      "learning_rate": 1.5225526616008912e-05,
      "loss": 11.7533,
      "step": 30388
    },
    {
      "epoch": 1.654802502161143,
      "grad_norm": 0.5205825535187503,
      "learning_rate": 1.5220849702372819e-05,
      "loss": 11.8398,
      "step": 30389
    },
    {
      "epoch": 1.654856956157726,
      "grad_norm": 0.5137468990488969,
      "learning_rate": 1.5216173447993654e-05,
      "loss": 11.7658,
      "step": 30390
    },
    {
      "epoch": 1.654911410154309,
      "grad_norm": 0.5846701781531567,
      "learning_rate": 1.5211497852907697e-05,
      "loss": 11.712,
      "step": 30391
    },
    {
      "epoch": 1.654965864150892,
      "grad_norm": 0.5596722748678072,
      "learning_rate": 1.5206822917151353e-05,
      "loss": 11.8297,
      "step": 30392
    },
    {
      "epoch": 1.655020318147475,
      "grad_norm": 0.6166217791286542,
      "learning_rate": 1.5202148640760993e-05,
      "loss": 11.888,
      "step": 30393
    },
    {
      "epoch": 1.655074772144058,
      "grad_norm": 0.5143980147394284,
      "learning_rate": 1.5197475023772933e-05,
      "loss": 11.7012,
      "step": 30394
    },
    {
      "epoch": 1.655129226140641,
      "grad_norm": 0.6011240633541712,
      "learning_rate": 1.5192802066223543e-05,
      "loss": 11.9945,
      "step": 30395
    },
    {
      "epoch": 1.655183680137224,
      "grad_norm": 0.5237221533972257,
      "learning_rate": 1.5188129768149129e-05,
      "loss": 11.7944,
      "step": 30396
    },
    {
      "epoch": 1.655238134133807,
      "grad_norm": 0.5219806364379025,
      "learning_rate": 1.518345812958606e-05,
      "loss": 11.7039,
      "step": 30397
    },
    {
      "epoch": 1.65529258813039,
      "grad_norm": 0.5315686986719056,
      "learning_rate": 1.5178787150570617e-05,
      "loss": 11.8195,
      "step": 30398
    },
    {
      "epoch": 1.6553470421269731,
      "grad_norm": 0.5243077468399364,
      "learning_rate": 1.5174116831139185e-05,
      "loss": 11.7269,
      "step": 30399
    },
    {
      "epoch": 1.6554014961235561,
      "grad_norm": 0.4962328515717563,
      "learning_rate": 1.5169447171328032e-05,
      "loss": 11.7835,
      "step": 30400
    },
    {
      "epoch": 1.6554559501201391,
      "grad_norm": 0.5297610904104818,
      "learning_rate": 1.5164778171173465e-05,
      "loss": 12.0246,
      "step": 30401
    },
    {
      "epoch": 1.6555104041167221,
      "grad_norm": 0.5402781603697728,
      "learning_rate": 1.5160109830711832e-05,
      "loss": 11.855,
      "step": 30402
    },
    {
      "epoch": 1.6555648581133051,
      "grad_norm": 0.7292855827356006,
      "learning_rate": 1.5155442149979393e-05,
      "loss": 11.8603,
      "step": 30403
    },
    {
      "epoch": 1.6556193121098883,
      "grad_norm": 0.574548793554654,
      "learning_rate": 1.5150775129012473e-05,
      "loss": 11.7707,
      "step": 30404
    },
    {
      "epoch": 1.6556737661064713,
      "grad_norm": 0.5715973146866069,
      "learning_rate": 1.5146108767847367e-05,
      "loss": 11.8918,
      "step": 30405
    },
    {
      "epoch": 1.6557282201030543,
      "grad_norm": 0.5483522380286712,
      "learning_rate": 1.5141443066520323e-05,
      "loss": 11.7617,
      "step": 30406
    },
    {
      "epoch": 1.6557826740996373,
      "grad_norm": 0.49708881440381997,
      "learning_rate": 1.5136778025067678e-05,
      "loss": 11.7585,
      "step": 30407
    },
    {
      "epoch": 1.6558371280962203,
      "grad_norm": 0.5767574518299988,
      "learning_rate": 1.5132113643525658e-05,
      "loss": 11.7842,
      "step": 30408
    },
    {
      "epoch": 1.6558915820928033,
      "grad_norm": 0.553466801046347,
      "learning_rate": 1.512744992193057e-05,
      "loss": 11.7271,
      "step": 30409
    },
    {
      "epoch": 1.6559460360893863,
      "grad_norm": 0.5508168131996588,
      "learning_rate": 1.5122786860318671e-05,
      "loss": 11.7983,
      "step": 30410
    },
    {
      "epoch": 1.6560004900859693,
      "grad_norm": 0.5412522124930113,
      "learning_rate": 1.511812445872619e-05,
      "loss": 11.8493,
      "step": 30411
    },
    {
      "epoch": 1.6560549440825523,
      "grad_norm": 0.762679716974236,
      "learning_rate": 1.511346271718943e-05,
      "loss": 11.753,
      "step": 30412
    },
    {
      "epoch": 1.6561093980791353,
      "grad_norm": 0.5304688621580921,
      "learning_rate": 1.5108801635744596e-05,
      "loss": 11.9053,
      "step": 30413
    },
    {
      "epoch": 1.6561638520757183,
      "grad_norm": 0.5631865236162934,
      "learning_rate": 1.510414121442798e-05,
      "loss": 11.7591,
      "step": 30414
    },
    {
      "epoch": 1.6562183060723012,
      "grad_norm": 0.5897639777853539,
      "learning_rate": 1.5099481453275777e-05,
      "loss": 11.7983,
      "step": 30415
    },
    {
      "epoch": 1.6562727600688842,
      "grad_norm": 0.5405331228809653,
      "learning_rate": 1.5094822352324246e-05,
      "loss": 11.7762,
      "step": 30416
    },
    {
      "epoch": 1.6563272140654672,
      "grad_norm": 0.5257861697223148,
      "learning_rate": 1.5090163911609633e-05,
      "loss": 11.7752,
      "step": 30417
    },
    {
      "epoch": 1.6563816680620502,
      "grad_norm": 0.5135910112359288,
      "learning_rate": 1.5085506131168125e-05,
      "loss": 11.6758,
      "step": 30418
    },
    {
      "epoch": 1.6564361220586332,
      "grad_norm": 0.5315099764161411,
      "learning_rate": 1.508084901103598e-05,
      "loss": 11.7313,
      "step": 30419
    },
    {
      "epoch": 1.6564905760552162,
      "grad_norm": 0.5490301029222859,
      "learning_rate": 1.5076192551249402e-05,
      "loss": 11.8143,
      "step": 30420
    },
    {
      "epoch": 1.6565450300517992,
      "grad_norm": 0.5209767159520465,
      "learning_rate": 1.5071536751844562e-05,
      "loss": 11.7725,
      "step": 30421
    },
    {
      "epoch": 1.6565994840483824,
      "grad_norm": 0.5816017991475442,
      "learning_rate": 1.506688161285772e-05,
      "loss": 11.8754,
      "step": 30422
    },
    {
      "epoch": 1.6566539380449654,
      "grad_norm": 0.5109652757838231,
      "learning_rate": 1.5062227134325024e-05,
      "loss": 11.5619,
      "step": 30423
    },
    {
      "epoch": 1.6567083920415484,
      "grad_norm": 0.5540279756642714,
      "learning_rate": 1.5057573316282725e-05,
      "loss": 11.7301,
      "step": 30424
    },
    {
      "epoch": 1.6567628460381314,
      "grad_norm": 0.524685815297852,
      "learning_rate": 1.5052920158766948e-05,
      "loss": 11.8003,
      "step": 30425
    },
    {
      "epoch": 1.6568173000347144,
      "grad_norm": 0.5097022926707467,
      "learning_rate": 1.5048267661813919e-05,
      "loss": 11.7585,
      "step": 30426
    },
    {
      "epoch": 1.6568717540312974,
      "grad_norm": 0.545251790768531,
      "learning_rate": 1.5043615825459823e-05,
      "loss": 11.8773,
      "step": 30427
    },
    {
      "epoch": 1.6569262080278806,
      "grad_norm": 0.5084444355519825,
      "learning_rate": 1.5038964649740806e-05,
      "loss": 11.8479,
      "step": 30428
    },
    {
      "epoch": 1.6569806620244636,
      "grad_norm": 0.5251794274872539,
      "learning_rate": 1.5034314134693061e-05,
      "loss": 11.6698,
      "step": 30429
    },
    {
      "epoch": 1.6570351160210466,
      "grad_norm": 0.6197933700010179,
      "learning_rate": 1.502966428035274e-05,
      "loss": 11.9312,
      "step": 30430
    },
    {
      "epoch": 1.6570895700176296,
      "grad_norm": 0.5446950480380597,
      "learning_rate": 1.5025015086755968e-05,
      "loss": 11.8244,
      "step": 30431
    },
    {
      "epoch": 1.6571440240142126,
      "grad_norm": 0.49786773049131056,
      "learning_rate": 1.5020366553938958e-05,
      "loss": 11.7073,
      "step": 30432
    },
    {
      "epoch": 1.6571984780107956,
      "grad_norm": 0.5520210459942693,
      "learning_rate": 1.5015718681937818e-05,
      "loss": 11.867,
      "step": 30433
    },
    {
      "epoch": 1.6572529320073786,
      "grad_norm": 0.5915950089523143,
      "learning_rate": 1.5011071470788706e-05,
      "loss": 11.8472,
      "step": 30434
    },
    {
      "epoch": 1.6573073860039615,
      "grad_norm": 0.5431760172320985,
      "learning_rate": 1.500642492052774e-05,
      "loss": 11.9128,
      "step": 30435
    },
    {
      "epoch": 1.6573618400005445,
      "grad_norm": 0.5358030956140224,
      "learning_rate": 1.5001779031191087e-05,
      "loss": 11.7488,
      "step": 30436
    },
    {
      "epoch": 1.6574162939971275,
      "grad_norm": 0.6157510286347626,
      "learning_rate": 1.4997133802814845e-05,
      "loss": 11.8955,
      "step": 30437
    },
    {
      "epoch": 1.6574707479937105,
      "grad_norm": 0.5774760449390086,
      "learning_rate": 1.4992489235435126e-05,
      "loss": 11.8392,
      "step": 30438
    },
    {
      "epoch": 1.6575252019902935,
      "grad_norm": 0.5401170132973687,
      "learning_rate": 1.4987845329088101e-05,
      "loss": 11.7123,
      "step": 30439
    },
    {
      "epoch": 1.6575796559868765,
      "grad_norm": 0.5495759029012905,
      "learning_rate": 1.4983202083809844e-05,
      "loss": 11.9677,
      "step": 30440
    },
    {
      "epoch": 1.6576341099834595,
      "grad_norm": 0.5041097356224132,
      "learning_rate": 1.4978559499636447e-05,
      "loss": 11.7353,
      "step": 30441
    },
    {
      "epoch": 1.6576885639800425,
      "grad_norm": 0.5175543807236078,
      "learning_rate": 1.4973917576604046e-05,
      "loss": 11.7952,
      "step": 30442
    },
    {
      "epoch": 1.6577430179766255,
      "grad_norm": 0.5684874834799495,
      "learning_rate": 1.4969276314748692e-05,
      "loss": 11.7802,
      "step": 30443
    },
    {
      "epoch": 1.6577974719732085,
      "grad_norm": 0.5264629618291342,
      "learning_rate": 1.4964635714106545e-05,
      "loss": 11.7771,
      "step": 30444
    },
    {
      "epoch": 1.6578519259697917,
      "grad_norm": 0.6101260725520791,
      "learning_rate": 1.495999577471362e-05,
      "loss": 11.7677,
      "step": 30445
    },
    {
      "epoch": 1.6579063799663747,
      "grad_norm": 0.5461996462767426,
      "learning_rate": 1.4955356496606042e-05,
      "loss": 11.8557,
      "step": 30446
    },
    {
      "epoch": 1.6579608339629577,
      "grad_norm": 0.5690858666358461,
      "learning_rate": 1.4950717879819864e-05,
      "loss": 11.7328,
      "step": 30447
    },
    {
      "epoch": 1.6580152879595407,
      "grad_norm": 0.5042772098131838,
      "learning_rate": 1.4946079924391165e-05,
      "loss": 11.7353,
      "step": 30448
    },
    {
      "epoch": 1.6580697419561237,
      "grad_norm": 0.5551249340875368,
      "learning_rate": 1.4941442630356028e-05,
      "loss": 11.8102,
      "step": 30449
    },
    {
      "epoch": 1.6581241959527067,
      "grad_norm": 0.6822976424863264,
      "learning_rate": 1.4936805997750503e-05,
      "loss": 11.9755,
      "step": 30450
    },
    {
      "epoch": 1.6581786499492899,
      "grad_norm": 0.5217195502125209,
      "learning_rate": 1.4932170026610636e-05,
      "loss": 11.7261,
      "step": 30451
    },
    {
      "epoch": 1.6582331039458729,
      "grad_norm": 0.5350002273576602,
      "learning_rate": 1.4927534716972468e-05,
      "loss": 11.8111,
      "step": 30452
    },
    {
      "epoch": 1.6582875579424559,
      "grad_norm": 0.5409593969053802,
      "learning_rate": 1.4922900068872048e-05,
      "loss": 11.8507,
      "step": 30453
    },
    {
      "epoch": 1.6583420119390389,
      "grad_norm": 0.5343336682199202,
      "learning_rate": 1.4918266082345444e-05,
      "loss": 11.8041,
      "step": 30454
    },
    {
      "epoch": 1.6583964659356218,
      "grad_norm": 0.5439666242708063,
      "learning_rate": 1.4913632757428652e-05,
      "loss": 11.8396,
      "step": 30455
    },
    {
      "epoch": 1.6584509199322048,
      "grad_norm": 0.5078023638394096,
      "learning_rate": 1.4909000094157743e-05,
      "loss": 11.7087,
      "step": 30456
    },
    {
      "epoch": 1.6585053739287878,
      "grad_norm": 0.5040080596064688,
      "learning_rate": 1.4904368092568699e-05,
      "loss": 11.7599,
      "step": 30457
    },
    {
      "epoch": 1.6585598279253708,
      "grad_norm": 0.5456376162897237,
      "learning_rate": 1.489973675269758e-05,
      "loss": 11.7348,
      "step": 30458
    },
    {
      "epoch": 1.6586142819219538,
      "grad_norm": 0.5226602620290465,
      "learning_rate": 1.4895106074580355e-05,
      "loss": 11.8768,
      "step": 30459
    },
    {
      "epoch": 1.6586687359185368,
      "grad_norm": 0.5421098455850706,
      "learning_rate": 1.4890476058253089e-05,
      "loss": 11.8537,
      "step": 30460
    },
    {
      "epoch": 1.6587231899151198,
      "grad_norm": 0.52238904348751,
      "learning_rate": 1.4885846703751749e-05,
      "loss": 11.8057,
      "step": 30461
    },
    {
      "epoch": 1.6587776439117028,
      "grad_norm": 0.5643276611856208,
      "learning_rate": 1.4881218011112308e-05,
      "loss": 11.84,
      "step": 30462
    },
    {
      "epoch": 1.6588320979082858,
      "grad_norm": 0.5108632330040422,
      "learning_rate": 1.4876589980370825e-05,
      "loss": 11.7637,
      "step": 30463
    },
    {
      "epoch": 1.6588865519048688,
      "grad_norm": 0.5748429324447424,
      "learning_rate": 1.4871962611563218e-05,
      "loss": 11.6599,
      "step": 30464
    },
    {
      "epoch": 1.6589410059014518,
      "grad_norm": 0.5685506754204068,
      "learning_rate": 1.486733590472551e-05,
      "loss": 11.7119,
      "step": 30465
    },
    {
      "epoch": 1.6589954598980348,
      "grad_norm": 0.5518315507953145,
      "learning_rate": 1.4862709859893708e-05,
      "loss": 11.8669,
      "step": 30466
    },
    {
      "epoch": 1.6590499138946178,
      "grad_norm": 0.5161797691907695,
      "learning_rate": 1.4858084477103717e-05,
      "loss": 11.8546,
      "step": 30467
    },
    {
      "epoch": 1.6591043678912007,
      "grad_norm": 0.5500540514779502,
      "learning_rate": 1.4853459756391564e-05,
      "loss": 11.891,
      "step": 30468
    },
    {
      "epoch": 1.659158821887784,
      "grad_norm": 0.5261406948857814,
      "learning_rate": 1.4848835697793174e-05,
      "loss": 11.8261,
      "step": 30469
    },
    {
      "epoch": 1.659213275884367,
      "grad_norm": 0.5221450123639628,
      "learning_rate": 1.484421230134453e-05,
      "loss": 11.8626,
      "step": 30470
    },
    {
      "epoch": 1.65926772988095,
      "grad_norm": 0.5370308777913975,
      "learning_rate": 1.483958956708157e-05,
      "loss": 11.7647,
      "step": 30471
    },
    {
      "epoch": 1.659322183877533,
      "grad_norm": 0.5451917737767658,
      "learning_rate": 1.483496749504022e-05,
      "loss": 11.7655,
      "step": 30472
    },
    {
      "epoch": 1.659376637874116,
      "grad_norm": 0.5512427098784394,
      "learning_rate": 1.4830346085256475e-05,
      "loss": 11.728,
      "step": 30473
    },
    {
      "epoch": 1.6594310918706991,
      "grad_norm": 0.48879493807625274,
      "learning_rate": 1.4825725337766217e-05,
      "loss": 11.7241,
      "step": 30474
    },
    {
      "epoch": 1.6594855458672821,
      "grad_norm": 0.5323193600760803,
      "learning_rate": 1.4821105252605404e-05,
      "loss": 11.6926,
      "step": 30475
    },
    {
      "epoch": 1.6595399998638651,
      "grad_norm": 0.5357552343034483,
      "learning_rate": 1.481648582980998e-05,
      "loss": 11.873,
      "step": 30476
    },
    {
      "epoch": 1.6595944538604481,
      "grad_norm": 0.5843543350729047,
      "learning_rate": 1.481186706941583e-05,
      "loss": 11.749,
      "step": 30477
    },
    {
      "epoch": 1.6596489078570311,
      "grad_norm": 0.6312598436209041,
      "learning_rate": 1.4807248971458898e-05,
      "loss": 11.9846,
      "step": 30478
    },
    {
      "epoch": 1.6597033618536141,
      "grad_norm": 0.6070430916918202,
      "learning_rate": 1.4802631535975076e-05,
      "loss": 11.8512,
      "step": 30479
    },
    {
      "epoch": 1.659757815850197,
      "grad_norm": 0.5799074330813278,
      "learning_rate": 1.47980147630003e-05,
      "loss": 11.8646,
      "step": 30480
    },
    {
      "epoch": 1.65981226984678,
      "grad_norm": 0.5673598018127015,
      "learning_rate": 1.4793398652570445e-05,
      "loss": 11.7869,
      "step": 30481
    },
    {
      "epoch": 1.659866723843363,
      "grad_norm": 0.5577040261014342,
      "learning_rate": 1.478878320472139e-05,
      "loss": 11.8044,
      "step": 30482
    },
    {
      "epoch": 1.659921177839946,
      "grad_norm": 0.5914851761444279,
      "learning_rate": 1.4784168419489065e-05,
      "loss": 11.926,
      "step": 30483
    },
    {
      "epoch": 1.659975631836529,
      "grad_norm": 0.5864248361351323,
      "learning_rate": 1.4779554296909315e-05,
      "loss": 11.8559,
      "step": 30484
    },
    {
      "epoch": 1.660030085833112,
      "grad_norm": 0.5727600462604755,
      "learning_rate": 1.477494083701807e-05,
      "loss": 11.7601,
      "step": 30485
    },
    {
      "epoch": 1.660084539829695,
      "grad_norm": 0.5837628149486116,
      "learning_rate": 1.4770328039851155e-05,
      "loss": 11.8826,
      "step": 30486
    },
    {
      "epoch": 1.660138993826278,
      "grad_norm": 0.5257373739101987,
      "learning_rate": 1.4765715905444455e-05,
      "loss": 11.7619,
      "step": 30487
    },
    {
      "epoch": 1.660193447822861,
      "grad_norm": 0.5569929326785522,
      "learning_rate": 1.4761104433833873e-05,
      "loss": 11.7252,
      "step": 30488
    },
    {
      "epoch": 1.660247901819444,
      "grad_norm": 0.5596793453691836,
      "learning_rate": 1.4756493625055211e-05,
      "loss": 11.5962,
      "step": 30489
    },
    {
      "epoch": 1.660302355816027,
      "grad_norm": 0.5268308993081,
      "learning_rate": 1.475188347914438e-05,
      "loss": 11.8873,
      "step": 30490
    },
    {
      "epoch": 1.66035680981261,
      "grad_norm": 0.5503512179842955,
      "learning_rate": 1.474727399613719e-05,
      "loss": 11.9062,
      "step": 30491
    },
    {
      "epoch": 1.6604112638091932,
      "grad_norm": 0.5652508792980745,
      "learning_rate": 1.474266517606947e-05,
      "loss": 11.92,
      "step": 30492
    },
    {
      "epoch": 1.6604657178057762,
      "grad_norm": 0.4919906257929225,
      "learning_rate": 1.4738057018977114e-05,
      "loss": 11.8253,
      "step": 30493
    },
    {
      "epoch": 1.6605201718023592,
      "grad_norm": 0.5038695577291233,
      "learning_rate": 1.4733449524895893e-05,
      "loss": 11.8345,
      "step": 30494
    },
    {
      "epoch": 1.6605746257989422,
      "grad_norm": 0.5230842360938279,
      "learning_rate": 1.47288426938617e-05,
      "loss": 11.7675,
      "step": 30495
    },
    {
      "epoch": 1.6606290797955252,
      "grad_norm": 0.5818535948817662,
      "learning_rate": 1.4724236525910296e-05,
      "loss": 11.876,
      "step": 30496
    },
    {
      "epoch": 1.6606835337921082,
      "grad_norm": 0.5106333140474635,
      "learning_rate": 1.471963102107753e-05,
      "loss": 11.5331,
      "step": 30497
    },
    {
      "epoch": 1.6607379877886914,
      "grad_norm": 0.5344151145727453,
      "learning_rate": 1.4715026179399239e-05,
      "loss": 11.7591,
      "step": 30498
    },
    {
      "epoch": 1.6607924417852744,
      "grad_norm": 0.500887120599743,
      "learning_rate": 1.4710422000911183e-05,
      "loss": 11.7601,
      "step": 30499
    },
    {
      "epoch": 1.6608468957818574,
      "grad_norm": 0.5596855613702822,
      "learning_rate": 1.4705818485649237e-05,
      "loss": 11.6184,
      "step": 30500
    },
    {
      "epoch": 1.6609013497784404,
      "grad_norm": 0.6390268041634257,
      "learning_rate": 1.4701215633649102e-05,
      "loss": 11.8665,
      "step": 30501
    },
    {
      "epoch": 1.6609558037750234,
      "grad_norm": 0.5300312005429197,
      "learning_rate": 1.4696613444946627e-05,
      "loss": 11.7044,
      "step": 30502
    },
    {
      "epoch": 1.6610102577716064,
      "grad_norm": 0.5479748636290885,
      "learning_rate": 1.4692011919577609e-05,
      "loss": 11.6887,
      "step": 30503
    },
    {
      "epoch": 1.6610647117681894,
      "grad_norm": 0.5865166959549861,
      "learning_rate": 1.4687411057577782e-05,
      "loss": 11.8067,
      "step": 30504
    },
    {
      "epoch": 1.6611191657647724,
      "grad_norm": 0.631559322131889,
      "learning_rate": 1.4682810858982986e-05,
      "loss": 11.8915,
      "step": 30505
    },
    {
      "epoch": 1.6611736197613554,
      "grad_norm": 0.6867109202353935,
      "learning_rate": 1.4678211323828939e-05,
      "loss": 11.8789,
      "step": 30506
    },
    {
      "epoch": 1.6612280737579384,
      "grad_norm": 0.5306137175143503,
      "learning_rate": 1.4673612452151441e-05,
      "loss": 11.8295,
      "step": 30507
    },
    {
      "epoch": 1.6612825277545213,
      "grad_norm": 0.5384906241963868,
      "learning_rate": 1.4669014243986224e-05,
      "loss": 11.7728,
      "step": 30508
    },
    {
      "epoch": 1.6613369817511043,
      "grad_norm": 0.5363580618189482,
      "learning_rate": 1.4664416699369065e-05,
      "loss": 11.643,
      "step": 30509
    },
    {
      "epoch": 1.6613914357476873,
      "grad_norm": 0.5276766457965232,
      "learning_rate": 1.4659819818335729e-05,
      "loss": 11.7706,
      "step": 30510
    },
    {
      "epoch": 1.6614458897442703,
      "grad_norm": 0.5972428001730457,
      "learning_rate": 1.4655223600921952e-05,
      "loss": 11.7991,
      "step": 30511
    },
    {
      "epoch": 1.6615003437408533,
      "grad_norm": 0.5463440321434727,
      "learning_rate": 1.4650628047163472e-05,
      "loss": 11.7979,
      "step": 30512
    },
    {
      "epoch": 1.6615547977374363,
      "grad_norm": 0.5086481462827724,
      "learning_rate": 1.4646033157095996e-05,
      "loss": 11.8116,
      "step": 30513
    },
    {
      "epoch": 1.6616092517340193,
      "grad_norm": 0.553074260326025,
      "learning_rate": 1.4641438930755268e-05,
      "loss": 11.7544,
      "step": 30514
    },
    {
      "epoch": 1.6616637057306025,
      "grad_norm": 0.5367150667301452,
      "learning_rate": 1.4636845368177065e-05,
      "loss": 11.8423,
      "step": 30515
    },
    {
      "epoch": 1.6617181597271855,
      "grad_norm": 0.56086017051629,
      "learning_rate": 1.4632252469397034e-05,
      "loss": 11.8746,
      "step": 30516
    },
    {
      "epoch": 1.6617726137237685,
      "grad_norm": 0.5090796710024545,
      "learning_rate": 1.4627660234450957e-05,
      "loss": 11.8088,
      "step": 30517
    },
    {
      "epoch": 1.6618270677203515,
      "grad_norm": 0.5647977886294014,
      "learning_rate": 1.4623068663374484e-05,
      "loss": 11.8778,
      "step": 30518
    },
    {
      "epoch": 1.6618815217169345,
      "grad_norm": 0.5852868407151046,
      "learning_rate": 1.4618477756203364e-05,
      "loss": 11.7201,
      "step": 30519
    },
    {
      "epoch": 1.6619359757135175,
      "grad_norm": 0.6172226511052394,
      "learning_rate": 1.4613887512973257e-05,
      "loss": 11.7215,
      "step": 30520
    },
    {
      "epoch": 1.6619904297101007,
      "grad_norm": 0.5306618304665179,
      "learning_rate": 1.4609297933719913e-05,
      "loss": 11.7218,
      "step": 30521
    },
    {
      "epoch": 1.6620448837066837,
      "grad_norm": 0.5640490787690716,
      "learning_rate": 1.460470901847898e-05,
      "loss": 11.8726,
      "step": 30522
    },
    {
      "epoch": 1.6620993377032667,
      "grad_norm": 0.589495949418483,
      "learning_rate": 1.460012076728613e-05,
      "loss": 11.8359,
      "step": 30523
    },
    {
      "epoch": 1.6621537916998497,
      "grad_norm": 0.5767911074678742,
      "learning_rate": 1.4595533180177057e-05,
      "loss": 11.8568,
      "step": 30524
    },
    {
      "epoch": 1.6622082456964327,
      "grad_norm": 0.5925122392277705,
      "learning_rate": 1.4590946257187465e-05,
      "loss": 11.7018,
      "step": 30525
    },
    {
      "epoch": 1.6622626996930157,
      "grad_norm": 0.5372430755545777,
      "learning_rate": 1.4586359998352984e-05,
      "loss": 11.6798,
      "step": 30526
    },
    {
      "epoch": 1.6623171536895986,
      "grad_norm": 0.5798060665421161,
      "learning_rate": 1.4581774403709303e-05,
      "loss": 11.8282,
      "step": 30527
    },
    {
      "epoch": 1.6623716076861816,
      "grad_norm": 0.5419721798492204,
      "learning_rate": 1.4577189473292053e-05,
      "loss": 11.8512,
      "step": 30528
    },
    {
      "epoch": 1.6624260616827646,
      "grad_norm": 0.5229119328459965,
      "learning_rate": 1.4572605207136925e-05,
      "loss": 11.8269,
      "step": 30529
    },
    {
      "epoch": 1.6624805156793476,
      "grad_norm": 0.5830340866341808,
      "learning_rate": 1.4568021605279525e-05,
      "loss": 11.8343,
      "step": 30530
    },
    {
      "epoch": 1.6625349696759306,
      "grad_norm": 0.5839352903112112,
      "learning_rate": 1.4563438667755536e-05,
      "loss": 11.7647,
      "step": 30531
    },
    {
      "epoch": 1.6625894236725136,
      "grad_norm": 0.5442136620128387,
      "learning_rate": 1.4558856394600573e-05,
      "loss": 11.7684,
      "step": 30532
    },
    {
      "epoch": 1.6626438776690966,
      "grad_norm": 0.5287246160487524,
      "learning_rate": 1.455427478585024e-05,
      "loss": 11.7827,
      "step": 30533
    },
    {
      "epoch": 1.6626983316656796,
      "grad_norm": 0.6048785959870898,
      "learning_rate": 1.4549693841540235e-05,
      "loss": 11.8265,
      "step": 30534
    },
    {
      "epoch": 1.6627527856622626,
      "grad_norm": 0.5371811579078356,
      "learning_rate": 1.4545113561706114e-05,
      "loss": 11.9294,
      "step": 30535
    },
    {
      "epoch": 1.6628072396588456,
      "grad_norm": 0.5189119070180531,
      "learning_rate": 1.4540533946383516e-05,
      "loss": 11.7807,
      "step": 30536
    },
    {
      "epoch": 1.6628616936554286,
      "grad_norm": 0.5601368480764733,
      "learning_rate": 1.4535954995608081e-05,
      "loss": 11.7246,
      "step": 30537
    },
    {
      "epoch": 1.6629161476520118,
      "grad_norm": 0.5494316509945332,
      "learning_rate": 1.4531376709415368e-05,
      "loss": 11.8384,
      "step": 30538
    },
    {
      "epoch": 1.6629706016485948,
      "grad_norm": 0.521643862118866,
      "learning_rate": 1.4526799087841037e-05,
      "loss": 11.7458,
      "step": 30539
    },
    {
      "epoch": 1.6630250556451778,
      "grad_norm": 0.5813182171851603,
      "learning_rate": 1.4522222130920616e-05,
      "loss": 11.8722,
      "step": 30540
    },
    {
      "epoch": 1.6630795096417608,
      "grad_norm": 0.515015291051856,
      "learning_rate": 1.4517645838689754e-05,
      "loss": 11.6671,
      "step": 30541
    },
    {
      "epoch": 1.6631339636383438,
      "grad_norm": 0.6300347934852151,
      "learning_rate": 1.451307021118401e-05,
      "loss": 11.933,
      "step": 30542
    },
    {
      "epoch": 1.6631884176349268,
      "grad_norm": 0.5629815631744695,
      "learning_rate": 1.4508495248438958e-05,
      "loss": 11.8467,
      "step": 30543
    },
    {
      "epoch": 1.66324287163151,
      "grad_norm": 0.5630607797950447,
      "learning_rate": 1.4503920950490202e-05,
      "loss": 11.7637,
      "step": 30544
    },
    {
      "epoch": 1.663297325628093,
      "grad_norm": 0.5409344508512145,
      "learning_rate": 1.4499347317373268e-05,
      "loss": 11.8202,
      "step": 30545
    },
    {
      "epoch": 1.663351779624676,
      "grad_norm": 0.5470298260052968,
      "learning_rate": 1.4494774349123741e-05,
      "loss": 11.834,
      "step": 30546
    },
    {
      "epoch": 1.663406233621259,
      "grad_norm": 0.5327489775189177,
      "learning_rate": 1.4490202045777224e-05,
      "loss": 11.6929,
      "step": 30547
    },
    {
      "epoch": 1.663460687617842,
      "grad_norm": 0.5302068927531721,
      "learning_rate": 1.4485630407369211e-05,
      "loss": 11.6785,
      "step": 30548
    },
    {
      "epoch": 1.663515141614425,
      "grad_norm": 0.523235391529584,
      "learning_rate": 1.4481059433935296e-05,
      "loss": 11.812,
      "step": 30549
    },
    {
      "epoch": 1.663569595611008,
      "grad_norm": 0.5708136537775274,
      "learning_rate": 1.4476489125510973e-05,
      "loss": 11.728,
      "step": 30550
    },
    {
      "epoch": 1.663624049607591,
      "grad_norm": 0.5440375571530032,
      "learning_rate": 1.4471919482131846e-05,
      "loss": 11.7661,
      "step": 30551
    },
    {
      "epoch": 1.663678503604174,
      "grad_norm": 0.5273290339034447,
      "learning_rate": 1.4467350503833421e-05,
      "loss": 11.7998,
      "step": 30552
    },
    {
      "epoch": 1.663732957600757,
      "grad_norm": 0.837060976833387,
      "learning_rate": 1.4462782190651191e-05,
      "loss": 11.6781,
      "step": 30553
    },
    {
      "epoch": 1.66378741159734,
      "grad_norm": 0.5517925530154327,
      "learning_rate": 1.4458214542620729e-05,
      "loss": 11.889,
      "step": 30554
    },
    {
      "epoch": 1.6638418655939229,
      "grad_norm": 0.5398113363513517,
      "learning_rate": 1.4453647559777527e-05,
      "loss": 11.8162,
      "step": 30555
    },
    {
      "epoch": 1.6638963195905059,
      "grad_norm": 0.48781454370799227,
      "learning_rate": 1.4449081242157126e-05,
      "loss": 11.6628,
      "step": 30556
    },
    {
      "epoch": 1.6639507735870889,
      "grad_norm": 0.5324761965955556,
      "learning_rate": 1.4444515589794982e-05,
      "loss": 11.7681,
      "step": 30557
    },
    {
      "epoch": 1.6640052275836719,
      "grad_norm": 0.5190118095315441,
      "learning_rate": 1.4439950602726648e-05,
      "loss": 11.7628,
      "step": 30558
    },
    {
      "epoch": 1.6640596815802549,
      "grad_norm": 0.562217111189288,
      "learning_rate": 1.4435386280987618e-05,
      "loss": 11.9103,
      "step": 30559
    },
    {
      "epoch": 1.6641141355768378,
      "grad_norm": 0.5883842957215601,
      "learning_rate": 1.443082262461336e-05,
      "loss": 11.9338,
      "step": 30560
    },
    {
      "epoch": 1.6641685895734208,
      "grad_norm": 0.6249725874539235,
      "learning_rate": 1.4426259633639416e-05,
      "loss": 11.8424,
      "step": 30561
    },
    {
      "epoch": 1.664223043570004,
      "grad_norm": 0.4985586351457106,
      "learning_rate": 1.4421697308101178e-05,
      "loss": 11.7519,
      "step": 30562
    },
    {
      "epoch": 1.664277497566587,
      "grad_norm": 0.5391981641148595,
      "learning_rate": 1.4417135648034186e-05,
      "loss": 11.8427,
      "step": 30563
    },
    {
      "epoch": 1.66433195156317,
      "grad_norm": 0.5288522672615051,
      "learning_rate": 1.441257465347391e-05,
      "loss": 11.7846,
      "step": 30564
    },
    {
      "epoch": 1.664386405559753,
      "grad_norm": 0.5518269616252821,
      "learning_rate": 1.4408014324455787e-05,
      "loss": 11.7347,
      "step": 30565
    },
    {
      "epoch": 1.664440859556336,
      "grad_norm": 0.6027580192488022,
      "learning_rate": 1.4403454661015326e-05,
      "loss": 11.8817,
      "step": 30566
    },
    {
      "epoch": 1.664495313552919,
      "grad_norm": 0.5729240918869509,
      "learning_rate": 1.4398895663187927e-05,
      "loss": 11.7502,
      "step": 30567
    },
    {
      "epoch": 1.6645497675495022,
      "grad_norm": 0.5349704726100336,
      "learning_rate": 1.43943373310091e-05,
      "loss": 11.6887,
      "step": 30568
    },
    {
      "epoch": 1.6646042215460852,
      "grad_norm": 0.6587897180079224,
      "learning_rate": 1.4389779664514235e-05,
      "loss": 11.7798,
      "step": 30569
    },
    {
      "epoch": 1.6646586755426682,
      "grad_norm": 0.5412107808277605,
      "learning_rate": 1.4385222663738796e-05,
      "loss": 11.8893,
      "step": 30570
    },
    {
      "epoch": 1.6647131295392512,
      "grad_norm": 0.5234676013799723,
      "learning_rate": 1.4380666328718274e-05,
      "loss": 11.8058,
      "step": 30571
    },
    {
      "epoch": 1.6647675835358342,
      "grad_norm": 0.5545345381822373,
      "learning_rate": 1.4376110659488006e-05,
      "loss": 11.8276,
      "step": 30572
    },
    {
      "epoch": 1.6648220375324172,
      "grad_norm": 0.5127483290730295,
      "learning_rate": 1.4371555656083457e-05,
      "loss": 11.7903,
      "step": 30573
    },
    {
      "epoch": 1.6648764915290002,
      "grad_norm": 0.6074805114055226,
      "learning_rate": 1.4367001318540075e-05,
      "loss": 11.7282,
      "step": 30574
    },
    {
      "epoch": 1.6649309455255832,
      "grad_norm": 0.524109377181651,
      "learning_rate": 1.4362447646893218e-05,
      "loss": 11.774,
      "step": 30575
    },
    {
      "epoch": 1.6649853995221662,
      "grad_norm": 0.5469659882901008,
      "learning_rate": 1.4357894641178371e-05,
      "loss": 11.7587,
      "step": 30576
    },
    {
      "epoch": 1.6650398535187492,
      "grad_norm": 0.5216704136518607,
      "learning_rate": 1.435334230143086e-05,
      "loss": 11.8727,
      "step": 30577
    },
    {
      "epoch": 1.6650943075153322,
      "grad_norm": 0.5255879142118899,
      "learning_rate": 1.4348790627686149e-05,
      "loss": 11.8424,
      "step": 30578
    },
    {
      "epoch": 1.6651487615119152,
      "grad_norm": 0.5120144482353609,
      "learning_rate": 1.4344239619979583e-05,
      "loss": 11.7848,
      "step": 30579
    },
    {
      "epoch": 1.6652032155084981,
      "grad_norm": 0.6554859104398678,
      "learning_rate": 1.433968927834658e-05,
      "loss": 11.7708,
      "step": 30580
    },
    {
      "epoch": 1.6652576695050811,
      "grad_norm": 0.5535995453122468,
      "learning_rate": 1.4335139602822557e-05,
      "loss": 11.6216,
      "step": 30581
    },
    {
      "epoch": 1.6653121235016641,
      "grad_norm": 0.5198100048389207,
      "learning_rate": 1.4330590593442817e-05,
      "loss": 11.5929,
      "step": 30582
    },
    {
      "epoch": 1.6653665774982471,
      "grad_norm": 0.5358411842760149,
      "learning_rate": 1.4326042250242789e-05,
      "loss": 11.8566,
      "step": 30583
    },
    {
      "epoch": 1.6654210314948301,
      "grad_norm": 0.5785564122028076,
      "learning_rate": 1.43214945732578e-05,
      "loss": 11.7175,
      "step": 30584
    },
    {
      "epoch": 1.6654754854914133,
      "grad_norm": 0.5435001051217152,
      "learning_rate": 1.431694756252323e-05,
      "loss": 11.7581,
      "step": 30585
    },
    {
      "epoch": 1.6655299394879963,
      "grad_norm": 0.5384336380334687,
      "learning_rate": 1.4312401218074478e-05,
      "loss": 11.7637,
      "step": 30586
    },
    {
      "epoch": 1.6655843934845793,
      "grad_norm": 0.5429834110530761,
      "learning_rate": 1.4307855539946847e-05,
      "loss": 11.8063,
      "step": 30587
    },
    {
      "epoch": 1.6656388474811623,
      "grad_norm": 0.5496962632683134,
      "learning_rate": 1.4303310528175717e-05,
      "loss": 11.8142,
      "step": 30588
    },
    {
      "epoch": 1.6656933014777453,
      "grad_norm": 0.5093237154352173,
      "learning_rate": 1.4298766182796386e-05,
      "loss": 11.7745,
      "step": 30589
    },
    {
      "epoch": 1.6657477554743283,
      "grad_norm": 0.5365372478799709,
      "learning_rate": 1.4294222503844257e-05,
      "loss": 11.8519,
      "step": 30590
    },
    {
      "epoch": 1.6658022094709115,
      "grad_norm": 0.5333590014600259,
      "learning_rate": 1.4289679491354613e-05,
      "loss": 11.8323,
      "step": 30591
    },
    {
      "epoch": 1.6658566634674945,
      "grad_norm": 0.5855486768965955,
      "learning_rate": 1.4285137145362781e-05,
      "loss": 11.8671,
      "step": 30592
    },
    {
      "epoch": 1.6659111174640775,
      "grad_norm": 0.5117628893373161,
      "learning_rate": 1.4280595465904123e-05,
      "loss": 11.8224,
      "step": 30593
    },
    {
      "epoch": 1.6659655714606605,
      "grad_norm": 0.5177405067725183,
      "learning_rate": 1.42760544530139e-05,
      "loss": 11.727,
      "step": 30594
    },
    {
      "epoch": 1.6660200254572435,
      "grad_norm": 0.590421014802051,
      "learning_rate": 1.4271514106727458e-05,
      "loss": 11.8403,
      "step": 30595
    },
    {
      "epoch": 1.6660744794538265,
      "grad_norm": 0.5205639485560986,
      "learning_rate": 1.4266974427080115e-05,
      "loss": 11.8655,
      "step": 30596
    },
    {
      "epoch": 1.6661289334504095,
      "grad_norm": 0.6257887917244348,
      "learning_rate": 1.4262435414107134e-05,
      "loss": 11.9091,
      "step": 30597
    },
    {
      "epoch": 1.6661833874469925,
      "grad_norm": 0.5213222188455704,
      "learning_rate": 1.4257897067843862e-05,
      "loss": 11.7735,
      "step": 30598
    },
    {
      "epoch": 1.6662378414435755,
      "grad_norm": 0.6016607595260474,
      "learning_rate": 1.4253359388325537e-05,
      "loss": 11.7408,
      "step": 30599
    },
    {
      "epoch": 1.6662922954401584,
      "grad_norm": 0.5728991967714202,
      "learning_rate": 1.4248822375587489e-05,
      "loss": 11.7922,
      "step": 30600
    },
    {
      "epoch": 1.6663467494367414,
      "grad_norm": 0.6421991607919589,
      "learning_rate": 1.4244286029664988e-05,
      "loss": 11.9538,
      "step": 30601
    },
    {
      "epoch": 1.6664012034333244,
      "grad_norm": 0.527930619823098,
      "learning_rate": 1.4239750350593273e-05,
      "loss": 11.8458,
      "step": 30602
    },
    {
      "epoch": 1.6664556574299074,
      "grad_norm": 0.5772038038487892,
      "learning_rate": 1.4235215338407658e-05,
      "loss": 11.7478,
      "step": 30603
    },
    {
      "epoch": 1.6665101114264904,
      "grad_norm": 0.5274345854094674,
      "learning_rate": 1.4230680993143376e-05,
      "loss": 11.7769,
      "step": 30604
    },
    {
      "epoch": 1.6665645654230734,
      "grad_norm": 0.5411191221571823,
      "learning_rate": 1.4226147314835714e-05,
      "loss": 11.7858,
      "step": 30605
    },
    {
      "epoch": 1.6666190194196564,
      "grad_norm": 0.4910491243074338,
      "learning_rate": 1.4221614303519904e-05,
      "loss": 11.7864,
      "step": 30606
    },
    {
      "epoch": 1.6666734734162394,
      "grad_norm": 0.5490921718782192,
      "learning_rate": 1.4217081959231204e-05,
      "loss": 11.8052,
      "step": 30607
    },
    {
      "epoch": 1.6667279274128226,
      "grad_norm": 0.5370539524170274,
      "learning_rate": 1.4212550282004878e-05,
      "loss": 11.8518,
      "step": 30608
    },
    {
      "epoch": 1.6667823814094056,
      "grad_norm": 0.5763943784481144,
      "learning_rate": 1.4208019271876128e-05,
      "loss": 11.8761,
      "step": 30609
    },
    {
      "epoch": 1.6668368354059886,
      "grad_norm": 0.5489631288190735,
      "learning_rate": 1.4203488928880226e-05,
      "loss": 11.7234,
      "step": 30610
    },
    {
      "epoch": 1.6668912894025716,
      "grad_norm": 0.6224037636667891,
      "learning_rate": 1.4198959253052391e-05,
      "loss": 11.8845,
      "step": 30611
    },
    {
      "epoch": 1.6669457433991546,
      "grad_norm": 0.5296441792635029,
      "learning_rate": 1.4194430244427802e-05,
      "loss": 11.8248,
      "step": 30612
    },
    {
      "epoch": 1.6670001973957376,
      "grad_norm": 0.6968604979288429,
      "learning_rate": 1.4189901903041746e-05,
      "loss": 11.8933,
      "step": 30613
    },
    {
      "epoch": 1.6670546513923208,
      "grad_norm": 0.5899802956762564,
      "learning_rate": 1.4185374228929382e-05,
      "loss": 11.9042,
      "step": 30614
    },
    {
      "epoch": 1.6671091053889038,
      "grad_norm": 0.5404336158935368,
      "learning_rate": 1.4180847222125959e-05,
      "loss": 11.834,
      "step": 30615
    },
    {
      "epoch": 1.6671635593854868,
      "grad_norm": 0.5609721932057188,
      "learning_rate": 1.4176320882666627e-05,
      "loss": 11.8251,
      "step": 30616
    },
    {
      "epoch": 1.6672180133820698,
      "grad_norm": 0.5526477556977242,
      "learning_rate": 1.4171795210586658e-05,
      "loss": 11.7074,
      "step": 30617
    },
    {
      "epoch": 1.6672724673786528,
      "grad_norm": 0.5548255554874686,
      "learning_rate": 1.4167270205921169e-05,
      "loss": 11.7813,
      "step": 30618
    },
    {
      "epoch": 1.6673269213752357,
      "grad_norm": 0.5542599397274358,
      "learning_rate": 1.4162745868705373e-05,
      "loss": 11.7384,
      "step": 30619
    },
    {
      "epoch": 1.6673813753718187,
      "grad_norm": 0.5542399215659537,
      "learning_rate": 1.4158222198974502e-05,
      "loss": 11.8433,
      "step": 30620
    },
    {
      "epoch": 1.6674358293684017,
      "grad_norm": 0.5338091516642297,
      "learning_rate": 1.415369919676368e-05,
      "loss": 11.8447,
      "step": 30621
    },
    {
      "epoch": 1.6674902833649847,
      "grad_norm": 0.5192170780996689,
      "learning_rate": 1.414917686210806e-05,
      "loss": 11.8986,
      "step": 30622
    },
    {
      "epoch": 1.6675447373615677,
      "grad_norm": 0.563720811981106,
      "learning_rate": 1.4144655195042877e-05,
      "loss": 11.8853,
      "step": 30623
    },
    {
      "epoch": 1.6675991913581507,
      "grad_norm": 0.5031909712320997,
      "learning_rate": 1.4140134195603216e-05,
      "loss": 11.7591,
      "step": 30624
    },
    {
      "epoch": 1.6676536453547337,
      "grad_norm": 0.6025047116870932,
      "learning_rate": 1.4135613863824304e-05,
      "loss": 11.7564,
      "step": 30625
    },
    {
      "epoch": 1.6677080993513167,
      "grad_norm": 0.4733156463199147,
      "learning_rate": 1.4131094199741224e-05,
      "loss": 11.8403,
      "step": 30626
    },
    {
      "epoch": 1.6677625533478997,
      "grad_norm": 0.5721359322113635,
      "learning_rate": 1.4126575203389181e-05,
      "loss": 11.8232,
      "step": 30627
    },
    {
      "epoch": 1.6678170073444827,
      "grad_norm": 0.534762221395335,
      "learning_rate": 1.412205687480328e-05,
      "loss": 11.779,
      "step": 30628
    },
    {
      "epoch": 1.6678714613410657,
      "grad_norm": 0.5322262854907802,
      "learning_rate": 1.411753921401865e-05,
      "loss": 11.7902,
      "step": 30629
    },
    {
      "epoch": 1.6679259153376487,
      "grad_norm": 0.5514355863362038,
      "learning_rate": 1.4113022221070472e-05,
      "loss": 11.8879,
      "step": 30630
    },
    {
      "epoch": 1.6679803693342317,
      "grad_norm": 0.502506217205064,
      "learning_rate": 1.410850589599383e-05,
      "loss": 11.7194,
      "step": 30631
    },
    {
      "epoch": 1.6680348233308149,
      "grad_norm": 0.5694251018926452,
      "learning_rate": 1.410399023882385e-05,
      "loss": 11.8151,
      "step": 30632
    },
    {
      "epoch": 1.6680892773273979,
      "grad_norm": 0.5125036313871394,
      "learning_rate": 1.4099475249595628e-05,
      "loss": 11.8312,
      "step": 30633
    },
    {
      "epoch": 1.6681437313239809,
      "grad_norm": 0.5450231604275171,
      "learning_rate": 1.4094960928344292e-05,
      "loss": 11.7312,
      "step": 30634
    },
    {
      "epoch": 1.6681981853205639,
      "grad_norm": 0.5849536903654996,
      "learning_rate": 1.4090447275104968e-05,
      "loss": 11.8918,
      "step": 30635
    },
    {
      "epoch": 1.6682526393171468,
      "grad_norm": 0.5251727664436304,
      "learning_rate": 1.4085934289912706e-05,
      "loss": 11.8149,
      "step": 30636
    },
    {
      "epoch": 1.66830709331373,
      "grad_norm": 0.6149918819302681,
      "learning_rate": 1.4081421972802655e-05,
      "loss": 11.8117,
      "step": 30637
    },
    {
      "epoch": 1.668361547310313,
      "grad_norm": 0.5665962930161643,
      "learning_rate": 1.4076910323809845e-05,
      "loss": 11.5848,
      "step": 30638
    },
    {
      "epoch": 1.668416001306896,
      "grad_norm": 0.5111726167768292,
      "learning_rate": 1.4072399342969422e-05,
      "loss": 11.8496,
      "step": 30639
    },
    {
      "epoch": 1.668470455303479,
      "grad_norm": 0.552531943121508,
      "learning_rate": 1.4067889030316406e-05,
      "loss": 11.6825,
      "step": 30640
    },
    {
      "epoch": 1.668524909300062,
      "grad_norm": 0.5175033775069464,
      "learning_rate": 1.4063379385885911e-05,
      "loss": 11.7762,
      "step": 30641
    },
    {
      "epoch": 1.668579363296645,
      "grad_norm": 0.6186484060029801,
      "learning_rate": 1.4058870409713e-05,
      "loss": 12.0445,
      "step": 30642
    },
    {
      "epoch": 1.668633817293228,
      "grad_norm": 0.5427269369432087,
      "learning_rate": 1.4054362101832696e-05,
      "loss": 11.8152,
      "step": 30643
    },
    {
      "epoch": 1.668688271289811,
      "grad_norm": 0.5530534526730182,
      "learning_rate": 1.4049854462280088e-05,
      "loss": 11.8655,
      "step": 30644
    },
    {
      "epoch": 1.668742725286394,
      "grad_norm": 0.5780781935152243,
      "learning_rate": 1.4045347491090254e-05,
      "loss": 11.7665,
      "step": 30645
    },
    {
      "epoch": 1.668797179282977,
      "grad_norm": 0.5181392305023769,
      "learning_rate": 1.4040841188298182e-05,
      "loss": 11.7054,
      "step": 30646
    },
    {
      "epoch": 1.66885163327956,
      "grad_norm": 0.636847789134479,
      "learning_rate": 1.4036335553938962e-05,
      "loss": 11.805,
      "step": 30647
    },
    {
      "epoch": 1.668906087276143,
      "grad_norm": 0.5660885977660897,
      "learning_rate": 1.4031830588047601e-05,
      "loss": 11.8311,
      "step": 30648
    },
    {
      "epoch": 1.668960541272726,
      "grad_norm": 0.6833788173018199,
      "learning_rate": 1.4027326290659159e-05,
      "loss": 11.8768,
      "step": 30649
    },
    {
      "epoch": 1.669014995269309,
      "grad_norm": 0.6139854201866528,
      "learning_rate": 1.4022822661808621e-05,
      "loss": 11.8321,
      "step": 30650
    },
    {
      "epoch": 1.669069449265892,
      "grad_norm": 0.5552076093213949,
      "learning_rate": 1.4018319701531035e-05,
      "loss": 11.7677,
      "step": 30651
    },
    {
      "epoch": 1.669123903262475,
      "grad_norm": 0.5209814904995312,
      "learning_rate": 1.4013817409861463e-05,
      "loss": 11.7055,
      "step": 30652
    },
    {
      "epoch": 1.669178357259058,
      "grad_norm": 0.5339816384169245,
      "learning_rate": 1.400931578683482e-05,
      "loss": 11.7872,
      "step": 30653
    },
    {
      "epoch": 1.669232811255641,
      "grad_norm": 0.5683988407237588,
      "learning_rate": 1.400481483248618e-05,
      "loss": 11.8208,
      "step": 30654
    },
    {
      "epoch": 1.6692872652522242,
      "grad_norm": 0.5907841922901259,
      "learning_rate": 1.4000314546850502e-05,
      "loss": 11.8629,
      "step": 30655
    },
    {
      "epoch": 1.6693417192488071,
      "grad_norm": 0.5407176440878831,
      "learning_rate": 1.3995814929962791e-05,
      "loss": 11.8693,
      "step": 30656
    },
    {
      "epoch": 1.6693961732453901,
      "grad_norm": 0.5387535016711819,
      "learning_rate": 1.3991315981858077e-05,
      "loss": 11.8178,
      "step": 30657
    },
    {
      "epoch": 1.6694506272419731,
      "grad_norm": 0.5622576450184463,
      "learning_rate": 1.3986817702571286e-05,
      "loss": 11.8224,
      "step": 30658
    },
    {
      "epoch": 1.6695050812385561,
      "grad_norm": 0.5118073203746123,
      "learning_rate": 1.3982320092137447e-05,
      "loss": 11.8534,
      "step": 30659
    },
    {
      "epoch": 1.6695595352351391,
      "grad_norm": 0.540542840608444,
      "learning_rate": 1.3977823150591496e-05,
      "loss": 11.8103,
      "step": 30660
    },
    {
      "epoch": 1.6696139892317223,
      "grad_norm": 0.5482135433940559,
      "learning_rate": 1.3973326877968429e-05,
      "loss": 11.7786,
      "step": 30661
    },
    {
      "epoch": 1.6696684432283053,
      "grad_norm": 0.553369146661749,
      "learning_rate": 1.3968831274303206e-05,
      "loss": 11.8318,
      "step": 30662
    },
    {
      "epoch": 1.6697228972248883,
      "grad_norm": 0.5073757333984068,
      "learning_rate": 1.3964336339630757e-05,
      "loss": 11.7957,
      "step": 30663
    },
    {
      "epoch": 1.6697773512214713,
      "grad_norm": 0.5055837732603168,
      "learning_rate": 1.3959842073986085e-05,
      "loss": 11.7707,
      "step": 30664
    },
    {
      "epoch": 1.6698318052180543,
      "grad_norm": 0.559876593933698,
      "learning_rate": 1.3955348477404072e-05,
      "loss": 11.7811,
      "step": 30665
    },
    {
      "epoch": 1.6698862592146373,
      "grad_norm": 0.5333976863578087,
      "learning_rate": 1.395085554991974e-05,
      "loss": 11.8541,
      "step": 30666
    },
    {
      "epoch": 1.6699407132112203,
      "grad_norm": 0.5454917870028871,
      "learning_rate": 1.3946363291567944e-05,
      "loss": 11.797,
      "step": 30667
    },
    {
      "epoch": 1.6699951672078033,
      "grad_norm": 0.530962618090314,
      "learning_rate": 1.3941871702383669e-05,
      "loss": 11.8457,
      "step": 30668
    },
    {
      "epoch": 1.6700496212043863,
      "grad_norm": 0.488430907721536,
      "learning_rate": 1.3937380782401855e-05,
      "loss": 11.7797,
      "step": 30669
    },
    {
      "epoch": 1.6701040752009693,
      "grad_norm": 0.5307821084548326,
      "learning_rate": 1.3932890531657373e-05,
      "loss": 11.796,
      "step": 30670
    },
    {
      "epoch": 1.6701585291975523,
      "grad_norm": 0.5233877365304789,
      "learning_rate": 1.3928400950185194e-05,
      "loss": 11.8846,
      "step": 30671
    },
    {
      "epoch": 1.6702129831941352,
      "grad_norm": 0.5273684616147256,
      "learning_rate": 1.3923912038020204e-05,
      "loss": 11.7697,
      "step": 30672
    },
    {
      "epoch": 1.6702674371907182,
      "grad_norm": 0.6265352816580725,
      "learning_rate": 1.3919423795197284e-05,
      "loss": 11.7663,
      "step": 30673
    },
    {
      "epoch": 1.6703218911873012,
      "grad_norm": 0.5377756673137262,
      "learning_rate": 1.3914936221751384e-05,
      "loss": 11.7565,
      "step": 30674
    },
    {
      "epoch": 1.6703763451838842,
      "grad_norm": 0.5160009649868486,
      "learning_rate": 1.3910449317717356e-05,
      "loss": 11.7871,
      "step": 30675
    },
    {
      "epoch": 1.6704307991804672,
      "grad_norm": 0.5258348428317202,
      "learning_rate": 1.3905963083130135e-05,
      "loss": 11.843,
      "step": 30676
    },
    {
      "epoch": 1.6704852531770502,
      "grad_norm": 0.5179978593831852,
      "learning_rate": 1.3901477518024552e-05,
      "loss": 11.7681,
      "step": 30677
    },
    {
      "epoch": 1.6705397071736334,
      "grad_norm": 0.5239347580002798,
      "learning_rate": 1.3896992622435523e-05,
      "loss": 11.7877,
      "step": 30678
    },
    {
      "epoch": 1.6705941611702164,
      "grad_norm": 0.5381226775521556,
      "learning_rate": 1.3892508396397941e-05,
      "loss": 11.7176,
      "step": 30679
    },
    {
      "epoch": 1.6706486151667994,
      "grad_norm": 0.5557597232340132,
      "learning_rate": 1.3888024839946635e-05,
      "loss": 11.8947,
      "step": 30680
    },
    {
      "epoch": 1.6707030691633824,
      "grad_norm": 0.5284770407387012,
      "learning_rate": 1.3883541953116508e-05,
      "loss": 11.7098,
      "step": 30681
    },
    {
      "epoch": 1.6707575231599654,
      "grad_norm": 0.5503145099653577,
      "learning_rate": 1.3879059735942401e-05,
      "loss": 11.8881,
      "step": 30682
    },
    {
      "epoch": 1.6708119771565484,
      "grad_norm": 0.6382435746497301,
      "learning_rate": 1.3874578188459153e-05,
      "loss": 11.8654,
      "step": 30683
    },
    {
      "epoch": 1.6708664311531316,
      "grad_norm": 0.5822616467219236,
      "learning_rate": 1.3870097310701636e-05,
      "loss": 11.9282,
      "step": 30684
    },
    {
      "epoch": 1.6709208851497146,
      "grad_norm": 0.5727434133047872,
      "learning_rate": 1.3865617102704676e-05,
      "loss": 11.7462,
      "step": 30685
    },
    {
      "epoch": 1.6709753391462976,
      "grad_norm": 0.515647104081518,
      "learning_rate": 1.3861137564503135e-05,
      "loss": 11.6965,
      "step": 30686
    },
    {
      "epoch": 1.6710297931428806,
      "grad_norm": 0.5183354230310832,
      "learning_rate": 1.385665869613182e-05,
      "loss": 11.7704,
      "step": 30687
    },
    {
      "epoch": 1.6710842471394636,
      "grad_norm": 0.6054209867642117,
      "learning_rate": 1.3852180497625588e-05,
      "loss": 11.7721,
      "step": 30688
    },
    {
      "epoch": 1.6711387011360466,
      "grad_norm": 0.5805249737938416,
      "learning_rate": 1.384770296901924e-05,
      "loss": 11.8295,
      "step": 30689
    },
    {
      "epoch": 1.6711931551326296,
      "grad_norm": 0.577409391021253,
      "learning_rate": 1.3843226110347584e-05,
      "loss": 11.714,
      "step": 30690
    },
    {
      "epoch": 1.6712476091292126,
      "grad_norm": 0.5656187536623469,
      "learning_rate": 1.3838749921645477e-05,
      "loss": 11.8524,
      "step": 30691
    },
    {
      "epoch": 1.6713020631257955,
      "grad_norm": 0.5594468803485352,
      "learning_rate": 1.3834274402947711e-05,
      "loss": 11.7565,
      "step": 30692
    },
    {
      "epoch": 1.6713565171223785,
      "grad_norm": 0.5624642442627585,
      "learning_rate": 1.3829799554289036e-05,
      "loss": 11.851,
      "step": 30693
    },
    {
      "epoch": 1.6714109711189615,
      "grad_norm": 0.6382464063271283,
      "learning_rate": 1.3825325375704323e-05,
      "loss": 11.7675,
      "step": 30694
    },
    {
      "epoch": 1.6714654251155445,
      "grad_norm": 0.5776129033826342,
      "learning_rate": 1.3820851867228313e-05,
      "loss": 11.8955,
      "step": 30695
    },
    {
      "epoch": 1.6715198791121275,
      "grad_norm": 0.5297254642489581,
      "learning_rate": 1.3816379028895832e-05,
      "loss": 11.8662,
      "step": 30696
    },
    {
      "epoch": 1.6715743331087105,
      "grad_norm": 0.5479233691876345,
      "learning_rate": 1.3811906860741608e-05,
      "loss": 11.8863,
      "step": 30697
    },
    {
      "epoch": 1.6716287871052935,
      "grad_norm": 0.6292991158702002,
      "learning_rate": 1.3807435362800481e-05,
      "loss": 11.7588,
      "step": 30698
    },
    {
      "epoch": 1.6716832411018765,
      "grad_norm": 0.517464277885571,
      "learning_rate": 1.3802964535107177e-05,
      "loss": 11.8557,
      "step": 30699
    },
    {
      "epoch": 1.6717376950984595,
      "grad_norm": 0.5444021249152353,
      "learning_rate": 1.3798494377696459e-05,
      "loss": 11.8088,
      "step": 30700
    },
    {
      "epoch": 1.6717921490950425,
      "grad_norm": 0.5918704828124932,
      "learning_rate": 1.379402489060314e-05,
      "loss": 11.7818,
      "step": 30701
    },
    {
      "epoch": 1.6718466030916257,
      "grad_norm": 0.5443720411630387,
      "learning_rate": 1.3789556073861931e-05,
      "loss": 11.9303,
      "step": 30702
    },
    {
      "epoch": 1.6719010570882087,
      "grad_norm": 0.5499557398483298,
      "learning_rate": 1.37850879275076e-05,
      "loss": 11.7787,
      "step": 30703
    },
    {
      "epoch": 1.6719555110847917,
      "grad_norm": 0.49798541994472717,
      "learning_rate": 1.3780620451574855e-05,
      "loss": 11.7193,
      "step": 30704
    },
    {
      "epoch": 1.6720099650813747,
      "grad_norm": 0.5402006959711375,
      "learning_rate": 1.3776153646098467e-05,
      "loss": 11.7147,
      "step": 30705
    },
    {
      "epoch": 1.6720644190779577,
      "grad_norm": 0.5664303938988359,
      "learning_rate": 1.3771687511113186e-05,
      "loss": 11.8356,
      "step": 30706
    },
    {
      "epoch": 1.6721188730745409,
      "grad_norm": 0.5531096500165288,
      "learning_rate": 1.3767222046653705e-05,
      "loss": 11.8854,
      "step": 30707
    },
    {
      "epoch": 1.6721733270711239,
      "grad_norm": 0.5545658808717064,
      "learning_rate": 1.3762757252754788e-05,
      "loss": 11.8637,
      "step": 30708
    },
    {
      "epoch": 1.6722277810677069,
      "grad_norm": 0.569745662900167,
      "learning_rate": 1.3758293129451116e-05,
      "loss": 11.884,
      "step": 30709
    },
    {
      "epoch": 1.6722822350642899,
      "grad_norm": 0.5800772003933736,
      "learning_rate": 1.375382967677743e-05,
      "loss": 11.8379,
      "step": 30710
    },
    {
      "epoch": 1.6723366890608728,
      "grad_norm": 0.5226995107064963,
      "learning_rate": 1.3749366894768412e-05,
      "loss": 11.8163,
      "step": 30711
    },
    {
      "epoch": 1.6723911430574558,
      "grad_norm": 0.5779024817809948,
      "learning_rate": 1.37449047834588e-05,
      "loss": 11.8702,
      "step": 30712
    },
    {
      "epoch": 1.6724455970540388,
      "grad_norm": 0.5495962175139693,
      "learning_rate": 1.374044334288328e-05,
      "loss": 11.742,
      "step": 30713
    },
    {
      "epoch": 1.6725000510506218,
      "grad_norm": 0.5564836727694991,
      "learning_rate": 1.3735982573076511e-05,
      "loss": 11.757,
      "step": 30714
    },
    {
      "epoch": 1.6725545050472048,
      "grad_norm": 0.5740085108021452,
      "learning_rate": 1.3731522474073233e-05,
      "loss": 11.7604,
      "step": 30715
    },
    {
      "epoch": 1.6726089590437878,
      "grad_norm": 0.5160515161411331,
      "learning_rate": 1.3727063045908084e-05,
      "loss": 11.8244,
      "step": 30716
    },
    {
      "epoch": 1.6726634130403708,
      "grad_norm": 0.543489470428873,
      "learning_rate": 1.3722604288615759e-05,
      "loss": 11.7809,
      "step": 30717
    },
    {
      "epoch": 1.6727178670369538,
      "grad_norm": 0.5175571927238948,
      "learning_rate": 1.3718146202230953e-05,
      "loss": 11.8779,
      "step": 30718
    },
    {
      "epoch": 1.6727723210335368,
      "grad_norm": 0.5198347335821604,
      "learning_rate": 1.371368878678828e-05,
      "loss": 11.7775,
      "step": 30719
    },
    {
      "epoch": 1.6728267750301198,
      "grad_norm": 0.5414702976639753,
      "learning_rate": 1.3709232042322472e-05,
      "loss": 11.8345,
      "step": 30720
    },
    {
      "epoch": 1.6728812290267028,
      "grad_norm": 0.6141249785605588,
      "learning_rate": 1.3704775968868111e-05,
      "loss": 11.9421,
      "step": 30721
    },
    {
      "epoch": 1.6729356830232858,
      "grad_norm": 0.7390792488911934,
      "learning_rate": 1.3700320566459912e-05,
      "loss": 11.7716,
      "step": 30722
    },
    {
      "epoch": 1.6729901370198688,
      "grad_norm": 0.5173868267241312,
      "learning_rate": 1.3695865835132493e-05,
      "loss": 11.7361,
      "step": 30723
    },
    {
      "epoch": 1.6730445910164518,
      "grad_norm": 0.5474029527500227,
      "learning_rate": 1.3691411774920471e-05,
      "loss": 11.8611,
      "step": 30724
    },
    {
      "epoch": 1.673099045013035,
      "grad_norm": 0.5514893474856272,
      "learning_rate": 1.3686958385858529e-05,
      "loss": 11.8024,
      "step": 30725
    },
    {
      "epoch": 1.673153499009618,
      "grad_norm": 0.577040005904867,
      "learning_rate": 1.368250566798125e-05,
      "loss": 11.7137,
      "step": 30726
    },
    {
      "epoch": 1.673207953006201,
      "grad_norm": 0.5475240789568734,
      "learning_rate": 1.3678053621323283e-05,
      "loss": 11.7408,
      "step": 30727
    },
    {
      "epoch": 1.673262407002784,
      "grad_norm": 0.530659666503884,
      "learning_rate": 1.367360224591927e-05,
      "loss": 11.8441,
      "step": 30728
    },
    {
      "epoch": 1.673316860999367,
      "grad_norm": 0.5611999792445183,
      "learning_rate": 1.3669151541803771e-05,
      "loss": 11.8717,
      "step": 30729
    },
    {
      "epoch": 1.67337131499595,
      "grad_norm": 0.5306339523517843,
      "learning_rate": 1.3664701509011457e-05,
      "loss": 11.806,
      "step": 30730
    },
    {
      "epoch": 1.6734257689925331,
      "grad_norm": 0.5219800102886839,
      "learning_rate": 1.3660252147576879e-05,
      "loss": 11.6249,
      "step": 30731
    },
    {
      "epoch": 1.6734802229891161,
      "grad_norm": 0.574986750708085,
      "learning_rate": 1.3655803457534688e-05,
      "loss": 11.7464,
      "step": 30732
    },
    {
      "epoch": 1.6735346769856991,
      "grad_norm": 0.6243519582976768,
      "learning_rate": 1.3651355438919444e-05,
      "loss": 11.9312,
      "step": 30733
    },
    {
      "epoch": 1.6735891309822821,
      "grad_norm": 0.5939908201828773,
      "learning_rate": 1.364690809176572e-05,
      "loss": 11.8336,
      "step": 30734
    },
    {
      "epoch": 1.6736435849788651,
      "grad_norm": 0.5744204053403791,
      "learning_rate": 1.3642461416108142e-05,
      "loss": 11.6633,
      "step": 30735
    },
    {
      "epoch": 1.673698038975448,
      "grad_norm": 0.5276219268542305,
      "learning_rate": 1.3638015411981242e-05,
      "loss": 11.7969,
      "step": 30736
    },
    {
      "epoch": 1.673752492972031,
      "grad_norm": 0.4895900857711987,
      "learning_rate": 1.3633570079419644e-05,
      "loss": 11.5451,
      "step": 30737
    },
    {
      "epoch": 1.673806946968614,
      "grad_norm": 0.6192604787760138,
      "learning_rate": 1.3629125418457867e-05,
      "loss": 11.8738,
      "step": 30738
    },
    {
      "epoch": 1.673861400965197,
      "grad_norm": 0.5183316054407755,
      "learning_rate": 1.3624681429130493e-05,
      "loss": 11.8157,
      "step": 30739
    },
    {
      "epoch": 1.67391585496178,
      "grad_norm": 0.5352572930781914,
      "learning_rate": 1.3620238111472095e-05,
      "loss": 11.7011,
      "step": 30740
    },
    {
      "epoch": 1.673970308958363,
      "grad_norm": 0.6608333963970415,
      "learning_rate": 1.3615795465517201e-05,
      "loss": 11.9022,
      "step": 30741
    },
    {
      "epoch": 1.674024762954946,
      "grad_norm": 0.5645761140980992,
      "learning_rate": 1.3611353491300383e-05,
      "loss": 11.8001,
      "step": 30742
    },
    {
      "epoch": 1.674079216951529,
      "grad_norm": 0.5175858894914743,
      "learning_rate": 1.360691218885617e-05,
      "loss": 11.7656,
      "step": 30743
    },
    {
      "epoch": 1.674133670948112,
      "grad_norm": 0.5380103375369542,
      "learning_rate": 1.3602471558219076e-05,
      "loss": 11.7123,
      "step": 30744
    },
    {
      "epoch": 1.674188124944695,
      "grad_norm": 0.5079453659542924,
      "learning_rate": 1.3598031599423666e-05,
      "loss": 11.7998,
      "step": 30745
    },
    {
      "epoch": 1.674242578941278,
      "grad_norm": 0.5437587357002618,
      "learning_rate": 1.3593592312504444e-05,
      "loss": 11.8777,
      "step": 30746
    },
    {
      "epoch": 1.674297032937861,
      "grad_norm": 0.5996576782136743,
      "learning_rate": 1.3589153697495948e-05,
      "loss": 11.8434,
      "step": 30747
    },
    {
      "epoch": 1.6743514869344442,
      "grad_norm": 0.5709378066021393,
      "learning_rate": 1.3584715754432664e-05,
      "loss": 11.79,
      "step": 30748
    },
    {
      "epoch": 1.6744059409310272,
      "grad_norm": 0.5601905928731062,
      "learning_rate": 1.3580278483349129e-05,
      "loss": 11.7178,
      "step": 30749
    },
    {
      "epoch": 1.6744603949276102,
      "grad_norm": 0.5323236474538788,
      "learning_rate": 1.3575841884279861e-05,
      "loss": 11.7674,
      "step": 30750
    },
    {
      "epoch": 1.6745148489241932,
      "grad_norm": 0.507929985239759,
      "learning_rate": 1.3571405957259309e-05,
      "loss": 11.7714,
      "step": 30751
    },
    {
      "epoch": 1.6745693029207762,
      "grad_norm": 0.48647216203174226,
      "learning_rate": 1.3566970702322058e-05,
      "loss": 11.8004,
      "step": 30752
    },
    {
      "epoch": 1.6746237569173592,
      "grad_norm": 0.5567592249661721,
      "learning_rate": 1.3562536119502477e-05,
      "loss": 11.5576,
      "step": 30753
    },
    {
      "epoch": 1.6746782109139424,
      "grad_norm": 0.611344391102703,
      "learning_rate": 1.3558102208835121e-05,
      "loss": 11.7727,
      "step": 30754
    },
    {
      "epoch": 1.6747326649105254,
      "grad_norm": 0.5741730647314339,
      "learning_rate": 1.3553668970354483e-05,
      "loss": 11.8161,
      "step": 30755
    },
    {
      "epoch": 1.6747871189071084,
      "grad_norm": 0.5089711773059751,
      "learning_rate": 1.3549236404094978e-05,
      "loss": 11.6753,
      "step": 30756
    },
    {
      "epoch": 1.6748415729036914,
      "grad_norm": 0.5474695100402294,
      "learning_rate": 1.3544804510091136e-05,
      "loss": 11.7483,
      "step": 30757
    },
    {
      "epoch": 1.6748960269002744,
      "grad_norm": 0.5201051846364081,
      "learning_rate": 1.3540373288377372e-05,
      "loss": 11.6236,
      "step": 30758
    },
    {
      "epoch": 1.6749504808968574,
      "grad_norm": 0.6008503605801372,
      "learning_rate": 1.3535942738988194e-05,
      "loss": 11.911,
      "step": 30759
    },
    {
      "epoch": 1.6750049348934404,
      "grad_norm": 0.5490633818844283,
      "learning_rate": 1.3531512861957995e-05,
      "loss": 11.8329,
      "step": 30760
    },
    {
      "epoch": 1.6750593888900234,
      "grad_norm": 0.5956673616462292,
      "learning_rate": 1.3527083657321248e-05,
      "loss": 11.6763,
      "step": 30761
    },
    {
      "epoch": 1.6751138428866064,
      "grad_norm": 0.5871668559190035,
      "learning_rate": 1.3522655125112449e-05,
      "loss": 11.8177,
      "step": 30762
    },
    {
      "epoch": 1.6751682968831894,
      "grad_norm": 0.5252319294898933,
      "learning_rate": 1.3518227265365945e-05,
      "loss": 11.8208,
      "step": 30763
    },
    {
      "epoch": 1.6752227508797723,
      "grad_norm": 0.6494343850464059,
      "learning_rate": 1.3513800078116234e-05,
      "loss": 11.8865,
      "step": 30764
    },
    {
      "epoch": 1.6752772048763553,
      "grad_norm": 0.5577113683637119,
      "learning_rate": 1.3509373563397688e-05,
      "loss": 11.8043,
      "step": 30765
    },
    {
      "epoch": 1.6753316588729383,
      "grad_norm": 0.5479970067312949,
      "learning_rate": 1.3504947721244753e-05,
      "loss": 11.8447,
      "step": 30766
    },
    {
      "epoch": 1.6753861128695213,
      "grad_norm": 0.5144120454661337,
      "learning_rate": 1.3500522551691885e-05,
      "loss": 11.7687,
      "step": 30767
    },
    {
      "epoch": 1.6754405668661043,
      "grad_norm": 0.5836956491599675,
      "learning_rate": 1.3496098054773432e-05,
      "loss": 11.8723,
      "step": 30768
    },
    {
      "epoch": 1.6754950208626873,
      "grad_norm": 0.5516998216749677,
      "learning_rate": 1.3491674230523842e-05,
      "loss": 11.7775,
      "step": 30769
    },
    {
      "epoch": 1.6755494748592703,
      "grad_norm": 0.6221994938327313,
      "learning_rate": 1.3487251078977492e-05,
      "loss": 11.9036,
      "step": 30770
    },
    {
      "epoch": 1.6756039288558535,
      "grad_norm": 0.4966220869176989,
      "learning_rate": 1.3482828600168795e-05,
      "loss": 11.7278,
      "step": 30771
    },
    {
      "epoch": 1.6756583828524365,
      "grad_norm": 0.5655745151714466,
      "learning_rate": 1.3478406794132137e-05,
      "loss": 11.791,
      "step": 30772
    },
    {
      "epoch": 1.6757128368490195,
      "grad_norm": 0.5737705742226658,
      "learning_rate": 1.3473985660901878e-05,
      "loss": 11.8163,
      "step": 30773
    },
    {
      "epoch": 1.6757672908456025,
      "grad_norm": 0.5724829551180888,
      "learning_rate": 1.3469565200512434e-05,
      "loss": 11.8458,
      "step": 30774
    },
    {
      "epoch": 1.6758217448421855,
      "grad_norm": 0.6026529526748534,
      "learning_rate": 1.3465145412998148e-05,
      "loss": 11.7957,
      "step": 30775
    },
    {
      "epoch": 1.6758761988387685,
      "grad_norm": 0.5668312546691957,
      "learning_rate": 1.34607262983934e-05,
      "loss": 11.7239,
      "step": 30776
    },
    {
      "epoch": 1.6759306528353517,
      "grad_norm": 0.5320837458259533,
      "learning_rate": 1.3456307856732575e-05,
      "loss": 11.7107,
      "step": 30777
    },
    {
      "epoch": 1.6759851068319347,
      "grad_norm": 0.5371011898510129,
      "learning_rate": 1.345189008804999e-05,
      "loss": 11.7957,
      "step": 30778
    },
    {
      "epoch": 1.6760395608285177,
      "grad_norm": 0.60628546475273,
      "learning_rate": 1.344747299238005e-05,
      "loss": 11.8693,
      "step": 30779
    },
    {
      "epoch": 1.6760940148251007,
      "grad_norm": 0.5945148256243175,
      "learning_rate": 1.344305656975705e-05,
      "loss": 11.8287,
      "step": 30780
    },
    {
      "epoch": 1.6761484688216837,
      "grad_norm": 0.5422130819635125,
      "learning_rate": 1.3438640820215376e-05,
      "loss": 11.7867,
      "step": 30781
    },
    {
      "epoch": 1.6762029228182667,
      "grad_norm": 0.502865469110512,
      "learning_rate": 1.3434225743789352e-05,
      "loss": 11.6306,
      "step": 30782
    },
    {
      "epoch": 1.6762573768148497,
      "grad_norm": 0.55338832640472,
      "learning_rate": 1.3429811340513287e-05,
      "loss": 11.7155,
      "step": 30783
    },
    {
      "epoch": 1.6763118308114326,
      "grad_norm": 0.5342679761102067,
      "learning_rate": 1.3425397610421541e-05,
      "loss": 11.7603,
      "step": 30784
    },
    {
      "epoch": 1.6763662848080156,
      "grad_norm": 0.5849556399975631,
      "learning_rate": 1.34209845535484e-05,
      "loss": 11.7956,
      "step": 30785
    },
    {
      "epoch": 1.6764207388045986,
      "grad_norm": 0.5567148521287311,
      "learning_rate": 1.3416572169928233e-05,
      "loss": 11.56,
      "step": 30786
    },
    {
      "epoch": 1.6764751928011816,
      "grad_norm": 0.5522869294628729,
      "learning_rate": 1.341216045959529e-05,
      "loss": 11.6991,
      "step": 30787
    },
    {
      "epoch": 1.6765296467977646,
      "grad_norm": 0.5534686532411146,
      "learning_rate": 1.3407749422583915e-05,
      "loss": 11.8312,
      "step": 30788
    },
    {
      "epoch": 1.6765841007943476,
      "grad_norm": 0.59627571237323,
      "learning_rate": 1.3403339058928422e-05,
      "loss": 11.9478,
      "step": 30789
    },
    {
      "epoch": 1.6766385547909306,
      "grad_norm": 0.4876074041422157,
      "learning_rate": 1.339892936866306e-05,
      "loss": 11.7527,
      "step": 30790
    },
    {
      "epoch": 1.6766930087875136,
      "grad_norm": 0.5432630204509983,
      "learning_rate": 1.339452035182217e-05,
      "loss": 11.819,
      "step": 30791
    },
    {
      "epoch": 1.6767474627840966,
      "grad_norm": 0.5341624086670461,
      "learning_rate": 1.3390112008439992e-05,
      "loss": 11.6182,
      "step": 30792
    },
    {
      "epoch": 1.6768019167806796,
      "grad_norm": 0.5474969809089916,
      "learning_rate": 1.3385704338550853e-05,
      "loss": 11.7691,
      "step": 30793
    },
    {
      "epoch": 1.6768563707772626,
      "grad_norm": 0.5223113431713742,
      "learning_rate": 1.3381297342189004e-05,
      "loss": 11.7281,
      "step": 30794
    },
    {
      "epoch": 1.6769108247738458,
      "grad_norm": 0.5546933604599444,
      "learning_rate": 1.3376891019388682e-05,
      "loss": 11.8756,
      "step": 30795
    },
    {
      "epoch": 1.6769652787704288,
      "grad_norm": 0.5150681125434099,
      "learning_rate": 1.3372485370184218e-05,
      "loss": 11.8583,
      "step": 30796
    },
    {
      "epoch": 1.6770197327670118,
      "grad_norm": 0.5345053336037571,
      "learning_rate": 1.3368080394609794e-05,
      "loss": 11.747,
      "step": 30797
    },
    {
      "epoch": 1.6770741867635948,
      "grad_norm": 0.5870743626595543,
      "learning_rate": 1.3363676092699718e-05,
      "loss": 11.9117,
      "step": 30798
    },
    {
      "epoch": 1.6771286407601778,
      "grad_norm": 0.5435534278676204,
      "learning_rate": 1.3359272464488248e-05,
      "loss": 11.8437,
      "step": 30799
    },
    {
      "epoch": 1.6771830947567608,
      "grad_norm": 0.4971165194270236,
      "learning_rate": 1.3354869510009583e-05,
      "loss": 11.7076,
      "step": 30800
    },
    {
      "epoch": 1.677237548753344,
      "grad_norm": 0.5501541279707652,
      "learning_rate": 1.3350467229298002e-05,
      "loss": 11.7002,
      "step": 30801
    },
    {
      "epoch": 1.677292002749927,
      "grad_norm": 0.5257373542374613,
      "learning_rate": 1.3346065622387694e-05,
      "loss": 11.7226,
      "step": 30802
    },
    {
      "epoch": 1.67734645674651,
      "grad_norm": 0.4854746856865274,
      "learning_rate": 1.3341664689312939e-05,
      "loss": 11.7435,
      "step": 30803
    },
    {
      "epoch": 1.677400910743093,
      "grad_norm": 0.5627802516140279,
      "learning_rate": 1.3337264430107921e-05,
      "loss": 11.8204,
      "step": 30804
    },
    {
      "epoch": 1.677455364739676,
      "grad_norm": 0.5372233772155028,
      "learning_rate": 1.3332864844806859e-05,
      "loss": 11.8013,
      "step": 30805
    },
    {
      "epoch": 1.677509818736259,
      "grad_norm": 0.5283210218300569,
      "learning_rate": 1.332846593344399e-05,
      "loss": 11.8013,
      "step": 30806
    },
    {
      "epoch": 1.677564272732842,
      "grad_norm": 0.4962134052480816,
      "learning_rate": 1.3324067696053488e-05,
      "loss": 11.6823,
      "step": 30807
    },
    {
      "epoch": 1.677618726729425,
      "grad_norm": 0.56457898796017,
      "learning_rate": 1.3319670132669593e-05,
      "loss": 11.7457,
      "step": 30808
    },
    {
      "epoch": 1.677673180726008,
      "grad_norm": 0.5815024347259764,
      "learning_rate": 1.3315273243326454e-05,
      "loss": 11.825,
      "step": 30809
    },
    {
      "epoch": 1.677727634722591,
      "grad_norm": 0.5140054589877632,
      "learning_rate": 1.3310877028058277e-05,
      "loss": 11.7688,
      "step": 30810
    },
    {
      "epoch": 1.677782088719174,
      "grad_norm": 0.5833100953870268,
      "learning_rate": 1.3306481486899292e-05,
      "loss": 11.7954,
      "step": 30811
    },
    {
      "epoch": 1.6778365427157569,
      "grad_norm": 0.6723215588131805,
      "learning_rate": 1.3302086619883625e-05,
      "loss": 12.0623,
      "step": 30812
    },
    {
      "epoch": 1.6778909967123399,
      "grad_norm": 0.530142616716068,
      "learning_rate": 1.3297692427045516e-05,
      "loss": 11.8834,
      "step": 30813
    },
    {
      "epoch": 1.6779454507089229,
      "grad_norm": 0.5224907506133561,
      "learning_rate": 1.329329890841904e-05,
      "loss": 11.7784,
      "step": 30814
    },
    {
      "epoch": 1.6779999047055059,
      "grad_norm": 0.5218275472829298,
      "learning_rate": 1.328890606403842e-05,
      "loss": 11.8317,
      "step": 30815
    },
    {
      "epoch": 1.6780543587020889,
      "grad_norm": 0.5489846031224762,
      "learning_rate": 1.3284513893937822e-05,
      "loss": 11.826,
      "step": 30816
    },
    {
      "epoch": 1.6781088126986718,
      "grad_norm": 0.5204790083568857,
      "learning_rate": 1.3280122398151363e-05,
      "loss": 11.7844,
      "step": 30817
    },
    {
      "epoch": 1.678163266695255,
      "grad_norm": 0.5194773603284347,
      "learning_rate": 1.3275731576713247e-05,
      "loss": 11.7438,
      "step": 30818
    },
    {
      "epoch": 1.678217720691838,
      "grad_norm": 0.5871451905002218,
      "learning_rate": 1.3271341429657557e-05,
      "loss": 11.5825,
      "step": 30819
    },
    {
      "epoch": 1.678272174688421,
      "grad_norm": 0.5159076570751091,
      "learning_rate": 1.3266951957018481e-05,
      "loss": 11.7884,
      "step": 30820
    },
    {
      "epoch": 1.678326628685004,
      "grad_norm": 0.6150612967492541,
      "learning_rate": 1.326256315883011e-05,
      "loss": 11.9415,
      "step": 30821
    },
    {
      "epoch": 1.678381082681587,
      "grad_norm": 0.4931136989370211,
      "learning_rate": 1.3258175035126596e-05,
      "loss": 11.8024,
      "step": 30822
    },
    {
      "epoch": 1.67843553667817,
      "grad_norm": 0.5725825767857251,
      "learning_rate": 1.3253787585942112e-05,
      "loss": 11.7904,
      "step": 30823
    },
    {
      "epoch": 1.6784899906747532,
      "grad_norm": 0.6390060461074529,
      "learning_rate": 1.3249400811310663e-05,
      "loss": 12.0097,
      "step": 30824
    },
    {
      "epoch": 1.6785444446713362,
      "grad_norm": 0.5314803990215526,
      "learning_rate": 1.3245014711266435e-05,
      "loss": 11.7944,
      "step": 30825
    },
    {
      "epoch": 1.6785988986679192,
      "grad_norm": 0.5038522679184477,
      "learning_rate": 1.3240629285843542e-05,
      "loss": 11.7378,
      "step": 30826
    },
    {
      "epoch": 1.6786533526645022,
      "grad_norm": 0.5454704986853873,
      "learning_rate": 1.3236244535076036e-05,
      "loss": 11.9079,
      "step": 30827
    },
    {
      "epoch": 1.6787078066610852,
      "grad_norm": 0.557951993907346,
      "learning_rate": 1.3231860458998069e-05,
      "loss": 11.8218,
      "step": 30828
    },
    {
      "epoch": 1.6787622606576682,
      "grad_norm": 0.555182406506991,
      "learning_rate": 1.3227477057643677e-05,
      "loss": 11.6933,
      "step": 30829
    },
    {
      "epoch": 1.6788167146542512,
      "grad_norm": 0.5695684130966864,
      "learning_rate": 1.3223094331047004e-05,
      "loss": 11.656,
      "step": 30830
    },
    {
      "epoch": 1.6788711686508342,
      "grad_norm": 0.5020493245916736,
      "learning_rate": 1.3218712279242084e-05,
      "loss": 11.8005,
      "step": 30831
    },
    {
      "epoch": 1.6789256226474172,
      "grad_norm": 0.5481198553243628,
      "learning_rate": 1.3214330902263006e-05,
      "loss": 11.8994,
      "step": 30832
    },
    {
      "epoch": 1.6789800766440002,
      "grad_norm": 0.5018290173423888,
      "learning_rate": 1.3209950200143884e-05,
      "loss": 11.7898,
      "step": 30833
    },
    {
      "epoch": 1.6790345306405832,
      "grad_norm": 0.502488793789876,
      "learning_rate": 1.3205570172918702e-05,
      "loss": 11.6313,
      "step": 30834
    },
    {
      "epoch": 1.6790889846371662,
      "grad_norm": 0.5579112840523157,
      "learning_rate": 1.320119082062159e-05,
      "loss": 11.8107,
      "step": 30835
    },
    {
      "epoch": 1.6791434386337492,
      "grad_norm": 0.6770367140611824,
      "learning_rate": 1.3196812143286551e-05,
      "loss": 11.8248,
      "step": 30836
    },
    {
      "epoch": 1.6791978926303321,
      "grad_norm": 0.5036100044091174,
      "learning_rate": 1.3192434140947652e-05,
      "loss": 11.7979,
      "step": 30837
    },
    {
      "epoch": 1.6792523466269151,
      "grad_norm": 0.5353437836758715,
      "learning_rate": 1.3188056813638971e-05,
      "loss": 11.7918,
      "step": 30838
    },
    {
      "epoch": 1.6793068006234981,
      "grad_norm": 0.5288200963388185,
      "learning_rate": 1.3183680161394495e-05,
      "loss": 11.6297,
      "step": 30839
    },
    {
      "epoch": 1.6793612546200811,
      "grad_norm": 0.5658695122389732,
      "learning_rate": 1.3179304184248298e-05,
      "loss": 11.8004,
      "step": 30840
    },
    {
      "epoch": 1.6794157086166643,
      "grad_norm": 0.544517652730449,
      "learning_rate": 1.3174928882234373e-05,
      "loss": 11.8579,
      "step": 30841
    },
    {
      "epoch": 1.6794701626132473,
      "grad_norm": 0.4962557362885047,
      "learning_rate": 1.3170554255386791e-05,
      "loss": 11.8731,
      "step": 30842
    },
    {
      "epoch": 1.6795246166098303,
      "grad_norm": 0.5188071007258321,
      "learning_rate": 1.3166180303739528e-05,
      "loss": 11.6215,
      "step": 30843
    },
    {
      "epoch": 1.6795790706064133,
      "grad_norm": 0.5188097641128784,
      "learning_rate": 1.316180702732659e-05,
      "loss": 11.8047,
      "step": 30844
    },
    {
      "epoch": 1.6796335246029963,
      "grad_norm": 0.5264623739112867,
      "learning_rate": 1.3157434426182025e-05,
      "loss": 11.8645,
      "step": 30845
    },
    {
      "epoch": 1.6796879785995793,
      "grad_norm": 0.5310097958857362,
      "learning_rate": 1.3153062500339796e-05,
      "loss": 11.7249,
      "step": 30846
    },
    {
      "epoch": 1.6797424325961625,
      "grad_norm": 0.5991591247823416,
      "learning_rate": 1.3148691249833922e-05,
      "loss": 11.8467,
      "step": 30847
    },
    {
      "epoch": 1.6797968865927455,
      "grad_norm": 0.5257352301499815,
      "learning_rate": 1.3144320674698396e-05,
      "loss": 11.8047,
      "step": 30848
    },
    {
      "epoch": 1.6798513405893285,
      "grad_norm": 0.5647495279696361,
      "learning_rate": 1.3139950774967192e-05,
      "loss": 11.8984,
      "step": 30849
    },
    {
      "epoch": 1.6799057945859115,
      "grad_norm": 0.5275975143919737,
      "learning_rate": 1.3135581550674315e-05,
      "loss": 11.709,
      "step": 30850
    },
    {
      "epoch": 1.6799602485824945,
      "grad_norm": 0.5936834863768895,
      "learning_rate": 1.3131213001853692e-05,
      "loss": 11.8145,
      "step": 30851
    },
    {
      "epoch": 1.6800147025790775,
      "grad_norm": 0.5436783695171021,
      "learning_rate": 1.3126845128539356e-05,
      "loss": 11.6715,
      "step": 30852
    },
    {
      "epoch": 1.6800691565756605,
      "grad_norm": 0.5401272411344255,
      "learning_rate": 1.3122477930765243e-05,
      "loss": 11.8293,
      "step": 30853
    },
    {
      "epoch": 1.6801236105722435,
      "grad_norm": 0.5071004671666826,
      "learning_rate": 1.311811140856528e-05,
      "loss": 11.7706,
      "step": 30854
    },
    {
      "epoch": 1.6801780645688265,
      "grad_norm": 0.5213372358570623,
      "learning_rate": 1.3113745561973478e-05,
      "loss": 11.8414,
      "step": 30855
    },
    {
      "epoch": 1.6802325185654094,
      "grad_norm": 0.5677481754254107,
      "learning_rate": 1.310938039102374e-05,
      "loss": 11.7646,
      "step": 30856
    },
    {
      "epoch": 1.6802869725619924,
      "grad_norm": 0.5898449492356662,
      "learning_rate": 1.310501589575005e-05,
      "loss": 11.7354,
      "step": 30857
    },
    {
      "epoch": 1.6803414265585754,
      "grad_norm": 0.5417890917817078,
      "learning_rate": 1.3100652076186314e-05,
      "loss": 11.7539,
      "step": 30858
    },
    {
      "epoch": 1.6803958805551584,
      "grad_norm": 0.5769759359254978,
      "learning_rate": 1.3096288932366474e-05,
      "loss": 11.7423,
      "step": 30859
    },
    {
      "epoch": 1.6804503345517414,
      "grad_norm": 0.6294268045759392,
      "learning_rate": 1.3091926464324488e-05,
      "loss": 11.8246,
      "step": 30860
    },
    {
      "epoch": 1.6805047885483244,
      "grad_norm": 0.5385612538260942,
      "learning_rate": 1.3087564672094243e-05,
      "loss": 11.7478,
      "step": 30861
    },
    {
      "epoch": 1.6805592425449074,
      "grad_norm": 0.5374740654419663,
      "learning_rate": 1.3083203555709677e-05,
      "loss": 11.7969,
      "step": 30862
    },
    {
      "epoch": 1.6806136965414904,
      "grad_norm": 0.527415784310097,
      "learning_rate": 1.3078843115204709e-05,
      "loss": 11.71,
      "step": 30863
    },
    {
      "epoch": 1.6806681505380734,
      "grad_norm": 0.5145511706149598,
      "learning_rate": 1.3074483350613209e-05,
      "loss": 11.73,
      "step": 30864
    },
    {
      "epoch": 1.6807226045346566,
      "grad_norm": 0.5551752795785434,
      "learning_rate": 1.307012426196912e-05,
      "loss": 11.866,
      "step": 30865
    },
    {
      "epoch": 1.6807770585312396,
      "grad_norm": 0.5474346226968603,
      "learning_rate": 1.3065765849306311e-05,
      "loss": 11.8227,
      "step": 30866
    },
    {
      "epoch": 1.6808315125278226,
      "grad_norm": 0.6086897915503154,
      "learning_rate": 1.3061408112658703e-05,
      "loss": 11.8433,
      "step": 30867
    },
    {
      "epoch": 1.6808859665244056,
      "grad_norm": 0.5770849409133813,
      "learning_rate": 1.3057051052060143e-05,
      "loss": 11.8486,
      "step": 30868
    },
    {
      "epoch": 1.6809404205209886,
      "grad_norm": 0.5439636611188174,
      "learning_rate": 1.3052694667544551e-05,
      "loss": 11.8306,
      "step": 30869
    },
    {
      "epoch": 1.6809948745175716,
      "grad_norm": 0.5498289907417206,
      "learning_rate": 1.3048338959145766e-05,
      "loss": 11.723,
      "step": 30870
    },
    {
      "epoch": 1.6810493285141548,
      "grad_norm": 0.5625196472309748,
      "learning_rate": 1.3043983926897684e-05,
      "loss": 11.8165,
      "step": 30871
    },
    {
      "epoch": 1.6811037825107378,
      "grad_norm": 0.5064244434293584,
      "learning_rate": 1.303962957083419e-05,
      "loss": 11.7715,
      "step": 30872
    },
    {
      "epoch": 1.6811582365073208,
      "grad_norm": 0.5110432577084042,
      "learning_rate": 1.3035275890989107e-05,
      "loss": 11.6278,
      "step": 30873
    },
    {
      "epoch": 1.6812126905039038,
      "grad_norm": 0.5213051899119718,
      "learning_rate": 1.3030922887396291e-05,
      "loss": 11.785,
      "step": 30874
    },
    {
      "epoch": 1.6812671445004868,
      "grad_norm": 0.5560173869588457,
      "learning_rate": 1.3026570560089624e-05,
      "loss": 11.6741,
      "step": 30875
    },
    {
      "epoch": 1.6813215984970697,
      "grad_norm": 0.5389905003713072,
      "learning_rate": 1.3022218909102901e-05,
      "loss": 11.8545,
      "step": 30876
    },
    {
      "epoch": 1.6813760524936527,
      "grad_norm": 0.5600843065563837,
      "learning_rate": 1.3017867934470018e-05,
      "loss": 11.7851,
      "step": 30877
    },
    {
      "epoch": 1.6814305064902357,
      "grad_norm": 0.5502635265922394,
      "learning_rate": 1.301351763622476e-05,
      "loss": 11.8169,
      "step": 30878
    },
    {
      "epoch": 1.6814849604868187,
      "grad_norm": 0.5284691962359876,
      "learning_rate": 1.3009168014400997e-05,
      "loss": 11.7361,
      "step": 30879
    },
    {
      "epoch": 1.6815394144834017,
      "grad_norm": 0.4789944056438769,
      "learning_rate": 1.3004819069032514e-05,
      "loss": 11.721,
      "step": 30880
    },
    {
      "epoch": 1.6815938684799847,
      "grad_norm": 0.5517250120970683,
      "learning_rate": 1.3000470800153141e-05,
      "loss": 11.7812,
      "step": 30881
    },
    {
      "epoch": 1.6816483224765677,
      "grad_norm": 0.5518930652654063,
      "learning_rate": 1.2996123207796718e-05,
      "loss": 11.8482,
      "step": 30882
    },
    {
      "epoch": 1.6817027764731507,
      "grad_norm": 0.5485477433553917,
      "learning_rate": 1.2991776291997037e-05,
      "loss": 11.802,
      "step": 30883
    },
    {
      "epoch": 1.6817572304697337,
      "grad_norm": 0.5263391224168971,
      "learning_rate": 1.2987430052787885e-05,
      "loss": 11.6938,
      "step": 30884
    },
    {
      "epoch": 1.6818116844663167,
      "grad_norm": 0.5706755975080879,
      "learning_rate": 1.2983084490203056e-05,
      "loss": 11.8174,
      "step": 30885
    },
    {
      "epoch": 1.6818661384628997,
      "grad_norm": 0.5285214526502554,
      "learning_rate": 1.2978739604276357e-05,
      "loss": 11.7778,
      "step": 30886
    },
    {
      "epoch": 1.6819205924594827,
      "grad_norm": 0.5909773457621915,
      "learning_rate": 1.2974395395041595e-05,
      "loss": 11.8949,
      "step": 30887
    },
    {
      "epoch": 1.6819750464560659,
      "grad_norm": 0.5747792388283205,
      "learning_rate": 1.2970051862532496e-05,
      "loss": 11.8241,
      "step": 30888
    },
    {
      "epoch": 1.6820295004526489,
      "grad_norm": 0.5925829241717875,
      "learning_rate": 1.29657090067829e-05,
      "loss": 11.8375,
      "step": 30889
    },
    {
      "epoch": 1.6820839544492319,
      "grad_norm": 0.5220072443568061,
      "learning_rate": 1.2961366827826515e-05,
      "loss": 11.8994,
      "step": 30890
    },
    {
      "epoch": 1.6821384084458149,
      "grad_norm": 0.5577779274779051,
      "learning_rate": 1.295702532569717e-05,
      "loss": 11.8704,
      "step": 30891
    },
    {
      "epoch": 1.6821928624423979,
      "grad_norm": 0.5051562087141184,
      "learning_rate": 1.2952684500428558e-05,
      "loss": 11.7841,
      "step": 30892
    },
    {
      "epoch": 1.6822473164389808,
      "grad_norm": 0.5518716408252478,
      "learning_rate": 1.294834435205451e-05,
      "loss": 11.7912,
      "step": 30893
    },
    {
      "epoch": 1.682301770435564,
      "grad_norm": 0.5066884801095428,
      "learning_rate": 1.2944004880608718e-05,
      "loss": 11.7375,
      "step": 30894
    },
    {
      "epoch": 1.682356224432147,
      "grad_norm": 0.5492426517534502,
      "learning_rate": 1.2939666086124936e-05,
      "loss": 11.8867,
      "step": 30895
    },
    {
      "epoch": 1.68241067842873,
      "grad_norm": 0.5693621790149476,
      "learning_rate": 1.29353279686369e-05,
      "loss": 11.8318,
      "step": 30896
    },
    {
      "epoch": 1.682465132425313,
      "grad_norm": 0.5165803167555809,
      "learning_rate": 1.2930990528178377e-05,
      "loss": 11.8264,
      "step": 30897
    },
    {
      "epoch": 1.682519586421896,
      "grad_norm": 0.5596161858222217,
      "learning_rate": 1.2926653764783047e-05,
      "loss": 11.8619,
      "step": 30898
    },
    {
      "epoch": 1.682574040418479,
      "grad_norm": 0.547827742362886,
      "learning_rate": 1.2922317678484697e-05,
      "loss": 11.7555,
      "step": 30899
    },
    {
      "epoch": 1.682628494415062,
      "grad_norm": 0.535879415983803,
      "learning_rate": 1.2917982269316975e-05,
      "loss": 11.7463,
      "step": 30900
    },
    {
      "epoch": 1.682682948411645,
      "grad_norm": 0.535413866828367,
      "learning_rate": 1.2913647537313644e-05,
      "loss": 11.8689,
      "step": 30901
    },
    {
      "epoch": 1.682737402408228,
      "grad_norm": 0.5805696529246693,
      "learning_rate": 1.2909313482508379e-05,
      "loss": 11.9161,
      "step": 30902
    },
    {
      "epoch": 1.682791856404811,
      "grad_norm": 0.5260264998409206,
      "learning_rate": 1.2904980104934917e-05,
      "loss": 11.8405,
      "step": 30903
    },
    {
      "epoch": 1.682846310401394,
      "grad_norm": 0.5854381143859504,
      "learning_rate": 1.2900647404626943e-05,
      "loss": 11.792,
      "step": 30904
    },
    {
      "epoch": 1.682900764397977,
      "grad_norm": 0.5253860655671488,
      "learning_rate": 1.2896315381618107e-05,
      "loss": 11.7813,
      "step": 30905
    },
    {
      "epoch": 1.68295521839456,
      "grad_norm": 0.554908637873242,
      "learning_rate": 1.2891984035942162e-05,
      "loss": 11.7716,
      "step": 30906
    },
    {
      "epoch": 1.683009672391143,
      "grad_norm": 0.5782336746781361,
      "learning_rate": 1.2887653367632735e-05,
      "loss": 11.6716,
      "step": 30907
    },
    {
      "epoch": 1.683064126387726,
      "grad_norm": 0.5665593192445647,
      "learning_rate": 1.2883323376723521e-05,
      "loss": 11.7545,
      "step": 30908
    },
    {
      "epoch": 1.683118580384309,
      "grad_norm": 0.6995263762196952,
      "learning_rate": 1.2878994063248218e-05,
      "loss": 11.9942,
      "step": 30909
    },
    {
      "epoch": 1.683173034380892,
      "grad_norm": 0.535471529613836,
      "learning_rate": 1.287466542724044e-05,
      "loss": 11.6526,
      "step": 30910
    },
    {
      "epoch": 1.6832274883774752,
      "grad_norm": 0.48303961442288756,
      "learning_rate": 1.2870337468733895e-05,
      "loss": 11.7168,
      "step": 30911
    },
    {
      "epoch": 1.6832819423740581,
      "grad_norm": 0.5542509936891705,
      "learning_rate": 1.28660101877622e-05,
      "loss": 11.8746,
      "step": 30912
    },
    {
      "epoch": 1.6833363963706411,
      "grad_norm": 0.5820860719517761,
      "learning_rate": 1.2861683584359041e-05,
      "loss": 11.7746,
      "step": 30913
    },
    {
      "epoch": 1.6833908503672241,
      "grad_norm": 0.6270610000900988,
      "learning_rate": 1.2857357658558056e-05,
      "loss": 11.691,
      "step": 30914
    },
    {
      "epoch": 1.6834453043638071,
      "grad_norm": 0.5736399326213332,
      "learning_rate": 1.285303241039284e-05,
      "loss": 11.8652,
      "step": 30915
    },
    {
      "epoch": 1.6834997583603901,
      "grad_norm": 0.5619805326157901,
      "learning_rate": 1.2848707839897078e-05,
      "loss": 11.7971,
      "step": 30916
    },
    {
      "epoch": 1.6835542123569733,
      "grad_norm": 0.479098233802878,
      "learning_rate": 1.2844383947104354e-05,
      "loss": 11.7201,
      "step": 30917
    },
    {
      "epoch": 1.6836086663535563,
      "grad_norm": 0.5459321419597545,
      "learning_rate": 1.2840060732048353e-05,
      "loss": 11.8616,
      "step": 30918
    },
    {
      "epoch": 1.6836631203501393,
      "grad_norm": 0.6303937182194855,
      "learning_rate": 1.2835738194762626e-05,
      "loss": 11.7714,
      "step": 30919
    },
    {
      "epoch": 1.6837175743467223,
      "grad_norm": 0.6063994481799019,
      "learning_rate": 1.283141633528081e-05,
      "loss": 11.8242,
      "step": 30920
    },
    {
      "epoch": 1.6837720283433053,
      "grad_norm": 0.5507727194406049,
      "learning_rate": 1.2827095153636548e-05,
      "loss": 11.8543,
      "step": 30921
    },
    {
      "epoch": 1.6838264823398883,
      "grad_norm": 0.5404127369170493,
      "learning_rate": 1.282277464986339e-05,
      "loss": 11.8502,
      "step": 30922
    },
    {
      "epoch": 1.6838809363364713,
      "grad_norm": 0.522691500546519,
      "learning_rate": 1.2818454823994974e-05,
      "loss": 11.7252,
      "step": 30923
    },
    {
      "epoch": 1.6839353903330543,
      "grad_norm": 0.5330287746979502,
      "learning_rate": 1.2814135676064887e-05,
      "loss": 11.8818,
      "step": 30924
    },
    {
      "epoch": 1.6839898443296373,
      "grad_norm": 0.5324819084756363,
      "learning_rate": 1.2809817206106667e-05,
      "loss": 11.7668,
      "step": 30925
    },
    {
      "epoch": 1.6840442983262203,
      "grad_norm": 0.5679799851672759,
      "learning_rate": 1.2805499414153954e-05,
      "loss": 11.7543,
      "step": 30926
    },
    {
      "epoch": 1.6840987523228033,
      "grad_norm": 0.5641950613952643,
      "learning_rate": 1.2801182300240277e-05,
      "loss": 11.8477,
      "step": 30927
    },
    {
      "epoch": 1.6841532063193863,
      "grad_norm": 0.5684721307222901,
      "learning_rate": 1.2796865864399254e-05,
      "loss": 11.9336,
      "step": 30928
    },
    {
      "epoch": 1.6842076603159692,
      "grad_norm": 0.5958884951547424,
      "learning_rate": 1.2792550106664415e-05,
      "loss": 11.8898,
      "step": 30929
    },
    {
      "epoch": 1.6842621143125522,
      "grad_norm": 0.5768648008155474,
      "learning_rate": 1.2788235027069318e-05,
      "loss": 11.8193,
      "step": 30930
    },
    {
      "epoch": 1.6843165683091352,
      "grad_norm": 0.5808263521474171,
      "learning_rate": 1.278392062564755e-05,
      "loss": 11.8038,
      "step": 30931
    },
    {
      "epoch": 1.6843710223057182,
      "grad_norm": 0.5670701911435472,
      "learning_rate": 1.2779606902432627e-05,
      "loss": 11.7317,
      "step": 30932
    },
    {
      "epoch": 1.6844254763023012,
      "grad_norm": 0.5128089869119639,
      "learning_rate": 1.2775293857458148e-05,
      "loss": 11.6964,
      "step": 30933
    },
    {
      "epoch": 1.6844799302988842,
      "grad_norm": 0.48434730096323336,
      "learning_rate": 1.2770981490757572e-05,
      "loss": 11.6702,
      "step": 30934
    },
    {
      "epoch": 1.6845343842954674,
      "grad_norm": 0.5543601897213132,
      "learning_rate": 1.2766669802364473e-05,
      "loss": 11.5906,
      "step": 30935
    },
    {
      "epoch": 1.6845888382920504,
      "grad_norm": 0.5149236057309247,
      "learning_rate": 1.276235879231239e-05,
      "loss": 11.8646,
      "step": 30936
    },
    {
      "epoch": 1.6846432922886334,
      "grad_norm": 0.5751090349175441,
      "learning_rate": 1.275804846063482e-05,
      "loss": 11.8037,
      "step": 30937
    },
    {
      "epoch": 1.6846977462852164,
      "grad_norm": 0.5163364506804589,
      "learning_rate": 1.2753738807365322e-05,
      "loss": 11.7896,
      "step": 30938
    },
    {
      "epoch": 1.6847522002817994,
      "grad_norm": 0.557297852222947,
      "learning_rate": 1.2749429832537352e-05,
      "loss": 11.8165,
      "step": 30939
    },
    {
      "epoch": 1.6848066542783826,
      "grad_norm": 0.5754185383229934,
      "learning_rate": 1.2745121536184468e-05,
      "loss": 11.8465,
      "step": 30940
    },
    {
      "epoch": 1.6848611082749656,
      "grad_norm": 0.5189316542291988,
      "learning_rate": 1.2740813918340144e-05,
      "loss": 11.6276,
      "step": 30941
    },
    {
      "epoch": 1.6849155622715486,
      "grad_norm": 0.5397203483892539,
      "learning_rate": 1.2736506979037866e-05,
      "loss": 11.814,
      "step": 30942
    },
    {
      "epoch": 1.6849700162681316,
      "grad_norm": 0.5248558922205516,
      "learning_rate": 1.2732200718311183e-05,
      "loss": 11.655,
      "step": 30943
    },
    {
      "epoch": 1.6850244702647146,
      "grad_norm": 0.569772147300932,
      "learning_rate": 1.2727895136193535e-05,
      "loss": 11.8354,
      "step": 30944
    },
    {
      "epoch": 1.6850789242612976,
      "grad_norm": 0.5107304223006778,
      "learning_rate": 1.2723590232718386e-05,
      "loss": 11.7712,
      "step": 30945
    },
    {
      "epoch": 1.6851333782578806,
      "grad_norm": 0.5517829892272162,
      "learning_rate": 1.2719286007919252e-05,
      "loss": 11.853,
      "step": 30946
    },
    {
      "epoch": 1.6851878322544636,
      "grad_norm": 0.5516534462677495,
      "learning_rate": 1.2714982461829571e-05,
      "loss": 11.9019,
      "step": 30947
    },
    {
      "epoch": 1.6852422862510466,
      "grad_norm": 0.5973626906897074,
      "learning_rate": 1.2710679594482855e-05,
      "loss": 11.8328,
      "step": 30948
    },
    {
      "epoch": 1.6852967402476295,
      "grad_norm": 0.486708149672646,
      "learning_rate": 1.2706377405912496e-05,
      "loss": 11.7596,
      "step": 30949
    },
    {
      "epoch": 1.6853511942442125,
      "grad_norm": 0.5392361116153961,
      "learning_rate": 1.270207589615201e-05,
      "loss": 11.8083,
      "step": 30950
    },
    {
      "epoch": 1.6854056482407955,
      "grad_norm": 0.6011225344433534,
      "learning_rate": 1.2697775065234807e-05,
      "loss": 11.8283,
      "step": 30951
    },
    {
      "epoch": 1.6854601022373785,
      "grad_norm": 0.4660341428272207,
      "learning_rate": 1.2693474913194347e-05,
      "loss": 11.7044,
      "step": 30952
    },
    {
      "epoch": 1.6855145562339615,
      "grad_norm": 0.5744173327336871,
      "learning_rate": 1.2689175440064083e-05,
      "loss": 11.8485,
      "step": 30953
    },
    {
      "epoch": 1.6855690102305445,
      "grad_norm": 0.49386697404015156,
      "learning_rate": 1.268487664587743e-05,
      "loss": 11.8396,
      "step": 30954
    },
    {
      "epoch": 1.6856234642271275,
      "grad_norm": 0.5649774813196295,
      "learning_rate": 1.268057853066783e-05,
      "loss": 11.7956,
      "step": 30955
    },
    {
      "epoch": 1.6856779182237105,
      "grad_norm": 0.5066835993067511,
      "learning_rate": 1.2676281094468667e-05,
      "loss": 11.9316,
      "step": 30956
    },
    {
      "epoch": 1.6857323722202935,
      "grad_norm": 0.5387885700361347,
      "learning_rate": 1.2671984337313381e-05,
      "loss": 11.809,
      "step": 30957
    },
    {
      "epoch": 1.6857868262168767,
      "grad_norm": 0.6158151256337171,
      "learning_rate": 1.266768825923541e-05,
      "loss": 11.7484,
      "step": 30958
    },
    {
      "epoch": 1.6858412802134597,
      "grad_norm": 0.5642077699526209,
      "learning_rate": 1.2663392860268131e-05,
      "loss": 11.7662,
      "step": 30959
    },
    {
      "epoch": 1.6858957342100427,
      "grad_norm": 0.5004019979844113,
      "learning_rate": 1.265909814044497e-05,
      "loss": 11.8103,
      "step": 30960
    },
    {
      "epoch": 1.6859501882066257,
      "grad_norm": 0.5326111269189654,
      "learning_rate": 1.265480409979929e-05,
      "loss": 11.7948,
      "step": 30961
    },
    {
      "epoch": 1.6860046422032087,
      "grad_norm": 0.503972500382625,
      "learning_rate": 1.2650510738364518e-05,
      "loss": 11.6654,
      "step": 30962
    },
    {
      "epoch": 1.6860590961997917,
      "grad_norm": 0.5222908121815988,
      "learning_rate": 1.2646218056174009e-05,
      "loss": 11.9125,
      "step": 30963
    },
    {
      "epoch": 1.6861135501963749,
      "grad_norm": 0.5209559935917166,
      "learning_rate": 1.2641926053261177e-05,
      "loss": 11.6175,
      "step": 30964
    },
    {
      "epoch": 1.6861680041929579,
      "grad_norm": 0.5519284255939755,
      "learning_rate": 1.2637634729659375e-05,
      "loss": 11.7419,
      "step": 30965
    },
    {
      "epoch": 1.6862224581895409,
      "grad_norm": 0.5532925618193903,
      "learning_rate": 1.2633344085401955e-05,
      "loss": 11.776,
      "step": 30966
    },
    {
      "epoch": 1.6862769121861239,
      "grad_norm": 0.5466818222668209,
      "learning_rate": 1.262905412052232e-05,
      "loss": 11.7007,
      "step": 30967
    },
    {
      "epoch": 1.6863313661827068,
      "grad_norm": 0.5760458209075585,
      "learning_rate": 1.2624764835053804e-05,
      "loss": 11.7261,
      "step": 30968
    },
    {
      "epoch": 1.6863858201792898,
      "grad_norm": 0.5607843698283759,
      "learning_rate": 1.2620476229029753e-05,
      "loss": 11.7256,
      "step": 30969
    },
    {
      "epoch": 1.6864402741758728,
      "grad_norm": 0.5876410562875248,
      "learning_rate": 1.2616188302483567e-05,
      "loss": 11.7445,
      "step": 30970
    },
    {
      "epoch": 1.6864947281724558,
      "grad_norm": 0.5333042910383061,
      "learning_rate": 1.2611901055448528e-05,
      "loss": 11.7903,
      "step": 30971
    },
    {
      "epoch": 1.6865491821690388,
      "grad_norm": 0.5375697937719466,
      "learning_rate": 1.260761448795802e-05,
      "loss": 11.795,
      "step": 30972
    },
    {
      "epoch": 1.6866036361656218,
      "grad_norm": 0.5320807055716856,
      "learning_rate": 1.2603328600045338e-05,
      "loss": 11.7174,
      "step": 30973
    },
    {
      "epoch": 1.6866580901622048,
      "grad_norm": 0.5754043768594431,
      "learning_rate": 1.259904339174386e-05,
      "loss": 11.7903,
      "step": 30974
    },
    {
      "epoch": 1.6867125441587878,
      "grad_norm": 0.5422487035895022,
      "learning_rate": 1.2594758863086865e-05,
      "loss": 11.8148,
      "step": 30975
    },
    {
      "epoch": 1.6867669981553708,
      "grad_norm": 0.5434807244508277,
      "learning_rate": 1.2590475014107661e-05,
      "loss": 11.8619,
      "step": 30976
    },
    {
      "epoch": 1.6868214521519538,
      "grad_norm": 0.5788963942280565,
      "learning_rate": 1.2586191844839612e-05,
      "loss": 11.8867,
      "step": 30977
    },
    {
      "epoch": 1.6868759061485368,
      "grad_norm": 0.6179581524250197,
      "learning_rate": 1.2581909355315968e-05,
      "loss": 11.7265,
      "step": 30978
    },
    {
      "epoch": 1.6869303601451198,
      "grad_norm": 0.6156278283170895,
      "learning_rate": 1.2577627545570059e-05,
      "loss": 11.7749,
      "step": 30979
    },
    {
      "epoch": 1.6869848141417028,
      "grad_norm": 0.5364324405623805,
      "learning_rate": 1.2573346415635201e-05,
      "loss": 11.7738,
      "step": 30980
    },
    {
      "epoch": 1.687039268138286,
      "grad_norm": 0.6067647389487533,
      "learning_rate": 1.2569065965544636e-05,
      "loss": 11.8974,
      "step": 30981
    },
    {
      "epoch": 1.687093722134869,
      "grad_norm": 0.5466475366529785,
      "learning_rate": 1.2564786195331702e-05,
      "loss": 11.771,
      "step": 30982
    },
    {
      "epoch": 1.687148176131452,
      "grad_norm": 0.5538879810462987,
      "learning_rate": 1.256050710502964e-05,
      "loss": 11.8921,
      "step": 30983
    },
    {
      "epoch": 1.687202630128035,
      "grad_norm": 0.7236317356514967,
      "learning_rate": 1.2556228694671746e-05,
      "loss": 11.8492,
      "step": 30984
    },
    {
      "epoch": 1.687257084124618,
      "grad_norm": 0.5389627890915959,
      "learning_rate": 1.2551950964291292e-05,
      "loss": 11.7868,
      "step": 30985
    },
    {
      "epoch": 1.687311538121201,
      "grad_norm": 0.5477972562887624,
      "learning_rate": 1.2547673913921499e-05,
      "loss": 11.661,
      "step": 30986
    },
    {
      "epoch": 1.6873659921177842,
      "grad_norm": 0.5549818238153731,
      "learning_rate": 1.2543397543595692e-05,
      "loss": 11.9308,
      "step": 30987
    },
    {
      "epoch": 1.6874204461143671,
      "grad_norm": 0.5392778514406316,
      "learning_rate": 1.2539121853347069e-05,
      "loss": 11.6602,
      "step": 30988
    },
    {
      "epoch": 1.6874749001109501,
      "grad_norm": 0.5639723548899724,
      "learning_rate": 1.2534846843208925e-05,
      "loss": 11.7974,
      "step": 30989
    },
    {
      "epoch": 1.6875293541075331,
      "grad_norm": 0.5938283471144208,
      "learning_rate": 1.2530572513214446e-05,
      "loss": 11.8409,
      "step": 30990
    },
    {
      "epoch": 1.6875838081041161,
      "grad_norm": 0.5400408892509663,
      "learning_rate": 1.2526298863396912e-05,
      "loss": 11.7953,
      "step": 30991
    },
    {
      "epoch": 1.6876382621006991,
      "grad_norm": 0.6553367130312735,
      "learning_rate": 1.2522025893789569e-05,
      "loss": 11.8586,
      "step": 30992
    },
    {
      "epoch": 1.687692716097282,
      "grad_norm": 0.5834404463377707,
      "learning_rate": 1.2517753604425608e-05,
      "loss": 11.7478,
      "step": 30993
    },
    {
      "epoch": 1.687747170093865,
      "grad_norm": 0.57803278579205,
      "learning_rate": 1.2513481995338284e-05,
      "loss": 11.8328,
      "step": 30994
    },
    {
      "epoch": 1.687801624090448,
      "grad_norm": 0.6286824435063781,
      "learning_rate": 1.2509211066560788e-05,
      "loss": 11.813,
      "step": 30995
    },
    {
      "epoch": 1.687856078087031,
      "grad_norm": 0.5277799842411208,
      "learning_rate": 1.250494081812632e-05,
      "loss": 11.7461,
      "step": 30996
    },
    {
      "epoch": 1.687910532083614,
      "grad_norm": 0.528310544407678,
      "learning_rate": 1.250067125006813e-05,
      "loss": 11.6591,
      "step": 30997
    },
    {
      "epoch": 1.687964986080197,
      "grad_norm": 0.5240383164639838,
      "learning_rate": 1.2496402362419369e-05,
      "loss": 11.8203,
      "step": 30998
    },
    {
      "epoch": 1.68801944007678,
      "grad_norm": 0.5499002892758581,
      "learning_rate": 1.2492134155213275e-05,
      "loss": 11.6102,
      "step": 30999
    },
    {
      "epoch": 1.688073894073363,
      "grad_norm": 0.5024318789840799,
      "learning_rate": 1.2487866628483014e-05,
      "loss": 11.7725,
      "step": 31000
    },
    {
      "epoch": 1.688128348069946,
      "grad_norm": 0.5838408608040243,
      "learning_rate": 1.2483599782261768e-05,
      "loss": 11.7318,
      "step": 31001
    },
    {
      "epoch": 1.688182802066529,
      "grad_norm": 0.6098789562032324,
      "learning_rate": 1.2479333616582744e-05,
      "loss": 11.8629,
      "step": 31002
    },
    {
      "epoch": 1.688237256063112,
      "grad_norm": 0.5656585730421902,
      "learning_rate": 1.2475068131479084e-05,
      "loss": 11.792,
      "step": 31003
    },
    {
      "epoch": 1.688291710059695,
      "grad_norm": 0.5392994469720828,
      "learning_rate": 1.2470803326984016e-05,
      "loss": 11.7874,
      "step": 31004
    },
    {
      "epoch": 1.6883461640562782,
      "grad_norm": 0.5252149296259485,
      "learning_rate": 1.2466539203130612e-05,
      "loss": 11.784,
      "step": 31005
    },
    {
      "epoch": 1.6884006180528612,
      "grad_norm": 0.507835131305577,
      "learning_rate": 1.246227575995208e-05,
      "loss": 11.7961,
      "step": 31006
    },
    {
      "epoch": 1.6884550720494442,
      "grad_norm": 0.5172020169125301,
      "learning_rate": 1.2458012997481594e-05,
      "loss": 11.8493,
      "step": 31007
    },
    {
      "epoch": 1.6885095260460272,
      "grad_norm": 0.580931403383934,
      "learning_rate": 1.245375091575226e-05,
      "loss": 11.8756,
      "step": 31008
    },
    {
      "epoch": 1.6885639800426102,
      "grad_norm": 0.5486886961906027,
      "learning_rate": 1.2449489514797264e-05,
      "loss": 11.7934,
      "step": 31009
    },
    {
      "epoch": 1.6886184340391934,
      "grad_norm": 0.47751340387962543,
      "learning_rate": 1.24452287946497e-05,
      "loss": 11.7613,
      "step": 31010
    },
    {
      "epoch": 1.6886728880357764,
      "grad_norm": 0.6112902242248104,
      "learning_rate": 1.244096875534273e-05,
      "loss": 11.7245,
      "step": 31011
    },
    {
      "epoch": 1.6887273420323594,
      "grad_norm": 0.5524698751702133,
      "learning_rate": 1.2436709396909451e-05,
      "loss": 11.817,
      "step": 31012
    },
    {
      "epoch": 1.6887817960289424,
      "grad_norm": 0.5503354050461463,
      "learning_rate": 1.2432450719383015e-05,
      "loss": 11.8147,
      "step": 31013
    },
    {
      "epoch": 1.6888362500255254,
      "grad_norm": 0.5508965272505504,
      "learning_rate": 1.242819272279656e-05,
      "loss": 11.6423,
      "step": 31014
    },
    {
      "epoch": 1.6888907040221084,
      "grad_norm": 0.5367811120380197,
      "learning_rate": 1.242393540718313e-05,
      "loss": 11.8595,
      "step": 31015
    },
    {
      "epoch": 1.6889451580186914,
      "grad_norm": 0.5561738870650693,
      "learning_rate": 1.2419678772575882e-05,
      "loss": 11.8342,
      "step": 31016
    },
    {
      "epoch": 1.6889996120152744,
      "grad_norm": 0.5301874112024781,
      "learning_rate": 1.2415422819007871e-05,
      "loss": 11.8866,
      "step": 31017
    },
    {
      "epoch": 1.6890540660118574,
      "grad_norm": 0.6467900675327487,
      "learning_rate": 1.2411167546512226e-05,
      "loss": 11.9131,
      "step": 31018
    },
    {
      "epoch": 1.6891085200084404,
      "grad_norm": 0.5411147604998521,
      "learning_rate": 1.2406912955122052e-05,
      "loss": 11.9355,
      "step": 31019
    },
    {
      "epoch": 1.6891629740050234,
      "grad_norm": 0.47049751414523905,
      "learning_rate": 1.240265904487039e-05,
      "loss": 11.7164,
      "step": 31020
    },
    {
      "epoch": 1.6892174280016063,
      "grad_norm": 0.5459693348498853,
      "learning_rate": 1.2398405815790371e-05,
      "loss": 11.8325,
      "step": 31021
    },
    {
      "epoch": 1.6892718819981893,
      "grad_norm": 0.5441759555883625,
      "learning_rate": 1.2394153267915009e-05,
      "loss": 11.8041,
      "step": 31022
    },
    {
      "epoch": 1.6893263359947723,
      "grad_norm": 0.5060146303218898,
      "learning_rate": 1.2389901401277426e-05,
      "loss": 11.7788,
      "step": 31023
    },
    {
      "epoch": 1.6893807899913553,
      "grad_norm": 0.6361849047081553,
      "learning_rate": 1.238565021591066e-05,
      "loss": 11.8789,
      "step": 31024
    },
    {
      "epoch": 1.6894352439879383,
      "grad_norm": 0.5062576448789375,
      "learning_rate": 1.2381399711847751e-05,
      "loss": 11.7049,
      "step": 31025
    },
    {
      "epoch": 1.6894896979845213,
      "grad_norm": 0.554103616111281,
      "learning_rate": 1.2377149889121797e-05,
      "loss": 11.7653,
      "step": 31026
    },
    {
      "epoch": 1.6895441519811043,
      "grad_norm": 0.4929483825124844,
      "learning_rate": 1.2372900747765791e-05,
      "loss": 11.6667,
      "step": 31027
    },
    {
      "epoch": 1.6895986059776875,
      "grad_norm": 0.5440798486434106,
      "learning_rate": 1.236865228781281e-05,
      "loss": 11.8324,
      "step": 31028
    },
    {
      "epoch": 1.6896530599742705,
      "grad_norm": 0.565908417010825,
      "learning_rate": 1.2364404509295902e-05,
      "loss": 11.7977,
      "step": 31029
    },
    {
      "epoch": 1.6897075139708535,
      "grad_norm": 0.5461142968907308,
      "learning_rate": 1.2360157412248064e-05,
      "loss": 11.7253,
      "step": 31030
    },
    {
      "epoch": 1.6897619679674365,
      "grad_norm": 0.5625146515850067,
      "learning_rate": 1.2355910996702358e-05,
      "loss": 11.9029,
      "step": 31031
    },
    {
      "epoch": 1.6898164219640195,
      "grad_norm": 0.4987479890051444,
      "learning_rate": 1.2351665262691758e-05,
      "loss": 11.644,
      "step": 31032
    },
    {
      "epoch": 1.6898708759606025,
      "grad_norm": 0.5498233010714532,
      "learning_rate": 1.2347420210249339e-05,
      "loss": 11.7239,
      "step": 31033
    },
    {
      "epoch": 1.6899253299571857,
      "grad_norm": 0.579645915344141,
      "learning_rate": 1.2343175839408072e-05,
      "loss": 11.8525,
      "step": 31034
    },
    {
      "epoch": 1.6899797839537687,
      "grad_norm": 0.5109973452372204,
      "learning_rate": 1.2338932150200955e-05,
      "loss": 11.7003,
      "step": 31035
    },
    {
      "epoch": 1.6900342379503517,
      "grad_norm": 0.5250080212971248,
      "learning_rate": 1.2334689142661015e-05,
      "loss": 11.7213,
      "step": 31036
    },
    {
      "epoch": 1.6900886919469347,
      "grad_norm": 0.5140151033662399,
      "learning_rate": 1.2330446816821217e-05,
      "loss": 11.7804,
      "step": 31037
    },
    {
      "epoch": 1.6901431459435177,
      "grad_norm": 0.5814524810375874,
      "learning_rate": 1.232620517271459e-05,
      "loss": 11.7545,
      "step": 31038
    },
    {
      "epoch": 1.6901975999401007,
      "grad_norm": 0.5844885144223078,
      "learning_rate": 1.2321964210374081e-05,
      "loss": 11.8049,
      "step": 31039
    },
    {
      "epoch": 1.6902520539366837,
      "grad_norm": 0.5273357932651959,
      "learning_rate": 1.231772392983267e-05,
      "loss": 11.825,
      "step": 31040
    },
    {
      "epoch": 1.6903065079332666,
      "grad_norm": 0.5230934720052108,
      "learning_rate": 1.2313484331123371e-05,
      "loss": 11.7785,
      "step": 31041
    },
    {
      "epoch": 1.6903609619298496,
      "grad_norm": 0.5066650193755428,
      "learning_rate": 1.2309245414279092e-05,
      "loss": 11.6719,
      "step": 31042
    },
    {
      "epoch": 1.6904154159264326,
      "grad_norm": 0.48386011572223003,
      "learning_rate": 1.230500717933285e-05,
      "loss": 11.7318,
      "step": 31043
    },
    {
      "epoch": 1.6904698699230156,
      "grad_norm": 0.5816903007676951,
      "learning_rate": 1.2300769626317587e-05,
      "loss": 11.7068,
      "step": 31044
    },
    {
      "epoch": 1.6905243239195986,
      "grad_norm": 0.5322336672555599,
      "learning_rate": 1.2296532755266222e-05,
      "loss": 11.8342,
      "step": 31045
    },
    {
      "epoch": 1.6905787779161816,
      "grad_norm": 0.5174156072071366,
      "learning_rate": 1.2292296566211737e-05,
      "loss": 11.7864,
      "step": 31046
    },
    {
      "epoch": 1.6906332319127646,
      "grad_norm": 0.6314504146664143,
      "learning_rate": 1.228806105918705e-05,
      "loss": 11.8131,
      "step": 31047
    },
    {
      "epoch": 1.6906876859093476,
      "grad_norm": 0.5293592883679874,
      "learning_rate": 1.2283826234225137e-05,
      "loss": 11.7786,
      "step": 31048
    },
    {
      "epoch": 1.6907421399059306,
      "grad_norm": 0.5495173292261814,
      "learning_rate": 1.2279592091358871e-05,
      "loss": 11.795,
      "step": 31049
    },
    {
      "epoch": 1.6907965939025136,
      "grad_norm": 0.5284385447491805,
      "learning_rate": 1.2275358630621214e-05,
      "loss": 11.7026,
      "step": 31050
    },
    {
      "epoch": 1.6908510478990968,
      "grad_norm": 0.5072889494758992,
      "learning_rate": 1.2271125852045095e-05,
      "loss": 11.8435,
      "step": 31051
    },
    {
      "epoch": 1.6909055018956798,
      "grad_norm": 0.541842245256738,
      "learning_rate": 1.2266893755663388e-05,
      "loss": 11.7366,
      "step": 31052
    },
    {
      "epoch": 1.6909599558922628,
      "grad_norm": 0.5077504448857384,
      "learning_rate": 1.2262662341509057e-05,
      "loss": 11.8134,
      "step": 31053
    },
    {
      "epoch": 1.6910144098888458,
      "grad_norm": 0.5042125936553743,
      "learning_rate": 1.2258431609614973e-05,
      "loss": 11.6919,
      "step": 31054
    },
    {
      "epoch": 1.6910688638854288,
      "grad_norm": 0.5543434981239634,
      "learning_rate": 1.2254201560014022e-05,
      "loss": 11.7729,
      "step": 31055
    },
    {
      "epoch": 1.6911233178820118,
      "grad_norm": 0.5241217795547016,
      "learning_rate": 1.2249972192739135e-05,
      "loss": 11.7454,
      "step": 31056
    },
    {
      "epoch": 1.691177771878595,
      "grad_norm": 0.5353220571388403,
      "learning_rate": 1.224574350782316e-05,
      "loss": 11.8802,
      "step": 31057
    },
    {
      "epoch": 1.691232225875178,
      "grad_norm": 0.5159713338830011,
      "learning_rate": 1.2241515505299016e-05,
      "loss": 11.7998,
      "step": 31058
    },
    {
      "epoch": 1.691286679871761,
      "grad_norm": 0.5840084513881622,
      "learning_rate": 1.2237288185199547e-05,
      "loss": 11.8563,
      "step": 31059
    },
    {
      "epoch": 1.691341133868344,
      "grad_norm": 0.5622320436290875,
      "learning_rate": 1.2233061547557667e-05,
      "loss": 11.7928,
      "step": 31060
    },
    {
      "epoch": 1.691395587864927,
      "grad_norm": 0.5405334827035437,
      "learning_rate": 1.2228835592406196e-05,
      "loss": 11.7847,
      "step": 31061
    },
    {
      "epoch": 1.69145004186151,
      "grad_norm": 0.5921790358345382,
      "learning_rate": 1.2224610319778018e-05,
      "loss": 11.7002,
      "step": 31062
    },
    {
      "epoch": 1.691504495858093,
      "grad_norm": 0.5404621715199007,
      "learning_rate": 1.2220385729706007e-05,
      "loss": 11.8761,
      "step": 31063
    },
    {
      "epoch": 1.691558949854676,
      "grad_norm": 0.6224685543905782,
      "learning_rate": 1.2216161822222993e-05,
      "loss": 11.8591,
      "step": 31064
    },
    {
      "epoch": 1.691613403851259,
      "grad_norm": 0.5297868569126959,
      "learning_rate": 1.2211938597361839e-05,
      "loss": 11.4864,
      "step": 31065
    },
    {
      "epoch": 1.691667857847842,
      "grad_norm": 0.7639102393028403,
      "learning_rate": 1.2207716055155349e-05,
      "loss": 11.7668,
      "step": 31066
    },
    {
      "epoch": 1.691722311844425,
      "grad_norm": 0.5075137608088227,
      "learning_rate": 1.2203494195636379e-05,
      "loss": 11.8143,
      "step": 31067
    },
    {
      "epoch": 1.691776765841008,
      "grad_norm": 0.5915385553745713,
      "learning_rate": 1.2199273018837775e-05,
      "loss": 11.8833,
      "step": 31068
    },
    {
      "epoch": 1.6918312198375909,
      "grad_norm": 0.5144594337393157,
      "learning_rate": 1.2195052524792339e-05,
      "loss": 11.7613,
      "step": 31069
    },
    {
      "epoch": 1.6918856738341739,
      "grad_norm": 0.4984695264655388,
      "learning_rate": 1.2190832713532917e-05,
      "loss": 11.7809,
      "step": 31070
    },
    {
      "epoch": 1.6919401278307569,
      "grad_norm": 0.5117894029324008,
      "learning_rate": 1.2186613585092277e-05,
      "loss": 11.7135,
      "step": 31071
    },
    {
      "epoch": 1.6919945818273399,
      "grad_norm": 0.5383639911433917,
      "learning_rate": 1.2182395139503289e-05,
      "loss": 11.8268,
      "step": 31072
    },
    {
      "epoch": 1.6920490358239229,
      "grad_norm": 0.5174906912576328,
      "learning_rate": 1.2178177376798693e-05,
      "loss": 11.8227,
      "step": 31073
    },
    {
      "epoch": 1.692103489820506,
      "grad_norm": 0.575912544748499,
      "learning_rate": 1.2173960297011344e-05,
      "loss": 11.6925,
      "step": 31074
    },
    {
      "epoch": 1.692157943817089,
      "grad_norm": 0.5235764455622189,
      "learning_rate": 1.2169743900174003e-05,
      "loss": 11.6668,
      "step": 31075
    },
    {
      "epoch": 1.692212397813672,
      "grad_norm": 0.5566910111289741,
      "learning_rate": 1.2165528186319453e-05,
      "loss": 11.8516,
      "step": 31076
    },
    {
      "epoch": 1.692266851810255,
      "grad_norm": 0.535604012565439,
      "learning_rate": 1.2161313155480469e-05,
      "loss": 11.8037,
      "step": 31077
    },
    {
      "epoch": 1.692321305806838,
      "grad_norm": 0.5247715048292585,
      "learning_rate": 1.2157098807689882e-05,
      "loss": 11.8749,
      "step": 31078
    },
    {
      "epoch": 1.692375759803421,
      "grad_norm": 0.568802881873654,
      "learning_rate": 1.2152885142980397e-05,
      "loss": 11.8276,
      "step": 31079
    },
    {
      "epoch": 1.6924302138000042,
      "grad_norm": 0.5621694856576721,
      "learning_rate": 1.2148672161384823e-05,
      "loss": 11.8172,
      "step": 31080
    },
    {
      "epoch": 1.6924846677965872,
      "grad_norm": 0.5564010752224365,
      "learning_rate": 1.2144459862935898e-05,
      "loss": 11.8701,
      "step": 31081
    },
    {
      "epoch": 1.6925391217931702,
      "grad_norm": 0.5462549488719423,
      "learning_rate": 1.2140248247666397e-05,
      "loss": 11.8552,
      "step": 31082
    },
    {
      "epoch": 1.6925935757897532,
      "grad_norm": 0.5615880949210886,
      "learning_rate": 1.213603731560905e-05,
      "loss": 11.7695,
      "step": 31083
    },
    {
      "epoch": 1.6926480297863362,
      "grad_norm": 0.5923126679584949,
      "learning_rate": 1.2131827066796608e-05,
      "loss": 11.7957,
      "step": 31084
    },
    {
      "epoch": 1.6927024837829192,
      "grad_norm": 0.5783788676073419,
      "learning_rate": 1.2127617501261845e-05,
      "loss": 11.691,
      "step": 31085
    },
    {
      "epoch": 1.6927569377795022,
      "grad_norm": 0.5420336454809602,
      "learning_rate": 1.2123408619037434e-05,
      "loss": 11.7871,
      "step": 31086
    },
    {
      "epoch": 1.6928113917760852,
      "grad_norm": 0.5781991952400373,
      "learning_rate": 1.211920042015615e-05,
      "loss": 11.773,
      "step": 31087
    },
    {
      "epoch": 1.6928658457726682,
      "grad_norm": 0.5485901242604088,
      "learning_rate": 1.2114992904650691e-05,
      "loss": 11.8089,
      "step": 31088
    },
    {
      "epoch": 1.6929202997692512,
      "grad_norm": 0.5246036213674602,
      "learning_rate": 1.211078607255377e-05,
      "loss": 11.8005,
      "step": 31089
    },
    {
      "epoch": 1.6929747537658342,
      "grad_norm": 0.651142675260831,
      "learning_rate": 1.2106579923898154e-05,
      "loss": 11.7684,
      "step": 31090
    },
    {
      "epoch": 1.6930292077624172,
      "grad_norm": 0.5184941021250798,
      "learning_rate": 1.2102374458716481e-05,
      "loss": 11.8613,
      "step": 31091
    },
    {
      "epoch": 1.6930836617590002,
      "grad_norm": 0.5544019442943551,
      "learning_rate": 1.2098169677041504e-05,
      "loss": 11.7862,
      "step": 31092
    },
    {
      "epoch": 1.6931381157555832,
      "grad_norm": 0.6027250340795689,
      "learning_rate": 1.2093965578905875e-05,
      "loss": 11.855,
      "step": 31093
    },
    {
      "epoch": 1.6931925697521661,
      "grad_norm": 0.6490606591017354,
      "learning_rate": 1.2089762164342344e-05,
      "loss": 11.853,
      "step": 31094
    },
    {
      "epoch": 1.6932470237487491,
      "grad_norm": 0.6449716678820732,
      "learning_rate": 1.2085559433383565e-05,
      "loss": 11.7931,
      "step": 31095
    },
    {
      "epoch": 1.6933014777453321,
      "grad_norm": 0.5988995642660392,
      "learning_rate": 1.2081357386062186e-05,
      "loss": 11.8962,
      "step": 31096
    },
    {
      "epoch": 1.6933559317419151,
      "grad_norm": 0.5489172557999267,
      "learning_rate": 1.2077156022410952e-05,
      "loss": 11.7662,
      "step": 31097
    },
    {
      "epoch": 1.6934103857384983,
      "grad_norm": 0.5388743072112442,
      "learning_rate": 1.2072955342462466e-05,
      "loss": 11.737,
      "step": 31098
    },
    {
      "epoch": 1.6934648397350813,
      "grad_norm": 0.5502970655509305,
      "learning_rate": 1.2068755346249427e-05,
      "loss": 11.8435,
      "step": 31099
    },
    {
      "epoch": 1.6935192937316643,
      "grad_norm": 0.5266090068269315,
      "learning_rate": 1.2064556033804508e-05,
      "loss": 11.656,
      "step": 31100
    },
    {
      "epoch": 1.6935737477282473,
      "grad_norm": 0.5333254402942086,
      "learning_rate": 1.206035740516034e-05,
      "loss": 11.8281,
      "step": 31101
    },
    {
      "epoch": 1.6936282017248303,
      "grad_norm": 0.5587803731499055,
      "learning_rate": 1.2056159460349592e-05,
      "loss": 11.842,
      "step": 31102
    },
    {
      "epoch": 1.6936826557214133,
      "grad_norm": 0.5887906236010659,
      "learning_rate": 1.2051962199404876e-05,
      "loss": 11.79,
      "step": 31103
    },
    {
      "epoch": 1.6937371097179965,
      "grad_norm": 0.5052092030785456,
      "learning_rate": 1.2047765622358875e-05,
      "loss": 11.7987,
      "step": 31104
    },
    {
      "epoch": 1.6937915637145795,
      "grad_norm": 0.5189554623571955,
      "learning_rate": 1.2043569729244198e-05,
      "loss": 11.7302,
      "step": 31105
    },
    {
      "epoch": 1.6938460177111625,
      "grad_norm": 0.536995774187871,
      "learning_rate": 1.203937452009345e-05,
      "loss": 11.9114,
      "step": 31106
    },
    {
      "epoch": 1.6939004717077455,
      "grad_norm": 0.5514921998254492,
      "learning_rate": 1.2035179994939295e-05,
      "loss": 11.733,
      "step": 31107
    },
    {
      "epoch": 1.6939549257043285,
      "grad_norm": 0.5376014961174918,
      "learning_rate": 1.2030986153814316e-05,
      "loss": 11.649,
      "step": 31108
    },
    {
      "epoch": 1.6940093797009115,
      "grad_norm": 0.6520060717512525,
      "learning_rate": 1.2026792996751157e-05,
      "loss": 11.9269,
      "step": 31109
    },
    {
      "epoch": 1.6940638336974945,
      "grad_norm": 0.5537771651695272,
      "learning_rate": 1.202260052378239e-05,
      "loss": 11.8471,
      "step": 31110
    },
    {
      "epoch": 1.6941182876940775,
      "grad_norm": 0.5484192303934142,
      "learning_rate": 1.2018408734940644e-05,
      "loss": 11.7694,
      "step": 31111
    },
    {
      "epoch": 1.6941727416906605,
      "grad_norm": 0.5503157610454154,
      "learning_rate": 1.2014217630258518e-05,
      "loss": 11.8576,
      "step": 31112
    },
    {
      "epoch": 1.6942271956872434,
      "grad_norm": 0.6120296218848519,
      "learning_rate": 1.201002720976857e-05,
      "loss": 11.7632,
      "step": 31113
    },
    {
      "epoch": 1.6942816496838264,
      "grad_norm": 0.5567956856022628,
      "learning_rate": 1.2005837473503434e-05,
      "loss": 11.9131,
      "step": 31114
    },
    {
      "epoch": 1.6943361036804094,
      "grad_norm": 0.5460710209147294,
      "learning_rate": 1.2001648421495648e-05,
      "loss": 11.6285,
      "step": 31115
    },
    {
      "epoch": 1.6943905576769924,
      "grad_norm": 0.48387214905013837,
      "learning_rate": 1.1997460053777799e-05,
      "loss": 11.6831,
      "step": 31116
    },
    {
      "epoch": 1.6944450116735754,
      "grad_norm": 0.6240826432684845,
      "learning_rate": 1.1993272370382469e-05,
      "loss": 11.9838,
      "step": 31117
    },
    {
      "epoch": 1.6944994656701584,
      "grad_norm": 0.5618401324838624,
      "learning_rate": 1.198908537134219e-05,
      "loss": 11.8195,
      "step": 31118
    },
    {
      "epoch": 1.6945539196667414,
      "grad_norm": 0.49165678973523896,
      "learning_rate": 1.1984899056689569e-05,
      "loss": 11.692,
      "step": 31119
    },
    {
      "epoch": 1.6946083736633244,
      "grad_norm": 0.5751672989876923,
      "learning_rate": 1.1980713426457102e-05,
      "loss": 11.9483,
      "step": 31120
    },
    {
      "epoch": 1.6946628276599076,
      "grad_norm": 0.5234085505286293,
      "learning_rate": 1.1976528480677396e-05,
      "loss": 11.6454,
      "step": 31121
    },
    {
      "epoch": 1.6947172816564906,
      "grad_norm": 0.5742276405421017,
      "learning_rate": 1.1972344219382947e-05,
      "loss": 11.7931,
      "step": 31122
    },
    {
      "epoch": 1.6947717356530736,
      "grad_norm": 0.5736738705491817,
      "learning_rate": 1.1968160642606308e-05,
      "loss": 11.8693,
      "step": 31123
    },
    {
      "epoch": 1.6948261896496566,
      "grad_norm": 0.5227934599039821,
      "learning_rate": 1.196397775038004e-05,
      "loss": 11.7932,
      "step": 31124
    },
    {
      "epoch": 1.6948806436462396,
      "grad_norm": 0.5585268892314317,
      "learning_rate": 1.195979554273664e-05,
      "loss": 11.8074,
      "step": 31125
    },
    {
      "epoch": 1.6949350976428226,
      "grad_norm": 0.5272249447494851,
      "learning_rate": 1.1955614019708616e-05,
      "loss": 11.8543,
      "step": 31126
    },
    {
      "epoch": 1.6949895516394058,
      "grad_norm": 0.6150872977246788,
      "learning_rate": 1.195143318132852e-05,
      "loss": 11.8697,
      "step": 31127
    },
    {
      "epoch": 1.6950440056359888,
      "grad_norm": 0.523893250614035,
      "learning_rate": 1.1947253027628824e-05,
      "loss": 11.7275,
      "step": 31128
    },
    {
      "epoch": 1.6950984596325718,
      "grad_norm": 0.5466708874336663,
      "learning_rate": 1.1943073558642082e-05,
      "loss": 11.6989,
      "step": 31129
    },
    {
      "epoch": 1.6951529136291548,
      "grad_norm": 0.5297247842485884,
      "learning_rate": 1.1938894774400743e-05,
      "loss": 11.7921,
      "step": 31130
    },
    {
      "epoch": 1.6952073676257378,
      "grad_norm": 0.668330003281475,
      "learning_rate": 1.1934716674937352e-05,
      "loss": 11.7513,
      "step": 31131
    },
    {
      "epoch": 1.6952618216223208,
      "grad_norm": 0.5890816140468341,
      "learning_rate": 1.1930539260284346e-05,
      "loss": 11.9074,
      "step": 31132
    },
    {
      "epoch": 1.6953162756189037,
      "grad_norm": 0.5921423566789797,
      "learning_rate": 1.1926362530474234e-05,
      "loss": 11.7829,
      "step": 31133
    },
    {
      "epoch": 1.6953707296154867,
      "grad_norm": 0.5604002994266877,
      "learning_rate": 1.1922186485539522e-05,
      "loss": 11.7451,
      "step": 31134
    },
    {
      "epoch": 1.6954251836120697,
      "grad_norm": 0.5386171188268375,
      "learning_rate": 1.1918011125512651e-05,
      "loss": 11.6358,
      "step": 31135
    },
    {
      "epoch": 1.6954796376086527,
      "grad_norm": 0.6047722369859541,
      "learning_rate": 1.1913836450426108e-05,
      "loss": 11.7773,
      "step": 31136
    },
    {
      "epoch": 1.6955340916052357,
      "grad_norm": 0.5636338677013298,
      "learning_rate": 1.190966246031232e-05,
      "loss": 11.8731,
      "step": 31137
    },
    {
      "epoch": 1.6955885456018187,
      "grad_norm": 0.5289540230140203,
      "learning_rate": 1.1905489155203764e-05,
      "loss": 11.6416,
      "step": 31138
    },
    {
      "epoch": 1.6956429995984017,
      "grad_norm": 0.5610172695635461,
      "learning_rate": 1.1901316535132922e-05,
      "loss": 11.8599,
      "step": 31139
    },
    {
      "epoch": 1.6956974535949847,
      "grad_norm": 0.5620621465242817,
      "learning_rate": 1.1897144600132203e-05,
      "loss": 11.8455,
      "step": 31140
    },
    {
      "epoch": 1.6957519075915677,
      "grad_norm": 0.5286016597676234,
      "learning_rate": 1.1892973350234072e-05,
      "loss": 11.7321,
      "step": 31141
    },
    {
      "epoch": 1.6958063615881507,
      "grad_norm": 0.5538988834899655,
      "learning_rate": 1.1888802785470943e-05,
      "loss": 11.7554,
      "step": 31142
    },
    {
      "epoch": 1.6958608155847337,
      "grad_norm": 0.5220134673972876,
      "learning_rate": 1.1884632905875281e-05,
      "loss": 11.6936,
      "step": 31143
    },
    {
      "epoch": 1.6959152695813169,
      "grad_norm": 0.5659510239305034,
      "learning_rate": 1.188046371147946e-05,
      "loss": 11.8246,
      "step": 31144
    },
    {
      "epoch": 1.6959697235778999,
      "grad_norm": 0.5261749517964932,
      "learning_rate": 1.1876295202315957e-05,
      "loss": 11.7013,
      "step": 31145
    },
    {
      "epoch": 1.6960241775744829,
      "grad_norm": 0.555824132148923,
      "learning_rate": 1.1872127378417163e-05,
      "loss": 11.9716,
      "step": 31146
    },
    {
      "epoch": 1.6960786315710659,
      "grad_norm": 0.5117819884497045,
      "learning_rate": 1.1867960239815467e-05,
      "loss": 11.8115,
      "step": 31147
    },
    {
      "epoch": 1.6961330855676489,
      "grad_norm": 0.5933197441541157,
      "learning_rate": 1.1863793786543288e-05,
      "loss": 11.7341,
      "step": 31148
    },
    {
      "epoch": 1.6961875395642318,
      "grad_norm": 0.5199894150538277,
      "learning_rate": 1.1859628018633052e-05,
      "loss": 11.7156,
      "step": 31149
    },
    {
      "epoch": 1.696241993560815,
      "grad_norm": 0.5452589369395606,
      "learning_rate": 1.1855462936117101e-05,
      "loss": 11.7479,
      "step": 31150
    },
    {
      "epoch": 1.696296447557398,
      "grad_norm": 0.5794796909162606,
      "learning_rate": 1.1851298539027889e-05,
      "loss": 11.7396,
      "step": 31151
    },
    {
      "epoch": 1.696350901553981,
      "grad_norm": 0.6460472069428692,
      "learning_rate": 1.184713482739772e-05,
      "loss": 11.6926,
      "step": 31152
    },
    {
      "epoch": 1.696405355550564,
      "grad_norm": 0.6811738661338196,
      "learning_rate": 1.184297180125905e-05,
      "loss": 11.8275,
      "step": 31153
    },
    {
      "epoch": 1.696459809547147,
      "grad_norm": 0.597358354833702,
      "learning_rate": 1.1838809460644195e-05,
      "loss": 11.8919,
      "step": 31154
    },
    {
      "epoch": 1.69651426354373,
      "grad_norm": 0.5613360676396864,
      "learning_rate": 1.1834647805585552e-05,
      "loss": 11.7637,
      "step": 31155
    },
    {
      "epoch": 1.696568717540313,
      "grad_norm": 0.5514983697671271,
      "learning_rate": 1.1830486836115483e-05,
      "loss": 11.7227,
      "step": 31156
    },
    {
      "epoch": 1.696623171536896,
      "grad_norm": 0.5435911315202359,
      "learning_rate": 1.1826326552266308e-05,
      "loss": 11.8893,
      "step": 31157
    },
    {
      "epoch": 1.696677625533479,
      "grad_norm": 0.5338506575659464,
      "learning_rate": 1.1822166954070434e-05,
      "loss": 11.6075,
      "step": 31158
    },
    {
      "epoch": 1.696732079530062,
      "grad_norm": 0.4986525568452677,
      "learning_rate": 1.1818008041560147e-05,
      "loss": 11.8006,
      "step": 31159
    },
    {
      "epoch": 1.696786533526645,
      "grad_norm": 0.5745697895053549,
      "learning_rate": 1.1813849814767818e-05,
      "loss": 11.898,
      "step": 31160
    },
    {
      "epoch": 1.696840987523228,
      "grad_norm": 0.5029913364089251,
      "learning_rate": 1.1809692273725803e-05,
      "loss": 11.7185,
      "step": 31161
    },
    {
      "epoch": 1.696895441519811,
      "grad_norm": 0.5740095031314133,
      "learning_rate": 1.1805535418466395e-05,
      "loss": 11.9046,
      "step": 31162
    },
    {
      "epoch": 1.696949895516394,
      "grad_norm": 0.5590567368795529,
      "learning_rate": 1.1801379249021948e-05,
      "loss": 11.7892,
      "step": 31163
    },
    {
      "epoch": 1.697004349512977,
      "grad_norm": 0.4763955732081574,
      "learning_rate": 1.1797223765424747e-05,
      "loss": 11.8285,
      "step": 31164
    },
    {
      "epoch": 1.69705880350956,
      "grad_norm": 0.5061884999654044,
      "learning_rate": 1.1793068967707154e-05,
      "loss": 11.7972,
      "step": 31165
    },
    {
      "epoch": 1.697113257506143,
      "grad_norm": 0.5523447234983317,
      "learning_rate": 1.1788914855901445e-05,
      "loss": 11.7633,
      "step": 31166
    },
    {
      "epoch": 1.697167711502726,
      "grad_norm": 0.5401459164104537,
      "learning_rate": 1.1784761430039904e-05,
      "loss": 11.7394,
      "step": 31167
    },
    {
      "epoch": 1.6972221654993092,
      "grad_norm": 0.5376616400063358,
      "learning_rate": 1.178060869015487e-05,
      "loss": 11.766,
      "step": 31168
    },
    {
      "epoch": 1.6972766194958921,
      "grad_norm": 0.5910258722714671,
      "learning_rate": 1.1776456636278598e-05,
      "loss": 11.7837,
      "step": 31169
    },
    {
      "epoch": 1.6973310734924751,
      "grad_norm": 0.5302378668243817,
      "learning_rate": 1.1772305268443417e-05,
      "loss": 11.8017,
      "step": 31170
    },
    {
      "epoch": 1.6973855274890581,
      "grad_norm": 0.6109088340488431,
      "learning_rate": 1.1768154586681568e-05,
      "loss": 11.8214,
      "step": 31171
    },
    {
      "epoch": 1.6974399814856411,
      "grad_norm": 0.5527344866619177,
      "learning_rate": 1.1764004591025346e-05,
      "loss": 11.6997,
      "step": 31172
    },
    {
      "epoch": 1.6974944354822243,
      "grad_norm": 0.625763945079985,
      "learning_rate": 1.1759855281507037e-05,
      "loss": 11.9015,
      "step": 31173
    },
    {
      "epoch": 1.6975488894788073,
      "grad_norm": 0.4966233521673928,
      "learning_rate": 1.1755706658158872e-05,
      "loss": 11.7,
      "step": 31174
    },
    {
      "epoch": 1.6976033434753903,
      "grad_norm": 0.5206706851319974,
      "learning_rate": 1.1751558721013157e-05,
      "loss": 11.7349,
      "step": 31175
    },
    {
      "epoch": 1.6976577974719733,
      "grad_norm": 0.5466083767639491,
      "learning_rate": 1.174741147010211e-05,
      "loss": 11.7965,
      "step": 31176
    },
    {
      "epoch": 1.6977122514685563,
      "grad_norm": 0.5555311654612501,
      "learning_rate": 1.1743264905457973e-05,
      "loss": 11.8305,
      "step": 31177
    },
    {
      "epoch": 1.6977667054651393,
      "grad_norm": 0.5827110532366812,
      "learning_rate": 1.1739119027113033e-05,
      "loss": 11.9338,
      "step": 31178
    },
    {
      "epoch": 1.6978211594617223,
      "grad_norm": 0.546713353054429,
      "learning_rate": 1.1734973835099483e-05,
      "loss": 11.8532,
      "step": 31179
    },
    {
      "epoch": 1.6978756134583053,
      "grad_norm": 0.5617743954800166,
      "learning_rate": 1.17308293294496e-05,
      "loss": 12.0388,
      "step": 31180
    },
    {
      "epoch": 1.6979300674548883,
      "grad_norm": 0.5302951201507142,
      "learning_rate": 1.1726685510195567e-05,
      "loss": 11.8557,
      "step": 31181
    },
    {
      "epoch": 1.6979845214514713,
      "grad_norm": 0.517914976928299,
      "learning_rate": 1.1722542377369639e-05,
      "loss": 11.8541,
      "step": 31182
    },
    {
      "epoch": 1.6980389754480543,
      "grad_norm": 0.5714738519216818,
      "learning_rate": 1.1718399931004043e-05,
      "loss": 11.8515,
      "step": 31183
    },
    {
      "epoch": 1.6980934294446373,
      "grad_norm": 0.6066301526252882,
      "learning_rate": 1.1714258171130954e-05,
      "loss": 11.713,
      "step": 31184
    },
    {
      "epoch": 1.6981478834412203,
      "grad_norm": 0.562355052368481,
      "learning_rate": 1.1710117097782635e-05,
      "loss": 11.8545,
      "step": 31185
    },
    {
      "epoch": 1.6982023374378032,
      "grad_norm": 0.5294204864785601,
      "learning_rate": 1.1705976710991206e-05,
      "loss": 11.7689,
      "step": 31186
    },
    {
      "epoch": 1.6982567914343862,
      "grad_norm": 0.5444777783774468,
      "learning_rate": 1.170183701078892e-05,
      "loss": 11.6812,
      "step": 31187
    },
    {
      "epoch": 1.6983112454309692,
      "grad_norm": 0.57113633311093,
      "learning_rate": 1.1697697997207979e-05,
      "loss": 11.7576,
      "step": 31188
    },
    {
      "epoch": 1.6983656994275522,
      "grad_norm": 0.5507576347384435,
      "learning_rate": 1.1693559670280518e-05,
      "loss": 11.5738,
      "step": 31189
    },
    {
      "epoch": 1.6984201534241352,
      "grad_norm": 0.5175553432450869,
      "learning_rate": 1.1689422030038765e-05,
      "loss": 11.7401,
      "step": 31190
    },
    {
      "epoch": 1.6984746074207184,
      "grad_norm": 0.5095825569788507,
      "learning_rate": 1.1685285076514863e-05,
      "loss": 11.6847,
      "step": 31191
    },
    {
      "epoch": 1.6985290614173014,
      "grad_norm": 0.5677070131155337,
      "learning_rate": 1.1681148809741016e-05,
      "loss": 11.8285,
      "step": 31192
    },
    {
      "epoch": 1.6985835154138844,
      "grad_norm": 0.5335949677262533,
      "learning_rate": 1.1677013229749334e-05,
      "loss": 11.7886,
      "step": 31193
    },
    {
      "epoch": 1.6986379694104674,
      "grad_norm": 0.560539418745713,
      "learning_rate": 1.1672878336572024e-05,
      "loss": 11.7421,
      "step": 31194
    },
    {
      "epoch": 1.6986924234070504,
      "grad_norm": 0.5705922297864177,
      "learning_rate": 1.166874413024126e-05,
      "loss": 11.8745,
      "step": 31195
    },
    {
      "epoch": 1.6987468774036334,
      "grad_norm": 0.554963106832216,
      "learning_rate": 1.1664610610789106e-05,
      "loss": 11.9591,
      "step": 31196
    },
    {
      "epoch": 1.6988013314002166,
      "grad_norm": 0.548341727094421,
      "learning_rate": 1.1660477778247758e-05,
      "loss": 11.7741,
      "step": 31197
    },
    {
      "epoch": 1.6988557853967996,
      "grad_norm": 0.5568708477835408,
      "learning_rate": 1.165634563264938e-05,
      "loss": 11.7128,
      "step": 31198
    },
    {
      "epoch": 1.6989102393933826,
      "grad_norm": 0.5605668718638176,
      "learning_rate": 1.1652214174026045e-05,
      "loss": 11.8779,
      "step": 31199
    },
    {
      "epoch": 1.6989646933899656,
      "grad_norm": 0.5776693095633945,
      "learning_rate": 1.1648083402409927e-05,
      "loss": 11.8465,
      "step": 31200
    },
    {
      "epoch": 1.6990191473865486,
      "grad_norm": 0.5691267610329024,
      "learning_rate": 1.1643953317833112e-05,
      "loss": 11.7833,
      "step": 31201
    },
    {
      "epoch": 1.6990736013831316,
      "grad_norm": 0.5488503787332141,
      "learning_rate": 1.1639823920327753e-05,
      "loss": 11.7771,
      "step": 31202
    },
    {
      "epoch": 1.6991280553797146,
      "grad_norm": 0.5433140690219086,
      "learning_rate": 1.1635695209925922e-05,
      "loss": 11.8082,
      "step": 31203
    },
    {
      "epoch": 1.6991825093762976,
      "grad_norm": 0.5317624685769947,
      "learning_rate": 1.1631567186659741e-05,
      "loss": 11.8112,
      "step": 31204
    },
    {
      "epoch": 1.6992369633728805,
      "grad_norm": 0.5585805141630426,
      "learning_rate": 1.1627439850561351e-05,
      "loss": 11.8755,
      "step": 31205
    },
    {
      "epoch": 1.6992914173694635,
      "grad_norm": 0.5323604788938309,
      "learning_rate": 1.1623313201662778e-05,
      "loss": 11.7716,
      "step": 31206
    },
    {
      "epoch": 1.6993458713660465,
      "grad_norm": 0.5593327894143041,
      "learning_rate": 1.1619187239996154e-05,
      "loss": 11.8379,
      "step": 31207
    },
    {
      "epoch": 1.6994003253626295,
      "grad_norm": 0.513432314602708,
      "learning_rate": 1.1615061965593533e-05,
      "loss": 11.8205,
      "step": 31208
    },
    {
      "epoch": 1.6994547793592125,
      "grad_norm": 0.5476954780496204,
      "learning_rate": 1.1610937378487008e-05,
      "loss": 11.826,
      "step": 31209
    },
    {
      "epoch": 1.6995092333557955,
      "grad_norm": 0.6067927322440799,
      "learning_rate": 1.1606813478708689e-05,
      "loss": 11.8313,
      "step": 31210
    },
    {
      "epoch": 1.6995636873523785,
      "grad_norm": 0.528739121928077,
      "learning_rate": 1.1602690266290583e-05,
      "loss": 11.7791,
      "step": 31211
    },
    {
      "epoch": 1.6996181413489615,
      "grad_norm": 0.4979464794477796,
      "learning_rate": 1.15985677412648e-05,
      "loss": 11.68,
      "step": 31212
    },
    {
      "epoch": 1.6996725953455445,
      "grad_norm": 0.5699731103190695,
      "learning_rate": 1.1594445903663365e-05,
      "loss": 11.783,
      "step": 31213
    },
    {
      "epoch": 1.6997270493421277,
      "grad_norm": 0.5240174059605114,
      "learning_rate": 1.1590324753518367e-05,
      "loss": 11.7972,
      "step": 31214
    },
    {
      "epoch": 1.6997815033387107,
      "grad_norm": 0.5348642485513088,
      "learning_rate": 1.1586204290861824e-05,
      "loss": 11.8107,
      "step": 31215
    },
    {
      "epoch": 1.6998359573352937,
      "grad_norm": 0.5532560565183934,
      "learning_rate": 1.1582084515725766e-05,
      "loss": 11.8337,
      "step": 31216
    },
    {
      "epoch": 1.6998904113318767,
      "grad_norm": 0.5466933677368346,
      "learning_rate": 1.1577965428142257e-05,
      "loss": 11.7897,
      "step": 31217
    },
    {
      "epoch": 1.6999448653284597,
      "grad_norm": 0.5572512944589181,
      "learning_rate": 1.1573847028143315e-05,
      "loss": 11.8394,
      "step": 31218
    },
    {
      "epoch": 1.6999993193250427,
      "grad_norm": 0.5551068691720326,
      "learning_rate": 1.1569729315760969e-05,
      "loss": 11.8603,
      "step": 31219
    },
    {
      "epoch": 1.7000537733216259,
      "grad_norm": 0.5692280362315412,
      "learning_rate": 1.1565612291027228e-05,
      "loss": 11.7198,
      "step": 31220
    },
    {
      "epoch": 1.7001082273182089,
      "grad_norm": 0.5044517711240781,
      "learning_rate": 1.1561495953974122e-05,
      "loss": 11.7405,
      "step": 31221
    },
    {
      "epoch": 1.7001626813147919,
      "grad_norm": 0.6662775759620126,
      "learning_rate": 1.1557380304633659e-05,
      "loss": 11.8044,
      "step": 31222
    },
    {
      "epoch": 1.7002171353113749,
      "grad_norm": 0.5468430216285007,
      "learning_rate": 1.1553265343037833e-05,
      "loss": 11.6875,
      "step": 31223
    },
    {
      "epoch": 1.7002715893079579,
      "grad_norm": 0.510436662117331,
      "learning_rate": 1.1549151069218655e-05,
      "loss": 11.7209,
      "step": 31224
    },
    {
      "epoch": 1.7003260433045408,
      "grad_norm": 0.571369443119801,
      "learning_rate": 1.1545037483208098e-05,
      "loss": 11.8318,
      "step": 31225
    },
    {
      "epoch": 1.7003804973011238,
      "grad_norm": 0.5187043516404863,
      "learning_rate": 1.1540924585038193e-05,
      "loss": 11.7688,
      "step": 31226
    },
    {
      "epoch": 1.7004349512977068,
      "grad_norm": 0.5264191141204417,
      "learning_rate": 1.153681237474088e-05,
      "loss": 11.8329,
      "step": 31227
    },
    {
      "epoch": 1.7004894052942898,
      "grad_norm": 0.5478201588307893,
      "learning_rate": 1.1532700852348132e-05,
      "loss": 11.8485,
      "step": 31228
    },
    {
      "epoch": 1.7005438592908728,
      "grad_norm": 0.5377739250236913,
      "learning_rate": 1.1528590017891961e-05,
      "loss": 11.6057,
      "step": 31229
    },
    {
      "epoch": 1.7005983132874558,
      "grad_norm": 0.5774197328841263,
      "learning_rate": 1.1524479871404293e-05,
      "loss": 11.7849,
      "step": 31230
    },
    {
      "epoch": 1.7006527672840388,
      "grad_norm": 0.5434830326628208,
      "learning_rate": 1.1520370412917103e-05,
      "loss": 11.7581,
      "step": 31231
    },
    {
      "epoch": 1.7007072212806218,
      "grad_norm": 0.5541688132906321,
      "learning_rate": 1.1516261642462366e-05,
      "loss": 11.781,
      "step": 31232
    },
    {
      "epoch": 1.7007616752772048,
      "grad_norm": 0.6036717618407496,
      "learning_rate": 1.1512153560072003e-05,
      "loss": 11.846,
      "step": 31233
    },
    {
      "epoch": 1.7008161292737878,
      "grad_norm": 0.5431214809691001,
      "learning_rate": 1.1508046165777997e-05,
      "loss": 11.79,
      "step": 31234
    },
    {
      "epoch": 1.7008705832703708,
      "grad_norm": 0.5744119753557512,
      "learning_rate": 1.1503939459612234e-05,
      "loss": 11.72,
      "step": 31235
    },
    {
      "epoch": 1.7009250372669538,
      "grad_norm": 0.6004856798370535,
      "learning_rate": 1.149983344160671e-05,
      "loss": 11.8934,
      "step": 31236
    },
    {
      "epoch": 1.7009794912635368,
      "grad_norm": 0.6745525890407313,
      "learning_rate": 1.1495728111793313e-05,
      "loss": 11.9227,
      "step": 31237
    },
    {
      "epoch": 1.70103394526012,
      "grad_norm": 0.5476790313785479,
      "learning_rate": 1.149162347020396e-05,
      "loss": 11.7708,
      "step": 31238
    },
    {
      "epoch": 1.701088399256703,
      "grad_norm": 0.5726140051205252,
      "learning_rate": 1.1487519516870604e-05,
      "loss": 11.7045,
      "step": 31239
    },
    {
      "epoch": 1.701142853253286,
      "grad_norm": 0.5206056900957082,
      "learning_rate": 1.148341625182512e-05,
      "loss": 11.7224,
      "step": 31240
    },
    {
      "epoch": 1.701197307249869,
      "grad_norm": 0.5515522092270686,
      "learning_rate": 1.1479313675099456e-05,
      "loss": 11.6924,
      "step": 31241
    },
    {
      "epoch": 1.701251761246452,
      "grad_norm": 0.5844488582553043,
      "learning_rate": 1.1475211786725482e-05,
      "loss": 11.8821,
      "step": 31242
    },
    {
      "epoch": 1.7013062152430352,
      "grad_norm": 0.5717821331759587,
      "learning_rate": 1.1471110586735101e-05,
      "loss": 11.7941,
      "step": 31243
    },
    {
      "epoch": 1.7013606692396182,
      "grad_norm": 0.5688372934165855,
      "learning_rate": 1.1467010075160223e-05,
      "loss": 11.965,
      "step": 31244
    },
    {
      "epoch": 1.7014151232362011,
      "grad_norm": 0.5840128309900645,
      "learning_rate": 1.1462910252032711e-05,
      "loss": 11.7605,
      "step": 31245
    },
    {
      "epoch": 1.7014695772327841,
      "grad_norm": 0.5091680797575212,
      "learning_rate": 1.1458811117384472e-05,
      "loss": 11.8584,
      "step": 31246
    },
    {
      "epoch": 1.7015240312293671,
      "grad_norm": 0.4746440996748854,
      "learning_rate": 1.145471267124737e-05,
      "loss": 11.6595,
      "step": 31247
    },
    {
      "epoch": 1.7015784852259501,
      "grad_norm": 0.4875825178627739,
      "learning_rate": 1.1450614913653246e-05,
      "loss": 11.6376,
      "step": 31248
    },
    {
      "epoch": 1.7016329392225331,
      "grad_norm": 0.5855410815440408,
      "learning_rate": 1.1446517844634009e-05,
      "loss": 11.8022,
      "step": 31249
    },
    {
      "epoch": 1.701687393219116,
      "grad_norm": 0.6982611521644404,
      "learning_rate": 1.1442421464221487e-05,
      "loss": 11.8648,
      "step": 31250
    },
    {
      "epoch": 1.701741847215699,
      "grad_norm": 0.578953360743314,
      "learning_rate": 1.1438325772447566e-05,
      "loss": 11.7539,
      "step": 31251
    },
    {
      "epoch": 1.701796301212282,
      "grad_norm": 0.5357224343657475,
      "learning_rate": 1.1434230769344046e-05,
      "loss": 11.9149,
      "step": 31252
    },
    {
      "epoch": 1.701850755208865,
      "grad_norm": 0.5423134189518319,
      "learning_rate": 1.1430136454942808e-05,
      "loss": 11.6547,
      "step": 31253
    },
    {
      "epoch": 1.701905209205448,
      "grad_norm": 0.5542201498977888,
      "learning_rate": 1.1426042829275706e-05,
      "loss": 11.7096,
      "step": 31254
    },
    {
      "epoch": 1.701959663202031,
      "grad_norm": 0.5109266878922101,
      "learning_rate": 1.1421949892374516e-05,
      "loss": 11.7829,
      "step": 31255
    },
    {
      "epoch": 1.702014117198614,
      "grad_norm": 0.5398111518041445,
      "learning_rate": 1.1417857644271158e-05,
      "loss": 11.7906,
      "step": 31256
    },
    {
      "epoch": 1.702068571195197,
      "grad_norm": 0.5654600725757558,
      "learning_rate": 1.1413766084997335e-05,
      "loss": 11.88,
      "step": 31257
    },
    {
      "epoch": 1.70212302519178,
      "grad_norm": 0.6032763644933257,
      "learning_rate": 1.1409675214584925e-05,
      "loss": 11.8079,
      "step": 31258
    },
    {
      "epoch": 1.702177479188363,
      "grad_norm": 0.5796129509260738,
      "learning_rate": 1.140558503306577e-05,
      "loss": 11.7964,
      "step": 31259
    },
    {
      "epoch": 1.702231933184946,
      "grad_norm": 0.5547492772467061,
      "learning_rate": 1.1401495540471607e-05,
      "loss": 11.7957,
      "step": 31260
    },
    {
      "epoch": 1.7022863871815292,
      "grad_norm": 0.5234253960011268,
      "learning_rate": 1.1397406736834305e-05,
      "loss": 11.7468,
      "step": 31261
    },
    {
      "epoch": 1.7023408411781122,
      "grad_norm": 0.5560471414455337,
      "learning_rate": 1.13933186221856e-05,
      "loss": 11.8825,
      "step": 31262
    },
    {
      "epoch": 1.7023952951746952,
      "grad_norm": 0.556439278089486,
      "learning_rate": 1.1389231196557337e-05,
      "loss": 11.8036,
      "step": 31263
    },
    {
      "epoch": 1.7024497491712782,
      "grad_norm": 0.5913091902537242,
      "learning_rate": 1.1385144459981245e-05,
      "loss": 11.8411,
      "step": 31264
    },
    {
      "epoch": 1.7025042031678612,
      "grad_norm": 0.5001200998457481,
      "learning_rate": 1.1381058412489132e-05,
      "loss": 11.801,
      "step": 31265
    },
    {
      "epoch": 1.7025586571644442,
      "grad_norm": 0.5743565103017917,
      "learning_rate": 1.1376973054112816e-05,
      "loss": 11.8137,
      "step": 31266
    },
    {
      "epoch": 1.7026131111610274,
      "grad_norm": 0.5366117216556852,
      "learning_rate": 1.1372888384883973e-05,
      "loss": 11.7674,
      "step": 31267
    },
    {
      "epoch": 1.7026675651576104,
      "grad_norm": 0.5544635367823163,
      "learning_rate": 1.1368804404834431e-05,
      "loss": 11.7378,
      "step": 31268
    },
    {
      "epoch": 1.7027220191541934,
      "grad_norm": 0.5414820553650712,
      "learning_rate": 1.1364721113995914e-05,
      "loss": 11.7492,
      "step": 31269
    },
    {
      "epoch": 1.7027764731507764,
      "grad_norm": 0.5411382474896068,
      "learning_rate": 1.1360638512400179e-05,
      "loss": 11.7101,
      "step": 31270
    },
    {
      "epoch": 1.7028309271473594,
      "grad_norm": 0.5423226520013353,
      "learning_rate": 1.1356556600079016e-05,
      "loss": 11.8066,
      "step": 31271
    },
    {
      "epoch": 1.7028853811439424,
      "grad_norm": 0.5846380472738664,
      "learning_rate": 1.135247537706411e-05,
      "loss": 11.8038,
      "step": 31272
    },
    {
      "epoch": 1.7029398351405254,
      "grad_norm": 0.5268004901861915,
      "learning_rate": 1.1348394843387244e-05,
      "loss": 11.8386,
      "step": 31273
    },
    {
      "epoch": 1.7029942891371084,
      "grad_norm": 0.5801831134073427,
      "learning_rate": 1.1344314999080107e-05,
      "loss": 11.8378,
      "step": 31274
    },
    {
      "epoch": 1.7030487431336914,
      "grad_norm": 0.5959497144288662,
      "learning_rate": 1.1340235844174463e-05,
      "loss": 11.8477,
      "step": 31275
    },
    {
      "epoch": 1.7031031971302744,
      "grad_norm": 0.5313650368151611,
      "learning_rate": 1.1336157378702018e-05,
      "loss": 11.729,
      "step": 31276
    },
    {
      "epoch": 1.7031576511268574,
      "grad_norm": 0.5720033144718367,
      "learning_rate": 1.1332079602694446e-05,
      "loss": 11.7933,
      "step": 31277
    },
    {
      "epoch": 1.7032121051234403,
      "grad_norm": 0.5260727974708785,
      "learning_rate": 1.1328002516183522e-05,
      "loss": 11.7659,
      "step": 31278
    },
    {
      "epoch": 1.7032665591200233,
      "grad_norm": 0.530997427908885,
      "learning_rate": 1.1323926119200912e-05,
      "loss": 11.6772,
      "step": 31279
    },
    {
      "epoch": 1.7033210131166063,
      "grad_norm": 0.4760019346679841,
      "learning_rate": 1.131985041177831e-05,
      "loss": 11.6817,
      "step": 31280
    },
    {
      "epoch": 1.7033754671131893,
      "grad_norm": 0.5251093535938762,
      "learning_rate": 1.1315775393947447e-05,
      "loss": 11.7829,
      "step": 31281
    },
    {
      "epoch": 1.7034299211097723,
      "grad_norm": 0.6078293290441132,
      "learning_rate": 1.1311701065739965e-05,
      "loss": 11.7932,
      "step": 31282
    },
    {
      "epoch": 1.7034843751063553,
      "grad_norm": 0.5385094091574454,
      "learning_rate": 1.1307627427187595e-05,
      "loss": 11.7556,
      "step": 31283
    },
    {
      "epoch": 1.7035388291029385,
      "grad_norm": 0.5637869318010038,
      "learning_rate": 1.1303554478321954e-05,
      "loss": 11.8293,
      "step": 31284
    },
    {
      "epoch": 1.7035932830995215,
      "grad_norm": 0.5672741725345258,
      "learning_rate": 1.1299482219174784e-05,
      "loss": 11.8961,
      "step": 31285
    },
    {
      "epoch": 1.7036477370961045,
      "grad_norm": 0.5191819917234376,
      "learning_rate": 1.1295410649777704e-05,
      "loss": 11.7897,
      "step": 31286
    },
    {
      "epoch": 1.7037021910926875,
      "grad_norm": 0.5211313978753638,
      "learning_rate": 1.1291339770162368e-05,
      "loss": 11.8367,
      "step": 31287
    },
    {
      "epoch": 1.7037566450892705,
      "grad_norm": 0.6807269684400129,
      "learning_rate": 1.1287269580360471e-05,
      "loss": 11.9386,
      "step": 31288
    },
    {
      "epoch": 1.7038110990858535,
      "grad_norm": 0.5270099005856907,
      "learning_rate": 1.1283200080403632e-05,
      "loss": 11.7162,
      "step": 31289
    },
    {
      "epoch": 1.7038655530824367,
      "grad_norm": 0.5911195333290961,
      "learning_rate": 1.1279131270323518e-05,
      "loss": 11.8088,
      "step": 31290
    },
    {
      "epoch": 1.7039200070790197,
      "grad_norm": 0.5900182950918653,
      "learning_rate": 1.1275063150151744e-05,
      "loss": 11.8025,
      "step": 31291
    },
    {
      "epoch": 1.7039744610756027,
      "grad_norm": 0.5600283388219753,
      "learning_rate": 1.1270995719919952e-05,
      "loss": 11.9367,
      "step": 31292
    },
    {
      "epoch": 1.7040289150721857,
      "grad_norm": 0.5049451539233827,
      "learning_rate": 1.1266928979659796e-05,
      "loss": 11.7135,
      "step": 31293
    },
    {
      "epoch": 1.7040833690687687,
      "grad_norm": 0.6294900510590399,
      "learning_rate": 1.1262862929402862e-05,
      "loss": 11.8334,
      "step": 31294
    },
    {
      "epoch": 1.7041378230653517,
      "grad_norm": 0.5269043995110961,
      "learning_rate": 1.1258797569180812e-05,
      "loss": 11.6721,
      "step": 31295
    },
    {
      "epoch": 1.7041922770619347,
      "grad_norm": 0.5255994567283027,
      "learning_rate": 1.1254732899025222e-05,
      "loss": 11.8429,
      "step": 31296
    },
    {
      "epoch": 1.7042467310585176,
      "grad_norm": 0.5035362961618848,
      "learning_rate": 1.1250668918967711e-05,
      "loss": 11.742,
      "step": 31297
    },
    {
      "epoch": 1.7043011850551006,
      "grad_norm": 0.5254803888413105,
      "learning_rate": 1.1246605629039886e-05,
      "loss": 11.9251,
      "step": 31298
    },
    {
      "epoch": 1.7043556390516836,
      "grad_norm": 0.50652490422515,
      "learning_rate": 1.1242543029273322e-05,
      "loss": 11.7973,
      "step": 31299
    },
    {
      "epoch": 1.7044100930482666,
      "grad_norm": 0.5293692893456606,
      "learning_rate": 1.123848111969965e-05,
      "loss": 11.8231,
      "step": 31300
    },
    {
      "epoch": 1.7044645470448496,
      "grad_norm": 0.4859020321205245,
      "learning_rate": 1.1234419900350413e-05,
      "loss": 11.6815,
      "step": 31301
    },
    {
      "epoch": 1.7045190010414326,
      "grad_norm": 0.49308515674546355,
      "learning_rate": 1.1230359371257215e-05,
      "loss": 11.7344,
      "step": 31302
    },
    {
      "epoch": 1.7045734550380156,
      "grad_norm": 0.5485698988111734,
      "learning_rate": 1.1226299532451657e-05,
      "loss": 11.7352,
      "step": 31303
    },
    {
      "epoch": 1.7046279090345986,
      "grad_norm": 0.5742764621579853,
      "learning_rate": 1.1222240383965243e-05,
      "loss": 11.7522,
      "step": 31304
    },
    {
      "epoch": 1.7046823630311816,
      "grad_norm": 0.4777860565935414,
      "learning_rate": 1.1218181925829607e-05,
      "loss": 11.7434,
      "step": 31305
    },
    {
      "epoch": 1.7047368170277646,
      "grad_norm": 0.5040213787829656,
      "learning_rate": 1.1214124158076278e-05,
      "loss": 11.8565,
      "step": 31306
    },
    {
      "epoch": 1.7047912710243478,
      "grad_norm": 0.5030122128680107,
      "learning_rate": 1.1210067080736786e-05,
      "loss": 11.6438,
      "step": 31307
    },
    {
      "epoch": 1.7048457250209308,
      "grad_norm": 0.5981916364110439,
      "learning_rate": 1.1206010693842727e-05,
      "loss": 11.9581,
      "step": 31308
    },
    {
      "epoch": 1.7049001790175138,
      "grad_norm": 0.5167265446436821,
      "learning_rate": 1.120195499742559e-05,
      "loss": 11.8135,
      "step": 31309
    },
    {
      "epoch": 1.7049546330140968,
      "grad_norm": 0.5268983740408758,
      "learning_rate": 1.1197899991516968e-05,
      "loss": 11.7719,
      "step": 31310
    },
    {
      "epoch": 1.7050090870106798,
      "grad_norm": 0.5874469871352266,
      "learning_rate": 1.119384567614835e-05,
      "loss": 11.7836,
      "step": 31311
    },
    {
      "epoch": 1.7050635410072628,
      "grad_norm": 0.5373827277132782,
      "learning_rate": 1.11897920513513e-05,
      "loss": 11.7336,
      "step": 31312
    },
    {
      "epoch": 1.705117995003846,
      "grad_norm": 0.5108124772621323,
      "learning_rate": 1.1185739117157301e-05,
      "loss": 11.8133,
      "step": 31313
    },
    {
      "epoch": 1.705172449000429,
      "grad_norm": 0.5876733779798438,
      "learning_rate": 1.1181686873597886e-05,
      "loss": 11.9087,
      "step": 31314
    },
    {
      "epoch": 1.705226902997012,
      "grad_norm": 0.5517447147814664,
      "learning_rate": 1.1177635320704594e-05,
      "loss": 11.9218,
      "step": 31315
    },
    {
      "epoch": 1.705281356993595,
      "grad_norm": 0.6209837257303116,
      "learning_rate": 1.1173584458508901e-05,
      "loss": 11.9754,
      "step": 31316
    },
    {
      "epoch": 1.705335810990178,
      "grad_norm": 0.5512831780592493,
      "learning_rate": 1.1169534287042283e-05,
      "loss": 11.6591,
      "step": 31317
    },
    {
      "epoch": 1.705390264986761,
      "grad_norm": 0.5985583538413535,
      "learning_rate": 1.1165484806336302e-05,
      "loss": 11.7803,
      "step": 31318
    },
    {
      "epoch": 1.705444718983344,
      "grad_norm": 0.575165712108634,
      "learning_rate": 1.1161436016422378e-05,
      "loss": 11.8899,
      "step": 31319
    },
    {
      "epoch": 1.705499172979927,
      "grad_norm": 0.5819721823764681,
      "learning_rate": 1.115738791733204e-05,
      "loss": 11.7869,
      "step": 31320
    },
    {
      "epoch": 1.70555362697651,
      "grad_norm": 0.5609460168421133,
      "learning_rate": 1.1153340509096744e-05,
      "loss": 11.7868,
      "step": 31321
    },
    {
      "epoch": 1.705608080973093,
      "grad_norm": 0.5771190839784344,
      "learning_rate": 1.1149293791747984e-05,
      "loss": 11.841,
      "step": 31322
    },
    {
      "epoch": 1.705662534969676,
      "grad_norm": 0.5188743409995102,
      "learning_rate": 1.114524776531719e-05,
      "loss": 11.8366,
      "step": 31323
    },
    {
      "epoch": 1.705716988966259,
      "grad_norm": 0.5913309838941305,
      "learning_rate": 1.1141202429835885e-05,
      "loss": 11.7614,
      "step": 31324
    },
    {
      "epoch": 1.7057714429628419,
      "grad_norm": 0.5552822084419874,
      "learning_rate": 1.1137157785335462e-05,
      "loss": 11.8259,
      "step": 31325
    },
    {
      "epoch": 1.7058258969594249,
      "grad_norm": 0.5318717769413936,
      "learning_rate": 1.1133113831847431e-05,
      "loss": 11.7007,
      "step": 31326
    },
    {
      "epoch": 1.7058803509560079,
      "grad_norm": 0.5681071697112415,
      "learning_rate": 1.1129070569403199e-05,
      "loss": 11.896,
      "step": 31327
    },
    {
      "epoch": 1.7059348049525909,
      "grad_norm": 0.5089217670051894,
      "learning_rate": 1.112502799803421e-05,
      "loss": 11.7835,
      "step": 31328
    },
    {
      "epoch": 1.7059892589491739,
      "grad_norm": 0.6497385507303329,
      "learning_rate": 1.1120986117771892e-05,
      "loss": 11.7724,
      "step": 31329
    },
    {
      "epoch": 1.7060437129457569,
      "grad_norm": 0.5022658093666803,
      "learning_rate": 1.1116944928647722e-05,
      "loss": 11.7518,
      "step": 31330
    },
    {
      "epoch": 1.70609816694234,
      "grad_norm": 0.5996853030420862,
      "learning_rate": 1.1112904430693071e-05,
      "loss": 11.7625,
      "step": 31331
    },
    {
      "epoch": 1.706152620938923,
      "grad_norm": 0.5378810116332772,
      "learning_rate": 1.1108864623939408e-05,
      "loss": 11.7291,
      "step": 31332
    },
    {
      "epoch": 1.706207074935506,
      "grad_norm": 0.5509921075397711,
      "learning_rate": 1.1104825508418094e-05,
      "loss": 11.8393,
      "step": 31333
    },
    {
      "epoch": 1.706261528932089,
      "grad_norm": 0.5440243687930099,
      "learning_rate": 1.110078708416058e-05,
      "loss": 11.7172,
      "step": 31334
    },
    {
      "epoch": 1.706315982928672,
      "grad_norm": 0.5716647565322295,
      "learning_rate": 1.1096749351198243e-05,
      "loss": 11.7178,
      "step": 31335
    },
    {
      "epoch": 1.706370436925255,
      "grad_norm": 0.59309097214083,
      "learning_rate": 1.1092712309562503e-05,
      "loss": 11.7863,
      "step": 31336
    },
    {
      "epoch": 1.7064248909218382,
      "grad_norm": 0.5770068869056281,
      "learning_rate": 1.1088675959284756e-05,
      "loss": 11.7634,
      "step": 31337
    },
    {
      "epoch": 1.7064793449184212,
      "grad_norm": 0.5370075067079356,
      "learning_rate": 1.1084640300396343e-05,
      "loss": 11.6557,
      "step": 31338
    },
    {
      "epoch": 1.7065337989150042,
      "grad_norm": 0.5773533412198194,
      "learning_rate": 1.1080605332928706e-05,
      "loss": 11.8865,
      "step": 31339
    },
    {
      "epoch": 1.7065882529115872,
      "grad_norm": 0.6064211803671088,
      "learning_rate": 1.1076571056913166e-05,
      "loss": 11.9042,
      "step": 31340
    },
    {
      "epoch": 1.7066427069081702,
      "grad_norm": 0.5828648957159318,
      "learning_rate": 1.1072537472381127e-05,
      "loss": 11.7946,
      "step": 31341
    },
    {
      "epoch": 1.7066971609047532,
      "grad_norm": 0.5706479761074192,
      "learning_rate": 1.106850457936397e-05,
      "loss": 11.8108,
      "step": 31342
    },
    {
      "epoch": 1.7067516149013362,
      "grad_norm": 0.5552683982108595,
      "learning_rate": 1.106447237789301e-05,
      "loss": 11.8471,
      "step": 31343
    },
    {
      "epoch": 1.7068060688979192,
      "grad_norm": 0.5314013550341861,
      "learning_rate": 1.1060440867999655e-05,
      "loss": 11.7732,
      "step": 31344
    },
    {
      "epoch": 1.7068605228945022,
      "grad_norm": 0.5327937711703757,
      "learning_rate": 1.1056410049715193e-05,
      "loss": 11.7761,
      "step": 31345
    },
    {
      "epoch": 1.7069149768910852,
      "grad_norm": 0.5575654136767733,
      "learning_rate": 1.1052379923071033e-05,
      "loss": 11.7641,
      "step": 31346
    },
    {
      "epoch": 1.7069694308876682,
      "grad_norm": 0.6278233816868286,
      "learning_rate": 1.1048350488098491e-05,
      "loss": 11.9154,
      "step": 31347
    },
    {
      "epoch": 1.7070238848842512,
      "grad_norm": 0.5061958734375708,
      "learning_rate": 1.1044321744828857e-05,
      "loss": 11.7967,
      "step": 31348
    },
    {
      "epoch": 1.7070783388808342,
      "grad_norm": 0.5369718885477847,
      "learning_rate": 1.1040293693293524e-05,
      "loss": 11.7805,
      "step": 31349
    },
    {
      "epoch": 1.7071327928774171,
      "grad_norm": 0.5771905146793501,
      "learning_rate": 1.103626633352376e-05,
      "loss": 11.8721,
      "step": 31350
    },
    {
      "epoch": 1.7071872468740001,
      "grad_norm": 0.5150195865782898,
      "learning_rate": 1.1032239665550915e-05,
      "loss": 11.7464,
      "step": 31351
    },
    {
      "epoch": 1.7072417008705831,
      "grad_norm": 0.5898189938885433,
      "learning_rate": 1.1028213689406309e-05,
      "loss": 11.8358,
      "step": 31352
    },
    {
      "epoch": 1.7072961548671661,
      "grad_norm": 0.6005633282977264,
      "learning_rate": 1.1024188405121216e-05,
      "loss": 11.8762,
      "step": 31353
    },
    {
      "epoch": 1.7073506088637493,
      "grad_norm": 0.7128329843736801,
      "learning_rate": 1.102016381272698e-05,
      "loss": 11.8238,
      "step": 31354
    },
    {
      "epoch": 1.7074050628603323,
      "grad_norm": 0.587798147854613,
      "learning_rate": 1.1016139912254841e-05,
      "loss": 11.9425,
      "step": 31355
    },
    {
      "epoch": 1.7074595168569153,
      "grad_norm": 0.5373226019184096,
      "learning_rate": 1.1012116703736153e-05,
      "loss": 11.7459,
      "step": 31356
    },
    {
      "epoch": 1.7075139708534983,
      "grad_norm": 0.5513145666790776,
      "learning_rate": 1.1008094187202166e-05,
      "loss": 11.6954,
      "step": 31357
    },
    {
      "epoch": 1.7075684248500813,
      "grad_norm": 0.5214057416926633,
      "learning_rate": 1.1004072362684136e-05,
      "loss": 11.7938,
      "step": 31358
    },
    {
      "epoch": 1.7076228788466643,
      "grad_norm": 0.4945392891809306,
      "learning_rate": 1.1000051230213393e-05,
      "loss": 11.7933,
      "step": 31359
    },
    {
      "epoch": 1.7076773328432475,
      "grad_norm": 0.5116429250355039,
      "learning_rate": 1.0996030789821143e-05,
      "loss": 11.8087,
      "step": 31360
    },
    {
      "epoch": 1.7077317868398305,
      "grad_norm": 0.5063420571560326,
      "learning_rate": 1.0992011041538719e-05,
      "loss": 11.8815,
      "step": 31361
    },
    {
      "epoch": 1.7077862408364135,
      "grad_norm": 0.5517296998105604,
      "learning_rate": 1.0987991985397317e-05,
      "loss": 11.768,
      "step": 31362
    },
    {
      "epoch": 1.7078406948329965,
      "grad_norm": 0.541663286380688,
      "learning_rate": 1.0983973621428212e-05,
      "loss": 11.7052,
      "step": 31363
    },
    {
      "epoch": 1.7078951488295795,
      "grad_norm": 0.538583345271869,
      "learning_rate": 1.097995594966268e-05,
      "loss": 11.7137,
      "step": 31364
    },
    {
      "epoch": 1.7079496028261625,
      "grad_norm": 0.5956429163476145,
      "learning_rate": 1.0975938970131915e-05,
      "loss": 11.8822,
      "step": 31365
    },
    {
      "epoch": 1.7080040568227455,
      "grad_norm": 0.528189305513491,
      "learning_rate": 1.0971922682867209e-05,
      "loss": 11.9011,
      "step": 31366
    },
    {
      "epoch": 1.7080585108193285,
      "grad_norm": 0.5457842011110622,
      "learning_rate": 1.0967907087899754e-05,
      "loss": 11.9135,
      "step": 31367
    },
    {
      "epoch": 1.7081129648159115,
      "grad_norm": 0.5762829539871385,
      "learning_rate": 1.0963892185260771e-05,
      "loss": 11.8354,
      "step": 31368
    },
    {
      "epoch": 1.7081674188124945,
      "grad_norm": 0.5604912212589517,
      "learning_rate": 1.0959877974981503e-05,
      "loss": 11.7911,
      "step": 31369
    },
    {
      "epoch": 1.7082218728090774,
      "grad_norm": 0.5599812857831126,
      "learning_rate": 1.0955864457093145e-05,
      "loss": 11.807,
      "step": 31370
    },
    {
      "epoch": 1.7082763268056604,
      "grad_norm": 0.5705282747501176,
      "learning_rate": 1.0951851631626931e-05,
      "loss": 11.7675,
      "step": 31371
    },
    {
      "epoch": 1.7083307808022434,
      "grad_norm": 0.5512783051901707,
      "learning_rate": 1.0947839498614032e-05,
      "loss": 11.8431,
      "step": 31372
    },
    {
      "epoch": 1.7083852347988264,
      "grad_norm": 0.5521971898170139,
      "learning_rate": 1.0943828058085693e-05,
      "loss": 11.774,
      "step": 31373
    },
    {
      "epoch": 1.7084396887954094,
      "grad_norm": 0.49511653281167395,
      "learning_rate": 1.0939817310073064e-05,
      "loss": 11.7015,
      "step": 31374
    },
    {
      "epoch": 1.7084941427919924,
      "grad_norm": 0.5341521834294026,
      "learning_rate": 1.0935807254607344e-05,
      "loss": 11.737,
      "step": 31375
    },
    {
      "epoch": 1.7085485967885754,
      "grad_norm": 0.49326756019797263,
      "learning_rate": 1.0931797891719742e-05,
      "loss": 11.6405,
      "step": 31376
    },
    {
      "epoch": 1.7086030507851586,
      "grad_norm": 0.5888327481424543,
      "learning_rate": 1.092778922144142e-05,
      "loss": 11.7765,
      "step": 31377
    },
    {
      "epoch": 1.7086575047817416,
      "grad_norm": 0.5394675366004322,
      "learning_rate": 1.0923781243803533e-05,
      "loss": 11.7522,
      "step": 31378
    },
    {
      "epoch": 1.7087119587783246,
      "grad_norm": 0.6782495984309581,
      "learning_rate": 1.0919773958837266e-05,
      "loss": 11.8773,
      "step": 31379
    },
    {
      "epoch": 1.7087664127749076,
      "grad_norm": 0.5195367539642743,
      "learning_rate": 1.091576736657377e-05,
      "loss": 11.8558,
      "step": 31380
    },
    {
      "epoch": 1.7088208667714906,
      "grad_norm": 0.5677555429094424,
      "learning_rate": 1.0911761467044212e-05,
      "loss": 11.7823,
      "step": 31381
    },
    {
      "epoch": 1.7088753207680736,
      "grad_norm": 0.5161460961865567,
      "learning_rate": 1.0907756260279734e-05,
      "loss": 11.8444,
      "step": 31382
    },
    {
      "epoch": 1.7089297747646568,
      "grad_norm": 0.5625333123438224,
      "learning_rate": 1.0903751746311485e-05,
      "loss": 11.7917,
      "step": 31383
    },
    {
      "epoch": 1.7089842287612398,
      "grad_norm": 0.5187215304958623,
      "learning_rate": 1.08997479251706e-05,
      "loss": 11.8216,
      "step": 31384
    },
    {
      "epoch": 1.7090386827578228,
      "grad_norm": 0.5733631636985802,
      "learning_rate": 1.0895744796888207e-05,
      "loss": 11.7785,
      "step": 31385
    },
    {
      "epoch": 1.7090931367544058,
      "grad_norm": 0.5444131973315395,
      "learning_rate": 1.0891742361495472e-05,
      "loss": 11.7844,
      "step": 31386
    },
    {
      "epoch": 1.7091475907509888,
      "grad_norm": 0.5296870045464985,
      "learning_rate": 1.0887740619023489e-05,
      "loss": 11.8324,
      "step": 31387
    },
    {
      "epoch": 1.7092020447475718,
      "grad_norm": 0.5362580175654252,
      "learning_rate": 1.0883739569503382e-05,
      "loss": 11.8728,
      "step": 31388
    },
    {
      "epoch": 1.7092564987441548,
      "grad_norm": 0.5050825661817893,
      "learning_rate": 1.0879739212966233e-05,
      "loss": 11.6665,
      "step": 31389
    },
    {
      "epoch": 1.7093109527407377,
      "grad_norm": 0.5089122476644603,
      "learning_rate": 1.0875739549443186e-05,
      "loss": 11.6489,
      "step": 31390
    },
    {
      "epoch": 1.7093654067373207,
      "grad_norm": 0.5137015693380984,
      "learning_rate": 1.087174057896535e-05,
      "loss": 11.7657,
      "step": 31391
    },
    {
      "epoch": 1.7094198607339037,
      "grad_norm": 0.6671127362235856,
      "learning_rate": 1.0867742301563788e-05,
      "loss": 11.8941,
      "step": 31392
    },
    {
      "epoch": 1.7094743147304867,
      "grad_norm": 0.5283914427623861,
      "learning_rate": 1.086374471726963e-05,
      "loss": 11.7407,
      "step": 31393
    },
    {
      "epoch": 1.7095287687270697,
      "grad_norm": 0.5791147656313416,
      "learning_rate": 1.085974782611392e-05,
      "loss": 11.9786,
      "step": 31394
    },
    {
      "epoch": 1.7095832227236527,
      "grad_norm": 0.5413001101306327,
      "learning_rate": 1.0855751628127775e-05,
      "loss": 11.7702,
      "step": 31395
    },
    {
      "epoch": 1.7096376767202357,
      "grad_norm": 0.5410949087423563,
      "learning_rate": 1.0851756123342239e-05,
      "loss": 11.8466,
      "step": 31396
    },
    {
      "epoch": 1.7096921307168187,
      "grad_norm": 0.5234902815370354,
      "learning_rate": 1.0847761311788418e-05,
      "loss": 11.8398,
      "step": 31397
    },
    {
      "epoch": 1.7097465847134017,
      "grad_norm": 0.5276060194192779,
      "learning_rate": 1.0843767193497356e-05,
      "loss": 11.8515,
      "step": 31398
    },
    {
      "epoch": 1.7098010387099847,
      "grad_norm": 0.5028690735956702,
      "learning_rate": 1.0839773768500095e-05,
      "loss": 11.8355,
      "step": 31399
    },
    {
      "epoch": 1.7098554927065677,
      "grad_norm": 0.5399868910470234,
      "learning_rate": 1.0835781036827697e-05,
      "loss": 11.8094,
      "step": 31400
    },
    {
      "epoch": 1.7099099467031509,
      "grad_norm": 0.6186854069007485,
      "learning_rate": 1.0831788998511238e-05,
      "loss": 11.8729,
      "step": 31401
    },
    {
      "epoch": 1.7099644006997339,
      "grad_norm": 0.5169071866840222,
      "learning_rate": 1.0827797653581728e-05,
      "loss": 11.7763,
      "step": 31402
    },
    {
      "epoch": 1.7100188546963169,
      "grad_norm": 0.5749566765483729,
      "learning_rate": 1.082380700207024e-05,
      "loss": 11.8186,
      "step": 31403
    },
    {
      "epoch": 1.7100733086928999,
      "grad_norm": 0.5346734206873114,
      "learning_rate": 1.0819817044007764e-05,
      "loss": 11.8243,
      "step": 31404
    },
    {
      "epoch": 1.7101277626894829,
      "grad_norm": 0.5721397827847031,
      "learning_rate": 1.0815827779425359e-05,
      "loss": 11.702,
      "step": 31405
    },
    {
      "epoch": 1.7101822166860658,
      "grad_norm": 0.5159035606249273,
      "learning_rate": 1.0811839208354014e-05,
      "loss": 11.8388,
      "step": 31406
    },
    {
      "epoch": 1.710236670682649,
      "grad_norm": 0.6457503850999484,
      "learning_rate": 1.0807851330824792e-05,
      "loss": 11.7218,
      "step": 31407
    },
    {
      "epoch": 1.710291124679232,
      "grad_norm": 0.5418568786335738,
      "learning_rate": 1.0803864146868669e-05,
      "loss": 11.9103,
      "step": 31408
    },
    {
      "epoch": 1.710345578675815,
      "grad_norm": 0.5107847895235661,
      "learning_rate": 1.0799877656516633e-05,
      "loss": 11.8648,
      "step": 31409
    },
    {
      "epoch": 1.710400032672398,
      "grad_norm": 0.5459476998412349,
      "learning_rate": 1.0795891859799734e-05,
      "loss": 11.758,
      "step": 31410
    },
    {
      "epoch": 1.710454486668981,
      "grad_norm": 0.6011175290544278,
      "learning_rate": 1.0791906756748916e-05,
      "loss": 11.8125,
      "step": 31411
    },
    {
      "epoch": 1.710508940665564,
      "grad_norm": 0.5648296862396707,
      "learning_rate": 1.0787922347395195e-05,
      "loss": 11.8706,
      "step": 31412
    },
    {
      "epoch": 1.710563394662147,
      "grad_norm": 0.6056109388881363,
      "learning_rate": 1.0783938631769564e-05,
      "loss": 11.8285,
      "step": 31413
    },
    {
      "epoch": 1.71061784865873,
      "grad_norm": 0.5171633534586365,
      "learning_rate": 1.077995560990297e-05,
      "loss": 11.8017,
      "step": 31414
    },
    {
      "epoch": 1.710672302655313,
      "grad_norm": 0.5490682626692884,
      "learning_rate": 1.0775973281826424e-05,
      "loss": 11.8877,
      "step": 31415
    },
    {
      "epoch": 1.710726756651896,
      "grad_norm": 0.5379869833675414,
      "learning_rate": 1.0771991647570856e-05,
      "loss": 11.7853,
      "step": 31416
    },
    {
      "epoch": 1.710781210648479,
      "grad_norm": 0.6122121634699095,
      "learning_rate": 1.0768010707167264e-05,
      "loss": 11.8365,
      "step": 31417
    },
    {
      "epoch": 1.710835664645062,
      "grad_norm": 0.6593521026631303,
      "learning_rate": 1.0764030460646579e-05,
      "loss": 11.9166,
      "step": 31418
    },
    {
      "epoch": 1.710890118641645,
      "grad_norm": 0.5613742894427809,
      "learning_rate": 1.0760050908039742e-05,
      "loss": 11.8424,
      "step": 31419
    },
    {
      "epoch": 1.710944572638228,
      "grad_norm": 0.5385458958570473,
      "learning_rate": 1.0756072049377742e-05,
      "loss": 11.6688,
      "step": 31420
    },
    {
      "epoch": 1.710999026634811,
      "grad_norm": 0.5693692277686349,
      "learning_rate": 1.075209388469146e-05,
      "loss": 11.8753,
      "step": 31421
    },
    {
      "epoch": 1.711053480631394,
      "grad_norm": 0.5054031846875073,
      "learning_rate": 1.0748116414011888e-05,
      "loss": 11.7141,
      "step": 31422
    },
    {
      "epoch": 1.711107934627977,
      "grad_norm": 0.5666660153708234,
      "learning_rate": 1.074413963736991e-05,
      "loss": 11.884,
      "step": 31423
    },
    {
      "epoch": 1.7111623886245602,
      "grad_norm": 0.5416718000326497,
      "learning_rate": 1.0740163554796478e-05,
      "loss": 11.8489,
      "step": 31424
    },
    {
      "epoch": 1.7112168426211432,
      "grad_norm": 0.5800941045494119,
      "learning_rate": 1.0736188166322513e-05,
      "loss": 11.8722,
      "step": 31425
    },
    {
      "epoch": 1.7112712966177261,
      "grad_norm": 0.5321302466771534,
      "learning_rate": 1.0732213471978902e-05,
      "loss": 11.7248,
      "step": 31426
    },
    {
      "epoch": 1.7113257506143091,
      "grad_norm": 0.6105716607849028,
      "learning_rate": 1.0728239471796586e-05,
      "loss": 11.9043,
      "step": 31427
    },
    {
      "epoch": 1.7113802046108921,
      "grad_norm": 0.512140905505257,
      "learning_rate": 1.072426616580645e-05,
      "loss": 11.7281,
      "step": 31428
    },
    {
      "epoch": 1.7114346586074751,
      "grad_norm": 0.5692570433891122,
      "learning_rate": 1.0720293554039374e-05,
      "loss": 11.8318,
      "step": 31429
    },
    {
      "epoch": 1.7114891126040583,
      "grad_norm": 0.5329991484708254,
      "learning_rate": 1.0716321636526295e-05,
      "loss": 11.7278,
      "step": 31430
    },
    {
      "epoch": 1.7115435666006413,
      "grad_norm": 0.5407219564206193,
      "learning_rate": 1.0712350413298045e-05,
      "loss": 11.8869,
      "step": 31431
    },
    {
      "epoch": 1.7115980205972243,
      "grad_norm": 0.5466139944200886,
      "learning_rate": 1.0708379884385545e-05,
      "loss": 11.8161,
      "step": 31432
    },
    {
      "epoch": 1.7116524745938073,
      "grad_norm": 0.5260275128237987,
      "learning_rate": 1.0704410049819647e-05,
      "loss": 11.7968,
      "step": 31433
    },
    {
      "epoch": 1.7117069285903903,
      "grad_norm": 0.5113930171897116,
      "learning_rate": 1.0700440909631226e-05,
      "loss": 11.7194,
      "step": 31434
    },
    {
      "epoch": 1.7117613825869733,
      "grad_norm": 0.5484905628324327,
      "learning_rate": 1.0696472463851181e-05,
      "loss": 11.7503,
      "step": 31435
    },
    {
      "epoch": 1.7118158365835563,
      "grad_norm": 0.6031907273961457,
      "learning_rate": 1.069250471251031e-05,
      "loss": 11.7069,
      "step": 31436
    },
    {
      "epoch": 1.7118702905801393,
      "grad_norm": 0.5901844988234785,
      "learning_rate": 1.0688537655639552e-05,
      "loss": 11.76,
      "step": 31437
    },
    {
      "epoch": 1.7119247445767223,
      "grad_norm": 0.5501451352658523,
      "learning_rate": 1.0684571293269652e-05,
      "loss": 11.7001,
      "step": 31438
    },
    {
      "epoch": 1.7119791985733053,
      "grad_norm": 0.5508893112889037,
      "learning_rate": 1.0680605625431506e-05,
      "loss": 11.7408,
      "step": 31439
    },
    {
      "epoch": 1.7120336525698883,
      "grad_norm": 0.5197146997984214,
      "learning_rate": 1.0676640652155979e-05,
      "loss": 11.7606,
      "step": 31440
    },
    {
      "epoch": 1.7120881065664713,
      "grad_norm": 0.6284286242143241,
      "learning_rate": 1.0672676373473845e-05,
      "loss": 11.8299,
      "step": 31441
    },
    {
      "epoch": 1.7121425605630542,
      "grad_norm": 0.5178726043955564,
      "learning_rate": 1.066871278941598e-05,
      "loss": 11.7827,
      "step": 31442
    },
    {
      "epoch": 1.7121970145596372,
      "grad_norm": 0.5233572969675926,
      "learning_rate": 1.066474990001316e-05,
      "loss": 11.7396,
      "step": 31443
    },
    {
      "epoch": 1.7122514685562202,
      "grad_norm": 0.5556031437831046,
      "learning_rate": 1.066078770529626e-05,
      "loss": 11.7558,
      "step": 31444
    },
    {
      "epoch": 1.7123059225528032,
      "grad_norm": 0.5105555233505923,
      "learning_rate": 1.0656826205296021e-05,
      "loss": 11.6405,
      "step": 31445
    },
    {
      "epoch": 1.7123603765493862,
      "grad_norm": 0.5691749437304826,
      "learning_rate": 1.0652865400043299e-05,
      "loss": 11.6957,
      "step": 31446
    },
    {
      "epoch": 1.7124148305459694,
      "grad_norm": 0.592482491596476,
      "learning_rate": 1.0648905289568912e-05,
      "loss": 11.9148,
      "step": 31447
    },
    {
      "epoch": 1.7124692845425524,
      "grad_norm": 0.5470973254217619,
      "learning_rate": 1.0644945873903579e-05,
      "loss": 11.8666,
      "step": 31448
    },
    {
      "epoch": 1.7125237385391354,
      "grad_norm": 0.5457120239878783,
      "learning_rate": 1.064098715307813e-05,
      "loss": 11.8624,
      "step": 31449
    },
    {
      "epoch": 1.7125781925357184,
      "grad_norm": 0.5779835299810903,
      "learning_rate": 1.0637029127123377e-05,
      "loss": 11.8025,
      "step": 31450
    },
    {
      "epoch": 1.7126326465323014,
      "grad_norm": 0.5365774391121249,
      "learning_rate": 1.063307179607006e-05,
      "loss": 11.7081,
      "step": 31451
    },
    {
      "epoch": 1.7126871005288844,
      "grad_norm": 0.5367586795865453,
      "learning_rate": 1.0629115159948966e-05,
      "loss": 11.7081,
      "step": 31452
    },
    {
      "epoch": 1.7127415545254676,
      "grad_norm": 0.5022251677931849,
      "learning_rate": 1.0625159218790847e-05,
      "loss": 11.6737,
      "step": 31453
    },
    {
      "epoch": 1.7127960085220506,
      "grad_norm": 0.5299995102224879,
      "learning_rate": 1.0621203972626504e-05,
      "loss": 11.7842,
      "step": 31454
    },
    {
      "epoch": 1.7128504625186336,
      "grad_norm": 0.5703988649620033,
      "learning_rate": 1.0617249421486642e-05,
      "loss": 11.8321,
      "step": 31455
    },
    {
      "epoch": 1.7129049165152166,
      "grad_norm": 0.5735867008308759,
      "learning_rate": 1.0613295565402038e-05,
      "loss": 11.8049,
      "step": 31456
    },
    {
      "epoch": 1.7129593705117996,
      "grad_norm": 0.5642239560494303,
      "learning_rate": 1.0609342404403477e-05,
      "loss": 11.7006,
      "step": 31457
    },
    {
      "epoch": 1.7130138245083826,
      "grad_norm": 0.5533283490365261,
      "learning_rate": 1.0605389938521626e-05,
      "loss": 11.826,
      "step": 31458
    },
    {
      "epoch": 1.7130682785049656,
      "grad_norm": 0.5759973187992862,
      "learning_rate": 1.060143816778727e-05,
      "loss": 11.9006,
      "step": 31459
    },
    {
      "epoch": 1.7131227325015486,
      "grad_norm": 0.5610058868292839,
      "learning_rate": 1.0597487092231096e-05,
      "loss": 11.8556,
      "step": 31460
    },
    {
      "epoch": 1.7131771864981316,
      "grad_norm": 0.6075491934975943,
      "learning_rate": 1.0593536711883866e-05,
      "loss": 11.8305,
      "step": 31461
    },
    {
      "epoch": 1.7132316404947145,
      "grad_norm": 0.5660814834145764,
      "learning_rate": 1.0589587026776304e-05,
      "loss": 11.6515,
      "step": 31462
    },
    {
      "epoch": 1.7132860944912975,
      "grad_norm": 0.5660420940437678,
      "learning_rate": 1.0585638036939083e-05,
      "loss": 11.661,
      "step": 31463
    },
    {
      "epoch": 1.7133405484878805,
      "grad_norm": 0.5647383866968578,
      "learning_rate": 1.0581689742402968e-05,
      "loss": 11.7554,
      "step": 31464
    },
    {
      "epoch": 1.7133950024844635,
      "grad_norm": 0.5337713886613089,
      "learning_rate": 1.05777421431986e-05,
      "loss": 11.8057,
      "step": 31465
    },
    {
      "epoch": 1.7134494564810465,
      "grad_norm": 0.6145542922416759,
      "learning_rate": 1.057379523935672e-05,
      "loss": 11.6535,
      "step": 31466
    },
    {
      "epoch": 1.7135039104776295,
      "grad_norm": 0.5585886103670066,
      "learning_rate": 1.056984903090802e-05,
      "loss": 11.8565,
      "step": 31467
    },
    {
      "epoch": 1.7135583644742125,
      "grad_norm": 0.5203912155534972,
      "learning_rate": 1.0565903517883135e-05,
      "loss": 11.7955,
      "step": 31468
    },
    {
      "epoch": 1.7136128184707955,
      "grad_norm": 0.5443404081948386,
      "learning_rate": 1.0561958700312812e-05,
      "loss": 11.8629,
      "step": 31469
    },
    {
      "epoch": 1.7136672724673785,
      "grad_norm": 0.5563624426346335,
      "learning_rate": 1.0558014578227671e-05,
      "loss": 11.7182,
      "step": 31470
    },
    {
      "epoch": 1.7137217264639617,
      "grad_norm": 0.5407574650049557,
      "learning_rate": 1.055407115165844e-05,
      "loss": 11.7053,
      "step": 31471
    },
    {
      "epoch": 1.7137761804605447,
      "grad_norm": 0.6411292787717409,
      "learning_rate": 1.0550128420635719e-05,
      "loss": 11.8317,
      "step": 31472
    },
    {
      "epoch": 1.7138306344571277,
      "grad_norm": 0.5362348810926708,
      "learning_rate": 1.0546186385190204e-05,
      "loss": 11.8486,
      "step": 31473
    },
    {
      "epoch": 1.7138850884537107,
      "grad_norm": 0.5584341649374477,
      "learning_rate": 1.0542245045352572e-05,
      "loss": 11.7925,
      "step": 31474
    },
    {
      "epoch": 1.7139395424502937,
      "grad_norm": 0.5238947057269379,
      "learning_rate": 1.053830440115341e-05,
      "loss": 11.8164,
      "step": 31475
    },
    {
      "epoch": 1.7139939964468769,
      "grad_norm": 0.627934181047507,
      "learning_rate": 1.0534364452623424e-05,
      "loss": 11.9633,
      "step": 31476
    },
    {
      "epoch": 1.7140484504434599,
      "grad_norm": 0.5471972408288592,
      "learning_rate": 1.0530425199793226e-05,
      "loss": 11.8643,
      "step": 31477
    },
    {
      "epoch": 1.7141029044400429,
      "grad_norm": 0.5573526478679477,
      "learning_rate": 1.0526486642693423e-05,
      "loss": 11.7174,
      "step": 31478
    },
    {
      "epoch": 1.7141573584366259,
      "grad_norm": 0.5196324123573932,
      "learning_rate": 1.0522548781354691e-05,
      "loss": 11.711,
      "step": 31479
    },
    {
      "epoch": 1.7142118124332089,
      "grad_norm": 0.5325676929214292,
      "learning_rate": 1.0518611615807594e-05,
      "loss": 11.7579,
      "step": 31480
    },
    {
      "epoch": 1.7142662664297919,
      "grad_norm": 0.574989754468607,
      "learning_rate": 1.0514675146082808e-05,
      "loss": 11.7042,
      "step": 31481
    },
    {
      "epoch": 1.7143207204263748,
      "grad_norm": 0.5475284050653473,
      "learning_rate": 1.0510739372210888e-05,
      "loss": 11.7333,
      "step": 31482
    },
    {
      "epoch": 1.7143751744229578,
      "grad_norm": 0.5267157853026788,
      "learning_rate": 1.0506804294222473e-05,
      "loss": 11.8675,
      "step": 31483
    },
    {
      "epoch": 1.7144296284195408,
      "grad_norm": 0.5417543576027039,
      "learning_rate": 1.0502869912148172e-05,
      "loss": 11.777,
      "step": 31484
    },
    {
      "epoch": 1.7144840824161238,
      "grad_norm": 0.5439329150693312,
      "learning_rate": 1.049893622601854e-05,
      "loss": 11.803,
      "step": 31485
    },
    {
      "epoch": 1.7145385364127068,
      "grad_norm": 0.6262666704501959,
      "learning_rate": 1.0495003235864209e-05,
      "loss": 11.8466,
      "step": 31486
    },
    {
      "epoch": 1.7145929904092898,
      "grad_norm": 0.5329814841722585,
      "learning_rate": 1.0491070941715752e-05,
      "loss": 11.748,
      "step": 31487
    },
    {
      "epoch": 1.7146474444058728,
      "grad_norm": 0.5956165880943686,
      "learning_rate": 1.0487139343603702e-05,
      "loss": 11.7892,
      "step": 31488
    },
    {
      "epoch": 1.7147018984024558,
      "grad_norm": 0.5532086445924472,
      "learning_rate": 1.04832084415587e-05,
      "loss": 11.7403,
      "step": 31489
    },
    {
      "epoch": 1.7147563523990388,
      "grad_norm": 0.5494991855615453,
      "learning_rate": 1.0479278235611267e-05,
      "loss": 11.7501,
      "step": 31490
    },
    {
      "epoch": 1.7148108063956218,
      "grad_norm": 0.5586482148519528,
      "learning_rate": 1.047534872579199e-05,
      "loss": 11.7576,
      "step": 31491
    },
    {
      "epoch": 1.7148652603922048,
      "grad_norm": 0.6106049331093408,
      "learning_rate": 1.0471419912131396e-05,
      "loss": 11.7508,
      "step": 31492
    },
    {
      "epoch": 1.7149197143887878,
      "grad_norm": 0.5640251305216247,
      "learning_rate": 1.0467491794660066e-05,
      "loss": 11.7891,
      "step": 31493
    },
    {
      "epoch": 1.714974168385371,
      "grad_norm": 0.5420770331907155,
      "learning_rate": 1.046356437340853e-05,
      "loss": 11.8908,
      "step": 31494
    },
    {
      "epoch": 1.715028622381954,
      "grad_norm": 0.5660027284710281,
      "learning_rate": 1.0459637648407328e-05,
      "loss": 11.8144,
      "step": 31495
    },
    {
      "epoch": 1.715083076378537,
      "grad_norm": 0.580241096977362,
      "learning_rate": 1.0455711619687014e-05,
      "loss": 11.8834,
      "step": 31496
    },
    {
      "epoch": 1.71513753037512,
      "grad_norm": 0.5168530731885738,
      "learning_rate": 1.0451786287278097e-05,
      "loss": 11.8216,
      "step": 31497
    },
    {
      "epoch": 1.715191984371703,
      "grad_norm": 0.6908020811749591,
      "learning_rate": 1.0447861651211099e-05,
      "loss": 11.9295,
      "step": 31498
    },
    {
      "epoch": 1.715246438368286,
      "grad_norm": 0.523079280413465,
      "learning_rate": 1.0443937711516571e-05,
      "loss": 11.7284,
      "step": 31499
    },
    {
      "epoch": 1.7153008923648692,
      "grad_norm": 0.6298982791860973,
      "learning_rate": 1.0440014468224968e-05,
      "loss": 11.745,
      "step": 31500
    },
    {
      "epoch": 1.7153553463614521,
      "grad_norm": 0.5342697318723727,
      "learning_rate": 1.0436091921366863e-05,
      "loss": 11.8669,
      "step": 31501
    },
    {
      "epoch": 1.7154098003580351,
      "grad_norm": 0.6044304543034712,
      "learning_rate": 1.0432170070972702e-05,
      "loss": 11.8372,
      "step": 31502
    },
    {
      "epoch": 1.7154642543546181,
      "grad_norm": 0.5340018605607262,
      "learning_rate": 1.0428248917073025e-05,
      "loss": 11.7431,
      "step": 31503
    },
    {
      "epoch": 1.7155187083512011,
      "grad_norm": 0.6007453410235755,
      "learning_rate": 1.0424328459698285e-05,
      "loss": 11.8048,
      "step": 31504
    },
    {
      "epoch": 1.7155731623477841,
      "grad_norm": 0.5575050090038742,
      "learning_rate": 1.0420408698878992e-05,
      "loss": 11.8508,
      "step": 31505
    },
    {
      "epoch": 1.7156276163443671,
      "grad_norm": 0.5591948108676068,
      "learning_rate": 1.0416489634645644e-05,
      "loss": 11.813,
      "step": 31506
    },
    {
      "epoch": 1.71568207034095,
      "grad_norm": 0.5707980019885782,
      "learning_rate": 1.0412571267028704e-05,
      "loss": 11.8364,
      "step": 31507
    },
    {
      "epoch": 1.715736524337533,
      "grad_norm": 0.5400907530965886,
      "learning_rate": 1.0408653596058615e-05,
      "loss": 11.7201,
      "step": 31508
    },
    {
      "epoch": 1.715790978334116,
      "grad_norm": 0.5476077115514427,
      "learning_rate": 1.0404736621765854e-05,
      "loss": 11.7321,
      "step": 31509
    },
    {
      "epoch": 1.715845432330699,
      "grad_norm": 0.5132813267877665,
      "learning_rate": 1.0400820344180884e-05,
      "loss": 11.8285,
      "step": 31510
    },
    {
      "epoch": 1.715899886327282,
      "grad_norm": 0.5453485699056855,
      "learning_rate": 1.039690476333418e-05,
      "loss": 11.8487,
      "step": 31511
    },
    {
      "epoch": 1.715954340323865,
      "grad_norm": 0.5177560533775238,
      "learning_rate": 1.0392989879256142e-05,
      "loss": 11.6971,
      "step": 31512
    },
    {
      "epoch": 1.716008794320448,
      "grad_norm": 0.5625019384813693,
      "learning_rate": 1.038907569197728e-05,
      "loss": 11.7111,
      "step": 31513
    },
    {
      "epoch": 1.716063248317031,
      "grad_norm": 0.5604026690973885,
      "learning_rate": 1.0385162201527954e-05,
      "loss": 11.8006,
      "step": 31514
    },
    {
      "epoch": 1.716117702313614,
      "grad_norm": 0.6216431233268295,
      "learning_rate": 1.0381249407938664e-05,
      "loss": 11.7218,
      "step": 31515
    },
    {
      "epoch": 1.716172156310197,
      "grad_norm": 0.5785475638762526,
      "learning_rate": 1.0377337311239787e-05,
      "loss": 11.8406,
      "step": 31516
    },
    {
      "epoch": 1.7162266103067803,
      "grad_norm": 0.6282864620618059,
      "learning_rate": 1.0373425911461764e-05,
      "loss": 11.9044,
      "step": 31517
    },
    {
      "epoch": 1.7162810643033632,
      "grad_norm": 0.5517508946106106,
      "learning_rate": 1.0369515208635061e-05,
      "loss": 11.6555,
      "step": 31518
    },
    {
      "epoch": 1.7163355182999462,
      "grad_norm": 0.5468075888658445,
      "learning_rate": 1.0365605202789986e-05,
      "loss": 11.6919,
      "step": 31519
    },
    {
      "epoch": 1.7163899722965292,
      "grad_norm": 0.5434999254704422,
      "learning_rate": 1.0361695893957001e-05,
      "loss": 11.801,
      "step": 31520
    },
    {
      "epoch": 1.7164444262931122,
      "grad_norm": 0.49435438127943954,
      "learning_rate": 1.035778728216652e-05,
      "loss": 11.8131,
      "step": 31521
    },
    {
      "epoch": 1.7164988802896952,
      "grad_norm": 0.5170349555973794,
      "learning_rate": 1.0353879367448905e-05,
      "loss": 11.6788,
      "step": 31522
    },
    {
      "epoch": 1.7165533342862784,
      "grad_norm": 0.5494413014674582,
      "learning_rate": 1.0349972149834574e-05,
      "loss": 11.8655,
      "step": 31523
    },
    {
      "epoch": 1.7166077882828614,
      "grad_norm": 0.532723027958204,
      "learning_rate": 1.0346065629353874e-05,
      "loss": 11.7205,
      "step": 31524
    },
    {
      "epoch": 1.7166622422794444,
      "grad_norm": 0.5396919734135337,
      "learning_rate": 1.034215980603721e-05,
      "loss": 11.6964,
      "step": 31525
    },
    {
      "epoch": 1.7167166962760274,
      "grad_norm": 0.5429105082128733,
      "learning_rate": 1.0338254679914939e-05,
      "loss": 11.8505,
      "step": 31526
    },
    {
      "epoch": 1.7167711502726104,
      "grad_norm": 0.5963999813032983,
      "learning_rate": 1.0334350251017455e-05,
      "loss": 11.7538,
      "step": 31527
    },
    {
      "epoch": 1.7168256042691934,
      "grad_norm": 0.6194946283530616,
      "learning_rate": 1.0330446519375104e-05,
      "loss": 11.8054,
      "step": 31528
    },
    {
      "epoch": 1.7168800582657764,
      "grad_norm": 0.4968158802263791,
      "learning_rate": 1.0326543485018214e-05,
      "loss": 11.7892,
      "step": 31529
    },
    {
      "epoch": 1.7169345122623594,
      "grad_norm": 0.5616479182415575,
      "learning_rate": 1.0322641147977185e-05,
      "loss": 11.7664,
      "step": 31530
    },
    {
      "epoch": 1.7169889662589424,
      "grad_norm": 0.5527259225655293,
      "learning_rate": 1.0318739508282305e-05,
      "loss": 11.869,
      "step": 31531
    },
    {
      "epoch": 1.7170434202555254,
      "grad_norm": 0.5166681923881729,
      "learning_rate": 1.0314838565963958e-05,
      "loss": 11.9058,
      "step": 31532
    },
    {
      "epoch": 1.7170978742521084,
      "grad_norm": 0.5237575025103077,
      "learning_rate": 1.0310938321052477e-05,
      "loss": 11.8324,
      "step": 31533
    },
    {
      "epoch": 1.7171523282486914,
      "grad_norm": 0.5178726626154517,
      "learning_rate": 1.030703877357817e-05,
      "loss": 11.8695,
      "step": 31534
    },
    {
      "epoch": 1.7172067822452743,
      "grad_norm": 0.5688868199917441,
      "learning_rate": 1.0303139923571393e-05,
      "loss": 11.833,
      "step": 31535
    },
    {
      "epoch": 1.7172612362418573,
      "grad_norm": 0.5133095572706747,
      "learning_rate": 1.0299241771062406e-05,
      "loss": 11.7907,
      "step": 31536
    },
    {
      "epoch": 1.7173156902384403,
      "grad_norm": 0.5517265001372719,
      "learning_rate": 1.029534431608159e-05,
      "loss": 11.8537,
      "step": 31537
    },
    {
      "epoch": 1.7173701442350233,
      "grad_norm": 0.6458168601278518,
      "learning_rate": 1.0291447558659218e-05,
      "loss": 11.8367,
      "step": 31538
    },
    {
      "epoch": 1.7174245982316063,
      "grad_norm": 0.6796053357856466,
      "learning_rate": 1.0287551498825575e-05,
      "loss": 11.7266,
      "step": 31539
    },
    {
      "epoch": 1.7174790522281893,
      "grad_norm": 0.5672932565031154,
      "learning_rate": 1.0283656136610997e-05,
      "loss": 11.7974,
      "step": 31540
    },
    {
      "epoch": 1.7175335062247725,
      "grad_norm": 0.5717451313671872,
      "learning_rate": 1.0279761472045735e-05,
      "loss": 11.744,
      "step": 31541
    },
    {
      "epoch": 1.7175879602213555,
      "grad_norm": 0.5334852185932999,
      "learning_rate": 1.0275867505160108e-05,
      "loss": 11.8491,
      "step": 31542
    },
    {
      "epoch": 1.7176424142179385,
      "grad_norm": 0.590529124136015,
      "learning_rate": 1.0271974235984372e-05,
      "loss": 11.8387,
      "step": 31543
    },
    {
      "epoch": 1.7176968682145215,
      "grad_norm": 0.5984197892987017,
      "learning_rate": 1.0268081664548802e-05,
      "loss": 11.8591,
      "step": 31544
    },
    {
      "epoch": 1.7177513222111045,
      "grad_norm": 0.5299065613184486,
      "learning_rate": 1.0264189790883693e-05,
      "loss": 11.5362,
      "step": 31545
    },
    {
      "epoch": 1.7178057762076877,
      "grad_norm": 0.5150687628781443,
      "learning_rate": 1.0260298615019281e-05,
      "loss": 11.6615,
      "step": 31546
    },
    {
      "epoch": 1.7178602302042707,
      "grad_norm": 0.575100554428203,
      "learning_rate": 1.0256408136985862e-05,
      "loss": 11.8952,
      "step": 31547
    },
    {
      "epoch": 1.7179146842008537,
      "grad_norm": 0.5113283756035647,
      "learning_rate": 1.0252518356813657e-05,
      "loss": 11.7354,
      "step": 31548
    },
    {
      "epoch": 1.7179691381974367,
      "grad_norm": 0.5525406438555145,
      "learning_rate": 1.0248629274532895e-05,
      "loss": 11.8523,
      "step": 31549
    },
    {
      "epoch": 1.7180235921940197,
      "grad_norm": 0.5285956182745611,
      "learning_rate": 1.0244740890173865e-05,
      "loss": 11.6948,
      "step": 31550
    },
    {
      "epoch": 1.7180780461906027,
      "grad_norm": 0.5017862131373222,
      "learning_rate": 1.0240853203766764e-05,
      "loss": 11.7097,
      "step": 31551
    },
    {
      "epoch": 1.7181325001871857,
      "grad_norm": 0.6261660800804799,
      "learning_rate": 1.023696621534187e-05,
      "loss": 11.8221,
      "step": 31552
    },
    {
      "epoch": 1.7181869541837687,
      "grad_norm": 0.5545612695863086,
      "learning_rate": 1.0233079924929346e-05,
      "loss": 11.922,
      "step": 31553
    },
    {
      "epoch": 1.7182414081803516,
      "grad_norm": 0.5104973443489229,
      "learning_rate": 1.0229194332559456e-05,
      "loss": 11.9069,
      "step": 31554
    },
    {
      "epoch": 1.7182958621769346,
      "grad_norm": 0.5420941901474664,
      "learning_rate": 1.0225309438262421e-05,
      "loss": 11.7832,
      "step": 31555
    },
    {
      "epoch": 1.7183503161735176,
      "grad_norm": 0.553564900845983,
      "learning_rate": 1.0221425242068417e-05,
      "loss": 11.9343,
      "step": 31556
    },
    {
      "epoch": 1.7184047701701006,
      "grad_norm": 0.5617533403323572,
      "learning_rate": 1.0217541744007687e-05,
      "loss": 11.7281,
      "step": 31557
    },
    {
      "epoch": 1.7184592241666836,
      "grad_norm": 0.5345195734226562,
      "learning_rate": 1.0213658944110404e-05,
      "loss": 11.7387,
      "step": 31558
    },
    {
      "epoch": 1.7185136781632666,
      "grad_norm": 0.5405528268589259,
      "learning_rate": 1.020977684240675e-05,
      "loss": 11.7239,
      "step": 31559
    },
    {
      "epoch": 1.7185681321598496,
      "grad_norm": 0.5258780473731511,
      "learning_rate": 1.0205895438926949e-05,
      "loss": 11.6759,
      "step": 31560
    },
    {
      "epoch": 1.7186225861564326,
      "grad_norm": 0.5641976631201479,
      "learning_rate": 1.0202014733701138e-05,
      "loss": 11.7516,
      "step": 31561
    },
    {
      "epoch": 1.7186770401530156,
      "grad_norm": 0.5717027721752065,
      "learning_rate": 1.0198134726759545e-05,
      "loss": 11.7368,
      "step": 31562
    },
    {
      "epoch": 1.7187314941495986,
      "grad_norm": 0.6088941775585593,
      "learning_rate": 1.0194255418132292e-05,
      "loss": 11.9148,
      "step": 31563
    },
    {
      "epoch": 1.7187859481461818,
      "grad_norm": 0.5165406297448353,
      "learning_rate": 1.01903768078496e-05,
      "loss": 11.7316,
      "step": 31564
    },
    {
      "epoch": 1.7188404021427648,
      "grad_norm": 0.5166590946357978,
      "learning_rate": 1.0186498895941566e-05,
      "loss": 11.7439,
      "step": 31565
    },
    {
      "epoch": 1.7188948561393478,
      "grad_norm": 0.5265213241462661,
      "learning_rate": 1.0182621682438386e-05,
      "loss": 11.8408,
      "step": 31566
    },
    {
      "epoch": 1.7189493101359308,
      "grad_norm": 0.5631334509084922,
      "learning_rate": 1.0178745167370218e-05,
      "loss": 11.8852,
      "step": 31567
    },
    {
      "epoch": 1.7190037641325138,
      "grad_norm": 0.579931971715743,
      "learning_rate": 1.017486935076719e-05,
      "loss": 11.8404,
      "step": 31568
    },
    {
      "epoch": 1.7190582181290968,
      "grad_norm": 0.6461344153584181,
      "learning_rate": 1.0170994232659425e-05,
      "loss": 11.8433,
      "step": 31569
    },
    {
      "epoch": 1.71911267212568,
      "grad_norm": 0.5633979225841509,
      "learning_rate": 1.0167119813077097e-05,
      "loss": 11.7002,
      "step": 31570
    },
    {
      "epoch": 1.719167126122263,
      "grad_norm": 0.5530773534330833,
      "learning_rate": 1.0163246092050283e-05,
      "loss": 11.7217,
      "step": 31571
    },
    {
      "epoch": 1.719221580118846,
      "grad_norm": 0.5637321739375956,
      "learning_rate": 1.0159373069609157e-05,
      "loss": 11.7768,
      "step": 31572
    },
    {
      "epoch": 1.719276034115429,
      "grad_norm": 0.5270648099352024,
      "learning_rate": 1.0155500745783797e-05,
      "loss": 11.725,
      "step": 31573
    },
    {
      "epoch": 1.719330488112012,
      "grad_norm": 0.5248390565762538,
      "learning_rate": 1.0151629120604345e-05,
      "loss": 11.8033,
      "step": 31574
    },
    {
      "epoch": 1.719384942108595,
      "grad_norm": 0.53525601514702,
      "learning_rate": 1.0147758194100864e-05,
      "loss": 11.7964,
      "step": 31575
    },
    {
      "epoch": 1.719439396105178,
      "grad_norm": 0.564528381781861,
      "learning_rate": 1.0143887966303512e-05,
      "loss": 11.7376,
      "step": 31576
    },
    {
      "epoch": 1.719493850101761,
      "grad_norm": 0.5347051331277457,
      "learning_rate": 1.0140018437242338e-05,
      "loss": 11.7006,
      "step": 31577
    },
    {
      "epoch": 1.719548304098344,
      "grad_norm": 0.5297048389539417,
      "learning_rate": 1.0136149606947466e-05,
      "loss": 11.7363,
      "step": 31578
    },
    {
      "epoch": 1.719602758094927,
      "grad_norm": 0.5562468619462452,
      "learning_rate": 1.0132281475448967e-05,
      "loss": 11.7608,
      "step": 31579
    },
    {
      "epoch": 1.71965721209151,
      "grad_norm": 0.5929683492025817,
      "learning_rate": 1.0128414042776901e-05,
      "loss": 11.92,
      "step": 31580
    },
    {
      "epoch": 1.719711666088093,
      "grad_norm": 0.5721271283055273,
      "learning_rate": 1.0124547308961352e-05,
      "loss": 11.6951,
      "step": 31581
    },
    {
      "epoch": 1.7197661200846759,
      "grad_norm": 0.5579167792201388,
      "learning_rate": 1.0120681274032417e-05,
      "loss": 11.7907,
      "step": 31582
    },
    {
      "epoch": 1.7198205740812589,
      "grad_norm": 0.497027615107683,
      "learning_rate": 1.0116815938020119e-05,
      "loss": 11.7754,
      "step": 31583
    },
    {
      "epoch": 1.7198750280778419,
      "grad_norm": 0.5914398622635827,
      "learning_rate": 1.0112951300954554e-05,
      "loss": 11.9825,
      "step": 31584
    },
    {
      "epoch": 1.7199294820744249,
      "grad_norm": 0.4773284932550976,
      "learning_rate": 1.0109087362865732e-05,
      "loss": 11.7669,
      "step": 31585
    },
    {
      "epoch": 1.7199839360710079,
      "grad_norm": 0.5543281504887684,
      "learning_rate": 1.0105224123783742e-05,
      "loss": 11.7636,
      "step": 31586
    },
    {
      "epoch": 1.720038390067591,
      "grad_norm": 0.5437165446321306,
      "learning_rate": 1.010136158373859e-05,
      "loss": 11.7309,
      "step": 31587
    },
    {
      "epoch": 1.720092844064174,
      "grad_norm": 0.5741886097642697,
      "learning_rate": 1.0097499742760342e-05,
      "loss": 11.8462,
      "step": 31588
    },
    {
      "epoch": 1.720147298060757,
      "grad_norm": 0.5216243349159138,
      "learning_rate": 1.0093638600879008e-05,
      "loss": 11.6872,
      "step": 31589
    },
    {
      "epoch": 1.72020175205734,
      "grad_norm": 0.5493658476374641,
      "learning_rate": 1.0089778158124596e-05,
      "loss": 11.7145,
      "step": 31590
    },
    {
      "epoch": 1.720256206053923,
      "grad_norm": 0.4942337410589276,
      "learning_rate": 1.0085918414527174e-05,
      "loss": 11.5876,
      "step": 31591
    },
    {
      "epoch": 1.720310660050506,
      "grad_norm": 0.49565025243563043,
      "learning_rate": 1.0082059370116714e-05,
      "loss": 11.7896,
      "step": 31592
    },
    {
      "epoch": 1.7203651140470892,
      "grad_norm": 0.5346386801223755,
      "learning_rate": 1.0078201024923228e-05,
      "loss": 11.7027,
      "step": 31593
    },
    {
      "epoch": 1.7204195680436722,
      "grad_norm": 0.5325015247587378,
      "learning_rate": 1.0074343378976758e-05,
      "loss": 11.767,
      "step": 31594
    },
    {
      "epoch": 1.7204740220402552,
      "grad_norm": 0.5011686389998581,
      "learning_rate": 1.0070486432307257e-05,
      "loss": 11.7314,
      "step": 31595
    },
    {
      "epoch": 1.7205284760368382,
      "grad_norm": 0.5931487508652592,
      "learning_rate": 1.0066630184944748e-05,
      "loss": 11.7359,
      "step": 31596
    },
    {
      "epoch": 1.7205829300334212,
      "grad_norm": 0.5578300443420816,
      "learning_rate": 1.0062774636919181e-05,
      "loss": 11.7218,
      "step": 31597
    },
    {
      "epoch": 1.7206373840300042,
      "grad_norm": 0.4853604650187636,
      "learning_rate": 1.005891978826059e-05,
      "loss": 11.648,
      "step": 31598
    },
    {
      "epoch": 1.7206918380265872,
      "grad_norm": 0.5662665237519126,
      "learning_rate": 1.0055065638998917e-05,
      "loss": 11.785,
      "step": 31599
    },
    {
      "epoch": 1.7207462920231702,
      "grad_norm": 0.583786061851586,
      "learning_rate": 1.0051212189164117e-05,
      "loss": 11.8159,
      "step": 31600
    },
    {
      "epoch": 1.7208007460197532,
      "grad_norm": 0.5268813172833314,
      "learning_rate": 1.0047359438786197e-05,
      "loss": 11.766,
      "step": 31601
    },
    {
      "epoch": 1.7208552000163362,
      "grad_norm": 0.5665939415294465,
      "learning_rate": 1.004350738789508e-05,
      "loss": 11.8127,
      "step": 31602
    },
    {
      "epoch": 1.7209096540129192,
      "grad_norm": 0.5773183052607285,
      "learning_rate": 1.0039656036520728e-05,
      "loss": 11.7662,
      "step": 31603
    },
    {
      "epoch": 1.7209641080095022,
      "grad_norm": 0.5510697257439507,
      "learning_rate": 1.003580538469312e-05,
      "loss": 11.8002,
      "step": 31604
    },
    {
      "epoch": 1.7210185620060852,
      "grad_norm": 0.580086336691622,
      "learning_rate": 1.0031955432442153e-05,
      "loss": 11.912,
      "step": 31605
    },
    {
      "epoch": 1.7210730160026682,
      "grad_norm": 0.5022698051681344,
      "learning_rate": 1.0028106179797813e-05,
      "loss": 11.6991,
      "step": 31606
    },
    {
      "epoch": 1.7211274699992511,
      "grad_norm": 0.5138518962521057,
      "learning_rate": 1.002425762678999e-05,
      "loss": 11.7953,
      "step": 31607
    },
    {
      "epoch": 1.7211819239958341,
      "grad_norm": 0.5711216960217186,
      "learning_rate": 1.0020409773448637e-05,
      "loss": 11.7749,
      "step": 31608
    },
    {
      "epoch": 1.7212363779924171,
      "grad_norm": 0.5005861228314973,
      "learning_rate": 1.0016562619803682e-05,
      "loss": 11.5601,
      "step": 31609
    },
    {
      "epoch": 1.7212908319890003,
      "grad_norm": 0.5337994292374386,
      "learning_rate": 1.0012716165884994e-05,
      "loss": 11.8073,
      "step": 31610
    },
    {
      "epoch": 1.7213452859855833,
      "grad_norm": 0.6007049351191095,
      "learning_rate": 1.0008870411722537e-05,
      "loss": 11.7589,
      "step": 31611
    },
    {
      "epoch": 1.7213997399821663,
      "grad_norm": 0.5666830835561676,
      "learning_rate": 1.0005025357346187e-05,
      "loss": 11.8024,
      "step": 31612
    },
    {
      "epoch": 1.7214541939787493,
      "grad_norm": 0.6016147626228205,
      "learning_rate": 1.0001181002785864e-05,
      "loss": 11.9061,
      "step": 31613
    },
    {
      "epoch": 1.7215086479753323,
      "grad_norm": 0.5700946700385966,
      "learning_rate": 9.997337348071423e-06,
      "loss": 11.7835,
      "step": 31614
    },
    {
      "epoch": 1.7215631019719153,
      "grad_norm": 0.6276406480298716,
      "learning_rate": 9.993494393232795e-06,
      "loss": 11.8186,
      "step": 31615
    },
    {
      "epoch": 1.7216175559684985,
      "grad_norm": 0.5382441499206801,
      "learning_rate": 9.989652138299854e-06,
      "loss": 11.6891,
      "step": 31616
    },
    {
      "epoch": 1.7216720099650815,
      "grad_norm": 0.5526851264588184,
      "learning_rate": 9.985810583302457e-06,
      "loss": 11.7511,
      "step": 31617
    },
    {
      "epoch": 1.7217264639616645,
      "grad_norm": 0.5077383918843621,
      "learning_rate": 9.981969728270524e-06,
      "loss": 11.8187,
      "step": 31618
    },
    {
      "epoch": 1.7217809179582475,
      "grad_norm": 0.559844989293298,
      "learning_rate": 9.978129573233875e-06,
      "loss": 11.7314,
      "step": 31619
    },
    {
      "epoch": 1.7218353719548305,
      "grad_norm": 0.5416457035722602,
      "learning_rate": 9.974290118222374e-06,
      "loss": 11.8057,
      "step": 31620
    },
    {
      "epoch": 1.7218898259514135,
      "grad_norm": 0.49255162000816144,
      "learning_rate": 9.970451363265909e-06,
      "loss": 11.821,
      "step": 31621
    },
    {
      "epoch": 1.7219442799479965,
      "grad_norm": 0.5756449643012322,
      "learning_rate": 9.9666133083943e-06,
      "loss": 11.7517,
      "step": 31622
    },
    {
      "epoch": 1.7219987339445795,
      "grad_norm": 0.554707897770957,
      "learning_rate": 9.962775953637416e-06,
      "loss": 11.7847,
      "step": 31623
    },
    {
      "epoch": 1.7220531879411625,
      "grad_norm": 0.5579892266497304,
      "learning_rate": 9.95893929902506e-06,
      "loss": 11.7983,
      "step": 31624
    },
    {
      "epoch": 1.7221076419377455,
      "grad_norm": 0.5721832657330594,
      "learning_rate": 9.955103344587125e-06,
      "loss": 11.78,
      "step": 31625
    },
    {
      "epoch": 1.7221620959343285,
      "grad_norm": 0.5286547776030073,
      "learning_rate": 9.951268090353382e-06,
      "loss": 11.8571,
      "step": 31626
    },
    {
      "epoch": 1.7222165499309114,
      "grad_norm": 0.6085563065468181,
      "learning_rate": 9.947433536353679e-06,
      "loss": 11.9021,
      "step": 31627
    },
    {
      "epoch": 1.7222710039274944,
      "grad_norm": 0.49485171895229996,
      "learning_rate": 9.943599682617865e-06,
      "loss": 11.7653,
      "step": 31628
    },
    {
      "epoch": 1.7223254579240774,
      "grad_norm": 0.5236511628822097,
      "learning_rate": 9.939766529175698e-06,
      "loss": 11.6727,
      "step": 31629
    },
    {
      "epoch": 1.7223799119206604,
      "grad_norm": 0.5440952775692672,
      "learning_rate": 9.935934076057008e-06,
      "loss": 11.8226,
      "step": 31630
    },
    {
      "epoch": 1.7224343659172434,
      "grad_norm": 0.5611291784323829,
      "learning_rate": 9.932102323291603e-06,
      "loss": 11.8492,
      "step": 31631
    },
    {
      "epoch": 1.7224888199138264,
      "grad_norm": 0.5407081392247196,
      "learning_rate": 9.928271270909273e-06,
      "loss": 11.7461,
      "step": 31632
    },
    {
      "epoch": 1.7225432739104094,
      "grad_norm": 0.5431104522580013,
      "learning_rate": 9.924440918939814e-06,
      "loss": 11.8053,
      "step": 31633
    },
    {
      "epoch": 1.7225977279069926,
      "grad_norm": 0.5648261138440481,
      "learning_rate": 9.920611267413005e-06,
      "loss": 11.7403,
      "step": 31634
    },
    {
      "epoch": 1.7226521819035756,
      "grad_norm": 0.6562552367201177,
      "learning_rate": 9.91678231635864e-06,
      "loss": 11.8344,
      "step": 31635
    },
    {
      "epoch": 1.7227066359001586,
      "grad_norm": 0.6069501439000724,
      "learning_rate": 9.912954065806468e-06,
      "loss": 11.7119,
      "step": 31636
    },
    {
      "epoch": 1.7227610898967416,
      "grad_norm": 0.5926715108090882,
      "learning_rate": 9.90912651578626e-06,
      "loss": 11.8739,
      "step": 31637
    },
    {
      "epoch": 1.7228155438933246,
      "grad_norm": 0.57333652997764,
      "learning_rate": 9.905299666327838e-06,
      "loss": 11.9253,
      "step": 31638
    },
    {
      "epoch": 1.7228699978899076,
      "grad_norm": 0.5179009446000575,
      "learning_rate": 9.90147351746088e-06,
      "loss": 11.7079,
      "step": 31639
    },
    {
      "epoch": 1.7229244518864908,
      "grad_norm": 0.5927678052247316,
      "learning_rate": 9.897648069215193e-06,
      "loss": 11.7076,
      "step": 31640
    },
    {
      "epoch": 1.7229789058830738,
      "grad_norm": 0.5420398795968947,
      "learning_rate": 9.893823321620488e-06,
      "loss": 11.8555,
      "step": 31641
    },
    {
      "epoch": 1.7230333598796568,
      "grad_norm": 0.7032982705210287,
      "learning_rate": 9.889999274706518e-06,
      "loss": 11.8216,
      "step": 31642
    },
    {
      "epoch": 1.7230878138762398,
      "grad_norm": 0.5456092851715043,
      "learning_rate": 9.886175928503038e-06,
      "loss": 11.8308,
      "step": 31643
    },
    {
      "epoch": 1.7231422678728228,
      "grad_norm": 0.5144023790602227,
      "learning_rate": 9.882353283039758e-06,
      "loss": 11.8272,
      "step": 31644
    },
    {
      "epoch": 1.7231967218694058,
      "grad_norm": 0.5219399078420015,
      "learning_rate": 9.87853133834643e-06,
      "loss": 11.7801,
      "step": 31645
    },
    {
      "epoch": 1.7232511758659887,
      "grad_norm": 0.5264655189569761,
      "learning_rate": 9.874710094452733e-06,
      "loss": 11.7471,
      "step": 31646
    },
    {
      "epoch": 1.7233056298625717,
      "grad_norm": 0.5354346228133944,
      "learning_rate": 9.870889551388419e-06,
      "loss": 11.7905,
      "step": 31647
    },
    {
      "epoch": 1.7233600838591547,
      "grad_norm": 0.5272933071145303,
      "learning_rate": 9.867069709183186e-06,
      "loss": 11.6881,
      "step": 31648
    },
    {
      "epoch": 1.7234145378557377,
      "grad_norm": 0.5897080081996334,
      "learning_rate": 9.863250567866721e-06,
      "loss": 11.7322,
      "step": 31649
    },
    {
      "epoch": 1.7234689918523207,
      "grad_norm": 0.5078046384524149,
      "learning_rate": 9.859432127468748e-06,
      "loss": 11.8635,
      "step": 31650
    },
    {
      "epoch": 1.7235234458489037,
      "grad_norm": 0.637272507312113,
      "learning_rate": 9.85561438801893e-06,
      "loss": 11.7572,
      "step": 31651
    },
    {
      "epoch": 1.7235778998454867,
      "grad_norm": 0.560055228726501,
      "learning_rate": 9.851797349546976e-06,
      "loss": 11.7605,
      "step": 31652
    },
    {
      "epoch": 1.7236323538420697,
      "grad_norm": 0.5062532132317071,
      "learning_rate": 9.847981012082574e-06,
      "loss": 11.6805,
      "step": 31653
    },
    {
      "epoch": 1.7236868078386527,
      "grad_norm": 0.5877989342666577,
      "learning_rate": 9.844165375655379e-06,
      "loss": 11.7646,
      "step": 31654
    },
    {
      "epoch": 1.7237412618352357,
      "grad_norm": 0.5330469216308327,
      "learning_rate": 9.840350440295088e-06,
      "loss": 11.6925,
      "step": 31655
    },
    {
      "epoch": 1.7237957158318187,
      "grad_norm": 0.5621134833465058,
      "learning_rate": 9.836536206031333e-06,
      "loss": 11.8281,
      "step": 31656
    },
    {
      "epoch": 1.723850169828402,
      "grad_norm": 0.5405881843850385,
      "learning_rate": 9.832722672893812e-06,
      "loss": 11.6365,
      "step": 31657
    },
    {
      "epoch": 1.7239046238249849,
      "grad_norm": 0.5139143593961488,
      "learning_rate": 9.82890984091216e-06,
      "loss": 11.7755,
      "step": 31658
    },
    {
      "epoch": 1.7239590778215679,
      "grad_norm": 0.5470150575024245,
      "learning_rate": 9.825097710116016e-06,
      "loss": 11.8374,
      "step": 31659
    },
    {
      "epoch": 1.7240135318181509,
      "grad_norm": 0.5799642034690877,
      "learning_rate": 9.821286280535048e-06,
      "loss": 11.8594,
      "step": 31660
    },
    {
      "epoch": 1.7240679858147339,
      "grad_norm": 0.5471452229170947,
      "learning_rate": 9.817475552198851e-06,
      "loss": 11.9307,
      "step": 31661
    },
    {
      "epoch": 1.7241224398113169,
      "grad_norm": 0.5604782843538757,
      "learning_rate": 9.813665525137117e-06,
      "loss": 11.8213,
      "step": 31662
    },
    {
      "epoch": 1.7241768938079,
      "grad_norm": 0.5269929859812372,
      "learning_rate": 9.80985619937943e-06,
      "loss": 11.797,
      "step": 31663
    },
    {
      "epoch": 1.724231347804483,
      "grad_norm": 0.5361352117910505,
      "learning_rate": 9.806047574955413e-06,
      "loss": 11.818,
      "step": 31664
    },
    {
      "epoch": 1.724285801801066,
      "grad_norm": 0.5594370648934125,
      "learning_rate": 9.80223965189473e-06,
      "loss": 11.7579,
      "step": 31665
    },
    {
      "epoch": 1.724340255797649,
      "grad_norm": 0.5921846916923454,
      "learning_rate": 9.798432430226923e-06,
      "loss": 11.9223,
      "step": 31666
    },
    {
      "epoch": 1.724394709794232,
      "grad_norm": 0.5474470645296436,
      "learning_rate": 9.794625909981659e-06,
      "loss": 11.7333,
      "step": 31667
    },
    {
      "epoch": 1.724449163790815,
      "grad_norm": 0.5149711932014519,
      "learning_rate": 9.790820091188502e-06,
      "loss": 11.8104,
      "step": 31668
    },
    {
      "epoch": 1.724503617787398,
      "grad_norm": 0.5614818687022267,
      "learning_rate": 9.787014973877062e-06,
      "loss": 11.8931,
      "step": 31669
    },
    {
      "epoch": 1.724558071783981,
      "grad_norm": 0.5239461447515954,
      "learning_rate": 9.783210558076928e-06,
      "loss": 11.7741,
      "step": 31670
    },
    {
      "epoch": 1.724612525780564,
      "grad_norm": 0.5675347104846662,
      "learning_rate": 9.77940684381765e-06,
      "loss": 11.7387,
      "step": 31671
    },
    {
      "epoch": 1.724666979777147,
      "grad_norm": 0.5974951287968702,
      "learning_rate": 9.775603831128865e-06,
      "loss": 11.8754,
      "step": 31672
    },
    {
      "epoch": 1.72472143377373,
      "grad_norm": 0.5715581040755615,
      "learning_rate": 9.77180152004009e-06,
      "loss": 11.872,
      "step": 31673
    },
    {
      "epoch": 1.724775887770313,
      "grad_norm": 0.5238339741704982,
      "learning_rate": 9.767999910580916e-06,
      "loss": 11.867,
      "step": 31674
    },
    {
      "epoch": 1.724830341766896,
      "grad_norm": 0.6017754906156638,
      "learning_rate": 9.764199002780927e-06,
      "loss": 11.8095,
      "step": 31675
    },
    {
      "epoch": 1.724884795763479,
      "grad_norm": 0.5378820399597718,
      "learning_rate": 9.760398796669646e-06,
      "loss": 11.8627,
      "step": 31676
    },
    {
      "epoch": 1.724939249760062,
      "grad_norm": 0.5688165944084097,
      "learning_rate": 9.756599292276646e-06,
      "loss": 11.9054,
      "step": 31677
    },
    {
      "epoch": 1.724993703756645,
      "grad_norm": 0.6024619360754615,
      "learning_rate": 9.752800489631453e-06,
      "loss": 11.9786,
      "step": 31678
    },
    {
      "epoch": 1.725048157753228,
      "grad_norm": 0.5228581285698074,
      "learning_rate": 9.74900238876364e-06,
      "loss": 11.7096,
      "step": 31679
    },
    {
      "epoch": 1.7251026117498112,
      "grad_norm": 0.5032368347871917,
      "learning_rate": 9.745204989702705e-06,
      "loss": 11.7223,
      "step": 31680
    },
    {
      "epoch": 1.7251570657463942,
      "grad_norm": 0.528420938205925,
      "learning_rate": 9.741408292478183e-06,
      "loss": 11.8796,
      "step": 31681
    },
    {
      "epoch": 1.7252115197429772,
      "grad_norm": 0.5160502954828745,
      "learning_rate": 9.737612297119625e-06,
      "loss": 11.7338,
      "step": 31682
    },
    {
      "epoch": 1.7252659737395601,
      "grad_norm": 0.5349271245456051,
      "learning_rate": 9.733817003656509e-06,
      "loss": 11.7757,
      "step": 31683
    },
    {
      "epoch": 1.7253204277361431,
      "grad_norm": 0.6173101001891643,
      "learning_rate": 9.7300224121184e-06,
      "loss": 11.8106,
      "step": 31684
    },
    {
      "epoch": 1.7253748817327261,
      "grad_norm": 0.5553358185095433,
      "learning_rate": 9.726228522534742e-06,
      "loss": 11.8653,
      "step": 31685
    },
    {
      "epoch": 1.7254293357293093,
      "grad_norm": 0.5359553888537437,
      "learning_rate": 9.722435334935077e-06,
      "loss": 11.7207,
      "step": 31686
    },
    {
      "epoch": 1.7254837897258923,
      "grad_norm": 0.5705403304738097,
      "learning_rate": 9.718642849348902e-06,
      "loss": 11.9162,
      "step": 31687
    },
    {
      "epoch": 1.7255382437224753,
      "grad_norm": 0.47644253008702425,
      "learning_rate": 9.714851065805697e-06,
      "loss": 11.6373,
      "step": 31688
    },
    {
      "epoch": 1.7255926977190583,
      "grad_norm": 0.5633304336926984,
      "learning_rate": 9.711059984334981e-06,
      "loss": 11.7309,
      "step": 31689
    },
    {
      "epoch": 1.7256471517156413,
      "grad_norm": 0.5367003899122796,
      "learning_rate": 9.707269604966162e-06,
      "loss": 11.7679,
      "step": 31690
    },
    {
      "epoch": 1.7257016057122243,
      "grad_norm": 0.5448976654034257,
      "learning_rate": 9.703479927728765e-06,
      "loss": 11.7898,
      "step": 31691
    },
    {
      "epoch": 1.7257560597088073,
      "grad_norm": 0.5791685649235563,
      "learning_rate": 9.699690952652275e-06,
      "loss": 11.8029,
      "step": 31692
    },
    {
      "epoch": 1.7258105137053903,
      "grad_norm": 0.5271671684126381,
      "learning_rate": 9.6959026797661e-06,
      "loss": 11.7772,
      "step": 31693
    },
    {
      "epoch": 1.7258649677019733,
      "grad_norm": 0.5772368975839575,
      "learning_rate": 9.692115109099754e-06,
      "loss": 11.8192,
      "step": 31694
    },
    {
      "epoch": 1.7259194216985563,
      "grad_norm": 0.5436844798966903,
      "learning_rate": 9.688328240682643e-06,
      "loss": 11.8391,
      "step": 31695
    },
    {
      "epoch": 1.7259738756951393,
      "grad_norm": 0.5332655215303967,
      "learning_rate": 9.684542074544256e-06,
      "loss": 11.8615,
      "step": 31696
    },
    {
      "epoch": 1.7260283296917223,
      "grad_norm": 0.5085579615833358,
      "learning_rate": 9.680756610714003e-06,
      "loss": 11.8357,
      "step": 31697
    },
    {
      "epoch": 1.7260827836883053,
      "grad_norm": 0.530327560360123,
      "learning_rate": 9.676971849221328e-06,
      "loss": 11.8331,
      "step": 31698
    },
    {
      "epoch": 1.7261372376848882,
      "grad_norm": 0.5679448804061265,
      "learning_rate": 9.673187790095706e-06,
      "loss": 11.8893,
      "step": 31699
    },
    {
      "epoch": 1.7261916916814712,
      "grad_norm": 0.5371532607973594,
      "learning_rate": 9.66940443336648e-06,
      "loss": 11.8694,
      "step": 31700
    },
    {
      "epoch": 1.7262461456780542,
      "grad_norm": 0.5364368743944987,
      "learning_rate": 9.665621779063127e-06,
      "loss": 11.811,
      "step": 31701
    },
    {
      "epoch": 1.7263005996746372,
      "grad_norm": 0.6120074762318003,
      "learning_rate": 9.661839827215058e-06,
      "loss": 11.9309,
      "step": 31702
    },
    {
      "epoch": 1.7263550536712202,
      "grad_norm": 0.5521704409754,
      "learning_rate": 9.658058577851658e-06,
      "loss": 11.7711,
      "step": 31703
    },
    {
      "epoch": 1.7264095076678034,
      "grad_norm": 0.49477285511138874,
      "learning_rate": 9.654278031002361e-06,
      "loss": 11.8737,
      "step": 31704
    },
    {
      "epoch": 1.7264639616643864,
      "grad_norm": 0.5873796201042932,
      "learning_rate": 9.650498186696522e-06,
      "loss": 11.9124,
      "step": 31705
    },
    {
      "epoch": 1.7265184156609694,
      "grad_norm": 0.5743880791251625,
      "learning_rate": 9.646719044963593e-06,
      "loss": 11.7838,
      "step": 31706
    },
    {
      "epoch": 1.7265728696575524,
      "grad_norm": 0.5289910373412918,
      "learning_rate": 9.642940605832906e-06,
      "loss": 11.7521,
      "step": 31707
    },
    {
      "epoch": 1.7266273236541354,
      "grad_norm": 0.5643291751214005,
      "learning_rate": 9.639162869333861e-06,
      "loss": 11.7548,
      "step": 31708
    },
    {
      "epoch": 1.7266817776507186,
      "grad_norm": 0.5648407876761691,
      "learning_rate": 9.63538583549588e-06,
      "loss": 11.8659,
      "step": 31709
    },
    {
      "epoch": 1.7267362316473016,
      "grad_norm": 0.6448956523557557,
      "learning_rate": 9.631609504348249e-06,
      "loss": 11.8444,
      "step": 31710
    },
    {
      "epoch": 1.7267906856438846,
      "grad_norm": 0.5195585514260834,
      "learning_rate": 9.627833875920411e-06,
      "loss": 11.7816,
      "step": 31711
    },
    {
      "epoch": 1.7268451396404676,
      "grad_norm": 0.5592209932938418,
      "learning_rate": 9.624058950241666e-06,
      "loss": 11.9012,
      "step": 31712
    },
    {
      "epoch": 1.7268995936370506,
      "grad_norm": 0.5576575172917734,
      "learning_rate": 9.62028472734139e-06,
      "loss": 12.0067,
      "step": 31713
    },
    {
      "epoch": 1.7269540476336336,
      "grad_norm": 0.577147121452383,
      "learning_rate": 9.616511207248957e-06,
      "loss": 11.7964,
      "step": 31714
    },
    {
      "epoch": 1.7270085016302166,
      "grad_norm": 0.5705233257587317,
      "learning_rate": 9.61273838999367e-06,
      "loss": 11.8475,
      "step": 31715
    },
    {
      "epoch": 1.7270629556267996,
      "grad_norm": 0.6145207127089715,
      "learning_rate": 9.608966275604913e-06,
      "loss": 11.8718,
      "step": 31716
    },
    {
      "epoch": 1.7271174096233826,
      "grad_norm": 0.5457264230962521,
      "learning_rate": 9.605194864111967e-06,
      "loss": 11.8942,
      "step": 31717
    },
    {
      "epoch": 1.7271718636199656,
      "grad_norm": 0.5731032537776786,
      "learning_rate": 9.601424155544214e-06,
      "loss": 11.7933,
      "step": 31718
    },
    {
      "epoch": 1.7272263176165485,
      "grad_norm": 0.5175332225203725,
      "learning_rate": 9.597654149930934e-06,
      "loss": 11.7578,
      "step": 31719
    },
    {
      "epoch": 1.7272807716131315,
      "grad_norm": 0.6721303858954103,
      "learning_rate": 9.593884847301437e-06,
      "loss": 12.0066,
      "step": 31720
    },
    {
      "epoch": 1.7273352256097145,
      "grad_norm": 0.6322488225549545,
      "learning_rate": 9.590116247685089e-06,
      "loss": 11.7774,
      "step": 31721
    },
    {
      "epoch": 1.7273896796062975,
      "grad_norm": 0.5490412921685415,
      "learning_rate": 9.586348351111118e-06,
      "loss": 11.6418,
      "step": 31722
    },
    {
      "epoch": 1.7274441336028805,
      "grad_norm": 0.5069457744941861,
      "learning_rate": 9.582581157608883e-06,
      "loss": 11.6878,
      "step": 31723
    },
    {
      "epoch": 1.7274985875994635,
      "grad_norm": 0.510792647693245,
      "learning_rate": 9.57881466720767e-06,
      "loss": 11.7458,
      "step": 31724
    },
    {
      "epoch": 1.7275530415960465,
      "grad_norm": 0.5589741024172136,
      "learning_rate": 9.575048879936732e-06,
      "loss": 11.8988,
      "step": 31725
    },
    {
      "epoch": 1.7276074955926295,
      "grad_norm": 0.5440348207592137,
      "learning_rate": 9.571283795825404e-06,
      "loss": 11.7806,
      "step": 31726
    },
    {
      "epoch": 1.7276619495892127,
      "grad_norm": 0.5228054075329174,
      "learning_rate": 9.567519414902926e-06,
      "loss": 11.7824,
      "step": 31727
    },
    {
      "epoch": 1.7277164035857957,
      "grad_norm": 0.5812598052893078,
      "learning_rate": 9.563755737198588e-06,
      "loss": 11.8072,
      "step": 31728
    },
    {
      "epoch": 1.7277708575823787,
      "grad_norm": 0.5124134013005092,
      "learning_rate": 9.559992762741666e-06,
      "loss": 11.7268,
      "step": 31729
    },
    {
      "epoch": 1.7278253115789617,
      "grad_norm": 0.5835164472508388,
      "learning_rate": 9.55623049156138e-06,
      "loss": 11.7904,
      "step": 31730
    },
    {
      "epoch": 1.7278797655755447,
      "grad_norm": 0.5085767572132961,
      "learning_rate": 9.55246892368703e-06,
      "loss": 11.8523,
      "step": 31731
    },
    {
      "epoch": 1.7279342195721277,
      "grad_norm": 0.576927234951472,
      "learning_rate": 9.548708059147827e-06,
      "loss": 11.7906,
      "step": 31732
    },
    {
      "epoch": 1.7279886735687109,
      "grad_norm": 0.597292849935191,
      "learning_rate": 9.544947897973066e-06,
      "loss": 11.7756,
      "step": 31733
    },
    {
      "epoch": 1.7280431275652939,
      "grad_norm": 0.5796608200635165,
      "learning_rate": 9.54118844019194e-06,
      "loss": 11.8446,
      "step": 31734
    },
    {
      "epoch": 1.7280975815618769,
      "grad_norm": 0.5473601123137064,
      "learning_rate": 9.5374296858337e-06,
      "loss": 11.9058,
      "step": 31735
    },
    {
      "epoch": 1.7281520355584599,
      "grad_norm": 0.6127335719945055,
      "learning_rate": 9.533671634927599e-06,
      "loss": 11.8116,
      "step": 31736
    },
    {
      "epoch": 1.7282064895550429,
      "grad_norm": 0.532967238794352,
      "learning_rate": 9.529914287502816e-06,
      "loss": 11.7317,
      "step": 31737
    },
    {
      "epoch": 1.7282609435516259,
      "grad_norm": 0.6523433605397009,
      "learning_rate": 9.526157643588618e-06,
      "loss": 11.8911,
      "step": 31738
    },
    {
      "epoch": 1.7283153975482088,
      "grad_norm": 0.5082996924963534,
      "learning_rate": 9.52240170321418e-06,
      "loss": 11.7689,
      "step": 31739
    },
    {
      "epoch": 1.7283698515447918,
      "grad_norm": 0.5489672555722254,
      "learning_rate": 9.518646466408709e-06,
      "loss": 11.8655,
      "step": 31740
    },
    {
      "epoch": 1.7284243055413748,
      "grad_norm": 0.47865030614903176,
      "learning_rate": 9.51489193320143e-06,
      "loss": 11.7239,
      "step": 31741
    },
    {
      "epoch": 1.7284787595379578,
      "grad_norm": 0.5008943163819279,
      "learning_rate": 9.511138103621508e-06,
      "loss": 11.6818,
      "step": 31742
    },
    {
      "epoch": 1.7285332135345408,
      "grad_norm": 0.5616135003176408,
      "learning_rate": 9.507384977698175e-06,
      "loss": 11.8098,
      "step": 31743
    },
    {
      "epoch": 1.7285876675311238,
      "grad_norm": 0.6008478960011499,
      "learning_rate": 9.503632555460574e-06,
      "loss": 11.8165,
      "step": 31744
    },
    {
      "epoch": 1.7286421215277068,
      "grad_norm": 0.5864006875525177,
      "learning_rate": 9.499880836937913e-06,
      "loss": 11.9344,
      "step": 31745
    },
    {
      "epoch": 1.7286965755242898,
      "grad_norm": 0.6029267443455991,
      "learning_rate": 9.496129822159338e-06,
      "loss": 11.7735,
      "step": 31746
    },
    {
      "epoch": 1.7287510295208728,
      "grad_norm": 0.5100562454781388,
      "learning_rate": 9.492379511154036e-06,
      "loss": 11.7855,
      "step": 31747
    },
    {
      "epoch": 1.7288054835174558,
      "grad_norm": 0.5145890233843773,
      "learning_rate": 9.488629903951197e-06,
      "loss": 11.7564,
      "step": 31748
    },
    {
      "epoch": 1.7288599375140388,
      "grad_norm": 0.5558111178106622,
      "learning_rate": 9.484881000579937e-06,
      "loss": 11.8101,
      "step": 31749
    },
    {
      "epoch": 1.728914391510622,
      "grad_norm": 0.5288825719725364,
      "learning_rate": 9.481132801069403e-06,
      "loss": 11.7869,
      "step": 31750
    },
    {
      "epoch": 1.728968845507205,
      "grad_norm": 0.5379104324708626,
      "learning_rate": 9.477385305448794e-06,
      "loss": 11.8337,
      "step": 31751
    },
    {
      "epoch": 1.729023299503788,
      "grad_norm": 0.6405719500892642,
      "learning_rate": 9.473638513747184e-06,
      "loss": 11.9674,
      "step": 31752
    },
    {
      "epoch": 1.729077753500371,
      "grad_norm": 0.614330814369286,
      "learning_rate": 9.469892425993764e-06,
      "loss": 11.7954,
      "step": 31753
    },
    {
      "epoch": 1.729132207496954,
      "grad_norm": 0.5079721712198427,
      "learning_rate": 9.466147042217632e-06,
      "loss": 11.652,
      "step": 31754
    },
    {
      "epoch": 1.729186661493537,
      "grad_norm": 0.5782492986472985,
      "learning_rate": 9.46240236244793e-06,
      "loss": 11.77,
      "step": 31755
    },
    {
      "epoch": 1.7292411154901202,
      "grad_norm": 0.5521552723003574,
      "learning_rate": 9.45865838671376e-06,
      "loss": 11.839,
      "step": 31756
    },
    {
      "epoch": 1.7292955694867032,
      "grad_norm": 0.5549373008724034,
      "learning_rate": 9.45491511504425e-06,
      "loss": 11.8042,
      "step": 31757
    },
    {
      "epoch": 1.7293500234832861,
      "grad_norm": 0.5215075361656176,
      "learning_rate": 9.451172547468512e-06,
      "loss": 11.6191,
      "step": 31758
    },
    {
      "epoch": 1.7294044774798691,
      "grad_norm": 0.5692800819500581,
      "learning_rate": 9.447430684015645e-06,
      "loss": 11.7545,
      "step": 31759
    },
    {
      "epoch": 1.7294589314764521,
      "grad_norm": 0.5437205367979392,
      "learning_rate": 9.44368952471475e-06,
      "loss": 11.8703,
      "step": 31760
    },
    {
      "epoch": 1.7295133854730351,
      "grad_norm": 0.5414735539849145,
      "learning_rate": 9.439949069594888e-06,
      "loss": 11.8,
      "step": 31761
    },
    {
      "epoch": 1.7295678394696181,
      "grad_norm": 0.5091608588791786,
      "learning_rate": 9.436209318685163e-06,
      "loss": 11.7333,
      "step": 31762
    },
    {
      "epoch": 1.729622293466201,
      "grad_norm": 0.5716410024395188,
      "learning_rate": 9.432470272014681e-06,
      "loss": 11.8264,
      "step": 31763
    },
    {
      "epoch": 1.729676747462784,
      "grad_norm": 0.5155399898054873,
      "learning_rate": 9.428731929612488e-06,
      "loss": 11.8019,
      "step": 31764
    },
    {
      "epoch": 1.729731201459367,
      "grad_norm": 0.5338373149824597,
      "learning_rate": 9.424994291507682e-06,
      "loss": 11.8044,
      "step": 31765
    },
    {
      "epoch": 1.72978565545595,
      "grad_norm": 0.5280839117958512,
      "learning_rate": 9.421257357729284e-06,
      "loss": 11.7311,
      "step": 31766
    },
    {
      "epoch": 1.729840109452533,
      "grad_norm": 0.5590267339098637,
      "learning_rate": 9.417521128306406e-06,
      "loss": 11.9218,
      "step": 31767
    },
    {
      "epoch": 1.729894563449116,
      "grad_norm": 0.5411376474594606,
      "learning_rate": 9.413785603268055e-06,
      "loss": 11.7603,
      "step": 31768
    },
    {
      "epoch": 1.729949017445699,
      "grad_norm": 0.5774859340004413,
      "learning_rate": 9.41005078264331e-06,
      "loss": 11.7246,
      "step": 31769
    },
    {
      "epoch": 1.730003471442282,
      "grad_norm": 0.5115390727274177,
      "learning_rate": 9.406316666461202e-06,
      "loss": 11.7678,
      "step": 31770
    },
    {
      "epoch": 1.730057925438865,
      "grad_norm": 0.5870685307456822,
      "learning_rate": 9.402583254750752e-06,
      "loss": 11.8025,
      "step": 31771
    },
    {
      "epoch": 1.730112379435448,
      "grad_norm": 0.5992386089876499,
      "learning_rate": 9.398850547541015e-06,
      "loss": 11.8757,
      "step": 31772
    },
    {
      "epoch": 1.730166833432031,
      "grad_norm": 0.5789352698885607,
      "learning_rate": 9.395118544861026e-06,
      "loss": 11.7844,
      "step": 31773
    },
    {
      "epoch": 1.7302212874286143,
      "grad_norm": 0.5334465536363548,
      "learning_rate": 9.391387246739758e-06,
      "loss": 11.6402,
      "step": 31774
    },
    {
      "epoch": 1.7302757414251972,
      "grad_norm": 0.5767553661427227,
      "learning_rate": 9.387656653206289e-06,
      "loss": 11.8415,
      "step": 31775
    },
    {
      "epoch": 1.7303301954217802,
      "grad_norm": 0.5284614688540364,
      "learning_rate": 9.383926764289574e-06,
      "loss": 11.7814,
      "step": 31776
    },
    {
      "epoch": 1.7303846494183632,
      "grad_norm": 0.5561196040121275,
      "learning_rate": 9.380197580018667e-06,
      "loss": 11.7798,
      "step": 31777
    },
    {
      "epoch": 1.7304391034149462,
      "grad_norm": 0.5335937976136347,
      "learning_rate": 9.376469100422513e-06,
      "loss": 11.4178,
      "step": 31778
    },
    {
      "epoch": 1.7304935574115294,
      "grad_norm": 0.5712066405402608,
      "learning_rate": 9.372741325530154e-06,
      "loss": 11.875,
      "step": 31779
    },
    {
      "epoch": 1.7305480114081124,
      "grad_norm": 0.5862121634867449,
      "learning_rate": 9.369014255370557e-06,
      "loss": 11.9227,
      "step": 31780
    },
    {
      "epoch": 1.7306024654046954,
      "grad_norm": 0.5074207143718371,
      "learning_rate": 9.365287889972686e-06,
      "loss": 11.6428,
      "step": 31781
    },
    {
      "epoch": 1.7306569194012784,
      "grad_norm": 0.5267269417926154,
      "learning_rate": 9.361562229365561e-06,
      "loss": 11.9004,
      "step": 31782
    },
    {
      "epoch": 1.7307113733978614,
      "grad_norm": 0.5822986060977571,
      "learning_rate": 9.357837273578096e-06,
      "loss": 11.7974,
      "step": 31783
    },
    {
      "epoch": 1.7307658273944444,
      "grad_norm": 0.5779219084655686,
      "learning_rate": 9.3541130226393e-06,
      "loss": 11.8296,
      "step": 31784
    },
    {
      "epoch": 1.7308202813910274,
      "grad_norm": 0.5272081635596604,
      "learning_rate": 9.350389476578137e-06,
      "loss": 11.6917,
      "step": 31785
    },
    {
      "epoch": 1.7308747353876104,
      "grad_norm": 0.5631118552644984,
      "learning_rate": 9.34666663542353e-06,
      "loss": 11.7668,
      "step": 31786
    },
    {
      "epoch": 1.7309291893841934,
      "grad_norm": 0.5549072649547676,
      "learning_rate": 9.342944499204465e-06,
      "loss": 11.6803,
      "step": 31787
    },
    {
      "epoch": 1.7309836433807764,
      "grad_norm": 0.7167830541880525,
      "learning_rate": 9.339223067949843e-06,
      "loss": 11.804,
      "step": 31788
    },
    {
      "epoch": 1.7310380973773594,
      "grad_norm": 0.5556018355130229,
      "learning_rate": 9.335502341688652e-06,
      "loss": 11.6957,
      "step": 31789
    },
    {
      "epoch": 1.7310925513739424,
      "grad_norm": 0.5446655057382435,
      "learning_rate": 9.33178232044979e-06,
      "loss": 11.8085,
      "step": 31790
    },
    {
      "epoch": 1.7311470053705253,
      "grad_norm": 0.5432132415105937,
      "learning_rate": 9.328063004262177e-06,
      "loss": 11.8292,
      "step": 31791
    },
    {
      "epoch": 1.7312014593671083,
      "grad_norm": 0.5688969977007066,
      "learning_rate": 9.324344393154783e-06,
      "loss": 11.8716,
      "step": 31792
    },
    {
      "epoch": 1.7312559133636913,
      "grad_norm": 0.5212004715882294,
      "learning_rate": 9.320626487156459e-06,
      "loss": 11.7897,
      "step": 31793
    },
    {
      "epoch": 1.7313103673602743,
      "grad_norm": 0.516186582647028,
      "learning_rate": 9.316909286296183e-06,
      "loss": 11.873,
      "step": 31794
    },
    {
      "epoch": 1.7313648213568573,
      "grad_norm": 0.5027105438080042,
      "learning_rate": 9.313192790602798e-06,
      "loss": 11.7418,
      "step": 31795
    },
    {
      "epoch": 1.7314192753534403,
      "grad_norm": 0.5349807289495488,
      "learning_rate": 9.309477000105237e-06,
      "loss": 11.6966,
      "step": 31796
    },
    {
      "epoch": 1.7314737293500235,
      "grad_norm": 0.5307829903497532,
      "learning_rate": 9.305761914832412e-06,
      "loss": 11.8554,
      "step": 31797
    },
    {
      "epoch": 1.7315281833466065,
      "grad_norm": 0.4943187880420814,
      "learning_rate": 9.302047534813174e-06,
      "loss": 11.7604,
      "step": 31798
    },
    {
      "epoch": 1.7315826373431895,
      "grad_norm": 0.537687020744414,
      "learning_rate": 9.298333860076435e-06,
      "loss": 11.8353,
      "step": 31799
    },
    {
      "epoch": 1.7316370913397725,
      "grad_norm": 0.5634131204889544,
      "learning_rate": 9.294620890651074e-06,
      "loss": 11.7512,
      "step": 31800
    },
    {
      "epoch": 1.7316915453363555,
      "grad_norm": 0.5880410043981958,
      "learning_rate": 9.29090862656593e-06,
      "loss": 11.9291,
      "step": 31801
    },
    {
      "epoch": 1.7317459993329385,
      "grad_norm": 0.5863121013597389,
      "learning_rate": 9.287197067849907e-06,
      "loss": 11.7841,
      "step": 31802
    },
    {
      "epoch": 1.7318004533295217,
      "grad_norm": 0.5473313432392998,
      "learning_rate": 9.283486214531833e-06,
      "loss": 11.8744,
      "step": 31803
    },
    {
      "epoch": 1.7318549073261047,
      "grad_norm": 0.535849865589238,
      "learning_rate": 9.27977606664061e-06,
      "loss": 11.8094,
      "step": 31804
    },
    {
      "epoch": 1.7319093613226877,
      "grad_norm": 0.5661076997331189,
      "learning_rate": 9.276066624205038e-06,
      "loss": 11.8197,
      "step": 31805
    },
    {
      "epoch": 1.7319638153192707,
      "grad_norm": 0.5627804408755753,
      "learning_rate": 9.27235788725399e-06,
      "loss": 11.7295,
      "step": 31806
    },
    {
      "epoch": 1.7320182693158537,
      "grad_norm": 0.5157923365260186,
      "learning_rate": 9.268649855816313e-06,
      "loss": 11.8166,
      "step": 31807
    },
    {
      "epoch": 1.7320727233124367,
      "grad_norm": 0.5736871896204039,
      "learning_rate": 9.264942529920817e-06,
      "loss": 11.8427,
      "step": 31808
    },
    {
      "epoch": 1.7321271773090197,
      "grad_norm": 0.5814916522615083,
      "learning_rate": 9.261235909596367e-06,
      "loss": 11.8641,
      "step": 31809
    },
    {
      "epoch": 1.7321816313056027,
      "grad_norm": 0.5461754129138099,
      "learning_rate": 9.25752999487176e-06,
      "loss": 11.8515,
      "step": 31810
    },
    {
      "epoch": 1.7322360853021856,
      "grad_norm": 0.5126673278482509,
      "learning_rate": 9.253824785775799e-06,
      "loss": 11.7771,
      "step": 31811
    },
    {
      "epoch": 1.7322905392987686,
      "grad_norm": 0.5560593970273288,
      "learning_rate": 9.250120282337326e-06,
      "loss": 11.7364,
      "step": 31812
    },
    {
      "epoch": 1.7323449932953516,
      "grad_norm": 0.5395423087362559,
      "learning_rate": 9.24641648458513e-06,
      "loss": 11.8862,
      "step": 31813
    },
    {
      "epoch": 1.7323994472919346,
      "grad_norm": 0.49090752907695995,
      "learning_rate": 9.24271339254803e-06,
      "loss": 11.7974,
      "step": 31814
    },
    {
      "epoch": 1.7324539012885176,
      "grad_norm": 0.4910345344878293,
      "learning_rate": 9.239011006254794e-06,
      "loss": 11.7537,
      "step": 31815
    },
    {
      "epoch": 1.7325083552851006,
      "grad_norm": 0.5468577576444289,
      "learning_rate": 9.235309325734242e-06,
      "loss": 11.8825,
      "step": 31816
    },
    {
      "epoch": 1.7325628092816836,
      "grad_norm": 0.541162242652814,
      "learning_rate": 9.23160835101513e-06,
      "loss": 11.8388,
      "step": 31817
    },
    {
      "epoch": 1.7326172632782666,
      "grad_norm": 0.6230660625624617,
      "learning_rate": 9.227908082126258e-06,
      "loss": 11.9004,
      "step": 31818
    },
    {
      "epoch": 1.7326717172748496,
      "grad_norm": 0.5382764596129828,
      "learning_rate": 9.22420851909641e-06,
      "loss": 11.9146,
      "step": 31819
    },
    {
      "epoch": 1.7327261712714328,
      "grad_norm": 0.5263115047747339,
      "learning_rate": 9.220509661954346e-06,
      "loss": 11.6744,
      "step": 31820
    },
    {
      "epoch": 1.7327806252680158,
      "grad_norm": 0.49533064461987614,
      "learning_rate": 9.216811510728795e-06,
      "loss": 11.765,
      "step": 31821
    },
    {
      "epoch": 1.7328350792645988,
      "grad_norm": 0.5375916987433108,
      "learning_rate": 9.213114065448559e-06,
      "loss": 11.7109,
      "step": 31822
    },
    {
      "epoch": 1.7328895332611818,
      "grad_norm": 0.5771549502666888,
      "learning_rate": 9.209417326142367e-06,
      "loss": 11.9013,
      "step": 31823
    },
    {
      "epoch": 1.7329439872577648,
      "grad_norm": 0.5788564287638096,
      "learning_rate": 9.205721292838976e-06,
      "loss": 11.7073,
      "step": 31824
    },
    {
      "epoch": 1.7329984412543478,
      "grad_norm": 0.5620934116215915,
      "learning_rate": 9.202025965567118e-06,
      "loss": 11.9465,
      "step": 31825
    },
    {
      "epoch": 1.733052895250931,
      "grad_norm": 0.5288264080461414,
      "learning_rate": 9.198331344355537e-06,
      "loss": 11.7418,
      "step": 31826
    },
    {
      "epoch": 1.733107349247514,
      "grad_norm": 0.5991099826495264,
      "learning_rate": 9.194637429232955e-06,
      "loss": 11.8112,
      "step": 31827
    },
    {
      "epoch": 1.733161803244097,
      "grad_norm": 0.49794012698116397,
      "learning_rate": 9.190944220228093e-06,
      "loss": 11.7288,
      "step": 31828
    },
    {
      "epoch": 1.73321625724068,
      "grad_norm": 0.5295308583277801,
      "learning_rate": 9.187251717369695e-06,
      "loss": 11.8134,
      "step": 31829
    },
    {
      "epoch": 1.733270711237263,
      "grad_norm": 0.5052311824340352,
      "learning_rate": 9.183559920686457e-06,
      "loss": 11.8452,
      "step": 31830
    },
    {
      "epoch": 1.733325165233846,
      "grad_norm": 0.524407725961568,
      "learning_rate": 9.179868830207084e-06,
      "loss": 11.6978,
      "step": 31831
    },
    {
      "epoch": 1.733379619230429,
      "grad_norm": 0.5772553659785641,
      "learning_rate": 9.17617844596027e-06,
      "loss": 11.7124,
      "step": 31832
    },
    {
      "epoch": 1.733434073227012,
      "grad_norm": 0.5215331440394445,
      "learning_rate": 9.172488767974718e-06,
      "loss": 11.688,
      "step": 31833
    },
    {
      "epoch": 1.733488527223595,
      "grad_norm": 0.5672522888550658,
      "learning_rate": 9.168799796279148e-06,
      "loss": 11.7624,
      "step": 31834
    },
    {
      "epoch": 1.733542981220178,
      "grad_norm": 0.5529480895985476,
      "learning_rate": 9.165111530902204e-06,
      "loss": 11.7447,
      "step": 31835
    },
    {
      "epoch": 1.733597435216761,
      "grad_norm": 0.5738203143307091,
      "learning_rate": 9.161423971872606e-06,
      "loss": 11.8681,
      "step": 31836
    },
    {
      "epoch": 1.733651889213344,
      "grad_norm": 0.48516725773577646,
      "learning_rate": 9.15773711921898e-06,
      "loss": 11.8038,
      "step": 31837
    },
    {
      "epoch": 1.733706343209927,
      "grad_norm": 0.559581261836742,
      "learning_rate": 9.154050972970052e-06,
      "loss": 11.8667,
      "step": 31838
    },
    {
      "epoch": 1.7337607972065099,
      "grad_norm": 0.5619177232601096,
      "learning_rate": 9.150365533154437e-06,
      "loss": 11.7478,
      "step": 31839
    },
    {
      "epoch": 1.7338152512030929,
      "grad_norm": 0.5469247544641608,
      "learning_rate": 9.146680799800834e-06,
      "loss": 11.8148,
      "step": 31840
    },
    {
      "epoch": 1.7338697051996759,
      "grad_norm": 0.5634893522774611,
      "learning_rate": 9.142996772937884e-06,
      "loss": 11.8232,
      "step": 31841
    },
    {
      "epoch": 1.7339241591962589,
      "grad_norm": 0.60737512667563,
      "learning_rate": 9.1393134525942e-06,
      "loss": 11.7279,
      "step": 31842
    },
    {
      "epoch": 1.733978613192842,
      "grad_norm": 0.5891923500901595,
      "learning_rate": 9.13563083879847e-06,
      "loss": 11.8399,
      "step": 31843
    },
    {
      "epoch": 1.734033067189425,
      "grad_norm": 0.549736623254649,
      "learning_rate": 9.131948931579303e-06,
      "loss": 11.6969,
      "step": 31844
    },
    {
      "epoch": 1.734087521186008,
      "grad_norm": 0.5923333227224441,
      "learning_rate": 9.128267730965334e-06,
      "loss": 11.9202,
      "step": 31845
    },
    {
      "epoch": 1.734141975182591,
      "grad_norm": 0.5112563342606398,
      "learning_rate": 9.124587236985216e-06,
      "loss": 11.8206,
      "step": 31846
    },
    {
      "epoch": 1.734196429179174,
      "grad_norm": 0.5577007296796052,
      "learning_rate": 9.120907449667527e-06,
      "loss": 11.7966,
      "step": 31847
    },
    {
      "epoch": 1.734250883175757,
      "grad_norm": 0.5229001121047716,
      "learning_rate": 9.11722836904092e-06,
      "loss": 11.7849,
      "step": 31848
    },
    {
      "epoch": 1.7343053371723403,
      "grad_norm": 0.5792622385692741,
      "learning_rate": 9.113549995133964e-06,
      "loss": 11.7612,
      "step": 31849
    },
    {
      "epoch": 1.7343597911689232,
      "grad_norm": 0.5338103589125868,
      "learning_rate": 9.10987232797531e-06,
      "loss": 11.7377,
      "step": 31850
    },
    {
      "epoch": 1.7344142451655062,
      "grad_norm": 0.5622271457193666,
      "learning_rate": 9.106195367593528e-06,
      "loss": 11.7544,
      "step": 31851
    },
    {
      "epoch": 1.7344686991620892,
      "grad_norm": 0.5177335513682215,
      "learning_rate": 9.102519114017194e-06,
      "loss": 11.7878,
      "step": 31852
    },
    {
      "epoch": 1.7345231531586722,
      "grad_norm": 0.5389755869777442,
      "learning_rate": 9.09884356727494e-06,
      "loss": 11.6675,
      "step": 31853
    },
    {
      "epoch": 1.7345776071552552,
      "grad_norm": 0.5404158838421443,
      "learning_rate": 9.095168727395298e-06,
      "loss": 11.8193,
      "step": 31854
    },
    {
      "epoch": 1.7346320611518382,
      "grad_norm": 0.5943710794913865,
      "learning_rate": 9.091494594406868e-06,
      "loss": 11.7008,
      "step": 31855
    },
    {
      "epoch": 1.7346865151484212,
      "grad_norm": 0.5339501381814628,
      "learning_rate": 9.087821168338239e-06,
      "loss": 11.6924,
      "step": 31856
    },
    {
      "epoch": 1.7347409691450042,
      "grad_norm": 0.5678115108341879,
      "learning_rate": 9.084148449217945e-06,
      "loss": 11.7249,
      "step": 31857
    },
    {
      "epoch": 1.7347954231415872,
      "grad_norm": 0.5668293287245962,
      "learning_rate": 9.080476437074569e-06,
      "loss": 11.8283,
      "step": 31858
    },
    {
      "epoch": 1.7348498771381702,
      "grad_norm": 0.5547633744492095,
      "learning_rate": 9.07680513193665e-06,
      "loss": 11.7417,
      "step": 31859
    },
    {
      "epoch": 1.7349043311347532,
      "grad_norm": 0.5436193098513749,
      "learning_rate": 9.07313453383275e-06,
      "loss": 11.6788,
      "step": 31860
    },
    {
      "epoch": 1.7349587851313362,
      "grad_norm": 0.5690684965834453,
      "learning_rate": 9.069464642791403e-06,
      "loss": 11.8594,
      "step": 31861
    },
    {
      "epoch": 1.7350132391279192,
      "grad_norm": 0.5921293204556438,
      "learning_rate": 9.065795458841143e-06,
      "loss": 11.8938,
      "step": 31862
    },
    {
      "epoch": 1.7350676931245022,
      "grad_norm": 0.5523228562514196,
      "learning_rate": 9.06212698201051e-06,
      "loss": 11.8822,
      "step": 31863
    },
    {
      "epoch": 1.7351221471210851,
      "grad_norm": 0.5686848873771777,
      "learning_rate": 9.058459212328018e-06,
      "loss": 11.8903,
      "step": 31864
    },
    {
      "epoch": 1.7351766011176681,
      "grad_norm": 0.5264078035947367,
      "learning_rate": 9.054792149822222e-06,
      "loss": 11.8165,
      "step": 31865
    },
    {
      "epoch": 1.7352310551142511,
      "grad_norm": 0.5488165106359091,
      "learning_rate": 9.051125794521587e-06,
      "loss": 11.7877,
      "step": 31866
    },
    {
      "epoch": 1.7352855091108343,
      "grad_norm": 0.5730019632075116,
      "learning_rate": 9.047460146454644e-06,
      "loss": 11.8723,
      "step": 31867
    },
    {
      "epoch": 1.7353399631074173,
      "grad_norm": 0.6138328700026215,
      "learning_rate": 9.04379520564993e-06,
      "loss": 11.7798,
      "step": 31868
    },
    {
      "epoch": 1.7353944171040003,
      "grad_norm": 0.6531888711113734,
      "learning_rate": 9.040130972135907e-06,
      "loss": 11.7869,
      "step": 31869
    },
    {
      "epoch": 1.7354488711005833,
      "grad_norm": 0.575646539004118,
      "learning_rate": 9.036467445941089e-06,
      "loss": 11.7694,
      "step": 31870
    },
    {
      "epoch": 1.7355033250971663,
      "grad_norm": 0.5919880430843911,
      "learning_rate": 9.03280462709395e-06,
      "loss": 11.8656,
      "step": 31871
    },
    {
      "epoch": 1.7355577790937493,
      "grad_norm": 0.524589233508476,
      "learning_rate": 9.029142515622968e-06,
      "loss": 11.7044,
      "step": 31872
    },
    {
      "epoch": 1.7356122330903325,
      "grad_norm": 0.5689243356154033,
      "learning_rate": 9.025481111556645e-06,
      "loss": 11.8084,
      "step": 31873
    },
    {
      "epoch": 1.7356666870869155,
      "grad_norm": 0.5010574702545381,
      "learning_rate": 9.021820414923421e-06,
      "loss": 11.6926,
      "step": 31874
    },
    {
      "epoch": 1.7357211410834985,
      "grad_norm": 0.5140298930794341,
      "learning_rate": 9.018160425751787e-06,
      "loss": 11.7479,
      "step": 31875
    },
    {
      "epoch": 1.7357755950800815,
      "grad_norm": 0.5598431460374912,
      "learning_rate": 9.014501144070187e-06,
      "loss": 11.8489,
      "step": 31876
    },
    {
      "epoch": 1.7358300490766645,
      "grad_norm": 0.5498528068954908,
      "learning_rate": 9.010842569907086e-06,
      "loss": 11.7145,
      "step": 31877
    },
    {
      "epoch": 1.7358845030732475,
      "grad_norm": 0.5128889158274546,
      "learning_rate": 9.00718470329095e-06,
      "loss": 11.8008,
      "step": 31878
    },
    {
      "epoch": 1.7359389570698305,
      "grad_norm": 0.5558278832011802,
      "learning_rate": 9.003527544250178e-06,
      "loss": 11.8823,
      "step": 31879
    },
    {
      "epoch": 1.7359934110664135,
      "grad_norm": 0.5685858614518955,
      "learning_rate": 8.999871092813272e-06,
      "loss": 11.8429,
      "step": 31880
    },
    {
      "epoch": 1.7360478650629965,
      "grad_norm": 0.5850625674664518,
      "learning_rate": 8.996215349008608e-06,
      "loss": 11.8085,
      "step": 31881
    },
    {
      "epoch": 1.7361023190595795,
      "grad_norm": 0.5071773126399107,
      "learning_rate": 8.992560312864617e-06,
      "loss": 11.804,
      "step": 31882
    },
    {
      "epoch": 1.7361567730561625,
      "grad_norm": 0.5433957365845743,
      "learning_rate": 8.988905984409768e-06,
      "loss": 11.7221,
      "step": 31883
    },
    {
      "epoch": 1.7362112270527454,
      "grad_norm": 0.501478910362739,
      "learning_rate": 8.985252363672426e-06,
      "loss": 11.7778,
      "step": 31884
    },
    {
      "epoch": 1.7362656810493284,
      "grad_norm": 0.5460184950510293,
      "learning_rate": 8.981599450681043e-06,
      "loss": 11.6809,
      "step": 31885
    },
    {
      "epoch": 1.7363201350459114,
      "grad_norm": 0.5794270108992531,
      "learning_rate": 8.977947245463991e-06,
      "loss": 11.7472,
      "step": 31886
    },
    {
      "epoch": 1.7363745890424944,
      "grad_norm": 0.513443854355982,
      "learning_rate": 8.97429574804971e-06,
      "loss": 11.6999,
      "step": 31887
    },
    {
      "epoch": 1.7364290430390774,
      "grad_norm": 0.5719591196016566,
      "learning_rate": 8.970644958466534e-06,
      "loss": 11.948,
      "step": 31888
    },
    {
      "epoch": 1.7364834970356604,
      "grad_norm": 0.5466542800558982,
      "learning_rate": 8.966994876742907e-06,
      "loss": 11.6543,
      "step": 31889
    },
    {
      "epoch": 1.7365379510322436,
      "grad_norm": 0.5855541183710491,
      "learning_rate": 8.963345502907216e-06,
      "loss": 11.8503,
      "step": 31890
    },
    {
      "epoch": 1.7365924050288266,
      "grad_norm": 0.4972010693033175,
      "learning_rate": 8.959696836987796e-06,
      "loss": 11.739,
      "step": 31891
    },
    {
      "epoch": 1.7366468590254096,
      "grad_norm": 0.6368720737829793,
      "learning_rate": 8.956048879013045e-06,
      "loss": 11.7997,
      "step": 31892
    },
    {
      "epoch": 1.7367013130219926,
      "grad_norm": 0.6061320673296801,
      "learning_rate": 8.95240162901132e-06,
      "loss": 11.8223,
      "step": 31893
    },
    {
      "epoch": 1.7367557670185756,
      "grad_norm": 0.5468653568716522,
      "learning_rate": 8.948755087010973e-06,
      "loss": 11.6784,
      "step": 31894
    },
    {
      "epoch": 1.7368102210151586,
      "grad_norm": 0.6787007297548442,
      "learning_rate": 8.945109253040407e-06,
      "loss": 11.9298,
      "step": 31895
    },
    {
      "epoch": 1.7368646750117418,
      "grad_norm": 0.4826222858430997,
      "learning_rate": 8.941464127127918e-06,
      "loss": 11.7985,
      "step": 31896
    },
    {
      "epoch": 1.7369191290083248,
      "grad_norm": 0.5567254485433407,
      "learning_rate": 8.937819709301898e-06,
      "loss": 11.7621,
      "step": 31897
    },
    {
      "epoch": 1.7369735830049078,
      "grad_norm": 0.5272206331795003,
      "learning_rate": 8.934175999590633e-06,
      "loss": 11.7968,
      "step": 31898
    },
    {
      "epoch": 1.7370280370014908,
      "grad_norm": 0.5289801539308546,
      "learning_rate": 8.930532998022512e-06,
      "loss": 11.7091,
      "step": 31899
    },
    {
      "epoch": 1.7370824909980738,
      "grad_norm": 0.5294484475452663,
      "learning_rate": 8.926890704625845e-06,
      "loss": 11.7452,
      "step": 31900
    },
    {
      "epoch": 1.7371369449946568,
      "grad_norm": 0.5283246401411331,
      "learning_rate": 8.923249119428922e-06,
      "loss": 11.758,
      "step": 31901
    },
    {
      "epoch": 1.7371913989912398,
      "grad_norm": 0.6418452457116163,
      "learning_rate": 8.919608242460108e-06,
      "loss": 11.8581,
      "step": 31902
    },
    {
      "epoch": 1.7372458529878227,
      "grad_norm": 0.5248087601600387,
      "learning_rate": 8.915968073747682e-06,
      "loss": 11.8406,
      "step": 31903
    },
    {
      "epoch": 1.7373003069844057,
      "grad_norm": 0.5388132701630212,
      "learning_rate": 8.912328613319953e-06,
      "loss": 11.8169,
      "step": 31904
    },
    {
      "epoch": 1.7373547609809887,
      "grad_norm": 0.5051992040933106,
      "learning_rate": 8.908689861205255e-06,
      "loss": 11.7077,
      "step": 31905
    },
    {
      "epoch": 1.7374092149775717,
      "grad_norm": 0.5759347931673632,
      "learning_rate": 8.905051817431853e-06,
      "loss": 11.8287,
      "step": 31906
    },
    {
      "epoch": 1.7374636689741547,
      "grad_norm": 0.4963267385136898,
      "learning_rate": 8.901414482028047e-06,
      "loss": 11.8085,
      "step": 31907
    },
    {
      "epoch": 1.7375181229707377,
      "grad_norm": 0.5020050777137453,
      "learning_rate": 8.897777855022105e-06,
      "loss": 11.7405,
      "step": 31908
    },
    {
      "epoch": 1.7375725769673207,
      "grad_norm": 0.5810868966399029,
      "learning_rate": 8.894141936442346e-06,
      "loss": 11.7657,
      "step": 31909
    },
    {
      "epoch": 1.7376270309639037,
      "grad_norm": 0.5735157174973503,
      "learning_rate": 8.890506726317005e-06,
      "loss": 11.7871,
      "step": 31910
    },
    {
      "epoch": 1.7376814849604867,
      "grad_norm": 0.5594504910573439,
      "learning_rate": 8.886872224674359e-06,
      "loss": 11.6382,
      "step": 31911
    },
    {
      "epoch": 1.7377359389570697,
      "grad_norm": 0.5037711618953534,
      "learning_rate": 8.883238431542684e-06,
      "loss": 11.791,
      "step": 31912
    },
    {
      "epoch": 1.737790392953653,
      "grad_norm": 0.5332720651748264,
      "learning_rate": 8.879605346950203e-06,
      "loss": 11.7786,
      "step": 31913
    },
    {
      "epoch": 1.7378448469502359,
      "grad_norm": 0.5461210602637525,
      "learning_rate": 8.875972970925229e-06,
      "loss": 11.8422,
      "step": 31914
    },
    {
      "epoch": 1.7378993009468189,
      "grad_norm": 0.5556292310379968,
      "learning_rate": 8.872341303495935e-06,
      "loss": 11.7714,
      "step": 31915
    },
    {
      "epoch": 1.7379537549434019,
      "grad_norm": 0.554320969484622,
      "learning_rate": 8.868710344690601e-06,
      "loss": 11.7083,
      "step": 31916
    },
    {
      "epoch": 1.7380082089399849,
      "grad_norm": 0.6041289850317021,
      "learning_rate": 8.86508009453748e-06,
      "loss": 11.7377,
      "step": 31917
    },
    {
      "epoch": 1.7380626629365679,
      "grad_norm": 0.5841586540888242,
      "learning_rate": 8.861450553064765e-06,
      "loss": 11.7625,
      "step": 31918
    },
    {
      "epoch": 1.738117116933151,
      "grad_norm": 0.5567303583460421,
      "learning_rate": 8.857821720300697e-06,
      "loss": 11.6293,
      "step": 31919
    },
    {
      "epoch": 1.738171570929734,
      "grad_norm": 0.5983747172622275,
      "learning_rate": 8.854193596273509e-06,
      "loss": 11.8934,
      "step": 31920
    },
    {
      "epoch": 1.738226024926317,
      "grad_norm": 0.7066976709933921,
      "learning_rate": 8.850566181011366e-06,
      "loss": 11.9972,
      "step": 31921
    },
    {
      "epoch": 1.7382804789229,
      "grad_norm": 0.5893838894800378,
      "learning_rate": 8.846939474542538e-06,
      "loss": 11.7745,
      "step": 31922
    },
    {
      "epoch": 1.738334932919483,
      "grad_norm": 0.5110780693151147,
      "learning_rate": 8.843313476895165e-06,
      "loss": 11.7757,
      "step": 31923
    },
    {
      "epoch": 1.738389386916066,
      "grad_norm": 0.6209174833375607,
      "learning_rate": 8.839688188097495e-06,
      "loss": 11.822,
      "step": 31924
    },
    {
      "epoch": 1.738443840912649,
      "grad_norm": 0.5819825506408214,
      "learning_rate": 8.83606360817768e-06,
      "loss": 11.7911,
      "step": 31925
    },
    {
      "epoch": 1.738498294909232,
      "grad_norm": 0.6126769758037125,
      "learning_rate": 8.832439737163923e-06,
      "loss": 11.7689,
      "step": 31926
    },
    {
      "epoch": 1.738552748905815,
      "grad_norm": 0.5648979642562577,
      "learning_rate": 8.82881657508442e-06,
      "loss": 11.8178,
      "step": 31927
    },
    {
      "epoch": 1.738607202902398,
      "grad_norm": 0.5358949345609733,
      "learning_rate": 8.825194121967307e-06,
      "loss": 11.7572,
      "step": 31928
    },
    {
      "epoch": 1.738661656898981,
      "grad_norm": 0.5717128670773614,
      "learning_rate": 8.821572377840803e-06,
      "loss": 11.8206,
      "step": 31929
    },
    {
      "epoch": 1.738716110895564,
      "grad_norm": 0.5853436900609554,
      "learning_rate": 8.817951342733032e-06,
      "loss": 11.8308,
      "step": 31930
    },
    {
      "epoch": 1.738770564892147,
      "grad_norm": 0.5471095256239284,
      "learning_rate": 8.81433101667215e-06,
      "loss": 11.7714,
      "step": 31931
    },
    {
      "epoch": 1.73882501888873,
      "grad_norm": 0.546559421596322,
      "learning_rate": 8.810711399686334e-06,
      "loss": 11.8059,
      "step": 31932
    },
    {
      "epoch": 1.738879472885313,
      "grad_norm": 0.5367329275204307,
      "learning_rate": 8.807092491803715e-06,
      "loss": 11.7581,
      "step": 31933
    },
    {
      "epoch": 1.738933926881896,
      "grad_norm": 0.5698202345816502,
      "learning_rate": 8.803474293052438e-06,
      "loss": 11.9099,
      "step": 31934
    },
    {
      "epoch": 1.738988380878479,
      "grad_norm": 0.5262262135844425,
      "learning_rate": 8.799856803460627e-06,
      "loss": 11.6944,
      "step": 31935
    },
    {
      "epoch": 1.739042834875062,
      "grad_norm": 0.5930909117868646,
      "learning_rate": 8.796240023056445e-06,
      "loss": 11.8409,
      "step": 31936
    },
    {
      "epoch": 1.7390972888716452,
      "grad_norm": 0.5720385926628261,
      "learning_rate": 8.792623951867985e-06,
      "loss": 11.9001,
      "step": 31937
    },
    {
      "epoch": 1.7391517428682282,
      "grad_norm": 0.587627436208008,
      "learning_rate": 8.789008589923364e-06,
      "loss": 11.8586,
      "step": 31938
    },
    {
      "epoch": 1.7392061968648111,
      "grad_norm": 0.5999198699587912,
      "learning_rate": 8.78539393725073e-06,
      "loss": 11.8785,
      "step": 31939
    },
    {
      "epoch": 1.7392606508613941,
      "grad_norm": 0.5202491532771909,
      "learning_rate": 8.781779993878169e-06,
      "loss": 11.5571,
      "step": 31940
    },
    {
      "epoch": 1.7393151048579771,
      "grad_norm": 0.6831119204493081,
      "learning_rate": 8.778166759833784e-06,
      "loss": 11.8553,
      "step": 31941
    },
    {
      "epoch": 1.7393695588545601,
      "grad_norm": 0.5254518224668623,
      "learning_rate": 8.774554235145648e-06,
      "loss": 11.8292,
      "step": 31942
    },
    {
      "epoch": 1.7394240128511433,
      "grad_norm": 0.5153392436022605,
      "learning_rate": 8.770942419841888e-06,
      "loss": 11.7068,
      "step": 31943
    },
    {
      "epoch": 1.7394784668477263,
      "grad_norm": 0.5577738292986892,
      "learning_rate": 8.767331313950588e-06,
      "loss": 11.6506,
      "step": 31944
    },
    {
      "epoch": 1.7395329208443093,
      "grad_norm": 0.5793486381665118,
      "learning_rate": 8.763720917499807e-06,
      "loss": 11.6731,
      "step": 31945
    },
    {
      "epoch": 1.7395873748408923,
      "grad_norm": 0.5667946436974896,
      "learning_rate": 8.760111230517653e-06,
      "loss": 11.8109,
      "step": 31946
    },
    {
      "epoch": 1.7396418288374753,
      "grad_norm": 0.5475771598830302,
      "learning_rate": 8.75650225303215e-06,
      "loss": 11.7654,
      "step": 31947
    },
    {
      "epoch": 1.7396962828340583,
      "grad_norm": 0.5202563000046072,
      "learning_rate": 8.75289398507141e-06,
      "loss": 11.8181,
      "step": 31948
    },
    {
      "epoch": 1.7397507368306413,
      "grad_norm": 0.5646446723735778,
      "learning_rate": 8.74928642666345e-06,
      "loss": 11.762,
      "step": 31949
    },
    {
      "epoch": 1.7398051908272243,
      "grad_norm": 0.6338854791429734,
      "learning_rate": 8.745679577836342e-06,
      "loss": 11.8404,
      "step": 31950
    },
    {
      "epoch": 1.7398596448238073,
      "grad_norm": 0.627681073375901,
      "learning_rate": 8.74207343861817e-06,
      "loss": 11.7785,
      "step": 31951
    },
    {
      "epoch": 1.7399140988203903,
      "grad_norm": 0.5419673933021651,
      "learning_rate": 8.738468009036893e-06,
      "loss": 11.724,
      "step": 31952
    },
    {
      "epoch": 1.7399685528169733,
      "grad_norm": 0.5212431761118159,
      "learning_rate": 8.73486328912061e-06,
      "loss": 11.6219,
      "step": 31953
    },
    {
      "epoch": 1.7400230068135563,
      "grad_norm": 0.60439168942001,
      "learning_rate": 8.731259278897341e-06,
      "loss": 11.8018,
      "step": 31954
    },
    {
      "epoch": 1.7400774608101393,
      "grad_norm": 0.5415075829565839,
      "learning_rate": 8.727655978395089e-06,
      "loss": 11.7804,
      "step": 31955
    },
    {
      "epoch": 1.7401319148067222,
      "grad_norm": 0.5351938086840057,
      "learning_rate": 8.724053387641906e-06,
      "loss": 11.5577,
      "step": 31956
    },
    {
      "epoch": 1.7401863688033052,
      "grad_norm": 0.621849052620925,
      "learning_rate": 8.720451506665783e-06,
      "loss": 11.7246,
      "step": 31957
    },
    {
      "epoch": 1.7402408227998882,
      "grad_norm": 0.5492982585708485,
      "learning_rate": 8.716850335494742e-06,
      "loss": 11.7424,
      "step": 31958
    },
    {
      "epoch": 1.7402952767964712,
      "grad_norm": 0.5920599598120099,
      "learning_rate": 8.71324987415677e-06,
      "loss": 11.8464,
      "step": 31959
    },
    {
      "epoch": 1.7403497307930544,
      "grad_norm": 0.4947891504692365,
      "learning_rate": 8.70965012267988e-06,
      "loss": 11.7205,
      "step": 31960
    },
    {
      "epoch": 1.7404041847896374,
      "grad_norm": 0.5455025699705034,
      "learning_rate": 8.706051081092092e-06,
      "loss": 11.6804,
      "step": 31961
    },
    {
      "epoch": 1.7404586387862204,
      "grad_norm": 0.557191036658413,
      "learning_rate": 8.70245274942132e-06,
      "loss": 11.6383,
      "step": 31962
    },
    {
      "epoch": 1.7405130927828034,
      "grad_norm": 0.5340196800062514,
      "learning_rate": 8.698855127695605e-06,
      "loss": 11.7639,
      "step": 31963
    },
    {
      "epoch": 1.7405675467793864,
      "grad_norm": 0.5767153667836261,
      "learning_rate": 8.695258215942893e-06,
      "loss": 11.9129,
      "step": 31964
    },
    {
      "epoch": 1.7406220007759694,
      "grad_norm": 0.5694295553676907,
      "learning_rate": 8.691662014191159e-06,
      "loss": 11.8254,
      "step": 31965
    },
    {
      "epoch": 1.7406764547725526,
      "grad_norm": 0.6535280581606852,
      "learning_rate": 8.688066522468397e-06,
      "loss": 11.91,
      "step": 31966
    },
    {
      "epoch": 1.7407309087691356,
      "grad_norm": 0.5542515877471047,
      "learning_rate": 8.684471740802514e-06,
      "loss": 11.7171,
      "step": 31967
    },
    {
      "epoch": 1.7407853627657186,
      "grad_norm": 0.5526873203872726,
      "learning_rate": 8.680877669221522e-06,
      "loss": 11.912,
      "step": 31968
    },
    {
      "epoch": 1.7408398167623016,
      "grad_norm": 0.6768706919925701,
      "learning_rate": 8.67728430775332e-06,
      "loss": 11.9056,
      "step": 31969
    },
    {
      "epoch": 1.7408942707588846,
      "grad_norm": 0.5126020406449916,
      "learning_rate": 8.673691656425885e-06,
      "loss": 11.8263,
      "step": 31970
    },
    {
      "epoch": 1.7409487247554676,
      "grad_norm": 0.5439267010646247,
      "learning_rate": 8.670099715267132e-06,
      "loss": 11.7099,
      "step": 31971
    },
    {
      "epoch": 1.7410031787520506,
      "grad_norm": 0.4976696273610106,
      "learning_rate": 8.666508484304992e-06,
      "loss": 11.7321,
      "step": 31972
    },
    {
      "epoch": 1.7410576327486336,
      "grad_norm": 0.5508934392557755,
      "learning_rate": 8.662917963567418e-06,
      "loss": 11.8543,
      "step": 31973
    },
    {
      "epoch": 1.7411120867452166,
      "grad_norm": 0.5740364595544567,
      "learning_rate": 8.65932815308228e-06,
      "loss": 11.842,
      "step": 31974
    },
    {
      "epoch": 1.7411665407417996,
      "grad_norm": 0.5215353855019623,
      "learning_rate": 8.655739052877532e-06,
      "loss": 11.7305,
      "step": 31975
    },
    {
      "epoch": 1.7412209947383825,
      "grad_norm": 0.5614963324268551,
      "learning_rate": 8.652150662981095e-06,
      "loss": 11.8178,
      "step": 31976
    },
    {
      "epoch": 1.7412754487349655,
      "grad_norm": 0.5653651352357203,
      "learning_rate": 8.648562983420839e-06,
      "loss": 11.9051,
      "step": 31977
    },
    {
      "epoch": 1.7413299027315485,
      "grad_norm": 0.5334463137538656,
      "learning_rate": 8.644976014224692e-06,
      "loss": 11.7374,
      "step": 31978
    },
    {
      "epoch": 1.7413843567281315,
      "grad_norm": 0.576985678978604,
      "learning_rate": 8.641389755420515e-06,
      "loss": 11.729,
      "step": 31979
    },
    {
      "epoch": 1.7414388107247145,
      "grad_norm": 0.5234658387895171,
      "learning_rate": 8.637804207036226e-06,
      "loss": 11.7497,
      "step": 31980
    },
    {
      "epoch": 1.7414932647212975,
      "grad_norm": 0.6089813444056918,
      "learning_rate": 8.634219369099694e-06,
      "loss": 11.8424,
      "step": 31981
    },
    {
      "epoch": 1.7415477187178805,
      "grad_norm": 0.5935612717936091,
      "learning_rate": 8.630635241638773e-06,
      "loss": 11.8468,
      "step": 31982
    },
    {
      "epoch": 1.7416021727144637,
      "grad_norm": 0.5886032923980896,
      "learning_rate": 8.627051824681376e-06,
      "loss": 11.8731,
      "step": 31983
    },
    {
      "epoch": 1.7416566267110467,
      "grad_norm": 0.5198412867645322,
      "learning_rate": 8.623469118255334e-06,
      "loss": 11.8138,
      "step": 31984
    },
    {
      "epoch": 1.7417110807076297,
      "grad_norm": 0.49262438340279746,
      "learning_rate": 8.619887122388525e-06,
      "loss": 11.775,
      "step": 31985
    },
    {
      "epoch": 1.7417655347042127,
      "grad_norm": 0.5374674231944098,
      "learning_rate": 8.616305837108795e-06,
      "loss": 11.7372,
      "step": 31986
    },
    {
      "epoch": 1.7418199887007957,
      "grad_norm": 0.5033764033578693,
      "learning_rate": 8.612725262443989e-06,
      "loss": 11.7487,
      "step": 31987
    },
    {
      "epoch": 1.7418744426973787,
      "grad_norm": 0.6043909876110837,
      "learning_rate": 8.60914539842198e-06,
      "loss": 11.8433,
      "step": 31988
    },
    {
      "epoch": 1.741928896693962,
      "grad_norm": 0.5455036964850147,
      "learning_rate": 8.605566245070552e-06,
      "loss": 11.7161,
      "step": 31989
    },
    {
      "epoch": 1.7419833506905449,
      "grad_norm": 0.5409583913962226,
      "learning_rate": 8.601987802417599e-06,
      "loss": 11.7858,
      "step": 31990
    },
    {
      "epoch": 1.7420378046871279,
      "grad_norm": 0.5815138530271966,
      "learning_rate": 8.598410070490915e-06,
      "loss": 11.7551,
      "step": 31991
    },
    {
      "epoch": 1.7420922586837109,
      "grad_norm": 0.5446835716470613,
      "learning_rate": 8.594833049318297e-06,
      "loss": 11.7497,
      "step": 31992
    },
    {
      "epoch": 1.7421467126802939,
      "grad_norm": 0.5786796657330281,
      "learning_rate": 8.591256738927611e-06,
      "loss": 11.717,
      "step": 31993
    },
    {
      "epoch": 1.7422011666768769,
      "grad_norm": 0.5645904802426776,
      "learning_rate": 8.587681139346615e-06,
      "loss": 11.7751,
      "step": 31994
    },
    {
      "epoch": 1.7422556206734598,
      "grad_norm": 0.5709068463833342,
      "learning_rate": 8.584106250603164e-06,
      "loss": 11.8462,
      "step": 31995
    },
    {
      "epoch": 1.7423100746700428,
      "grad_norm": 0.5340152565702709,
      "learning_rate": 8.580532072725012e-06,
      "loss": 11.7117,
      "step": 31996
    },
    {
      "epoch": 1.7423645286666258,
      "grad_norm": 0.5387086854060601,
      "learning_rate": 8.576958605740004e-06,
      "loss": 11.817,
      "step": 31997
    },
    {
      "epoch": 1.7424189826632088,
      "grad_norm": 0.5140277553920145,
      "learning_rate": 8.573385849675863e-06,
      "loss": 11.7447,
      "step": 31998
    },
    {
      "epoch": 1.7424734366597918,
      "grad_norm": 0.5629323046688717,
      "learning_rate": 8.56981380456041e-06,
      "loss": 11.8892,
      "step": 31999
    },
    {
      "epoch": 1.7425278906563748,
      "grad_norm": 0.5431528813466309,
      "learning_rate": 8.566242470421448e-06,
      "loss": 11.7875,
      "step": 32000
    },
    {
      "epoch": 1.7425823446529578,
      "grad_norm": 0.5154491026682231,
      "learning_rate": 8.562671847286707e-06,
      "loss": 11.6852,
      "step": 32001
    },
    {
      "epoch": 1.7426367986495408,
      "grad_norm": 0.5892238812937414,
      "learning_rate": 8.559101935183944e-06,
      "loss": 11.7823,
      "step": 32002
    },
    {
      "epoch": 1.7426912526461238,
      "grad_norm": 0.5680434408571875,
      "learning_rate": 8.555532734140959e-06,
      "loss": 11.7483,
      "step": 32003
    },
    {
      "epoch": 1.7427457066427068,
      "grad_norm": 0.5022921114212071,
      "learning_rate": 8.551964244185474e-06,
      "loss": 11.7121,
      "step": 32004
    },
    {
      "epoch": 1.7428001606392898,
      "grad_norm": 0.5543481251024199,
      "learning_rate": 8.548396465345265e-06,
      "loss": 11.7405,
      "step": 32005
    },
    {
      "epoch": 1.7428546146358728,
      "grad_norm": 0.5417179749654835,
      "learning_rate": 8.544829397648046e-06,
      "loss": 11.7747,
      "step": 32006
    },
    {
      "epoch": 1.742909068632456,
      "grad_norm": 0.5116342225880156,
      "learning_rate": 8.541263041121584e-06,
      "loss": 11.7312,
      "step": 32007
    },
    {
      "epoch": 1.742963522629039,
      "grad_norm": 0.512291100772166,
      "learning_rate": 8.537697395793586e-06,
      "loss": 11.7384,
      "step": 32008
    },
    {
      "epoch": 1.743017976625622,
      "grad_norm": 0.6432909811509387,
      "learning_rate": 8.534132461691779e-06,
      "loss": 11.7537,
      "step": 32009
    },
    {
      "epoch": 1.743072430622205,
      "grad_norm": 0.5673547760782457,
      "learning_rate": 8.530568238843928e-06,
      "loss": 11.782,
      "step": 32010
    },
    {
      "epoch": 1.743126884618788,
      "grad_norm": 0.5607694313388282,
      "learning_rate": 8.52700472727771e-06,
      "loss": 11.6333,
      "step": 32011
    },
    {
      "epoch": 1.7431813386153712,
      "grad_norm": 0.4913740142745817,
      "learning_rate": 8.523441927020848e-06,
      "loss": 11.7803,
      "step": 32012
    },
    {
      "epoch": 1.7432357926119542,
      "grad_norm": 0.5694761407372012,
      "learning_rate": 8.519879838101031e-06,
      "loss": 11.7153,
      "step": 32013
    },
    {
      "epoch": 1.7432902466085372,
      "grad_norm": 0.6028894497907186,
      "learning_rate": 8.516318460545958e-06,
      "loss": 11.9751,
      "step": 32014
    },
    {
      "epoch": 1.7433447006051201,
      "grad_norm": 0.5507114254422816,
      "learning_rate": 8.512757794383353e-06,
      "loss": 11.8017,
      "step": 32015
    },
    {
      "epoch": 1.7433991546017031,
      "grad_norm": 0.5972184955355894,
      "learning_rate": 8.50919783964087e-06,
      "loss": 11.739,
      "step": 32016
    },
    {
      "epoch": 1.7434536085982861,
      "grad_norm": 0.5857228924141316,
      "learning_rate": 8.505638596346233e-06,
      "loss": 11.6848,
      "step": 32017
    },
    {
      "epoch": 1.7435080625948691,
      "grad_norm": 0.562031072838741,
      "learning_rate": 8.502080064527063e-06,
      "loss": 11.7777,
      "step": 32018
    },
    {
      "epoch": 1.7435625165914521,
      "grad_norm": 0.7034782977047336,
      "learning_rate": 8.498522244211093e-06,
      "loss": 11.835,
      "step": 32019
    },
    {
      "epoch": 1.743616970588035,
      "grad_norm": 0.5649368271146247,
      "learning_rate": 8.494965135425937e-06,
      "loss": 11.8731,
      "step": 32020
    },
    {
      "epoch": 1.743671424584618,
      "grad_norm": 0.5436977892164904,
      "learning_rate": 8.491408738199291e-06,
      "loss": 11.8892,
      "step": 32021
    },
    {
      "epoch": 1.743725878581201,
      "grad_norm": 0.5559043028084236,
      "learning_rate": 8.487853052558791e-06,
      "loss": 11.9061,
      "step": 32022
    },
    {
      "epoch": 1.743780332577784,
      "grad_norm": 0.5133931781143667,
      "learning_rate": 8.484298078532083e-06,
      "loss": 11.7108,
      "step": 32023
    },
    {
      "epoch": 1.743834786574367,
      "grad_norm": 0.5480535266969013,
      "learning_rate": 8.480743816146818e-06,
      "loss": 11.7956,
      "step": 32024
    },
    {
      "epoch": 1.74388924057095,
      "grad_norm": 0.5323917940609546,
      "learning_rate": 8.477190265430668e-06,
      "loss": 11.8297,
      "step": 32025
    },
    {
      "epoch": 1.743943694567533,
      "grad_norm": 0.5391719565254227,
      "learning_rate": 8.473637426411196e-06,
      "loss": 11.6837,
      "step": 32026
    },
    {
      "epoch": 1.743998148564116,
      "grad_norm": 0.5212312196844444,
      "learning_rate": 8.470085299116103e-06,
      "loss": 11.7863,
      "step": 32027
    },
    {
      "epoch": 1.744052602560699,
      "grad_norm": 0.5708364673038417,
      "learning_rate": 8.466533883572947e-06,
      "loss": 11.8323,
      "step": 32028
    },
    {
      "epoch": 1.744107056557282,
      "grad_norm": 0.5751547927852145,
      "learning_rate": 8.462983179809391e-06,
      "loss": 11.8152,
      "step": 32029
    },
    {
      "epoch": 1.7441615105538653,
      "grad_norm": 0.48722597878671126,
      "learning_rate": 8.459433187853016e-06,
      "loss": 11.8481,
      "step": 32030
    },
    {
      "epoch": 1.7442159645504483,
      "grad_norm": 0.5742185937981873,
      "learning_rate": 8.455883907731465e-06,
      "loss": 11.7642,
      "step": 32031
    },
    {
      "epoch": 1.7442704185470312,
      "grad_norm": 0.6257192235986521,
      "learning_rate": 8.452335339472305e-06,
      "loss": 11.8823,
      "step": 32032
    },
    {
      "epoch": 1.7443248725436142,
      "grad_norm": 0.527745318790352,
      "learning_rate": 8.448787483103116e-06,
      "loss": 11.7927,
      "step": 32033
    },
    {
      "epoch": 1.7443793265401972,
      "grad_norm": 0.6222700022992099,
      "learning_rate": 8.445240338651527e-06,
      "loss": 11.871,
      "step": 32034
    },
    {
      "epoch": 1.7444337805367802,
      "grad_norm": 0.5487261637021156,
      "learning_rate": 8.441693906145088e-06,
      "loss": 11.5871,
      "step": 32035
    },
    {
      "epoch": 1.7444882345333634,
      "grad_norm": 0.5047141327693643,
      "learning_rate": 8.438148185611395e-06,
      "loss": 11.6439,
      "step": 32036
    },
    {
      "epoch": 1.7445426885299464,
      "grad_norm": 0.5784583151736736,
      "learning_rate": 8.434603177078027e-06,
      "loss": 11.7275,
      "step": 32037
    },
    {
      "epoch": 1.7445971425265294,
      "grad_norm": 0.5229644104779019,
      "learning_rate": 8.43105888057253e-06,
      "loss": 11.6475,
      "step": 32038
    },
    {
      "epoch": 1.7446515965231124,
      "grad_norm": 0.5699041812977644,
      "learning_rate": 8.427515296122491e-06,
      "loss": 11.7752,
      "step": 32039
    },
    {
      "epoch": 1.7447060505196954,
      "grad_norm": 0.5957268317513674,
      "learning_rate": 8.423972423755433e-06,
      "loss": 11.9105,
      "step": 32040
    },
    {
      "epoch": 1.7447605045162784,
      "grad_norm": 0.539938225679523,
      "learning_rate": 8.420430263498935e-06,
      "loss": 11.7252,
      "step": 32041
    },
    {
      "epoch": 1.7448149585128614,
      "grad_norm": 0.5709212899677252,
      "learning_rate": 8.41688881538053e-06,
      "loss": 11.739,
      "step": 32042
    },
    {
      "epoch": 1.7448694125094444,
      "grad_norm": 0.5266446459566502,
      "learning_rate": 8.41334807942774e-06,
      "loss": 11.7513,
      "step": 32043
    },
    {
      "epoch": 1.7449238665060274,
      "grad_norm": 0.5677609274402388,
      "learning_rate": 8.409808055668134e-06,
      "loss": 11.8223,
      "step": 32044
    },
    {
      "epoch": 1.7449783205026104,
      "grad_norm": 0.5532136403402126,
      "learning_rate": 8.406268744129209e-06,
      "loss": 11.8705,
      "step": 32045
    },
    {
      "epoch": 1.7450327744991934,
      "grad_norm": 0.5151713012528072,
      "learning_rate": 8.4027301448385e-06,
      "loss": 11.7093,
      "step": 32046
    },
    {
      "epoch": 1.7450872284957764,
      "grad_norm": 0.6215215255628963,
      "learning_rate": 8.399192257823518e-06,
      "loss": 11.7937,
      "step": 32047
    },
    {
      "epoch": 1.7451416824923593,
      "grad_norm": 0.6201131430276449,
      "learning_rate": 8.395655083111776e-06,
      "loss": 11.8427,
      "step": 32048
    },
    {
      "epoch": 1.7451961364889423,
      "grad_norm": 0.5209650258702061,
      "learning_rate": 8.392118620730794e-06,
      "loss": 11.7063,
      "step": 32049
    },
    {
      "epoch": 1.7452505904855253,
      "grad_norm": 0.5612328279856644,
      "learning_rate": 8.38858287070805e-06,
      "loss": 11.9087,
      "step": 32050
    },
    {
      "epoch": 1.7453050444821083,
      "grad_norm": 0.564227927612779,
      "learning_rate": 8.385047833071058e-06,
      "loss": 11.6975,
      "step": 32051
    },
    {
      "epoch": 1.7453594984786913,
      "grad_norm": 0.567497662223573,
      "learning_rate": 8.381513507847306e-06,
      "loss": 11.8284,
      "step": 32052
    },
    {
      "epoch": 1.7454139524752745,
      "grad_norm": 0.5363259550569481,
      "learning_rate": 8.377979895064248e-06,
      "loss": 11.7159,
      "step": 32053
    },
    {
      "epoch": 1.7454684064718575,
      "grad_norm": 0.590642468655495,
      "learning_rate": 8.374446994749396e-06,
      "loss": 11.7312,
      "step": 32054
    },
    {
      "epoch": 1.7455228604684405,
      "grad_norm": 0.5465002820785123,
      "learning_rate": 8.370914806930198e-06,
      "loss": 11.6307,
      "step": 32055
    },
    {
      "epoch": 1.7455773144650235,
      "grad_norm": 0.5888990470494139,
      "learning_rate": 8.367383331634148e-06,
      "loss": 11.8342,
      "step": 32056
    },
    {
      "epoch": 1.7456317684616065,
      "grad_norm": 0.5746100887956731,
      "learning_rate": 8.363852568888686e-06,
      "loss": 11.6764,
      "step": 32057
    },
    {
      "epoch": 1.7456862224581895,
      "grad_norm": 0.5887223194598696,
      "learning_rate": 8.360322518721265e-06,
      "loss": 11.8901,
      "step": 32058
    },
    {
      "epoch": 1.7457406764547727,
      "grad_norm": 0.5108734790458672,
      "learning_rate": 8.35679318115935e-06,
      "loss": 11.7426,
      "step": 32059
    },
    {
      "epoch": 1.7457951304513557,
      "grad_norm": 0.5690170304757952,
      "learning_rate": 8.353264556230378e-06,
      "loss": 11.6658,
      "step": 32060
    },
    {
      "epoch": 1.7458495844479387,
      "grad_norm": 0.5176612902090224,
      "learning_rate": 8.349736643961813e-06,
      "loss": 11.6612,
      "step": 32061
    },
    {
      "epoch": 1.7459040384445217,
      "grad_norm": 0.5799500284494544,
      "learning_rate": 8.346209444381048e-06,
      "loss": 11.8357,
      "step": 32062
    },
    {
      "epoch": 1.7459584924411047,
      "grad_norm": 0.5396245797154937,
      "learning_rate": 8.342682957515513e-06,
      "loss": 11.8793,
      "step": 32063
    },
    {
      "epoch": 1.7460129464376877,
      "grad_norm": 0.5222696040881784,
      "learning_rate": 8.339157183392666e-06,
      "loss": 11.7805,
      "step": 32064
    },
    {
      "epoch": 1.7460674004342707,
      "grad_norm": 0.550740507352698,
      "learning_rate": 8.335632122039893e-06,
      "loss": 11.7638,
      "step": 32065
    },
    {
      "epoch": 1.7461218544308537,
      "grad_norm": 0.5574886696018079,
      "learning_rate": 8.332107773484633e-06,
      "loss": 11.7661,
      "step": 32066
    },
    {
      "epoch": 1.7461763084274367,
      "grad_norm": 0.6567091073380188,
      "learning_rate": 8.328584137754259e-06,
      "loss": 11.9116,
      "step": 32067
    },
    {
      "epoch": 1.7462307624240196,
      "grad_norm": 0.5897784852784564,
      "learning_rate": 8.325061214876195e-06,
      "loss": 11.6664,
      "step": 32068
    },
    {
      "epoch": 1.7462852164206026,
      "grad_norm": 0.5467779883960694,
      "learning_rate": 8.32153900487782e-06,
      "loss": 11.6983,
      "step": 32069
    },
    {
      "epoch": 1.7463396704171856,
      "grad_norm": 0.5084861598687963,
      "learning_rate": 8.318017507786535e-06,
      "loss": 11.6828,
      "step": 32070
    },
    {
      "epoch": 1.7463941244137686,
      "grad_norm": 0.5764218627558026,
      "learning_rate": 8.31449672362975e-06,
      "loss": 11.6875,
      "step": 32071
    },
    {
      "epoch": 1.7464485784103516,
      "grad_norm": 0.5343029023386447,
      "learning_rate": 8.310976652434776e-06,
      "loss": 11.8479,
      "step": 32072
    },
    {
      "epoch": 1.7465030324069346,
      "grad_norm": 0.5455757670649288,
      "learning_rate": 8.307457294229038e-06,
      "loss": 11.7445,
      "step": 32073
    },
    {
      "epoch": 1.7465574864035176,
      "grad_norm": 0.5637366874826736,
      "learning_rate": 8.303938649039888e-06,
      "loss": 11.5759,
      "step": 32074
    },
    {
      "epoch": 1.7466119404001006,
      "grad_norm": 0.541970138016841,
      "learning_rate": 8.300420716894686e-06,
      "loss": 11.8475,
      "step": 32075
    },
    {
      "epoch": 1.7466663943966836,
      "grad_norm": 0.568620399398183,
      "learning_rate": 8.296903497820808e-06,
      "loss": 11.8766,
      "step": 32076
    },
    {
      "epoch": 1.7467208483932668,
      "grad_norm": 0.5413186586308619,
      "learning_rate": 8.293386991845553e-06,
      "loss": 11.8918,
      "step": 32077
    },
    {
      "epoch": 1.7467753023898498,
      "grad_norm": 0.5654869300695888,
      "learning_rate": 8.289871198996334e-06,
      "loss": 11.8282,
      "step": 32078
    },
    {
      "epoch": 1.7468297563864328,
      "grad_norm": 0.599210642396545,
      "learning_rate": 8.286356119300432e-06,
      "loss": 11.8196,
      "step": 32079
    },
    {
      "epoch": 1.7468842103830158,
      "grad_norm": 0.5681529131139145,
      "learning_rate": 8.28284175278521e-06,
      "loss": 11.7322,
      "step": 32080
    },
    {
      "epoch": 1.7469386643795988,
      "grad_norm": 0.5764930737237418,
      "learning_rate": 8.279328099478023e-06,
      "loss": 11.8125,
      "step": 32081
    },
    {
      "epoch": 1.746993118376182,
      "grad_norm": 0.6045864425024923,
      "learning_rate": 8.27581515940612e-06,
      "loss": 11.7184,
      "step": 32082
    },
    {
      "epoch": 1.747047572372765,
      "grad_norm": 0.5662377367856796,
      "learning_rate": 8.272302932596888e-06,
      "loss": 11.9217,
      "step": 32083
    },
    {
      "epoch": 1.747102026369348,
      "grad_norm": 0.6057743589010771,
      "learning_rate": 8.268791419077592e-06,
      "loss": 11.8011,
      "step": 32084
    },
    {
      "epoch": 1.747156480365931,
      "grad_norm": 0.5613648064944241,
      "learning_rate": 8.265280618875559e-06,
      "loss": 11.8049,
      "step": 32085
    },
    {
      "epoch": 1.747210934362514,
      "grad_norm": 0.48816779325586207,
      "learning_rate": 8.261770532018098e-06,
      "loss": 11.6951,
      "step": 32086
    },
    {
      "epoch": 1.747265388359097,
      "grad_norm": 0.6015002130304541,
      "learning_rate": 8.258261158532487e-06,
      "loss": 11.8388,
      "step": 32087
    },
    {
      "epoch": 1.74731984235568,
      "grad_norm": 0.5379049160926599,
      "learning_rate": 8.254752498446028e-06,
      "loss": 11.8134,
      "step": 32088
    },
    {
      "epoch": 1.747374296352263,
      "grad_norm": 0.536591873539549,
      "learning_rate": 8.251244551785987e-06,
      "loss": 11.7769,
      "step": 32089
    },
    {
      "epoch": 1.747428750348846,
      "grad_norm": 0.5370086668121891,
      "learning_rate": 8.247737318579673e-06,
      "loss": 11.8478,
      "step": 32090
    },
    {
      "epoch": 1.747483204345429,
      "grad_norm": 0.5393090808096153,
      "learning_rate": 8.244230798854347e-06,
      "loss": 11.6601,
      "step": 32091
    },
    {
      "epoch": 1.747537658342012,
      "grad_norm": 0.6406910306153049,
      "learning_rate": 8.24072499263724e-06,
      "loss": 11.8621,
      "step": 32092
    },
    {
      "epoch": 1.747592112338595,
      "grad_norm": 0.5317494391056862,
      "learning_rate": 8.237219899955662e-06,
      "loss": 11.7992,
      "step": 32093
    },
    {
      "epoch": 1.747646566335178,
      "grad_norm": 0.5756798275330044,
      "learning_rate": 8.233715520836837e-06,
      "loss": 11.7773,
      "step": 32094
    },
    {
      "epoch": 1.747701020331761,
      "grad_norm": 0.5776534959828046,
      "learning_rate": 8.230211855308057e-06,
      "loss": 11.8673,
      "step": 32095
    },
    {
      "epoch": 1.7477554743283439,
      "grad_norm": 0.551942844959109,
      "learning_rate": 8.226708903396507e-06,
      "loss": 11.7723,
      "step": 32096
    },
    {
      "epoch": 1.7478099283249269,
      "grad_norm": 0.5698823976918336,
      "learning_rate": 8.223206665129468e-06,
      "loss": 11.649,
      "step": 32097
    },
    {
      "epoch": 1.7478643823215099,
      "grad_norm": 0.5585603105322928,
      "learning_rate": 8.219705140534173e-06,
      "loss": 11.837,
      "step": 32098
    },
    {
      "epoch": 1.7479188363180929,
      "grad_norm": 0.5582876530335303,
      "learning_rate": 8.216204329637834e-06,
      "loss": 11.7641,
      "step": 32099
    },
    {
      "epoch": 1.747973290314676,
      "grad_norm": 0.6105735570584604,
      "learning_rate": 8.212704232467694e-06,
      "loss": 11.7557,
      "step": 32100
    },
    {
      "epoch": 1.748027744311259,
      "grad_norm": 0.5067331004793603,
      "learning_rate": 8.209204849050944e-06,
      "loss": 11.7947,
      "step": 32101
    },
    {
      "epoch": 1.748082198307842,
      "grad_norm": 0.5115845141990338,
      "learning_rate": 8.205706179414829e-06,
      "loss": 11.7104,
      "step": 32102
    },
    {
      "epoch": 1.748136652304425,
      "grad_norm": 0.5998195876428078,
      "learning_rate": 8.202208223586538e-06,
      "loss": 11.8674,
      "step": 32103
    },
    {
      "epoch": 1.748191106301008,
      "grad_norm": 0.5431466281900563,
      "learning_rate": 8.198710981593249e-06,
      "loss": 11.8214,
      "step": 32104
    },
    {
      "epoch": 1.748245560297591,
      "grad_norm": 0.5681776478049104,
      "learning_rate": 8.195214453462196e-06,
      "loss": 11.7089,
      "step": 32105
    },
    {
      "epoch": 1.7483000142941743,
      "grad_norm": 0.5113870119080456,
      "learning_rate": 8.191718639220536e-06,
      "loss": 11.7082,
      "step": 32106
    },
    {
      "epoch": 1.7483544682907572,
      "grad_norm": 0.5602182821619474,
      "learning_rate": 8.188223538895456e-06,
      "loss": 11.7769,
      "step": 32107
    },
    {
      "epoch": 1.7484089222873402,
      "grad_norm": 0.5190456682225881,
      "learning_rate": 8.184729152514182e-06,
      "loss": 11.6851,
      "step": 32108
    },
    {
      "epoch": 1.7484633762839232,
      "grad_norm": 0.5102597510300653,
      "learning_rate": 8.181235480103822e-06,
      "loss": 11.7892,
      "step": 32109
    },
    {
      "epoch": 1.7485178302805062,
      "grad_norm": 0.5306065048933504,
      "learning_rate": 8.1777425216916e-06,
      "loss": 11.8607,
      "step": 32110
    },
    {
      "epoch": 1.7485722842770892,
      "grad_norm": 0.5225156835424806,
      "learning_rate": 8.174250277304628e-06,
      "loss": 11.7497,
      "step": 32111
    },
    {
      "epoch": 1.7486267382736722,
      "grad_norm": 0.5087099348069014,
      "learning_rate": 8.170758746970097e-06,
      "loss": 11.7138,
      "step": 32112
    },
    {
      "epoch": 1.7486811922702552,
      "grad_norm": 0.5325545484352462,
      "learning_rate": 8.167267930715161e-06,
      "loss": 11.6696,
      "step": 32113
    },
    {
      "epoch": 1.7487356462668382,
      "grad_norm": 0.5834730509714643,
      "learning_rate": 8.163777828566921e-06,
      "loss": 11.8353,
      "step": 32114
    },
    {
      "epoch": 1.7487901002634212,
      "grad_norm": 0.5603422351374289,
      "learning_rate": 8.160288440552565e-06,
      "loss": 11.6679,
      "step": 32115
    },
    {
      "epoch": 1.7488445542600042,
      "grad_norm": 0.509828969330385,
      "learning_rate": 8.156799766699186e-06,
      "loss": 11.6698,
      "step": 32116
    },
    {
      "epoch": 1.7488990082565872,
      "grad_norm": 0.5374525294611626,
      "learning_rate": 8.153311807033958e-06,
      "loss": 11.6937,
      "step": 32117
    },
    {
      "epoch": 1.7489534622531702,
      "grad_norm": 0.554043448104786,
      "learning_rate": 8.149824561583962e-06,
      "loss": 11.734,
      "step": 32118
    },
    {
      "epoch": 1.7490079162497532,
      "grad_norm": 0.5397036647979724,
      "learning_rate": 8.146338030376332e-06,
      "loss": 11.7771,
      "step": 32119
    },
    {
      "epoch": 1.7490623702463362,
      "grad_norm": 0.5295951749240384,
      "learning_rate": 8.14285221343819e-06,
      "loss": 11.7543,
      "step": 32120
    },
    {
      "epoch": 1.7491168242429191,
      "grad_norm": 0.5525421236400031,
      "learning_rate": 8.139367110796626e-06,
      "loss": 11.6253,
      "step": 32121
    },
    {
      "epoch": 1.7491712782395021,
      "grad_norm": 0.5072333533669457,
      "learning_rate": 8.135882722478772e-06,
      "loss": 11.7333,
      "step": 32122
    },
    {
      "epoch": 1.7492257322360854,
      "grad_norm": 0.4910920049852444,
      "learning_rate": 8.132399048511685e-06,
      "loss": 11.7351,
      "step": 32123
    },
    {
      "epoch": 1.7492801862326683,
      "grad_norm": 0.5801308156761865,
      "learning_rate": 8.128916088922467e-06,
      "loss": 11.6644,
      "step": 32124
    },
    {
      "epoch": 1.7493346402292513,
      "grad_norm": 0.6177879020362912,
      "learning_rate": 8.125433843738206e-06,
      "loss": 11.7532,
      "step": 32125
    },
    {
      "epoch": 1.7493890942258343,
      "grad_norm": 0.5496992994543061,
      "learning_rate": 8.12195231298597e-06,
      "loss": 11.8126,
      "step": 32126
    },
    {
      "epoch": 1.7494435482224173,
      "grad_norm": 0.5578115138425902,
      "learning_rate": 8.118471496692859e-06,
      "loss": 11.8202,
      "step": 32127
    },
    {
      "epoch": 1.7494980022190003,
      "grad_norm": 0.5027642151831847,
      "learning_rate": 8.114991394885908e-06,
      "loss": 11.6968,
      "step": 32128
    },
    {
      "epoch": 1.7495524562155835,
      "grad_norm": 0.51993964938592,
      "learning_rate": 8.111512007592181e-06,
      "loss": 11.7713,
      "step": 32129
    },
    {
      "epoch": 1.7496069102121665,
      "grad_norm": 0.5210930704701255,
      "learning_rate": 8.108033334838771e-06,
      "loss": 11.7823,
      "step": 32130
    },
    {
      "epoch": 1.7496613642087495,
      "grad_norm": 0.532312018316325,
      "learning_rate": 8.104555376652689e-06,
      "loss": 11.5907,
      "step": 32131
    },
    {
      "epoch": 1.7497158182053325,
      "grad_norm": 0.5969973555377257,
      "learning_rate": 8.101078133061025e-06,
      "loss": 11.8562,
      "step": 32132
    },
    {
      "epoch": 1.7497702722019155,
      "grad_norm": 0.5720486340891017,
      "learning_rate": 8.097601604090765e-06,
      "loss": 11.6376,
      "step": 32133
    },
    {
      "epoch": 1.7498247261984985,
      "grad_norm": 0.5672396675148712,
      "learning_rate": 8.094125789768959e-06,
      "loss": 11.8103,
      "step": 32134
    },
    {
      "epoch": 1.7498791801950815,
      "grad_norm": 0.5272982053436494,
      "learning_rate": 8.090650690122659e-06,
      "loss": 11.6549,
      "step": 32135
    },
    {
      "epoch": 1.7499336341916645,
      "grad_norm": 0.5139232150430204,
      "learning_rate": 8.087176305178856e-06,
      "loss": 11.7585,
      "step": 32136
    },
    {
      "epoch": 1.7499880881882475,
      "grad_norm": 0.5509708901767132,
      "learning_rate": 8.083702634964595e-06,
      "loss": 11.7857,
      "step": 32137
    },
    {
      "epoch": 1.7500425421848305,
      "grad_norm": 0.5744520180512087,
      "learning_rate": 8.080229679506868e-06,
      "loss": 11.7313,
      "step": 32138
    },
    {
      "epoch": 1.7500969961814135,
      "grad_norm": 0.5655000383945293,
      "learning_rate": 8.076757438832706e-06,
      "loss": 11.9291,
      "step": 32139
    },
    {
      "epoch": 1.7501514501779964,
      "grad_norm": 0.5862390754138455,
      "learning_rate": 8.073285912969075e-06,
      "loss": 11.8236,
      "step": 32140
    },
    {
      "epoch": 1.7502059041745794,
      "grad_norm": 0.5722237955712688,
      "learning_rate": 8.06981510194299e-06,
      "loss": 11.9508,
      "step": 32141
    },
    {
      "epoch": 1.7502603581711624,
      "grad_norm": 0.5175385689487015,
      "learning_rate": 8.066345005781473e-06,
      "loss": 11.7662,
      "step": 32142
    },
    {
      "epoch": 1.7503148121677454,
      "grad_norm": 0.5907442384318156,
      "learning_rate": 8.062875624511446e-06,
      "loss": 11.767,
      "step": 32143
    },
    {
      "epoch": 1.7503692661643284,
      "grad_norm": 0.5211366070284962,
      "learning_rate": 8.05940695815992e-06,
      "loss": 11.8284,
      "step": 32144
    },
    {
      "epoch": 1.7504237201609114,
      "grad_norm": 0.5708486403797793,
      "learning_rate": 8.055939006753855e-06,
      "loss": 11.8896,
      "step": 32145
    },
    {
      "epoch": 1.7504781741574946,
      "grad_norm": 0.5930456298382185,
      "learning_rate": 8.052471770320236e-06,
      "loss": 11.7559,
      "step": 32146
    },
    {
      "epoch": 1.7505326281540776,
      "grad_norm": 0.5351082534180633,
      "learning_rate": 8.049005248886021e-06,
      "loss": 11.7867,
      "step": 32147
    },
    {
      "epoch": 1.7505870821506606,
      "grad_norm": 0.5900071507189579,
      "learning_rate": 8.045539442478144e-06,
      "loss": 11.8933,
      "step": 32148
    },
    {
      "epoch": 1.7506415361472436,
      "grad_norm": 0.5572461041020155,
      "learning_rate": 8.042074351123596e-06,
      "loss": 11.9106,
      "step": 32149
    },
    {
      "epoch": 1.7506959901438266,
      "grad_norm": 0.6163914801817144,
      "learning_rate": 8.038609974849276e-06,
      "loss": 11.9965,
      "step": 32150
    },
    {
      "epoch": 1.7507504441404096,
      "grad_norm": 0.7564647393300263,
      "learning_rate": 8.035146313682173e-06,
      "loss": 11.8039,
      "step": 32151
    },
    {
      "epoch": 1.7508048981369928,
      "grad_norm": 0.48232305983174045,
      "learning_rate": 8.03168336764919e-06,
      "loss": 11.7294,
      "step": 32152
    },
    {
      "epoch": 1.7508593521335758,
      "grad_norm": 0.4935820149526705,
      "learning_rate": 8.028221136777237e-06,
      "loss": 11.8181,
      "step": 32153
    },
    {
      "epoch": 1.7509138061301588,
      "grad_norm": 0.5416017889021689,
      "learning_rate": 8.024759621093281e-06,
      "loss": 11.8525,
      "step": 32154
    },
    {
      "epoch": 1.7509682601267418,
      "grad_norm": 0.5129958786548087,
      "learning_rate": 8.021298820624212e-06,
      "loss": 11.7635,
      "step": 32155
    },
    {
      "epoch": 1.7510227141233248,
      "grad_norm": 0.5283839041939262,
      "learning_rate": 8.017838735396932e-06,
      "loss": 11.7942,
      "step": 32156
    },
    {
      "epoch": 1.7510771681199078,
      "grad_norm": 0.5214212576874,
      "learning_rate": 8.014379365438396e-06,
      "loss": 11.8274,
      "step": 32157
    },
    {
      "epoch": 1.7511316221164908,
      "grad_norm": 0.5444012761202118,
      "learning_rate": 8.010920710775437e-06,
      "loss": 11.7991,
      "step": 32158
    },
    {
      "epoch": 1.7511860761130738,
      "grad_norm": 0.5768476318510125,
      "learning_rate": 8.007462771435015e-06,
      "loss": 11.79,
      "step": 32159
    },
    {
      "epoch": 1.7512405301096567,
      "grad_norm": 0.5243840732132846,
      "learning_rate": 8.00400554744397e-06,
      "loss": 11.6581,
      "step": 32160
    },
    {
      "epoch": 1.7512949841062397,
      "grad_norm": 0.5432265632365408,
      "learning_rate": 8.00054903882922e-06,
      "loss": 11.8268,
      "step": 32161
    },
    {
      "epoch": 1.7513494381028227,
      "grad_norm": 0.49905626161452704,
      "learning_rate": 7.997093245617638e-06,
      "loss": 11.8295,
      "step": 32162
    },
    {
      "epoch": 1.7514038920994057,
      "grad_norm": 0.5509428819257189,
      "learning_rate": 7.99363816783606e-06,
      "loss": 11.8169,
      "step": 32163
    },
    {
      "epoch": 1.7514583460959887,
      "grad_norm": 0.5070706101941347,
      "learning_rate": 7.990183805511398e-06,
      "loss": 11.6974,
      "step": 32164
    },
    {
      "epoch": 1.7515128000925717,
      "grad_norm": 0.5655243994198613,
      "learning_rate": 7.986730158670485e-06,
      "loss": 11.8078,
      "step": 32165
    },
    {
      "epoch": 1.7515672540891547,
      "grad_norm": 0.6123305922211647,
      "learning_rate": 7.983277227340203e-06,
      "loss": 11.9129,
      "step": 32166
    },
    {
      "epoch": 1.7516217080857377,
      "grad_norm": 0.538501649298484,
      "learning_rate": 7.979825011547381e-06,
      "loss": 11.8663,
      "step": 32167
    },
    {
      "epoch": 1.7516761620823207,
      "grad_norm": 0.5830568664258794,
      "learning_rate": 7.976373511318857e-06,
      "loss": 11.7276,
      "step": 32168
    },
    {
      "epoch": 1.7517306160789037,
      "grad_norm": 0.5367117301081419,
      "learning_rate": 7.972922726681508e-06,
      "loss": 11.7983,
      "step": 32169
    },
    {
      "epoch": 1.751785070075487,
      "grad_norm": 0.528729286250279,
      "learning_rate": 7.969472657662136e-06,
      "loss": 11.6916,
      "step": 32170
    },
    {
      "epoch": 1.7518395240720699,
      "grad_norm": 0.48942031309086037,
      "learning_rate": 7.966023304287585e-06,
      "loss": 11.7504,
      "step": 32171
    },
    {
      "epoch": 1.7518939780686529,
      "grad_norm": 0.5260973860763183,
      "learning_rate": 7.962574666584676e-06,
      "loss": 11.8194,
      "step": 32172
    },
    {
      "epoch": 1.7519484320652359,
      "grad_norm": 0.6013993876569956,
      "learning_rate": 7.959126744580203e-06,
      "loss": 11.9182,
      "step": 32173
    },
    {
      "epoch": 1.7520028860618189,
      "grad_norm": 0.5291298509488411,
      "learning_rate": 7.955679538301008e-06,
      "loss": 11.6776,
      "step": 32174
    },
    {
      "epoch": 1.7520573400584019,
      "grad_norm": 0.5724929330919186,
      "learning_rate": 7.952233047773871e-06,
      "loss": 11.6852,
      "step": 32175
    },
    {
      "epoch": 1.752111794054985,
      "grad_norm": 0.5458027967153894,
      "learning_rate": 7.948787273025626e-06,
      "loss": 11.8398,
      "step": 32176
    },
    {
      "epoch": 1.752166248051568,
      "grad_norm": 0.5948731986463153,
      "learning_rate": 7.945342214083029e-06,
      "loss": 11.8019,
      "step": 32177
    },
    {
      "epoch": 1.752220702048151,
      "grad_norm": 0.5175419612460501,
      "learning_rate": 7.941897870972881e-06,
      "loss": 11.8234,
      "step": 32178
    },
    {
      "epoch": 1.752275156044734,
      "grad_norm": 0.5192168255454845,
      "learning_rate": 7.938454243722004e-06,
      "loss": 11.7962,
      "step": 32179
    },
    {
      "epoch": 1.752329610041317,
      "grad_norm": 0.5530401944175701,
      "learning_rate": 7.935011332357112e-06,
      "loss": 11.8509,
      "step": 32180
    },
    {
      "epoch": 1.7523840640379,
      "grad_norm": 0.5483546100052389,
      "learning_rate": 7.931569136905048e-06,
      "loss": 11.8307,
      "step": 32181
    },
    {
      "epoch": 1.752438518034483,
      "grad_norm": 0.5308029554244464,
      "learning_rate": 7.928127657392526e-06,
      "loss": 11.7503,
      "step": 32182
    },
    {
      "epoch": 1.752492972031066,
      "grad_norm": 0.6463200199516159,
      "learning_rate": 7.92468689384631e-06,
      "loss": 11.9085,
      "step": 32183
    },
    {
      "epoch": 1.752547426027649,
      "grad_norm": 0.6052874888046295,
      "learning_rate": 7.921246846293195e-06,
      "loss": 11.8176,
      "step": 32184
    },
    {
      "epoch": 1.752601880024232,
      "grad_norm": 0.5277665412074546,
      "learning_rate": 7.917807514759879e-06,
      "loss": 11.7246,
      "step": 32185
    },
    {
      "epoch": 1.752656334020815,
      "grad_norm": 0.5577879226699427,
      "learning_rate": 7.914368899273161e-06,
      "loss": 11.8003,
      "step": 32186
    },
    {
      "epoch": 1.752710788017398,
      "grad_norm": 0.5877162949416089,
      "learning_rate": 7.910930999859734e-06,
      "loss": 11.836,
      "step": 32187
    },
    {
      "epoch": 1.752765242013981,
      "grad_norm": 0.5768428727662831,
      "learning_rate": 7.907493816546362e-06,
      "loss": 11.7091,
      "step": 32188
    },
    {
      "epoch": 1.752819696010564,
      "grad_norm": 0.5081975865111498,
      "learning_rate": 7.90405734935975e-06,
      "loss": 11.7753,
      "step": 32189
    },
    {
      "epoch": 1.752874150007147,
      "grad_norm": 0.5409616280016912,
      "learning_rate": 7.90062159832663e-06,
      "loss": 11.8314,
      "step": 32190
    },
    {
      "epoch": 1.75292860400373,
      "grad_norm": 0.6280703923892754,
      "learning_rate": 7.897186563473735e-06,
      "loss": 11.8578,
      "step": 32191
    },
    {
      "epoch": 1.752983058000313,
      "grad_norm": 0.5890986189124914,
      "learning_rate": 7.893752244827768e-06,
      "loss": 11.787,
      "step": 32192
    },
    {
      "epoch": 1.7530375119968962,
      "grad_norm": 0.5070755262227107,
      "learning_rate": 7.89031864241543e-06,
      "loss": 11.7177,
      "step": 32193
    },
    {
      "epoch": 1.7530919659934792,
      "grad_norm": 0.49700882147831205,
      "learning_rate": 7.886885756263407e-06,
      "loss": 11.8008,
      "step": 32194
    },
    {
      "epoch": 1.7531464199900622,
      "grad_norm": 0.5594109306667665,
      "learning_rate": 7.883453586398404e-06,
      "loss": 11.7856,
      "step": 32195
    },
    {
      "epoch": 1.7532008739866451,
      "grad_norm": 0.5707705400748605,
      "learning_rate": 7.88002213284712e-06,
      "loss": 11.7651,
      "step": 32196
    },
    {
      "epoch": 1.7532553279832281,
      "grad_norm": 0.52791498707753,
      "learning_rate": 7.876591395636234e-06,
      "loss": 11.799,
      "step": 32197
    },
    {
      "epoch": 1.7533097819798111,
      "grad_norm": 0.5885877943751425,
      "learning_rate": 7.873161374792426e-06,
      "loss": 11.6264,
      "step": 32198
    },
    {
      "epoch": 1.7533642359763943,
      "grad_norm": 0.5577006781823926,
      "learning_rate": 7.86973207034235e-06,
      "loss": 11.71,
      "step": 32199
    },
    {
      "epoch": 1.7534186899729773,
      "grad_norm": 0.5008494017413169,
      "learning_rate": 7.86630348231271e-06,
      "loss": 11.7965,
      "step": 32200
    },
    {
      "epoch": 1.7534731439695603,
      "grad_norm": 0.5661881923445657,
      "learning_rate": 7.862875610730125e-06,
      "loss": 11.884,
      "step": 32201
    },
    {
      "epoch": 1.7535275979661433,
      "grad_norm": 0.6334994442026655,
      "learning_rate": 7.859448455621288e-06,
      "loss": 11.8528,
      "step": 32202
    },
    {
      "epoch": 1.7535820519627263,
      "grad_norm": 0.5008431638330966,
      "learning_rate": 7.856022017012832e-06,
      "loss": 11.7394,
      "step": 32203
    },
    {
      "epoch": 1.7536365059593093,
      "grad_norm": 0.5261220463483217,
      "learning_rate": 7.852596294931391e-06,
      "loss": 11.7669,
      "step": 32204
    },
    {
      "epoch": 1.7536909599558923,
      "grad_norm": 0.5660318987309142,
      "learning_rate": 7.8491712894036e-06,
      "loss": 11.7858,
      "step": 32205
    },
    {
      "epoch": 1.7537454139524753,
      "grad_norm": 0.5383905058425613,
      "learning_rate": 7.845747000456138e-06,
      "loss": 11.7916,
      "step": 32206
    },
    {
      "epoch": 1.7537998679490583,
      "grad_norm": 0.5486025497378929,
      "learning_rate": 7.842323428115572e-06,
      "loss": 11.8162,
      "step": 32207
    },
    {
      "epoch": 1.7538543219456413,
      "grad_norm": 0.5545640399730635,
      "learning_rate": 7.838900572408581e-06,
      "loss": 11.812,
      "step": 32208
    },
    {
      "epoch": 1.7539087759422243,
      "grad_norm": 0.5135894571875912,
      "learning_rate": 7.835478433361732e-06,
      "loss": 11.7503,
      "step": 32209
    },
    {
      "epoch": 1.7539632299388073,
      "grad_norm": 0.5727064257373675,
      "learning_rate": 7.83205701100168e-06,
      "loss": 11.7242,
      "step": 32210
    },
    {
      "epoch": 1.7540176839353903,
      "grad_norm": 0.48632272167826107,
      "learning_rate": 7.828636305354986e-06,
      "loss": 11.8623,
      "step": 32211
    },
    {
      "epoch": 1.7540721379319733,
      "grad_norm": 0.5449425851250413,
      "learning_rate": 7.825216316448292e-06,
      "loss": 11.7668,
      "step": 32212
    },
    {
      "epoch": 1.7541265919285562,
      "grad_norm": 0.5409883419744331,
      "learning_rate": 7.821797044308177e-06,
      "loss": 11.8818,
      "step": 32213
    },
    {
      "epoch": 1.7541810459251392,
      "grad_norm": 0.5330606175454291,
      "learning_rate": 7.818378488961208e-06,
      "loss": 11.8365,
      "step": 32214
    },
    {
      "epoch": 1.7542354999217222,
      "grad_norm": 0.5497375834604316,
      "learning_rate": 7.814960650434011e-06,
      "loss": 11.8318,
      "step": 32215
    },
    {
      "epoch": 1.7542899539183054,
      "grad_norm": 0.6446595210524717,
      "learning_rate": 7.811543528753106e-06,
      "loss": 11.8336,
      "step": 32216
    },
    {
      "epoch": 1.7543444079148884,
      "grad_norm": 0.5698699944627266,
      "learning_rate": 7.808127123945108e-06,
      "loss": 11.6559,
      "step": 32217
    },
    {
      "epoch": 1.7543988619114714,
      "grad_norm": 0.522611867715547,
      "learning_rate": 7.804711436036593e-06,
      "loss": 11.7201,
      "step": 32218
    },
    {
      "epoch": 1.7544533159080544,
      "grad_norm": 0.5714231990359632,
      "learning_rate": 7.801296465054087e-06,
      "loss": 11.8032,
      "step": 32219
    },
    {
      "epoch": 1.7545077699046374,
      "grad_norm": 0.540221801484937,
      "learning_rate": 7.797882211024177e-06,
      "loss": 11.8662,
      "step": 32220
    },
    {
      "epoch": 1.7545622239012204,
      "grad_norm": 0.5585542781002193,
      "learning_rate": 7.794468673973376e-06,
      "loss": 11.6747,
      "step": 32221
    },
    {
      "epoch": 1.7546166778978036,
      "grad_norm": 0.5320363297074058,
      "learning_rate": 7.791055853928264e-06,
      "loss": 11.7608,
      "step": 32222
    },
    {
      "epoch": 1.7546711318943866,
      "grad_norm": 0.5462074931922277,
      "learning_rate": 7.787643750915374e-06,
      "loss": 11.559,
      "step": 32223
    },
    {
      "epoch": 1.7547255858909696,
      "grad_norm": 0.535820239283221,
      "learning_rate": 7.784232364961208e-06,
      "loss": 11.7968,
      "step": 32224
    },
    {
      "epoch": 1.7547800398875526,
      "grad_norm": 0.5547533369282873,
      "learning_rate": 7.780821696092333e-06,
      "loss": 11.877,
      "step": 32225
    },
    {
      "epoch": 1.7548344938841356,
      "grad_norm": 0.6763425151967221,
      "learning_rate": 7.777411744335238e-06,
      "loss": 11.8638,
      "step": 32226
    },
    {
      "epoch": 1.7548889478807186,
      "grad_norm": 0.5207995028623225,
      "learning_rate": 7.77400250971645e-06,
      "loss": 11.7608,
      "step": 32227
    },
    {
      "epoch": 1.7549434018773016,
      "grad_norm": 0.5806136414228242,
      "learning_rate": 7.77059399226251e-06,
      "loss": 11.7124,
      "step": 32228
    },
    {
      "epoch": 1.7549978558738846,
      "grad_norm": 0.5977322860435186,
      "learning_rate": 7.767186191999876e-06,
      "loss": 11.6959,
      "step": 32229
    },
    {
      "epoch": 1.7550523098704676,
      "grad_norm": 0.5416504793801568,
      "learning_rate": 7.763779108955094e-06,
      "loss": 11.7832,
      "step": 32230
    },
    {
      "epoch": 1.7551067638670506,
      "grad_norm": 0.567089416905431,
      "learning_rate": 7.76037274315461e-06,
      "loss": 11.8185,
      "step": 32231
    },
    {
      "epoch": 1.7551612178636335,
      "grad_norm": 0.5591491809986737,
      "learning_rate": 7.756967094624946e-06,
      "loss": 11.7368,
      "step": 32232
    },
    {
      "epoch": 1.7552156718602165,
      "grad_norm": 0.5325861996181607,
      "learning_rate": 7.753562163392592e-06,
      "loss": 11.7578,
      "step": 32233
    },
    {
      "epoch": 1.7552701258567995,
      "grad_norm": 0.5506910665304571,
      "learning_rate": 7.750157949483983e-06,
      "loss": 11.8254,
      "step": 32234
    },
    {
      "epoch": 1.7553245798533825,
      "grad_norm": 0.5056868603709823,
      "learning_rate": 7.746754452925631e-06,
      "loss": 11.7153,
      "step": 32235
    },
    {
      "epoch": 1.7553790338499655,
      "grad_norm": 0.5344924082242175,
      "learning_rate": 7.74335167374397e-06,
      "loss": 11.7105,
      "step": 32236
    },
    {
      "epoch": 1.7554334878465485,
      "grad_norm": 0.5599740681400854,
      "learning_rate": 7.739949611965491e-06,
      "loss": 11.8362,
      "step": 32237
    },
    {
      "epoch": 1.7554879418431315,
      "grad_norm": 0.5271103339849061,
      "learning_rate": 7.736548267616628e-06,
      "loss": 11.5224,
      "step": 32238
    },
    {
      "epoch": 1.7555423958397145,
      "grad_norm": 0.5685243285199002,
      "learning_rate": 7.733147640723837e-06,
      "loss": 11.946,
      "step": 32239
    },
    {
      "epoch": 1.7555968498362977,
      "grad_norm": 0.5300780559426779,
      "learning_rate": 7.729747731313574e-06,
      "loss": 11.7598,
      "step": 32240
    },
    {
      "epoch": 1.7556513038328807,
      "grad_norm": 0.5029465535898692,
      "learning_rate": 7.726348539412254e-06,
      "loss": 11.7824,
      "step": 32241
    },
    {
      "epoch": 1.7557057578294637,
      "grad_norm": 0.5929412511542949,
      "learning_rate": 7.72295006504633e-06,
      "loss": 11.836,
      "step": 32242
    },
    {
      "epoch": 1.7557602118260467,
      "grad_norm": 0.565940440276524,
      "learning_rate": 7.719552308242239e-06,
      "loss": 11.7975,
      "step": 32243
    },
    {
      "epoch": 1.7558146658226297,
      "grad_norm": 0.5574182071318897,
      "learning_rate": 7.716155269026349e-06,
      "loss": 11.9,
      "step": 32244
    },
    {
      "epoch": 1.7558691198192127,
      "grad_norm": 0.49909450072414313,
      "learning_rate": 7.712758947425147e-06,
      "loss": 11.7809,
      "step": 32245
    },
    {
      "epoch": 1.755923573815796,
      "grad_norm": 0.5235263892340182,
      "learning_rate": 7.709363343464982e-06,
      "loss": 11.7689,
      "step": 32246
    },
    {
      "epoch": 1.7559780278123789,
      "grad_norm": 0.5664352797917843,
      "learning_rate": 7.705968457172297e-06,
      "loss": 11.7751,
      "step": 32247
    },
    {
      "epoch": 1.7560324818089619,
      "grad_norm": 0.5830060694994326,
      "learning_rate": 7.702574288573461e-06,
      "loss": 11.8712,
      "step": 32248
    },
    {
      "epoch": 1.7560869358055449,
      "grad_norm": 0.5566821559668119,
      "learning_rate": 7.699180837694908e-06,
      "loss": 11.8874,
      "step": 32249
    },
    {
      "epoch": 1.7561413898021279,
      "grad_norm": 0.5491128300321498,
      "learning_rate": 7.695788104562984e-06,
      "loss": 11.9275,
      "step": 32250
    },
    {
      "epoch": 1.7561958437987109,
      "grad_norm": 0.5680822424895551,
      "learning_rate": 7.692396089204101e-06,
      "loss": 11.8738,
      "step": 32251
    },
    {
      "epoch": 1.7562502977952938,
      "grad_norm": 0.6692234037475192,
      "learning_rate": 7.68900479164464e-06,
      "loss": 11.7829,
      "step": 32252
    },
    {
      "epoch": 1.7563047517918768,
      "grad_norm": 0.5513177088985287,
      "learning_rate": 7.685614211910951e-06,
      "loss": 11.7933,
      "step": 32253
    },
    {
      "epoch": 1.7563592057884598,
      "grad_norm": 0.6010922202352793,
      "learning_rate": 7.682224350029387e-06,
      "loss": 11.7873,
      "step": 32254
    },
    {
      "epoch": 1.7564136597850428,
      "grad_norm": 0.5554814044252768,
      "learning_rate": 7.678835206026357e-06,
      "loss": 11.7286,
      "step": 32255
    },
    {
      "epoch": 1.7564681137816258,
      "grad_norm": 0.5534693096387924,
      "learning_rate": 7.675446779928163e-06,
      "loss": 11.8511,
      "step": 32256
    },
    {
      "epoch": 1.7565225677782088,
      "grad_norm": 0.5657564163925257,
      "learning_rate": 7.672059071761196e-06,
      "loss": 11.8573,
      "step": 32257
    },
    {
      "epoch": 1.7565770217747918,
      "grad_norm": 0.5257295790399134,
      "learning_rate": 7.668672081551765e-06,
      "loss": 11.8729,
      "step": 32258
    },
    {
      "epoch": 1.7566314757713748,
      "grad_norm": 0.5803789243855803,
      "learning_rate": 7.665285809326239e-06,
      "loss": 11.8916,
      "step": 32259
    },
    {
      "epoch": 1.7566859297679578,
      "grad_norm": 0.6112079479631757,
      "learning_rate": 7.66190025511091e-06,
      "loss": 11.8393,
      "step": 32260
    },
    {
      "epoch": 1.7567403837645408,
      "grad_norm": 0.5415121407551673,
      "learning_rate": 7.658515418932144e-06,
      "loss": 11.868,
      "step": 32261
    },
    {
      "epoch": 1.7567948377611238,
      "grad_norm": 0.5504172781674589,
      "learning_rate": 7.655131300816254e-06,
      "loss": 11.7321,
      "step": 32262
    },
    {
      "epoch": 1.756849291757707,
      "grad_norm": 0.5419586328626931,
      "learning_rate": 7.651747900789551e-06,
      "loss": 11.7968,
      "step": 32263
    },
    {
      "epoch": 1.75690374575429,
      "grad_norm": 0.49878985276194937,
      "learning_rate": 7.64836521887835e-06,
      "loss": 11.8255,
      "step": 32264
    },
    {
      "epoch": 1.756958199750873,
      "grad_norm": 0.5384417709443856,
      "learning_rate": 7.644983255108928e-06,
      "loss": 11.6968,
      "step": 32265
    },
    {
      "epoch": 1.757012653747456,
      "grad_norm": 0.5099488674337725,
      "learning_rate": 7.64160200950761e-06,
      "loss": 11.6918,
      "step": 32266
    },
    {
      "epoch": 1.757067107744039,
      "grad_norm": 0.5744231889499453,
      "learning_rate": 7.638221482100694e-06,
      "loss": 11.7467,
      "step": 32267
    },
    {
      "epoch": 1.757121561740622,
      "grad_norm": 0.6547415147829072,
      "learning_rate": 7.63484167291444e-06,
      "loss": 11.7437,
      "step": 32268
    },
    {
      "epoch": 1.7571760157372052,
      "grad_norm": 0.5257627177962527,
      "learning_rate": 7.631462581975158e-06,
      "loss": 11.6708,
      "step": 32269
    },
    {
      "epoch": 1.7572304697337882,
      "grad_norm": 0.5846229757023956,
      "learning_rate": 7.628084209309106e-06,
      "loss": 11.8139,
      "step": 32270
    },
    {
      "epoch": 1.7572849237303712,
      "grad_norm": 0.5961961171614157,
      "learning_rate": 7.624706554942573e-06,
      "loss": 11.8151,
      "step": 32271
    },
    {
      "epoch": 1.7573393777269541,
      "grad_norm": 0.5271634897816913,
      "learning_rate": 7.621329618901796e-06,
      "loss": 11.8484,
      "step": 32272
    },
    {
      "epoch": 1.7573938317235371,
      "grad_norm": 0.490433808307365,
      "learning_rate": 7.617953401213074e-06,
      "loss": 11.7982,
      "step": 32273
    },
    {
      "epoch": 1.7574482857201201,
      "grad_norm": 0.5442882276878024,
      "learning_rate": 7.614577901902631e-06,
      "loss": 11.7319,
      "step": 32274
    },
    {
      "epoch": 1.7575027397167031,
      "grad_norm": 0.5778531847508808,
      "learning_rate": 7.6112031209967125e-06,
      "loss": 11.8007,
      "step": 32275
    },
    {
      "epoch": 1.7575571937132861,
      "grad_norm": 0.494578925015559,
      "learning_rate": 7.607829058521576e-06,
      "loss": 11.7508,
      "step": 32276
    },
    {
      "epoch": 1.757611647709869,
      "grad_norm": 0.531911092636844,
      "learning_rate": 7.604455714503467e-06,
      "loss": 11.7154,
      "step": 32277
    },
    {
      "epoch": 1.757666101706452,
      "grad_norm": 0.5440454427937312,
      "learning_rate": 7.601083088968586e-06,
      "loss": 11.8183,
      "step": 32278
    },
    {
      "epoch": 1.757720555703035,
      "grad_norm": 0.4721846534624302,
      "learning_rate": 7.59771118194319e-06,
      "loss": 11.5942,
      "step": 32279
    },
    {
      "epoch": 1.757775009699618,
      "grad_norm": 0.5711021111911527,
      "learning_rate": 7.594339993453481e-06,
      "loss": 11.7782,
      "step": 32280
    },
    {
      "epoch": 1.757829463696201,
      "grad_norm": 0.5456321476474515,
      "learning_rate": 7.590969523525693e-06,
      "loss": 11.8084,
      "step": 32281
    },
    {
      "epoch": 1.757883917692784,
      "grad_norm": 0.6153892555574109,
      "learning_rate": 7.587599772186005e-06,
      "loss": 11.7252,
      "step": 32282
    },
    {
      "epoch": 1.757938371689367,
      "grad_norm": 0.5864348175280548,
      "learning_rate": 7.584230739460663e-06,
      "loss": 11.8638,
      "step": 32283
    },
    {
      "epoch": 1.75799282568595,
      "grad_norm": 0.5530913718208338,
      "learning_rate": 7.5808624253758345e-06,
      "loss": 11.7893,
      "step": 32284
    },
    {
      "epoch": 1.758047279682533,
      "grad_norm": 0.5258033004590266,
      "learning_rate": 7.577494829957698e-06,
      "loss": 11.8255,
      "step": 32285
    },
    {
      "epoch": 1.7581017336791163,
      "grad_norm": 0.5209343895028691,
      "learning_rate": 7.57412795323249e-06,
      "loss": 11.7554,
      "step": 32286
    },
    {
      "epoch": 1.7581561876756993,
      "grad_norm": 0.6293390912174309,
      "learning_rate": 7.570761795226333e-06,
      "loss": 11.704,
      "step": 32287
    },
    {
      "epoch": 1.7582106416722822,
      "grad_norm": 0.5319007120541434,
      "learning_rate": 7.567396355965439e-06,
      "loss": 11.8324,
      "step": 32288
    },
    {
      "epoch": 1.7582650956688652,
      "grad_norm": 0.580508535233131,
      "learning_rate": 7.564031635475988e-06,
      "loss": 11.9045,
      "step": 32289
    },
    {
      "epoch": 1.7583195496654482,
      "grad_norm": 0.5068231410167017,
      "learning_rate": 7.560667633784113e-06,
      "loss": 11.7702,
      "step": 32290
    },
    {
      "epoch": 1.7583740036620312,
      "grad_norm": 0.5250683827358754,
      "learning_rate": 7.5573043509160055e-06,
      "loss": 11.7798,
      "step": 32291
    },
    {
      "epoch": 1.7584284576586144,
      "grad_norm": 0.5100140078716555,
      "learning_rate": 7.553941786897778e-06,
      "loss": 11.7662,
      "step": 32292
    },
    {
      "epoch": 1.7584829116551974,
      "grad_norm": 0.5553137085994898,
      "learning_rate": 7.5505799417556315e-06,
      "loss": 11.6263,
      "step": 32293
    },
    {
      "epoch": 1.7585373656517804,
      "grad_norm": 0.49266713690707487,
      "learning_rate": 7.5472188155156666e-06,
      "loss": 11.682,
      "step": 32294
    },
    {
      "epoch": 1.7585918196483634,
      "grad_norm": 0.6093620570986684,
      "learning_rate": 7.54385840820403e-06,
      "loss": 11.726,
      "step": 32295
    },
    {
      "epoch": 1.7586462736449464,
      "grad_norm": 0.5638044436141385,
      "learning_rate": 7.540498719846856e-06,
      "loss": 11.7721,
      "step": 32296
    },
    {
      "epoch": 1.7587007276415294,
      "grad_norm": 0.5268638154805048,
      "learning_rate": 7.537139750470268e-06,
      "loss": 11.7714,
      "step": 32297
    },
    {
      "epoch": 1.7587551816381124,
      "grad_norm": 0.5547729103595103,
      "learning_rate": 7.5337815001003895e-06,
      "loss": 11.6223,
      "step": 32298
    },
    {
      "epoch": 1.7588096356346954,
      "grad_norm": 0.5643332879019087,
      "learning_rate": 7.530423968763323e-06,
      "loss": 11.7611,
      "step": 32299
    },
    {
      "epoch": 1.7588640896312784,
      "grad_norm": 0.5092644379332668,
      "learning_rate": 7.52706715648519e-06,
      "loss": 11.7484,
      "step": 32300
    },
    {
      "epoch": 1.7589185436278614,
      "grad_norm": 0.6123002738172678,
      "learning_rate": 7.523711063292105e-06,
      "loss": 12.0026,
      "step": 32301
    },
    {
      "epoch": 1.7589729976244444,
      "grad_norm": 0.5364123042148184,
      "learning_rate": 7.520355689210134e-06,
      "loss": 11.615,
      "step": 32302
    },
    {
      "epoch": 1.7590274516210274,
      "grad_norm": 0.5853044926867247,
      "learning_rate": 7.517001034265403e-06,
      "loss": 11.8361,
      "step": 32303
    },
    {
      "epoch": 1.7590819056176104,
      "grad_norm": 0.5177055192904207,
      "learning_rate": 7.513647098483978e-06,
      "loss": 11.8876,
      "step": 32304
    },
    {
      "epoch": 1.7591363596141933,
      "grad_norm": 0.709516664642074,
      "learning_rate": 7.5102938818919274e-06,
      "loss": 11.888,
      "step": 32305
    },
    {
      "epoch": 1.7591908136107763,
      "grad_norm": 0.5587832619399481,
      "learning_rate": 7.506941384515365e-06,
      "loss": 11.7161,
      "step": 32306
    },
    {
      "epoch": 1.7592452676073593,
      "grad_norm": 0.5509328295572602,
      "learning_rate": 7.503589606380312e-06,
      "loss": 11.8065,
      "step": 32307
    },
    {
      "epoch": 1.7592997216039423,
      "grad_norm": 0.5763956979013817,
      "learning_rate": 7.5002385475128835e-06,
      "loss": 11.7853,
      "step": 32308
    },
    {
      "epoch": 1.7593541756005253,
      "grad_norm": 0.5897447799682455,
      "learning_rate": 7.4968882079390905e-06,
      "loss": 11.8323,
      "step": 32309
    },
    {
      "epoch": 1.7594086295971085,
      "grad_norm": 0.5614212276382561,
      "learning_rate": 7.4935385876850114e-06,
      "loss": 11.8205,
      "step": 32310
    },
    {
      "epoch": 1.7594630835936915,
      "grad_norm": 0.5580453990328673,
      "learning_rate": 7.4901896867766944e-06,
      "loss": 11.8253,
      "step": 32311
    },
    {
      "epoch": 1.7595175375902745,
      "grad_norm": 0.5768845641338056,
      "learning_rate": 7.486841505240172e-06,
      "loss": 11.8223,
      "step": 32312
    },
    {
      "epoch": 1.7595719915868575,
      "grad_norm": 0.5870514910026379,
      "learning_rate": 7.483494043101514e-06,
      "loss": 11.848,
      "step": 32313
    },
    {
      "epoch": 1.7596264455834405,
      "grad_norm": 0.5314168646702158,
      "learning_rate": 7.4801473003866864e-06,
      "loss": 11.86,
      "step": 32314
    },
    {
      "epoch": 1.7596808995800237,
      "grad_norm": 0.620104276143737,
      "learning_rate": 7.476801277121748e-06,
      "loss": 11.8212,
      "step": 32315
    },
    {
      "epoch": 1.7597353535766067,
      "grad_norm": 0.5435303021810113,
      "learning_rate": 7.473455973332743e-06,
      "loss": 11.7835,
      "step": 32316
    },
    {
      "epoch": 1.7597898075731897,
      "grad_norm": 0.5963718671984589,
      "learning_rate": 7.470111389045642e-06,
      "loss": 11.7939,
      "step": 32317
    },
    {
      "epoch": 1.7598442615697727,
      "grad_norm": 0.5304119456924522,
      "learning_rate": 7.466767524286477e-06,
      "loss": 11.8243,
      "step": 32318
    },
    {
      "epoch": 1.7598987155663557,
      "grad_norm": 0.553422404781632,
      "learning_rate": 7.463424379081241e-06,
      "loss": 11.9248,
      "step": 32319
    },
    {
      "epoch": 1.7599531695629387,
      "grad_norm": 0.5249786498460067,
      "learning_rate": 7.460081953455955e-06,
      "loss": 11.8219,
      "step": 32320
    },
    {
      "epoch": 1.7600076235595217,
      "grad_norm": 0.5083800382337074,
      "learning_rate": 7.456740247436567e-06,
      "loss": 11.8487,
      "step": 32321
    },
    {
      "epoch": 1.7600620775561047,
      "grad_norm": 0.5440220910278729,
      "learning_rate": 7.4533992610490875e-06,
      "loss": 11.7141,
      "step": 32322
    },
    {
      "epoch": 1.7601165315526877,
      "grad_norm": 0.48060071553130085,
      "learning_rate": 7.450058994319531e-06,
      "loss": 11.7604,
      "step": 32323
    },
    {
      "epoch": 1.7601709855492707,
      "grad_norm": 0.5719979749730558,
      "learning_rate": 7.446719447273798e-06,
      "loss": 11.8195,
      "step": 32324
    },
    {
      "epoch": 1.7602254395458536,
      "grad_norm": 0.535382825884724,
      "learning_rate": 7.443380619937901e-06,
      "loss": 11.7395,
      "step": 32325
    },
    {
      "epoch": 1.7602798935424366,
      "grad_norm": 0.6219871745107087,
      "learning_rate": 7.440042512337808e-06,
      "loss": 11.9031,
      "step": 32326
    },
    {
      "epoch": 1.7603343475390196,
      "grad_norm": 0.5571690019897455,
      "learning_rate": 7.4367051244994545e-06,
      "loss": 11.8527,
      "step": 32327
    },
    {
      "epoch": 1.7603888015356026,
      "grad_norm": 0.5644400894009494,
      "learning_rate": 7.43336845644882e-06,
      "loss": 11.9148,
      "step": 32328
    },
    {
      "epoch": 1.7604432555321856,
      "grad_norm": 0.5039294177055477,
      "learning_rate": 7.430032508211826e-06,
      "loss": 11.8055,
      "step": 32329
    },
    {
      "epoch": 1.7604977095287686,
      "grad_norm": 0.5505615147334901,
      "learning_rate": 7.426697279814432e-06,
      "loss": 11.7359,
      "step": 32330
    },
    {
      "epoch": 1.7605521635253516,
      "grad_norm": 0.5532087565581637,
      "learning_rate": 7.4233627712825495e-06,
      "loss": 11.7698,
      "step": 32331
    },
    {
      "epoch": 1.7606066175219346,
      "grad_norm": 0.5392862202807782,
      "learning_rate": 7.420028982642124e-06,
      "loss": 11.7708,
      "step": 32332
    },
    {
      "epoch": 1.7606610715185178,
      "grad_norm": 0.5398926937844774,
      "learning_rate": 7.416695913919114e-06,
      "loss": 11.8584,
      "step": 32333
    },
    {
      "epoch": 1.7607155255151008,
      "grad_norm": 0.5723571536725904,
      "learning_rate": 7.413363565139364e-06,
      "loss": 11.8024,
      "step": 32334
    },
    {
      "epoch": 1.7607699795116838,
      "grad_norm": 0.5295461305273266,
      "learning_rate": 7.410031936328854e-06,
      "loss": 11.6689,
      "step": 32335
    },
    {
      "epoch": 1.7608244335082668,
      "grad_norm": 0.5634649371573618,
      "learning_rate": 7.40670102751343e-06,
      "loss": 11.6344,
      "step": 32336
    },
    {
      "epoch": 1.7608788875048498,
      "grad_norm": 0.5427299651954265,
      "learning_rate": 7.403370838719037e-06,
      "loss": 11.6804,
      "step": 32337
    },
    {
      "epoch": 1.7609333415014328,
      "grad_norm": 0.5157469887731284,
      "learning_rate": 7.400041369971577e-06,
      "loss": 11.682,
      "step": 32338
    },
    {
      "epoch": 1.760987795498016,
      "grad_norm": 0.6181380236442666,
      "learning_rate": 7.3967126212969085e-06,
      "loss": 11.8238,
      "step": 32339
    },
    {
      "epoch": 1.761042249494599,
      "grad_norm": 0.5783597891575023,
      "learning_rate": 7.393384592720942e-06,
      "loss": 11.7492,
      "step": 32340
    },
    {
      "epoch": 1.761096703491182,
      "grad_norm": 0.5322414404189337,
      "learning_rate": 7.390057284269536e-06,
      "loss": 11.8629,
      "step": 32341
    },
    {
      "epoch": 1.761151157487765,
      "grad_norm": 0.5611333027155997,
      "learning_rate": 7.3867306959685915e-06,
      "loss": 11.8159,
      "step": 32342
    },
    {
      "epoch": 1.761205611484348,
      "grad_norm": 0.5706571472987123,
      "learning_rate": 7.383404827843965e-06,
      "loss": 11.838,
      "step": 32343
    },
    {
      "epoch": 1.761260065480931,
      "grad_norm": 0.6170261038711772,
      "learning_rate": 7.380079679921493e-06,
      "loss": 11.876,
      "step": 32344
    },
    {
      "epoch": 1.761314519477514,
      "grad_norm": 0.5252955293287406,
      "learning_rate": 7.376755252227063e-06,
      "loss": 11.843,
      "step": 32345
    },
    {
      "epoch": 1.761368973474097,
      "grad_norm": 0.5390535813073829,
      "learning_rate": 7.373431544786513e-06,
      "loss": 11.7998,
      "step": 32346
    },
    {
      "epoch": 1.76142342747068,
      "grad_norm": 0.537408796287597,
      "learning_rate": 7.3701085576256986e-06,
      "loss": 11.8537,
      "step": 32347
    },
    {
      "epoch": 1.761477881467263,
      "grad_norm": 0.5563103088766296,
      "learning_rate": 7.366786290770445e-06,
      "loss": 11.7829,
      "step": 32348
    },
    {
      "epoch": 1.761532335463846,
      "grad_norm": 0.5364377866093762,
      "learning_rate": 7.363464744246596e-06,
      "loss": 11.8085,
      "step": 32349
    },
    {
      "epoch": 1.761586789460429,
      "grad_norm": 0.5351507697297039,
      "learning_rate": 7.360143918079987e-06,
      "loss": 11.8374,
      "step": 32350
    },
    {
      "epoch": 1.761641243457012,
      "grad_norm": 0.5204197436056024,
      "learning_rate": 7.356823812296432e-06,
      "loss": 11.8803,
      "step": 32351
    },
    {
      "epoch": 1.7616956974535949,
      "grad_norm": 0.5480532340536863,
      "learning_rate": 7.353504426921754e-06,
      "loss": 11.7533,
      "step": 32352
    },
    {
      "epoch": 1.7617501514501779,
      "grad_norm": 0.5295869301956416,
      "learning_rate": 7.350185761981776e-06,
      "loss": 11.8076,
      "step": 32353
    },
    {
      "epoch": 1.7618046054467609,
      "grad_norm": 0.4870461952741857,
      "learning_rate": 7.346867817502267e-06,
      "loss": 11.8195,
      "step": 32354
    },
    {
      "epoch": 1.7618590594433439,
      "grad_norm": 0.5677620771240898,
      "learning_rate": 7.3435505935090635e-06,
      "loss": 11.7581,
      "step": 32355
    },
    {
      "epoch": 1.761913513439927,
      "grad_norm": 0.5574819696135239,
      "learning_rate": 7.340234090027942e-06,
      "loss": 11.6624,
      "step": 32356
    },
    {
      "epoch": 1.76196796743651,
      "grad_norm": 0.590936554342043,
      "learning_rate": 7.336918307084717e-06,
      "loss": 11.7542,
      "step": 32357
    },
    {
      "epoch": 1.762022421433093,
      "grad_norm": 0.5000063304337666,
      "learning_rate": 7.333603244705134e-06,
      "loss": 11.8136,
      "step": 32358
    },
    {
      "epoch": 1.762076875429676,
      "grad_norm": 0.5676488735060264,
      "learning_rate": 7.330288902914983e-06,
      "loss": 11.8305,
      "step": 32359
    },
    {
      "epoch": 1.762131329426259,
      "grad_norm": 0.6944711710574692,
      "learning_rate": 7.326975281740078e-06,
      "loss": 11.6905,
      "step": 32360
    },
    {
      "epoch": 1.762185783422842,
      "grad_norm": 0.5067506588848002,
      "learning_rate": 7.323662381206131e-06,
      "loss": 11.891,
      "step": 32361
    },
    {
      "epoch": 1.7622402374194253,
      "grad_norm": 0.5903404600956154,
      "learning_rate": 7.320350201338944e-06,
      "loss": 11.7509,
      "step": 32362
    },
    {
      "epoch": 1.7622946914160083,
      "grad_norm": 0.6223132012732371,
      "learning_rate": 7.31703874216425e-06,
      "loss": 11.8712,
      "step": 32363
    },
    {
      "epoch": 1.7623491454125912,
      "grad_norm": 0.5699465982485366,
      "learning_rate": 7.3137280037077985e-06,
      "loss": 11.7497,
      "step": 32364
    },
    {
      "epoch": 1.7624035994091742,
      "grad_norm": 0.5382094454685953,
      "learning_rate": 7.310417985995355e-06,
      "loss": 11.7891,
      "step": 32365
    },
    {
      "epoch": 1.7624580534057572,
      "grad_norm": 0.9520201525640292,
      "learning_rate": 7.307108689052633e-06,
      "loss": 11.7321,
      "step": 32366
    },
    {
      "epoch": 1.7625125074023402,
      "grad_norm": 0.6391005015733988,
      "learning_rate": 7.30380011290539e-06,
      "loss": 11.9134,
      "step": 32367
    },
    {
      "epoch": 1.7625669613989232,
      "grad_norm": 0.581569098823044,
      "learning_rate": 7.300492257579327e-06,
      "loss": 11.9759,
      "step": 32368
    },
    {
      "epoch": 1.7626214153955062,
      "grad_norm": 0.5519937761675575,
      "learning_rate": 7.297185123100192e-06,
      "loss": 11.9339,
      "step": 32369
    },
    {
      "epoch": 1.7626758693920892,
      "grad_norm": 0.5469487552409232,
      "learning_rate": 7.293878709493684e-06,
      "loss": 11.7986,
      "step": 32370
    },
    {
      "epoch": 1.7627303233886722,
      "grad_norm": 0.5260441176529977,
      "learning_rate": 7.290573016785518e-06,
      "loss": 11.6944,
      "step": 32371
    },
    {
      "epoch": 1.7627847773852552,
      "grad_norm": 0.5805763485213549,
      "learning_rate": 7.287268045001428e-06,
      "loss": 11.8504,
      "step": 32372
    },
    {
      "epoch": 1.7628392313818382,
      "grad_norm": 0.5655698371088128,
      "learning_rate": 7.283963794167081e-06,
      "loss": 11.7975,
      "step": 32373
    },
    {
      "epoch": 1.7628936853784212,
      "grad_norm": 0.5660394550762327,
      "learning_rate": 7.28066026430817e-06,
      "loss": 11.7062,
      "step": 32374
    },
    {
      "epoch": 1.7629481393750042,
      "grad_norm": 0.4911911745799803,
      "learning_rate": 7.277357455450407e-06,
      "loss": 11.85,
      "step": 32375
    },
    {
      "epoch": 1.7630025933715872,
      "grad_norm": 0.5423764304606848,
      "learning_rate": 7.274055367619437e-06,
      "loss": 11.7692,
      "step": 32376
    },
    {
      "epoch": 1.7630570473681701,
      "grad_norm": 0.5731570304029355,
      "learning_rate": 7.270754000840985e-06,
      "loss": 11.6807,
      "step": 32377
    },
    {
      "epoch": 1.7631115013647531,
      "grad_norm": 0.6037863315480049,
      "learning_rate": 7.267453355140685e-06,
      "loss": 11.5989,
      "step": 32378
    },
    {
      "epoch": 1.7631659553613361,
      "grad_norm": 0.536057241950262,
      "learning_rate": 7.26415343054424e-06,
      "loss": 11.7712,
      "step": 32379
    },
    {
      "epoch": 1.7632204093579193,
      "grad_norm": 0.5426477716803638,
      "learning_rate": 7.260854227077274e-06,
      "loss": 11.7726,
      "step": 32380
    },
    {
      "epoch": 1.7632748633545023,
      "grad_norm": 0.48845889683067467,
      "learning_rate": 7.257555744765454e-06,
      "loss": 11.7433,
      "step": 32381
    },
    {
      "epoch": 1.7633293173510853,
      "grad_norm": 0.9140697807030868,
      "learning_rate": 7.25425798363445e-06,
      "loss": 11.7978,
      "step": 32382
    },
    {
      "epoch": 1.7633837713476683,
      "grad_norm": 0.4784156197793373,
      "learning_rate": 7.250960943709884e-06,
      "loss": 11.7587,
      "step": 32383
    },
    {
      "epoch": 1.7634382253442513,
      "grad_norm": 0.6121897958137692,
      "learning_rate": 7.247664625017403e-06,
      "loss": 11.7678,
      "step": 32384
    },
    {
      "epoch": 1.7634926793408345,
      "grad_norm": 0.5203835716886616,
      "learning_rate": 7.244369027582621e-06,
      "loss": 11.8362,
      "step": 32385
    },
    {
      "epoch": 1.7635471333374175,
      "grad_norm": 0.7471726687272454,
      "learning_rate": 7.241074151431182e-06,
      "loss": 11.7367,
      "step": 32386
    },
    {
      "epoch": 1.7636015873340005,
      "grad_norm": 0.5778906960302164,
      "learning_rate": 7.237779996588723e-06,
      "loss": 11.8412,
      "step": 32387
    },
    {
      "epoch": 1.7636560413305835,
      "grad_norm": 0.6550759964786931,
      "learning_rate": 7.234486563080823e-06,
      "loss": 11.8367,
      "step": 32388
    },
    {
      "epoch": 1.7637104953271665,
      "grad_norm": 0.5475345367993918,
      "learning_rate": 7.231193850933127e-06,
      "loss": 11.8007,
      "step": 32389
    },
    {
      "epoch": 1.7637649493237495,
      "grad_norm": 0.5899456756389151,
      "learning_rate": 7.227901860171215e-06,
      "loss": 11.8182,
      "step": 32390
    },
    {
      "epoch": 1.7638194033203325,
      "grad_norm": 0.5973111699437943,
      "learning_rate": 7.224610590820713e-06,
      "loss": 11.8454,
      "step": 32391
    },
    {
      "epoch": 1.7638738573169155,
      "grad_norm": 0.5544637859212342,
      "learning_rate": 7.221320042907176e-06,
      "loss": 11.7157,
      "step": 32392
    },
    {
      "epoch": 1.7639283113134985,
      "grad_norm": 0.5580414332870065,
      "learning_rate": 7.2180302164562175e-06,
      "loss": 11.7504,
      "step": 32393
    },
    {
      "epoch": 1.7639827653100815,
      "grad_norm": 0.5193774247888508,
      "learning_rate": 7.214741111493439e-06,
      "loss": 11.6611,
      "step": 32394
    },
    {
      "epoch": 1.7640372193066645,
      "grad_norm": 0.5531527796970839,
      "learning_rate": 7.211452728044377e-06,
      "loss": 11.7538,
      "step": 32395
    },
    {
      "epoch": 1.7640916733032475,
      "grad_norm": 0.5379305994824084,
      "learning_rate": 7.20816506613462e-06,
      "loss": 11.7033,
      "step": 32396
    },
    {
      "epoch": 1.7641461272998304,
      "grad_norm": 0.5246291274011378,
      "learning_rate": 7.2048781257897265e-06,
      "loss": 11.8177,
      "step": 32397
    },
    {
      "epoch": 1.7642005812964134,
      "grad_norm": 0.5485662663006204,
      "learning_rate": 7.2015919070352545e-06,
      "loss": 11.7589,
      "step": 32398
    },
    {
      "epoch": 1.7642550352929964,
      "grad_norm": 0.5439113601867139,
      "learning_rate": 7.198306409896794e-06,
      "loss": 11.8224,
      "step": 32399
    },
    {
      "epoch": 1.7643094892895794,
      "grad_norm": 0.5510783818630199,
      "learning_rate": 7.195021634399834e-06,
      "loss": 11.9455,
      "step": 32400
    },
    {
      "epoch": 1.7643639432861624,
      "grad_norm": 0.5078001366956315,
      "learning_rate": 7.191737580569979e-06,
      "loss": 11.7833,
      "step": 32401
    },
    {
      "epoch": 1.7644183972827454,
      "grad_norm": 0.5463870607280911,
      "learning_rate": 7.1884542484327076e-06,
      "loss": 11.6649,
      "step": 32402
    },
    {
      "epoch": 1.7644728512793286,
      "grad_norm": 0.5293179509605374,
      "learning_rate": 7.18517163801361e-06,
      "loss": 11.7558,
      "step": 32403
    },
    {
      "epoch": 1.7645273052759116,
      "grad_norm": 0.5661948285546391,
      "learning_rate": 7.181889749338178e-06,
      "loss": 11.8284,
      "step": 32404
    },
    {
      "epoch": 1.7645817592724946,
      "grad_norm": 0.4774295334887779,
      "learning_rate": 7.178608582431912e-06,
      "loss": 11.7392,
      "step": 32405
    },
    {
      "epoch": 1.7646362132690776,
      "grad_norm": 0.5569778331117314,
      "learning_rate": 7.1753281373203805e-06,
      "loss": 11.7874,
      "step": 32406
    },
    {
      "epoch": 1.7646906672656606,
      "grad_norm": 0.49605450321583283,
      "learning_rate": 7.172048414029042e-06,
      "loss": 11.7344,
      "step": 32407
    },
    {
      "epoch": 1.7647451212622436,
      "grad_norm": 0.5286431509239733,
      "learning_rate": 7.1687694125834205e-06,
      "loss": 11.8066,
      "step": 32408
    },
    {
      "epoch": 1.7647995752588268,
      "grad_norm": 0.6143050511893186,
      "learning_rate": 7.165491133009039e-06,
      "loss": 11.8055,
      "step": 32409
    },
    {
      "epoch": 1.7648540292554098,
      "grad_norm": 0.5104935075402033,
      "learning_rate": 7.1622135753313445e-06,
      "loss": 11.7688,
      "step": 32410
    },
    {
      "epoch": 1.7649084832519928,
      "grad_norm": 0.5085736202065824,
      "learning_rate": 7.158936739575861e-06,
      "loss": 11.7869,
      "step": 32411
    },
    {
      "epoch": 1.7649629372485758,
      "grad_norm": 0.5548308218267389,
      "learning_rate": 7.155660625768046e-06,
      "loss": 11.647,
      "step": 32412
    },
    {
      "epoch": 1.7650173912451588,
      "grad_norm": 0.5182332690613815,
      "learning_rate": 7.152385233933389e-06,
      "loss": 11.8057,
      "step": 32413
    },
    {
      "epoch": 1.7650718452417418,
      "grad_norm": 0.5407855924686837,
      "learning_rate": 7.149110564097372e-06,
      "loss": 11.7418,
      "step": 32414
    },
    {
      "epoch": 1.7651262992383248,
      "grad_norm": 0.6337329798454913,
      "learning_rate": 7.145836616285406e-06,
      "loss": 11.7943,
      "step": 32415
    },
    {
      "epoch": 1.7651807532349078,
      "grad_norm": 0.48449606329339867,
      "learning_rate": 7.142563390523016e-06,
      "loss": 11.5754,
      "step": 32416
    },
    {
      "epoch": 1.7652352072314907,
      "grad_norm": 0.5302133558003101,
      "learning_rate": 7.139290886835604e-06,
      "loss": 11.8318,
      "step": 32417
    },
    {
      "epoch": 1.7652896612280737,
      "grad_norm": 0.5462321802071552,
      "learning_rate": 7.136019105248659e-06,
      "loss": 11.7628,
      "step": 32418
    },
    {
      "epoch": 1.7653441152246567,
      "grad_norm": 0.593250827650166,
      "learning_rate": 7.132748045787585e-06,
      "loss": 11.936,
      "step": 32419
    },
    {
      "epoch": 1.7653985692212397,
      "grad_norm": 0.534377515132765,
      "learning_rate": 7.1294777084778385e-06,
      "loss": 11.8307,
      "step": 32420
    },
    {
      "epoch": 1.7654530232178227,
      "grad_norm": 0.5957141896429868,
      "learning_rate": 7.126208093344855e-06,
      "loss": 11.8671,
      "step": 32421
    },
    {
      "epoch": 1.7655074772144057,
      "grad_norm": 0.5253310196150951,
      "learning_rate": 7.122939200414047e-06,
      "loss": 11.5939,
      "step": 32422
    },
    {
      "epoch": 1.7655619312109887,
      "grad_norm": 0.5727741432569967,
      "learning_rate": 7.119671029710851e-06,
      "loss": 11.8215,
      "step": 32423
    },
    {
      "epoch": 1.7656163852075717,
      "grad_norm": 0.5645552107529307,
      "learning_rate": 7.116403581260666e-06,
      "loss": 11.5887,
      "step": 32424
    },
    {
      "epoch": 1.7656708392041547,
      "grad_norm": 0.5728926466432189,
      "learning_rate": 7.113136855088887e-06,
      "loss": 11.6244,
      "step": 32425
    },
    {
      "epoch": 1.765725293200738,
      "grad_norm": 0.5505402048947162,
      "learning_rate": 7.109870851220946e-06,
      "loss": 11.8847,
      "step": 32426
    },
    {
      "epoch": 1.765779747197321,
      "grad_norm": 0.48449133413234013,
      "learning_rate": 7.106605569682223e-06,
      "loss": 11.8173,
      "step": 32427
    },
    {
      "epoch": 1.7658342011939039,
      "grad_norm": 0.6641475685607001,
      "learning_rate": 7.103341010498121e-06,
      "loss": 11.8476,
      "step": 32428
    },
    {
      "epoch": 1.7658886551904869,
      "grad_norm": 0.5200225627804707,
      "learning_rate": 7.100077173693998e-06,
      "loss": 11.7741,
      "step": 32429
    },
    {
      "epoch": 1.7659431091870699,
      "grad_norm": 0.6478267440136369,
      "learning_rate": 7.0968140592952645e-06,
      "loss": 11.7211,
      "step": 32430
    },
    {
      "epoch": 1.7659975631836529,
      "grad_norm": 0.526273367586206,
      "learning_rate": 7.093551667327292e-06,
      "loss": 11.7279,
      "step": 32431
    },
    {
      "epoch": 1.766052017180236,
      "grad_norm": 0.5293466186121332,
      "learning_rate": 7.090289997815436e-06,
      "loss": 11.6945,
      "step": 32432
    },
    {
      "epoch": 1.766106471176819,
      "grad_norm": 0.5234387758570058,
      "learning_rate": 7.087029050785077e-06,
      "loss": 11.785,
      "step": 32433
    },
    {
      "epoch": 1.766160925173402,
      "grad_norm": 0.5608956079550506,
      "learning_rate": 7.083768826261561e-06,
      "loss": 11.7521,
      "step": 32434
    },
    {
      "epoch": 1.766215379169985,
      "grad_norm": 0.6132855594715134,
      "learning_rate": 7.080509324270224e-06,
      "loss": 11.8305,
      "step": 32435
    },
    {
      "epoch": 1.766269833166568,
      "grad_norm": 0.579739800056128,
      "learning_rate": 7.077250544836444e-06,
      "loss": 11.8372,
      "step": 32436
    },
    {
      "epoch": 1.766324287163151,
      "grad_norm": 0.4848196679170275,
      "learning_rate": 7.073992487985537e-06,
      "loss": 11.6512,
      "step": 32437
    },
    {
      "epoch": 1.766378741159734,
      "grad_norm": 0.5769411461503011,
      "learning_rate": 7.070735153742858e-06,
      "loss": 11.7398,
      "step": 32438
    },
    {
      "epoch": 1.766433195156317,
      "grad_norm": 0.5400025851720152,
      "learning_rate": 7.06747854213371e-06,
      "loss": 11.7714,
      "step": 32439
    },
    {
      "epoch": 1.7664876491529,
      "grad_norm": 0.5430651435837082,
      "learning_rate": 7.0642226531834495e-06,
      "loss": 11.7308,
      "step": 32440
    },
    {
      "epoch": 1.766542103149483,
      "grad_norm": 0.5546201747900835,
      "learning_rate": 7.0609674869173695e-06,
      "loss": 11.8682,
      "step": 32441
    },
    {
      "epoch": 1.766596557146066,
      "grad_norm": 0.4697497270887615,
      "learning_rate": 7.057713043360781e-06,
      "loss": 11.7482,
      "step": 32442
    },
    {
      "epoch": 1.766651011142649,
      "grad_norm": 0.5527719939885997,
      "learning_rate": 7.054459322539031e-06,
      "loss": 11.689,
      "step": 32443
    },
    {
      "epoch": 1.766705465139232,
      "grad_norm": 0.5539698824192684,
      "learning_rate": 7.051206324477389e-06,
      "loss": 11.8499,
      "step": 32444
    },
    {
      "epoch": 1.766759919135815,
      "grad_norm": 0.5806791927372109,
      "learning_rate": 7.047954049201144e-06,
      "loss": 11.8532,
      "step": 32445
    },
    {
      "epoch": 1.766814373132398,
      "grad_norm": 0.5164966670585809,
      "learning_rate": 7.044702496735589e-06,
      "loss": 11.7236,
      "step": 32446
    },
    {
      "epoch": 1.766868827128981,
      "grad_norm": 0.5388727806756906,
      "learning_rate": 7.041451667106014e-06,
      "loss": 11.8655,
      "step": 32447
    },
    {
      "epoch": 1.766923281125564,
      "grad_norm": 0.5258345518559903,
      "learning_rate": 7.038201560337721e-06,
      "loss": 11.8075,
      "step": 32448
    },
    {
      "epoch": 1.7669777351221472,
      "grad_norm": 0.5586393142600838,
      "learning_rate": 7.034952176455945e-06,
      "loss": 11.7938,
      "step": 32449
    },
    {
      "epoch": 1.7670321891187302,
      "grad_norm": 0.6787701241470055,
      "learning_rate": 7.031703515485988e-06,
      "loss": 11.8707,
      "step": 32450
    },
    {
      "epoch": 1.7670866431153132,
      "grad_norm": 0.5443562415419549,
      "learning_rate": 7.028455577453074e-06,
      "loss": 11.8743,
      "step": 32451
    },
    {
      "epoch": 1.7671410971118962,
      "grad_norm": 0.548410018421957,
      "learning_rate": 7.025208362382496e-06,
      "loss": 11.7629,
      "step": 32452
    },
    {
      "epoch": 1.7671955511084791,
      "grad_norm": 0.5662985835500658,
      "learning_rate": 7.021961870299476e-06,
      "loss": 11.868,
      "step": 32453
    },
    {
      "epoch": 1.7672500051050621,
      "grad_norm": 0.5611080212631334,
      "learning_rate": 7.0187161012292945e-06,
      "loss": 11.789,
      "step": 32454
    },
    {
      "epoch": 1.7673044591016454,
      "grad_norm": 0.5441373707393591,
      "learning_rate": 7.015471055197164e-06,
      "loss": 11.756,
      "step": 32455
    },
    {
      "epoch": 1.7673589130982283,
      "grad_norm": 0.6761327254056887,
      "learning_rate": 7.0122267322282995e-06,
      "loss": 11.894,
      "step": 32456
    },
    {
      "epoch": 1.7674133670948113,
      "grad_norm": 0.563502754538389,
      "learning_rate": 7.008983132347968e-06,
      "loss": 11.7751,
      "step": 32457
    },
    {
      "epoch": 1.7674678210913943,
      "grad_norm": 0.5391698601908398,
      "learning_rate": 7.005740255581395e-06,
      "loss": 11.8214,
      "step": 32458
    },
    {
      "epoch": 1.7675222750879773,
      "grad_norm": 0.5756762889962632,
      "learning_rate": 7.002498101953758e-06,
      "loss": 11.7541,
      "step": 32459
    },
    {
      "epoch": 1.7675767290845603,
      "grad_norm": 0.5510747624109077,
      "learning_rate": 6.999256671490306e-06,
      "loss": 11.9001,
      "step": 32460
    },
    {
      "epoch": 1.7676311830811433,
      "grad_norm": 0.5600344165461107,
      "learning_rate": 6.996015964216218e-06,
      "loss": 11.6302,
      "step": 32461
    },
    {
      "epoch": 1.7676856370777263,
      "grad_norm": 0.5105209825942861,
      "learning_rate": 6.992775980156718e-06,
      "loss": 11.6263,
      "step": 32462
    },
    {
      "epoch": 1.7677400910743093,
      "grad_norm": 0.5287327792460506,
      "learning_rate": 6.989536719336976e-06,
      "loss": 11.7988,
      "step": 32463
    },
    {
      "epoch": 1.7677945450708923,
      "grad_norm": 0.5235574865193675,
      "learning_rate": 6.986298181782213e-06,
      "loss": 11.7791,
      "step": 32464
    },
    {
      "epoch": 1.7678489990674753,
      "grad_norm": 0.5248584359930557,
      "learning_rate": 6.983060367517591e-06,
      "loss": 11.5514,
      "step": 32465
    },
    {
      "epoch": 1.7679034530640583,
      "grad_norm": 0.53106990569008,
      "learning_rate": 6.979823276568276e-06,
      "loss": 11.73,
      "step": 32466
    },
    {
      "epoch": 1.7679579070606413,
      "grad_norm": 0.5624229679426564,
      "learning_rate": 6.97658690895947e-06,
      "loss": 11.788,
      "step": 32467
    },
    {
      "epoch": 1.7680123610572243,
      "grad_norm": 0.5983849167658366,
      "learning_rate": 6.97335126471631e-06,
      "loss": 11.8528,
      "step": 32468
    },
    {
      "epoch": 1.7680668150538073,
      "grad_norm": 0.49099023311629675,
      "learning_rate": 6.9701163438639635e-06,
      "loss": 11.8061,
      "step": 32469
    },
    {
      "epoch": 1.7681212690503902,
      "grad_norm": 0.5159192516763268,
      "learning_rate": 6.9668821464276224e-06,
      "loss": 11.8415,
      "step": 32470
    },
    {
      "epoch": 1.7681757230469732,
      "grad_norm": 0.5505523919006803,
      "learning_rate": 6.9636486724323765e-06,
      "loss": 11.741,
      "step": 32471
    },
    {
      "epoch": 1.7682301770435562,
      "grad_norm": 0.5282648311414133,
      "learning_rate": 6.960415921903429e-06,
      "loss": 11.7232,
      "step": 32472
    },
    {
      "epoch": 1.7682846310401394,
      "grad_norm": 0.5803095678248418,
      "learning_rate": 6.95718389486587e-06,
      "loss": 11.7988,
      "step": 32473
    },
    {
      "epoch": 1.7683390850367224,
      "grad_norm": 0.5393963369431248,
      "learning_rate": 6.95395259134487e-06,
      "loss": 11.7687,
      "step": 32474
    },
    {
      "epoch": 1.7683935390333054,
      "grad_norm": 0.5067067948482463,
      "learning_rate": 6.95072201136554e-06,
      "loss": 11.7722,
      "step": 32475
    },
    {
      "epoch": 1.7684479930298884,
      "grad_norm": 0.5390073898022799,
      "learning_rate": 6.947492154952972e-06,
      "loss": 11.7714,
      "step": 32476
    },
    {
      "epoch": 1.7685024470264714,
      "grad_norm": 0.5290798418720969,
      "learning_rate": 6.944263022132336e-06,
      "loss": 11.8138,
      "step": 32477
    },
    {
      "epoch": 1.7685569010230544,
      "grad_norm": 0.47873354783519273,
      "learning_rate": 6.941034612928699e-06,
      "loss": 11.7539,
      "step": 32478
    },
    {
      "epoch": 1.7686113550196376,
      "grad_norm": 0.551475868516879,
      "learning_rate": 6.9378069273671855e-06,
      "loss": 11.7834,
      "step": 32479
    },
    {
      "epoch": 1.7686658090162206,
      "grad_norm": 0.5259224440001655,
      "learning_rate": 6.934579965472898e-06,
      "loss": 11.6581,
      "step": 32480
    },
    {
      "epoch": 1.7687202630128036,
      "grad_norm": 0.5675922206340825,
      "learning_rate": 6.931353727270917e-06,
      "loss": 11.7332,
      "step": 32481
    },
    {
      "epoch": 1.7687747170093866,
      "grad_norm": 0.4966813392031055,
      "learning_rate": 6.928128212786356e-06,
      "loss": 11.7096,
      "step": 32482
    },
    {
      "epoch": 1.7688291710059696,
      "grad_norm": 0.5851621807802333,
      "learning_rate": 6.92490342204426e-06,
      "loss": 11.7999,
      "step": 32483
    },
    {
      "epoch": 1.7688836250025526,
      "grad_norm": 0.5693229515501992,
      "learning_rate": 6.921679355069732e-06,
      "loss": 11.8676,
      "step": 32484
    },
    {
      "epoch": 1.7689380789991356,
      "grad_norm": 0.5154227529447211,
      "learning_rate": 6.918456011887842e-06,
      "loss": 11.7555,
      "step": 32485
    },
    {
      "epoch": 1.7689925329957186,
      "grad_norm": 0.5550616391724824,
      "learning_rate": 6.915233392523623e-06,
      "loss": 11.8038,
      "step": 32486
    },
    {
      "epoch": 1.7690469869923016,
      "grad_norm": 0.5692104693139884,
      "learning_rate": 6.91201149700218e-06,
      "loss": 11.762,
      "step": 32487
    },
    {
      "epoch": 1.7691014409888846,
      "grad_norm": 0.5375555291794875,
      "learning_rate": 6.908790325348536e-06,
      "loss": 11.7449,
      "step": 32488
    },
    {
      "epoch": 1.7691558949854675,
      "grad_norm": 0.5655107579397572,
      "learning_rate": 6.905569877587759e-06,
      "loss": 11.812,
      "step": 32489
    },
    {
      "epoch": 1.7692103489820505,
      "grad_norm": 0.5203234674230789,
      "learning_rate": 6.902350153744874e-06,
      "loss": 11.7405,
      "step": 32490
    },
    {
      "epoch": 1.7692648029786335,
      "grad_norm": 0.5442782451858077,
      "learning_rate": 6.8991311538449175e-06,
      "loss": 11.8278,
      "step": 32491
    },
    {
      "epoch": 1.7693192569752165,
      "grad_norm": 0.5938194500026797,
      "learning_rate": 6.895912877912958e-06,
      "loss": 11.8093,
      "step": 32492
    },
    {
      "epoch": 1.7693737109717995,
      "grad_norm": 0.49089936157676045,
      "learning_rate": 6.892695325973963e-06,
      "loss": 11.7251,
      "step": 32493
    },
    {
      "epoch": 1.7694281649683825,
      "grad_norm": 0.5462910053179499,
      "learning_rate": 6.8894784980530255e-06,
      "loss": 11.7654,
      "step": 32494
    },
    {
      "epoch": 1.7694826189649655,
      "grad_norm": 0.5416389054162575,
      "learning_rate": 6.88626239417508e-06,
      "loss": 11.8812,
      "step": 32495
    },
    {
      "epoch": 1.7695370729615487,
      "grad_norm": 0.5772196206579886,
      "learning_rate": 6.883047014365173e-06,
      "loss": 11.8,
      "step": 32496
    },
    {
      "epoch": 1.7695915269581317,
      "grad_norm": 0.5731945478590326,
      "learning_rate": 6.8798323586483305e-06,
      "loss": 11.8409,
      "step": 32497
    },
    {
      "epoch": 1.7696459809547147,
      "grad_norm": 0.5237249111482807,
      "learning_rate": 6.876618427049509e-06,
      "loss": 11.6249,
      "step": 32498
    },
    {
      "epoch": 1.7697004349512977,
      "grad_norm": 0.6041684509377229,
      "learning_rate": 6.873405219593732e-06,
      "loss": 11.8516,
      "step": 32499
    },
    {
      "epoch": 1.7697548889478807,
      "grad_norm": 0.5715753407255886,
      "learning_rate": 6.870192736305958e-06,
      "loss": 11.8486,
      "step": 32500
    },
    {
      "epoch": 1.7698093429444637,
      "grad_norm": 0.6189144514321696,
      "learning_rate": 6.866980977211201e-06,
      "loss": 11.8001,
      "step": 32501
    },
    {
      "epoch": 1.769863796941047,
      "grad_norm": 0.5777239824397306,
      "learning_rate": 6.8637699423344085e-06,
      "loss": 11.8568,
      "step": 32502
    },
    {
      "epoch": 1.7699182509376299,
      "grad_norm": 0.8850594492937024,
      "learning_rate": 6.860559631700558e-06,
      "loss": 11.7279,
      "step": 32503
    },
    {
      "epoch": 1.7699727049342129,
      "grad_norm": 0.5851050335484635,
      "learning_rate": 6.857350045334643e-06,
      "loss": 11.8075,
      "step": 32504
    },
    {
      "epoch": 1.7700271589307959,
      "grad_norm": 0.6013720013585722,
      "learning_rate": 6.854141183261564e-06,
      "loss": 11.9136,
      "step": 32505
    },
    {
      "epoch": 1.7700816129273789,
      "grad_norm": 0.509443011363841,
      "learning_rate": 6.850933045506302e-06,
      "loss": 11.6482,
      "step": 32506
    },
    {
      "epoch": 1.7701360669239619,
      "grad_norm": 0.5679414901449624,
      "learning_rate": 6.847725632093815e-06,
      "loss": 11.5742,
      "step": 32507
    },
    {
      "epoch": 1.7701905209205449,
      "grad_norm": 0.5392368385181602,
      "learning_rate": 6.844518943049027e-06,
      "loss": 11.8295,
      "step": 32508
    },
    {
      "epoch": 1.7702449749171278,
      "grad_norm": 0.5751937584515109,
      "learning_rate": 6.841312978396896e-06,
      "loss": 11.8178,
      "step": 32509
    },
    {
      "epoch": 1.7702994289137108,
      "grad_norm": 0.5864386990441026,
      "learning_rate": 6.838107738162325e-06,
      "loss": 11.8829,
      "step": 32510
    },
    {
      "epoch": 1.7703538829102938,
      "grad_norm": 0.5505025726782754,
      "learning_rate": 6.83490322237027e-06,
      "loss": 11.8084,
      "step": 32511
    },
    {
      "epoch": 1.7704083369068768,
      "grad_norm": 0.5154488770079803,
      "learning_rate": 6.831699431045602e-06,
      "loss": 11.8203,
      "step": 32512
    },
    {
      "epoch": 1.7704627909034598,
      "grad_norm": 0.5465051550207456,
      "learning_rate": 6.828496364213277e-06,
      "loss": 11.7996,
      "step": 32513
    },
    {
      "epoch": 1.7705172449000428,
      "grad_norm": 0.5741377442620422,
      "learning_rate": 6.825294021898221e-06,
      "loss": 11.6627,
      "step": 32514
    },
    {
      "epoch": 1.7705716988966258,
      "grad_norm": 0.5494894114487702,
      "learning_rate": 6.822092404125258e-06,
      "loss": 11.8403,
      "step": 32515
    },
    {
      "epoch": 1.7706261528932088,
      "grad_norm": 0.5090681600776872,
      "learning_rate": 6.818891510919356e-06,
      "loss": 11.7853,
      "step": 32516
    },
    {
      "epoch": 1.7706806068897918,
      "grad_norm": 0.5263071241968802,
      "learning_rate": 6.815691342305364e-06,
      "loss": 11.7875,
      "step": 32517
    },
    {
      "epoch": 1.7707350608863748,
      "grad_norm": 0.549212220543552,
      "learning_rate": 6.812491898308193e-06,
      "loss": 11.7381,
      "step": 32518
    },
    {
      "epoch": 1.770789514882958,
      "grad_norm": 0.5314527347005255,
      "learning_rate": 6.809293178952714e-06,
      "loss": 11.6578,
      "step": 32519
    },
    {
      "epoch": 1.770843968879541,
      "grad_norm": 0.5252082284132608,
      "learning_rate": 6.806095184263783e-06,
      "loss": 11.8465,
      "step": 32520
    },
    {
      "epoch": 1.770898422876124,
      "grad_norm": 0.49797130645623167,
      "learning_rate": 6.802897914266315e-06,
      "loss": 11.8213,
      "step": 32521
    },
    {
      "epoch": 1.770952876872707,
      "grad_norm": 0.5635534863376405,
      "learning_rate": 6.799701368985112e-06,
      "loss": 11.9435,
      "step": 32522
    },
    {
      "epoch": 1.77100733086929,
      "grad_norm": 0.5629515308452587,
      "learning_rate": 6.7965055484450865e-06,
      "loss": 11.8811,
      "step": 32523
    },
    {
      "epoch": 1.771061784865873,
      "grad_norm": 0.5315500683196582,
      "learning_rate": 6.7933104526710534e-06,
      "loss": 11.709,
      "step": 32524
    },
    {
      "epoch": 1.7711162388624562,
      "grad_norm": 0.5764388303605275,
      "learning_rate": 6.790116081687858e-06,
      "loss": 11.8658,
      "step": 32525
    },
    {
      "epoch": 1.7711706928590392,
      "grad_norm": 0.6623134247580197,
      "learning_rate": 6.786922435520371e-06,
      "loss": 11.9887,
      "step": 32526
    },
    {
      "epoch": 1.7712251468556222,
      "grad_norm": 0.5157210021703791,
      "learning_rate": 6.783729514193382e-06,
      "loss": 11.7958,
      "step": 32527
    },
    {
      "epoch": 1.7712796008522051,
      "grad_norm": 0.5376612123642904,
      "learning_rate": 6.78053731773175e-06,
      "loss": 11.816,
      "step": 32528
    },
    {
      "epoch": 1.7713340548487881,
      "grad_norm": 0.5261152942844866,
      "learning_rate": 6.77734584616031e-06,
      "loss": 11.7756,
      "step": 32529
    },
    {
      "epoch": 1.7713885088453711,
      "grad_norm": 0.5143438841387615,
      "learning_rate": 6.7741550995038535e-06,
      "loss": 11.7241,
      "step": 32530
    },
    {
      "epoch": 1.7714429628419541,
      "grad_norm": 0.6092866657952518,
      "learning_rate": 6.770965077787206e-06,
      "loss": 11.7917,
      "step": 32531
    },
    {
      "epoch": 1.7714974168385371,
      "grad_norm": 0.5320070727590585,
      "learning_rate": 6.767775781035157e-06,
      "loss": 11.7141,
      "step": 32532
    },
    {
      "epoch": 1.7715518708351201,
      "grad_norm": 0.5772470246873229,
      "learning_rate": 6.764587209272533e-06,
      "loss": 11.7088,
      "step": 32533
    },
    {
      "epoch": 1.771606324831703,
      "grad_norm": 0.5277213074435889,
      "learning_rate": 6.76139936252409e-06,
      "loss": 11.6816,
      "step": 32534
    },
    {
      "epoch": 1.771660778828286,
      "grad_norm": 0.5219917950417122,
      "learning_rate": 6.7582122408146656e-06,
      "loss": 11.7806,
      "step": 32535
    },
    {
      "epoch": 1.771715232824869,
      "grad_norm": 0.5715357613118955,
      "learning_rate": 6.755025844169027e-06,
      "loss": 11.6965,
      "step": 32536
    },
    {
      "epoch": 1.771769686821452,
      "grad_norm": 0.5850917935202838,
      "learning_rate": 6.751840172611923e-06,
      "loss": 11.5886,
      "step": 32537
    },
    {
      "epoch": 1.771824140818035,
      "grad_norm": 0.6101323911637545,
      "learning_rate": 6.748655226168155e-06,
      "loss": 11.836,
      "step": 32538
    },
    {
      "epoch": 1.771878594814618,
      "grad_norm": 0.5412089774253621,
      "learning_rate": 6.745471004862481e-06,
      "loss": 11.7079,
      "step": 32539
    },
    {
      "epoch": 1.771933048811201,
      "grad_norm": 0.5299501793833984,
      "learning_rate": 6.742287508719647e-06,
      "loss": 11.7622,
      "step": 32540
    },
    {
      "epoch": 1.771987502807784,
      "grad_norm": 0.5232663481832034,
      "learning_rate": 6.739104737764457e-06,
      "loss": 11.6454,
      "step": 32541
    },
    {
      "epoch": 1.772041956804367,
      "grad_norm": 0.5395042908687733,
      "learning_rate": 6.735922692021601e-06,
      "loss": 11.8339,
      "step": 32542
    },
    {
      "epoch": 1.7720964108009503,
      "grad_norm": 0.5607707941876608,
      "learning_rate": 6.732741371515871e-06,
      "loss": 11.7649,
      "step": 32543
    },
    {
      "epoch": 1.7721508647975333,
      "grad_norm": 0.50940726724126,
      "learning_rate": 6.72956077627197e-06,
      "loss": 11.5963,
      "step": 32544
    },
    {
      "epoch": 1.7722053187941162,
      "grad_norm": 0.5687962796539477,
      "learning_rate": 6.7263809063146554e-06,
      "loss": 11.8172,
      "step": 32545
    },
    {
      "epoch": 1.7722597727906992,
      "grad_norm": 0.5552807747158219,
      "learning_rate": 6.723201761668651e-06,
      "loss": 11.7434,
      "step": 32546
    },
    {
      "epoch": 1.7723142267872822,
      "grad_norm": 0.6312128068106516,
      "learning_rate": 6.72002334235865e-06,
      "loss": 11.7108,
      "step": 32547
    },
    {
      "epoch": 1.7723686807838654,
      "grad_norm": 0.5364946785921043,
      "learning_rate": 6.71684564840942e-06,
      "loss": 11.7466,
      "step": 32548
    },
    {
      "epoch": 1.7724231347804484,
      "grad_norm": 0.4954720453777132,
      "learning_rate": 6.7136686798456084e-06,
      "loss": 11.7325,
      "step": 32549
    },
    {
      "epoch": 1.7724775887770314,
      "grad_norm": 0.5507651694484306,
      "learning_rate": 6.710492436691984e-06,
      "loss": 11.8744,
      "step": 32550
    },
    {
      "epoch": 1.7725320427736144,
      "grad_norm": 0.5603768587860525,
      "learning_rate": 6.707316918973194e-06,
      "loss": 11.8327,
      "step": 32551
    },
    {
      "epoch": 1.7725864967701974,
      "grad_norm": 0.5773301010353931,
      "learning_rate": 6.704142126713953e-06,
      "loss": 11.7658,
      "step": 32552
    },
    {
      "epoch": 1.7726409507667804,
      "grad_norm": 0.5225682569384548,
      "learning_rate": 6.700968059938962e-06,
      "loss": 11.8519,
      "step": 32553
    },
    {
      "epoch": 1.7726954047633634,
      "grad_norm": 0.5220744324181364,
      "learning_rate": 6.697794718672878e-06,
      "loss": 11.7751,
      "step": 32554
    },
    {
      "epoch": 1.7727498587599464,
      "grad_norm": 0.5522113729084689,
      "learning_rate": 6.694622102940396e-06,
      "loss": 11.8958,
      "step": 32555
    },
    {
      "epoch": 1.7728043127565294,
      "grad_norm": 0.5953742583947584,
      "learning_rate": 6.691450212766193e-06,
      "loss": 11.8135,
      "step": 32556
    },
    {
      "epoch": 1.7728587667531124,
      "grad_norm": 0.582818102037976,
      "learning_rate": 6.688279048174895e-06,
      "loss": 11.8136,
      "step": 32557
    },
    {
      "epoch": 1.7729132207496954,
      "grad_norm": 0.5999572791971642,
      "learning_rate": 6.685108609191204e-06,
      "loss": 11.8104,
      "step": 32558
    },
    {
      "epoch": 1.7729676747462784,
      "grad_norm": 0.510340518078583,
      "learning_rate": 6.681938895839746e-06,
      "loss": 11.7312,
      "step": 32559
    },
    {
      "epoch": 1.7730221287428614,
      "grad_norm": 0.7048680747222791,
      "learning_rate": 6.678769908145188e-06,
      "loss": 11.8127,
      "step": 32560
    },
    {
      "epoch": 1.7730765827394444,
      "grad_norm": 0.5453801199230138,
      "learning_rate": 6.675601646132157e-06,
      "loss": 11.8925,
      "step": 32561
    },
    {
      "epoch": 1.7731310367360273,
      "grad_norm": 0.5304706086726125,
      "learning_rate": 6.672434109825299e-06,
      "loss": 11.8602,
      "step": 32562
    },
    {
      "epoch": 1.7731854907326103,
      "grad_norm": 0.6165774646844658,
      "learning_rate": 6.66926729924926e-06,
      "loss": 11.7511,
      "step": 32563
    },
    {
      "epoch": 1.7732399447291933,
      "grad_norm": 0.5789662439924932,
      "learning_rate": 6.6661012144286325e-06,
      "loss": 11.8659,
      "step": 32564
    },
    {
      "epoch": 1.7732943987257763,
      "grad_norm": 0.590542330161939,
      "learning_rate": 6.662935855388086e-06,
      "loss": 11.7068,
      "step": 32565
    },
    {
      "epoch": 1.7733488527223595,
      "grad_norm": 0.5549403780924586,
      "learning_rate": 6.659771222152167e-06,
      "loss": 11.6931,
      "step": 32566
    },
    {
      "epoch": 1.7734033067189425,
      "grad_norm": 0.6011244737428332,
      "learning_rate": 6.656607314745522e-06,
      "loss": 11.8375,
      "step": 32567
    },
    {
      "epoch": 1.7734577607155255,
      "grad_norm": 0.5470194606503198,
      "learning_rate": 6.653444133192777e-06,
      "loss": 11.8163,
      "step": 32568
    },
    {
      "epoch": 1.7735122147121085,
      "grad_norm": 0.5263834449077107,
      "learning_rate": 6.650281677518477e-06,
      "loss": 11.7583,
      "step": 32569
    },
    {
      "epoch": 1.7735666687086915,
      "grad_norm": 0.6066447590285262,
      "learning_rate": 6.64711994774726e-06,
      "loss": 11.8415,
      "step": 32570
    },
    {
      "epoch": 1.7736211227052745,
      "grad_norm": 0.564971141513668,
      "learning_rate": 6.6439589439036834e-06,
      "loss": 11.7612,
      "step": 32571
    },
    {
      "epoch": 1.7736755767018577,
      "grad_norm": 0.5426395113191236,
      "learning_rate": 6.640798666012349e-06,
      "loss": 11.8211,
      "step": 32572
    },
    {
      "epoch": 1.7737300306984407,
      "grad_norm": 0.5355788596137272,
      "learning_rate": 6.637639114097804e-06,
      "loss": 11.7272,
      "step": 32573
    },
    {
      "epoch": 1.7737844846950237,
      "grad_norm": 0.5417248876945642,
      "learning_rate": 6.6344802881846416e-06,
      "loss": 11.7728,
      "step": 32574
    },
    {
      "epoch": 1.7738389386916067,
      "grad_norm": 0.5176878213694106,
      "learning_rate": 6.63132218829744e-06,
      "loss": 11.5263,
      "step": 32575
    },
    {
      "epoch": 1.7738933926881897,
      "grad_norm": 0.538084768549536,
      "learning_rate": 6.628164814460702e-06,
      "loss": 11.8603,
      "step": 32576
    },
    {
      "epoch": 1.7739478466847727,
      "grad_norm": 0.5634966133398733,
      "learning_rate": 6.62500816669902e-06,
      "loss": 11.8694,
      "step": 32577
    },
    {
      "epoch": 1.7740023006813557,
      "grad_norm": 0.5087286585919892,
      "learning_rate": 6.621852245036952e-06,
      "loss": 11.6815,
      "step": 32578
    },
    {
      "epoch": 1.7740567546779387,
      "grad_norm": 0.5553469008752997,
      "learning_rate": 6.618697049499001e-06,
      "loss": 11.7431,
      "step": 32579
    },
    {
      "epoch": 1.7741112086745217,
      "grad_norm": 0.5456400676770684,
      "learning_rate": 6.615542580109735e-06,
      "loss": 11.762,
      "step": 32580
    },
    {
      "epoch": 1.7741656626711046,
      "grad_norm": 0.536257816136381,
      "learning_rate": 6.612388836893657e-06,
      "loss": 11.523,
      "step": 32581
    },
    {
      "epoch": 1.7742201166676876,
      "grad_norm": 0.5835429825597319,
      "learning_rate": 6.609235819875314e-06,
      "loss": 11.7725,
      "step": 32582
    },
    {
      "epoch": 1.7742745706642706,
      "grad_norm": 0.508417587120594,
      "learning_rate": 6.60608352907921e-06,
      "loss": 11.8408,
      "step": 32583
    },
    {
      "epoch": 1.7743290246608536,
      "grad_norm": 0.5429689687452576,
      "learning_rate": 6.6029319645298575e-06,
      "loss": 11.8121,
      "step": 32584
    },
    {
      "epoch": 1.7743834786574366,
      "grad_norm": 0.5955028609000584,
      "learning_rate": 6.599781126251792e-06,
      "loss": 11.7714,
      "step": 32585
    },
    {
      "epoch": 1.7744379326540196,
      "grad_norm": 0.6138863765080108,
      "learning_rate": 6.596631014269472e-06,
      "loss": 11.8127,
      "step": 32586
    },
    {
      "epoch": 1.7744923866506026,
      "grad_norm": 0.5274485178907716,
      "learning_rate": 6.593481628607423e-06,
      "loss": 11.7936,
      "step": 32587
    },
    {
      "epoch": 1.7745468406471856,
      "grad_norm": 0.6001581529946779,
      "learning_rate": 6.590332969290114e-06,
      "loss": 11.8609,
      "step": 32588
    },
    {
      "epoch": 1.7746012946437688,
      "grad_norm": 0.5261086031578917,
      "learning_rate": 6.587185036342036e-06,
      "loss": 11.8238,
      "step": 32589
    },
    {
      "epoch": 1.7746557486403518,
      "grad_norm": 0.5738828339982922,
      "learning_rate": 6.584037829787692e-06,
      "loss": 11.8126,
      "step": 32590
    },
    {
      "epoch": 1.7747102026369348,
      "grad_norm": 0.5554205995944045,
      "learning_rate": 6.580891349651519e-06,
      "loss": 11.8371,
      "step": 32591
    },
    {
      "epoch": 1.7747646566335178,
      "grad_norm": 0.5183023933811208,
      "learning_rate": 6.577745595958018e-06,
      "loss": 11.7964,
      "step": 32592
    },
    {
      "epoch": 1.7748191106301008,
      "grad_norm": 0.5084647466055651,
      "learning_rate": 6.574600568731615e-06,
      "loss": 11.7346,
      "step": 32593
    },
    {
      "epoch": 1.7748735646266838,
      "grad_norm": 0.5481947987741224,
      "learning_rate": 6.571456267996801e-06,
      "loss": 11.7844,
      "step": 32594
    },
    {
      "epoch": 1.774928018623267,
      "grad_norm": 0.5891789659361069,
      "learning_rate": 6.568312693778011e-06,
      "loss": 11.7306,
      "step": 32595
    },
    {
      "epoch": 1.77498247261985,
      "grad_norm": 0.5276265723583782,
      "learning_rate": 6.5651698460996834e-06,
      "loss": 11.7326,
      "step": 32596
    },
    {
      "epoch": 1.775036926616433,
      "grad_norm": 0.5704125501566568,
      "learning_rate": 6.562027724986264e-06,
      "loss": 11.6956,
      "step": 32597
    },
    {
      "epoch": 1.775091380613016,
      "grad_norm": 0.562789157110727,
      "learning_rate": 6.558886330462189e-06,
      "loss": 11.789,
      "step": 32598
    },
    {
      "epoch": 1.775145834609599,
      "grad_norm": 0.5928072781857274,
      "learning_rate": 6.555745662551882e-06,
      "loss": 11.6489,
      "step": 32599
    },
    {
      "epoch": 1.775200288606182,
      "grad_norm": 0.5326225760551547,
      "learning_rate": 6.5526057212797585e-06,
      "loss": 11.7364,
      "step": 32600
    },
    {
      "epoch": 1.775254742602765,
      "grad_norm": 0.5278390274475373,
      "learning_rate": 6.549466506670244e-06,
      "loss": 11.7725,
      "step": 32601
    },
    {
      "epoch": 1.775309196599348,
      "grad_norm": 0.5295545231723466,
      "learning_rate": 6.546328018747772e-06,
      "loss": 11.7202,
      "step": 32602
    },
    {
      "epoch": 1.775363650595931,
      "grad_norm": 0.4988386655783545,
      "learning_rate": 6.543190257536702e-06,
      "loss": 11.8298,
      "step": 32603
    },
    {
      "epoch": 1.775418104592514,
      "grad_norm": 0.5897230884922778,
      "learning_rate": 6.5400532230614706e-06,
      "loss": 11.8236,
      "step": 32604
    },
    {
      "epoch": 1.775472558589097,
      "grad_norm": 0.5374701214690483,
      "learning_rate": 6.536916915346469e-06,
      "loss": 11.7702,
      "step": 32605
    },
    {
      "epoch": 1.77552701258568,
      "grad_norm": 0.5796462809143641,
      "learning_rate": 6.533781334416056e-06,
      "loss": 11.8336,
      "step": 32606
    },
    {
      "epoch": 1.775581466582263,
      "grad_norm": 0.5378564911576558,
      "learning_rate": 6.5306464802946445e-06,
      "loss": 11.6341,
      "step": 32607
    },
    {
      "epoch": 1.775635920578846,
      "grad_norm": 0.5606812729941905,
      "learning_rate": 6.527512353006593e-06,
      "loss": 11.9215,
      "step": 32608
    },
    {
      "epoch": 1.7756903745754289,
      "grad_norm": 0.5763508853752508,
      "learning_rate": 6.524378952576282e-06,
      "loss": 11.539,
      "step": 32609
    },
    {
      "epoch": 1.7757448285720119,
      "grad_norm": 0.6068411877977646,
      "learning_rate": 6.52124627902807e-06,
      "loss": 11.7649,
      "step": 32610
    },
    {
      "epoch": 1.7757992825685949,
      "grad_norm": 0.6007422795674634,
      "learning_rate": 6.518114332386327e-06,
      "loss": 11.8686,
      "step": 32611
    },
    {
      "epoch": 1.7758537365651779,
      "grad_norm": 0.5722685691181251,
      "learning_rate": 6.5149831126754105e-06,
      "loss": 11.9874,
      "step": 32612
    },
    {
      "epoch": 1.775908190561761,
      "grad_norm": 0.551065592349112,
      "learning_rate": 6.511852619919656e-06,
      "loss": 11.6544,
      "step": 32613
    },
    {
      "epoch": 1.775962644558344,
      "grad_norm": 0.5352143631889089,
      "learning_rate": 6.508722854143423e-06,
      "loss": 11.653,
      "step": 32614
    },
    {
      "epoch": 1.776017098554927,
      "grad_norm": 0.5635059400041645,
      "learning_rate": 6.505593815371047e-06,
      "loss": 11.8178,
      "step": 32615
    },
    {
      "epoch": 1.77607155255151,
      "grad_norm": 0.5749195072114188,
      "learning_rate": 6.502465503626831e-06,
      "loss": 11.843,
      "step": 32616
    },
    {
      "epoch": 1.776126006548093,
      "grad_norm": 0.5009668994357556,
      "learning_rate": 6.499337918935133e-06,
      "loss": 11.7927,
      "step": 32617
    },
    {
      "epoch": 1.7761804605446763,
      "grad_norm": 0.504484548633708,
      "learning_rate": 6.496211061320256e-06,
      "loss": 11.7142,
      "step": 32618
    },
    {
      "epoch": 1.7762349145412593,
      "grad_norm": 0.5485633307366743,
      "learning_rate": 6.493084930806537e-06,
      "loss": 11.7661,
      "step": 32619
    },
    {
      "epoch": 1.7762893685378423,
      "grad_norm": 0.4947471269227897,
      "learning_rate": 6.489959527418255e-06,
      "loss": 11.7114,
      "step": 32620
    },
    {
      "epoch": 1.7763438225344252,
      "grad_norm": 0.5299280750765318,
      "learning_rate": 6.486834851179746e-06,
      "loss": 11.7021,
      "step": 32621
    },
    {
      "epoch": 1.7763982765310082,
      "grad_norm": 0.6046416560381348,
      "learning_rate": 6.4837109021152696e-06,
      "loss": 11.8864,
      "step": 32622
    },
    {
      "epoch": 1.7764527305275912,
      "grad_norm": 0.5971586014993002,
      "learning_rate": 6.48058768024915e-06,
      "loss": 11.8263,
      "step": 32623
    },
    {
      "epoch": 1.7765071845241742,
      "grad_norm": 0.5535618459704909,
      "learning_rate": 6.477465185605669e-06,
      "loss": 11.8389,
      "step": 32624
    },
    {
      "epoch": 1.7765616385207572,
      "grad_norm": 0.5770403531098282,
      "learning_rate": 6.474343418209106e-06,
      "loss": 11.7517,
      "step": 32625
    },
    {
      "epoch": 1.7766160925173402,
      "grad_norm": 0.5941900106701521,
      "learning_rate": 6.471222378083719e-06,
      "loss": 11.784,
      "step": 32626
    },
    {
      "epoch": 1.7766705465139232,
      "grad_norm": 0.5260095947151255,
      "learning_rate": 6.468102065253812e-06,
      "loss": 11.7068,
      "step": 32627
    },
    {
      "epoch": 1.7767250005105062,
      "grad_norm": 0.4935738656016135,
      "learning_rate": 6.464982479743598e-06,
      "loss": 11.7772,
      "step": 32628
    },
    {
      "epoch": 1.7767794545070892,
      "grad_norm": 0.5438689668500842,
      "learning_rate": 6.461863621577391e-06,
      "loss": 11.6524,
      "step": 32629
    },
    {
      "epoch": 1.7768339085036722,
      "grad_norm": 0.5956595402759075,
      "learning_rate": 6.458745490779405e-06,
      "loss": 11.7141,
      "step": 32630
    },
    {
      "epoch": 1.7768883625002552,
      "grad_norm": 0.530747206389311,
      "learning_rate": 6.4556280873739214e-06,
      "loss": 11.8797,
      "step": 32631
    },
    {
      "epoch": 1.7769428164968382,
      "grad_norm": 0.5550826770474203,
      "learning_rate": 6.452511411385131e-06,
      "loss": 11.8071,
      "step": 32632
    },
    {
      "epoch": 1.7769972704934212,
      "grad_norm": 0.51211618541263,
      "learning_rate": 6.449395462837315e-06,
      "loss": 11.6899,
      "step": 32633
    },
    {
      "epoch": 1.7770517244900041,
      "grad_norm": 0.5216564382793045,
      "learning_rate": 6.446280241754687e-06,
      "loss": 11.7083,
      "step": 32634
    },
    {
      "epoch": 1.7771061784865871,
      "grad_norm": 0.5227773404881256,
      "learning_rate": 6.443165748161484e-06,
      "loss": 11.791,
      "step": 32635
    },
    {
      "epoch": 1.7771606324831704,
      "grad_norm": 0.5125099603910652,
      "learning_rate": 6.440051982081918e-06,
      "loss": 11.7994,
      "step": 32636
    },
    {
      "epoch": 1.7772150864797533,
      "grad_norm": 0.5557595719417294,
      "learning_rate": 6.436938943540172e-06,
      "loss": 11.596,
      "step": 32637
    },
    {
      "epoch": 1.7772695404763363,
      "grad_norm": 0.5539884593410619,
      "learning_rate": 6.433826632560491e-06,
      "loss": 11.7835,
      "step": 32638
    },
    {
      "epoch": 1.7773239944729193,
      "grad_norm": 0.5064855042708822,
      "learning_rate": 6.430715049167069e-06,
      "loss": 11.7443,
      "step": 32639
    },
    {
      "epoch": 1.7773784484695023,
      "grad_norm": 0.5309002760722017,
      "learning_rate": 6.427604193384085e-06,
      "loss": 11.9209,
      "step": 32640
    },
    {
      "epoch": 1.7774329024660853,
      "grad_norm": 0.5919438412692817,
      "learning_rate": 6.424494065235764e-06,
      "loss": 11.6732,
      "step": 32641
    },
    {
      "epoch": 1.7774873564626685,
      "grad_norm": 0.5355567293841104,
      "learning_rate": 6.421384664746244e-06,
      "loss": 11.7462,
      "step": 32642
    },
    {
      "epoch": 1.7775418104592515,
      "grad_norm": 0.5255919046413203,
      "learning_rate": 6.418275991939759e-06,
      "loss": 11.8555,
      "step": 32643
    },
    {
      "epoch": 1.7775962644558345,
      "grad_norm": 0.5583327883722515,
      "learning_rate": 6.415168046840425e-06,
      "loss": 11.7699,
      "step": 32644
    },
    {
      "epoch": 1.7776507184524175,
      "grad_norm": 0.5817085883094335,
      "learning_rate": 6.412060829472466e-06,
      "loss": 11.7111,
      "step": 32645
    },
    {
      "epoch": 1.7777051724490005,
      "grad_norm": 0.5088954993976289,
      "learning_rate": 6.408954339860007e-06,
      "loss": 11.8435,
      "step": 32646
    },
    {
      "epoch": 1.7777596264455835,
      "grad_norm": 0.5786548787336618,
      "learning_rate": 6.405848578027207e-06,
      "loss": 11.7142,
      "step": 32647
    },
    {
      "epoch": 1.7778140804421665,
      "grad_norm": 0.5514046851765381,
      "learning_rate": 6.402743543998235e-06,
      "loss": 11.6604,
      "step": 32648
    },
    {
      "epoch": 1.7778685344387495,
      "grad_norm": 0.5248882071075454,
      "learning_rate": 6.399639237797206e-06,
      "loss": 11.727,
      "step": 32649
    },
    {
      "epoch": 1.7779229884353325,
      "grad_norm": 0.5662896027745242,
      "learning_rate": 6.396535659448288e-06,
      "loss": 11.8725,
      "step": 32650
    },
    {
      "epoch": 1.7779774424319155,
      "grad_norm": 0.5455941611714284,
      "learning_rate": 6.393432808975619e-06,
      "loss": 11.8148,
      "step": 32651
    },
    {
      "epoch": 1.7780318964284985,
      "grad_norm": 0.5795116195227646,
      "learning_rate": 6.3903306864032895e-06,
      "loss": 11.7415,
      "step": 32652
    },
    {
      "epoch": 1.7780863504250815,
      "grad_norm": 0.5435402068006697,
      "learning_rate": 6.38722929175547e-06,
      "loss": 11.7942,
      "step": 32653
    },
    {
      "epoch": 1.7781408044216644,
      "grad_norm": 0.4898854798438047,
      "learning_rate": 6.384128625056241e-06,
      "loss": 11.6952,
      "step": 32654
    },
    {
      "epoch": 1.7781952584182474,
      "grad_norm": 0.5502141616198264,
      "learning_rate": 6.381028686329738e-06,
      "loss": 11.687,
      "step": 32655
    },
    {
      "epoch": 1.7782497124148304,
      "grad_norm": 0.5715235364798035,
      "learning_rate": 6.377929475600064e-06,
      "loss": 11.6627,
      "step": 32656
    },
    {
      "epoch": 1.7783041664114134,
      "grad_norm": 0.5565606715258251,
      "learning_rate": 6.37483099289129e-06,
      "loss": 11.7002,
      "step": 32657
    },
    {
      "epoch": 1.7783586204079964,
      "grad_norm": 0.559697312513193,
      "learning_rate": 6.371733238227539e-06,
      "loss": 11.8468,
      "step": 32658
    },
    {
      "epoch": 1.7784130744045796,
      "grad_norm": 0.5666534709032173,
      "learning_rate": 6.368636211632883e-06,
      "loss": 11.7649,
      "step": 32659
    },
    {
      "epoch": 1.7784675284011626,
      "grad_norm": 0.5574114448826537,
      "learning_rate": 6.3655399131314245e-06,
      "loss": 11.7239,
      "step": 32660
    },
    {
      "epoch": 1.7785219823977456,
      "grad_norm": 0.5925011940755665,
      "learning_rate": 6.362444342747232e-06,
      "loss": 11.8493,
      "step": 32661
    },
    {
      "epoch": 1.7785764363943286,
      "grad_norm": 0.5269764138423191,
      "learning_rate": 6.359349500504364e-06,
      "loss": 11.7507,
      "step": 32662
    },
    {
      "epoch": 1.7786308903909116,
      "grad_norm": 0.5403476584558234,
      "learning_rate": 6.356255386426912e-06,
      "loss": 11.8175,
      "step": 32663
    },
    {
      "epoch": 1.7786853443874946,
      "grad_norm": 0.61565991339433,
      "learning_rate": 6.353162000538903e-06,
      "loss": 12.0279,
      "step": 32664
    },
    {
      "epoch": 1.7787397983840778,
      "grad_norm": 0.5591293243317734,
      "learning_rate": 6.350069342864429e-06,
      "loss": 11.7781,
      "step": 32665
    },
    {
      "epoch": 1.7787942523806608,
      "grad_norm": 0.5662041809359237,
      "learning_rate": 6.346977413427524e-06,
      "loss": 11.9033,
      "step": 32666
    },
    {
      "epoch": 1.7788487063772438,
      "grad_norm": 0.5844082084590769,
      "learning_rate": 6.343886212252215e-06,
      "loss": 11.8189,
      "step": 32667
    },
    {
      "epoch": 1.7789031603738268,
      "grad_norm": 0.7302541912977605,
      "learning_rate": 6.340795739362559e-06,
      "loss": 11.8814,
      "step": 32668
    },
    {
      "epoch": 1.7789576143704098,
      "grad_norm": 0.5493051291355272,
      "learning_rate": 6.337705994782573e-06,
      "loss": 11.7421,
      "step": 32669
    },
    {
      "epoch": 1.7790120683669928,
      "grad_norm": 0.5271900879138697,
      "learning_rate": 6.3346169785363115e-06,
      "loss": 11.7093,
      "step": 32670
    },
    {
      "epoch": 1.7790665223635758,
      "grad_norm": 0.5556375173986846,
      "learning_rate": 6.331528690647759e-06,
      "loss": 11.7206,
      "step": 32671
    },
    {
      "epoch": 1.7791209763601588,
      "grad_norm": 0.5979757578752126,
      "learning_rate": 6.328441131140938e-06,
      "loss": 11.7581,
      "step": 32672
    },
    {
      "epoch": 1.7791754303567417,
      "grad_norm": 0.5278252118688105,
      "learning_rate": 6.325354300039887e-06,
      "loss": 11.8388,
      "step": 32673
    },
    {
      "epoch": 1.7792298843533247,
      "grad_norm": 0.5879647844127346,
      "learning_rate": 6.322268197368586e-06,
      "loss": 11.7815,
      "step": 32674
    },
    {
      "epoch": 1.7792843383499077,
      "grad_norm": 0.5437098437859124,
      "learning_rate": 6.3191828231510375e-06,
      "loss": 11.7574,
      "step": 32675
    },
    {
      "epoch": 1.7793387923464907,
      "grad_norm": 0.5610889898473906,
      "learning_rate": 6.316098177411245e-06,
      "loss": 11.7519,
      "step": 32676
    },
    {
      "epoch": 1.7793932463430737,
      "grad_norm": 0.5288993721382848,
      "learning_rate": 6.313014260173167e-06,
      "loss": 11.641,
      "step": 32677
    },
    {
      "epoch": 1.7794477003396567,
      "grad_norm": 0.5735405654787178,
      "learning_rate": 6.309931071460806e-06,
      "loss": 11.7809,
      "step": 32678
    },
    {
      "epoch": 1.7795021543362397,
      "grad_norm": 0.550847871540742,
      "learning_rate": 6.3068486112981325e-06,
      "loss": 11.7952,
      "step": 32679
    },
    {
      "epoch": 1.7795566083328227,
      "grad_norm": 0.6224556952124183,
      "learning_rate": 6.303766879709117e-06,
      "loss": 11.8511,
      "step": 32680
    },
    {
      "epoch": 1.7796110623294057,
      "grad_norm": 0.49699209659575416,
      "learning_rate": 6.300685876717716e-06,
      "loss": 11.7905,
      "step": 32681
    },
    {
      "epoch": 1.779665516325989,
      "grad_norm": 0.5197056696772931,
      "learning_rate": 6.2976056023479e-06,
      "loss": 11.7658,
      "step": 32682
    },
    {
      "epoch": 1.779719970322572,
      "grad_norm": 0.5299205418333491,
      "learning_rate": 6.294526056623617e-06,
      "loss": 11.7874,
      "step": 32683
    },
    {
      "epoch": 1.779774424319155,
      "grad_norm": 0.5006191990495225,
      "learning_rate": 6.291447239568815e-06,
      "loss": 11.6852,
      "step": 32684
    },
    {
      "epoch": 1.7798288783157379,
      "grad_norm": 0.5241928736389975,
      "learning_rate": 6.28836915120744e-06,
      "loss": 11.7525,
      "step": 32685
    },
    {
      "epoch": 1.7798833323123209,
      "grad_norm": 0.5512105472519098,
      "learning_rate": 6.28529179156343e-06,
      "loss": 11.8235,
      "step": 32686
    },
    {
      "epoch": 1.7799377863089039,
      "grad_norm": 0.5758030156783916,
      "learning_rate": 6.282215160660676e-06,
      "loss": 11.8157,
      "step": 32687
    },
    {
      "epoch": 1.779992240305487,
      "grad_norm": 0.5578829898472907,
      "learning_rate": 6.279139258523159e-06,
      "loss": 11.6964,
      "step": 32688
    },
    {
      "epoch": 1.78004669430207,
      "grad_norm": 0.5467188643625759,
      "learning_rate": 6.27606408517476e-06,
      "loss": 11.7781,
      "step": 32689
    },
    {
      "epoch": 1.780101148298653,
      "grad_norm": 0.6195338947234599,
      "learning_rate": 6.2729896406394155e-06,
      "loss": 11.8318,
      "step": 32690
    },
    {
      "epoch": 1.780155602295236,
      "grad_norm": 0.555715689687519,
      "learning_rate": 6.269915924941006e-06,
      "loss": 11.7176,
      "step": 32691
    },
    {
      "epoch": 1.780210056291819,
      "grad_norm": 0.5294021269516758,
      "learning_rate": 6.266842938103468e-06,
      "loss": 11.8282,
      "step": 32692
    },
    {
      "epoch": 1.780264510288402,
      "grad_norm": 0.5760740376196016,
      "learning_rate": 6.2637706801506605e-06,
      "loss": 11.7588,
      "step": 32693
    },
    {
      "epoch": 1.780318964284985,
      "grad_norm": 0.5384086038782483,
      "learning_rate": 6.2606991511064865e-06,
      "loss": 11.783,
      "step": 32694
    },
    {
      "epoch": 1.780373418281568,
      "grad_norm": 0.5761383915638604,
      "learning_rate": 6.257628350994859e-06,
      "loss": 11.8015,
      "step": 32695
    },
    {
      "epoch": 1.780427872278151,
      "grad_norm": 0.546854857346783,
      "learning_rate": 6.254558279839628e-06,
      "loss": 11.8423,
      "step": 32696
    },
    {
      "epoch": 1.780482326274734,
      "grad_norm": 0.5394379748626792,
      "learning_rate": 6.251488937664674e-06,
      "loss": 11.6775,
      "step": 32697
    },
    {
      "epoch": 1.780536780271317,
      "grad_norm": 0.5814153801090587,
      "learning_rate": 6.248420324493853e-06,
      "loss": 11.7156,
      "step": 32698
    },
    {
      "epoch": 1.7805912342679,
      "grad_norm": 0.5081416254563221,
      "learning_rate": 6.245352440351037e-06,
      "loss": 11.8127,
      "step": 32699
    },
    {
      "epoch": 1.780645688264483,
      "grad_norm": 0.5885158572104521,
      "learning_rate": 6.242285285260097e-06,
      "loss": 11.7697,
      "step": 32700
    },
    {
      "epoch": 1.780700142261066,
      "grad_norm": 0.5903081283855978,
      "learning_rate": 6.2392188592448554e-06,
      "loss": 11.8192,
      "step": 32701
    },
    {
      "epoch": 1.780754596257649,
      "grad_norm": 0.5813850653756711,
      "learning_rate": 6.236153162329184e-06,
      "loss": 11.6511,
      "step": 32702
    },
    {
      "epoch": 1.780809050254232,
      "grad_norm": 0.6015448142938231,
      "learning_rate": 6.233088194536896e-06,
      "loss": 11.8689,
      "step": 32703
    },
    {
      "epoch": 1.780863504250815,
      "grad_norm": 0.5436128411710512,
      "learning_rate": 6.2300239558918506e-06,
      "loss": 11.7093,
      "step": 32704
    },
    {
      "epoch": 1.780917958247398,
      "grad_norm": 0.601270612565341,
      "learning_rate": 6.226960446417851e-06,
      "loss": 11.7905,
      "step": 32705
    },
    {
      "epoch": 1.7809724122439812,
      "grad_norm": 0.5890192276838048,
      "learning_rate": 6.223897666138756e-06,
      "loss": 11.9191,
      "step": 32706
    },
    {
      "epoch": 1.7810268662405642,
      "grad_norm": 0.5455178899138761,
      "learning_rate": 6.220835615078346e-06,
      "loss": 11.7916,
      "step": 32707
    },
    {
      "epoch": 1.7810813202371472,
      "grad_norm": 0.552575852189147,
      "learning_rate": 6.217774293260436e-06,
      "loss": 11.7924,
      "step": 32708
    },
    {
      "epoch": 1.7811357742337302,
      "grad_norm": 0.526128252261358,
      "learning_rate": 6.2147137007088405e-06,
      "loss": 11.794,
      "step": 32709
    },
    {
      "epoch": 1.7811902282303131,
      "grad_norm": 0.5799430183200035,
      "learning_rate": 6.2116538374473735e-06,
      "loss": 11.7877,
      "step": 32710
    },
    {
      "epoch": 1.7812446822268961,
      "grad_norm": 0.5552499412949681,
      "learning_rate": 6.208594703499804e-06,
      "loss": 11.8429,
      "step": 32711
    },
    {
      "epoch": 1.7812991362234794,
      "grad_norm": 0.5965677348838014,
      "learning_rate": 6.2055362988899355e-06,
      "loss": 11.8804,
      "step": 32712
    },
    {
      "epoch": 1.7813535902200623,
      "grad_norm": 0.5188102393525604,
      "learning_rate": 6.202478623641539e-06,
      "loss": 11.7638,
      "step": 32713
    },
    {
      "epoch": 1.7814080442166453,
      "grad_norm": 0.5312133620617769,
      "learning_rate": 6.1994216777784166e-06,
      "loss": 11.6473,
      "step": 32714
    },
    {
      "epoch": 1.7814624982132283,
      "grad_norm": 0.5897830592166103,
      "learning_rate": 6.196365461324305e-06,
      "loss": 11.805,
      "step": 32715
    },
    {
      "epoch": 1.7815169522098113,
      "grad_norm": 0.546088818681991,
      "learning_rate": 6.193309974302997e-06,
      "loss": 11.8009,
      "step": 32716
    },
    {
      "epoch": 1.7815714062063943,
      "grad_norm": 0.5243351059626951,
      "learning_rate": 6.190255216738239e-06,
      "loss": 11.7343,
      "step": 32717
    },
    {
      "epoch": 1.7816258602029773,
      "grad_norm": 0.5256419405680276,
      "learning_rate": 6.187201188653779e-06,
      "loss": 11.7348,
      "step": 32718
    },
    {
      "epoch": 1.7816803141995603,
      "grad_norm": 0.5594804599033242,
      "learning_rate": 6.184147890073388e-06,
      "loss": 11.692,
      "step": 32719
    },
    {
      "epoch": 1.7817347681961433,
      "grad_norm": 0.5321778004033472,
      "learning_rate": 6.18109532102078e-06,
      "loss": 11.7293,
      "step": 32720
    },
    {
      "epoch": 1.7817892221927263,
      "grad_norm": 0.5011377673136107,
      "learning_rate": 6.178043481519713e-06,
      "loss": 11.7447,
      "step": 32721
    },
    {
      "epoch": 1.7818436761893093,
      "grad_norm": 0.5403480612341243,
      "learning_rate": 6.174992371593924e-06,
      "loss": 11.8208,
      "step": 32722
    },
    {
      "epoch": 1.7818981301858923,
      "grad_norm": 0.5525945969178169,
      "learning_rate": 6.171941991267116e-06,
      "loss": 11.8392,
      "step": 32723
    },
    {
      "epoch": 1.7819525841824753,
      "grad_norm": 0.5598211056078931,
      "learning_rate": 6.168892340563037e-06,
      "loss": 11.7512,
      "step": 32724
    },
    {
      "epoch": 1.7820070381790583,
      "grad_norm": 0.5632965586119917,
      "learning_rate": 6.165843419505379e-06,
      "loss": 11.696,
      "step": 32725
    },
    {
      "epoch": 1.7820614921756412,
      "grad_norm": 0.5739009722718315,
      "learning_rate": 6.162795228117868e-06,
      "loss": 11.8494,
      "step": 32726
    },
    {
      "epoch": 1.7821159461722242,
      "grad_norm": 0.5474179586942003,
      "learning_rate": 6.159747766424195e-06,
      "loss": 11.8151,
      "step": 32727
    },
    {
      "epoch": 1.7821704001688072,
      "grad_norm": 0.5117156862463175,
      "learning_rate": 6.156701034448042e-06,
      "loss": 11.7808,
      "step": 32728
    },
    {
      "epoch": 1.7822248541653904,
      "grad_norm": 0.5650488556079817,
      "learning_rate": 6.153655032213135e-06,
      "loss": 11.7281,
      "step": 32729
    },
    {
      "epoch": 1.7822793081619734,
      "grad_norm": 0.584714363894475,
      "learning_rate": 6.150609759743131e-06,
      "loss": 11.7672,
      "step": 32730
    },
    {
      "epoch": 1.7823337621585564,
      "grad_norm": 0.5345534779849421,
      "learning_rate": 6.147565217061735e-06,
      "loss": 11.8112,
      "step": 32731
    },
    {
      "epoch": 1.7823882161551394,
      "grad_norm": 0.5670964215084656,
      "learning_rate": 6.144521404192616e-06,
      "loss": 11.8892,
      "step": 32732
    },
    {
      "epoch": 1.7824426701517224,
      "grad_norm": 0.6037630502503243,
      "learning_rate": 6.141478321159422e-06,
      "loss": 12.0556,
      "step": 32733
    },
    {
      "epoch": 1.7824971241483054,
      "grad_norm": 0.5704912551789066,
      "learning_rate": 6.138435967985845e-06,
      "loss": 11.7784,
      "step": 32734
    },
    {
      "epoch": 1.7825515781448886,
      "grad_norm": 0.4837017365684751,
      "learning_rate": 6.135394344695522e-06,
      "loss": 11.7829,
      "step": 32735
    },
    {
      "epoch": 1.7826060321414716,
      "grad_norm": 0.5633150706322234,
      "learning_rate": 6.132353451312134e-06,
      "loss": 11.7697,
      "step": 32736
    },
    {
      "epoch": 1.7826604861380546,
      "grad_norm": 0.5994231024040529,
      "learning_rate": 6.129313287859295e-06,
      "loss": 11.8579,
      "step": 32737
    },
    {
      "epoch": 1.7827149401346376,
      "grad_norm": 0.5598219524080285,
      "learning_rate": 6.126273854360653e-06,
      "loss": 11.8211,
      "step": 32738
    },
    {
      "epoch": 1.7827693941312206,
      "grad_norm": 0.5437292911132128,
      "learning_rate": 6.123235150839857e-06,
      "loss": 11.7411,
      "step": 32739
    },
    {
      "epoch": 1.7828238481278036,
      "grad_norm": 0.5355352760056432,
      "learning_rate": 6.120197177320508e-06,
      "loss": 11.8826,
      "step": 32740
    },
    {
      "epoch": 1.7828783021243866,
      "grad_norm": 0.5801272054397876,
      "learning_rate": 6.117159933826277e-06,
      "loss": 11.8695,
      "step": 32741
    },
    {
      "epoch": 1.7829327561209696,
      "grad_norm": 0.6026029119665858,
      "learning_rate": 6.114123420380724e-06,
      "loss": 11.7992,
      "step": 32742
    },
    {
      "epoch": 1.7829872101175526,
      "grad_norm": 0.5090448052521881,
      "learning_rate": 6.111087637007506e-06,
      "loss": 11.8298,
      "step": 32743
    },
    {
      "epoch": 1.7830416641141356,
      "grad_norm": 0.5809190171254183,
      "learning_rate": 6.1080525837302284e-06,
      "loss": 11.8446,
      "step": 32744
    },
    {
      "epoch": 1.7830961181107186,
      "grad_norm": 0.5084379390934818,
      "learning_rate": 6.105018260572459e-06,
      "loss": 11.6902,
      "step": 32745
    },
    {
      "epoch": 1.7831505721073015,
      "grad_norm": 0.5102844516896258,
      "learning_rate": 6.101984667557836e-06,
      "loss": 11.7845,
      "step": 32746
    },
    {
      "epoch": 1.7832050261038845,
      "grad_norm": 0.5393110489205621,
      "learning_rate": 6.098951804709918e-06,
      "loss": 11.7556,
      "step": 32747
    },
    {
      "epoch": 1.7832594801004675,
      "grad_norm": 0.5610236594888275,
      "learning_rate": 6.095919672052286e-06,
      "loss": 11.8582,
      "step": 32748
    },
    {
      "epoch": 1.7833139340970505,
      "grad_norm": 0.5295984301667501,
      "learning_rate": 6.0928882696085435e-06,
      "loss": 11.8565,
      "step": 32749
    },
    {
      "epoch": 1.7833683880936335,
      "grad_norm": 0.5455010219919253,
      "learning_rate": 6.0898575974022375e-06,
      "loss": 11.7454,
      "step": 32750
    },
    {
      "epoch": 1.7834228420902165,
      "grad_norm": 0.532496244299336,
      "learning_rate": 6.0868276554569725e-06,
      "loss": 11.7405,
      "step": 32751
    },
    {
      "epoch": 1.7834772960867997,
      "grad_norm": 0.5289465348200701,
      "learning_rate": 6.083798443796251e-06,
      "loss": 11.7201,
      "step": 32752
    },
    {
      "epoch": 1.7835317500833827,
      "grad_norm": 0.5679209360527092,
      "learning_rate": 6.080769962443689e-06,
      "loss": 11.7737,
      "step": 32753
    },
    {
      "epoch": 1.7835862040799657,
      "grad_norm": 0.521273573393573,
      "learning_rate": 6.077742211422799e-06,
      "loss": 11.6698,
      "step": 32754
    },
    {
      "epoch": 1.7836406580765487,
      "grad_norm": 0.5747886977133413,
      "learning_rate": 6.07471519075713e-06,
      "loss": 11.8493,
      "step": 32755
    },
    {
      "epoch": 1.7836951120731317,
      "grad_norm": 0.5486564983341604,
      "learning_rate": 6.071688900470251e-06,
      "loss": 11.5643,
      "step": 32756
    },
    {
      "epoch": 1.7837495660697147,
      "grad_norm": 0.5807463120178638,
      "learning_rate": 6.068663340585656e-06,
      "loss": 11.7643,
      "step": 32757
    },
    {
      "epoch": 1.783804020066298,
      "grad_norm": 0.6299894467839595,
      "learning_rate": 6.065638511126881e-06,
      "loss": 11.8122,
      "step": 32758
    },
    {
      "epoch": 1.783858474062881,
      "grad_norm": 0.5433739098086517,
      "learning_rate": 6.062614412117462e-06,
      "loss": 11.8107,
      "step": 32759
    },
    {
      "epoch": 1.7839129280594639,
      "grad_norm": 0.5579663249929967,
      "learning_rate": 6.059591043580892e-06,
      "loss": 11.7298,
      "step": 32760
    },
    {
      "epoch": 1.7839673820560469,
      "grad_norm": 0.5379817593054868,
      "learning_rate": 6.056568405540719e-06,
      "loss": 11.8564,
      "step": 32761
    },
    {
      "epoch": 1.7840218360526299,
      "grad_norm": 0.5656778190490366,
      "learning_rate": 6.053546498020401e-06,
      "loss": 11.7822,
      "step": 32762
    },
    {
      "epoch": 1.7840762900492129,
      "grad_norm": 0.517132446100167,
      "learning_rate": 6.050525321043487e-06,
      "loss": 11.677,
      "step": 32763
    },
    {
      "epoch": 1.7841307440457959,
      "grad_norm": 0.5526752874623392,
      "learning_rate": 6.047504874633414e-06,
      "loss": 11.683,
      "step": 32764
    },
    {
      "epoch": 1.7841851980423789,
      "grad_norm": 0.6635846424445987,
      "learning_rate": 6.0444851588137065e-06,
      "loss": 11.8886,
      "step": 32765
    },
    {
      "epoch": 1.7842396520389618,
      "grad_norm": 0.5028456486693429,
      "learning_rate": 6.041466173607868e-06,
      "loss": 11.7425,
      "step": 32766
    },
    {
      "epoch": 1.7842941060355448,
      "grad_norm": 0.5358399233934722,
      "learning_rate": 6.038447919039314e-06,
      "loss": 11.8024,
      "step": 32767
    },
    {
      "epoch": 1.7843485600321278,
      "grad_norm": 0.6545409848659228,
      "learning_rate": 6.03543039513157e-06,
      "loss": 11.932,
      "step": 32768
    },
    {
      "epoch": 1.7844030140287108,
      "grad_norm": 0.5409977093844731,
      "learning_rate": 6.032413601908049e-06,
      "loss": 11.7231,
      "step": 32769
    },
    {
      "epoch": 1.7844574680252938,
      "grad_norm": 0.5547535912506245,
      "learning_rate": 6.029397539392246e-06,
      "loss": 11.8589,
      "step": 32770
    },
    {
      "epoch": 1.7845119220218768,
      "grad_norm": 0.5452665682740018,
      "learning_rate": 6.026382207607617e-06,
      "loss": 11.8653,
      "step": 32771
    },
    {
      "epoch": 1.7845663760184598,
      "grad_norm": 0.5795592234860748,
      "learning_rate": 6.02336760657759e-06,
      "loss": 11.8091,
      "step": 32772
    },
    {
      "epoch": 1.7846208300150428,
      "grad_norm": 0.5758528998235982,
      "learning_rate": 6.020353736325635e-06,
      "loss": 11.7532,
      "step": 32773
    },
    {
      "epoch": 1.7846752840116258,
      "grad_norm": 0.5091401476597546,
      "learning_rate": 6.017340596875143e-06,
      "loss": 11.6873,
      "step": 32774
    },
    {
      "epoch": 1.7847297380082088,
      "grad_norm": 0.531788250732534,
      "learning_rate": 6.014328188249596e-06,
      "loss": 11.8192,
      "step": 32775
    },
    {
      "epoch": 1.784784192004792,
      "grad_norm": 0.5296134574860314,
      "learning_rate": 6.011316510472386e-06,
      "loss": 11.7253,
      "step": 32776
    },
    {
      "epoch": 1.784838646001375,
      "grad_norm": 0.527924853212081,
      "learning_rate": 6.00830556356693e-06,
      "loss": 11.8661,
      "step": 32777
    },
    {
      "epoch": 1.784893099997958,
      "grad_norm": 0.5520828888280412,
      "learning_rate": 6.005295347556672e-06,
      "loss": 11.7979,
      "step": 32778
    },
    {
      "epoch": 1.784947553994541,
      "grad_norm": 0.5351865975895744,
      "learning_rate": 6.0022858624649735e-06,
      "loss": 11.7371,
      "step": 32779
    },
    {
      "epoch": 1.785002007991124,
      "grad_norm": 0.537562345824475,
      "learning_rate": 5.999277108315271e-06,
      "loss": 11.749,
      "step": 32780
    },
    {
      "epoch": 1.785056461987707,
      "grad_norm": 0.6614506627680491,
      "learning_rate": 5.9962690851309675e-06,
      "loss": 11.9624,
      "step": 32781
    },
    {
      "epoch": 1.7851109159842902,
      "grad_norm": 0.532982334405565,
      "learning_rate": 5.993261792935423e-06,
      "loss": 11.8047,
      "step": 32782
    },
    {
      "epoch": 1.7851653699808732,
      "grad_norm": 0.565314211049647,
      "learning_rate": 5.990255231752062e-06,
      "loss": 11.8521,
      "step": 32783
    },
    {
      "epoch": 1.7852198239774562,
      "grad_norm": 0.559505079002129,
      "learning_rate": 5.9872494016042225e-06,
      "loss": 11.9165,
      "step": 32784
    },
    {
      "epoch": 1.7852742779740391,
      "grad_norm": 0.49870783123113654,
      "learning_rate": 5.984244302515307e-06,
      "loss": 11.6201,
      "step": 32785
    },
    {
      "epoch": 1.7853287319706221,
      "grad_norm": 0.5204790619077684,
      "learning_rate": 5.981239934508686e-06,
      "loss": 11.8264,
      "step": 32786
    },
    {
      "epoch": 1.7853831859672051,
      "grad_norm": 0.5996319298658219,
      "learning_rate": 5.9782362976076865e-06,
      "loss": 11.8328,
      "step": 32787
    },
    {
      "epoch": 1.7854376399637881,
      "grad_norm": 0.5637128176984325,
      "learning_rate": 5.97523339183571e-06,
      "loss": 11.8389,
      "step": 32788
    },
    {
      "epoch": 1.7854920939603711,
      "grad_norm": 0.5489482003779558,
      "learning_rate": 5.972231217216062e-06,
      "loss": 11.7052,
      "step": 32789
    },
    {
      "epoch": 1.785546547956954,
      "grad_norm": 0.6024242860078325,
      "learning_rate": 5.969229773772134e-06,
      "loss": 11.9011,
      "step": 32790
    },
    {
      "epoch": 1.785601001953537,
      "grad_norm": 0.5263928789986556,
      "learning_rate": 5.966229061527229e-06,
      "loss": 11.8157,
      "step": 32791
    },
    {
      "epoch": 1.78565545595012,
      "grad_norm": 0.5241021630678591,
      "learning_rate": 5.963229080504696e-06,
      "loss": 11.7763,
      "step": 32792
    },
    {
      "epoch": 1.785709909946703,
      "grad_norm": 0.5216179403005902,
      "learning_rate": 5.960229830727882e-06,
      "loss": 11.7932,
      "step": 32793
    },
    {
      "epoch": 1.785764363943286,
      "grad_norm": 0.5423555008380151,
      "learning_rate": 5.957231312220069e-06,
      "loss": 11.8024,
      "step": 32794
    },
    {
      "epoch": 1.785818817939869,
      "grad_norm": 0.5718442422207454,
      "learning_rate": 5.9542335250046155e-06,
      "loss": 11.5255,
      "step": 32795
    },
    {
      "epoch": 1.785873271936452,
      "grad_norm": 0.5243168265178102,
      "learning_rate": 5.951236469104815e-06,
      "loss": 11.7026,
      "step": 32796
    },
    {
      "epoch": 1.785927725933035,
      "grad_norm": 0.48359023598502987,
      "learning_rate": 5.9482401445439595e-06,
      "loss": 11.7673,
      "step": 32797
    },
    {
      "epoch": 1.785982179929618,
      "grad_norm": 0.5389200397432837,
      "learning_rate": 5.945244551345364e-06,
      "loss": 11.7954,
      "step": 32798
    },
    {
      "epoch": 1.7860366339262013,
      "grad_norm": 0.5026237159577307,
      "learning_rate": 5.94224968953232e-06,
      "loss": 11.6851,
      "step": 32799
    },
    {
      "epoch": 1.7860910879227843,
      "grad_norm": 0.5157097858037699,
      "learning_rate": 5.939255559128121e-06,
      "loss": 11.7758,
      "step": 32800
    },
    {
      "epoch": 1.7861455419193673,
      "grad_norm": 0.5293425019336567,
      "learning_rate": 5.936262160156025e-06,
      "loss": 11.7399,
      "step": 32801
    },
    {
      "epoch": 1.7861999959159502,
      "grad_norm": 0.4931659322533404,
      "learning_rate": 5.933269492639359e-06,
      "loss": 11.6566,
      "step": 32802
    },
    {
      "epoch": 1.7862544499125332,
      "grad_norm": 0.5246199651328718,
      "learning_rate": 5.930277556601338e-06,
      "loss": 11.7588,
      "step": 32803
    },
    {
      "epoch": 1.7863089039091162,
      "grad_norm": 0.545291086389174,
      "learning_rate": 5.927286352065253e-06,
      "loss": 11.7092,
      "step": 32804
    },
    {
      "epoch": 1.7863633579056994,
      "grad_norm": 0.6090685558430966,
      "learning_rate": 5.9242958790543865e-06,
      "loss": 11.859,
      "step": 32805
    },
    {
      "epoch": 1.7864178119022824,
      "grad_norm": 0.5063771339523607,
      "learning_rate": 5.921306137591975e-06,
      "loss": 11.637,
      "step": 32806
    },
    {
      "epoch": 1.7864722658988654,
      "grad_norm": 0.5353859595848508,
      "learning_rate": 5.918317127701245e-06,
      "loss": 11.8884,
      "step": 32807
    },
    {
      "epoch": 1.7865267198954484,
      "grad_norm": 0.5738267069233514,
      "learning_rate": 5.915328849405466e-06,
      "loss": 11.8586,
      "step": 32808
    },
    {
      "epoch": 1.7865811738920314,
      "grad_norm": 0.5902060104006173,
      "learning_rate": 5.912341302727864e-06,
      "loss": 11.7625,
      "step": 32809
    },
    {
      "epoch": 1.7866356278886144,
      "grad_norm": 0.5447056975591531,
      "learning_rate": 5.909354487691687e-06,
      "loss": 11.8551,
      "step": 32810
    },
    {
      "epoch": 1.7866900818851974,
      "grad_norm": 0.6176259087339802,
      "learning_rate": 5.9063684043201285e-06,
      "loss": 11.75,
      "step": 32811
    },
    {
      "epoch": 1.7867445358817804,
      "grad_norm": 0.5873460302544379,
      "learning_rate": 5.903383052636446e-06,
      "loss": 11.7537,
      "step": 32812
    },
    {
      "epoch": 1.7867989898783634,
      "grad_norm": 0.5253352394469801,
      "learning_rate": 5.900398432663823e-06,
      "loss": 11.84,
      "step": 32813
    },
    {
      "epoch": 1.7868534438749464,
      "grad_norm": 0.583531651090751,
      "learning_rate": 5.897414544425483e-06,
      "loss": 11.9039,
      "step": 32814
    },
    {
      "epoch": 1.7869078978715294,
      "grad_norm": 0.5496762027681773,
      "learning_rate": 5.894431387944644e-06,
      "loss": 11.6629,
      "step": 32815
    },
    {
      "epoch": 1.7869623518681124,
      "grad_norm": 0.5588490667752386,
      "learning_rate": 5.891448963244483e-06,
      "loss": 11.7954,
      "step": 32816
    },
    {
      "epoch": 1.7870168058646954,
      "grad_norm": 0.5534679792393347,
      "learning_rate": 5.888467270348208e-06,
      "loss": 11.7441,
      "step": 32817
    },
    {
      "epoch": 1.7870712598612783,
      "grad_norm": 0.5075673098860951,
      "learning_rate": 5.885486309278964e-06,
      "loss": 11.7683,
      "step": 32818
    },
    {
      "epoch": 1.7871257138578613,
      "grad_norm": 0.582522103117588,
      "learning_rate": 5.882506080059968e-06,
      "loss": 11.7187,
      "step": 32819
    },
    {
      "epoch": 1.7871801678544443,
      "grad_norm": 0.560677545645884,
      "learning_rate": 5.879526582714412e-06,
      "loss": 11.7653,
      "step": 32820
    },
    {
      "epoch": 1.7872346218510273,
      "grad_norm": 0.5288387157816162,
      "learning_rate": 5.876547817265421e-06,
      "loss": 11.7938,
      "step": 32821
    },
    {
      "epoch": 1.7872890758476105,
      "grad_norm": 0.5107213676622757,
      "learning_rate": 5.873569783736188e-06,
      "loss": 11.748,
      "step": 32822
    },
    {
      "epoch": 1.7873435298441935,
      "grad_norm": 0.5409498270161708,
      "learning_rate": 5.87059248214985e-06,
      "loss": 11.8192,
      "step": 32823
    },
    {
      "epoch": 1.7873979838407765,
      "grad_norm": 0.4982023751157804,
      "learning_rate": 5.867615912529589e-06,
      "loss": 11.7029,
      "step": 32824
    },
    {
      "epoch": 1.7874524378373595,
      "grad_norm": 0.5016788576366078,
      "learning_rate": 5.864640074898509e-06,
      "loss": 11.7278,
      "step": 32825
    },
    {
      "epoch": 1.7875068918339425,
      "grad_norm": 0.6166524688272519,
      "learning_rate": 5.861664969279779e-06,
      "loss": 11.8109,
      "step": 32826
    },
    {
      "epoch": 1.7875613458305255,
      "grad_norm": 0.6151538621448444,
      "learning_rate": 5.858690595696559e-06,
      "loss": 11.8117,
      "step": 32827
    },
    {
      "epoch": 1.7876157998271087,
      "grad_norm": 0.5488559382736814,
      "learning_rate": 5.855716954171919e-06,
      "loss": 11.7301,
      "step": 32828
    },
    {
      "epoch": 1.7876702538236917,
      "grad_norm": 0.5040495965532082,
      "learning_rate": 5.852744044729008e-06,
      "loss": 11.8371,
      "step": 32829
    },
    {
      "epoch": 1.7877247078202747,
      "grad_norm": 0.5197784734437446,
      "learning_rate": 5.849771867390974e-06,
      "loss": 11.7459,
      "step": 32830
    },
    {
      "epoch": 1.7877791618168577,
      "grad_norm": 0.5692484554854838,
      "learning_rate": 5.846800422180876e-06,
      "loss": 11.7582,
      "step": 32831
    },
    {
      "epoch": 1.7878336158134407,
      "grad_norm": 0.5860774792490339,
      "learning_rate": 5.843829709121862e-06,
      "loss": 11.8864,
      "step": 32832
    },
    {
      "epoch": 1.7878880698100237,
      "grad_norm": 0.5093837756154763,
      "learning_rate": 5.840859728237013e-06,
      "loss": 11.6715,
      "step": 32833
    },
    {
      "epoch": 1.7879425238066067,
      "grad_norm": 0.529759685057493,
      "learning_rate": 5.837890479549435e-06,
      "loss": 11.8237,
      "step": 32834
    },
    {
      "epoch": 1.7879969778031897,
      "grad_norm": 0.5358648716126443,
      "learning_rate": 5.834921963082207e-06,
      "loss": 11.7788,
      "step": 32835
    },
    {
      "epoch": 1.7880514317997727,
      "grad_norm": 0.49667954301496087,
      "learning_rate": 5.831954178858412e-06,
      "loss": 11.6285,
      "step": 32836
    },
    {
      "epoch": 1.7881058857963557,
      "grad_norm": 0.5044001812051533,
      "learning_rate": 5.828987126901164e-06,
      "loss": 11.6813,
      "step": 32837
    },
    {
      "epoch": 1.7881603397929386,
      "grad_norm": 0.49883837962208377,
      "learning_rate": 5.826020807233467e-06,
      "loss": 11.8703,
      "step": 32838
    },
    {
      "epoch": 1.7882147937895216,
      "grad_norm": 0.520880660530782,
      "learning_rate": 5.823055219878448e-06,
      "loss": 11.7585,
      "step": 32839
    },
    {
      "epoch": 1.7882692477861046,
      "grad_norm": 0.5862275949805851,
      "learning_rate": 5.820090364859132e-06,
      "loss": 11.7042,
      "step": 32840
    },
    {
      "epoch": 1.7883237017826876,
      "grad_norm": 0.6167287607529381,
      "learning_rate": 5.817126242198578e-06,
      "loss": 11.7581,
      "step": 32841
    },
    {
      "epoch": 1.7883781557792706,
      "grad_norm": 0.5315434795855483,
      "learning_rate": 5.814162851919869e-06,
      "loss": 11.567,
      "step": 32842
    },
    {
      "epoch": 1.7884326097758536,
      "grad_norm": 0.5537450764862275,
      "learning_rate": 5.811200194046007e-06,
      "loss": 11.7587,
      "step": 32843
    },
    {
      "epoch": 1.7884870637724366,
      "grad_norm": 0.5531575408292612,
      "learning_rate": 5.808238268600064e-06,
      "loss": 11.929,
      "step": 32844
    },
    {
      "epoch": 1.7885415177690196,
      "grad_norm": 0.5542420036282248,
      "learning_rate": 5.8052770756050315e-06,
      "loss": 11.7609,
      "step": 32845
    },
    {
      "epoch": 1.7885959717656028,
      "grad_norm": 0.5390362602659625,
      "learning_rate": 5.80231661508398e-06,
      "loss": 11.7488,
      "step": 32846
    },
    {
      "epoch": 1.7886504257621858,
      "grad_norm": 0.5517606227076052,
      "learning_rate": 5.799356887059915e-06,
      "loss": 11.844,
      "step": 32847
    },
    {
      "epoch": 1.7887048797587688,
      "grad_norm": 0.5451060092260289,
      "learning_rate": 5.7963978915558384e-06,
      "loss": 11.7586,
      "step": 32848
    },
    {
      "epoch": 1.7887593337553518,
      "grad_norm": 0.5593841263572491,
      "learning_rate": 5.793439628594777e-06,
      "loss": 11.7846,
      "step": 32849
    },
    {
      "epoch": 1.7888137877519348,
      "grad_norm": 0.5517176516321998,
      "learning_rate": 5.7904820981997125e-06,
      "loss": 11.7455,
      "step": 32850
    },
    {
      "epoch": 1.788868241748518,
      "grad_norm": 0.5971910759972061,
      "learning_rate": 5.787525300393681e-06,
      "loss": 11.8001,
      "step": 32851
    },
    {
      "epoch": 1.788922695745101,
      "grad_norm": 0.5457154284841771,
      "learning_rate": 5.784569235199633e-06,
      "loss": 11.7141,
      "step": 32852
    },
    {
      "epoch": 1.788977149741684,
      "grad_norm": 0.5036840400789974,
      "learning_rate": 5.78161390264057e-06,
      "loss": 11.8536,
      "step": 32853
    },
    {
      "epoch": 1.789031603738267,
      "grad_norm": 0.5905699640468144,
      "learning_rate": 5.778659302739497e-06,
      "loss": 11.6751,
      "step": 32854
    },
    {
      "epoch": 1.78908605773485,
      "grad_norm": 0.4858515758353979,
      "learning_rate": 5.775705435519351e-06,
      "loss": 11.6447,
      "step": 32855
    },
    {
      "epoch": 1.789140511731433,
      "grad_norm": 0.49867023838384755,
      "learning_rate": 5.772752301003148e-06,
      "loss": 11.6954,
      "step": 32856
    },
    {
      "epoch": 1.789194965728016,
      "grad_norm": 0.5234473695025423,
      "learning_rate": 5.769799899213812e-06,
      "loss": 11.8778,
      "step": 32857
    },
    {
      "epoch": 1.789249419724599,
      "grad_norm": 0.5285744211556234,
      "learning_rate": 5.766848230174304e-06,
      "loss": 11.8652,
      "step": 32858
    },
    {
      "epoch": 1.789303873721182,
      "grad_norm": 0.5694143238664334,
      "learning_rate": 5.763897293907605e-06,
      "loss": 11.6405,
      "step": 32859
    },
    {
      "epoch": 1.789358327717765,
      "grad_norm": 0.5514348461301628,
      "learning_rate": 5.760947090436619e-06,
      "loss": 11.7148,
      "step": 32860
    },
    {
      "epoch": 1.789412781714348,
      "grad_norm": 0.5305600285038201,
      "learning_rate": 5.757997619784339e-06,
      "loss": 11.7894,
      "step": 32861
    },
    {
      "epoch": 1.789467235710931,
      "grad_norm": 0.5262665324909412,
      "learning_rate": 5.755048881973657e-06,
      "loss": 11.8313,
      "step": 32862
    },
    {
      "epoch": 1.789521689707514,
      "grad_norm": 0.5583189119528796,
      "learning_rate": 5.752100877027511e-06,
      "loss": 11.8179,
      "step": 32863
    },
    {
      "epoch": 1.789576143704097,
      "grad_norm": 0.5139461141436407,
      "learning_rate": 5.74915360496886e-06,
      "loss": 11.6874,
      "step": 32864
    },
    {
      "epoch": 1.78963059770068,
      "grad_norm": 0.531811869166695,
      "learning_rate": 5.746207065820575e-06,
      "loss": 11.9065,
      "step": 32865
    },
    {
      "epoch": 1.7896850516972629,
      "grad_norm": 0.5573450083285999,
      "learning_rate": 5.743261259605603e-06,
      "loss": 11.7443,
      "step": 32866
    },
    {
      "epoch": 1.7897395056938459,
      "grad_norm": 0.5363820158103323,
      "learning_rate": 5.740316186346839e-06,
      "loss": 11.7429,
      "step": 32867
    },
    {
      "epoch": 1.7897939596904289,
      "grad_norm": 0.5264741953493456,
      "learning_rate": 5.737371846067174e-06,
      "loss": 11.5681,
      "step": 32868
    },
    {
      "epoch": 1.789848413687012,
      "grad_norm": 0.5184990745929842,
      "learning_rate": 5.734428238789524e-06,
      "loss": 11.8289,
      "step": 32869
    },
    {
      "epoch": 1.789902867683595,
      "grad_norm": 0.5524714354586424,
      "learning_rate": 5.73148536453676e-06,
      "loss": 11.6553,
      "step": 32870
    },
    {
      "epoch": 1.789957321680178,
      "grad_norm": 0.5484058375687554,
      "learning_rate": 5.728543223331784e-06,
      "loss": 11.8601,
      "step": 32871
    },
    {
      "epoch": 1.790011775676761,
      "grad_norm": 0.5695995821207973,
      "learning_rate": 5.725601815197445e-06,
      "loss": 11.7591,
      "step": 32872
    },
    {
      "epoch": 1.790066229673344,
      "grad_norm": 0.57874390247035,
      "learning_rate": 5.72266114015666e-06,
      "loss": 11.7745,
      "step": 32873
    },
    {
      "epoch": 1.790120683669927,
      "grad_norm": 0.531637325728104,
      "learning_rate": 5.719721198232253e-06,
      "loss": 11.7771,
      "step": 32874
    },
    {
      "epoch": 1.7901751376665103,
      "grad_norm": 0.5462136666414714,
      "learning_rate": 5.716781989447106e-06,
      "loss": 11.6699,
      "step": 32875
    },
    {
      "epoch": 1.7902295916630933,
      "grad_norm": 0.503406205559153,
      "learning_rate": 5.713843513824091e-06,
      "loss": 11.6755,
      "step": 32876
    },
    {
      "epoch": 1.7902840456596762,
      "grad_norm": 0.4759854353265222,
      "learning_rate": 5.710905771386043e-06,
      "loss": 11.6307,
      "step": 32877
    },
    {
      "epoch": 1.7903384996562592,
      "grad_norm": 0.5647717676050596,
      "learning_rate": 5.70796876215578e-06,
      "loss": 11.7854,
      "step": 32878
    },
    {
      "epoch": 1.7903929536528422,
      "grad_norm": 0.5050902283551958,
      "learning_rate": 5.705032486156181e-06,
      "loss": 11.5763,
      "step": 32879
    },
    {
      "epoch": 1.7904474076494252,
      "grad_norm": 0.5145242576764114,
      "learning_rate": 5.702096943410052e-06,
      "loss": 11.7488,
      "step": 32880
    },
    {
      "epoch": 1.7905018616460082,
      "grad_norm": 0.5428305930845646,
      "learning_rate": 5.69916213394025e-06,
      "loss": 11.7431,
      "step": 32881
    },
    {
      "epoch": 1.7905563156425912,
      "grad_norm": 0.5071416773790963,
      "learning_rate": 5.696228057769559e-06,
      "loss": 11.7854,
      "step": 32882
    },
    {
      "epoch": 1.7906107696391742,
      "grad_norm": 0.5550151678472932,
      "learning_rate": 5.693294714920816e-06,
      "loss": 11.8018,
      "step": 32883
    },
    {
      "epoch": 1.7906652236357572,
      "grad_norm": 0.5399487750150528,
      "learning_rate": 5.690362105416825e-06,
      "loss": 11.7867,
      "step": 32884
    },
    {
      "epoch": 1.7907196776323402,
      "grad_norm": 0.5485015365277324,
      "learning_rate": 5.6874302292803995e-06,
      "loss": 11.682,
      "step": 32885
    },
    {
      "epoch": 1.7907741316289232,
      "grad_norm": 0.7571456953560544,
      "learning_rate": 5.684499086534345e-06,
      "loss": 11.9283,
      "step": 32886
    },
    {
      "epoch": 1.7908285856255062,
      "grad_norm": 0.5273593563894615,
      "learning_rate": 5.681568677201432e-06,
      "loss": 11.7999,
      "step": 32887
    },
    {
      "epoch": 1.7908830396220892,
      "grad_norm": 0.626783140819843,
      "learning_rate": 5.678639001304464e-06,
      "loss": 11.9423,
      "step": 32888
    },
    {
      "epoch": 1.7909374936186722,
      "grad_norm": 0.5420035564587004,
      "learning_rate": 5.675710058866202e-06,
      "loss": 11.8615,
      "step": 32889
    },
    {
      "epoch": 1.7909919476152552,
      "grad_norm": 0.518446198273281,
      "learning_rate": 5.672781849909436e-06,
      "loss": 11.6747,
      "step": 32890
    },
    {
      "epoch": 1.7910464016118381,
      "grad_norm": 0.5310628896367843,
      "learning_rate": 5.669854374456962e-06,
      "loss": 11.7012,
      "step": 32891
    },
    {
      "epoch": 1.7911008556084214,
      "grad_norm": 0.5225573995709512,
      "learning_rate": 5.666927632531494e-06,
      "loss": 11.8082,
      "step": 32892
    },
    {
      "epoch": 1.7911553096050044,
      "grad_norm": 0.5447897525909504,
      "learning_rate": 5.664001624155835e-06,
      "loss": 11.8022,
      "step": 32893
    },
    {
      "epoch": 1.7912097636015873,
      "grad_norm": 0.5324193042929436,
      "learning_rate": 5.661076349352701e-06,
      "loss": 11.9058,
      "step": 32894
    },
    {
      "epoch": 1.7912642175981703,
      "grad_norm": 0.6308899564451923,
      "learning_rate": 5.658151808144874e-06,
      "loss": 11.6467,
      "step": 32895
    },
    {
      "epoch": 1.7913186715947533,
      "grad_norm": 0.5388995465943365,
      "learning_rate": 5.655228000555069e-06,
      "loss": 11.7834,
      "step": 32896
    },
    {
      "epoch": 1.7913731255913363,
      "grad_norm": 0.5674606228068833,
      "learning_rate": 5.652304926606045e-06,
      "loss": 11.7996,
      "step": 32897
    },
    {
      "epoch": 1.7914275795879195,
      "grad_norm": 0.5652502506867298,
      "learning_rate": 5.649382586320517e-06,
      "loss": 11.7887,
      "step": 32898
    },
    {
      "epoch": 1.7914820335845025,
      "grad_norm": 0.5457927536965536,
      "learning_rate": 5.646460979721202e-06,
      "loss": 11.6952,
      "step": 32899
    },
    {
      "epoch": 1.7915364875810855,
      "grad_norm": 0.6501923471868624,
      "learning_rate": 5.643540106830825e-06,
      "loss": 11.9492,
      "step": 32900
    },
    {
      "epoch": 1.7915909415776685,
      "grad_norm": 0.5184715250681738,
      "learning_rate": 5.640619967672123e-06,
      "loss": 11.789,
      "step": 32901
    },
    {
      "epoch": 1.7916453955742515,
      "grad_norm": 0.5407044796019927,
      "learning_rate": 5.637700562267767e-06,
      "loss": 11.8146,
      "step": 32902
    },
    {
      "epoch": 1.7916998495708345,
      "grad_norm": 0.5577398170497451,
      "learning_rate": 5.634781890640484e-06,
      "loss": 11.7791,
      "step": 32903
    },
    {
      "epoch": 1.7917543035674175,
      "grad_norm": 0.564337721593817,
      "learning_rate": 5.631863952812955e-06,
      "loss": 11.9121,
      "step": 32904
    },
    {
      "epoch": 1.7918087575640005,
      "grad_norm": 0.5491338197300244,
      "learning_rate": 5.628946748807895e-06,
      "loss": 11.7338,
      "step": 32905
    },
    {
      "epoch": 1.7918632115605835,
      "grad_norm": 0.529156402740374,
      "learning_rate": 5.626030278647954e-06,
      "loss": 11.7318,
      "step": 32906
    },
    {
      "epoch": 1.7919176655571665,
      "grad_norm": 0.6292734936144189,
      "learning_rate": 5.623114542355845e-06,
      "loss": 11.7865,
      "step": 32907
    },
    {
      "epoch": 1.7919721195537495,
      "grad_norm": 0.558882568823229,
      "learning_rate": 5.620199539954218e-06,
      "loss": 11.9001,
      "step": 32908
    },
    {
      "epoch": 1.7920265735503325,
      "grad_norm": 0.5287236011991581,
      "learning_rate": 5.6172852714657335e-06,
      "loss": 11.6772,
      "step": 32909
    },
    {
      "epoch": 1.7920810275469155,
      "grad_norm": 0.5153919322195507,
      "learning_rate": 5.614371736913082e-06,
      "loss": 11.6605,
      "step": 32910
    },
    {
      "epoch": 1.7921354815434984,
      "grad_norm": 0.49298910148421154,
      "learning_rate": 5.6114589363188915e-06,
      "loss": 11.7694,
      "step": 32911
    },
    {
      "epoch": 1.7921899355400814,
      "grad_norm": 0.5639422226384082,
      "learning_rate": 5.608546869705822e-06,
      "loss": 11.7764,
      "step": 32912
    },
    {
      "epoch": 1.7922443895366644,
      "grad_norm": 0.48621126242864204,
      "learning_rate": 5.605635537096543e-06,
      "loss": 11.6721,
      "step": 32913
    },
    {
      "epoch": 1.7922988435332474,
      "grad_norm": 0.5956402582652661,
      "learning_rate": 5.602724938513649e-06,
      "loss": 11.8766,
      "step": 32914
    },
    {
      "epoch": 1.7923532975298304,
      "grad_norm": 0.578792789996769,
      "learning_rate": 5.599815073979819e-06,
      "loss": 11.8459,
      "step": 32915
    },
    {
      "epoch": 1.7924077515264136,
      "grad_norm": 0.520250136008144,
      "learning_rate": 5.5969059435176386e-06,
      "loss": 11.8288,
      "step": 32916
    },
    {
      "epoch": 1.7924622055229966,
      "grad_norm": 0.6073175168381555,
      "learning_rate": 5.593997547149765e-06,
      "loss": 11.8408,
      "step": 32917
    },
    {
      "epoch": 1.7925166595195796,
      "grad_norm": 0.5155050623384161,
      "learning_rate": 5.5910898848987925e-06,
      "loss": 11.831,
      "step": 32918
    },
    {
      "epoch": 1.7925711135161626,
      "grad_norm": 0.5066378974759215,
      "learning_rate": 5.5881829567873355e-06,
      "loss": 11.7648,
      "step": 32919
    },
    {
      "epoch": 1.7926255675127456,
      "grad_norm": 0.530946138852541,
      "learning_rate": 5.585276762838009e-06,
      "loss": 11.7931,
      "step": 32920
    },
    {
      "epoch": 1.7926800215093288,
      "grad_norm": 0.5880682300761637,
      "learning_rate": 5.582371303073386e-06,
      "loss": 11.8818,
      "step": 32921
    },
    {
      "epoch": 1.7927344755059118,
      "grad_norm": 0.7158473795371252,
      "learning_rate": 5.57946657751609e-06,
      "loss": 11.8053,
      "step": 32922
    },
    {
      "epoch": 1.7927889295024948,
      "grad_norm": 0.5643935947936998,
      "learning_rate": 5.576562586188694e-06,
      "loss": 11.7546,
      "step": 32923
    },
    {
      "epoch": 1.7928433834990778,
      "grad_norm": 0.5462105170578148,
      "learning_rate": 5.573659329113767e-06,
      "loss": 11.7147,
      "step": 32924
    },
    {
      "epoch": 1.7928978374956608,
      "grad_norm": 0.5741835540156626,
      "learning_rate": 5.570756806313926e-06,
      "loss": 11.9257,
      "step": 32925
    },
    {
      "epoch": 1.7929522914922438,
      "grad_norm": 0.5454007195822214,
      "learning_rate": 5.567855017811696e-06,
      "loss": 11.7167,
      "step": 32926
    },
    {
      "epoch": 1.7930067454888268,
      "grad_norm": 0.565170862047559,
      "learning_rate": 5.564953963629671e-06,
      "loss": 11.7812,
      "step": 32927
    },
    {
      "epoch": 1.7930611994854098,
      "grad_norm": 0.5751964823867437,
      "learning_rate": 5.562053643790411e-06,
      "loss": 11.8629,
      "step": 32928
    },
    {
      "epoch": 1.7931156534819928,
      "grad_norm": 0.6041347164794797,
      "learning_rate": 5.5591540583164406e-06,
      "loss": 11.8195,
      "step": 32929
    },
    {
      "epoch": 1.7931701074785757,
      "grad_norm": 0.542418557889861,
      "learning_rate": 5.556255207230343e-06,
      "loss": 11.802,
      "step": 32930
    },
    {
      "epoch": 1.7932245614751587,
      "grad_norm": 0.5974718855396798,
      "learning_rate": 5.553357090554623e-06,
      "loss": 11.7662,
      "step": 32931
    },
    {
      "epoch": 1.7932790154717417,
      "grad_norm": 0.49823910135020305,
      "learning_rate": 5.550459708311862e-06,
      "loss": 11.7352,
      "step": 32932
    },
    {
      "epoch": 1.7933334694683247,
      "grad_norm": 0.5628189693596657,
      "learning_rate": 5.547563060524541e-06,
      "loss": 11.711,
      "step": 32933
    },
    {
      "epoch": 1.7933879234649077,
      "grad_norm": 0.5985189818794429,
      "learning_rate": 5.54466714721521e-06,
      "loss": 11.6791,
      "step": 32934
    },
    {
      "epoch": 1.7934423774614907,
      "grad_norm": 0.549031945653635,
      "learning_rate": 5.541771968406406e-06,
      "loss": 11.8105,
      "step": 32935
    },
    {
      "epoch": 1.7934968314580737,
      "grad_norm": 0.585048829188477,
      "learning_rate": 5.538877524120611e-06,
      "loss": 11.8103,
      "step": 32936
    },
    {
      "epoch": 1.7935512854546567,
      "grad_norm": 0.5897362453287507,
      "learning_rate": 5.5359838143803635e-06,
      "loss": 11.8408,
      "step": 32937
    },
    {
      "epoch": 1.7936057394512397,
      "grad_norm": 0.519749619448922,
      "learning_rate": 5.533090839208133e-06,
      "loss": 11.714,
      "step": 32938
    },
    {
      "epoch": 1.793660193447823,
      "grad_norm": 0.5693986335915681,
      "learning_rate": 5.5301985986264234e-06,
      "loss": 11.7528,
      "step": 32939
    },
    {
      "epoch": 1.793714647444406,
      "grad_norm": 0.5171493502925723,
      "learning_rate": 5.527307092657741e-06,
      "loss": 11.6295,
      "step": 32940
    },
    {
      "epoch": 1.793769101440989,
      "grad_norm": 0.5438200903069883,
      "learning_rate": 5.5244163213245545e-06,
      "loss": 11.8122,
      "step": 32941
    },
    {
      "epoch": 1.7938235554375719,
      "grad_norm": 0.6501297040214571,
      "learning_rate": 5.521526284649359e-06,
      "loss": 11.8396,
      "step": 32942
    },
    {
      "epoch": 1.7938780094341549,
      "grad_norm": 0.5484339027683268,
      "learning_rate": 5.518636982654612e-06,
      "loss": 11.7793,
      "step": 32943
    },
    {
      "epoch": 1.7939324634307379,
      "grad_norm": 0.549753147222448,
      "learning_rate": 5.515748415362798e-06,
      "loss": 11.893,
      "step": 32944
    },
    {
      "epoch": 1.793986917427321,
      "grad_norm": 0.5140662256940166,
      "learning_rate": 5.512860582796353e-06,
      "loss": 11.7629,
      "step": 32945
    },
    {
      "epoch": 1.794041371423904,
      "grad_norm": 0.5795490634534858,
      "learning_rate": 5.5099734849777595e-06,
      "loss": 11.7468,
      "step": 32946
    },
    {
      "epoch": 1.794095825420487,
      "grad_norm": 0.5368867896303627,
      "learning_rate": 5.5070871219294776e-06,
      "loss": 11.808,
      "step": 32947
    },
    {
      "epoch": 1.79415027941707,
      "grad_norm": 0.5882965945355642,
      "learning_rate": 5.50420149367391e-06,
      "loss": 11.8415,
      "step": 32948
    },
    {
      "epoch": 1.794204733413653,
      "grad_norm": 0.5701701114734583,
      "learning_rate": 5.501316600233508e-06,
      "loss": 11.7892,
      "step": 32949
    },
    {
      "epoch": 1.794259187410236,
      "grad_norm": 0.5466941626380739,
      "learning_rate": 5.4984324416307405e-06,
      "loss": 11.8151,
      "step": 32950
    },
    {
      "epoch": 1.794313641406819,
      "grad_norm": 0.5454520733639197,
      "learning_rate": 5.495549017887991e-06,
      "loss": 11.7103,
      "step": 32951
    },
    {
      "epoch": 1.794368095403402,
      "grad_norm": 0.5774645190721264,
      "learning_rate": 5.492666329027718e-06,
      "loss": 11.7959,
      "step": 32952
    },
    {
      "epoch": 1.794422549399985,
      "grad_norm": 0.5382498296973779,
      "learning_rate": 5.4897843750723045e-06,
      "loss": 11.7658,
      "step": 32953
    },
    {
      "epoch": 1.794477003396568,
      "grad_norm": 0.5640751589546652,
      "learning_rate": 5.486903156044187e-06,
      "loss": 11.8879,
      "step": 32954
    },
    {
      "epoch": 1.794531457393151,
      "grad_norm": 0.5619623001192191,
      "learning_rate": 5.48402267196575e-06,
      "loss": 11.9502,
      "step": 32955
    },
    {
      "epoch": 1.794585911389734,
      "grad_norm": 0.6720658821772526,
      "learning_rate": 5.481142922859428e-06,
      "loss": 11.8907,
      "step": 32956
    },
    {
      "epoch": 1.794640365386317,
      "grad_norm": 0.5309193801337244,
      "learning_rate": 5.4782639087475714e-06,
      "loss": 11.8927,
      "step": 32957
    },
    {
      "epoch": 1.7946948193829,
      "grad_norm": 0.5060859659602768,
      "learning_rate": 5.475385629652585e-06,
      "loss": 11.7418,
      "step": 32958
    },
    {
      "epoch": 1.794749273379483,
      "grad_norm": 0.5500296358704544,
      "learning_rate": 5.472508085596861e-06,
      "loss": 11.5995,
      "step": 32959
    },
    {
      "epoch": 1.794803727376066,
      "grad_norm": 0.5415131342335241,
      "learning_rate": 5.469631276602749e-06,
      "loss": 11.7777,
      "step": 32960
    },
    {
      "epoch": 1.794858181372649,
      "grad_norm": 0.4773191893379602,
      "learning_rate": 5.4667552026926415e-06,
      "loss": 11.7708,
      "step": 32961
    },
    {
      "epoch": 1.7949126353692322,
      "grad_norm": 0.6345859726488672,
      "learning_rate": 5.46387986388891e-06,
      "loss": 11.9852,
      "step": 32962
    },
    {
      "epoch": 1.7949670893658152,
      "grad_norm": 0.5511800044115259,
      "learning_rate": 5.461005260213892e-06,
      "loss": 11.807,
      "step": 32963
    },
    {
      "epoch": 1.7950215433623982,
      "grad_norm": 0.6016446725644914,
      "learning_rate": 5.458131391689958e-06,
      "loss": 11.8363,
      "step": 32964
    },
    {
      "epoch": 1.7950759973589812,
      "grad_norm": 0.5373914682548222,
      "learning_rate": 5.455258258339446e-06,
      "loss": 11.782,
      "step": 32965
    },
    {
      "epoch": 1.7951304513555641,
      "grad_norm": 0.49659643496513534,
      "learning_rate": 5.452385860184705e-06,
      "loss": 11.7047,
      "step": 32966
    },
    {
      "epoch": 1.7951849053521471,
      "grad_norm": 0.6386390216313624,
      "learning_rate": 5.449514197248051e-06,
      "loss": 11.6199,
      "step": 32967
    },
    {
      "epoch": 1.7952393593487304,
      "grad_norm": 0.6626104470852224,
      "learning_rate": 5.4466432695518545e-06,
      "loss": 11.8896,
      "step": 32968
    },
    {
      "epoch": 1.7952938133453133,
      "grad_norm": 0.5813216714212438,
      "learning_rate": 5.443773077118419e-06,
      "loss": 11.8182,
      "step": 32969
    },
    {
      "epoch": 1.7953482673418963,
      "grad_norm": 0.5478349669179279,
      "learning_rate": 5.4409036199700395e-06,
      "loss": 11.8171,
      "step": 32970
    },
    {
      "epoch": 1.7954027213384793,
      "grad_norm": 0.55359895922794,
      "learning_rate": 5.438034898129063e-06,
      "loss": 11.6894,
      "step": 32971
    },
    {
      "epoch": 1.7954571753350623,
      "grad_norm": 0.5285163082513964,
      "learning_rate": 5.4351669116177725e-06,
      "loss": 11.7452,
      "step": 32972
    },
    {
      "epoch": 1.7955116293316453,
      "grad_norm": 0.5531286289009327,
      "learning_rate": 5.432299660458484e-06,
      "loss": 11.7209,
      "step": 32973
    },
    {
      "epoch": 1.7955660833282283,
      "grad_norm": 0.533561977470842,
      "learning_rate": 5.429433144673512e-06,
      "loss": 11.7901,
      "step": 32974
    },
    {
      "epoch": 1.7956205373248113,
      "grad_norm": 0.5161039121828637,
      "learning_rate": 5.426567364285107e-06,
      "loss": 11.6132,
      "step": 32975
    },
    {
      "epoch": 1.7956749913213943,
      "grad_norm": 0.5068217370561073,
      "learning_rate": 5.423702319315593e-06,
      "loss": 11.7355,
      "step": 32976
    },
    {
      "epoch": 1.7957294453179773,
      "grad_norm": 0.6415120804812983,
      "learning_rate": 5.42083800978721e-06,
      "loss": 11.9336,
      "step": 32977
    },
    {
      "epoch": 1.7957838993145603,
      "grad_norm": 0.526911813800498,
      "learning_rate": 5.417974435722273e-06,
      "loss": 11.7279,
      "step": 32978
    },
    {
      "epoch": 1.7958383533111433,
      "grad_norm": 0.4895660059007519,
      "learning_rate": 5.415111597143019e-06,
      "loss": 11.8068,
      "step": 32979
    },
    {
      "epoch": 1.7958928073077263,
      "grad_norm": 0.5158891977356518,
      "learning_rate": 5.412249494071709e-06,
      "loss": 11.8079,
      "step": 32980
    },
    {
      "epoch": 1.7959472613043093,
      "grad_norm": 0.53320740878039,
      "learning_rate": 5.4093881265306235e-06,
      "loss": 11.7425,
      "step": 32981
    },
    {
      "epoch": 1.7960017153008923,
      "grad_norm": 0.5302737779417936,
      "learning_rate": 5.406527494541991e-06,
      "loss": 11.7888,
      "step": 32982
    },
    {
      "epoch": 1.7960561692974752,
      "grad_norm": 0.5772101317185414,
      "learning_rate": 5.4036675981280485e-06,
      "loss": 11.8371,
      "step": 32983
    },
    {
      "epoch": 1.7961106232940582,
      "grad_norm": 0.5354878640147748,
      "learning_rate": 5.400808437311078e-06,
      "loss": 11.9267,
      "step": 32984
    },
    {
      "epoch": 1.7961650772906415,
      "grad_norm": 0.5921124983555275,
      "learning_rate": 5.397950012113273e-06,
      "loss": 11.8837,
      "step": 32985
    },
    {
      "epoch": 1.7962195312872244,
      "grad_norm": 0.5211389425899624,
      "learning_rate": 5.395092322556883e-06,
      "loss": 11.8441,
      "step": 32986
    },
    {
      "epoch": 1.7962739852838074,
      "grad_norm": 0.5216618274321483,
      "learning_rate": 5.3922353686641e-06,
      "loss": 11.6793,
      "step": 32987
    },
    {
      "epoch": 1.7963284392803904,
      "grad_norm": 0.5172682682443953,
      "learning_rate": 5.389379150457186e-06,
      "loss": 11.7679,
      "step": 32988
    },
    {
      "epoch": 1.7963828932769734,
      "grad_norm": 0.4961430726988221,
      "learning_rate": 5.38652366795831e-06,
      "loss": 11.7149,
      "step": 32989
    },
    {
      "epoch": 1.7964373472735564,
      "grad_norm": 0.6063725149669045,
      "learning_rate": 5.383668921189689e-06,
      "loss": 11.8792,
      "step": 32990
    },
    {
      "epoch": 1.7964918012701396,
      "grad_norm": 0.5702487715606085,
      "learning_rate": 5.380814910173548e-06,
      "loss": 11.7312,
      "step": 32991
    },
    {
      "epoch": 1.7965462552667226,
      "grad_norm": 0.5801547243275779,
      "learning_rate": 5.377961634932027e-06,
      "loss": 11.7412,
      "step": 32992
    },
    {
      "epoch": 1.7966007092633056,
      "grad_norm": 0.5527701141612907,
      "learning_rate": 5.375109095487374e-06,
      "loss": 11.7354,
      "step": 32993
    },
    {
      "epoch": 1.7966551632598886,
      "grad_norm": 0.5735356711884958,
      "learning_rate": 5.372257291861715e-06,
      "loss": 11.8257,
      "step": 32994
    },
    {
      "epoch": 1.7967096172564716,
      "grad_norm": 0.5822056748790422,
      "learning_rate": 5.369406224077256e-06,
      "loss": 11.6295,
      "step": 32995
    },
    {
      "epoch": 1.7967640712530546,
      "grad_norm": 0.5841452217221534,
      "learning_rate": 5.366555892156178e-06,
      "loss": 11.8433,
      "step": 32996
    },
    {
      "epoch": 1.7968185252496376,
      "grad_norm": 0.5432889074077022,
      "learning_rate": 5.363706296120618e-06,
      "loss": 11.7912,
      "step": 32997
    },
    {
      "epoch": 1.7968729792462206,
      "grad_norm": 0.5221669384389692,
      "learning_rate": 5.360857435992772e-06,
      "loss": 11.8043,
      "step": 32998
    },
    {
      "epoch": 1.7969274332428036,
      "grad_norm": 0.5864699388637166,
      "learning_rate": 5.358009311794754e-06,
      "loss": 11.8788,
      "step": 32999
    },
    {
      "epoch": 1.7969818872393866,
      "grad_norm": 0.5059116582391517,
      "learning_rate": 5.355161923548724e-06,
      "loss": 11.686,
      "step": 33000
    },
    {
      "epoch": 1.7970363412359696,
      "grad_norm": 0.5752081796614689,
      "learning_rate": 5.352315271276831e-06,
      "loss": 11.7916,
      "step": 33001
    },
    {
      "epoch": 1.7970907952325526,
      "grad_norm": 0.5372043231782221,
      "learning_rate": 5.349469355001202e-06,
      "loss": 11.8713,
      "step": 33002
    },
    {
      "epoch": 1.7971452492291355,
      "grad_norm": 0.5803842771590753,
      "learning_rate": 5.346624174743986e-06,
      "loss": 11.9202,
      "step": 33003
    },
    {
      "epoch": 1.7971997032257185,
      "grad_norm": 0.5032991441482636,
      "learning_rate": 5.343779730527277e-06,
      "loss": 11.6306,
      "step": 33004
    },
    {
      "epoch": 1.7972541572223015,
      "grad_norm": 0.5803015960911285,
      "learning_rate": 5.340936022373222e-06,
      "loss": 11.8288,
      "step": 33005
    },
    {
      "epoch": 1.7973086112188845,
      "grad_norm": 0.5305174714243117,
      "learning_rate": 5.338093050303905e-06,
      "loss": 11.6913,
      "step": 33006
    },
    {
      "epoch": 1.7973630652154675,
      "grad_norm": 0.6319059823594511,
      "learning_rate": 5.335250814341464e-06,
      "loss": 12.0044,
      "step": 33007
    },
    {
      "epoch": 1.7974175192120505,
      "grad_norm": 0.5374725158357982,
      "learning_rate": 5.332409314508003e-06,
      "loss": 11.6781,
      "step": 33008
    },
    {
      "epoch": 1.7974719732086337,
      "grad_norm": 0.5578158580914994,
      "learning_rate": 5.329568550825581e-06,
      "loss": 11.8052,
      "step": 33009
    },
    {
      "epoch": 1.7975264272052167,
      "grad_norm": 0.5957287702480674,
      "learning_rate": 5.3267285233163045e-06,
      "loss": 11.8027,
      "step": 33010
    },
    {
      "epoch": 1.7975808812017997,
      "grad_norm": 0.4860268760649681,
      "learning_rate": 5.3238892320022886e-06,
      "loss": 11.8393,
      "step": 33011
    },
    {
      "epoch": 1.7976353351983827,
      "grad_norm": 0.5436182457278895,
      "learning_rate": 5.3210506769055705e-06,
      "loss": 11.7452,
      "step": 33012
    },
    {
      "epoch": 1.7976897891949657,
      "grad_norm": 0.5570223204020921,
      "learning_rate": 5.318212858048244e-06,
      "loss": 11.787,
      "step": 33013
    },
    {
      "epoch": 1.7977442431915487,
      "grad_norm": 0.5986976690922344,
      "learning_rate": 5.315375775452369e-06,
      "loss": 11.8338,
      "step": 33014
    },
    {
      "epoch": 1.797798697188132,
      "grad_norm": 0.551169818946494,
      "learning_rate": 5.312539429140018e-06,
      "loss": 11.8283,
      "step": 33015
    },
    {
      "epoch": 1.797853151184715,
      "grad_norm": 0.6176384875769115,
      "learning_rate": 5.309703819133238e-06,
      "loss": 11.869,
      "step": 33016
    },
    {
      "epoch": 1.7979076051812979,
      "grad_norm": 0.5242391249554645,
      "learning_rate": 5.306868945454068e-06,
      "loss": 11.8052,
      "step": 33017
    },
    {
      "epoch": 1.7979620591778809,
      "grad_norm": 0.5367222734114467,
      "learning_rate": 5.304034808124591e-06,
      "loss": 11.7946,
      "step": 33018
    },
    {
      "epoch": 1.7980165131744639,
      "grad_norm": 0.5272415705511996,
      "learning_rate": 5.3012014071668e-06,
      "loss": 11.882,
      "step": 33019
    },
    {
      "epoch": 1.7980709671710469,
      "grad_norm": 0.5580051244643481,
      "learning_rate": 5.298368742602766e-06,
      "loss": 11.6822,
      "step": 33020
    },
    {
      "epoch": 1.7981254211676299,
      "grad_norm": 0.6239567137405468,
      "learning_rate": 5.295536814454472e-06,
      "loss": 11.8083,
      "step": 33021
    },
    {
      "epoch": 1.7981798751642128,
      "grad_norm": 0.5463087922129438,
      "learning_rate": 5.292705622743977e-06,
      "loss": 11.5613,
      "step": 33022
    },
    {
      "epoch": 1.7982343291607958,
      "grad_norm": 0.5552737913614896,
      "learning_rate": 5.289875167493286e-06,
      "loss": 11.7769,
      "step": 33023
    },
    {
      "epoch": 1.7982887831573788,
      "grad_norm": 0.5147142660490316,
      "learning_rate": 5.287045448724404e-06,
      "loss": 11.8173,
      "step": 33024
    },
    {
      "epoch": 1.7983432371539618,
      "grad_norm": 0.5788489928901949,
      "learning_rate": 5.284216466459357e-06,
      "loss": 11.7764,
      "step": 33025
    },
    {
      "epoch": 1.7983976911505448,
      "grad_norm": 0.5810937995182615,
      "learning_rate": 5.281388220720107e-06,
      "loss": 11.8385,
      "step": 33026
    },
    {
      "epoch": 1.7984521451471278,
      "grad_norm": 0.541854898260314,
      "learning_rate": 5.27856071152868e-06,
      "loss": 11.7527,
      "step": 33027
    },
    {
      "epoch": 1.7985065991437108,
      "grad_norm": 0.4937860521646309,
      "learning_rate": 5.275733938907046e-06,
      "loss": 11.7261,
      "step": 33028
    },
    {
      "epoch": 1.7985610531402938,
      "grad_norm": 0.5547554882226353,
      "learning_rate": 5.272907902877189e-06,
      "loss": 11.7142,
      "step": 33029
    },
    {
      "epoch": 1.7986155071368768,
      "grad_norm": 0.544133630270464,
      "learning_rate": 5.27008260346109e-06,
      "loss": 11.8562,
      "step": 33030
    },
    {
      "epoch": 1.7986699611334598,
      "grad_norm": 0.5407408427436463,
      "learning_rate": 5.2672580406807e-06,
      "loss": 11.8199,
      "step": 33031
    },
    {
      "epoch": 1.798724415130043,
      "grad_norm": 0.5375801738999948,
      "learning_rate": 5.2644342145580005e-06,
      "loss": 11.7831,
      "step": 33032
    },
    {
      "epoch": 1.798778869126626,
      "grad_norm": 0.6789230656775038,
      "learning_rate": 5.261611125114963e-06,
      "loss": 11.8353,
      "step": 33033
    },
    {
      "epoch": 1.798833323123209,
      "grad_norm": 0.6411747950919314,
      "learning_rate": 5.258788772373513e-06,
      "loss": 11.7864,
      "step": 33034
    },
    {
      "epoch": 1.798887777119792,
      "grad_norm": 0.510044093749449,
      "learning_rate": 5.255967156355623e-06,
      "loss": 11.863,
      "step": 33035
    },
    {
      "epoch": 1.798942231116375,
      "grad_norm": 0.5360965215681892,
      "learning_rate": 5.253146277083199e-06,
      "loss": 11.6881,
      "step": 33036
    },
    {
      "epoch": 1.798996685112958,
      "grad_norm": 0.5428324202304564,
      "learning_rate": 5.250326134578221e-06,
      "loss": 11.6235,
      "step": 33037
    },
    {
      "epoch": 1.7990511391095412,
      "grad_norm": 0.49448376491651336,
      "learning_rate": 5.247506728862595e-06,
      "loss": 11.7862,
      "step": 33038
    },
    {
      "epoch": 1.7991055931061242,
      "grad_norm": 0.5633106550018803,
      "learning_rate": 5.244688059958225e-06,
      "loss": 11.8028,
      "step": 33039
    },
    {
      "epoch": 1.7991600471027072,
      "grad_norm": 0.6164474620025211,
      "learning_rate": 5.241870127887072e-06,
      "loss": 11.7571,
      "step": 33040
    },
    {
      "epoch": 1.7992145010992902,
      "grad_norm": 0.556047598832963,
      "learning_rate": 5.239052932671018e-06,
      "loss": 11.7722,
      "step": 33041
    },
    {
      "epoch": 1.7992689550958731,
      "grad_norm": 0.561233804075412,
      "learning_rate": 5.23623647433199e-06,
      "loss": 11.833,
      "step": 33042
    },
    {
      "epoch": 1.7993234090924561,
      "grad_norm": 0.5363343879344251,
      "learning_rate": 5.23342075289186e-06,
      "loss": 11.8106,
      "step": 33043
    },
    {
      "epoch": 1.7993778630890391,
      "grad_norm": 0.5180780725819338,
      "learning_rate": 5.230605768372554e-06,
      "loss": 11.7159,
      "step": 33044
    },
    {
      "epoch": 1.7994323170856221,
      "grad_norm": 0.5917444056953549,
      "learning_rate": 5.227791520795955e-06,
      "loss": 11.8704,
      "step": 33045
    },
    {
      "epoch": 1.7994867710822051,
      "grad_norm": 0.6058205122319649,
      "learning_rate": 5.2249780101839345e-06,
      "loss": 11.776,
      "step": 33046
    },
    {
      "epoch": 1.799541225078788,
      "grad_norm": 0.5910864790269835,
      "learning_rate": 5.222165236558385e-06,
      "loss": 11.8462,
      "step": 33047
    },
    {
      "epoch": 1.799595679075371,
      "grad_norm": 0.5409314574371767,
      "learning_rate": 5.2193531999411795e-06,
      "loss": 11.8283,
      "step": 33048
    },
    {
      "epoch": 1.799650133071954,
      "grad_norm": 0.703583447428384,
      "learning_rate": 5.2165419003541545e-06,
      "loss": 11.8901,
      "step": 33049
    },
    {
      "epoch": 1.799704587068537,
      "grad_norm": 0.5390945416622668,
      "learning_rate": 5.213731337819217e-06,
      "loss": 11.7955,
      "step": 33050
    },
    {
      "epoch": 1.79975904106512,
      "grad_norm": 0.49991864252572077,
      "learning_rate": 5.210921512358191e-06,
      "loss": 11.7977,
      "step": 33051
    },
    {
      "epoch": 1.799813495061703,
      "grad_norm": 0.5359739836329991,
      "learning_rate": 5.2081124239929395e-06,
      "loss": 11.7865,
      "step": 33052
    },
    {
      "epoch": 1.799867949058286,
      "grad_norm": 0.6323550774870728,
      "learning_rate": 5.205304072745299e-06,
      "loss": 11.9618,
      "step": 33053
    },
    {
      "epoch": 1.799922403054869,
      "grad_norm": 0.5009089060164308,
      "learning_rate": 5.202496458637118e-06,
      "loss": 11.7868,
      "step": 33054
    },
    {
      "epoch": 1.7999768570514523,
      "grad_norm": 0.5397962049504392,
      "learning_rate": 5.199689581690204e-06,
      "loss": 11.7871,
      "step": 33055
    },
    {
      "epoch": 1.8000313110480353,
      "grad_norm": 0.7182613204472554,
      "learning_rate": 5.196883441926415e-06,
      "loss": 11.7472,
      "step": 33056
    },
    {
      "epoch": 1.8000857650446183,
      "grad_norm": 0.5485250475107362,
      "learning_rate": 5.194078039367556e-06,
      "loss": 11.6487,
      "step": 33057
    },
    {
      "epoch": 1.8001402190412013,
      "grad_norm": 0.5951127871722497,
      "learning_rate": 5.191273374035455e-06,
      "loss": 11.9036,
      "step": 33058
    },
    {
      "epoch": 1.8001946730377842,
      "grad_norm": 0.5071896827292842,
      "learning_rate": 5.1884694459519045e-06,
      "loss": 11.6455,
      "step": 33059
    },
    {
      "epoch": 1.8002491270343672,
      "grad_norm": 0.5661822993466582,
      "learning_rate": 5.185666255138721e-06,
      "loss": 11.8508,
      "step": 33060
    },
    {
      "epoch": 1.8003035810309505,
      "grad_norm": 0.5037982708150142,
      "learning_rate": 5.182863801617677e-06,
      "loss": 11.8576,
      "step": 33061
    },
    {
      "epoch": 1.8003580350275334,
      "grad_norm": 0.5182435953116797,
      "learning_rate": 5.180062085410609e-06,
      "loss": 11.7623,
      "step": 33062
    },
    {
      "epoch": 1.8004124890241164,
      "grad_norm": 0.6543159068382637,
      "learning_rate": 5.177261106539255e-06,
      "loss": 11.9147,
      "step": 33063
    },
    {
      "epoch": 1.8004669430206994,
      "grad_norm": 0.573379508773424,
      "learning_rate": 5.174460865025443e-06,
      "loss": 11.6139,
      "step": 33064
    },
    {
      "epoch": 1.8005213970172824,
      "grad_norm": 0.5455433176042846,
      "learning_rate": 5.171661360890911e-06,
      "loss": 11.6488,
      "step": 33065
    },
    {
      "epoch": 1.8005758510138654,
      "grad_norm": 0.616186317030893,
      "learning_rate": 5.168862594157442e-06,
      "loss": 11.7815,
      "step": 33066
    },
    {
      "epoch": 1.8006303050104484,
      "grad_norm": 0.5576784896864911,
      "learning_rate": 5.166064564846818e-06,
      "loss": 11.7695,
      "step": 33067
    },
    {
      "epoch": 1.8006847590070314,
      "grad_norm": 0.558524115267277,
      "learning_rate": 5.163267272980776e-06,
      "loss": 11.9219,
      "step": 33068
    },
    {
      "epoch": 1.8007392130036144,
      "grad_norm": 0.5776225595460162,
      "learning_rate": 5.160470718581068e-06,
      "loss": 11.8872,
      "step": 33069
    },
    {
      "epoch": 1.8007936670001974,
      "grad_norm": 0.5695110058292474,
      "learning_rate": 5.157674901669441e-06,
      "loss": 11.6353,
      "step": 33070
    },
    {
      "epoch": 1.8008481209967804,
      "grad_norm": 0.68544646777409,
      "learning_rate": 5.154879822267633e-06,
      "loss": 11.8049,
      "step": 33071
    },
    {
      "epoch": 1.8009025749933634,
      "grad_norm": 0.7136796969335726,
      "learning_rate": 5.152085480397395e-06,
      "loss": 11.7655,
      "step": 33072
    },
    {
      "epoch": 1.8009570289899464,
      "grad_norm": 0.5283364656274697,
      "learning_rate": 5.149291876080431e-06,
      "loss": 11.7,
      "step": 33073
    },
    {
      "epoch": 1.8010114829865294,
      "grad_norm": 0.5487399564934956,
      "learning_rate": 5.146499009338501e-06,
      "loss": 11.7431,
      "step": 33074
    },
    {
      "epoch": 1.8010659369831123,
      "grad_norm": 0.573375986776589,
      "learning_rate": 5.143706880193289e-06,
      "loss": 11.6611,
      "step": 33075
    },
    {
      "epoch": 1.8011203909796953,
      "grad_norm": 0.5127462304835643,
      "learning_rate": 5.1409154886665315e-06,
      "loss": 11.7831,
      "step": 33076
    },
    {
      "epoch": 1.8011748449762783,
      "grad_norm": 0.5341124749966027,
      "learning_rate": 5.138124834779901e-06,
      "loss": 11.7424,
      "step": 33077
    },
    {
      "epoch": 1.8012292989728613,
      "grad_norm": 0.558461180509717,
      "learning_rate": 5.135334918555146e-06,
      "loss": 11.5907,
      "step": 33078
    },
    {
      "epoch": 1.8012837529694445,
      "grad_norm": 0.5695077286896717,
      "learning_rate": 5.132545740013928e-06,
      "loss": 11.812,
      "step": 33079
    },
    {
      "epoch": 1.8013382069660275,
      "grad_norm": 0.6236338186733865,
      "learning_rate": 5.129757299177928e-06,
      "loss": 11.8106,
      "step": 33080
    },
    {
      "epoch": 1.8013926609626105,
      "grad_norm": 0.6048148154238496,
      "learning_rate": 5.126969596068853e-06,
      "loss": 11.9336,
      "step": 33081
    },
    {
      "epoch": 1.8014471149591935,
      "grad_norm": 0.6021874516734276,
      "learning_rate": 5.124182630708385e-06,
      "loss": 11.8336,
      "step": 33082
    },
    {
      "epoch": 1.8015015689557765,
      "grad_norm": 0.5510384474859819,
      "learning_rate": 5.1213964031181615e-06,
      "loss": 11.8096,
      "step": 33083
    },
    {
      "epoch": 1.8015560229523597,
      "grad_norm": 0.5608582089557207,
      "learning_rate": 5.118610913319888e-06,
      "loss": 11.8161,
      "step": 33084
    },
    {
      "epoch": 1.8016104769489427,
      "grad_norm": 0.5780941252311154,
      "learning_rate": 5.115826161335202e-06,
      "loss": 11.7778,
      "step": 33085
    },
    {
      "epoch": 1.8016649309455257,
      "grad_norm": 0.5049369266089829,
      "learning_rate": 5.113042147185765e-06,
      "loss": 11.7887,
      "step": 33086
    },
    {
      "epoch": 1.8017193849421087,
      "grad_norm": 0.572217096407932,
      "learning_rate": 5.110258870893225e-06,
      "loss": 11.8803,
      "step": 33087
    },
    {
      "epoch": 1.8017738389386917,
      "grad_norm": 0.5332739464871372,
      "learning_rate": 5.1074763324792215e-06,
      "loss": 11.8783,
      "step": 33088
    },
    {
      "epoch": 1.8018282929352747,
      "grad_norm": 0.584070318484261,
      "learning_rate": 5.104694531965415e-06,
      "loss": 11.7572,
      "step": 33089
    },
    {
      "epoch": 1.8018827469318577,
      "grad_norm": 0.5883338984393851,
      "learning_rate": 5.101913469373387e-06,
      "loss": 11.7152,
      "step": 33090
    },
    {
      "epoch": 1.8019372009284407,
      "grad_norm": 0.5086451157941407,
      "learning_rate": 5.099133144724821e-06,
      "loss": 11.7435,
      "step": 33091
    },
    {
      "epoch": 1.8019916549250237,
      "grad_norm": 0.5571779434488101,
      "learning_rate": 5.096353558041289e-06,
      "loss": 11.8092,
      "step": 33092
    },
    {
      "epoch": 1.8020461089216067,
      "grad_norm": 0.5312922849609283,
      "learning_rate": 5.0935747093444285e-06,
      "loss": 11.6488,
      "step": 33093
    },
    {
      "epoch": 1.8021005629181897,
      "grad_norm": 0.5166841817620795,
      "learning_rate": 5.090796598655867e-06,
      "loss": 11.718,
      "step": 33094
    },
    {
      "epoch": 1.8021550169147726,
      "grad_norm": 0.5906951796143799,
      "learning_rate": 5.0880192259971645e-06,
      "loss": 11.7187,
      "step": 33095
    },
    {
      "epoch": 1.8022094709113556,
      "grad_norm": 0.4857076515708823,
      "learning_rate": 5.0852425913899605e-06,
      "loss": 11.7128,
      "step": 33096
    },
    {
      "epoch": 1.8022639249079386,
      "grad_norm": 0.548613000878737,
      "learning_rate": 5.082466694855803e-06,
      "loss": 11.8502,
      "step": 33097
    },
    {
      "epoch": 1.8023183789045216,
      "grad_norm": 0.5440445244085107,
      "learning_rate": 5.079691536416331e-06,
      "loss": 11.7426,
      "step": 33098
    },
    {
      "epoch": 1.8023728329011046,
      "grad_norm": 0.5765165755015134,
      "learning_rate": 5.0769171160930824e-06,
      "loss": 11.9014,
      "step": 33099
    },
    {
      "epoch": 1.8024272868976876,
      "grad_norm": 0.5193150501470704,
      "learning_rate": 5.074143433907641e-06,
      "loss": 11.7797,
      "step": 33100
    },
    {
      "epoch": 1.8024817408942706,
      "grad_norm": 0.6264881221743966,
      "learning_rate": 5.071370489881589e-06,
      "loss": 11.857,
      "step": 33101
    },
    {
      "epoch": 1.8025361948908538,
      "grad_norm": 0.5292029358386698,
      "learning_rate": 5.068598284036474e-06,
      "loss": 11.6965,
      "step": 33102
    },
    {
      "epoch": 1.8025906488874368,
      "grad_norm": 0.5577462168844834,
      "learning_rate": 5.065826816393848e-06,
      "loss": 11.718,
      "step": 33103
    },
    {
      "epoch": 1.8026451028840198,
      "grad_norm": 0.5124702474738054,
      "learning_rate": 5.063056086975293e-06,
      "loss": 11.7704,
      "step": 33104
    },
    {
      "epoch": 1.8026995568806028,
      "grad_norm": 0.5172760410745203,
      "learning_rate": 5.0602860958023136e-06,
      "loss": 11.7581,
      "step": 33105
    },
    {
      "epoch": 1.8027540108771858,
      "grad_norm": 0.5681498001389979,
      "learning_rate": 5.057516842896492e-06,
      "loss": 11.7469,
      "step": 33106
    },
    {
      "epoch": 1.8028084648737688,
      "grad_norm": 0.553499130275587,
      "learning_rate": 5.054748328279324e-06,
      "loss": 11.725,
      "step": 33107
    },
    {
      "epoch": 1.802862918870352,
      "grad_norm": 0.5267487531792995,
      "learning_rate": 5.051980551972369e-06,
      "loss": 11.8036,
      "step": 33108
    },
    {
      "epoch": 1.802917372866935,
      "grad_norm": 0.5835299047872015,
      "learning_rate": 5.049213513997142e-06,
      "loss": 11.7708,
      "step": 33109
    },
    {
      "epoch": 1.802971826863518,
      "grad_norm": 0.564457165664405,
      "learning_rate": 5.046447214375138e-06,
      "loss": 11.9303,
      "step": 33110
    },
    {
      "epoch": 1.803026280860101,
      "grad_norm": 0.5626347562366967,
      "learning_rate": 5.043681653127885e-06,
      "loss": 11.5876,
      "step": 33111
    },
    {
      "epoch": 1.803080734856684,
      "grad_norm": 0.5262466057129509,
      "learning_rate": 5.040916830276887e-06,
      "loss": 11.8207,
      "step": 33112
    },
    {
      "epoch": 1.803135188853267,
      "grad_norm": 0.5368449488581447,
      "learning_rate": 5.03815274584366e-06,
      "loss": 11.7659,
      "step": 33113
    },
    {
      "epoch": 1.80318964284985,
      "grad_norm": 0.6010632105142093,
      "learning_rate": 5.035389399849666e-06,
      "loss": 11.9371,
      "step": 33114
    },
    {
      "epoch": 1.803244096846433,
      "grad_norm": 0.5916678779008685,
      "learning_rate": 5.032626792316408e-06,
      "loss": 11.7108,
      "step": 33115
    },
    {
      "epoch": 1.803298550843016,
      "grad_norm": 0.5369634232869531,
      "learning_rate": 5.029864923265382e-06,
      "loss": 11.8294,
      "step": 33116
    },
    {
      "epoch": 1.803353004839599,
      "grad_norm": 0.5089388708774236,
      "learning_rate": 5.027103792718036e-06,
      "loss": 11.7105,
      "step": 33117
    },
    {
      "epoch": 1.803407458836182,
      "grad_norm": 0.5112987221801994,
      "learning_rate": 5.024343400695874e-06,
      "loss": 11.6755,
      "step": 33118
    },
    {
      "epoch": 1.803461912832765,
      "grad_norm": 0.5754809501452369,
      "learning_rate": 5.021583747220349e-06,
      "loss": 11.843,
      "step": 33119
    },
    {
      "epoch": 1.803516366829348,
      "grad_norm": 0.569141860179444,
      "learning_rate": 5.018824832312907e-06,
      "loss": 11.7935,
      "step": 33120
    },
    {
      "epoch": 1.803570820825931,
      "grad_norm": 0.5524580090942568,
      "learning_rate": 5.016066655995022e-06,
      "loss": 11.7522,
      "step": 33121
    },
    {
      "epoch": 1.803625274822514,
      "grad_norm": 0.5185444764335322,
      "learning_rate": 5.013309218288109e-06,
      "loss": 11.7835,
      "step": 33122
    },
    {
      "epoch": 1.8036797288190969,
      "grad_norm": 0.5046861956728936,
      "learning_rate": 5.0105525192136515e-06,
      "loss": 11.6846,
      "step": 33123
    },
    {
      "epoch": 1.8037341828156799,
      "grad_norm": 0.549262063097627,
      "learning_rate": 5.007796558793054e-06,
      "loss": 11.7959,
      "step": 33124
    },
    {
      "epoch": 1.803788636812263,
      "grad_norm": 0.5706250564929886,
      "learning_rate": 5.005041337047778e-06,
      "loss": 11.6666,
      "step": 33125
    },
    {
      "epoch": 1.803843090808846,
      "grad_norm": 0.5280845153105259,
      "learning_rate": 5.002286853999216e-06,
      "loss": 11.7222,
      "step": 33126
    },
    {
      "epoch": 1.803897544805429,
      "grad_norm": 0.6090800605198194,
      "learning_rate": 4.999533109668797e-06,
      "loss": 11.8456,
      "step": 33127
    },
    {
      "epoch": 1.803951998802012,
      "grad_norm": 0.5449261964947809,
      "learning_rate": 4.996780104077958e-06,
      "loss": 11.8051,
      "step": 33128
    },
    {
      "epoch": 1.804006452798595,
      "grad_norm": 0.5253676105361571,
      "learning_rate": 4.994027837248094e-06,
      "loss": 11.826,
      "step": 33129
    },
    {
      "epoch": 1.804060906795178,
      "grad_norm": 0.5381580407631475,
      "learning_rate": 4.991276309200588e-06,
      "loss": 11.8203,
      "step": 33130
    },
    {
      "epoch": 1.8041153607917613,
      "grad_norm": 0.5776018075832223,
      "learning_rate": 4.988525519956855e-06,
      "loss": 11.8983,
      "step": 33131
    },
    {
      "epoch": 1.8041698147883443,
      "grad_norm": 0.5193345683995854,
      "learning_rate": 4.985775469538268e-06,
      "loss": 11.7223,
      "step": 33132
    },
    {
      "epoch": 1.8042242687849273,
      "grad_norm": 0.5355267333610721,
      "learning_rate": 4.983026157966242e-06,
      "loss": 11.7857,
      "step": 33133
    },
    {
      "epoch": 1.8042787227815102,
      "grad_norm": 0.5441305285446537,
      "learning_rate": 4.980277585262128e-06,
      "loss": 11.8036,
      "step": 33134
    },
    {
      "epoch": 1.8043331767780932,
      "grad_norm": 0.557721790054812,
      "learning_rate": 4.977529751447318e-06,
      "loss": 11.7417,
      "step": 33135
    },
    {
      "epoch": 1.8043876307746762,
      "grad_norm": 0.5690619179987269,
      "learning_rate": 4.974782656543164e-06,
      "loss": 11.7234,
      "step": 33136
    },
    {
      "epoch": 1.8044420847712592,
      "grad_norm": 0.5062107400952486,
      "learning_rate": 4.9720363005710365e-06,
      "loss": 11.7422,
      "step": 33137
    },
    {
      "epoch": 1.8044965387678422,
      "grad_norm": 0.5485248627223632,
      "learning_rate": 4.969290683552297e-06,
      "loss": 11.6907,
      "step": 33138
    },
    {
      "epoch": 1.8045509927644252,
      "grad_norm": 0.5305898060488378,
      "learning_rate": 4.966545805508293e-06,
      "loss": 11.7244,
      "step": 33139
    },
    {
      "epoch": 1.8046054467610082,
      "grad_norm": 0.5429745541015821,
      "learning_rate": 4.963801666460377e-06,
      "loss": 11.7969,
      "step": 33140
    },
    {
      "epoch": 1.8046599007575912,
      "grad_norm": 0.592253568178117,
      "learning_rate": 4.961058266429852e-06,
      "loss": 11.6663,
      "step": 33141
    },
    {
      "epoch": 1.8047143547541742,
      "grad_norm": 0.5559710956921947,
      "learning_rate": 4.95831560543808e-06,
      "loss": 11.7606,
      "step": 33142
    },
    {
      "epoch": 1.8047688087507572,
      "grad_norm": 0.6063945203602211,
      "learning_rate": 4.955573683506387e-06,
      "loss": 11.7917,
      "step": 33143
    },
    {
      "epoch": 1.8048232627473402,
      "grad_norm": 0.5676914394759548,
      "learning_rate": 4.9528325006560905e-06,
      "loss": 11.7848,
      "step": 33144
    },
    {
      "epoch": 1.8048777167439232,
      "grad_norm": 0.5441310715019334,
      "learning_rate": 4.950092056908518e-06,
      "loss": 11.789,
      "step": 33145
    },
    {
      "epoch": 1.8049321707405062,
      "grad_norm": 0.562846309247268,
      "learning_rate": 4.947352352284962e-06,
      "loss": 11.8551,
      "step": 33146
    },
    {
      "epoch": 1.8049866247370892,
      "grad_norm": 0.5843194000520776,
      "learning_rate": 4.94461338680674e-06,
      "loss": 11.7459,
      "step": 33147
    },
    {
      "epoch": 1.8050410787336721,
      "grad_norm": 0.5762016328871316,
      "learning_rate": 4.941875160495135e-06,
      "loss": 11.8354,
      "step": 33148
    },
    {
      "epoch": 1.8050955327302554,
      "grad_norm": 0.5579528750347603,
      "learning_rate": 4.939137673371452e-06,
      "loss": 11.811,
      "step": 33149
    },
    {
      "epoch": 1.8051499867268384,
      "grad_norm": 0.5200221724236684,
      "learning_rate": 4.936400925456997e-06,
      "loss": 11.7344,
      "step": 33150
    },
    {
      "epoch": 1.8052044407234213,
      "grad_norm": 0.6106854710543946,
      "learning_rate": 4.933664916773007e-06,
      "loss": 11.8581,
      "step": 33151
    },
    {
      "epoch": 1.8052588947200043,
      "grad_norm": 0.554498844373309,
      "learning_rate": 4.930929647340776e-06,
      "loss": 11.8799,
      "step": 33152
    },
    {
      "epoch": 1.8053133487165873,
      "grad_norm": 0.5525246288148445,
      "learning_rate": 4.9281951171816e-06,
      "loss": 11.6564,
      "step": 33153
    },
    {
      "epoch": 1.8053678027131705,
      "grad_norm": 0.535704884413912,
      "learning_rate": 4.925461326316705e-06,
      "loss": 11.6345,
      "step": 33154
    },
    {
      "epoch": 1.8054222567097535,
      "grad_norm": 0.5484366647238916,
      "learning_rate": 4.922728274767374e-06,
      "loss": 11.9068,
      "step": 33155
    },
    {
      "epoch": 1.8054767107063365,
      "grad_norm": 0.5386424025094163,
      "learning_rate": 4.919995962554846e-06,
      "loss": 11.7773,
      "step": 33156
    },
    {
      "epoch": 1.8055311647029195,
      "grad_norm": 0.5393011097664493,
      "learning_rate": 4.9172643897003936e-06,
      "loss": 11.7834,
      "step": 33157
    },
    {
      "epoch": 1.8055856186995025,
      "grad_norm": 0.5000349905754421,
      "learning_rate": 4.9145335562252204e-06,
      "loss": 11.6803,
      "step": 33158
    },
    {
      "epoch": 1.8056400726960855,
      "grad_norm": 0.5346719670032192,
      "learning_rate": 4.911803462150588e-06,
      "loss": 11.715,
      "step": 33159
    },
    {
      "epoch": 1.8056945266926685,
      "grad_norm": 0.5394760021643212,
      "learning_rate": 4.9090741074977245e-06,
      "loss": 11.715,
      "step": 33160
    },
    {
      "epoch": 1.8057489806892515,
      "grad_norm": 0.5571987610832586,
      "learning_rate": 4.906345492287834e-06,
      "loss": 11.8169,
      "step": 33161
    },
    {
      "epoch": 1.8058034346858345,
      "grad_norm": 0.5534356781641624,
      "learning_rate": 4.903617616542156e-06,
      "loss": 11.6854,
      "step": 33162
    },
    {
      "epoch": 1.8058578886824175,
      "grad_norm": 0.5355276617833473,
      "learning_rate": 4.900890480281883e-06,
      "loss": 11.8339,
      "step": 33163
    },
    {
      "epoch": 1.8059123426790005,
      "grad_norm": 0.529139284611768,
      "learning_rate": 4.898164083528245e-06,
      "loss": 11.6885,
      "step": 33164
    },
    {
      "epoch": 1.8059667966755835,
      "grad_norm": 0.6012888361537658,
      "learning_rate": 4.895438426302435e-06,
      "loss": 11.7846,
      "step": 33165
    },
    {
      "epoch": 1.8060212506721665,
      "grad_norm": 0.5267517422127683,
      "learning_rate": 4.892713508625635e-06,
      "loss": 11.7775,
      "step": 33166
    },
    {
      "epoch": 1.8060757046687494,
      "grad_norm": 0.5625057355380759,
      "learning_rate": 4.8899893305190514e-06,
      "loss": 11.8971,
      "step": 33167
    },
    {
      "epoch": 1.8061301586653324,
      "grad_norm": 0.5216984893198304,
      "learning_rate": 4.887265892003856e-06,
      "loss": 11.816,
      "step": 33168
    },
    {
      "epoch": 1.8061846126619154,
      "grad_norm": 0.6246677665710533,
      "learning_rate": 4.884543193101232e-06,
      "loss": 11.8072,
      "step": 33169
    },
    {
      "epoch": 1.8062390666584984,
      "grad_norm": 0.5301392317448178,
      "learning_rate": 4.881821233832362e-06,
      "loss": 11.6522,
      "step": 33170
    },
    {
      "epoch": 1.8062935206550814,
      "grad_norm": 0.5864101392981528,
      "learning_rate": 4.879100014218385e-06,
      "loss": 11.7581,
      "step": 33171
    },
    {
      "epoch": 1.8063479746516646,
      "grad_norm": 0.6302461544781593,
      "learning_rate": 4.876379534280495e-06,
      "loss": 11.8697,
      "step": 33172
    },
    {
      "epoch": 1.8064024286482476,
      "grad_norm": 0.5126815062616424,
      "learning_rate": 4.873659794039809e-06,
      "loss": 11.8067,
      "step": 33173
    },
    {
      "epoch": 1.8064568826448306,
      "grad_norm": 0.5338329742520125,
      "learning_rate": 4.87094079351752e-06,
      "loss": 11.8254,
      "step": 33174
    },
    {
      "epoch": 1.8065113366414136,
      "grad_norm": 0.5558821238255813,
      "learning_rate": 4.868222532734734e-06,
      "loss": 11.8499,
      "step": 33175
    },
    {
      "epoch": 1.8065657906379966,
      "grad_norm": 0.5297772250345244,
      "learning_rate": 4.86550501171259e-06,
      "loss": 11.6535,
      "step": 33176
    },
    {
      "epoch": 1.8066202446345796,
      "grad_norm": 0.5301268819352307,
      "learning_rate": 4.862788230472259e-06,
      "loss": 11.5577,
      "step": 33177
    },
    {
      "epoch": 1.8066746986311628,
      "grad_norm": 0.5195226774698151,
      "learning_rate": 4.860072189034826e-06,
      "loss": 11.604,
      "step": 33178
    },
    {
      "epoch": 1.8067291526277458,
      "grad_norm": 0.5873862330329568,
      "learning_rate": 4.857356887421438e-06,
      "loss": 11.8314,
      "step": 33179
    },
    {
      "epoch": 1.8067836066243288,
      "grad_norm": 0.5355624033162096,
      "learning_rate": 4.854642325653202e-06,
      "loss": 11.5848,
      "step": 33180
    },
    {
      "epoch": 1.8068380606209118,
      "grad_norm": 0.5334199011435603,
      "learning_rate": 4.851928503751202e-06,
      "loss": 11.6847,
      "step": 33181
    },
    {
      "epoch": 1.8068925146174948,
      "grad_norm": 0.5424485365186009,
      "learning_rate": 4.849215421736586e-06,
      "loss": 11.7574,
      "step": 33182
    },
    {
      "epoch": 1.8069469686140778,
      "grad_norm": 0.5519686488107474,
      "learning_rate": 4.846503079630404e-06,
      "loss": 11.7667,
      "step": 33183
    },
    {
      "epoch": 1.8070014226106608,
      "grad_norm": 0.5372690316473078,
      "learning_rate": 4.843791477453785e-06,
      "loss": 11.7687,
      "step": 33184
    },
    {
      "epoch": 1.8070558766072438,
      "grad_norm": 0.5206480337445053,
      "learning_rate": 4.8410806152278e-06,
      "loss": 11.6573,
      "step": 33185
    },
    {
      "epoch": 1.8071103306038268,
      "grad_norm": 0.5600968336725373,
      "learning_rate": 4.838370492973521e-06,
      "loss": 11.7819,
      "step": 33186
    },
    {
      "epoch": 1.8071647846004097,
      "grad_norm": 0.5411745921831451,
      "learning_rate": 4.835661110712042e-06,
      "loss": 11.8349,
      "step": 33187
    },
    {
      "epoch": 1.8072192385969927,
      "grad_norm": 0.5557196344027634,
      "learning_rate": 4.832952468464413e-06,
      "loss": 11.7677,
      "step": 33188
    },
    {
      "epoch": 1.8072736925935757,
      "grad_norm": 0.5819464925228438,
      "learning_rate": 4.830244566251729e-06,
      "loss": 11.7541,
      "step": 33189
    },
    {
      "epoch": 1.8073281465901587,
      "grad_norm": 0.573373616125001,
      "learning_rate": 4.827537404095006e-06,
      "loss": 11.7272,
      "step": 33190
    },
    {
      "epoch": 1.8073826005867417,
      "grad_norm": 0.6061551048176003,
      "learning_rate": 4.824830982015305e-06,
      "loss": 11.8376,
      "step": 33191
    },
    {
      "epoch": 1.8074370545833247,
      "grad_norm": 0.5699583363847336,
      "learning_rate": 4.822125300033686e-06,
      "loss": 11.5983,
      "step": 33192
    },
    {
      "epoch": 1.8074915085799077,
      "grad_norm": 0.5608869196218322,
      "learning_rate": 4.819420358171178e-06,
      "loss": 11.8257,
      "step": 33193
    },
    {
      "epoch": 1.8075459625764907,
      "grad_norm": 0.5604810907891347,
      "learning_rate": 4.816716156448831e-06,
      "loss": 11.6896,
      "step": 33194
    },
    {
      "epoch": 1.807600416573074,
      "grad_norm": 0.5146847407936258,
      "learning_rate": 4.814012694887649e-06,
      "loss": 11.7387,
      "step": 33195
    },
    {
      "epoch": 1.807654870569657,
      "grad_norm": 0.506587606170563,
      "learning_rate": 4.811309973508682e-06,
      "loss": 11.7663,
      "step": 33196
    },
    {
      "epoch": 1.80770932456624,
      "grad_norm": 0.5397408178174148,
      "learning_rate": 4.808607992332914e-06,
      "loss": 11.8642,
      "step": 33197
    },
    {
      "epoch": 1.8077637785628229,
      "grad_norm": 0.541070921712911,
      "learning_rate": 4.805906751381373e-06,
      "loss": 11.7178,
      "step": 33198
    },
    {
      "epoch": 1.8078182325594059,
      "grad_norm": 0.5703390458843115,
      "learning_rate": 4.803206250675097e-06,
      "loss": 11.8087,
      "step": 33199
    },
    {
      "epoch": 1.8078726865559889,
      "grad_norm": 0.5797365627006066,
      "learning_rate": 4.800506490235013e-06,
      "loss": 11.7641,
      "step": 33200
    },
    {
      "epoch": 1.807927140552572,
      "grad_norm": 0.5177289704755285,
      "learning_rate": 4.797807470082172e-06,
      "loss": 11.7672,
      "step": 33201
    },
    {
      "epoch": 1.807981594549155,
      "grad_norm": 0.5838144505341726,
      "learning_rate": 4.795109190237557e-06,
      "loss": 11.8404,
      "step": 33202
    },
    {
      "epoch": 1.808036048545738,
      "grad_norm": 0.5445052167707678,
      "learning_rate": 4.792411650722117e-06,
      "loss": 11.8278,
      "step": 33203
    },
    {
      "epoch": 1.808090502542321,
      "grad_norm": 0.5908857135536393,
      "learning_rate": 4.78971485155687e-06,
      "loss": 11.8719,
      "step": 33204
    },
    {
      "epoch": 1.808144956538904,
      "grad_norm": 0.48831478538696144,
      "learning_rate": 4.787018792762743e-06,
      "loss": 11.7465,
      "step": 33205
    },
    {
      "epoch": 1.808199410535487,
      "grad_norm": 0.548509157319017,
      "learning_rate": 4.7843234743607525e-06,
      "loss": 11.6314,
      "step": 33206
    },
    {
      "epoch": 1.80825386453207,
      "grad_norm": 0.561310458372122,
      "learning_rate": 4.781628896371815e-06,
      "loss": 11.8118,
      "step": 33207
    },
    {
      "epoch": 1.808308318528653,
      "grad_norm": 0.5234777153094029,
      "learning_rate": 4.778935058816902e-06,
      "loss": 11.7851,
      "step": 33208
    },
    {
      "epoch": 1.808362772525236,
      "grad_norm": 0.6093699230494634,
      "learning_rate": 4.776241961716965e-06,
      "loss": 11.8338,
      "step": 33209
    },
    {
      "epoch": 1.808417226521819,
      "grad_norm": 0.5867070046528813,
      "learning_rate": 4.773549605092931e-06,
      "loss": 11.8577,
      "step": 33210
    },
    {
      "epoch": 1.808471680518402,
      "grad_norm": 0.5452169022255674,
      "learning_rate": 4.77085798896576e-06,
      "loss": 11.9047,
      "step": 33211
    },
    {
      "epoch": 1.808526134514985,
      "grad_norm": 0.5196413237297726,
      "learning_rate": 4.7681671133563476e-06,
      "loss": 11.5996,
      "step": 33212
    },
    {
      "epoch": 1.808580588511568,
      "grad_norm": 0.6086607823386528,
      "learning_rate": 4.765476978285633e-06,
      "loss": 11.8549,
      "step": 33213
    },
    {
      "epoch": 1.808635042508151,
      "grad_norm": 0.5089794525742093,
      "learning_rate": 4.762787583774564e-06,
      "loss": 11.678,
      "step": 33214
    },
    {
      "epoch": 1.808689496504734,
      "grad_norm": 0.574101671690218,
      "learning_rate": 4.760098929844003e-06,
      "loss": 11.6728,
      "step": 33215
    },
    {
      "epoch": 1.808743950501317,
      "grad_norm": 0.5626940699284111,
      "learning_rate": 4.757411016514912e-06,
      "loss": 11.8825,
      "step": 33216
    },
    {
      "epoch": 1.8087984044979,
      "grad_norm": 0.5615339223466838,
      "learning_rate": 4.754723843808151e-06,
      "loss": 11.8007,
      "step": 33217
    },
    {
      "epoch": 1.8088528584944832,
      "grad_norm": 0.5455314024715735,
      "learning_rate": 4.752037411744637e-06,
      "loss": 11.8554,
      "step": 33218
    },
    {
      "epoch": 1.8089073124910662,
      "grad_norm": 0.5402356367933193,
      "learning_rate": 4.749351720345252e-06,
      "loss": 11.8262,
      "step": 33219
    },
    {
      "epoch": 1.8089617664876492,
      "grad_norm": 0.5582377215152176,
      "learning_rate": 4.74666676963087e-06,
      "loss": 11.8435,
      "step": 33220
    },
    {
      "epoch": 1.8090162204842322,
      "grad_norm": 0.588828318117417,
      "learning_rate": 4.743982559622395e-06,
      "loss": 11.7713,
      "step": 33221
    },
    {
      "epoch": 1.8090706744808152,
      "grad_norm": 0.5557691878795636,
      "learning_rate": 4.741299090340678e-06,
      "loss": 11.8135,
      "step": 33222
    },
    {
      "epoch": 1.8091251284773981,
      "grad_norm": 0.6213764815908573,
      "learning_rate": 4.7386163618066026e-06,
      "loss": 11.8986,
      "step": 33223
    },
    {
      "epoch": 1.8091795824739814,
      "grad_norm": 0.5406688724320587,
      "learning_rate": 4.735934374041007e-06,
      "loss": 11.8455,
      "step": 33224
    },
    {
      "epoch": 1.8092340364705644,
      "grad_norm": 0.5716837753206888,
      "learning_rate": 4.733253127064763e-06,
      "loss": 11.7499,
      "step": 33225
    },
    {
      "epoch": 1.8092884904671473,
      "grad_norm": 0.5579331798827912,
      "learning_rate": 4.730572620898732e-06,
      "loss": 11.7505,
      "step": 33226
    },
    {
      "epoch": 1.8093429444637303,
      "grad_norm": 0.5274949203688987,
      "learning_rate": 4.727892855563731e-06,
      "loss": 11.5504,
      "step": 33227
    },
    {
      "epoch": 1.8093973984603133,
      "grad_norm": 0.5336049351114056,
      "learning_rate": 4.7252138310806324e-06,
      "loss": 11.8481,
      "step": 33228
    },
    {
      "epoch": 1.8094518524568963,
      "grad_norm": 0.6252957727507026,
      "learning_rate": 4.722535547470242e-06,
      "loss": 11.8298,
      "step": 33229
    },
    {
      "epoch": 1.8095063064534793,
      "grad_norm": 0.5727061752869124,
      "learning_rate": 4.719858004753375e-06,
      "loss": 11.8586,
      "step": 33230
    },
    {
      "epoch": 1.8095607604500623,
      "grad_norm": 0.5170299752091735,
      "learning_rate": 4.717181202950893e-06,
      "loss": 11.8325,
      "step": 33231
    },
    {
      "epoch": 1.8096152144466453,
      "grad_norm": 0.5299384660261783,
      "learning_rate": 4.71450514208357e-06,
      "loss": 11.7124,
      "step": 33232
    },
    {
      "epoch": 1.8096696684432283,
      "grad_norm": 0.5499786496689046,
      "learning_rate": 4.711829822172254e-06,
      "loss": 11.8386,
      "step": 33233
    },
    {
      "epoch": 1.8097241224398113,
      "grad_norm": 0.5593279097519885,
      "learning_rate": 4.709155243237706e-06,
      "loss": 11.8838,
      "step": 33234
    },
    {
      "epoch": 1.8097785764363943,
      "grad_norm": 0.5756440750572177,
      "learning_rate": 4.706481405300756e-06,
      "loss": 11.7977,
      "step": 33235
    },
    {
      "epoch": 1.8098330304329773,
      "grad_norm": 0.5310462041528016,
      "learning_rate": 4.703808308382196e-06,
      "loss": 11.7635,
      "step": 33236
    },
    {
      "epoch": 1.8098874844295603,
      "grad_norm": 0.5094785712825856,
      "learning_rate": 4.701135952502788e-06,
      "loss": 11.7599,
      "step": 33237
    },
    {
      "epoch": 1.8099419384261433,
      "grad_norm": 0.528188634335068,
      "learning_rate": 4.69846433768335e-06,
      "loss": 11.7453,
      "step": 33238
    },
    {
      "epoch": 1.8099963924227263,
      "grad_norm": 0.5171714685422232,
      "learning_rate": 4.695793463944631e-06,
      "loss": 11.6009,
      "step": 33239
    },
    {
      "epoch": 1.8100508464193092,
      "grad_norm": 0.5283440178362762,
      "learning_rate": 4.693123331307392e-06,
      "loss": 11.7741,
      "step": 33240
    },
    {
      "epoch": 1.8101053004158922,
      "grad_norm": 0.5691858803247282,
      "learning_rate": 4.690453939792427e-06,
      "loss": 11.7403,
      "step": 33241
    },
    {
      "epoch": 1.8101597544124755,
      "grad_norm": 0.575870998897387,
      "learning_rate": 4.687785289420454e-06,
      "loss": 11.7369,
      "step": 33242
    },
    {
      "epoch": 1.8102142084090584,
      "grad_norm": 0.583534904636657,
      "learning_rate": 4.6851173802122675e-06,
      "loss": 11.8699,
      "step": 33243
    },
    {
      "epoch": 1.8102686624056414,
      "grad_norm": 0.5322452685176067,
      "learning_rate": 4.6824502121885714e-06,
      "loss": 11.733,
      "step": 33244
    },
    {
      "epoch": 1.8103231164022244,
      "grad_norm": 0.5102198165469266,
      "learning_rate": 4.6797837853701394e-06,
      "loss": 11.7132,
      "step": 33245
    },
    {
      "epoch": 1.8103775703988074,
      "grad_norm": 0.5272270547549536,
      "learning_rate": 4.677118099777688e-06,
      "loss": 11.7703,
      "step": 33246
    },
    {
      "epoch": 1.8104320243953904,
      "grad_norm": 0.5301405200023217,
      "learning_rate": 4.674453155431946e-06,
      "loss": 11.7402,
      "step": 33247
    },
    {
      "epoch": 1.8104864783919736,
      "grad_norm": 0.5920261063214396,
      "learning_rate": 4.671788952353662e-06,
      "loss": 11.7978,
      "step": 33248
    },
    {
      "epoch": 1.8105409323885566,
      "grad_norm": 0.4839884667414224,
      "learning_rate": 4.669125490563531e-06,
      "loss": 11.7457,
      "step": 33249
    },
    {
      "epoch": 1.8105953863851396,
      "grad_norm": 0.6332621013155387,
      "learning_rate": 4.666462770082247e-06,
      "loss": 11.7544,
      "step": 33250
    },
    {
      "epoch": 1.8106498403817226,
      "grad_norm": 0.5678331737352504,
      "learning_rate": 4.663800790930561e-06,
      "loss": 11.7125,
      "step": 33251
    },
    {
      "epoch": 1.8107042943783056,
      "grad_norm": 0.5643788351233826,
      "learning_rate": 4.661139553129123e-06,
      "loss": 11.8097,
      "step": 33252
    },
    {
      "epoch": 1.8107587483748886,
      "grad_norm": 0.5241273300555964,
      "learning_rate": 4.658479056698672e-06,
      "loss": 11.6924,
      "step": 33253
    },
    {
      "epoch": 1.8108132023714716,
      "grad_norm": 0.521911469572689,
      "learning_rate": 4.655819301659869e-06,
      "loss": 11.7294,
      "step": 33254
    },
    {
      "epoch": 1.8108676563680546,
      "grad_norm": 0.5678487776401085,
      "learning_rate": 4.653160288033409e-06,
      "loss": 11.8871,
      "step": 33255
    },
    {
      "epoch": 1.8109221103646376,
      "grad_norm": 0.49279398776295474,
      "learning_rate": 4.650502015839953e-06,
      "loss": 11.7573,
      "step": 33256
    },
    {
      "epoch": 1.8109765643612206,
      "grad_norm": 0.49438942981577944,
      "learning_rate": 4.647844485100184e-06,
      "loss": 11.7171,
      "step": 33257
    },
    {
      "epoch": 1.8110310183578036,
      "grad_norm": 0.7163185688421985,
      "learning_rate": 4.645187695834774e-06,
      "loss": 11.7511,
      "step": 33258
    },
    {
      "epoch": 1.8110854723543865,
      "grad_norm": 0.5259278219870365,
      "learning_rate": 4.642531648064374e-06,
      "loss": 11.7328,
      "step": 33259
    },
    {
      "epoch": 1.8111399263509695,
      "grad_norm": 0.5057567915034931,
      "learning_rate": 4.639876341809657e-06,
      "loss": 11.7507,
      "step": 33260
    },
    {
      "epoch": 1.8111943803475525,
      "grad_norm": 0.5143063553239388,
      "learning_rate": 4.637221777091227e-06,
      "loss": 11.8169,
      "step": 33261
    },
    {
      "epoch": 1.8112488343441355,
      "grad_norm": 0.5147328992008211,
      "learning_rate": 4.634567953929758e-06,
      "loss": 11.813,
      "step": 33262
    },
    {
      "epoch": 1.8113032883407185,
      "grad_norm": 0.5632237736043315,
      "learning_rate": 4.6319148723459e-06,
      "loss": 11.6644,
      "step": 33263
    },
    {
      "epoch": 1.8113577423373015,
      "grad_norm": 0.580805442319458,
      "learning_rate": 4.6292625323602346e-06,
      "loss": 11.7563,
      "step": 33264
    },
    {
      "epoch": 1.8114121963338847,
      "grad_norm": 0.5617976680277361,
      "learning_rate": 4.626610933993447e-06,
      "loss": 11.7837,
      "step": 33265
    },
    {
      "epoch": 1.8114666503304677,
      "grad_norm": 0.649113014473685,
      "learning_rate": 4.62396007726611e-06,
      "loss": 11.8386,
      "step": 33266
    },
    {
      "epoch": 1.8115211043270507,
      "grad_norm": 0.5411429635084574,
      "learning_rate": 4.621309962198861e-06,
      "loss": 11.7343,
      "step": 33267
    },
    {
      "epoch": 1.8115755583236337,
      "grad_norm": 0.5173265291460204,
      "learning_rate": 4.618660588812284e-06,
      "loss": 11.7112,
      "step": 33268
    },
    {
      "epoch": 1.8116300123202167,
      "grad_norm": 0.6377719460421658,
      "learning_rate": 4.616011957127009e-06,
      "loss": 11.8151,
      "step": 33269
    },
    {
      "epoch": 1.8116844663167997,
      "grad_norm": 0.47039995371666304,
      "learning_rate": 4.613364067163639e-06,
      "loss": 11.8002,
      "step": 33270
    },
    {
      "epoch": 1.811738920313383,
      "grad_norm": 0.5676157980270421,
      "learning_rate": 4.610716918942726e-06,
      "loss": 11.788,
      "step": 33271
    },
    {
      "epoch": 1.811793374309966,
      "grad_norm": 0.5539329546890714,
      "learning_rate": 4.6080705124848854e-06,
      "loss": 11.7338,
      "step": 33272
    },
    {
      "epoch": 1.811847828306549,
      "grad_norm": 0.6242612650651835,
      "learning_rate": 4.60542484781068e-06,
      "loss": 11.7645,
      "step": 33273
    },
    {
      "epoch": 1.8119022823031319,
      "grad_norm": 0.5256584462171094,
      "learning_rate": 4.602779924940681e-06,
      "loss": 11.6497,
      "step": 33274
    },
    {
      "epoch": 1.8119567362997149,
      "grad_norm": 0.5091817870369126,
      "learning_rate": 4.600135743895473e-06,
      "loss": 11.6957,
      "step": 33275
    },
    {
      "epoch": 1.8120111902962979,
      "grad_norm": 0.6179767147898236,
      "learning_rate": 4.597492304695605e-06,
      "loss": 11.9228,
      "step": 33276
    },
    {
      "epoch": 1.8120656442928809,
      "grad_norm": 0.547809004844969,
      "learning_rate": 4.5948496073616395e-06,
      "loss": 11.8367,
      "step": 33277
    },
    {
      "epoch": 1.8121200982894639,
      "grad_norm": 0.5145807759496834,
      "learning_rate": 4.592207651914115e-06,
      "loss": 11.7819,
      "step": 33278
    },
    {
      "epoch": 1.8121745522860468,
      "grad_norm": 0.5736815125760386,
      "learning_rate": 4.589566438373594e-06,
      "loss": 11.7776,
      "step": 33279
    },
    {
      "epoch": 1.8122290062826298,
      "grad_norm": 0.6243906740683176,
      "learning_rate": 4.586925966760602e-06,
      "loss": 11.7867,
      "step": 33280
    },
    {
      "epoch": 1.8122834602792128,
      "grad_norm": 0.5354100915325727,
      "learning_rate": 4.584286237095669e-06,
      "loss": 11.7204,
      "step": 33281
    },
    {
      "epoch": 1.8123379142757958,
      "grad_norm": 0.5680149612435472,
      "learning_rate": 4.581647249399335e-06,
      "loss": 11.6994,
      "step": 33282
    },
    {
      "epoch": 1.8123923682723788,
      "grad_norm": 0.5192362144096057,
      "learning_rate": 4.579009003692103e-06,
      "loss": 11.7376,
      "step": 33283
    },
    {
      "epoch": 1.8124468222689618,
      "grad_norm": 0.5867927421074767,
      "learning_rate": 4.576371499994503e-06,
      "loss": 11.8128,
      "step": 33284
    },
    {
      "epoch": 1.8125012762655448,
      "grad_norm": 0.5088532240817195,
      "learning_rate": 4.573734738327052e-06,
      "loss": 11.7488,
      "step": 33285
    },
    {
      "epoch": 1.8125557302621278,
      "grad_norm": 0.5126047136330019,
      "learning_rate": 4.57109871871022e-06,
      "loss": 11.6943,
      "step": 33286
    },
    {
      "epoch": 1.8126101842587108,
      "grad_norm": 0.5445520444535854,
      "learning_rate": 4.56846344116455e-06,
      "loss": 11.7375,
      "step": 33287
    },
    {
      "epoch": 1.812664638255294,
      "grad_norm": 0.5235453073984584,
      "learning_rate": 4.565828905710501e-06,
      "loss": 11.7828,
      "step": 33288
    },
    {
      "epoch": 1.812719092251877,
      "grad_norm": 0.6082597676176608,
      "learning_rate": 4.563195112368579e-06,
      "loss": 11.7967,
      "step": 33289
    },
    {
      "epoch": 1.81277354624846,
      "grad_norm": 0.5184887605400904,
      "learning_rate": 4.560562061159257e-06,
      "loss": 11.853,
      "step": 33290
    },
    {
      "epoch": 1.812828000245043,
      "grad_norm": 0.5212808673359198,
      "learning_rate": 4.557929752102996e-06,
      "loss": 11.7158,
      "step": 33291
    },
    {
      "epoch": 1.812882454241626,
      "grad_norm": 0.5108478368583238,
      "learning_rate": 4.555298185220291e-06,
      "loss": 11.7638,
      "step": 33292
    },
    {
      "epoch": 1.812936908238209,
      "grad_norm": 0.5577709139267213,
      "learning_rate": 4.552667360531582e-06,
      "loss": 11.7567,
      "step": 33293
    },
    {
      "epoch": 1.8129913622347922,
      "grad_norm": 0.5426431823864006,
      "learning_rate": 4.5500372780573505e-06,
      "loss": 11.8768,
      "step": 33294
    },
    {
      "epoch": 1.8130458162313752,
      "grad_norm": 0.5837656771242806,
      "learning_rate": 4.5474079378180265e-06,
      "loss": 11.7182,
      "step": 33295
    },
    {
      "epoch": 1.8131002702279582,
      "grad_norm": 0.5553445735645262,
      "learning_rate": 4.54477933983406e-06,
      "loss": 11.6822,
      "step": 33296
    },
    {
      "epoch": 1.8131547242245412,
      "grad_norm": 0.5262483855216749,
      "learning_rate": 4.5421514841259115e-06,
      "loss": 11.7623,
      "step": 33297
    },
    {
      "epoch": 1.8132091782211242,
      "grad_norm": 0.5803607492229638,
      "learning_rate": 4.539524370713988e-06,
      "loss": 11.7194,
      "step": 33298
    },
    {
      "epoch": 1.8132636322177071,
      "grad_norm": 0.5781905249474677,
      "learning_rate": 4.53689799961875e-06,
      "loss": 11.9124,
      "step": 33299
    },
    {
      "epoch": 1.8133180862142901,
      "grad_norm": 0.5236121957406582,
      "learning_rate": 4.5342723708606036e-06,
      "loss": 11.81,
      "step": 33300
    },
    {
      "epoch": 1.8133725402108731,
      "grad_norm": 0.4992174506719999,
      "learning_rate": 4.531647484459945e-06,
      "loss": 11.7106,
      "step": 33301
    },
    {
      "epoch": 1.8134269942074561,
      "grad_norm": 0.5549655380744889,
      "learning_rate": 4.529023340437222e-06,
      "loss": 11.6963,
      "step": 33302
    },
    {
      "epoch": 1.8134814482040391,
      "grad_norm": 0.5603578588330804,
      "learning_rate": 4.5263999388128085e-06,
      "loss": 11.8157,
      "step": 33303
    },
    {
      "epoch": 1.813535902200622,
      "grad_norm": 0.5322873348611824,
      "learning_rate": 4.523777279607133e-06,
      "loss": 11.7214,
      "step": 33304
    },
    {
      "epoch": 1.813590356197205,
      "grad_norm": 0.5531494564247507,
      "learning_rate": 4.521155362840568e-06,
      "loss": 11.7389,
      "step": 33305
    },
    {
      "epoch": 1.813644810193788,
      "grad_norm": 0.6302713973539736,
      "learning_rate": 4.518534188533508e-06,
      "loss": 11.6865,
      "step": 33306
    },
    {
      "epoch": 1.813699264190371,
      "grad_norm": 0.49506686958018375,
      "learning_rate": 4.515913756706347e-06,
      "loss": 11.724,
      "step": 33307
    },
    {
      "epoch": 1.813753718186954,
      "grad_norm": 0.5194636773562015,
      "learning_rate": 4.513294067379448e-06,
      "loss": 11.7975,
      "step": 33308
    },
    {
      "epoch": 1.813808172183537,
      "grad_norm": 0.5697393268501594,
      "learning_rate": 4.510675120573204e-06,
      "loss": 11.8829,
      "step": 33309
    },
    {
      "epoch": 1.81386262618012,
      "grad_norm": 0.5224354794587687,
      "learning_rate": 4.508056916307957e-06,
      "loss": 11.7556,
      "step": 33310
    },
    {
      "epoch": 1.813917080176703,
      "grad_norm": 0.5405521840368618,
      "learning_rate": 4.505439454604054e-06,
      "loss": 11.7114,
      "step": 33311
    },
    {
      "epoch": 1.8139715341732863,
      "grad_norm": 0.5638519595378328,
      "learning_rate": 4.5028227354818935e-06,
      "loss": 11.8164,
      "step": 33312
    },
    {
      "epoch": 1.8140259881698693,
      "grad_norm": 0.568896695552425,
      "learning_rate": 4.500206758961778e-06,
      "loss": 11.7192,
      "step": 33313
    },
    {
      "epoch": 1.8140804421664523,
      "grad_norm": 0.5615458323351968,
      "learning_rate": 4.497591525064082e-06,
      "loss": 11.7677,
      "step": 33314
    },
    {
      "epoch": 1.8141348961630352,
      "grad_norm": 0.5293997293990172,
      "learning_rate": 4.4949770338091225e-06,
      "loss": 11.8222,
      "step": 33315
    },
    {
      "epoch": 1.8141893501596182,
      "grad_norm": 0.568010811307435,
      "learning_rate": 4.49236328521725e-06,
      "loss": 11.9389,
      "step": 33316
    },
    {
      "epoch": 1.8142438041562012,
      "grad_norm": 0.49981054646930256,
      "learning_rate": 4.489750279308757e-06,
      "loss": 11.8114,
      "step": 33317
    },
    {
      "epoch": 1.8142982581527844,
      "grad_norm": 0.5225249272932014,
      "learning_rate": 4.4871380161039865e-06,
      "loss": 11.7547,
      "step": 33318
    },
    {
      "epoch": 1.8143527121493674,
      "grad_norm": 0.5493646771922226,
      "learning_rate": 4.484526495623265e-06,
      "loss": 11.5598,
      "step": 33319
    },
    {
      "epoch": 1.8144071661459504,
      "grad_norm": 0.6062000021701177,
      "learning_rate": 4.481915717886886e-06,
      "loss": 11.743,
      "step": 33320
    },
    {
      "epoch": 1.8144616201425334,
      "grad_norm": 0.5753678436648219,
      "learning_rate": 4.479305682915136e-06,
      "loss": 11.9379,
      "step": 33321
    },
    {
      "epoch": 1.8145160741391164,
      "grad_norm": 0.6455252697951465,
      "learning_rate": 4.476696390728318e-06,
      "loss": 11.7879,
      "step": 33322
    },
    {
      "epoch": 1.8145705281356994,
      "grad_norm": 0.5407366107188332,
      "learning_rate": 4.47408784134673e-06,
      "loss": 11.8897,
      "step": 33323
    },
    {
      "epoch": 1.8146249821322824,
      "grad_norm": 0.5738336666237459,
      "learning_rate": 4.471480034790676e-06,
      "loss": 11.8631,
      "step": 33324
    },
    {
      "epoch": 1.8146794361288654,
      "grad_norm": 0.5893288333801068,
      "learning_rate": 4.468872971080384e-06,
      "loss": 11.8426,
      "step": 33325
    },
    {
      "epoch": 1.8147338901254484,
      "grad_norm": 0.5472307563950716,
      "learning_rate": 4.466266650236184e-06,
      "loss": 11.7547,
      "step": 33326
    },
    {
      "epoch": 1.8147883441220314,
      "grad_norm": 0.6296197773275323,
      "learning_rate": 4.463661072278291e-06,
      "loss": 11.8189,
      "step": 33327
    },
    {
      "epoch": 1.8148427981186144,
      "grad_norm": 0.5841100972519099,
      "learning_rate": 4.461056237227001e-06,
      "loss": 11.7946,
      "step": 33328
    },
    {
      "epoch": 1.8148972521151974,
      "grad_norm": 0.5151973204027636,
      "learning_rate": 4.458452145102554e-06,
      "loss": 11.7281,
      "step": 33329
    },
    {
      "epoch": 1.8149517061117804,
      "grad_norm": 0.5503912756346909,
      "learning_rate": 4.4558487959252215e-06,
      "loss": 11.8112,
      "step": 33330
    },
    {
      "epoch": 1.8150061601083634,
      "grad_norm": 0.5226465804257504,
      "learning_rate": 4.453246189715232e-06,
      "loss": 11.8009,
      "step": 33331
    },
    {
      "epoch": 1.8150606141049463,
      "grad_norm": 0.6119326188396066,
      "learning_rate": 4.450644326492803e-06,
      "loss": 11.8591,
      "step": 33332
    },
    {
      "epoch": 1.8151150681015293,
      "grad_norm": 0.5231959641530465,
      "learning_rate": 4.448043206278196e-06,
      "loss": 11.7045,
      "step": 33333
    },
    {
      "epoch": 1.8151695220981123,
      "grad_norm": 0.4931963916129974,
      "learning_rate": 4.445442829091629e-06,
      "loss": 11.7716,
      "step": 33334
    },
    {
      "epoch": 1.8152239760946955,
      "grad_norm": 0.5697086660320094,
      "learning_rate": 4.442843194953317e-06,
      "loss": 11.9071,
      "step": 33335
    },
    {
      "epoch": 1.8152784300912785,
      "grad_norm": 0.5087142488913186,
      "learning_rate": 4.440244303883501e-06,
      "loss": 11.7693,
      "step": 33336
    },
    {
      "epoch": 1.8153328840878615,
      "grad_norm": 0.5295033186600981,
      "learning_rate": 4.437646155902353e-06,
      "loss": 11.8257,
      "step": 33337
    },
    {
      "epoch": 1.8153873380844445,
      "grad_norm": 0.5766055544264328,
      "learning_rate": 4.435048751030102e-06,
      "loss": 11.8032,
      "step": 33338
    },
    {
      "epoch": 1.8154417920810275,
      "grad_norm": 0.5235030491962843,
      "learning_rate": 4.432452089286931e-06,
      "loss": 11.8001,
      "step": 33339
    },
    {
      "epoch": 1.8154962460776105,
      "grad_norm": 0.5517239769619563,
      "learning_rate": 4.429856170693047e-06,
      "loss": 11.7633,
      "step": 33340
    },
    {
      "epoch": 1.8155507000741937,
      "grad_norm": 0.5235495340085781,
      "learning_rate": 4.427260995268634e-06,
      "loss": 11.7956,
      "step": 33341
    },
    {
      "epoch": 1.8156051540707767,
      "grad_norm": 0.47897288269388455,
      "learning_rate": 4.424666563033853e-06,
      "loss": 11.6569,
      "step": 33342
    },
    {
      "epoch": 1.8156596080673597,
      "grad_norm": 0.5624732894903519,
      "learning_rate": 4.422072874008909e-06,
      "loss": 11.9135,
      "step": 33343
    },
    {
      "epoch": 1.8157140620639427,
      "grad_norm": 0.5635515144033085,
      "learning_rate": 4.419479928213932e-06,
      "loss": 11.6486,
      "step": 33344
    },
    {
      "epoch": 1.8157685160605257,
      "grad_norm": 0.5859999162358663,
      "learning_rate": 4.4168877256691165e-06,
      "loss": 11.7451,
      "step": 33345
    },
    {
      "epoch": 1.8158229700571087,
      "grad_norm": 0.5564169979815358,
      "learning_rate": 4.414296266394624e-06,
      "loss": 11.8639,
      "step": 33346
    },
    {
      "epoch": 1.8158774240536917,
      "grad_norm": 0.5819972700378305,
      "learning_rate": 4.411705550410572e-06,
      "loss": 11.7479,
      "step": 33347
    },
    {
      "epoch": 1.8159318780502747,
      "grad_norm": 0.5720018007402299,
      "learning_rate": 4.409115577737155e-06,
      "loss": 11.8476,
      "step": 33348
    },
    {
      "epoch": 1.8159863320468577,
      "grad_norm": 0.5472262518933869,
      "learning_rate": 4.406526348394458e-06,
      "loss": 11.8056,
      "step": 33349
    },
    {
      "epoch": 1.8160407860434407,
      "grad_norm": 0.5890280566237442,
      "learning_rate": 4.403937862402663e-06,
      "loss": 11.7987,
      "step": 33350
    },
    {
      "epoch": 1.8160952400400237,
      "grad_norm": 0.5684000945037254,
      "learning_rate": 4.401350119781877e-06,
      "loss": 11.8233,
      "step": 33351
    },
    {
      "epoch": 1.8161496940366066,
      "grad_norm": 0.5102991266086635,
      "learning_rate": 4.398763120552218e-06,
      "loss": 11.6719,
      "step": 33352
    },
    {
      "epoch": 1.8162041480331896,
      "grad_norm": 0.5504354368638585,
      "learning_rate": 4.396176864733815e-06,
      "loss": 11.8069,
      "step": 33353
    },
    {
      "epoch": 1.8162586020297726,
      "grad_norm": 0.608197532885495,
      "learning_rate": 4.39359135234676e-06,
      "loss": 11.7731,
      "step": 33354
    },
    {
      "epoch": 1.8163130560263556,
      "grad_norm": 0.6193116912963961,
      "learning_rate": 4.3910065834111725e-06,
      "loss": 11.8017,
      "step": 33355
    },
    {
      "epoch": 1.8163675100229386,
      "grad_norm": 0.5551117452595495,
      "learning_rate": 4.38842255794717e-06,
      "loss": 11.8228,
      "step": 33356
    },
    {
      "epoch": 1.8164219640195216,
      "grad_norm": 0.5526334834554533,
      "learning_rate": 4.385839275974812e-06,
      "loss": 11.7001,
      "step": 33357
    },
    {
      "epoch": 1.8164764180161048,
      "grad_norm": 0.5410470492028854,
      "learning_rate": 4.3832567375142186e-06,
      "loss": 11.769,
      "step": 33358
    },
    {
      "epoch": 1.8165308720126878,
      "grad_norm": 0.4824851732505219,
      "learning_rate": 4.380674942585428e-06,
      "loss": 11.7198,
      "step": 33359
    },
    {
      "epoch": 1.8165853260092708,
      "grad_norm": 0.5371286258570362,
      "learning_rate": 4.378093891208568e-06,
      "loss": 11.7387,
      "step": 33360
    },
    {
      "epoch": 1.8166397800058538,
      "grad_norm": 0.5261341840184692,
      "learning_rate": 4.37551358340369e-06,
      "loss": 11.6994,
      "step": 33361
    },
    {
      "epoch": 1.8166942340024368,
      "grad_norm": 0.506506395276939,
      "learning_rate": 4.372934019190833e-06,
      "loss": 11.7797,
      "step": 33362
    },
    {
      "epoch": 1.8167486879990198,
      "grad_norm": 0.5681386074969045,
      "learning_rate": 4.370355198590103e-06,
      "loss": 11.7945,
      "step": 33363
    },
    {
      "epoch": 1.816803141995603,
      "grad_norm": 0.5927926920329334,
      "learning_rate": 4.367777121621508e-06,
      "loss": 11.9358,
      "step": 33364
    },
    {
      "epoch": 1.816857595992186,
      "grad_norm": 0.5492525892142982,
      "learning_rate": 4.365199788305119e-06,
      "loss": 11.8749,
      "step": 33365
    },
    {
      "epoch": 1.816912049988769,
      "grad_norm": 0.5868734938464674,
      "learning_rate": 4.362623198660975e-06,
      "loss": 11.8656,
      "step": 33366
    },
    {
      "epoch": 1.816966503985352,
      "grad_norm": 0.5737197849021911,
      "learning_rate": 4.360047352709107e-06,
      "loss": 11.6498,
      "step": 33367
    },
    {
      "epoch": 1.817020957981935,
      "grad_norm": 0.5010726442825639,
      "learning_rate": 4.357472250469563e-06,
      "loss": 11.7842,
      "step": 33368
    },
    {
      "epoch": 1.817075411978518,
      "grad_norm": 0.5260934326391239,
      "learning_rate": 4.354897891962339e-06,
      "loss": 11.763,
      "step": 33369
    },
    {
      "epoch": 1.817129865975101,
      "grad_norm": 0.5762939373781866,
      "learning_rate": 4.352324277207498e-06,
      "loss": 11.8411,
      "step": 33370
    },
    {
      "epoch": 1.817184319971684,
      "grad_norm": 0.46623385380449567,
      "learning_rate": 4.349751406224989e-06,
      "loss": 11.7587,
      "step": 33371
    },
    {
      "epoch": 1.817238773968267,
      "grad_norm": 0.5648296635165347,
      "learning_rate": 4.3471792790348634e-06,
      "loss": 11.7265,
      "step": 33372
    },
    {
      "epoch": 1.81729322796485,
      "grad_norm": 0.5351545931153535,
      "learning_rate": 4.344607895657127e-06,
      "loss": 11.7363,
      "step": 33373
    },
    {
      "epoch": 1.817347681961433,
      "grad_norm": 0.5870113109642406,
      "learning_rate": 4.342037256111742e-06,
      "loss": 11.701,
      "step": 33374
    },
    {
      "epoch": 1.817402135958016,
      "grad_norm": 0.538544264144728,
      "learning_rate": 4.3394673604187255e-06,
      "loss": 11.7449,
      "step": 33375
    },
    {
      "epoch": 1.817456589954599,
      "grad_norm": 0.5717901168737194,
      "learning_rate": 4.33689820859805e-06,
      "loss": 11.8415,
      "step": 33376
    },
    {
      "epoch": 1.817511043951182,
      "grad_norm": 0.6177129381709924,
      "learning_rate": 4.3343298006697005e-06,
      "loss": 11.6388,
      "step": 33377
    },
    {
      "epoch": 1.817565497947765,
      "grad_norm": 0.5550373898347775,
      "learning_rate": 4.331762136653639e-06,
      "loss": 11.7749,
      "step": 33378
    },
    {
      "epoch": 1.817619951944348,
      "grad_norm": 0.4833398613095589,
      "learning_rate": 4.329195216569837e-06,
      "loss": 11.5995,
      "step": 33379
    },
    {
      "epoch": 1.8176744059409309,
      "grad_norm": 0.5039485652407718,
      "learning_rate": 4.32662904043829e-06,
      "loss": 11.8422,
      "step": 33380
    },
    {
      "epoch": 1.8177288599375139,
      "grad_norm": 0.5287346972619532,
      "learning_rate": 4.324063608278883e-06,
      "loss": 11.8981,
      "step": 33381
    },
    {
      "epoch": 1.817783313934097,
      "grad_norm": 0.5433745401187149,
      "learning_rate": 4.3214989201116105e-06,
      "loss": 11.7002,
      "step": 33382
    },
    {
      "epoch": 1.81783776793068,
      "grad_norm": 0.5623893756865836,
      "learning_rate": 4.318934975956413e-06,
      "loss": 11.7466,
      "step": 33383
    },
    {
      "epoch": 1.817892221927263,
      "grad_norm": 0.5229503510293498,
      "learning_rate": 4.316371775833217e-06,
      "loss": 11.7498,
      "step": 33384
    },
    {
      "epoch": 1.817946675923846,
      "grad_norm": 0.5196069408964588,
      "learning_rate": 4.3138093197619634e-06,
      "loss": 11.6792,
      "step": 33385
    },
    {
      "epoch": 1.818001129920429,
      "grad_norm": 0.5270375830935387,
      "learning_rate": 4.311247607762569e-06,
      "loss": 11.7954,
      "step": 33386
    },
    {
      "epoch": 1.8180555839170123,
      "grad_norm": 0.523893051620918,
      "learning_rate": 4.308686639854975e-06,
      "loss": 11.7321,
      "step": 33387
    },
    {
      "epoch": 1.8181100379135953,
      "grad_norm": 0.5614365727548793,
      "learning_rate": 4.306126416059064e-06,
      "loss": 11.7834,
      "step": 33388
    },
    {
      "epoch": 1.8181644919101783,
      "grad_norm": 0.513632033751421,
      "learning_rate": 4.303566936394765e-06,
      "loss": 11.7146,
      "step": 33389
    },
    {
      "epoch": 1.8182189459067613,
      "grad_norm": 0.5477296502040427,
      "learning_rate": 4.301008200882006e-06,
      "loss": 11.7553,
      "step": 33390
    },
    {
      "epoch": 1.8182733999033442,
      "grad_norm": 0.5785223420716445,
      "learning_rate": 4.298450209540628e-06,
      "loss": 11.7638,
      "step": 33391
    },
    {
      "epoch": 1.8183278538999272,
      "grad_norm": 0.5166285606492488,
      "learning_rate": 4.295892962390558e-06,
      "loss": 11.7441,
      "step": 33392
    },
    {
      "epoch": 1.8183823078965102,
      "grad_norm": 0.6215707218826039,
      "learning_rate": 4.293336459451669e-06,
      "loss": 11.8865,
      "step": 33393
    },
    {
      "epoch": 1.8184367618930932,
      "grad_norm": 0.5193871759628252,
      "learning_rate": 4.290780700743846e-06,
      "loss": 11.753,
      "step": 33394
    },
    {
      "epoch": 1.8184912158896762,
      "grad_norm": 0.5457247022914097,
      "learning_rate": 4.288225686286962e-06,
      "loss": 11.8888,
      "step": 33395
    },
    {
      "epoch": 1.8185456698862592,
      "grad_norm": 0.6053692002563366,
      "learning_rate": 4.285671416100889e-06,
      "loss": 11.7425,
      "step": 33396
    },
    {
      "epoch": 1.8186001238828422,
      "grad_norm": 0.5378080864852702,
      "learning_rate": 4.28311789020549e-06,
      "loss": 11.7059,
      "step": 33397
    },
    {
      "epoch": 1.8186545778794252,
      "grad_norm": 0.5961398522834784,
      "learning_rate": 4.280565108620604e-06,
      "loss": 11.8281,
      "step": 33398
    },
    {
      "epoch": 1.8187090318760082,
      "grad_norm": 0.616068474741855,
      "learning_rate": 4.278013071366116e-06,
      "loss": 11.9312,
      "step": 33399
    },
    {
      "epoch": 1.8187634858725912,
      "grad_norm": 0.5498839603826776,
      "learning_rate": 4.275461778461831e-06,
      "loss": 11.7726,
      "step": 33400
    },
    {
      "epoch": 1.8188179398691742,
      "grad_norm": 0.5951827338815259,
      "learning_rate": 4.272911229927634e-06,
      "loss": 11.759,
      "step": 33401
    },
    {
      "epoch": 1.8188723938657572,
      "grad_norm": 0.5330808372606821,
      "learning_rate": 4.27036142578332e-06,
      "loss": 11.8161,
      "step": 33402
    },
    {
      "epoch": 1.8189268478623402,
      "grad_norm": 0.5826470579002819,
      "learning_rate": 4.267812366048718e-06,
      "loss": 11.8479,
      "step": 33403
    },
    {
      "epoch": 1.8189813018589231,
      "grad_norm": 0.510313617563817,
      "learning_rate": 4.265264050743667e-06,
      "loss": 11.7532,
      "step": 33404
    },
    {
      "epoch": 1.8190357558555064,
      "grad_norm": 0.5416770344837316,
      "learning_rate": 4.2627164798879845e-06,
      "loss": 11.8805,
      "step": 33405
    },
    {
      "epoch": 1.8190902098520894,
      "grad_norm": 0.5713216715113655,
      "learning_rate": 4.260169653501467e-06,
      "loss": 11.8938,
      "step": 33406
    },
    {
      "epoch": 1.8191446638486723,
      "grad_norm": 0.5185567635271205,
      "learning_rate": 4.25762357160393e-06,
      "loss": 11.7382,
      "step": 33407
    },
    {
      "epoch": 1.8191991178452553,
      "grad_norm": 0.5088800254469613,
      "learning_rate": 4.25507823421516e-06,
      "loss": 11.7215,
      "step": 33408
    },
    {
      "epoch": 1.8192535718418383,
      "grad_norm": 0.5321922930209237,
      "learning_rate": 4.252533641354972e-06,
      "loss": 11.7429,
      "step": 33409
    },
    {
      "epoch": 1.8193080258384213,
      "grad_norm": 0.5342018890374852,
      "learning_rate": 4.249989793043119e-06,
      "loss": 11.6897,
      "step": 33410
    },
    {
      "epoch": 1.8193624798350045,
      "grad_norm": 0.523277589494354,
      "learning_rate": 4.247446689299428e-06,
      "loss": 11.7073,
      "step": 33411
    },
    {
      "epoch": 1.8194169338315875,
      "grad_norm": 0.5845380724600674,
      "learning_rate": 4.24490433014364e-06,
      "loss": 11.7868,
      "step": 33412
    },
    {
      "epoch": 1.8194713878281705,
      "grad_norm": 0.5011164408601148,
      "learning_rate": 4.242362715595527e-06,
      "loss": 11.8852,
      "step": 33413
    },
    {
      "epoch": 1.8195258418247535,
      "grad_norm": 0.5536522211802043,
      "learning_rate": 4.239821845674874e-06,
      "loss": 11.8753,
      "step": 33414
    },
    {
      "epoch": 1.8195802958213365,
      "grad_norm": 0.49836783003234875,
      "learning_rate": 4.23728172040141e-06,
      "loss": 11.7772,
      "step": 33415
    },
    {
      "epoch": 1.8196347498179195,
      "grad_norm": 0.5547445071794451,
      "learning_rate": 4.2347423397948945e-06,
      "loss": 11.8077,
      "step": 33416
    },
    {
      "epoch": 1.8196892038145025,
      "grad_norm": 0.5113337040094863,
      "learning_rate": 4.232203703875104e-06,
      "loss": 11.6234,
      "step": 33417
    },
    {
      "epoch": 1.8197436578110855,
      "grad_norm": 0.5791002775489162,
      "learning_rate": 4.229665812661742e-06,
      "loss": 11.9064,
      "step": 33418
    },
    {
      "epoch": 1.8197981118076685,
      "grad_norm": 0.504620041642399,
      "learning_rate": 4.227128666174574e-06,
      "loss": 11.7447,
      "step": 33419
    },
    {
      "epoch": 1.8198525658042515,
      "grad_norm": 0.6257757387641968,
      "learning_rate": 4.224592264433302e-06,
      "loss": 11.8402,
      "step": 33420
    },
    {
      "epoch": 1.8199070198008345,
      "grad_norm": 0.6169789961814764,
      "learning_rate": 4.222056607457669e-06,
      "loss": 11.8489,
      "step": 33421
    },
    {
      "epoch": 1.8199614737974175,
      "grad_norm": 0.7099164133328735,
      "learning_rate": 4.219521695267392e-06,
      "loss": 11.6361,
      "step": 33422
    },
    {
      "epoch": 1.8200159277940005,
      "grad_norm": 0.5359550502325594,
      "learning_rate": 4.216987527882166e-06,
      "loss": 11.8551,
      "step": 33423
    },
    {
      "epoch": 1.8200703817905834,
      "grad_norm": 0.5362127744955114,
      "learning_rate": 4.214454105321719e-06,
      "loss": 11.7809,
      "step": 33424
    },
    {
      "epoch": 1.8201248357871664,
      "grad_norm": 0.5881300916325793,
      "learning_rate": 4.211921427605725e-06,
      "loss": 11.7678,
      "step": 33425
    },
    {
      "epoch": 1.8201792897837494,
      "grad_norm": 0.5299651412201789,
      "learning_rate": 4.209389494753902e-06,
      "loss": 11.7097,
      "step": 33426
    },
    {
      "epoch": 1.8202337437803324,
      "grad_norm": 0.5394203768634799,
      "learning_rate": 4.206858306785922e-06,
      "loss": 11.7668,
      "step": 33427
    },
    {
      "epoch": 1.8202881977769156,
      "grad_norm": 0.55487418992176,
      "learning_rate": 4.20432786372148e-06,
      "loss": 11.7554,
      "step": 33428
    },
    {
      "epoch": 1.8203426517734986,
      "grad_norm": 0.5553023300732464,
      "learning_rate": 4.201798165580262e-06,
      "loss": 11.747,
      "step": 33429
    },
    {
      "epoch": 1.8203971057700816,
      "grad_norm": 0.5267832576306941,
      "learning_rate": 4.199269212381906e-06,
      "loss": 11.8648,
      "step": 33430
    },
    {
      "epoch": 1.8204515597666646,
      "grad_norm": 0.5247482544173551,
      "learning_rate": 4.196741004146121e-06,
      "loss": 11.6874,
      "step": 33431
    },
    {
      "epoch": 1.8205060137632476,
      "grad_norm": 0.5166248630521795,
      "learning_rate": 4.194213540892544e-06,
      "loss": 11.8875,
      "step": 33432
    },
    {
      "epoch": 1.8205604677598306,
      "grad_norm": 0.507621012519032,
      "learning_rate": 4.191686822640806e-06,
      "loss": 11.7607,
      "step": 33433
    },
    {
      "epoch": 1.8206149217564138,
      "grad_norm": 0.5069945546515978,
      "learning_rate": 4.1891608494106e-06,
      "loss": 11.7674,
      "step": 33434
    },
    {
      "epoch": 1.8206693757529968,
      "grad_norm": 0.6204043248303384,
      "learning_rate": 4.186635621221524e-06,
      "loss": 11.8956,
      "step": 33435
    },
    {
      "epoch": 1.8207238297495798,
      "grad_norm": 0.5070988944301096,
      "learning_rate": 4.184111138093261e-06,
      "loss": 11.7003,
      "step": 33436
    },
    {
      "epoch": 1.8207782837461628,
      "grad_norm": 0.5401162562355685,
      "learning_rate": 4.181587400045406e-06,
      "loss": 11.772,
      "step": 33437
    },
    {
      "epoch": 1.8208327377427458,
      "grad_norm": 0.6120771082563683,
      "learning_rate": 4.179064407097588e-06,
      "loss": 11.8083,
      "step": 33438
    },
    {
      "epoch": 1.8208871917393288,
      "grad_norm": 0.5474527864695609,
      "learning_rate": 4.176542159269448e-06,
      "loss": 11.8253,
      "step": 33439
    },
    {
      "epoch": 1.8209416457359118,
      "grad_norm": 0.4984756252724798,
      "learning_rate": 4.17402065658058e-06,
      "loss": 11.7179,
      "step": 33440
    },
    {
      "epoch": 1.8209960997324948,
      "grad_norm": 0.5595777782453509,
      "learning_rate": 4.171499899050612e-06,
      "loss": 11.8789,
      "step": 33441
    },
    {
      "epoch": 1.8210505537290778,
      "grad_norm": 0.551713785230797,
      "learning_rate": 4.16897988669912e-06,
      "loss": 11.7899,
      "step": 33442
    },
    {
      "epoch": 1.8211050077256608,
      "grad_norm": 0.5359306983219716,
      "learning_rate": 4.166460619545698e-06,
      "loss": 11.6985,
      "step": 33443
    },
    {
      "epoch": 1.8211594617222437,
      "grad_norm": 0.5220655482790552,
      "learning_rate": 4.1639420976099745e-06,
      "loss": 11.7098,
      "step": 33444
    },
    {
      "epoch": 1.8212139157188267,
      "grad_norm": 0.5894974152081649,
      "learning_rate": 4.1614243209114886e-06,
      "loss": 11.7912,
      "step": 33445
    },
    {
      "epoch": 1.8212683697154097,
      "grad_norm": 0.549315203766272,
      "learning_rate": 4.158907289469849e-06,
      "loss": 11.715,
      "step": 33446
    },
    {
      "epoch": 1.8213228237119927,
      "grad_norm": 0.5489235924790603,
      "learning_rate": 4.156391003304616e-06,
      "loss": 11.8682,
      "step": 33447
    },
    {
      "epoch": 1.8213772777085757,
      "grad_norm": 0.587180068057914,
      "learning_rate": 4.153875462435363e-06,
      "loss": 11.7768,
      "step": 33448
    },
    {
      "epoch": 1.8214317317051587,
      "grad_norm": 0.6152518409633797,
      "learning_rate": 4.151360666881643e-06,
      "loss": 11.8247,
      "step": 33449
    },
    {
      "epoch": 1.8214861857017417,
      "grad_norm": 0.5084308337759688,
      "learning_rate": 4.1488466166630046e-06,
      "loss": 11.7028,
      "step": 33450
    },
    {
      "epoch": 1.8215406396983247,
      "grad_norm": 0.585755402914931,
      "learning_rate": 4.146333311799044e-06,
      "loss": 11.7917,
      "step": 33451
    },
    {
      "epoch": 1.821595093694908,
      "grad_norm": 0.5121071994847662,
      "learning_rate": 4.1438207523092466e-06,
      "loss": 11.7782,
      "step": 33452
    },
    {
      "epoch": 1.821649547691491,
      "grad_norm": 0.510229864272492,
      "learning_rate": 4.141308938213162e-06,
      "loss": 11.8306,
      "step": 33453
    },
    {
      "epoch": 1.821704001688074,
      "grad_norm": 0.5967912187830556,
      "learning_rate": 4.1387978695303645e-06,
      "loss": 11.6707,
      "step": 33454
    },
    {
      "epoch": 1.8217584556846569,
      "grad_norm": 0.592826957125138,
      "learning_rate": 4.1362875462803264e-06,
      "loss": 11.9532,
      "step": 33455
    },
    {
      "epoch": 1.8218129096812399,
      "grad_norm": 0.6119896503146464,
      "learning_rate": 4.13377796848261e-06,
      "loss": 11.7899,
      "step": 33456
    },
    {
      "epoch": 1.821867363677823,
      "grad_norm": 0.5436330551330625,
      "learning_rate": 4.1312691361566905e-06,
      "loss": 11.8297,
      "step": 33457
    },
    {
      "epoch": 1.821921817674406,
      "grad_norm": 0.5556863848375361,
      "learning_rate": 4.128761049322127e-06,
      "loss": 11.7746,
      "step": 33458
    },
    {
      "epoch": 1.821976271670989,
      "grad_norm": 0.6291168622184845,
      "learning_rate": 4.126253707998373e-06,
      "loss": 11.8717,
      "step": 33459
    },
    {
      "epoch": 1.822030725667572,
      "grad_norm": 0.5158533575860381,
      "learning_rate": 4.123747112204945e-06,
      "loss": 11.7635,
      "step": 33460
    },
    {
      "epoch": 1.822085179664155,
      "grad_norm": 0.5252329388785419,
      "learning_rate": 4.121241261961372e-06,
      "loss": 11.8638,
      "step": 33461
    },
    {
      "epoch": 1.822139633660738,
      "grad_norm": 0.5511530083840109,
      "learning_rate": 4.118736157287073e-06,
      "loss": 11.8423,
      "step": 33462
    },
    {
      "epoch": 1.822194087657321,
      "grad_norm": 0.5093725510377453,
      "learning_rate": 4.116231798201586e-06,
      "loss": 11.7447,
      "step": 33463
    },
    {
      "epoch": 1.822248541653904,
      "grad_norm": 0.5513784833495462,
      "learning_rate": 4.11372818472433e-06,
      "loss": 11.8521,
      "step": 33464
    },
    {
      "epoch": 1.822302995650487,
      "grad_norm": 0.5737361005938579,
      "learning_rate": 4.111225316874823e-06,
      "loss": 11.8044,
      "step": 33465
    },
    {
      "epoch": 1.82235744964707,
      "grad_norm": 0.5736580509093636,
      "learning_rate": 4.108723194672504e-06,
      "loss": 11.7777,
      "step": 33466
    },
    {
      "epoch": 1.822411903643653,
      "grad_norm": 0.4762929827191769,
      "learning_rate": 4.106221818136835e-06,
      "loss": 11.7381,
      "step": 33467
    },
    {
      "epoch": 1.822466357640236,
      "grad_norm": 0.5685099642399056,
      "learning_rate": 4.103721187287268e-06,
      "loss": 11.7572,
      "step": 33468
    },
    {
      "epoch": 1.822520811636819,
      "grad_norm": 0.5643956750201643,
      "learning_rate": 4.101221302143244e-06,
      "loss": 11.7392,
      "step": 33469
    },
    {
      "epoch": 1.822575265633402,
      "grad_norm": 0.5591510287579061,
      "learning_rate": 4.0987221627242114e-06,
      "loss": 11.706,
      "step": 33470
    },
    {
      "epoch": 1.822629719629985,
      "grad_norm": 0.5173483899573453,
      "learning_rate": 4.0962237690496005e-06,
      "loss": 11.6633,
      "step": 33471
    },
    {
      "epoch": 1.822684173626568,
      "grad_norm": 0.5755603656146498,
      "learning_rate": 4.093726121138819e-06,
      "loss": 11.7247,
      "step": 33472
    },
    {
      "epoch": 1.822738627623151,
      "grad_norm": 0.5355424571927285,
      "learning_rate": 4.0912292190113275e-06,
      "loss": 11.6119,
      "step": 33473
    },
    {
      "epoch": 1.822793081619734,
      "grad_norm": 0.5545457762756087,
      "learning_rate": 4.088733062686501e-06,
      "loss": 11.7037,
      "step": 33474
    },
    {
      "epoch": 1.8228475356163172,
      "grad_norm": 0.5973165584493365,
      "learning_rate": 4.086237652183789e-06,
      "loss": 11.7259,
      "step": 33475
    },
    {
      "epoch": 1.8229019896129002,
      "grad_norm": 0.5586578768783266,
      "learning_rate": 4.083742987522565e-06,
      "loss": 11.7125,
      "step": 33476
    },
    {
      "epoch": 1.8229564436094832,
      "grad_norm": 0.535720181937925,
      "learning_rate": 4.081249068722237e-06,
      "loss": 11.7799,
      "step": 33477
    },
    {
      "epoch": 1.8230108976060662,
      "grad_norm": 0.5093407659815583,
      "learning_rate": 4.078755895802222e-06,
      "loss": 11.8156,
      "step": 33478
    },
    {
      "epoch": 1.8230653516026492,
      "grad_norm": 0.5401439310621149,
      "learning_rate": 4.076263468781871e-06,
      "loss": 11.7707,
      "step": 33479
    },
    {
      "epoch": 1.8231198055992321,
      "grad_norm": 0.5588026189918949,
      "learning_rate": 4.073771787680591e-06,
      "loss": 11.7741,
      "step": 33480
    },
    {
      "epoch": 1.8231742595958154,
      "grad_norm": 0.7520636812294781,
      "learning_rate": 4.0712808525177445e-06,
      "loss": 11.898,
      "step": 33481
    },
    {
      "epoch": 1.8232287135923984,
      "grad_norm": 0.5100512927896358,
      "learning_rate": 4.068790663312705e-06,
      "loss": 11.6486,
      "step": 33482
    },
    {
      "epoch": 1.8232831675889813,
      "grad_norm": 0.5180993532143796,
      "learning_rate": 4.066301220084834e-06,
      "loss": 11.7956,
      "step": 33483
    },
    {
      "epoch": 1.8233376215855643,
      "grad_norm": 0.552233970687271,
      "learning_rate": 4.063812522853494e-06,
      "loss": 11.7794,
      "step": 33484
    },
    {
      "epoch": 1.8233920755821473,
      "grad_norm": 0.5829374629722772,
      "learning_rate": 4.061324571638048e-06,
      "loss": 11.8612,
      "step": 33485
    },
    {
      "epoch": 1.8234465295787303,
      "grad_norm": 0.5354389819448515,
      "learning_rate": 4.058837366457813e-06,
      "loss": 11.827,
      "step": 33486
    },
    {
      "epoch": 1.8235009835753133,
      "grad_norm": 0.5289263660728635,
      "learning_rate": 4.056350907332141e-06,
      "loss": 11.8566,
      "step": 33487
    },
    {
      "epoch": 1.8235554375718963,
      "grad_norm": 0.5190264478511732,
      "learning_rate": 4.053865194280404e-06,
      "loss": 11.7686,
      "step": 33488
    },
    {
      "epoch": 1.8236098915684793,
      "grad_norm": 0.6100448974671006,
      "learning_rate": 4.051380227321877e-06,
      "loss": 11.7739,
      "step": 33489
    },
    {
      "epoch": 1.8236643455650623,
      "grad_norm": 0.5570678020624251,
      "learning_rate": 4.048896006475922e-06,
      "loss": 11.9415,
      "step": 33490
    },
    {
      "epoch": 1.8237187995616453,
      "grad_norm": 0.6030830455316337,
      "learning_rate": 4.046412531761845e-06,
      "loss": 11.8224,
      "step": 33491
    },
    {
      "epoch": 1.8237732535582283,
      "grad_norm": 0.5958088040953632,
      "learning_rate": 4.043929803198943e-06,
      "loss": 11.8535,
      "step": 33492
    },
    {
      "epoch": 1.8238277075548113,
      "grad_norm": 0.5204784015270262,
      "learning_rate": 4.041447820806543e-06,
      "loss": 11.8078,
      "step": 33493
    },
    {
      "epoch": 1.8238821615513943,
      "grad_norm": 0.5771124407140996,
      "learning_rate": 4.038966584603932e-06,
      "loss": 11.6476,
      "step": 33494
    },
    {
      "epoch": 1.8239366155479773,
      "grad_norm": 0.5696700008076531,
      "learning_rate": 4.036486094610414e-06,
      "loss": 11.7869,
      "step": 33495
    },
    {
      "epoch": 1.8239910695445603,
      "grad_norm": 0.555014384169201,
      "learning_rate": 4.034006350845265e-06,
      "loss": 11.8537,
      "step": 33496
    },
    {
      "epoch": 1.8240455235411432,
      "grad_norm": 0.5408362132387518,
      "learning_rate": 4.031527353327791e-06,
      "loss": 11.7032,
      "step": 33497
    },
    {
      "epoch": 1.8240999775377265,
      "grad_norm": 0.5366653302165357,
      "learning_rate": 4.029049102077231e-06,
      "loss": 11.7905,
      "step": 33498
    },
    {
      "epoch": 1.8241544315343095,
      "grad_norm": 0.5751795087361428,
      "learning_rate": 4.026571597112893e-06,
      "loss": 11.7215,
      "step": 33499
    },
    {
      "epoch": 1.8242088855308924,
      "grad_norm": 0.5635483618024176,
      "learning_rate": 4.0240948384540286e-06,
      "loss": 11.8326,
      "step": 33500
    },
    {
      "epoch": 1.8242633395274754,
      "grad_norm": 0.5132812454468167,
      "learning_rate": 4.02161882611991e-06,
      "loss": 11.744,
      "step": 33501
    },
    {
      "epoch": 1.8243177935240584,
      "grad_norm": 0.7318438973514497,
      "learning_rate": 4.019143560129757e-06,
      "loss": 11.9016,
      "step": 33502
    },
    {
      "epoch": 1.8243722475206414,
      "grad_norm": 0.5615510206887165,
      "learning_rate": 4.016669040502863e-06,
      "loss": 11.8107,
      "step": 33503
    },
    {
      "epoch": 1.8244267015172246,
      "grad_norm": 0.6124006959764241,
      "learning_rate": 4.014195267258425e-06,
      "loss": 11.89,
      "step": 33504
    },
    {
      "epoch": 1.8244811555138076,
      "grad_norm": 0.5255157333733381,
      "learning_rate": 4.011722240415716e-06,
      "loss": 11.7617,
      "step": 33505
    },
    {
      "epoch": 1.8245356095103906,
      "grad_norm": 0.5741861787237976,
      "learning_rate": 4.009249959993943e-06,
      "loss": 11.8035,
      "step": 33506
    },
    {
      "epoch": 1.8245900635069736,
      "grad_norm": 0.557906623715009,
      "learning_rate": 4.006778426012348e-06,
      "loss": 11.767,
      "step": 33507
    },
    {
      "epoch": 1.8246445175035566,
      "grad_norm": 0.510337835917614,
      "learning_rate": 4.004307638490135e-06,
      "loss": 11.8081,
      "step": 33508
    },
    {
      "epoch": 1.8246989715001396,
      "grad_norm": 0.5642408994260448,
      "learning_rate": 4.001837597446523e-06,
      "loss": 11.8833,
      "step": 33509
    },
    {
      "epoch": 1.8247534254967226,
      "grad_norm": 0.580121439087195,
      "learning_rate": 3.99936830290073e-06,
      "loss": 11.8494,
      "step": 33510
    },
    {
      "epoch": 1.8248078794933056,
      "grad_norm": 0.5428320757164858,
      "learning_rate": 3.996899754871952e-06,
      "loss": 11.7593,
      "step": 33511
    },
    {
      "epoch": 1.8248623334898886,
      "grad_norm": 0.5673038505201914,
      "learning_rate": 3.994431953379374e-06,
      "loss": 11.8183,
      "step": 33512
    },
    {
      "epoch": 1.8249167874864716,
      "grad_norm": 0.5365934862740096,
      "learning_rate": 3.991964898442191e-06,
      "loss": 11.565,
      "step": 33513
    },
    {
      "epoch": 1.8249712414830546,
      "grad_norm": 0.526265695858374,
      "learning_rate": 3.989498590079577e-06,
      "loss": 11.7453,
      "step": 33514
    },
    {
      "epoch": 1.8250256954796376,
      "grad_norm": 0.5468074167815248,
      "learning_rate": 3.987033028310749e-06,
      "loss": 11.8768,
      "step": 33515
    },
    {
      "epoch": 1.8250801494762205,
      "grad_norm": 0.5590618180762743,
      "learning_rate": 3.984568213154827e-06,
      "loss": 11.7486,
      "step": 33516
    },
    {
      "epoch": 1.8251346034728035,
      "grad_norm": 0.5274691560683105,
      "learning_rate": 3.982104144631027e-06,
      "loss": 11.6954,
      "step": 33517
    },
    {
      "epoch": 1.8251890574693865,
      "grad_norm": 0.5301098752998251,
      "learning_rate": 3.979640822758468e-06,
      "loss": 11.8421,
      "step": 33518
    },
    {
      "epoch": 1.8252435114659695,
      "grad_norm": 0.6038447826816482,
      "learning_rate": 3.977178247556346e-06,
      "loss": 11.795,
      "step": 33519
    },
    {
      "epoch": 1.8252979654625525,
      "grad_norm": 0.5864158844167162,
      "learning_rate": 3.974716419043767e-06,
      "loss": 11.7993,
      "step": 33520
    },
    {
      "epoch": 1.8253524194591357,
      "grad_norm": 0.5217450081365466,
      "learning_rate": 3.972255337239916e-06,
      "loss": 11.757,
      "step": 33521
    },
    {
      "epoch": 1.8254068734557187,
      "grad_norm": 0.5495316015868976,
      "learning_rate": 3.969795002163912e-06,
      "loss": 11.7986,
      "step": 33522
    },
    {
      "epoch": 1.8254613274523017,
      "grad_norm": 0.5359550676514172,
      "learning_rate": 3.967335413834872e-06,
      "loss": 11.8066,
      "step": 33523
    },
    {
      "epoch": 1.8255157814488847,
      "grad_norm": 0.528853344261235,
      "learning_rate": 3.964876572271947e-06,
      "loss": 11.7869,
      "step": 33524
    },
    {
      "epoch": 1.8255702354454677,
      "grad_norm": 0.5651961918332623,
      "learning_rate": 3.962418477494234e-06,
      "loss": 11.8611,
      "step": 33525
    },
    {
      "epoch": 1.8256246894420507,
      "grad_norm": 0.6233616935130436,
      "learning_rate": 3.959961129520862e-06,
      "loss": 11.7343,
      "step": 33526
    },
    {
      "epoch": 1.825679143438634,
      "grad_norm": 0.5257494402201819,
      "learning_rate": 3.957504528370948e-06,
      "loss": 11.6419,
      "step": 33527
    },
    {
      "epoch": 1.825733597435217,
      "grad_norm": 0.5632429576399808,
      "learning_rate": 3.955048674063577e-06,
      "loss": 11.8467,
      "step": 33528
    },
    {
      "epoch": 1.8257880514318,
      "grad_norm": 0.5665512604678933,
      "learning_rate": 3.952593566617868e-06,
      "loss": 11.8056,
      "step": 33529
    },
    {
      "epoch": 1.825842505428383,
      "grad_norm": 0.4963751260642664,
      "learning_rate": 3.950139206052883e-06,
      "loss": 11.6963,
      "step": 33530
    },
    {
      "epoch": 1.8258969594249659,
      "grad_norm": 0.6040634228372773,
      "learning_rate": 3.94768559238774e-06,
      "loss": 11.6905,
      "step": 33531
    },
    {
      "epoch": 1.8259514134215489,
      "grad_norm": 0.5819911824782974,
      "learning_rate": 3.94523272564149e-06,
      "loss": 11.8623,
      "step": 33532
    },
    {
      "epoch": 1.8260058674181319,
      "grad_norm": 0.5712258398644969,
      "learning_rate": 3.942780605833218e-06,
      "loss": 11.8323,
      "step": 33533
    },
    {
      "epoch": 1.8260603214147149,
      "grad_norm": 0.5260499541865548,
      "learning_rate": 3.940329232981999e-06,
      "loss": 11.7525,
      "step": 33534
    },
    {
      "epoch": 1.8261147754112979,
      "grad_norm": 0.6315230831867228,
      "learning_rate": 3.937878607106882e-06,
      "loss": 11.7957,
      "step": 33535
    },
    {
      "epoch": 1.8261692294078808,
      "grad_norm": 0.5747869563676422,
      "learning_rate": 3.93542872822692e-06,
      "loss": 11.7983,
      "step": 33536
    },
    {
      "epoch": 1.8262236834044638,
      "grad_norm": 0.5092012466463031,
      "learning_rate": 3.932979596361197e-06,
      "loss": 11.7679,
      "step": 33537
    },
    {
      "epoch": 1.8262781374010468,
      "grad_norm": 0.5208759833375679,
      "learning_rate": 3.930531211528721e-06,
      "loss": 11.8191,
      "step": 33538
    },
    {
      "epoch": 1.8263325913976298,
      "grad_norm": 0.510786388246234,
      "learning_rate": 3.928083573748554e-06,
      "loss": 11.7865,
      "step": 33539
    },
    {
      "epoch": 1.8263870453942128,
      "grad_norm": 0.5496315114605621,
      "learning_rate": 3.925636683039713e-06,
      "loss": 11.7737,
      "step": 33540
    },
    {
      "epoch": 1.8264414993907958,
      "grad_norm": 0.5115296641116465,
      "learning_rate": 3.92319053942124e-06,
      "loss": 11.5915,
      "step": 33541
    },
    {
      "epoch": 1.8264959533873788,
      "grad_norm": 0.5194726769345599,
      "learning_rate": 3.920745142912152e-06,
      "loss": 11.7564,
      "step": 33542
    },
    {
      "epoch": 1.8265504073839618,
      "grad_norm": 0.5129502256690269,
      "learning_rate": 3.9183004935314575e-06,
      "loss": 11.8047,
      "step": 33543
    },
    {
      "epoch": 1.8266048613805448,
      "grad_norm": 0.5156117474702039,
      "learning_rate": 3.915856591298172e-06,
      "loss": 11.6586,
      "step": 33544
    },
    {
      "epoch": 1.826659315377128,
      "grad_norm": 0.5090345224098513,
      "learning_rate": 3.913413436231284e-06,
      "loss": 11.7061,
      "step": 33545
    },
    {
      "epoch": 1.826713769373711,
      "grad_norm": 0.5375689023583371,
      "learning_rate": 3.91097102834983e-06,
      "loss": 11.7355,
      "step": 33546
    },
    {
      "epoch": 1.826768223370294,
      "grad_norm": 0.5301205373658786,
      "learning_rate": 3.908529367672764e-06,
      "loss": 11.7086,
      "step": 33547
    },
    {
      "epoch": 1.826822677366877,
      "grad_norm": 0.5711261212450155,
      "learning_rate": 3.906088454219081e-06,
      "loss": 11.7186,
      "step": 33548
    },
    {
      "epoch": 1.82687713136346,
      "grad_norm": 0.552201056906126,
      "learning_rate": 3.9036482880077885e-06,
      "loss": 11.7314,
      "step": 33549
    },
    {
      "epoch": 1.826931585360043,
      "grad_norm": 0.6254665150234314,
      "learning_rate": 3.901208869057838e-06,
      "loss": 11.717,
      "step": 33550
    },
    {
      "epoch": 1.8269860393566262,
      "grad_norm": 0.5310749922950447,
      "learning_rate": 3.898770197388202e-06,
      "loss": 11.8462,
      "step": 33551
    },
    {
      "epoch": 1.8270404933532092,
      "grad_norm": 0.5839481534108598,
      "learning_rate": 3.896332273017844e-06,
      "loss": 11.8887,
      "step": 33552
    },
    {
      "epoch": 1.8270949473497922,
      "grad_norm": 0.5488042552274396,
      "learning_rate": 3.8938950959657164e-06,
      "loss": 11.7305,
      "step": 33553
    },
    {
      "epoch": 1.8271494013463752,
      "grad_norm": 0.5482019480639584,
      "learning_rate": 3.891458666250791e-06,
      "loss": 11.8,
      "step": 33554
    },
    {
      "epoch": 1.8272038553429581,
      "grad_norm": 0.5318702393227676,
      "learning_rate": 3.889022983891988e-06,
      "loss": 11.6851,
      "step": 33555
    },
    {
      "epoch": 1.8272583093395411,
      "grad_norm": 0.5674620351831604,
      "learning_rate": 3.886588048908268e-06,
      "loss": 11.6989,
      "step": 33556
    },
    {
      "epoch": 1.8273127633361241,
      "grad_norm": 0.5604734315062843,
      "learning_rate": 3.8841538613185516e-06,
      "loss": 11.7631,
      "step": 33557
    },
    {
      "epoch": 1.8273672173327071,
      "grad_norm": 0.5664493019605271,
      "learning_rate": 3.881720421141766e-06,
      "loss": 11.774,
      "step": 33558
    },
    {
      "epoch": 1.8274216713292901,
      "grad_norm": 0.5485116698974144,
      "learning_rate": 3.8792877283968635e-06,
      "loss": 11.8739,
      "step": 33559
    },
    {
      "epoch": 1.8274761253258731,
      "grad_norm": 0.5468985098187719,
      "learning_rate": 3.8768557831027175e-06,
      "loss": 11.7811,
      "step": 33560
    },
    {
      "epoch": 1.827530579322456,
      "grad_norm": 0.56447714643943,
      "learning_rate": 3.87442458527828e-06,
      "loss": 11.8985,
      "step": 33561
    },
    {
      "epoch": 1.827585033319039,
      "grad_norm": 0.5239203982580789,
      "learning_rate": 3.871994134942436e-06,
      "loss": 11.786,
      "step": 33562
    },
    {
      "epoch": 1.827639487315622,
      "grad_norm": 0.6344256070612694,
      "learning_rate": 3.86956443211407e-06,
      "loss": 11.7964,
      "step": 33563
    },
    {
      "epoch": 1.827693941312205,
      "grad_norm": 0.5252318449258297,
      "learning_rate": 3.867135476812101e-06,
      "loss": 11.6403,
      "step": 33564
    },
    {
      "epoch": 1.827748395308788,
      "grad_norm": 0.5342986990063621,
      "learning_rate": 3.864707269055401e-06,
      "loss": 11.6848,
      "step": 33565
    },
    {
      "epoch": 1.827802849305371,
      "grad_norm": 0.5265795921368792,
      "learning_rate": 3.862279808862878e-06,
      "loss": 11.7295,
      "step": 33566
    },
    {
      "epoch": 1.827857303301954,
      "grad_norm": 0.5534938744120571,
      "learning_rate": 3.859853096253363e-06,
      "loss": 11.8545,
      "step": 33567
    },
    {
      "epoch": 1.8279117572985373,
      "grad_norm": 0.5314140467853407,
      "learning_rate": 3.8574271312457725e-06,
      "loss": 11.7534,
      "step": 33568
    },
    {
      "epoch": 1.8279662112951203,
      "grad_norm": 0.589235670157932,
      "learning_rate": 3.8550019138589465e-06,
      "loss": 11.8212,
      "step": 33569
    },
    {
      "epoch": 1.8280206652917033,
      "grad_norm": 0.6138541162915487,
      "learning_rate": 3.852577444111738e-06,
      "loss": 11.7081,
      "step": 33570
    },
    {
      "epoch": 1.8280751192882863,
      "grad_norm": 0.5913178943141056,
      "learning_rate": 3.850153722023031e-06,
      "loss": 11.7072,
      "step": 33571
    },
    {
      "epoch": 1.8281295732848692,
      "grad_norm": 0.5667341265681017,
      "learning_rate": 3.8477307476116445e-06,
      "loss": 11.7691,
      "step": 33572
    },
    {
      "epoch": 1.8281840272814522,
      "grad_norm": 0.5785024779146959,
      "learning_rate": 3.84530852089644e-06,
      "loss": 11.9016,
      "step": 33573
    },
    {
      "epoch": 1.8282384812780355,
      "grad_norm": 0.5261711522291102,
      "learning_rate": 3.842887041896226e-06,
      "loss": 11.7926,
      "step": 33574
    },
    {
      "epoch": 1.8282929352746184,
      "grad_norm": 0.5538186553914405,
      "learning_rate": 3.8404663106298426e-06,
      "loss": 11.6854,
      "step": 33575
    },
    {
      "epoch": 1.8283473892712014,
      "grad_norm": 0.5154450321743397,
      "learning_rate": 3.8380463271161294e-06,
      "loss": 11.8107,
      "step": 33576
    },
    {
      "epoch": 1.8284018432677844,
      "grad_norm": 0.5768646424999689,
      "learning_rate": 3.835627091373883e-06,
      "loss": 11.7498,
      "step": 33577
    },
    {
      "epoch": 1.8284562972643674,
      "grad_norm": 0.5540500575422419,
      "learning_rate": 3.833208603421945e-06,
      "loss": 11.6689,
      "step": 33578
    },
    {
      "epoch": 1.8285107512609504,
      "grad_norm": 0.5255612682975356,
      "learning_rate": 3.830790863279088e-06,
      "loss": 11.6593,
      "step": 33579
    },
    {
      "epoch": 1.8285652052575334,
      "grad_norm": 0.6457394825812683,
      "learning_rate": 3.828373870964153e-06,
      "loss": 11.7615,
      "step": 33580
    },
    {
      "epoch": 1.8286196592541164,
      "grad_norm": 0.5497840083945347,
      "learning_rate": 3.82595762649588e-06,
      "loss": 11.8321,
      "step": 33581
    },
    {
      "epoch": 1.8286741132506994,
      "grad_norm": 0.5115940301137489,
      "learning_rate": 3.8235421298931096e-06,
      "loss": 11.9497,
      "step": 33582
    },
    {
      "epoch": 1.8287285672472824,
      "grad_norm": 0.5415655013355506,
      "learning_rate": 3.8211273811746055e-06,
      "loss": 11.6561,
      "step": 33583
    },
    {
      "epoch": 1.8287830212438654,
      "grad_norm": 0.5508433141977883,
      "learning_rate": 3.8187133803591295e-06,
      "loss": 11.7448,
      "step": 33584
    },
    {
      "epoch": 1.8288374752404484,
      "grad_norm": 0.5045897152001088,
      "learning_rate": 3.816300127465466e-06,
      "loss": 11.7792,
      "step": 33585
    },
    {
      "epoch": 1.8288919292370314,
      "grad_norm": 0.5137576802657025,
      "learning_rate": 3.8138876225124022e-06,
      "loss": 11.7979,
      "step": 33586
    },
    {
      "epoch": 1.8289463832336144,
      "grad_norm": 0.5264990376744149,
      "learning_rate": 3.8114758655186657e-06,
      "loss": 11.6629,
      "step": 33587
    },
    {
      "epoch": 1.8290008372301974,
      "grad_norm": 0.5882370370095509,
      "learning_rate": 3.8090648565030305e-06,
      "loss": 11.8454,
      "step": 33588
    },
    {
      "epoch": 1.8290552912267803,
      "grad_norm": 0.49267080795308765,
      "learning_rate": 3.806654595484227e-06,
      "loss": 11.7577,
      "step": 33589
    },
    {
      "epoch": 1.8291097452233633,
      "grad_norm": 0.5659817523041979,
      "learning_rate": 3.804245082481017e-06,
      "loss": 11.845,
      "step": 33590
    },
    {
      "epoch": 1.8291641992199466,
      "grad_norm": 0.5002647537304712,
      "learning_rate": 3.8018363175121306e-06,
      "loss": 11.7127,
      "step": 33591
    },
    {
      "epoch": 1.8292186532165295,
      "grad_norm": 0.537488999759871,
      "learning_rate": 3.7994283005962974e-06,
      "loss": 11.7534,
      "step": 33592
    },
    {
      "epoch": 1.8292731072131125,
      "grad_norm": 0.5781595998247397,
      "learning_rate": 3.7970210317522457e-06,
      "loss": 11.7543,
      "step": 33593
    },
    {
      "epoch": 1.8293275612096955,
      "grad_norm": 0.514201095862873,
      "learning_rate": 3.794614510998684e-06,
      "loss": 11.7786,
      "step": 33594
    },
    {
      "epoch": 1.8293820152062785,
      "grad_norm": 0.47920266652534765,
      "learning_rate": 3.7922087383543414e-06,
      "loss": 11.6605,
      "step": 33595
    },
    {
      "epoch": 1.8294364692028615,
      "grad_norm": 0.5740164054785348,
      "learning_rate": 3.7898037138379027e-06,
      "loss": 11.8035,
      "step": 33596
    },
    {
      "epoch": 1.8294909231994447,
      "grad_norm": 0.5942858605802906,
      "learning_rate": 3.7873994374680864e-06,
      "loss": 11.7615,
      "step": 33597
    },
    {
      "epoch": 1.8295453771960277,
      "grad_norm": 0.5675241724473631,
      "learning_rate": 3.7849959092636e-06,
      "loss": 11.7158,
      "step": 33598
    },
    {
      "epoch": 1.8295998311926107,
      "grad_norm": 0.5415942452650854,
      "learning_rate": 3.782593129243117e-06,
      "loss": 11.7951,
      "step": 33599
    },
    {
      "epoch": 1.8296542851891937,
      "grad_norm": 0.5467909468926304,
      "learning_rate": 3.780191097425323e-06,
      "loss": 11.6424,
      "step": 33600
    },
    {
      "epoch": 1.8297087391857767,
      "grad_norm": 0.5691203385637891,
      "learning_rate": 3.7777898138289025e-06,
      "loss": 11.7199,
      "step": 33601
    },
    {
      "epoch": 1.8297631931823597,
      "grad_norm": 0.5605891023426018,
      "learning_rate": 3.7753892784725188e-06,
      "loss": 11.8701,
      "step": 33602
    },
    {
      "epoch": 1.8298176471789427,
      "grad_norm": 0.584924898795815,
      "learning_rate": 3.7729894913748564e-06,
      "loss": 11.902,
      "step": 33603
    },
    {
      "epoch": 1.8298721011755257,
      "grad_norm": 0.644131364469211,
      "learning_rate": 3.7705904525545567e-06,
      "loss": 11.7686,
      "step": 33604
    },
    {
      "epoch": 1.8299265551721087,
      "grad_norm": 0.5328483942849503,
      "learning_rate": 3.7681921620302818e-06,
      "loss": 11.6705,
      "step": 33605
    },
    {
      "epoch": 1.8299810091686917,
      "grad_norm": 0.5084748882255641,
      "learning_rate": 3.765794619820684e-06,
      "loss": 11.7388,
      "step": 33606
    },
    {
      "epoch": 1.8300354631652747,
      "grad_norm": 0.5387853995100667,
      "learning_rate": 3.7633978259444035e-06,
      "loss": 11.8504,
      "step": 33607
    },
    {
      "epoch": 1.8300899171618576,
      "grad_norm": 0.552683848284675,
      "learning_rate": 3.7610017804200815e-06,
      "loss": 11.8648,
      "step": 33608
    },
    {
      "epoch": 1.8301443711584406,
      "grad_norm": 0.5490437201936529,
      "learning_rate": 3.758606483266347e-06,
      "loss": 11.7925,
      "step": 33609
    },
    {
      "epoch": 1.8301988251550236,
      "grad_norm": 0.5589677545008368,
      "learning_rate": 3.7562119345018408e-06,
      "loss": 11.797,
      "step": 33610
    },
    {
      "epoch": 1.8302532791516066,
      "grad_norm": 0.5426710002220848,
      "learning_rate": 3.7538181341451593e-06,
      "loss": 11.8016,
      "step": 33611
    },
    {
      "epoch": 1.8303077331481896,
      "grad_norm": 0.5816641511814663,
      "learning_rate": 3.751425082214932e-06,
      "loss": 11.7449,
      "step": 33612
    },
    {
      "epoch": 1.8303621871447726,
      "grad_norm": 0.54028323069145,
      "learning_rate": 3.7490327787297664e-06,
      "loss": 11.8756,
      "step": 33613
    },
    {
      "epoch": 1.8304166411413556,
      "grad_norm": 0.5437505885248093,
      "learning_rate": 3.746641223708258e-06,
      "loss": 11.7894,
      "step": 33614
    },
    {
      "epoch": 1.8304710951379388,
      "grad_norm": 0.5107128114373266,
      "learning_rate": 3.744250417169015e-06,
      "loss": 11.7212,
      "step": 33615
    },
    {
      "epoch": 1.8305255491345218,
      "grad_norm": 0.553345160010139,
      "learning_rate": 3.7418603591306112e-06,
      "loss": 11.7638,
      "step": 33616
    },
    {
      "epoch": 1.8305800031311048,
      "grad_norm": 0.5201727159599808,
      "learning_rate": 3.7394710496116648e-06,
      "loss": 11.8593,
      "step": 33617
    },
    {
      "epoch": 1.8306344571276878,
      "grad_norm": 0.5548311846824785,
      "learning_rate": 3.7370824886307056e-06,
      "loss": 11.7792,
      "step": 33618
    },
    {
      "epoch": 1.8306889111242708,
      "grad_norm": 0.4999218151319178,
      "learning_rate": 3.734694676206352e-06,
      "loss": 11.7706,
      "step": 33619
    },
    {
      "epoch": 1.830743365120854,
      "grad_norm": 0.5730300704645586,
      "learning_rate": 3.7323076123571556e-06,
      "loss": 11.7754,
      "step": 33620
    },
    {
      "epoch": 1.830797819117437,
      "grad_norm": 0.620387799682463,
      "learning_rate": 3.7299212971016797e-06,
      "loss": 11.8119,
      "step": 33621
    },
    {
      "epoch": 1.83085227311402,
      "grad_norm": 0.5691015568130691,
      "learning_rate": 3.7275357304584978e-06,
      "loss": 11.8631,
      "step": 33622
    },
    {
      "epoch": 1.830906727110603,
      "grad_norm": 0.6034637461747832,
      "learning_rate": 3.725150912446118e-06,
      "loss": 11.8568,
      "step": 33623
    },
    {
      "epoch": 1.830961181107186,
      "grad_norm": 0.4916739862755106,
      "learning_rate": 3.722766843083114e-06,
      "loss": 11.644,
      "step": 33624
    },
    {
      "epoch": 1.831015635103769,
      "grad_norm": 0.5358512196182962,
      "learning_rate": 3.7203835223880268e-06,
      "loss": 11.7572,
      "step": 33625
    },
    {
      "epoch": 1.831070089100352,
      "grad_norm": 0.5223640356404262,
      "learning_rate": 3.7180009503793743e-06,
      "loss": 11.7622,
      "step": 33626
    },
    {
      "epoch": 1.831124543096935,
      "grad_norm": 0.5357213005973265,
      "learning_rate": 3.715619127075709e-06,
      "loss": 11.7943,
      "step": 33627
    },
    {
      "epoch": 1.831178997093518,
      "grad_norm": 0.5403553467227992,
      "learning_rate": 3.713238052495516e-06,
      "loss": 11.7865,
      "step": 33628
    },
    {
      "epoch": 1.831233451090101,
      "grad_norm": 0.555072468847666,
      "learning_rate": 3.7108577266573464e-06,
      "loss": 11.7744,
      "step": 33629
    },
    {
      "epoch": 1.831287905086684,
      "grad_norm": 0.5400711145683039,
      "learning_rate": 3.708478149579686e-06,
      "loss": 11.7561,
      "step": 33630
    },
    {
      "epoch": 1.831342359083267,
      "grad_norm": 0.5609842697305087,
      "learning_rate": 3.7060993212810535e-06,
      "loss": 11.7684,
      "step": 33631
    },
    {
      "epoch": 1.83139681307985,
      "grad_norm": 0.5558519570080996,
      "learning_rate": 3.703721241779956e-06,
      "loss": 11.7791,
      "step": 33632
    },
    {
      "epoch": 1.831451267076433,
      "grad_norm": 0.5742799670780467,
      "learning_rate": 3.7013439110948454e-06,
      "loss": 11.6923,
      "step": 33633
    },
    {
      "epoch": 1.831505721073016,
      "grad_norm": 0.4924399750888138,
      "learning_rate": 3.6989673292442407e-06,
      "loss": 11.8236,
      "step": 33634
    },
    {
      "epoch": 1.831560175069599,
      "grad_norm": 0.5755479277658104,
      "learning_rate": 3.6965914962466153e-06,
      "loss": 11.6702,
      "step": 33635
    },
    {
      "epoch": 1.8316146290661819,
      "grad_norm": 0.5668670997137093,
      "learning_rate": 3.694216412120444e-06,
      "loss": 11.7997,
      "step": 33636
    },
    {
      "epoch": 1.8316690830627649,
      "grad_norm": 0.5988314770585237,
      "learning_rate": 3.691842076884211e-06,
      "loss": 11.7937,
      "step": 33637
    },
    {
      "epoch": 1.831723537059348,
      "grad_norm": 0.5620322342811402,
      "learning_rate": 3.689468490556347e-06,
      "loss": 11.8267,
      "step": 33638
    },
    {
      "epoch": 1.831777991055931,
      "grad_norm": 0.5555223649173702,
      "learning_rate": 3.687095653155337e-06,
      "loss": 11.7555,
      "step": 33639
    },
    {
      "epoch": 1.831832445052514,
      "grad_norm": 0.5262201788231075,
      "learning_rate": 3.6847235646996102e-06,
      "loss": 11.7956,
      "step": 33640
    },
    {
      "epoch": 1.831886899049097,
      "grad_norm": 0.6097477835124032,
      "learning_rate": 3.6823522252076306e-06,
      "loss": 11.8023,
      "step": 33641
    },
    {
      "epoch": 1.83194135304568,
      "grad_norm": 0.5737853532295223,
      "learning_rate": 3.67998163469786e-06,
      "loss": 11.8284,
      "step": 33642
    },
    {
      "epoch": 1.831995807042263,
      "grad_norm": 0.5420299003831286,
      "learning_rate": 3.6776117931886733e-06,
      "loss": 11.757,
      "step": 33643
    },
    {
      "epoch": 1.8320502610388463,
      "grad_norm": 0.5757652343362268,
      "learning_rate": 3.6752427006985446e-06,
      "loss": 11.8109,
      "step": 33644
    },
    {
      "epoch": 1.8321047150354293,
      "grad_norm": 0.5118663299731981,
      "learning_rate": 3.67287435724587e-06,
      "loss": 11.6178,
      "step": 33645
    },
    {
      "epoch": 1.8321591690320123,
      "grad_norm": 0.5685451668477773,
      "learning_rate": 3.67050676284908e-06,
      "loss": 11.8246,
      "step": 33646
    },
    {
      "epoch": 1.8322136230285953,
      "grad_norm": 0.5323133237344017,
      "learning_rate": 3.668139917526592e-06,
      "loss": 11.8171,
      "step": 33647
    },
    {
      "epoch": 1.8322680770251782,
      "grad_norm": 0.5140507649072633,
      "learning_rate": 3.665773821296792e-06,
      "loss": 11.7333,
      "step": 33648
    },
    {
      "epoch": 1.8323225310217612,
      "grad_norm": 0.5631979179573443,
      "learning_rate": 3.6634084741781094e-06,
      "loss": 11.7532,
      "step": 33649
    },
    {
      "epoch": 1.8323769850183442,
      "grad_norm": 0.5506797082488566,
      "learning_rate": 3.6610438761888965e-06,
      "loss": 11.8504,
      "step": 33650
    },
    {
      "epoch": 1.8324314390149272,
      "grad_norm": 0.514061050701803,
      "learning_rate": 3.658680027347583e-06,
      "loss": 11.7024,
      "step": 33651
    },
    {
      "epoch": 1.8324858930115102,
      "grad_norm": 0.523363066878204,
      "learning_rate": 3.6563169276725206e-06,
      "loss": 11.801,
      "step": 33652
    },
    {
      "epoch": 1.8325403470080932,
      "grad_norm": 0.5703393082583804,
      "learning_rate": 3.6539545771820837e-06,
      "loss": 11.7958,
      "step": 33653
    },
    {
      "epoch": 1.8325948010046762,
      "grad_norm": 0.5645968935853388,
      "learning_rate": 3.6515929758946798e-06,
      "loss": 11.8249,
      "step": 33654
    },
    {
      "epoch": 1.8326492550012592,
      "grad_norm": 0.5269401939551209,
      "learning_rate": 3.6492321238286166e-06,
      "loss": 11.8765,
      "step": 33655
    },
    {
      "epoch": 1.8327037089978422,
      "grad_norm": 0.5474930751105823,
      "learning_rate": 3.64687202100229e-06,
      "loss": 11.814,
      "step": 33656
    },
    {
      "epoch": 1.8327581629944252,
      "grad_norm": 0.5282810706860984,
      "learning_rate": 3.6445126674340636e-06,
      "loss": 11.8933,
      "step": 33657
    },
    {
      "epoch": 1.8328126169910082,
      "grad_norm": 0.5192044567011344,
      "learning_rate": 3.6421540631422447e-06,
      "loss": 11.6787,
      "step": 33658
    },
    {
      "epoch": 1.8328670709875912,
      "grad_norm": 0.607667276788634,
      "learning_rate": 3.6397962081452074e-06,
      "loss": 11.9255,
      "step": 33659
    },
    {
      "epoch": 1.8329215249841742,
      "grad_norm": 0.5863241837845117,
      "learning_rate": 3.6374391024612597e-06,
      "loss": 11.7884,
      "step": 33660
    },
    {
      "epoch": 1.8329759789807574,
      "grad_norm": 0.564734192758178,
      "learning_rate": 3.6350827461087646e-06,
      "loss": 11.6631,
      "step": 33661
    },
    {
      "epoch": 1.8330304329773404,
      "grad_norm": 0.5771472327379159,
      "learning_rate": 3.632727139106018e-06,
      "loss": 11.7665,
      "step": 33662
    },
    {
      "epoch": 1.8330848869739234,
      "grad_norm": 0.5001199290655215,
      "learning_rate": 3.6303722814713503e-06,
      "loss": 11.6977,
      "step": 33663
    },
    {
      "epoch": 1.8331393409705063,
      "grad_norm": 0.5356334823763569,
      "learning_rate": 3.628018173223069e-06,
      "loss": 11.8373,
      "step": 33664
    },
    {
      "epoch": 1.8331937949670893,
      "grad_norm": 0.5407584876816743,
      "learning_rate": 3.6256648143794703e-06,
      "loss": 11.7917,
      "step": 33665
    },
    {
      "epoch": 1.8332482489636723,
      "grad_norm": 0.5262477456202952,
      "learning_rate": 3.623312204958873e-06,
      "loss": 11.8158,
      "step": 33666
    },
    {
      "epoch": 1.8333027029602555,
      "grad_norm": 0.5384037636619359,
      "learning_rate": 3.6209603449795515e-06,
      "loss": 11.7585,
      "step": 33667
    },
    {
      "epoch": 1.8333571569568385,
      "grad_norm": 0.5770060144243758,
      "learning_rate": 3.6186092344598023e-06,
      "loss": 11.7487,
      "step": 33668
    },
    {
      "epoch": 1.8334116109534215,
      "grad_norm": 0.48745336548330126,
      "learning_rate": 3.6162588734179326e-06,
      "loss": 11.8087,
      "step": 33669
    },
    {
      "epoch": 1.8334660649500045,
      "grad_norm": 0.5086000950835574,
      "learning_rate": 3.6139092618721727e-06,
      "loss": 11.7118,
      "step": 33670
    },
    {
      "epoch": 1.8335205189465875,
      "grad_norm": 0.5192043453566273,
      "learning_rate": 3.611560399840841e-06,
      "loss": 11.6702,
      "step": 33671
    },
    {
      "epoch": 1.8335749729431705,
      "grad_norm": 0.5433328063871803,
      "learning_rate": 3.6092122873421783e-06,
      "loss": 11.8374,
      "step": 33672
    },
    {
      "epoch": 1.8336294269397535,
      "grad_norm": 0.5404471055371471,
      "learning_rate": 3.6068649243944264e-06,
      "loss": 11.7247,
      "step": 33673
    },
    {
      "epoch": 1.8336838809363365,
      "grad_norm": 0.5311974378110095,
      "learning_rate": 3.604518311015881e-06,
      "loss": 11.655,
      "step": 33674
    },
    {
      "epoch": 1.8337383349329195,
      "grad_norm": 0.5458367952834732,
      "learning_rate": 3.6021724472247387e-06,
      "loss": 11.7666,
      "step": 33675
    },
    {
      "epoch": 1.8337927889295025,
      "grad_norm": 0.6863974465711044,
      "learning_rate": 3.599827333039296e-06,
      "loss": 11.8068,
      "step": 33676
    },
    {
      "epoch": 1.8338472429260855,
      "grad_norm": 0.49571684168626645,
      "learning_rate": 3.59748296847775e-06,
      "loss": 11.6708,
      "step": 33677
    },
    {
      "epoch": 1.8339016969226685,
      "grad_norm": 0.5016990640118166,
      "learning_rate": 3.5951393535583413e-06,
      "loss": 11.6921,
      "step": 33678
    },
    {
      "epoch": 1.8339561509192515,
      "grad_norm": 0.5088780451289755,
      "learning_rate": 3.592796488299299e-06,
      "loss": 11.7353,
      "step": 33679
    },
    {
      "epoch": 1.8340106049158345,
      "grad_norm": 0.5834436289501197,
      "learning_rate": 3.5904543727188322e-06,
      "loss": 11.8659,
      "step": 33680
    },
    {
      "epoch": 1.8340650589124174,
      "grad_norm": 0.5486892229223033,
      "learning_rate": 3.5881130068351698e-06,
      "loss": 11.7854,
      "step": 33681
    },
    {
      "epoch": 1.8341195129090004,
      "grad_norm": 0.5295833476695455,
      "learning_rate": 3.5857723906665197e-06,
      "loss": 11.7739,
      "step": 33682
    },
    {
      "epoch": 1.8341739669055834,
      "grad_norm": 0.5252285010143786,
      "learning_rate": 3.5834325242310453e-06,
      "loss": 11.7262,
      "step": 33683
    },
    {
      "epoch": 1.8342284209021664,
      "grad_norm": 0.6019713085614004,
      "learning_rate": 3.581093407546987e-06,
      "loss": 11.7305,
      "step": 33684
    },
    {
      "epoch": 1.8342828748987496,
      "grad_norm": 0.5351547807703942,
      "learning_rate": 3.5787550406325086e-06,
      "loss": 11.7171,
      "step": 33685
    },
    {
      "epoch": 1.8343373288953326,
      "grad_norm": 0.5282363304151417,
      "learning_rate": 3.576417423505807e-06,
      "loss": 11.6159,
      "step": 33686
    },
    {
      "epoch": 1.8343917828919156,
      "grad_norm": 0.5251180636048071,
      "learning_rate": 3.5740805561850445e-06,
      "loss": 11.9031,
      "step": 33687
    },
    {
      "epoch": 1.8344462368884986,
      "grad_norm": 0.622908629364273,
      "learning_rate": 3.571744438688418e-06,
      "loss": 12.027,
      "step": 33688
    },
    {
      "epoch": 1.8345006908850816,
      "grad_norm": 0.5142626164041925,
      "learning_rate": 3.5694090710340576e-06,
      "loss": 11.6893,
      "step": 33689
    },
    {
      "epoch": 1.8345551448816648,
      "grad_norm": 0.5283159116676002,
      "learning_rate": 3.567074453240149e-06,
      "loss": 11.6854,
      "step": 33690
    },
    {
      "epoch": 1.8346095988782478,
      "grad_norm": 0.5511602178658775,
      "learning_rate": 3.564740585324855e-06,
      "loss": 11.6854,
      "step": 33691
    },
    {
      "epoch": 1.8346640528748308,
      "grad_norm": 0.6081058223194803,
      "learning_rate": 3.562407467306295e-06,
      "loss": 11.8995,
      "step": 33692
    },
    {
      "epoch": 1.8347185068714138,
      "grad_norm": 0.6237050189656914,
      "learning_rate": 3.560075099202642e-06,
      "loss": 11.8137,
      "step": 33693
    },
    {
      "epoch": 1.8347729608679968,
      "grad_norm": 0.5849995900037084,
      "learning_rate": 3.5577434810320055e-06,
      "loss": 11.8858,
      "step": 33694
    },
    {
      "epoch": 1.8348274148645798,
      "grad_norm": 0.5450710006351104,
      "learning_rate": 3.5554126128125256e-06,
      "loss": 11.9415,
      "step": 33695
    },
    {
      "epoch": 1.8348818688611628,
      "grad_norm": 0.8510014482508347,
      "learning_rate": 3.5530824945623542e-06,
      "loss": 11.8374,
      "step": 33696
    },
    {
      "epoch": 1.8349363228577458,
      "grad_norm": 0.5490630740266585,
      "learning_rate": 3.5507531262995553e-06,
      "loss": 11.8302,
      "step": 33697
    },
    {
      "epoch": 1.8349907768543288,
      "grad_norm": 0.6000585345043453,
      "learning_rate": 3.5484245080423027e-06,
      "loss": 11.9073,
      "step": 33698
    },
    {
      "epoch": 1.8350452308509118,
      "grad_norm": 0.5315158199283085,
      "learning_rate": 3.5460966398086602e-06,
      "loss": 11.8954,
      "step": 33699
    },
    {
      "epoch": 1.8350996848474947,
      "grad_norm": 0.5903353983840656,
      "learning_rate": 3.543769521616758e-06,
      "loss": 11.7407,
      "step": 33700
    },
    {
      "epoch": 1.8351541388440777,
      "grad_norm": 0.5577473741100369,
      "learning_rate": 3.541443153484658e-06,
      "loss": 11.8393,
      "step": 33701
    },
    {
      "epoch": 1.8352085928406607,
      "grad_norm": 0.5673069611908624,
      "learning_rate": 3.5391175354304807e-06,
      "loss": 11.7718,
      "step": 33702
    },
    {
      "epoch": 1.8352630468372437,
      "grad_norm": 0.6519382976389545,
      "learning_rate": 3.5367926674723216e-06,
      "loss": 11.9164,
      "step": 33703
    },
    {
      "epoch": 1.8353175008338267,
      "grad_norm": 0.5968121862080892,
      "learning_rate": 3.5344685496282227e-06,
      "loss": 11.7602,
      "step": 33704
    },
    {
      "epoch": 1.8353719548304097,
      "grad_norm": 0.6036063004325238,
      "learning_rate": 3.532145181916269e-06,
      "loss": 11.6866,
      "step": 33705
    },
    {
      "epoch": 1.8354264088269927,
      "grad_norm": 0.5781734630937158,
      "learning_rate": 3.5298225643545457e-06,
      "loss": 11.8314,
      "step": 33706
    },
    {
      "epoch": 1.8354808628235757,
      "grad_norm": 0.5272934985316827,
      "learning_rate": 3.5275006969610835e-06,
      "loss": 11.7894,
      "step": 33707
    },
    {
      "epoch": 1.835535316820159,
      "grad_norm": 0.5774879294105703,
      "learning_rate": 3.5251795797539676e-06,
      "loss": 11.7908,
      "step": 33708
    },
    {
      "epoch": 1.835589770816742,
      "grad_norm": 0.5483932207367126,
      "learning_rate": 3.5228592127512285e-06,
      "loss": 11.7713,
      "step": 33709
    },
    {
      "epoch": 1.835644224813325,
      "grad_norm": 0.5514752429617766,
      "learning_rate": 3.5205395959709286e-06,
      "loss": 11.7251,
      "step": 33710
    },
    {
      "epoch": 1.835698678809908,
      "grad_norm": 0.5734875429848707,
      "learning_rate": 3.5182207294310654e-06,
      "loss": 11.7756,
      "step": 33711
    },
    {
      "epoch": 1.8357531328064909,
      "grad_norm": 0.5350863732956618,
      "learning_rate": 3.5159026131497132e-06,
      "loss": 11.7566,
      "step": 33712
    },
    {
      "epoch": 1.8358075868030739,
      "grad_norm": 0.5154652665206887,
      "learning_rate": 3.5135852471449016e-06,
      "loss": 11.7158,
      "step": 33713
    },
    {
      "epoch": 1.835862040799657,
      "grad_norm": 0.5818234333088478,
      "learning_rate": 3.5112686314346054e-06,
      "loss": 11.6471,
      "step": 33714
    },
    {
      "epoch": 1.83591649479624,
      "grad_norm": 0.5309050324678231,
      "learning_rate": 3.508952766036877e-06,
      "loss": 11.8418,
      "step": 33715
    },
    {
      "epoch": 1.835970948792823,
      "grad_norm": 0.5316833087325932,
      "learning_rate": 3.506637650969702e-06,
      "loss": 11.5931,
      "step": 33716
    },
    {
      "epoch": 1.836025402789406,
      "grad_norm": 0.508372623903475,
      "learning_rate": 3.5043232862510987e-06,
      "loss": 11.7174,
      "step": 33717
    },
    {
      "epoch": 1.836079856785989,
      "grad_norm": 0.5095887549123126,
      "learning_rate": 3.5020096718990756e-06,
      "loss": 11.6387,
      "step": 33718
    },
    {
      "epoch": 1.836134310782572,
      "grad_norm": 0.5077160094496781,
      "learning_rate": 3.499696807931585e-06,
      "loss": 11.7219,
      "step": 33719
    },
    {
      "epoch": 1.836188764779155,
      "grad_norm": 0.5156147150890515,
      "learning_rate": 3.4973846943666568e-06,
      "loss": 11.8598,
      "step": 33720
    },
    {
      "epoch": 1.836243218775738,
      "grad_norm": 0.5657735402184215,
      "learning_rate": 3.4950733312222315e-06,
      "loss": 11.7291,
      "step": 33721
    },
    {
      "epoch": 1.836297672772321,
      "grad_norm": 0.5027216645238088,
      "learning_rate": 3.492762718516307e-06,
      "loss": 11.7495,
      "step": 33722
    },
    {
      "epoch": 1.836352126768904,
      "grad_norm": 0.575887213552128,
      "learning_rate": 3.490452856266857e-06,
      "loss": 11.8109,
      "step": 33723
    },
    {
      "epoch": 1.836406580765487,
      "grad_norm": 0.6160079531206806,
      "learning_rate": 3.488143744491801e-06,
      "loss": 11.9721,
      "step": 33724
    },
    {
      "epoch": 1.83646103476207,
      "grad_norm": 0.5444621174267315,
      "learning_rate": 3.4858353832091463e-06,
      "loss": 11.9605,
      "step": 33725
    },
    {
      "epoch": 1.836515488758653,
      "grad_norm": 0.5286723973743807,
      "learning_rate": 3.483527772436812e-06,
      "loss": 11.7713,
      "step": 33726
    },
    {
      "epoch": 1.836569942755236,
      "grad_norm": 0.5236218450241388,
      "learning_rate": 3.481220912192762e-06,
      "loss": 11.827,
      "step": 33727
    },
    {
      "epoch": 1.836624396751819,
      "grad_norm": 0.517060319718189,
      "learning_rate": 3.4789148024949035e-06,
      "loss": 11.77,
      "step": 33728
    },
    {
      "epoch": 1.836678850748402,
      "grad_norm": 0.5859984114605573,
      "learning_rate": 3.4766094433612006e-06,
      "loss": 11.8325,
      "step": 33729
    },
    {
      "epoch": 1.836733304744985,
      "grad_norm": 0.6051276245372441,
      "learning_rate": 3.474304834809583e-06,
      "loss": 11.9341,
      "step": 33730
    },
    {
      "epoch": 1.8367877587415682,
      "grad_norm": 0.5844962191411933,
      "learning_rate": 3.4720009768579365e-06,
      "loss": 11.7248,
      "step": 33731
    },
    {
      "epoch": 1.8368422127381512,
      "grad_norm": 0.5563580638775164,
      "learning_rate": 3.4696978695242132e-06,
      "loss": 11.7365,
      "step": 33732
    },
    {
      "epoch": 1.8368966667347342,
      "grad_norm": 0.4947205237272991,
      "learning_rate": 3.46739551282631e-06,
      "loss": 11.7052,
      "step": 33733
    },
    {
      "epoch": 1.8369511207313172,
      "grad_norm": 0.6067653619978699,
      "learning_rate": 3.465093906782124e-06,
      "loss": 11.8687,
      "step": 33734
    },
    {
      "epoch": 1.8370055747279002,
      "grad_norm": 0.5250863734251067,
      "learning_rate": 3.462793051409552e-06,
      "loss": 11.6991,
      "step": 33735
    },
    {
      "epoch": 1.8370600287244832,
      "grad_norm": 0.5854503019613355,
      "learning_rate": 3.46049294672649e-06,
      "loss": 11.7324,
      "step": 33736
    },
    {
      "epoch": 1.8371144827210664,
      "grad_norm": 0.5107861565301598,
      "learning_rate": 3.4581935927508357e-06,
      "loss": 11.7062,
      "step": 33737
    },
    {
      "epoch": 1.8371689367176494,
      "grad_norm": 0.6099423778392281,
      "learning_rate": 3.455894989500441e-06,
      "loss": 11.8401,
      "step": 33738
    },
    {
      "epoch": 1.8372233907142324,
      "grad_norm": 0.5488953774915811,
      "learning_rate": 3.453597136993203e-06,
      "loss": 11.752,
      "step": 33739
    },
    {
      "epoch": 1.8372778447108153,
      "grad_norm": 0.5356423523111601,
      "learning_rate": 3.4513000352469848e-06,
      "loss": 11.7573,
      "step": 33740
    },
    {
      "epoch": 1.8373322987073983,
      "grad_norm": 0.5279503509526772,
      "learning_rate": 3.44900368427965e-06,
      "loss": 11.8028,
      "step": 33741
    },
    {
      "epoch": 1.8373867527039813,
      "grad_norm": 0.5434130940980396,
      "learning_rate": 3.4467080841090628e-06,
      "loss": 11.7407,
      "step": 33742
    },
    {
      "epoch": 1.8374412067005643,
      "grad_norm": 0.5766316413889403,
      "learning_rate": 3.4444132347530635e-06,
      "loss": 11.8449,
      "step": 33743
    },
    {
      "epoch": 1.8374956606971473,
      "grad_norm": 0.652101556792841,
      "learning_rate": 3.4421191362294824e-06,
      "loss": 11.93,
      "step": 33744
    },
    {
      "epoch": 1.8375501146937303,
      "grad_norm": 0.5960265972082481,
      "learning_rate": 3.4398257885561945e-06,
      "loss": 11.8512,
      "step": 33745
    },
    {
      "epoch": 1.8376045686903133,
      "grad_norm": 0.5640952024080866,
      "learning_rate": 3.4375331917510077e-06,
      "loss": 11.7464,
      "step": 33746
    },
    {
      "epoch": 1.8376590226868963,
      "grad_norm": 0.5098863373297707,
      "learning_rate": 3.435241345831752e-06,
      "loss": 11.7593,
      "step": 33747
    },
    {
      "epoch": 1.8377134766834793,
      "grad_norm": 0.5095515741351909,
      "learning_rate": 3.4329502508162583e-06,
      "loss": 11.6392,
      "step": 33748
    },
    {
      "epoch": 1.8377679306800623,
      "grad_norm": 0.542932740237173,
      "learning_rate": 3.4306599067223443e-06,
      "loss": 11.7267,
      "step": 33749
    },
    {
      "epoch": 1.8378223846766453,
      "grad_norm": 0.565006808595657,
      "learning_rate": 3.428370313567797e-06,
      "loss": 11.6211,
      "step": 33750
    },
    {
      "epoch": 1.8378768386732283,
      "grad_norm": 0.6050045356351809,
      "learning_rate": 3.426081471370435e-06,
      "loss": 11.8577,
      "step": 33751
    },
    {
      "epoch": 1.8379312926698113,
      "grad_norm": 0.583321263599445,
      "learning_rate": 3.423793380148077e-06,
      "loss": 11.7218,
      "step": 33752
    },
    {
      "epoch": 1.8379857466663942,
      "grad_norm": 0.5872353698009192,
      "learning_rate": 3.4215060399184986e-06,
      "loss": 11.7619,
      "step": 33753
    },
    {
      "epoch": 1.8380402006629775,
      "grad_norm": 0.5766620994994406,
      "learning_rate": 3.4192194506994733e-06,
      "loss": 11.9911,
      "step": 33754
    },
    {
      "epoch": 1.8380946546595605,
      "grad_norm": 0.5074922654203639,
      "learning_rate": 3.4169336125087994e-06,
      "loss": 11.6697,
      "step": 33755
    },
    {
      "epoch": 1.8381491086561434,
      "grad_norm": 0.5260217142404257,
      "learning_rate": 3.414648525364239e-06,
      "loss": 11.7082,
      "step": 33756
    },
    {
      "epoch": 1.8382035626527264,
      "grad_norm": 0.5470496184281105,
      "learning_rate": 3.41236418928359e-06,
      "loss": 11.7668,
      "step": 33757
    },
    {
      "epoch": 1.8382580166493094,
      "grad_norm": 0.5724853788041439,
      "learning_rate": 3.410080604284571e-06,
      "loss": 11.7921,
      "step": 33758
    },
    {
      "epoch": 1.8383124706458924,
      "grad_norm": 0.6009617179526612,
      "learning_rate": 3.4077977703849794e-06,
      "loss": 11.7534,
      "step": 33759
    },
    {
      "epoch": 1.8383669246424756,
      "grad_norm": 0.5783049663565869,
      "learning_rate": 3.405515687602534e-06,
      "loss": 11.736,
      "step": 33760
    },
    {
      "epoch": 1.8384213786390586,
      "grad_norm": 0.5730574125204992,
      "learning_rate": 3.4032343559549984e-06,
      "loss": 11.7719,
      "step": 33761
    },
    {
      "epoch": 1.8384758326356416,
      "grad_norm": 0.519669808350654,
      "learning_rate": 3.400953775460136e-06,
      "loss": 11.7636,
      "step": 33762
    },
    {
      "epoch": 1.8385302866322246,
      "grad_norm": 0.5964376024612283,
      "learning_rate": 3.398673946135644e-06,
      "loss": 11.8903,
      "step": 33763
    },
    {
      "epoch": 1.8385847406288076,
      "grad_norm": 0.5672682021379412,
      "learning_rate": 3.396394867999264e-06,
      "loss": 11.7035,
      "step": 33764
    },
    {
      "epoch": 1.8386391946253906,
      "grad_norm": 0.5265068891237084,
      "learning_rate": 3.3941165410687147e-06,
      "loss": 11.8036,
      "step": 33765
    },
    {
      "epoch": 1.8386936486219736,
      "grad_norm": 0.516687143355546,
      "learning_rate": 3.3918389653617043e-06,
      "loss": 11.7173,
      "step": 33766
    },
    {
      "epoch": 1.8387481026185566,
      "grad_norm": 0.5810793523202998,
      "learning_rate": 3.3895621408959745e-06,
      "loss": 11.8949,
      "step": 33767
    },
    {
      "epoch": 1.8388025566151396,
      "grad_norm": 0.578112063505665,
      "learning_rate": 3.3872860676891995e-06,
      "loss": 11.6485,
      "step": 33768
    },
    {
      "epoch": 1.8388570106117226,
      "grad_norm": 0.5520149915577371,
      "learning_rate": 3.3850107457590983e-06,
      "loss": 11.7525,
      "step": 33769
    },
    {
      "epoch": 1.8389114646083056,
      "grad_norm": 0.6627735492956288,
      "learning_rate": 3.3827361751233465e-06,
      "loss": 11.7934,
      "step": 33770
    },
    {
      "epoch": 1.8389659186048886,
      "grad_norm": 0.5528380821301336,
      "learning_rate": 3.3804623557996516e-06,
      "loss": 11.6964,
      "step": 33771
    },
    {
      "epoch": 1.8390203726014716,
      "grad_norm": 0.5441781566992544,
      "learning_rate": 3.3781892878056777e-06,
      "loss": 11.8651,
      "step": 33772
    },
    {
      "epoch": 1.8390748265980545,
      "grad_norm": 0.5268215498162374,
      "learning_rate": 3.37591697115911e-06,
      "loss": 11.8521,
      "step": 33773
    },
    {
      "epoch": 1.8391292805946375,
      "grad_norm": 0.5350371230736668,
      "learning_rate": 3.3736454058776236e-06,
      "loss": 11.8748,
      "step": 33774
    },
    {
      "epoch": 1.8391837345912205,
      "grad_norm": 0.5694064059618958,
      "learning_rate": 3.37137459197886e-06,
      "loss": 11.8134,
      "step": 33775
    },
    {
      "epoch": 1.8392381885878035,
      "grad_norm": 0.5457504647782292,
      "learning_rate": 3.3691045294805047e-06,
      "loss": 11.8112,
      "step": 33776
    },
    {
      "epoch": 1.8392926425843865,
      "grad_norm": 0.5434586038544973,
      "learning_rate": 3.3668352184001885e-06,
      "loss": 11.7694,
      "step": 33777
    },
    {
      "epoch": 1.8393470965809697,
      "grad_norm": 0.48286760906165666,
      "learning_rate": 3.3645666587555635e-06,
      "loss": 11.637,
      "step": 33778
    },
    {
      "epoch": 1.8394015505775527,
      "grad_norm": 0.5622188256202755,
      "learning_rate": 3.3622988505642826e-06,
      "loss": 11.798,
      "step": 33779
    },
    {
      "epoch": 1.8394560045741357,
      "grad_norm": 0.5629060755557711,
      "learning_rate": 3.360031793843965e-06,
      "loss": 11.7861,
      "step": 33780
    },
    {
      "epoch": 1.8395104585707187,
      "grad_norm": 0.6039799015888979,
      "learning_rate": 3.3577654886122524e-06,
      "loss": 11.7994,
      "step": 33781
    },
    {
      "epoch": 1.8395649125673017,
      "grad_norm": 0.5314482109570983,
      "learning_rate": 3.3554999348867633e-06,
      "loss": 11.7301,
      "step": 33782
    },
    {
      "epoch": 1.8396193665638847,
      "grad_norm": 0.5141102963073315,
      "learning_rate": 3.3532351326851174e-06,
      "loss": 11.7489,
      "step": 33783
    },
    {
      "epoch": 1.839673820560468,
      "grad_norm": 0.5368867589358306,
      "learning_rate": 3.3509710820249228e-06,
      "loss": 11.7278,
      "step": 33784
    },
    {
      "epoch": 1.839728274557051,
      "grad_norm": 0.5577536904149286,
      "learning_rate": 3.3487077829237655e-06,
      "loss": 11.8324,
      "step": 33785
    },
    {
      "epoch": 1.839782728553634,
      "grad_norm": 0.5442647554330254,
      "learning_rate": 3.346445235399287e-06,
      "loss": 11.7692,
      "step": 33786
    },
    {
      "epoch": 1.8398371825502169,
      "grad_norm": 0.6038344235159461,
      "learning_rate": 3.3441834394690507e-06,
      "loss": 11.76,
      "step": 33787
    },
    {
      "epoch": 1.8398916365467999,
      "grad_norm": 0.5063113196907719,
      "learning_rate": 3.341922395150643e-06,
      "loss": 11.8317,
      "step": 33788
    },
    {
      "epoch": 1.8399460905433829,
      "grad_norm": 0.5978406466617064,
      "learning_rate": 3.3396621024616714e-06,
      "loss": 11.8848,
      "step": 33789
    },
    {
      "epoch": 1.8400005445399659,
      "grad_norm": 0.5054431701294958,
      "learning_rate": 3.337402561419689e-06,
      "loss": 11.8459,
      "step": 33790
    },
    {
      "epoch": 1.8400549985365489,
      "grad_norm": 0.5702263084054048,
      "learning_rate": 3.3351437720422818e-06,
      "loss": 11.7451,
      "step": 33791
    },
    {
      "epoch": 1.8401094525331319,
      "grad_norm": 0.6097616800921006,
      "learning_rate": 3.3328857343470023e-06,
      "loss": 11.9097,
      "step": 33792
    },
    {
      "epoch": 1.8401639065297148,
      "grad_norm": 0.5396892024496525,
      "learning_rate": 3.330628448351414e-06,
      "loss": 11.7652,
      "step": 33793
    },
    {
      "epoch": 1.8402183605262978,
      "grad_norm": 0.6901732418284905,
      "learning_rate": 3.328371914073081e-06,
      "loss": 11.855,
      "step": 33794
    },
    {
      "epoch": 1.8402728145228808,
      "grad_norm": 0.597373235272736,
      "learning_rate": 3.3261161315295218e-06,
      "loss": 11.8066,
      "step": 33795
    },
    {
      "epoch": 1.8403272685194638,
      "grad_norm": 0.6055857919287467,
      "learning_rate": 3.3238611007383124e-06,
      "loss": 11.7263,
      "step": 33796
    },
    {
      "epoch": 1.8403817225160468,
      "grad_norm": 0.5341443804936415,
      "learning_rate": 3.3216068217169606e-06,
      "loss": 11.8828,
      "step": 33797
    },
    {
      "epoch": 1.8404361765126298,
      "grad_norm": 0.5368310268770802,
      "learning_rate": 3.3193532944830185e-06,
      "loss": 11.8119,
      "step": 33798
    },
    {
      "epoch": 1.8404906305092128,
      "grad_norm": 0.5166187888250507,
      "learning_rate": 3.317100519053984e-06,
      "loss": 11.6499,
      "step": 33799
    },
    {
      "epoch": 1.8405450845057958,
      "grad_norm": 0.5043569772314318,
      "learning_rate": 3.314848495447387e-06,
      "loss": 11.7501,
      "step": 33800
    },
    {
      "epoch": 1.840599538502379,
      "grad_norm": 0.5723431816021484,
      "learning_rate": 3.312597223680758e-06,
      "loss": 11.8072,
      "step": 33801
    },
    {
      "epoch": 1.840653992498962,
      "grad_norm": 0.5232662325831565,
      "learning_rate": 3.310346703771583e-06,
      "loss": 11.7075,
      "step": 33802
    },
    {
      "epoch": 1.840708446495545,
      "grad_norm": 0.5849260386456648,
      "learning_rate": 3.3080969357373703e-06,
      "loss": 11.8087,
      "step": 33803
    },
    {
      "epoch": 1.840762900492128,
      "grad_norm": 0.5810891250532745,
      "learning_rate": 3.305847919595606e-06,
      "loss": 11.6649,
      "step": 33804
    },
    {
      "epoch": 1.840817354488711,
      "grad_norm": 0.5432501955593974,
      "learning_rate": 3.3035996553637762e-06,
      "loss": 11.7504,
      "step": 33805
    },
    {
      "epoch": 1.840871808485294,
      "grad_norm": 0.5981020426794402,
      "learning_rate": 3.3013521430593884e-06,
      "loss": 11.736,
      "step": 33806
    },
    {
      "epoch": 1.8409262624818772,
      "grad_norm": 0.540309161120785,
      "learning_rate": 3.2991053826998853e-06,
      "loss": 11.7747,
      "step": 33807
    },
    {
      "epoch": 1.8409807164784602,
      "grad_norm": 0.5358183372556647,
      "learning_rate": 3.2968593743027744e-06,
      "loss": 11.8613,
      "step": 33808
    },
    {
      "epoch": 1.8410351704750432,
      "grad_norm": 0.6148948096617408,
      "learning_rate": 3.294614117885486e-06,
      "loss": 11.7305,
      "step": 33809
    },
    {
      "epoch": 1.8410896244716262,
      "grad_norm": 0.5049933350510138,
      "learning_rate": 3.2923696134654957e-06,
      "loss": 11.7738,
      "step": 33810
    },
    {
      "epoch": 1.8411440784682092,
      "grad_norm": 0.5862344141498278,
      "learning_rate": 3.290125861060267e-06,
      "loss": 11.7892,
      "step": 33811
    },
    {
      "epoch": 1.8411985324647921,
      "grad_norm": 0.5401173404633814,
      "learning_rate": 3.287882860687219e-06,
      "loss": 11.766,
      "step": 33812
    },
    {
      "epoch": 1.8412529864613751,
      "grad_norm": 0.5384693365841544,
      "learning_rate": 3.2856406123638496e-06,
      "loss": 11.72,
      "step": 33813
    },
    {
      "epoch": 1.8413074404579581,
      "grad_norm": 0.6024943728394331,
      "learning_rate": 3.283399116107533e-06,
      "loss": 11.8491,
      "step": 33814
    },
    {
      "epoch": 1.8413618944545411,
      "grad_norm": 0.5547636411800385,
      "learning_rate": 3.281158371935711e-06,
      "loss": 11.8517,
      "step": 33815
    },
    {
      "epoch": 1.8414163484511241,
      "grad_norm": 0.5204289649108008,
      "learning_rate": 3.278918379865847e-06,
      "loss": 11.7569,
      "step": 33816
    },
    {
      "epoch": 1.8414708024477071,
      "grad_norm": 0.5472886969893096,
      "learning_rate": 3.2766791399153175e-06,
      "loss": 11.6995,
      "step": 33817
    },
    {
      "epoch": 1.84152525644429,
      "grad_norm": 0.5206538728662238,
      "learning_rate": 3.2744406521015627e-06,
      "loss": 11.7028,
      "step": 33818
    },
    {
      "epoch": 1.841579710440873,
      "grad_norm": 0.596221335886329,
      "learning_rate": 3.272202916441969e-06,
      "loss": 11.9266,
      "step": 33819
    },
    {
      "epoch": 1.841634164437456,
      "grad_norm": 0.5045741931754221,
      "learning_rate": 3.2699659329539557e-06,
      "loss": 11.7785,
      "step": 33820
    },
    {
      "epoch": 1.841688618434039,
      "grad_norm": 0.5872171316368902,
      "learning_rate": 3.267729701654898e-06,
      "loss": 11.7731,
      "step": 33821
    },
    {
      "epoch": 1.841743072430622,
      "grad_norm": 0.5898172680419017,
      "learning_rate": 3.265494222562193e-06,
      "loss": 11.9146,
      "step": 33822
    },
    {
      "epoch": 1.841797526427205,
      "grad_norm": 0.5707552746372736,
      "learning_rate": 3.2632594956932603e-06,
      "loss": 11.6587,
      "step": 33823
    },
    {
      "epoch": 1.8418519804237883,
      "grad_norm": 0.52788861099732,
      "learning_rate": 3.2610255210654082e-06,
      "loss": 11.7568,
      "step": 33824
    },
    {
      "epoch": 1.8419064344203713,
      "grad_norm": 0.5824919257319955,
      "learning_rate": 3.2587922986960674e-06,
      "loss": 11.8044,
      "step": 33825
    },
    {
      "epoch": 1.8419608884169543,
      "grad_norm": 0.5163323818054658,
      "learning_rate": 3.2565598286025566e-06,
      "loss": 11.7038,
      "step": 33826
    },
    {
      "epoch": 1.8420153424135373,
      "grad_norm": 0.5501110666957616,
      "learning_rate": 3.2543281108022736e-06,
      "loss": 11.6753,
      "step": 33827
    },
    {
      "epoch": 1.8420697964101203,
      "grad_norm": 0.5496418342032785,
      "learning_rate": 3.2520971453125715e-06,
      "loss": 11.7697,
      "step": 33828
    },
    {
      "epoch": 1.8421242504067032,
      "grad_norm": 0.5626608410252125,
      "learning_rate": 3.2498669321507692e-06,
      "loss": 11.8011,
      "step": 33829
    },
    {
      "epoch": 1.8421787044032865,
      "grad_norm": 0.5368191917332996,
      "learning_rate": 3.2476374713342304e-06,
      "loss": 11.8879,
      "step": 33830
    },
    {
      "epoch": 1.8422331583998695,
      "grad_norm": 0.5143398354378745,
      "learning_rate": 3.2454087628802863e-06,
      "loss": 11.7049,
      "step": 33831
    },
    {
      "epoch": 1.8422876123964524,
      "grad_norm": 0.4836503017569531,
      "learning_rate": 3.2431808068062786e-06,
      "loss": 11.7691,
      "step": 33832
    },
    {
      "epoch": 1.8423420663930354,
      "grad_norm": 0.542519706301802,
      "learning_rate": 3.240953603129515e-06,
      "loss": 11.8129,
      "step": 33833
    },
    {
      "epoch": 1.8423965203896184,
      "grad_norm": 0.5948890937753298,
      "learning_rate": 3.238727151867338e-06,
      "loss": 11.9013,
      "step": 33834
    },
    {
      "epoch": 1.8424509743862014,
      "grad_norm": 0.6205790799442409,
      "learning_rate": 3.236501453037033e-06,
      "loss": 11.7787,
      "step": 33835
    },
    {
      "epoch": 1.8425054283827844,
      "grad_norm": 0.5399811529053937,
      "learning_rate": 3.23427650665592e-06,
      "loss": 11.7706,
      "step": 33836
    },
    {
      "epoch": 1.8425598823793674,
      "grad_norm": 0.5837166679594895,
      "learning_rate": 3.232052312741296e-06,
      "loss": 11.8098,
      "step": 33837
    },
    {
      "epoch": 1.8426143363759504,
      "grad_norm": 0.5234130122805322,
      "learning_rate": 3.2298288713104695e-06,
      "loss": 11.7363,
      "step": 33838
    },
    {
      "epoch": 1.8426687903725334,
      "grad_norm": 0.5453969860580936,
      "learning_rate": 3.227606182380716e-06,
      "loss": 11.8542,
      "step": 33839
    },
    {
      "epoch": 1.8427232443691164,
      "grad_norm": 0.5675145898980608,
      "learning_rate": 3.2253842459693318e-06,
      "loss": 11.7664,
      "step": 33840
    },
    {
      "epoch": 1.8427776983656994,
      "grad_norm": 0.5845154079088531,
      "learning_rate": 3.2231630620935814e-06,
      "loss": 11.8287,
      "step": 33841
    },
    {
      "epoch": 1.8428321523622824,
      "grad_norm": 0.5210611882694561,
      "learning_rate": 3.2209426307707515e-06,
      "loss": 11.5919,
      "step": 33842
    },
    {
      "epoch": 1.8428866063588654,
      "grad_norm": 0.576083556110766,
      "learning_rate": 3.218722952018094e-06,
      "loss": 11.8641,
      "step": 33843
    },
    {
      "epoch": 1.8429410603554484,
      "grad_norm": 0.506464158582543,
      "learning_rate": 3.2165040258528844e-06,
      "loss": 11.8904,
      "step": 33844
    },
    {
      "epoch": 1.8429955143520313,
      "grad_norm": 0.5790154077641596,
      "learning_rate": 3.2142858522923757e-06,
      "loss": 11.6134,
      "step": 33845
    },
    {
      "epoch": 1.8430499683486143,
      "grad_norm": 0.5140118067808708,
      "learning_rate": 3.2120684313537985e-06,
      "loss": 11.9575,
      "step": 33846
    },
    {
      "epoch": 1.8431044223451973,
      "grad_norm": 0.5090120031802001,
      "learning_rate": 3.2098517630544168e-06,
      "loss": 11.7169,
      "step": 33847
    },
    {
      "epoch": 1.8431588763417805,
      "grad_norm": 0.5135841133725867,
      "learning_rate": 3.2076358474114498e-06,
      "loss": 11.6483,
      "step": 33848
    },
    {
      "epoch": 1.8432133303383635,
      "grad_norm": 0.5670370227024023,
      "learning_rate": 3.2054206844421397e-06,
      "loss": 11.8771,
      "step": 33849
    },
    {
      "epoch": 1.8432677843349465,
      "grad_norm": 0.5503407544068725,
      "learning_rate": 3.203206274163717e-06,
      "loss": 11.8097,
      "step": 33850
    },
    {
      "epoch": 1.8433222383315295,
      "grad_norm": 0.5307854885737221,
      "learning_rate": 3.20099261659339e-06,
      "loss": 11.7264,
      "step": 33851
    },
    {
      "epoch": 1.8433766923281125,
      "grad_norm": 0.49715861760956076,
      "learning_rate": 3.1987797117483786e-06,
      "loss": 11.7581,
      "step": 33852
    },
    {
      "epoch": 1.8434311463246955,
      "grad_norm": 0.5831303357984639,
      "learning_rate": 3.196567559645891e-06,
      "loss": 11.9091,
      "step": 33853
    },
    {
      "epoch": 1.8434856003212787,
      "grad_norm": 0.5464664626966996,
      "learning_rate": 3.194356160303136e-06,
      "loss": 11.7487,
      "step": 33854
    },
    {
      "epoch": 1.8435400543178617,
      "grad_norm": 0.5775957701878826,
      "learning_rate": 3.192145513737299e-06,
      "loss": 11.8228,
      "step": 33855
    },
    {
      "epoch": 1.8435945083144447,
      "grad_norm": 0.619429479039788,
      "learning_rate": 3.189935619965567e-06,
      "loss": 11.7778,
      "step": 33856
    },
    {
      "epoch": 1.8436489623110277,
      "grad_norm": 0.5248449823582793,
      "learning_rate": 3.1877264790051377e-06,
      "loss": 11.7079,
      "step": 33857
    },
    {
      "epoch": 1.8437034163076107,
      "grad_norm": 0.5542833636643758,
      "learning_rate": 3.185518090873174e-06,
      "loss": 11.8628,
      "step": 33858
    },
    {
      "epoch": 1.8437578703041937,
      "grad_norm": 0.5158652188160496,
      "learning_rate": 3.183310455586852e-06,
      "loss": 11.7808,
      "step": 33859
    },
    {
      "epoch": 1.8438123243007767,
      "grad_norm": 0.544956279425335,
      "learning_rate": 3.1811035731633576e-06,
      "loss": 11.8081,
      "step": 33860
    },
    {
      "epoch": 1.8438667782973597,
      "grad_norm": 0.6838284135909499,
      "learning_rate": 3.1788974436198328e-06,
      "loss": 11.8219,
      "step": 33861
    },
    {
      "epoch": 1.8439212322939427,
      "grad_norm": 0.5159639415411428,
      "learning_rate": 3.1766920669734414e-06,
      "loss": 11.7425,
      "step": 33862
    },
    {
      "epoch": 1.8439756862905257,
      "grad_norm": 0.5035792350524115,
      "learning_rate": 3.1744874432413253e-06,
      "loss": 11.7774,
      "step": 33863
    },
    {
      "epoch": 1.8440301402871087,
      "grad_norm": 0.5575584635097174,
      "learning_rate": 3.1722835724406374e-06,
      "loss": 11.8778,
      "step": 33864
    },
    {
      "epoch": 1.8440845942836916,
      "grad_norm": 0.521595897479169,
      "learning_rate": 3.1700804545885087e-06,
      "loss": 11.8565,
      "step": 33865
    },
    {
      "epoch": 1.8441390482802746,
      "grad_norm": 0.5325072597234949,
      "learning_rate": 3.16787808970207e-06,
      "loss": 11.6101,
      "step": 33866
    },
    {
      "epoch": 1.8441935022768576,
      "grad_norm": 0.526921449884404,
      "learning_rate": 3.1656764777984625e-06,
      "loss": 11.785,
      "step": 33867
    },
    {
      "epoch": 1.8442479562734406,
      "grad_norm": 0.5604604463304292,
      "learning_rate": 3.1634756188947736e-06,
      "loss": 11.8091,
      "step": 33868
    },
    {
      "epoch": 1.8443024102700236,
      "grad_norm": 0.5237033650127261,
      "learning_rate": 3.161275513008155e-06,
      "loss": 11.7208,
      "step": 33869
    },
    {
      "epoch": 1.8443568642666066,
      "grad_norm": 0.6297718900730161,
      "learning_rate": 3.159076160155683e-06,
      "loss": 11.7882,
      "step": 33870
    },
    {
      "epoch": 1.8444113182631898,
      "grad_norm": 0.5738234071997348,
      "learning_rate": 3.1568775603544766e-06,
      "loss": 11.8193,
      "step": 33871
    },
    {
      "epoch": 1.8444657722597728,
      "grad_norm": 0.5641880449837706,
      "learning_rate": 3.1546797136216443e-06,
      "loss": 11.8338,
      "step": 33872
    },
    {
      "epoch": 1.8445202262563558,
      "grad_norm": 0.5413760584223677,
      "learning_rate": 3.1524826199742506e-06,
      "loss": 11.6959,
      "step": 33873
    },
    {
      "epoch": 1.8445746802529388,
      "grad_norm": 0.5616691097512236,
      "learning_rate": 3.1502862794294152e-06,
      "loss": 11.6755,
      "step": 33874
    },
    {
      "epoch": 1.8446291342495218,
      "grad_norm": 0.6246207200975681,
      "learning_rate": 3.1480906920041798e-06,
      "loss": 11.8428,
      "step": 33875
    },
    {
      "epoch": 1.8446835882461048,
      "grad_norm": 0.5401450050853716,
      "learning_rate": 3.1458958577156195e-06,
      "loss": 11.6716,
      "step": 33876
    },
    {
      "epoch": 1.844738042242688,
      "grad_norm": 0.5416520671529061,
      "learning_rate": 3.143701776580832e-06,
      "loss": 11.7409,
      "step": 33877
    },
    {
      "epoch": 1.844792496239271,
      "grad_norm": 0.49940488635913727,
      "learning_rate": 3.141508448616859e-06,
      "loss": 11.6558,
      "step": 33878
    },
    {
      "epoch": 1.844846950235854,
      "grad_norm": 0.5851468047936453,
      "learning_rate": 3.1393158738407645e-06,
      "loss": 11.7389,
      "step": 33879
    },
    {
      "epoch": 1.844901404232437,
      "grad_norm": 0.5432004950819189,
      "learning_rate": 3.13712405226958e-06,
      "loss": 11.9062,
      "step": 33880
    },
    {
      "epoch": 1.84495585822902,
      "grad_norm": 0.5686325387673025,
      "learning_rate": 3.1349329839203802e-06,
      "loss": 11.8332,
      "step": 33881
    },
    {
      "epoch": 1.845010312225603,
      "grad_norm": 0.5388373122055787,
      "learning_rate": 3.1327426688101733e-06,
      "loss": 11.6179,
      "step": 33882
    },
    {
      "epoch": 1.845064766222186,
      "grad_norm": 0.519201245880715,
      "learning_rate": 3.1305531069560025e-06,
      "loss": 11.8168,
      "step": 33883
    },
    {
      "epoch": 1.845119220218769,
      "grad_norm": 0.566674166251352,
      "learning_rate": 3.1283642983749085e-06,
      "loss": 11.7421,
      "step": 33884
    },
    {
      "epoch": 1.845173674215352,
      "grad_norm": 0.5437773880958092,
      "learning_rate": 3.1261762430838894e-06,
      "loss": 11.8667,
      "step": 33885
    },
    {
      "epoch": 1.845228128211935,
      "grad_norm": 0.5301400094827246,
      "learning_rate": 3.1239889410999644e-06,
      "loss": 11.7364,
      "step": 33886
    },
    {
      "epoch": 1.845282582208518,
      "grad_norm": 0.5028840908671239,
      "learning_rate": 3.121802392440165e-06,
      "loss": 11.8648,
      "step": 33887
    },
    {
      "epoch": 1.845337036205101,
      "grad_norm": 0.5392734574427442,
      "learning_rate": 3.1196165971214553e-06,
      "loss": 11.8577,
      "step": 33888
    },
    {
      "epoch": 1.845391490201684,
      "grad_norm": 0.528793753122846,
      "learning_rate": 3.1174315551608768e-06,
      "loss": 11.8459,
      "step": 33889
    },
    {
      "epoch": 1.845445944198267,
      "grad_norm": 0.5099363291544728,
      "learning_rate": 3.1152472665753717e-06,
      "loss": 11.8106,
      "step": 33890
    },
    {
      "epoch": 1.84550039819485,
      "grad_norm": 0.5899545497919645,
      "learning_rate": 3.113063731381971e-06,
      "loss": 11.7961,
      "step": 33891
    },
    {
      "epoch": 1.845554852191433,
      "grad_norm": 0.5196418330767255,
      "learning_rate": 3.1108809495976275e-06,
      "loss": 11.6222,
      "step": 33892
    },
    {
      "epoch": 1.8456093061880159,
      "grad_norm": 0.5522045974661178,
      "learning_rate": 3.108698921239317e-06,
      "loss": 11.6702,
      "step": 33893
    },
    {
      "epoch": 1.845663760184599,
      "grad_norm": 0.5347161207556804,
      "learning_rate": 3.1065176463240364e-06,
      "loss": 11.7245,
      "step": 33894
    },
    {
      "epoch": 1.845718214181182,
      "grad_norm": 0.5985366129011804,
      "learning_rate": 3.104337124868706e-06,
      "loss": 11.9679,
      "step": 33895
    },
    {
      "epoch": 1.845772668177765,
      "grad_norm": 0.5567806245290867,
      "learning_rate": 3.1021573568903007e-06,
      "loss": 11.8281,
      "step": 33896
    },
    {
      "epoch": 1.845827122174348,
      "grad_norm": 0.4984262442091683,
      "learning_rate": 3.0999783424057626e-06,
      "loss": 11.6549,
      "step": 33897
    },
    {
      "epoch": 1.845881576170931,
      "grad_norm": 0.5305676714975427,
      "learning_rate": 3.0978000814320452e-06,
      "loss": 11.8301,
      "step": 33898
    },
    {
      "epoch": 1.845936030167514,
      "grad_norm": 0.5835378027254757,
      "learning_rate": 3.09562257398609e-06,
      "loss": 11.9314,
      "step": 33899
    },
    {
      "epoch": 1.8459904841640973,
      "grad_norm": 0.538974559621088,
      "learning_rate": 3.093445820084817e-06,
      "loss": 11.8596,
      "step": 33900
    },
    {
      "epoch": 1.8460449381606803,
      "grad_norm": 0.6174274770695499,
      "learning_rate": 3.0912698197451796e-06,
      "loss": 11.8054,
      "step": 33901
    },
    {
      "epoch": 1.8460993921572633,
      "grad_norm": 0.4943718810476187,
      "learning_rate": 3.0890945729840524e-06,
      "loss": 11.7585,
      "step": 33902
    },
    {
      "epoch": 1.8461538461538463,
      "grad_norm": 0.5133081439897508,
      "learning_rate": 3.0869200798184004e-06,
      "loss": 11.8115,
      "step": 33903
    },
    {
      "epoch": 1.8462083001504292,
      "grad_norm": 0.57591531044245,
      "learning_rate": 3.08474634026511e-06,
      "loss": 11.8596,
      "step": 33904
    },
    {
      "epoch": 1.8462627541470122,
      "grad_norm": 0.5128937353643369,
      "learning_rate": 3.082573354341067e-06,
      "loss": 11.854,
      "step": 33905
    },
    {
      "epoch": 1.8463172081435952,
      "grad_norm": 0.5434425857331038,
      "learning_rate": 3.0804011220632033e-06,
      "loss": 11.7711,
      "step": 33906
    },
    {
      "epoch": 1.8463716621401782,
      "grad_norm": 0.6067532377719885,
      "learning_rate": 3.0782296434483825e-06,
      "loss": 11.7875,
      "step": 33907
    },
    {
      "epoch": 1.8464261161367612,
      "grad_norm": 0.5375928356579653,
      "learning_rate": 3.0760589185135026e-06,
      "loss": 11.8202,
      "step": 33908
    },
    {
      "epoch": 1.8464805701333442,
      "grad_norm": 0.5511883926322493,
      "learning_rate": 3.0738889472754493e-06,
      "loss": 11.8206,
      "step": 33909
    },
    {
      "epoch": 1.8465350241299272,
      "grad_norm": 0.5489226508382511,
      "learning_rate": 3.071719729751077e-06,
      "loss": 11.8662,
      "step": 33910
    },
    {
      "epoch": 1.8465894781265102,
      "grad_norm": 0.48990277090348355,
      "learning_rate": 3.0695512659572823e-06,
      "loss": 11.7124,
      "step": 33911
    },
    {
      "epoch": 1.8466439321230932,
      "grad_norm": 0.5500921390007175,
      "learning_rate": 3.0673835559109075e-06,
      "loss": 11.6711,
      "step": 33912
    },
    {
      "epoch": 1.8466983861196762,
      "grad_norm": 0.53862484676813,
      "learning_rate": 3.0652165996288174e-06,
      "loss": 11.8484,
      "step": 33913
    },
    {
      "epoch": 1.8467528401162592,
      "grad_norm": 0.5906898883728187,
      "learning_rate": 3.0630503971278646e-06,
      "loss": 11.8781,
      "step": 33914
    },
    {
      "epoch": 1.8468072941128422,
      "grad_norm": 0.6049249207166347,
      "learning_rate": 3.060884948424869e-06,
      "loss": 11.8281,
      "step": 33915
    },
    {
      "epoch": 1.8468617481094252,
      "grad_norm": 0.5412072270855967,
      "learning_rate": 3.0587202535367065e-06,
      "loss": 11.5532,
      "step": 33916
    },
    {
      "epoch": 1.8469162021060082,
      "grad_norm": 0.512322497233974,
      "learning_rate": 3.0565563124801745e-06,
      "loss": 11.7961,
      "step": 33917
    },
    {
      "epoch": 1.8469706561025914,
      "grad_norm": 0.5129580644807363,
      "learning_rate": 3.054393125272137e-06,
      "loss": 11.6918,
      "step": 33918
    },
    {
      "epoch": 1.8470251100991744,
      "grad_norm": 0.5634917997525085,
      "learning_rate": 3.052230691929381e-06,
      "loss": 11.814,
      "step": 33919
    },
    {
      "epoch": 1.8470795640957574,
      "grad_norm": 0.5260412748198389,
      "learning_rate": 3.050069012468737e-06,
      "loss": 11.7143,
      "step": 33920
    },
    {
      "epoch": 1.8471340180923403,
      "grad_norm": 0.5614031038340227,
      "learning_rate": 3.0479080869070253e-06,
      "loss": 11.8278,
      "step": 33921
    },
    {
      "epoch": 1.8471884720889233,
      "grad_norm": 0.5044524238585407,
      "learning_rate": 3.0457479152610324e-06,
      "loss": 11.7716,
      "step": 33922
    },
    {
      "epoch": 1.8472429260855066,
      "grad_norm": 0.4937868742172702,
      "learning_rate": 3.043588497547567e-06,
      "loss": 11.7005,
      "step": 33923
    },
    {
      "epoch": 1.8472973800820895,
      "grad_norm": 0.5206714157371767,
      "learning_rate": 3.041429833783427e-06,
      "loss": 11.7411,
      "step": 33924
    },
    {
      "epoch": 1.8473518340786725,
      "grad_norm": 0.5995436217804876,
      "learning_rate": 3.039271923985365e-06,
      "loss": 11.7636,
      "step": 33925
    },
    {
      "epoch": 1.8474062880752555,
      "grad_norm": 0.6395912019597665,
      "learning_rate": 3.037114768170202e-06,
      "loss": 11.7916,
      "step": 33926
    },
    {
      "epoch": 1.8474607420718385,
      "grad_norm": 0.5854515514875622,
      "learning_rate": 3.03495836635469e-06,
      "loss": 11.8173,
      "step": 33927
    },
    {
      "epoch": 1.8475151960684215,
      "grad_norm": 0.5693612498369074,
      "learning_rate": 3.0328027185556052e-06,
      "loss": 11.802,
      "step": 33928
    },
    {
      "epoch": 1.8475696500650045,
      "grad_norm": 0.5205507286134287,
      "learning_rate": 3.0306478247897008e-06,
      "loss": 11.7203,
      "step": 33929
    },
    {
      "epoch": 1.8476241040615875,
      "grad_norm": 0.5217218520518706,
      "learning_rate": 3.028493685073752e-06,
      "loss": 11.8448,
      "step": 33930
    },
    {
      "epoch": 1.8476785580581705,
      "grad_norm": 0.5338324715648306,
      "learning_rate": 3.0263402994244794e-06,
      "loss": 11.682,
      "step": 33931
    },
    {
      "epoch": 1.8477330120547535,
      "grad_norm": 0.5709604337486814,
      "learning_rate": 3.0241876678586577e-06,
      "loss": 11.8978,
      "step": 33932
    },
    {
      "epoch": 1.8477874660513365,
      "grad_norm": 0.5203366203077877,
      "learning_rate": 3.0220357903930297e-06,
      "loss": 11.8205,
      "step": 33933
    },
    {
      "epoch": 1.8478419200479195,
      "grad_norm": 0.5372597897618722,
      "learning_rate": 3.0198846670443037e-06,
      "loss": 11.8024,
      "step": 33934
    },
    {
      "epoch": 1.8478963740445025,
      "grad_norm": 0.543726202110921,
      "learning_rate": 3.017734297829211e-06,
      "loss": 11.7345,
      "step": 33935
    },
    {
      "epoch": 1.8479508280410855,
      "grad_norm": 0.5597512685153038,
      "learning_rate": 3.015584682764494e-06,
      "loss": 11.8128,
      "step": 33936
    },
    {
      "epoch": 1.8480052820376685,
      "grad_norm": 0.6448671965687806,
      "learning_rate": 3.013435821866839e-06,
      "loss": 11.6734,
      "step": 33937
    },
    {
      "epoch": 1.8480597360342514,
      "grad_norm": 0.5558892060875587,
      "learning_rate": 3.0112877151529884e-06,
      "loss": 11.7819,
      "step": 33938
    },
    {
      "epoch": 1.8481141900308344,
      "grad_norm": 0.5321331313157706,
      "learning_rate": 3.0091403626396177e-06,
      "loss": 11.7569,
      "step": 33939
    },
    {
      "epoch": 1.8481686440274174,
      "grad_norm": 0.6038253896564069,
      "learning_rate": 3.0069937643434354e-06,
      "loss": 11.7305,
      "step": 33940
    },
    {
      "epoch": 1.8482230980240006,
      "grad_norm": 0.5640110249845319,
      "learning_rate": 3.0048479202811398e-06,
      "loss": 11.7231,
      "step": 33941
    },
    {
      "epoch": 1.8482775520205836,
      "grad_norm": 0.5658637186911623,
      "learning_rate": 3.002702830469406e-06,
      "loss": 11.7794,
      "step": 33942
    },
    {
      "epoch": 1.8483320060171666,
      "grad_norm": 0.6092022829773311,
      "learning_rate": 3.000558494924932e-06,
      "loss": 11.894,
      "step": 33943
    },
    {
      "epoch": 1.8483864600137496,
      "grad_norm": 0.5247351580081084,
      "learning_rate": 2.9984149136643823e-06,
      "loss": 11.7366,
      "step": 33944
    },
    {
      "epoch": 1.8484409140103326,
      "grad_norm": 0.5024333038604187,
      "learning_rate": 2.9962720867044326e-06,
      "loss": 11.7122,
      "step": 33945
    },
    {
      "epoch": 1.8484953680069156,
      "grad_norm": 0.536140477341649,
      "learning_rate": 2.9941300140617246e-06,
      "loss": 11.7737,
      "step": 33946
    },
    {
      "epoch": 1.8485498220034988,
      "grad_norm": 0.5158297429179122,
      "learning_rate": 2.991988695752923e-06,
      "loss": 11.6573,
      "step": 33947
    },
    {
      "epoch": 1.8486042760000818,
      "grad_norm": 0.48013086023660484,
      "learning_rate": 2.9898481317947036e-06,
      "loss": 11.7769,
      "step": 33948
    },
    {
      "epoch": 1.8486587299966648,
      "grad_norm": 0.569550321927307,
      "learning_rate": 2.987708322203675e-06,
      "loss": 11.7834,
      "step": 33949
    },
    {
      "epoch": 1.8487131839932478,
      "grad_norm": 0.5193882620624866,
      "learning_rate": 2.985569266996513e-06,
      "loss": 11.8766,
      "step": 33950
    },
    {
      "epoch": 1.8487676379898308,
      "grad_norm": 0.5274351535341895,
      "learning_rate": 2.9834309661898264e-06,
      "loss": 11.6403,
      "step": 33951
    },
    {
      "epoch": 1.8488220919864138,
      "grad_norm": 0.5220389440061838,
      "learning_rate": 2.9812934198002463e-06,
      "loss": 11.7715,
      "step": 33952
    },
    {
      "epoch": 1.8488765459829968,
      "grad_norm": 0.5481192506398179,
      "learning_rate": 2.979156627844404e-06,
      "loss": 11.7027,
      "step": 33953
    },
    {
      "epoch": 1.8489309999795798,
      "grad_norm": 0.5675706386525854,
      "learning_rate": 2.977020590338908e-06,
      "loss": 11.8059,
      "step": 33954
    },
    {
      "epoch": 1.8489854539761628,
      "grad_norm": 0.5724062075791675,
      "learning_rate": 2.974885307300379e-06,
      "loss": 11.7369,
      "step": 33955
    },
    {
      "epoch": 1.8490399079727458,
      "grad_norm": 0.6104988207407752,
      "learning_rate": 2.9727507787454035e-06,
      "loss": 11.8075,
      "step": 33956
    },
    {
      "epoch": 1.8490943619693287,
      "grad_norm": 0.5096959022771089,
      "learning_rate": 2.970617004690579e-06,
      "loss": 11.7302,
      "step": 33957
    },
    {
      "epoch": 1.8491488159659117,
      "grad_norm": 0.5879912890201504,
      "learning_rate": 2.968483985152526e-06,
      "loss": 11.8342,
      "step": 33958
    },
    {
      "epoch": 1.8492032699624947,
      "grad_norm": 0.5704977956925356,
      "learning_rate": 2.9663517201478084e-06,
      "loss": 11.8169,
      "step": 33959
    },
    {
      "epoch": 1.8492577239590777,
      "grad_norm": 0.5562458096238031,
      "learning_rate": 2.964220209693014e-06,
      "loss": 11.6883,
      "step": 33960
    },
    {
      "epoch": 1.8493121779556607,
      "grad_norm": 0.5260987530055947,
      "learning_rate": 2.9620894538047175e-06,
      "loss": 11.6639,
      "step": 33961
    },
    {
      "epoch": 1.8493666319522437,
      "grad_norm": 0.538078585004661,
      "learning_rate": 2.9599594524994834e-06,
      "loss": 11.8035,
      "step": 33962
    },
    {
      "epoch": 1.8494210859488267,
      "grad_norm": 0.5188453130396761,
      "learning_rate": 2.9578302057938766e-06,
      "loss": 11.5551,
      "step": 33963
    },
    {
      "epoch": 1.84947553994541,
      "grad_norm": 0.5379328298430154,
      "learning_rate": 2.9557017137044617e-06,
      "loss": 11.7338,
      "step": 33964
    },
    {
      "epoch": 1.849529993941993,
      "grad_norm": 0.5754936847753943,
      "learning_rate": 2.953573976247781e-06,
      "loss": 11.719,
      "step": 33965
    },
    {
      "epoch": 1.849584447938576,
      "grad_norm": 0.5249416751920055,
      "learning_rate": 2.9514469934403876e-06,
      "loss": 11.6886,
      "step": 33966
    },
    {
      "epoch": 1.849638901935159,
      "grad_norm": 0.5380695841891844,
      "learning_rate": 2.949320765298813e-06,
      "loss": 11.7186,
      "step": 33967
    },
    {
      "epoch": 1.849693355931742,
      "grad_norm": 0.5157242662529193,
      "learning_rate": 2.9471952918395883e-06,
      "loss": 11.6475,
      "step": 33968
    },
    {
      "epoch": 1.8497478099283249,
      "grad_norm": 0.599758824239076,
      "learning_rate": 2.945070573079256e-06,
      "loss": 11.8665,
      "step": 33969
    },
    {
      "epoch": 1.849802263924908,
      "grad_norm": 0.54743572491124,
      "learning_rate": 2.942946609034336e-06,
      "loss": 11.7804,
      "step": 33970
    },
    {
      "epoch": 1.849856717921491,
      "grad_norm": 0.5556212574903687,
      "learning_rate": 2.940823399721326e-06,
      "loss": 11.7047,
      "step": 33971
    },
    {
      "epoch": 1.849911171918074,
      "grad_norm": 0.5043351902681977,
      "learning_rate": 2.938700945156769e-06,
      "loss": 11.8355,
      "step": 33972
    },
    {
      "epoch": 1.849965625914657,
      "grad_norm": 0.5355679036669441,
      "learning_rate": 2.9365792453571404e-06,
      "loss": 11.8139,
      "step": 33973
    },
    {
      "epoch": 1.85002007991124,
      "grad_norm": 0.5586999119186564,
      "learning_rate": 2.934458300338949e-06,
      "loss": 11.7065,
      "step": 33974
    },
    {
      "epoch": 1.850074533907823,
      "grad_norm": 0.5536597163442772,
      "learning_rate": 2.9323381101186933e-06,
      "loss": 11.7894,
      "step": 33975
    },
    {
      "epoch": 1.850128987904406,
      "grad_norm": 0.5836834594475118,
      "learning_rate": 2.9302186747128478e-06,
      "loss": 11.8819,
      "step": 33976
    },
    {
      "epoch": 1.850183441900989,
      "grad_norm": 0.5914927375099355,
      "learning_rate": 2.9280999941379005e-06,
      "loss": 11.6999,
      "step": 33977
    },
    {
      "epoch": 1.850237895897572,
      "grad_norm": 0.4948145150744575,
      "learning_rate": 2.9259820684103267e-06,
      "loss": 11.7372,
      "step": 33978
    },
    {
      "epoch": 1.850292349894155,
      "grad_norm": 0.5262052599923139,
      "learning_rate": 2.923864897546602e-06,
      "loss": 11.7061,
      "step": 33979
    },
    {
      "epoch": 1.850346803890738,
      "grad_norm": 0.6282740310847362,
      "learning_rate": 2.9217484815631803e-06,
      "loss": 11.8636,
      "step": 33980
    },
    {
      "epoch": 1.850401257887321,
      "grad_norm": 0.508782728275568,
      "learning_rate": 2.9196328204765145e-06,
      "loss": 11.8079,
      "step": 33981
    },
    {
      "epoch": 1.850455711883904,
      "grad_norm": 0.5279491061551861,
      "learning_rate": 2.9175179143030697e-06,
      "loss": 11.7281,
      "step": 33982
    },
    {
      "epoch": 1.850510165880487,
      "grad_norm": 0.49618291423348704,
      "learning_rate": 2.915403763059288e-06,
      "loss": 11.7713,
      "step": 33983
    },
    {
      "epoch": 1.85056461987707,
      "grad_norm": 0.5385394130196661,
      "learning_rate": 2.913290366761612e-06,
      "loss": 11.8213,
      "step": 33984
    },
    {
      "epoch": 1.850619073873653,
      "grad_norm": 0.5479233683614301,
      "learning_rate": 2.9111777254264728e-06,
      "loss": 11.7341,
      "step": 33985
    },
    {
      "epoch": 1.850673527870236,
      "grad_norm": 0.49936015882159296,
      "learning_rate": 2.9090658390702907e-06,
      "loss": 11.7457,
      "step": 33986
    },
    {
      "epoch": 1.850727981866819,
      "grad_norm": 0.5808296453329583,
      "learning_rate": 2.9069547077094972e-06,
      "loss": 11.773,
      "step": 33987
    },
    {
      "epoch": 1.8507824358634022,
      "grad_norm": 0.5293009174682723,
      "learning_rate": 2.904844331360501e-06,
      "loss": 11.8333,
      "step": 33988
    },
    {
      "epoch": 1.8508368898599852,
      "grad_norm": 0.5841705642277881,
      "learning_rate": 2.902734710039723e-06,
      "loss": 11.8187,
      "step": 33989
    },
    {
      "epoch": 1.8508913438565682,
      "grad_norm": 0.5649771813488085,
      "learning_rate": 2.9006258437635605e-06,
      "loss": 11.7901,
      "step": 33990
    },
    {
      "epoch": 1.8509457978531512,
      "grad_norm": 0.5302723487249812,
      "learning_rate": 2.8985177325484113e-06,
      "loss": 11.7943,
      "step": 33991
    },
    {
      "epoch": 1.8510002518497342,
      "grad_norm": 0.5866540247940978,
      "learning_rate": 2.8964103764106855e-06,
      "loss": 11.8323,
      "step": 33992
    },
    {
      "epoch": 1.8510547058463174,
      "grad_norm": 0.5551975874233183,
      "learning_rate": 2.894303775366736e-06,
      "loss": 11.9083,
      "step": 33993
    },
    {
      "epoch": 1.8511091598429004,
      "grad_norm": 0.5270400714667117,
      "learning_rate": 2.8921979294329825e-06,
      "loss": 11.8537,
      "step": 33994
    },
    {
      "epoch": 1.8511636138394834,
      "grad_norm": 0.5135052772250565,
      "learning_rate": 2.8900928386257906e-06,
      "loss": 11.71,
      "step": 33995
    },
    {
      "epoch": 1.8512180678360663,
      "grad_norm": 0.5771000790059908,
      "learning_rate": 2.887988502961503e-06,
      "loss": 11.7572,
      "step": 33996
    },
    {
      "epoch": 1.8512725218326493,
      "grad_norm": 0.5483909976668098,
      "learning_rate": 2.8858849224565164e-06,
      "loss": 11.8167,
      "step": 33997
    },
    {
      "epoch": 1.8513269758292323,
      "grad_norm": 0.5226034055040285,
      "learning_rate": 2.8837820971271634e-06,
      "loss": 11.7273,
      "step": 33998
    },
    {
      "epoch": 1.8513814298258153,
      "grad_norm": 0.541963955985011,
      "learning_rate": 2.881680026989808e-06,
      "loss": 11.7224,
      "step": 33999
    },
    {
      "epoch": 1.8514358838223983,
      "grad_norm": 0.5181706792953558,
      "learning_rate": 2.879578712060793e-06,
      "loss": 11.8442,
      "step": 34000
    },
    {
      "epoch": 1.8514903378189813,
      "grad_norm": 0.5399078815227443,
      "learning_rate": 2.877478152356472e-06,
      "loss": 11.7581,
      "step": 34001
    },
    {
      "epoch": 1.8515447918155643,
      "grad_norm": 0.5359867665277845,
      "learning_rate": 2.8753783478931653e-06,
      "loss": 11.7962,
      "step": 34002
    },
    {
      "epoch": 1.8515992458121473,
      "grad_norm": 0.5435760006232293,
      "learning_rate": 2.8732792986871925e-06,
      "loss": 11.7628,
      "step": 34003
    },
    {
      "epoch": 1.8516536998087303,
      "grad_norm": 0.555593989298099,
      "learning_rate": 2.8711810047549082e-06,
      "loss": 11.8577,
      "step": 34004
    },
    {
      "epoch": 1.8517081538053133,
      "grad_norm": 0.5285585549972435,
      "learning_rate": 2.8690834661125988e-06,
      "loss": 11.7192,
      "step": 34005
    },
    {
      "epoch": 1.8517626078018963,
      "grad_norm": 0.5503586250886979,
      "learning_rate": 2.8669866827765844e-06,
      "loss": 11.6285,
      "step": 34006
    },
    {
      "epoch": 1.8518170617984793,
      "grad_norm": 0.5363522402626031,
      "learning_rate": 2.864890654763175e-06,
      "loss": 11.7311,
      "step": 34007
    },
    {
      "epoch": 1.8518715157950623,
      "grad_norm": 0.5241239077671972,
      "learning_rate": 2.8627953820886567e-06,
      "loss": 11.5543,
      "step": 34008
    },
    {
      "epoch": 1.8519259697916453,
      "grad_norm": 0.5527898725009992,
      "learning_rate": 2.860700864769339e-06,
      "loss": 11.9241,
      "step": 34009
    },
    {
      "epoch": 1.8519804237882282,
      "grad_norm": 0.5655947024463986,
      "learning_rate": 2.8586071028214976e-06,
      "loss": 11.8699,
      "step": 34010
    },
    {
      "epoch": 1.8520348777848115,
      "grad_norm": 0.5864008975548566,
      "learning_rate": 2.85651409626142e-06,
      "loss": 11.8148,
      "step": 34011
    },
    {
      "epoch": 1.8520893317813945,
      "grad_norm": 0.5272738235541787,
      "learning_rate": 2.8544218451053816e-06,
      "loss": 11.7192,
      "step": 34012
    },
    {
      "epoch": 1.8521437857779774,
      "grad_norm": 0.545272321828681,
      "learning_rate": 2.8523303493696364e-06,
      "loss": 11.8962,
      "step": 34013
    },
    {
      "epoch": 1.8521982397745604,
      "grad_norm": 0.5912552431684458,
      "learning_rate": 2.850239609070482e-06,
      "loss": 11.7549,
      "step": 34014
    },
    {
      "epoch": 1.8522526937711434,
      "grad_norm": 0.4986214073980574,
      "learning_rate": 2.8481496242241502e-06,
      "loss": 11.8233,
      "step": 34015
    },
    {
      "epoch": 1.8523071477677264,
      "grad_norm": 0.5315990421146979,
      "learning_rate": 2.8460603948469057e-06,
      "loss": 11.805,
      "step": 34016
    },
    {
      "epoch": 1.8523616017643096,
      "grad_norm": 0.5277944879919373,
      "learning_rate": 2.8439719209549687e-06,
      "loss": 11.8399,
      "step": 34017
    },
    {
      "epoch": 1.8524160557608926,
      "grad_norm": 0.5307441067634918,
      "learning_rate": 2.841884202564604e-06,
      "loss": 11.8005,
      "step": 34018
    },
    {
      "epoch": 1.8524705097574756,
      "grad_norm": 0.5361733699957829,
      "learning_rate": 2.839797239692055e-06,
      "loss": 11.7477,
      "step": 34019
    },
    {
      "epoch": 1.8525249637540586,
      "grad_norm": 0.50796473172156,
      "learning_rate": 2.8377110323535293e-06,
      "loss": 11.3824,
      "step": 34020
    },
    {
      "epoch": 1.8525794177506416,
      "grad_norm": 0.5777575613399936,
      "learning_rate": 2.83562558056526e-06,
      "loss": 11.7881,
      "step": 34021
    },
    {
      "epoch": 1.8526338717472246,
      "grad_norm": 0.5001863028163757,
      "learning_rate": 2.833540884343455e-06,
      "loss": 11.6736,
      "step": 34022
    },
    {
      "epoch": 1.8526883257438076,
      "grad_norm": 0.5182572141374138,
      "learning_rate": 2.831456943704336e-06,
      "loss": 11.8024,
      "step": 34023
    },
    {
      "epoch": 1.8527427797403906,
      "grad_norm": 0.6047831763564729,
      "learning_rate": 2.829373758664089e-06,
      "loss": 11.8276,
      "step": 34024
    },
    {
      "epoch": 1.8527972337369736,
      "grad_norm": 0.5913716437327828,
      "learning_rate": 2.8272913292389457e-06,
      "loss": 11.8893,
      "step": 34025
    },
    {
      "epoch": 1.8528516877335566,
      "grad_norm": 0.5208796587409614,
      "learning_rate": 2.8252096554450824e-06,
      "loss": 11.7287,
      "step": 34026
    },
    {
      "epoch": 1.8529061417301396,
      "grad_norm": 0.535002590497656,
      "learning_rate": 2.8231287372986635e-06,
      "loss": 11.6704,
      "step": 34027
    },
    {
      "epoch": 1.8529605957267226,
      "grad_norm": 0.5129719394648435,
      "learning_rate": 2.82104857481591e-06,
      "loss": 11.6365,
      "step": 34028
    },
    {
      "epoch": 1.8530150497233056,
      "grad_norm": 0.5661020644053443,
      "learning_rate": 2.818969168012975e-06,
      "loss": 11.6811,
      "step": 34029
    },
    {
      "epoch": 1.8530695037198885,
      "grad_norm": 0.5854461566190753,
      "learning_rate": 2.8168905169060233e-06,
      "loss": 11.6504,
      "step": 34030
    },
    {
      "epoch": 1.8531239577164715,
      "grad_norm": 0.5553709904444278,
      "learning_rate": 2.8148126215112425e-06,
      "loss": 11.7144,
      "step": 34031
    },
    {
      "epoch": 1.8531784117130545,
      "grad_norm": 0.5810422576221066,
      "learning_rate": 2.812735481844764e-06,
      "loss": 11.8553,
      "step": 34032
    },
    {
      "epoch": 1.8532328657096375,
      "grad_norm": 0.5812792611574698,
      "learning_rate": 2.810659097922763e-06,
      "loss": 11.8353,
      "step": 34033
    },
    {
      "epoch": 1.8532873197062207,
      "grad_norm": 0.5702078686054134,
      "learning_rate": 2.8085834697613722e-06,
      "loss": 11.7031,
      "step": 34034
    },
    {
      "epoch": 1.8533417737028037,
      "grad_norm": 0.5366748486364615,
      "learning_rate": 2.8065085973767445e-06,
      "loss": 11.8238,
      "step": 34035
    },
    {
      "epoch": 1.8533962276993867,
      "grad_norm": 0.5374314304561377,
      "learning_rate": 2.804434480785001e-06,
      "loss": 11.8128,
      "step": 34036
    },
    {
      "epoch": 1.8534506816959697,
      "grad_norm": 0.5532755574199162,
      "learning_rate": 2.8023611200022616e-06,
      "loss": 11.6945,
      "step": 34037
    },
    {
      "epoch": 1.8535051356925527,
      "grad_norm": 0.5401358258063391,
      "learning_rate": 2.80028851504468e-06,
      "loss": 11.8425,
      "step": 34038
    },
    {
      "epoch": 1.8535595896891357,
      "grad_norm": 0.5763647670588261,
      "learning_rate": 2.798216665928333e-06,
      "loss": 11.7155,
      "step": 34039
    },
    {
      "epoch": 1.853614043685719,
      "grad_norm": 0.5766449647815882,
      "learning_rate": 2.796145572669362e-06,
      "loss": 11.7909,
      "step": 34040
    },
    {
      "epoch": 1.853668497682302,
      "grad_norm": 0.5725141589151067,
      "learning_rate": 2.7940752352838773e-06,
      "loss": 11.8821,
      "step": 34041
    },
    {
      "epoch": 1.853722951678885,
      "grad_norm": 0.6125597265303008,
      "learning_rate": 2.7920056537879547e-06,
      "loss": 11.8031,
      "step": 34042
    },
    {
      "epoch": 1.853777405675468,
      "grad_norm": 0.5361428075674681,
      "learning_rate": 2.7899368281977034e-06,
      "loss": 11.7827,
      "step": 34043
    },
    {
      "epoch": 1.8538318596720509,
      "grad_norm": 0.5198248211482072,
      "learning_rate": 2.7878687585291995e-06,
      "loss": 11.7022,
      "step": 34044
    },
    {
      "epoch": 1.8538863136686339,
      "grad_norm": 0.5079659696727752,
      "learning_rate": 2.7858014447985414e-06,
      "loss": 11.6905,
      "step": 34045
    },
    {
      "epoch": 1.8539407676652169,
      "grad_norm": 0.5385895893421295,
      "learning_rate": 2.783734887021783e-06,
      "loss": 11.7728,
      "step": 34046
    },
    {
      "epoch": 1.8539952216617999,
      "grad_norm": 0.5560019379130211,
      "learning_rate": 2.781669085215011e-06,
      "loss": 11.7799,
      "step": 34047
    },
    {
      "epoch": 1.8540496756583829,
      "grad_norm": 0.5357157989539586,
      "learning_rate": 2.779604039394279e-06,
      "loss": 11.7575,
      "step": 34048
    },
    {
      "epoch": 1.8541041296549658,
      "grad_norm": 0.5371506830369353,
      "learning_rate": 2.7775397495756527e-06,
      "loss": 11.7611,
      "step": 34049
    },
    {
      "epoch": 1.8541585836515488,
      "grad_norm": 0.5161986694474797,
      "learning_rate": 2.7754762157751857e-06,
      "loss": 11.6191,
      "step": 34050
    },
    {
      "epoch": 1.8542130376481318,
      "grad_norm": 0.5605193402336714,
      "learning_rate": 2.7734134380089093e-06,
      "loss": 11.8143,
      "step": 34051
    },
    {
      "epoch": 1.8542674916447148,
      "grad_norm": 0.564727523872469,
      "learning_rate": 2.771351416292878e-06,
      "loss": 11.8044,
      "step": 34052
    },
    {
      "epoch": 1.8543219456412978,
      "grad_norm": 0.6114687241225171,
      "learning_rate": 2.7692901506431334e-06,
      "loss": 11.8897,
      "step": 34053
    },
    {
      "epoch": 1.8543763996378808,
      "grad_norm": 0.5843686735740672,
      "learning_rate": 2.767229641075675e-06,
      "loss": 11.6551,
      "step": 34054
    },
    {
      "epoch": 1.8544308536344638,
      "grad_norm": 0.5647573311623332,
      "learning_rate": 2.765169887606567e-06,
      "loss": 11.725,
      "step": 34055
    },
    {
      "epoch": 1.8544853076310468,
      "grad_norm": 0.5802162322802504,
      "learning_rate": 2.7631108902517964e-06,
      "loss": 11.7947,
      "step": 34056
    },
    {
      "epoch": 1.85453976162763,
      "grad_norm": 0.652859205557321,
      "learning_rate": 2.7610526490273738e-06,
      "loss": 11.7925,
      "step": 34057
    },
    {
      "epoch": 1.854594215624213,
      "grad_norm": 0.5545669111004263,
      "learning_rate": 2.7589951639493296e-06,
      "loss": 11.7274,
      "step": 34058
    },
    {
      "epoch": 1.854648669620796,
      "grad_norm": 0.5583813046794736,
      "learning_rate": 2.7569384350336293e-06,
      "loss": 11.7297,
      "step": 34059
    },
    {
      "epoch": 1.854703123617379,
      "grad_norm": 0.5484538012243809,
      "learning_rate": 2.754882462296293e-06,
      "loss": 11.6868,
      "step": 34060
    },
    {
      "epoch": 1.854757577613962,
      "grad_norm": 0.532367958735592,
      "learning_rate": 2.7528272457532865e-06,
      "loss": 11.7704,
      "step": 34061
    },
    {
      "epoch": 1.854812031610545,
      "grad_norm": 0.47254148780927535,
      "learning_rate": 2.7507727854206076e-06,
      "loss": 11.7237,
      "step": 34062
    },
    {
      "epoch": 1.8548664856071282,
      "grad_norm": 0.558732675017002,
      "learning_rate": 2.748719081314244e-06,
      "loss": 11.7849,
      "step": 34063
    },
    {
      "epoch": 1.8549209396037112,
      "grad_norm": 0.5876116258967456,
      "learning_rate": 2.7466661334501266e-06,
      "loss": 11.9338,
      "step": 34064
    },
    {
      "epoch": 1.8549753936002942,
      "grad_norm": 0.5205508541048451,
      "learning_rate": 2.7446139418442763e-06,
      "loss": 11.8928,
      "step": 34065
    },
    {
      "epoch": 1.8550298475968772,
      "grad_norm": 0.5501540084999282,
      "learning_rate": 2.7425625065125917e-06,
      "loss": 11.7058,
      "step": 34066
    },
    {
      "epoch": 1.8550843015934602,
      "grad_norm": 0.5275910836032256,
      "learning_rate": 2.7405118274710482e-06,
      "loss": 11.9045,
      "step": 34067
    },
    {
      "epoch": 1.8551387555900432,
      "grad_norm": 0.515798708229121,
      "learning_rate": 2.738461904735612e-06,
      "loss": 11.6111,
      "step": 34068
    },
    {
      "epoch": 1.8551932095866261,
      "grad_norm": 0.5428290845476124,
      "learning_rate": 2.7364127383221914e-06,
      "loss": 11.8164,
      "step": 34069
    },
    {
      "epoch": 1.8552476635832091,
      "grad_norm": 0.6048802628482521,
      "learning_rate": 2.7343643282467413e-06,
      "loss": 11.7328,
      "step": 34070
    },
    {
      "epoch": 1.8553021175797921,
      "grad_norm": 0.5607030518340594,
      "learning_rate": 2.7323166745251814e-06,
      "loss": 11.7448,
      "step": 34071
    },
    {
      "epoch": 1.8553565715763751,
      "grad_norm": 0.4982570699389137,
      "learning_rate": 2.7302697771734553e-06,
      "loss": 11.7343,
      "step": 34072
    },
    {
      "epoch": 1.8554110255729581,
      "grad_norm": 0.4921227757229111,
      "learning_rate": 2.72822363620745e-06,
      "loss": 11.6926,
      "step": 34073
    },
    {
      "epoch": 1.855465479569541,
      "grad_norm": 0.5654418146983228,
      "learning_rate": 2.7261782516430854e-06,
      "loss": 11.5966,
      "step": 34074
    },
    {
      "epoch": 1.855519933566124,
      "grad_norm": 0.7153251081472566,
      "learning_rate": 2.7241336234962944e-06,
      "loss": 11.7875,
      "step": 34075
    },
    {
      "epoch": 1.855574387562707,
      "grad_norm": 0.49568571008554196,
      "learning_rate": 2.7220897517829303e-06,
      "loss": 11.7951,
      "step": 34076
    },
    {
      "epoch": 1.85562884155929,
      "grad_norm": 0.5459610465500246,
      "learning_rate": 2.720046636518925e-06,
      "loss": 11.7583,
      "step": 34077
    },
    {
      "epoch": 1.855683295555873,
      "grad_norm": 0.52765938688209,
      "learning_rate": 2.718004277720143e-06,
      "loss": 11.7692,
      "step": 34078
    },
    {
      "epoch": 1.855737749552456,
      "grad_norm": 0.5497646074992447,
      "learning_rate": 2.7159626754024615e-06,
      "loss": 11.7232,
      "step": 34079
    },
    {
      "epoch": 1.855792203549039,
      "grad_norm": 0.510688996790669,
      "learning_rate": 2.713921829581789e-06,
      "loss": 11.6517,
      "step": 34080
    },
    {
      "epoch": 1.8558466575456223,
      "grad_norm": 0.4958188034929215,
      "learning_rate": 2.7118817402739695e-06,
      "loss": 11.7161,
      "step": 34081
    },
    {
      "epoch": 1.8559011115422053,
      "grad_norm": 0.49682322626932107,
      "learning_rate": 2.7098424074948782e-06,
      "loss": 11.7348,
      "step": 34082
    },
    {
      "epoch": 1.8559555655387883,
      "grad_norm": 0.5405829211041254,
      "learning_rate": 2.707803831260347e-06,
      "loss": 11.6823,
      "step": 34083
    },
    {
      "epoch": 1.8560100195353713,
      "grad_norm": 0.5487267444143137,
      "learning_rate": 2.7057660115862748e-06,
      "loss": 11.717,
      "step": 34084
    },
    {
      "epoch": 1.8560644735319543,
      "grad_norm": 0.5829987937994934,
      "learning_rate": 2.703728948488471e-06,
      "loss": 11.6097,
      "step": 34085
    },
    {
      "epoch": 1.8561189275285372,
      "grad_norm": 0.5683107597925958,
      "learning_rate": 2.7016926419827892e-06,
      "loss": 11.5456,
      "step": 34086
    },
    {
      "epoch": 1.8561733815251205,
      "grad_norm": 0.5247377964798963,
      "learning_rate": 2.6996570920850615e-06,
      "loss": 11.6565,
      "step": 34087
    },
    {
      "epoch": 1.8562278355217035,
      "grad_norm": 0.5438516098119133,
      "learning_rate": 2.697622298811109e-06,
      "loss": 11.7745,
      "step": 34088
    },
    {
      "epoch": 1.8562822895182864,
      "grad_norm": 0.5375365510810982,
      "learning_rate": 2.6955882621767627e-06,
      "loss": 11.7952,
      "step": 34089
    },
    {
      "epoch": 1.8563367435148694,
      "grad_norm": 0.5472913557677537,
      "learning_rate": 2.6935549821978545e-06,
      "loss": 11.8205,
      "step": 34090
    },
    {
      "epoch": 1.8563911975114524,
      "grad_norm": 0.5682658780083004,
      "learning_rate": 2.6915224588901723e-06,
      "loss": 11.8902,
      "step": 34091
    },
    {
      "epoch": 1.8564456515080354,
      "grad_norm": 0.5987486409175985,
      "learning_rate": 2.689490692269536e-06,
      "loss": 11.7818,
      "step": 34092
    },
    {
      "epoch": 1.8565001055046184,
      "grad_norm": 0.5172981863585994,
      "learning_rate": 2.687459682351734e-06,
      "loss": 11.7375,
      "step": 34093
    },
    {
      "epoch": 1.8565545595012014,
      "grad_norm": 0.5655138168649811,
      "learning_rate": 2.685429429152575e-06,
      "loss": 11.6837,
      "step": 34094
    },
    {
      "epoch": 1.8566090134977844,
      "grad_norm": 0.566723333506846,
      "learning_rate": 2.6833999326878244e-06,
      "loss": 11.8561,
      "step": 34095
    },
    {
      "epoch": 1.8566634674943674,
      "grad_norm": 0.508191075909932,
      "learning_rate": 2.681371192973281e-06,
      "loss": 11.499,
      "step": 34096
    },
    {
      "epoch": 1.8567179214909504,
      "grad_norm": 0.5461943753485724,
      "learning_rate": 2.6793432100247094e-06,
      "loss": 11.6532,
      "step": 34097
    },
    {
      "epoch": 1.8567723754875334,
      "grad_norm": 0.5362114674594104,
      "learning_rate": 2.6773159838578867e-06,
      "loss": 11.7125,
      "step": 34098
    },
    {
      "epoch": 1.8568268294841164,
      "grad_norm": 0.566289879168616,
      "learning_rate": 2.675289514488588e-06,
      "loss": 11.7581,
      "step": 34099
    },
    {
      "epoch": 1.8568812834806994,
      "grad_norm": 0.5526048342528732,
      "learning_rate": 2.673263801932546e-06,
      "loss": 11.8534,
      "step": 34100
    },
    {
      "epoch": 1.8569357374772824,
      "grad_norm": 0.5463096012916056,
      "learning_rate": 2.671238846205526e-06,
      "loss": 11.8857,
      "step": 34101
    },
    {
      "epoch": 1.8569901914738653,
      "grad_norm": 0.5366637513217796,
      "learning_rate": 2.6692146473232816e-06,
      "loss": 11.8138,
      "step": 34102
    },
    {
      "epoch": 1.8570446454704483,
      "grad_norm": 0.6373991308782505,
      "learning_rate": 2.667191205301545e-06,
      "loss": 11.8892,
      "step": 34103
    },
    {
      "epoch": 1.8570990994670316,
      "grad_norm": 0.5667846566917036,
      "learning_rate": 2.6651685201560583e-06,
      "loss": 11.7853,
      "step": 34104
    },
    {
      "epoch": 1.8571535534636145,
      "grad_norm": 0.5419256198203309,
      "learning_rate": 2.6631465919025433e-06,
      "loss": 11.6945,
      "step": 34105
    },
    {
      "epoch": 1.8572080074601975,
      "grad_norm": 0.5776209773524162,
      "learning_rate": 2.661125420556709e-06,
      "loss": 11.7391,
      "step": 34106
    },
    {
      "epoch": 1.8572624614567805,
      "grad_norm": 0.5877709735118014,
      "learning_rate": 2.6591050061342994e-06,
      "loss": 11.8664,
      "step": 34107
    },
    {
      "epoch": 1.8573169154533635,
      "grad_norm": 0.5876932201534659,
      "learning_rate": 2.6570853486510115e-06,
      "loss": 11.7654,
      "step": 34108
    },
    {
      "epoch": 1.8573713694499465,
      "grad_norm": 0.5421207030376842,
      "learning_rate": 2.6550664481225564e-06,
      "loss": 11.6486,
      "step": 34109
    },
    {
      "epoch": 1.8574258234465297,
      "grad_norm": 0.5299217307624858,
      "learning_rate": 2.6530483045646205e-06,
      "loss": 11.6776,
      "step": 34110
    },
    {
      "epoch": 1.8574802774431127,
      "grad_norm": 0.5534014212715822,
      "learning_rate": 2.6510309179929025e-06,
      "loss": 11.808,
      "step": 34111
    },
    {
      "epoch": 1.8575347314396957,
      "grad_norm": 0.5490184703527806,
      "learning_rate": 2.6490142884231016e-06,
      "loss": 11.7646,
      "step": 34112
    },
    {
      "epoch": 1.8575891854362787,
      "grad_norm": 0.5146089147151446,
      "learning_rate": 2.6469984158708825e-06,
      "loss": 11.8306,
      "step": 34113
    },
    {
      "epoch": 1.8576436394328617,
      "grad_norm": 0.5469835749781836,
      "learning_rate": 2.6449833003519444e-06,
      "loss": 11.8555,
      "step": 34114
    },
    {
      "epoch": 1.8576980934294447,
      "grad_norm": 0.5818258261115045,
      "learning_rate": 2.6429689418819403e-06,
      "loss": 11.7264,
      "step": 34115
    },
    {
      "epoch": 1.8577525474260277,
      "grad_norm": 0.5738309420067346,
      "learning_rate": 2.6409553404765365e-06,
      "loss": 11.7786,
      "step": 34116
    },
    {
      "epoch": 1.8578070014226107,
      "grad_norm": 0.5532584211809648,
      "learning_rate": 2.6389424961513866e-06,
      "loss": 11.7642,
      "step": 34117
    },
    {
      "epoch": 1.8578614554191937,
      "grad_norm": 0.5342596315396536,
      "learning_rate": 2.6369304089221447e-06,
      "loss": 11.7343,
      "step": 34118
    },
    {
      "epoch": 1.8579159094157767,
      "grad_norm": 0.6193558740509785,
      "learning_rate": 2.6349190788044652e-06,
      "loss": 11.7922,
      "step": 34119
    },
    {
      "epoch": 1.8579703634123597,
      "grad_norm": 0.5782423910365604,
      "learning_rate": 2.63290850581398e-06,
      "loss": 11.7755,
      "step": 34120
    },
    {
      "epoch": 1.8580248174089427,
      "grad_norm": 0.5145585207300134,
      "learning_rate": 2.6308986899663323e-06,
      "loss": 11.6257,
      "step": 34121
    },
    {
      "epoch": 1.8580792714055256,
      "grad_norm": 0.5637859053969433,
      "learning_rate": 2.6288896312771315e-06,
      "loss": 11.871,
      "step": 34122
    },
    {
      "epoch": 1.8581337254021086,
      "grad_norm": 0.4941110555645744,
      "learning_rate": 2.626881329762021e-06,
      "loss": 11.6426,
      "step": 34123
    },
    {
      "epoch": 1.8581881793986916,
      "grad_norm": 0.5744780816570595,
      "learning_rate": 2.62487378543661e-06,
      "loss": 11.8303,
      "step": 34124
    },
    {
      "epoch": 1.8582426333952746,
      "grad_norm": 0.598086808279827,
      "learning_rate": 2.622866998316509e-06,
      "loss": 11.6945,
      "step": 34125
    },
    {
      "epoch": 1.8582970873918576,
      "grad_norm": 0.5973845623077279,
      "learning_rate": 2.6208609684173493e-06,
      "loss": 11.6114,
      "step": 34126
    },
    {
      "epoch": 1.8583515413884408,
      "grad_norm": 0.5666796603852995,
      "learning_rate": 2.618855695754674e-06,
      "loss": 11.8497,
      "step": 34127
    },
    {
      "epoch": 1.8584059953850238,
      "grad_norm": 0.5428249603268561,
      "learning_rate": 2.6168511803441154e-06,
      "loss": 11.62,
      "step": 34128
    },
    {
      "epoch": 1.8584604493816068,
      "grad_norm": 0.5146541338848857,
      "learning_rate": 2.6148474222012608e-06,
      "loss": 11.7109,
      "step": 34129
    },
    {
      "epoch": 1.8585149033781898,
      "grad_norm": 0.5506291667570146,
      "learning_rate": 2.612844421341676e-06,
      "loss": 11.8329,
      "step": 34130
    },
    {
      "epoch": 1.8585693573747728,
      "grad_norm": 0.5131219078540313,
      "learning_rate": 2.610842177780948e-06,
      "loss": 11.8199,
      "step": 34131
    },
    {
      "epoch": 1.8586238113713558,
      "grad_norm": 0.567544050071326,
      "learning_rate": 2.608840691534642e-06,
      "loss": 11.7734,
      "step": 34132
    },
    {
      "epoch": 1.858678265367939,
      "grad_norm": 0.5105797380134522,
      "learning_rate": 2.6068399626183236e-06,
      "loss": 11.7357,
      "step": 34133
    },
    {
      "epoch": 1.858732719364522,
      "grad_norm": 0.5662576253554312,
      "learning_rate": 2.6048399910475473e-06,
      "loss": 11.7998,
      "step": 34134
    },
    {
      "epoch": 1.858787173361105,
      "grad_norm": 0.5163416056105855,
      "learning_rate": 2.602840776837867e-06,
      "loss": 11.7284,
      "step": 34135
    },
    {
      "epoch": 1.858841627357688,
      "grad_norm": 0.536067522575811,
      "learning_rate": 2.6008423200048483e-06,
      "loss": 11.8133,
      "step": 34136
    },
    {
      "epoch": 1.858896081354271,
      "grad_norm": 0.5424786276923304,
      "learning_rate": 2.5988446205640004e-06,
      "loss": 11.7822,
      "step": 34137
    },
    {
      "epoch": 1.858950535350854,
      "grad_norm": 0.5252799566672407,
      "learning_rate": 2.596847678530867e-06,
      "loss": 11.6414,
      "step": 34138
    },
    {
      "epoch": 1.859004989347437,
      "grad_norm": 0.5157648847759491,
      "learning_rate": 2.5948514939209912e-06,
      "loss": 11.6841,
      "step": 34139
    },
    {
      "epoch": 1.85905944334402,
      "grad_norm": 0.5395751269361226,
      "learning_rate": 2.5928560667498713e-06,
      "loss": 11.7793,
      "step": 34140
    },
    {
      "epoch": 1.859113897340603,
      "grad_norm": 0.5638058456493286,
      "learning_rate": 2.590861397033051e-06,
      "loss": 11.7167,
      "step": 34141
    },
    {
      "epoch": 1.859168351337186,
      "grad_norm": 0.5847364246820539,
      "learning_rate": 2.5888674847860175e-06,
      "loss": 11.6694,
      "step": 34142
    },
    {
      "epoch": 1.859222805333769,
      "grad_norm": 0.5103492098472038,
      "learning_rate": 2.5868743300242916e-06,
      "loss": 11.7623,
      "step": 34143
    },
    {
      "epoch": 1.859277259330352,
      "grad_norm": 0.5412922834575296,
      "learning_rate": 2.584881932763361e-06,
      "loss": 11.7553,
      "step": 34144
    },
    {
      "epoch": 1.859331713326935,
      "grad_norm": 0.5538789553502556,
      "learning_rate": 2.582890293018725e-06,
      "loss": 11.7341,
      "step": 34145
    },
    {
      "epoch": 1.859386167323518,
      "grad_norm": 0.5653406075227116,
      "learning_rate": 2.5808994108058925e-06,
      "loss": 11.7556,
      "step": 34146
    },
    {
      "epoch": 1.859440621320101,
      "grad_norm": 0.6198355925560457,
      "learning_rate": 2.5789092861403076e-06,
      "loss": 11.7568,
      "step": 34147
    },
    {
      "epoch": 1.859495075316684,
      "grad_norm": 0.5829700476640527,
      "learning_rate": 2.5769199190374683e-06,
      "loss": 11.8438,
      "step": 34148
    },
    {
      "epoch": 1.859549529313267,
      "grad_norm": 0.6709208864682427,
      "learning_rate": 2.574931309512818e-06,
      "loss": 11.7601,
      "step": 34149
    },
    {
      "epoch": 1.8596039833098499,
      "grad_norm": 0.576368123322428,
      "learning_rate": 2.5729434575818557e-06,
      "loss": 11.7127,
      "step": 34150
    },
    {
      "epoch": 1.859658437306433,
      "grad_norm": 0.5050539730265304,
      "learning_rate": 2.570956363260024e-06,
      "loss": 11.732,
      "step": 34151
    },
    {
      "epoch": 1.859712891303016,
      "grad_norm": 0.6046256561023708,
      "learning_rate": 2.5689700265627668e-06,
      "loss": 11.8598,
      "step": 34152
    },
    {
      "epoch": 1.859767345299599,
      "grad_norm": 0.5269572616654264,
      "learning_rate": 2.5669844475055492e-06,
      "loss": 11.6571,
      "step": 34153
    },
    {
      "epoch": 1.859821799296182,
      "grad_norm": 0.603658806901523,
      "learning_rate": 2.564999626103781e-06,
      "loss": 11.8524,
      "step": 34154
    },
    {
      "epoch": 1.859876253292765,
      "grad_norm": 0.5321904211696481,
      "learning_rate": 2.563015562372939e-06,
      "loss": 11.7855,
      "step": 34155
    },
    {
      "epoch": 1.8599307072893483,
      "grad_norm": 0.5061557068441046,
      "learning_rate": 2.5610322563284216e-06,
      "loss": 11.8365,
      "step": 34156
    },
    {
      "epoch": 1.8599851612859313,
      "grad_norm": 0.5852691174880951,
      "learning_rate": 2.559049707985639e-06,
      "loss": 11.7605,
      "step": 34157
    },
    {
      "epoch": 1.8600396152825143,
      "grad_norm": 0.5499744580013146,
      "learning_rate": 2.5570679173600564e-06,
      "loss": 11.8352,
      "step": 34158
    },
    {
      "epoch": 1.8600940692790973,
      "grad_norm": 0.5637411575606855,
      "learning_rate": 2.5550868844670283e-06,
      "loss": 11.775,
      "step": 34159
    },
    {
      "epoch": 1.8601485232756803,
      "grad_norm": 0.5444245383974731,
      "learning_rate": 2.5531066093219978e-06,
      "loss": 11.6991,
      "step": 34160
    },
    {
      "epoch": 1.8602029772722632,
      "grad_norm": 0.5262136559054372,
      "learning_rate": 2.551127091940353e-06,
      "loss": 11.7058,
      "step": 34161
    },
    {
      "epoch": 1.8602574312688462,
      "grad_norm": 0.6000498145071027,
      "learning_rate": 2.549148332337481e-06,
      "loss": 11.7554,
      "step": 34162
    },
    {
      "epoch": 1.8603118852654292,
      "grad_norm": 0.5196297525263693,
      "learning_rate": 2.5471703305287807e-06,
      "loss": 11.7863,
      "step": 34163
    },
    {
      "epoch": 1.8603663392620122,
      "grad_norm": 0.5747606300573432,
      "learning_rate": 2.5451930865296183e-06,
      "loss": 11.728,
      "step": 34164
    },
    {
      "epoch": 1.8604207932585952,
      "grad_norm": 0.5196508024223219,
      "learning_rate": 2.5432166003553914e-06,
      "loss": 11.71,
      "step": 34165
    },
    {
      "epoch": 1.8604752472551782,
      "grad_norm": 0.5213264834311555,
      "learning_rate": 2.5412408720214443e-06,
      "loss": 11.8188,
      "step": 34166
    },
    {
      "epoch": 1.8605297012517612,
      "grad_norm": 0.5298991715536169,
      "learning_rate": 2.539265901543153e-06,
      "loss": 11.8251,
      "step": 34167
    },
    {
      "epoch": 1.8605841552483442,
      "grad_norm": 0.5340164841269298,
      "learning_rate": 2.5372916889358835e-06,
      "loss": 11.7761,
      "step": 34168
    },
    {
      "epoch": 1.8606386092449272,
      "grad_norm": 0.5569992179735073,
      "learning_rate": 2.535318234214967e-06,
      "loss": 11.7586,
      "step": 34169
    },
    {
      "epoch": 1.8606930632415102,
      "grad_norm": 0.5866682630661555,
      "learning_rate": 2.5333455373957705e-06,
      "loss": 11.8225,
      "step": 34170
    },
    {
      "epoch": 1.8607475172380932,
      "grad_norm": 0.5428021841429663,
      "learning_rate": 2.5313735984936137e-06,
      "loss": 11.7837,
      "step": 34171
    },
    {
      "epoch": 1.8608019712346762,
      "grad_norm": 0.5527744381721219,
      "learning_rate": 2.529402417523841e-06,
      "loss": 11.699,
      "step": 34172
    },
    {
      "epoch": 1.8608564252312592,
      "grad_norm": 0.5296158164996029,
      "learning_rate": 2.527431994501783e-06,
      "loss": 11.7278,
      "step": 34173
    },
    {
      "epoch": 1.8609108792278424,
      "grad_norm": 0.576018565837309,
      "learning_rate": 2.5254623294427626e-06,
      "loss": 11.8261,
      "step": 34174
    },
    {
      "epoch": 1.8609653332244254,
      "grad_norm": 0.5740271902952466,
      "learning_rate": 2.5234934223621e-06,
      "loss": 11.7174,
      "step": 34175
    },
    {
      "epoch": 1.8610197872210084,
      "grad_norm": 0.4834014411927382,
      "learning_rate": 2.521525273275094e-06,
      "loss": 11.8187,
      "step": 34176
    },
    {
      "epoch": 1.8610742412175914,
      "grad_norm": 0.5667849462284822,
      "learning_rate": 2.5195578821970545e-06,
      "loss": 11.766,
      "step": 34177
    },
    {
      "epoch": 1.8611286952141743,
      "grad_norm": 0.5830541716647009,
      "learning_rate": 2.517591249143281e-06,
      "loss": 11.7631,
      "step": 34178
    },
    {
      "epoch": 1.8611831492107573,
      "grad_norm": 0.5263017125893658,
      "learning_rate": 2.5156253741290603e-06,
      "loss": 11.7173,
      "step": 34179
    },
    {
      "epoch": 1.8612376032073406,
      "grad_norm": 0.5570995837286824,
      "learning_rate": 2.513660257169692e-06,
      "loss": 11.7829,
      "step": 34180
    },
    {
      "epoch": 1.8612920572039235,
      "grad_norm": 0.543429935542524,
      "learning_rate": 2.5116958982804418e-06,
      "loss": 11.812,
      "step": 34181
    },
    {
      "epoch": 1.8613465112005065,
      "grad_norm": 0.5425701658297523,
      "learning_rate": 2.509732297476608e-06,
      "loss": 11.9944,
      "step": 34182
    },
    {
      "epoch": 1.8614009651970895,
      "grad_norm": 0.5524225480267694,
      "learning_rate": 2.507769454773434e-06,
      "loss": 11.8507,
      "step": 34183
    },
    {
      "epoch": 1.8614554191936725,
      "grad_norm": 0.5906724294506779,
      "learning_rate": 2.5058073701861864e-06,
      "loss": 11.8047,
      "step": 34184
    },
    {
      "epoch": 1.8615098731902555,
      "grad_norm": 0.5278422417604384,
      "learning_rate": 2.50384604373014e-06,
      "loss": 11.7586,
      "step": 34185
    },
    {
      "epoch": 1.8615643271868385,
      "grad_norm": 0.5351336110155367,
      "learning_rate": 2.501885475420551e-06,
      "loss": 11.7011,
      "step": 34186
    },
    {
      "epoch": 1.8616187811834215,
      "grad_norm": 0.5940612577929096,
      "learning_rate": 2.499925665272629e-06,
      "loss": 11.7994,
      "step": 34187
    },
    {
      "epoch": 1.8616732351800045,
      "grad_norm": 0.5494946382919955,
      "learning_rate": 2.49796661330165e-06,
      "loss": 11.827,
      "step": 34188
    },
    {
      "epoch": 1.8617276891765875,
      "grad_norm": 0.5304235029645848,
      "learning_rate": 2.4960083195228244e-06,
      "loss": 11.8589,
      "step": 34189
    },
    {
      "epoch": 1.8617821431731705,
      "grad_norm": 0.5175006356071478,
      "learning_rate": 2.494050783951396e-06,
      "loss": 11.7035,
      "step": 34190
    },
    {
      "epoch": 1.8618365971697535,
      "grad_norm": 0.5503796944672893,
      "learning_rate": 2.4920940066025635e-06,
      "loss": 11.8346,
      "step": 34191
    },
    {
      "epoch": 1.8618910511663365,
      "grad_norm": 0.5345422841907969,
      "learning_rate": 2.490137987491581e-06,
      "loss": 11.7668,
      "step": 34192
    },
    {
      "epoch": 1.8619455051629195,
      "grad_norm": 0.5448337000854672,
      "learning_rate": 2.4881827266336255e-06,
      "loss": 11.8119,
      "step": 34193
    },
    {
      "epoch": 1.8619999591595024,
      "grad_norm": 0.5575270582841075,
      "learning_rate": 2.486228224043918e-06,
      "loss": 11.9021,
      "step": 34194
    },
    {
      "epoch": 1.8620544131560854,
      "grad_norm": 0.525244067361692,
      "learning_rate": 2.484274479737658e-06,
      "loss": 11.6603,
      "step": 34195
    },
    {
      "epoch": 1.8621088671526684,
      "grad_norm": 0.5677293472556686,
      "learning_rate": 2.482321493730033e-06,
      "loss": 11.7894,
      "step": 34196
    },
    {
      "epoch": 1.8621633211492516,
      "grad_norm": 0.5553831084579393,
      "learning_rate": 2.480369266036231e-06,
      "loss": 11.6958,
      "step": 34197
    },
    {
      "epoch": 1.8622177751458346,
      "grad_norm": 0.5331850581872086,
      "learning_rate": 2.4784177966714172e-06,
      "loss": 11.9201,
      "step": 34198
    },
    {
      "epoch": 1.8622722291424176,
      "grad_norm": 0.5978388936682295,
      "learning_rate": 2.476467085650791e-06,
      "loss": 11.8646,
      "step": 34199
    },
    {
      "epoch": 1.8623266831390006,
      "grad_norm": 0.5316424538573272,
      "learning_rate": 2.4745171329895065e-06,
      "loss": 11.7897,
      "step": 34200
    },
    {
      "epoch": 1.8623811371355836,
      "grad_norm": 0.5333208931118412,
      "learning_rate": 2.47256793870273e-06,
      "loss": 11.8553,
      "step": 34201
    },
    {
      "epoch": 1.8624355911321666,
      "grad_norm": 0.5304161429780652,
      "learning_rate": 2.470619502805638e-06,
      "loss": 11.6802,
      "step": 34202
    },
    {
      "epoch": 1.8624900451287498,
      "grad_norm": 0.5419035587078287,
      "learning_rate": 2.46867182531334e-06,
      "loss": 11.8046,
      "step": 34203
    },
    {
      "epoch": 1.8625444991253328,
      "grad_norm": 0.5318397569660418,
      "learning_rate": 2.466724906241025e-06,
      "loss": 11.8833,
      "step": 34204
    },
    {
      "epoch": 1.8625989531219158,
      "grad_norm": 0.5124298265960896,
      "learning_rate": 2.464778745603802e-06,
      "loss": 11.6758,
      "step": 34205
    },
    {
      "epoch": 1.8626534071184988,
      "grad_norm": 0.5815998586177793,
      "learning_rate": 2.462833343416826e-06,
      "loss": 11.9073,
      "step": 34206
    },
    {
      "epoch": 1.8627078611150818,
      "grad_norm": 0.5996786164517552,
      "learning_rate": 2.4608886996952186e-06,
      "loss": 11.8514,
      "step": 34207
    },
    {
      "epoch": 1.8627623151116648,
      "grad_norm": 0.5399291136870131,
      "learning_rate": 2.4589448144540783e-06,
      "loss": 11.8548,
      "step": 34208
    },
    {
      "epoch": 1.8628167691082478,
      "grad_norm": 0.585185179304002,
      "learning_rate": 2.457001687708549e-06,
      "loss": 11.6782,
      "step": 34209
    },
    {
      "epoch": 1.8628712231048308,
      "grad_norm": 0.5246050621343618,
      "learning_rate": 2.4550593194737404e-06,
      "loss": 11.7869,
      "step": 34210
    },
    {
      "epoch": 1.8629256771014138,
      "grad_norm": 0.5029747833277742,
      "learning_rate": 2.4531177097647408e-06,
      "loss": 11.8503,
      "step": 34211
    },
    {
      "epoch": 1.8629801310979968,
      "grad_norm": 0.6242831355913958,
      "learning_rate": 2.45117685859666e-06,
      "loss": 11.8906,
      "step": 34212
    },
    {
      "epoch": 1.8630345850945798,
      "grad_norm": 0.5425892746415845,
      "learning_rate": 2.449236765984586e-06,
      "loss": 11.8247,
      "step": 34213
    },
    {
      "epoch": 1.8630890390911627,
      "grad_norm": 0.5155520131370054,
      "learning_rate": 2.447297431943607e-06,
      "loss": 11.8579,
      "step": 34214
    },
    {
      "epoch": 1.8631434930877457,
      "grad_norm": 0.5262307966690408,
      "learning_rate": 2.4453588564887996e-06,
      "loss": 11.6946,
      "step": 34215
    },
    {
      "epoch": 1.8631979470843287,
      "grad_norm": 0.5198146549994429,
      "learning_rate": 2.443421039635252e-06,
      "loss": 11.7159,
      "step": 34216
    },
    {
      "epoch": 1.8632524010809117,
      "grad_norm": 0.5144561516247748,
      "learning_rate": 2.4414839813980183e-06,
      "loss": 11.7513,
      "step": 34217
    },
    {
      "epoch": 1.8633068550774947,
      "grad_norm": 0.5160035491790321,
      "learning_rate": 2.439547681792154e-06,
      "loss": 11.7505,
      "step": 34218
    },
    {
      "epoch": 1.8633613090740777,
      "grad_norm": 0.4955868170545605,
      "learning_rate": 2.4376121408327458e-06,
      "loss": 11.7747,
      "step": 34219
    },
    {
      "epoch": 1.8634157630706607,
      "grad_norm": 0.5103738312816297,
      "learning_rate": 2.435677358534816e-06,
      "loss": 11.6458,
      "step": 34220
    },
    {
      "epoch": 1.863470217067244,
      "grad_norm": 0.5371214183596122,
      "learning_rate": 2.4337433349134077e-06,
      "loss": 11.5142,
      "step": 34221
    },
    {
      "epoch": 1.863524671063827,
      "grad_norm": 0.570535075632023,
      "learning_rate": 2.4318100699835977e-06,
      "loss": 11.8633,
      "step": 34222
    },
    {
      "epoch": 1.86357912506041,
      "grad_norm": 0.5840620998834162,
      "learning_rate": 2.4298775637603745e-06,
      "loss": 11.7923,
      "step": 34223
    },
    {
      "epoch": 1.863633579056993,
      "grad_norm": 0.5727108335166309,
      "learning_rate": 2.427945816258803e-06,
      "loss": 11.8758,
      "step": 34224
    },
    {
      "epoch": 1.8636880330535759,
      "grad_norm": 0.5794853517027942,
      "learning_rate": 2.426014827493872e-06,
      "loss": 11.7628,
      "step": 34225
    },
    {
      "epoch": 1.863742487050159,
      "grad_norm": 0.5708225316638832,
      "learning_rate": 2.424084597480636e-06,
      "loss": 11.7865,
      "step": 34226
    },
    {
      "epoch": 1.863796941046742,
      "grad_norm": 0.6114812305209636,
      "learning_rate": 2.4221551262340716e-06,
      "loss": 11.8621,
      "step": 34227
    },
    {
      "epoch": 1.863851395043325,
      "grad_norm": 0.5540456319701818,
      "learning_rate": 2.4202264137691776e-06,
      "loss": 11.6821,
      "step": 34228
    },
    {
      "epoch": 1.863905849039908,
      "grad_norm": 0.5730136453861171,
      "learning_rate": 2.418298460100987e-06,
      "loss": 11.814,
      "step": 34229
    },
    {
      "epoch": 1.863960303036491,
      "grad_norm": 0.523589859419443,
      "learning_rate": 2.4163712652444547e-06,
      "loss": 11.6111,
      "step": 34230
    },
    {
      "epoch": 1.864014757033074,
      "grad_norm": 0.6112924156925106,
      "learning_rate": 2.4144448292146017e-06,
      "loss": 11.7233,
      "step": 34231
    },
    {
      "epoch": 1.864069211029657,
      "grad_norm": 0.5374203790311808,
      "learning_rate": 2.4125191520263713e-06,
      "loss": 11.7915,
      "step": 34232
    },
    {
      "epoch": 1.86412366502624,
      "grad_norm": 0.5362833093421417,
      "learning_rate": 2.410594233694763e-06,
      "loss": 11.7705,
      "step": 34233
    },
    {
      "epoch": 1.864178119022823,
      "grad_norm": 0.5197452888719586,
      "learning_rate": 2.4086700742347535e-06,
      "loss": 11.7186,
      "step": 34234
    },
    {
      "epoch": 1.864232573019406,
      "grad_norm": 0.5190742130390831,
      "learning_rate": 2.406746673661275e-06,
      "loss": 11.5895,
      "step": 34235
    },
    {
      "epoch": 1.864287027015989,
      "grad_norm": 0.6293038113589222,
      "learning_rate": 2.4048240319893055e-06,
      "loss": 11.8104,
      "step": 34236
    },
    {
      "epoch": 1.864341481012572,
      "grad_norm": 0.5279810593944538,
      "learning_rate": 2.402902149233799e-06,
      "loss": 11.6538,
      "step": 34237
    },
    {
      "epoch": 1.864395935009155,
      "grad_norm": 0.527408394474537,
      "learning_rate": 2.4009810254096875e-06,
      "loss": 11.8652,
      "step": 34238
    },
    {
      "epoch": 1.864450389005738,
      "grad_norm": 0.5098192001087642,
      "learning_rate": 2.3990606605319156e-06,
      "loss": 11.6551,
      "step": 34239
    },
    {
      "epoch": 1.864504843002321,
      "grad_norm": 0.5070566259114763,
      "learning_rate": 2.397141054615415e-06,
      "loss": 11.735,
      "step": 34240
    },
    {
      "epoch": 1.864559296998904,
      "grad_norm": 0.49966074639794694,
      "learning_rate": 2.3952222076751184e-06,
      "loss": 11.7903,
      "step": 34241
    },
    {
      "epoch": 1.864613750995487,
      "grad_norm": 0.5308729556639347,
      "learning_rate": 2.3933041197259253e-06,
      "loss": 11.7152,
      "step": 34242
    },
    {
      "epoch": 1.86466820499207,
      "grad_norm": 0.5165496154658284,
      "learning_rate": 2.391386790782779e-06,
      "loss": 11.671,
      "step": 34243
    },
    {
      "epoch": 1.8647226589886532,
      "grad_norm": 0.6427412163321673,
      "learning_rate": 2.389470220860579e-06,
      "loss": 11.8475,
      "step": 34244
    },
    {
      "epoch": 1.8647771129852362,
      "grad_norm": 0.5266160782859761,
      "learning_rate": 2.387554409974224e-06,
      "loss": 11.7488,
      "step": 34245
    },
    {
      "epoch": 1.8648315669818192,
      "grad_norm": 0.5867827464115407,
      "learning_rate": 2.385639358138636e-06,
      "loss": 11.9609,
      "step": 34246
    },
    {
      "epoch": 1.8648860209784022,
      "grad_norm": 0.5885272033914,
      "learning_rate": 2.3837250653686583e-06,
      "loss": 11.8016,
      "step": 34247
    },
    {
      "epoch": 1.8649404749749852,
      "grad_norm": 0.5648847339188333,
      "learning_rate": 2.3818115316792122e-06,
      "loss": 11.7979,
      "step": 34248
    },
    {
      "epoch": 1.8649949289715682,
      "grad_norm": 0.6350035996609225,
      "learning_rate": 2.379898757085175e-06,
      "loss": 11.8398,
      "step": 34249
    },
    {
      "epoch": 1.8650493829681514,
      "grad_norm": 0.524575375399674,
      "learning_rate": 2.3779867416014125e-06,
      "loss": 11.899,
      "step": 34250
    },
    {
      "epoch": 1.8651038369647344,
      "grad_norm": 0.5214220943150758,
      "learning_rate": 2.3760754852428015e-06,
      "loss": 11.8004,
      "step": 34251
    },
    {
      "epoch": 1.8651582909613174,
      "grad_norm": 0.5475231764212184,
      "learning_rate": 2.374164988024197e-06,
      "loss": 11.7604,
      "step": 34252
    },
    {
      "epoch": 1.8652127449579003,
      "grad_norm": 0.5265529373533434,
      "learning_rate": 2.3722552499604645e-06,
      "loss": 11.6525,
      "step": 34253
    },
    {
      "epoch": 1.8652671989544833,
      "grad_norm": 0.5207911192734828,
      "learning_rate": 2.370346271066426e-06,
      "loss": 11.7003,
      "step": 34254
    },
    {
      "epoch": 1.8653216529510663,
      "grad_norm": 0.6038112220169792,
      "learning_rate": 2.368438051356958e-06,
      "loss": 11.872,
      "step": 34255
    },
    {
      "epoch": 1.8653761069476493,
      "grad_norm": 0.5391677969810922,
      "learning_rate": 2.366530590846905e-06,
      "loss": 11.6137,
      "step": 34256
    },
    {
      "epoch": 1.8654305609442323,
      "grad_norm": 0.5426360851693741,
      "learning_rate": 2.364623889551065e-06,
      "loss": 11.8784,
      "step": 34257
    },
    {
      "epoch": 1.8654850149408153,
      "grad_norm": 0.5505850702167205,
      "learning_rate": 2.3627179474842833e-06,
      "loss": 11.7918,
      "step": 34258
    },
    {
      "epoch": 1.8655394689373983,
      "grad_norm": 0.5028812290055151,
      "learning_rate": 2.360812764661391e-06,
      "loss": 11.6313,
      "step": 34259
    },
    {
      "epoch": 1.8655939229339813,
      "grad_norm": 0.5912852801999976,
      "learning_rate": 2.3589083410971768e-06,
      "loss": 11.7307,
      "step": 34260
    },
    {
      "epoch": 1.8656483769305643,
      "grad_norm": 0.5048018768131038,
      "learning_rate": 2.3570046768064847e-06,
      "loss": 11.8072,
      "step": 34261
    },
    {
      "epoch": 1.8657028309271473,
      "grad_norm": 0.5317571589028683,
      "learning_rate": 2.35510177180408e-06,
      "loss": 11.6593,
      "step": 34262
    },
    {
      "epoch": 1.8657572849237303,
      "grad_norm": 0.4986764309703747,
      "learning_rate": 2.353199626104796e-06,
      "loss": 11.7164,
      "step": 34263
    },
    {
      "epoch": 1.8658117389203133,
      "grad_norm": 0.5505031809051203,
      "learning_rate": 2.3512982397233987e-06,
      "loss": 11.7329,
      "step": 34264
    },
    {
      "epoch": 1.8658661929168963,
      "grad_norm": 0.5392223950234498,
      "learning_rate": 2.3493976126746754e-06,
      "loss": 11.7521,
      "step": 34265
    },
    {
      "epoch": 1.8659206469134793,
      "grad_norm": 0.5083800362848726,
      "learning_rate": 2.347497744973437e-06,
      "loss": 11.8374,
      "step": 34266
    },
    {
      "epoch": 1.8659751009100625,
      "grad_norm": 0.5482753870093451,
      "learning_rate": 2.345598636634405e-06,
      "loss": 11.7542,
      "step": 34267
    },
    {
      "epoch": 1.8660295549066455,
      "grad_norm": 0.580260195311508,
      "learning_rate": 2.34370028767239e-06,
      "loss": 11.9381,
      "step": 34268
    },
    {
      "epoch": 1.8660840089032285,
      "grad_norm": 0.5747469685504321,
      "learning_rate": 2.3418026981021357e-06,
      "loss": 11.7568,
      "step": 34269
    },
    {
      "epoch": 1.8661384628998114,
      "grad_norm": 0.5508804485965543,
      "learning_rate": 2.3399058679383855e-06,
      "loss": 11.816,
      "step": 34270
    },
    {
      "epoch": 1.8661929168963944,
      "grad_norm": 0.5502687580181147,
      "learning_rate": 2.338009797195928e-06,
      "loss": 11.8374,
      "step": 34271
    },
    {
      "epoch": 1.8662473708929774,
      "grad_norm": 0.5356402709789219,
      "learning_rate": 2.3361144858894734e-06,
      "loss": 11.7471,
      "step": 34272
    },
    {
      "epoch": 1.8663018248895606,
      "grad_norm": 0.7388047894645445,
      "learning_rate": 2.334219934033777e-06,
      "loss": 11.7528,
      "step": 34273
    },
    {
      "epoch": 1.8663562788861436,
      "grad_norm": 0.5753619368989512,
      "learning_rate": 2.332326141643548e-06,
      "loss": 11.817,
      "step": 34274
    },
    {
      "epoch": 1.8664107328827266,
      "grad_norm": 0.7071034474573465,
      "learning_rate": 2.3304331087335542e-06,
      "loss": 11.8966,
      "step": 34275
    },
    {
      "epoch": 1.8664651868793096,
      "grad_norm": 0.5041905253307493,
      "learning_rate": 2.328540835318471e-06,
      "loss": 11.6612,
      "step": 34276
    },
    {
      "epoch": 1.8665196408758926,
      "grad_norm": 0.5531223309782023,
      "learning_rate": 2.3266493214130437e-06,
      "loss": 11.7908,
      "step": 34277
    },
    {
      "epoch": 1.8665740948724756,
      "grad_norm": 0.5583570288124222,
      "learning_rate": 2.3247585670319703e-06,
      "loss": 11.788,
      "step": 34278
    },
    {
      "epoch": 1.8666285488690586,
      "grad_norm": 0.5290098880893309,
      "learning_rate": 2.322868572189951e-06,
      "loss": 11.6824,
      "step": 34279
    },
    {
      "epoch": 1.8666830028656416,
      "grad_norm": 0.5187326875973255,
      "learning_rate": 2.3209793369016965e-06,
      "loss": 11.7978,
      "step": 34280
    },
    {
      "epoch": 1.8667374568622246,
      "grad_norm": 0.5672294546787048,
      "learning_rate": 2.3190908611818717e-06,
      "loss": 11.8205,
      "step": 34281
    },
    {
      "epoch": 1.8667919108588076,
      "grad_norm": 0.4957828950353944,
      "learning_rate": 2.3172031450451883e-06,
      "loss": 11.5603,
      "step": 34282
    },
    {
      "epoch": 1.8668463648553906,
      "grad_norm": 0.5370713289251441,
      "learning_rate": 2.3153161885063113e-06,
      "loss": 11.702,
      "step": 34283
    },
    {
      "epoch": 1.8669008188519736,
      "grad_norm": 0.5764498160337324,
      "learning_rate": 2.3134299915799184e-06,
      "loss": 11.6951,
      "step": 34284
    },
    {
      "epoch": 1.8669552728485566,
      "grad_norm": 0.5379404055186005,
      "learning_rate": 2.3115445542806757e-06,
      "loss": 11.7268,
      "step": 34285
    },
    {
      "epoch": 1.8670097268451396,
      "grad_norm": 0.5687889360300659,
      "learning_rate": 2.309659876623238e-06,
      "loss": 11.808,
      "step": 34286
    },
    {
      "epoch": 1.8670641808417225,
      "grad_norm": 0.545817895197542,
      "learning_rate": 2.3077759586222825e-06,
      "loss": 11.6021,
      "step": 34287
    },
    {
      "epoch": 1.8671186348383055,
      "grad_norm": 0.5231374223122174,
      "learning_rate": 2.305892800292442e-06,
      "loss": 11.7889,
      "step": 34288
    },
    {
      "epoch": 1.8671730888348885,
      "grad_norm": 0.5538354207890446,
      "learning_rate": 2.304010401648349e-06,
      "loss": 11.7504,
      "step": 34289
    },
    {
      "epoch": 1.8672275428314717,
      "grad_norm": 0.5949262545531886,
      "learning_rate": 2.3021287627046695e-06,
      "loss": 11.7684,
      "step": 34290
    },
    {
      "epoch": 1.8672819968280547,
      "grad_norm": 0.5877613945347738,
      "learning_rate": 2.3002478834760035e-06,
      "loss": 11.9002,
      "step": 34291
    },
    {
      "epoch": 1.8673364508246377,
      "grad_norm": 0.5546132023592178,
      "learning_rate": 2.298367763976994e-06,
      "loss": 11.8494,
      "step": 34292
    },
    {
      "epoch": 1.8673909048212207,
      "grad_norm": 0.5340674487277636,
      "learning_rate": 2.296488404222286e-06,
      "loss": 11.8007,
      "step": 34293
    },
    {
      "epoch": 1.8674453588178037,
      "grad_norm": 0.5522883664425916,
      "learning_rate": 2.294609804226444e-06,
      "loss": 11.8901,
      "step": 34294
    },
    {
      "epoch": 1.8674998128143867,
      "grad_norm": 0.5625909293605665,
      "learning_rate": 2.292731964004113e-06,
      "loss": 11.8434,
      "step": 34295
    },
    {
      "epoch": 1.86755426681097,
      "grad_norm": 0.5467337569973905,
      "learning_rate": 2.290854883569882e-06,
      "loss": 11.8364,
      "step": 34296
    },
    {
      "epoch": 1.867608720807553,
      "grad_norm": 0.5375065484188294,
      "learning_rate": 2.2889785629383486e-06,
      "loss": 11.8107,
      "step": 34297
    },
    {
      "epoch": 1.867663174804136,
      "grad_norm": 0.5255140464611013,
      "learning_rate": 2.2871030021241134e-06,
      "loss": 11.6934,
      "step": 34298
    },
    {
      "epoch": 1.867717628800719,
      "grad_norm": 0.49817229426717313,
      "learning_rate": 2.2852282011417424e-06,
      "loss": 11.7442,
      "step": 34299
    },
    {
      "epoch": 1.867772082797302,
      "grad_norm": 0.5203996708735958,
      "learning_rate": 2.283354160005824e-06,
      "loss": 11.8311,
      "step": 34300
    },
    {
      "epoch": 1.8678265367938849,
      "grad_norm": 0.5673202159230402,
      "learning_rate": 2.2814808787309348e-06,
      "loss": 11.7362,
      "step": 34301
    },
    {
      "epoch": 1.8678809907904679,
      "grad_norm": 1.0016004923503519,
      "learning_rate": 2.2796083573316306e-06,
      "loss": 11.822,
      "step": 34302
    },
    {
      "epoch": 1.8679354447870509,
      "grad_norm": 0.5657832098027313,
      "learning_rate": 2.277736595822477e-06,
      "loss": 11.8573,
      "step": 34303
    },
    {
      "epoch": 1.8679898987836339,
      "grad_norm": 0.5360344128760475,
      "learning_rate": 2.2758655942180407e-06,
      "loss": 11.737,
      "step": 34304
    },
    {
      "epoch": 1.8680443527802169,
      "grad_norm": 0.5493261703465715,
      "learning_rate": 2.273995352532865e-06,
      "loss": 11.6219,
      "step": 34305
    },
    {
      "epoch": 1.8680988067767998,
      "grad_norm": 0.5148775070202533,
      "learning_rate": 2.272125870781472e-06,
      "loss": 11.7599,
      "step": 34306
    },
    {
      "epoch": 1.8681532607733828,
      "grad_norm": 0.5306871466967943,
      "learning_rate": 2.2702571489784388e-06,
      "loss": 11.7628,
      "step": 34307
    },
    {
      "epoch": 1.8682077147699658,
      "grad_norm": 0.5516223979441167,
      "learning_rate": 2.2683891871382646e-06,
      "loss": 11.7774,
      "step": 34308
    },
    {
      "epoch": 1.8682621687665488,
      "grad_norm": 0.5264760278860711,
      "learning_rate": 2.2665219852754825e-06,
      "loss": 11.7439,
      "step": 34309
    },
    {
      "epoch": 1.8683166227631318,
      "grad_norm": 0.5794411286558803,
      "learning_rate": 2.264655543404626e-06,
      "loss": 11.6774,
      "step": 34310
    },
    {
      "epoch": 1.8683710767597148,
      "grad_norm": 0.701812098298003,
      "learning_rate": 2.2627898615401933e-06,
      "loss": 11.8844,
      "step": 34311
    },
    {
      "epoch": 1.8684255307562978,
      "grad_norm": 0.5253276827866118,
      "learning_rate": 2.260924939696696e-06,
      "loss": 11.7294,
      "step": 34312
    },
    {
      "epoch": 1.8684799847528808,
      "grad_norm": 0.5835653874817696,
      "learning_rate": 2.259060777888633e-06,
      "loss": 11.8683,
      "step": 34313
    },
    {
      "epoch": 1.868534438749464,
      "grad_norm": 0.5507583791240651,
      "learning_rate": 2.2571973761305045e-06,
      "loss": 11.7279,
      "step": 34314
    },
    {
      "epoch": 1.868588892746047,
      "grad_norm": 0.5577230961812434,
      "learning_rate": 2.255334734436809e-06,
      "loss": 11.753,
      "step": 34315
    },
    {
      "epoch": 1.86864334674263,
      "grad_norm": 0.5313683324233448,
      "learning_rate": 2.253472852822014e-06,
      "loss": 11.6531,
      "step": 34316
    },
    {
      "epoch": 1.868697800739213,
      "grad_norm": 0.6020399393248772,
      "learning_rate": 2.251611731300629e-06,
      "loss": 11.8591,
      "step": 34317
    },
    {
      "epoch": 1.868752254735796,
      "grad_norm": 0.5795915911399782,
      "learning_rate": 2.2497513698870766e-06,
      "loss": 11.7954,
      "step": 34318
    },
    {
      "epoch": 1.868806708732379,
      "grad_norm": 0.526170584593981,
      "learning_rate": 2.2478917685958554e-06,
      "loss": 11.7472,
      "step": 34319
    },
    {
      "epoch": 1.8688611627289622,
      "grad_norm": 0.5060685953506894,
      "learning_rate": 2.2460329274414328e-06,
      "loss": 11.7706,
      "step": 34320
    },
    {
      "epoch": 1.8689156167255452,
      "grad_norm": 0.49720238683926316,
      "learning_rate": 2.2441748464382295e-06,
      "loss": 11.7317,
      "step": 34321
    },
    {
      "epoch": 1.8689700707221282,
      "grad_norm": 0.7488114605314281,
      "learning_rate": 2.2423175256007345e-06,
      "loss": 11.7128,
      "step": 34322
    },
    {
      "epoch": 1.8690245247187112,
      "grad_norm": 0.5379971433796815,
      "learning_rate": 2.2404609649433692e-06,
      "loss": 11.749,
      "step": 34323
    },
    {
      "epoch": 1.8690789787152942,
      "grad_norm": 0.5231756967868846,
      "learning_rate": 2.2386051644805783e-06,
      "loss": 11.7986,
      "step": 34324
    },
    {
      "epoch": 1.8691334327118772,
      "grad_norm": 0.5753825588422949,
      "learning_rate": 2.2367501242267717e-06,
      "loss": 11.7385,
      "step": 34325
    },
    {
      "epoch": 1.8691878867084601,
      "grad_norm": 0.5608086685297307,
      "learning_rate": 2.234895844196394e-06,
      "loss": 11.7776,
      "step": 34326
    },
    {
      "epoch": 1.8692423407050431,
      "grad_norm": 0.5394895111798916,
      "learning_rate": 2.233042324403889e-06,
      "loss": 11.7824,
      "step": 34327
    },
    {
      "epoch": 1.8692967947016261,
      "grad_norm": 0.5689546395263193,
      "learning_rate": 2.2311895648636117e-06,
      "loss": 11.7004,
      "step": 34328
    },
    {
      "epoch": 1.8693512486982091,
      "grad_norm": 0.600586015788296,
      "learning_rate": 2.229337565590006e-06,
      "loss": 11.8067,
      "step": 34329
    },
    {
      "epoch": 1.8694057026947921,
      "grad_norm": 0.5158412661360852,
      "learning_rate": 2.227486326597483e-06,
      "loss": 11.6357,
      "step": 34330
    },
    {
      "epoch": 1.869460156691375,
      "grad_norm": 0.50905046108615,
      "learning_rate": 2.225635847900409e-06,
      "loss": 11.7642,
      "step": 34331
    },
    {
      "epoch": 1.869514610687958,
      "grad_norm": 0.5485737885268168,
      "learning_rate": 2.2237861295131946e-06,
      "loss": 11.8075,
      "step": 34332
    },
    {
      "epoch": 1.869569064684541,
      "grad_norm": 0.5234525241867425,
      "learning_rate": 2.2219371714502058e-06,
      "loss": 11.772,
      "step": 34333
    },
    {
      "epoch": 1.869623518681124,
      "grad_norm": 0.5333371854004794,
      "learning_rate": 2.2200889737258423e-06,
      "loss": 11.7838,
      "step": 34334
    },
    {
      "epoch": 1.869677972677707,
      "grad_norm": 0.5365931389998904,
      "learning_rate": 2.2182415363544482e-06,
      "loss": 11.8084,
      "step": 34335
    },
    {
      "epoch": 1.86973242667429,
      "grad_norm": 0.6032294290204636,
      "learning_rate": 2.2163948593504236e-06,
      "loss": 11.8715,
      "step": 34336
    },
    {
      "epoch": 1.8697868806708733,
      "grad_norm": 0.5684661051567209,
      "learning_rate": 2.2145489427281117e-06,
      "loss": 11.6568,
      "step": 34337
    },
    {
      "epoch": 1.8698413346674563,
      "grad_norm": 0.5391462109251062,
      "learning_rate": 2.2127037865018573e-06,
      "loss": 11.7007,
      "step": 34338
    },
    {
      "epoch": 1.8698957886640393,
      "grad_norm": 0.5558384979744345,
      "learning_rate": 2.2108593906860155e-06,
      "loss": 11.662,
      "step": 34339
    },
    {
      "epoch": 1.8699502426606223,
      "grad_norm": 0.590831257116328,
      "learning_rate": 2.2090157552949297e-06,
      "loss": 11.8627,
      "step": 34340
    },
    {
      "epoch": 1.8700046966572053,
      "grad_norm": 0.5359014405334612,
      "learning_rate": 2.2071728803429337e-06,
      "loss": 11.73,
      "step": 34341
    },
    {
      "epoch": 1.8700591506537882,
      "grad_norm": 0.5653957194860302,
      "learning_rate": 2.205330765844382e-06,
      "loss": 11.8693,
      "step": 34342
    },
    {
      "epoch": 1.8701136046503715,
      "grad_norm": 0.5562694823055013,
      "learning_rate": 2.203489411813553e-06,
      "loss": 11.6833,
      "step": 34343
    },
    {
      "epoch": 1.8701680586469545,
      "grad_norm": 0.5149356771779124,
      "learning_rate": 2.2016488182648122e-06,
      "loss": 11.7022,
      "step": 34344
    },
    {
      "epoch": 1.8702225126435374,
      "grad_norm": 0.5422724576588746,
      "learning_rate": 2.199808985212437e-06,
      "loss": 11.5619,
      "step": 34345
    },
    {
      "epoch": 1.8702769666401204,
      "grad_norm": 0.5864744721307434,
      "learning_rate": 2.19796991267075e-06,
      "loss": 11.6507,
      "step": 34346
    },
    {
      "epoch": 1.8703314206367034,
      "grad_norm": 0.46228481529677695,
      "learning_rate": 2.196131600654061e-06,
      "loss": 11.7805,
      "step": 34347
    },
    {
      "epoch": 1.8703858746332864,
      "grad_norm": 0.5508542909946582,
      "learning_rate": 2.194294049176637e-06,
      "loss": 11.8564,
      "step": 34348
    },
    {
      "epoch": 1.8704403286298694,
      "grad_norm": 0.5355241792843812,
      "learning_rate": 2.1924572582528003e-06,
      "loss": 11.8426,
      "step": 34349
    },
    {
      "epoch": 1.8704947826264524,
      "grad_norm": 0.6120102724229697,
      "learning_rate": 2.1906212278968053e-06,
      "loss": 11.7656,
      "step": 34350
    },
    {
      "epoch": 1.8705492366230354,
      "grad_norm": 0.5116721589038303,
      "learning_rate": 2.188785958122963e-06,
      "loss": 11.6986,
      "step": 34351
    },
    {
      "epoch": 1.8706036906196184,
      "grad_norm": 0.5128777813311106,
      "learning_rate": 2.1869514489454955e-06,
      "loss": 11.6956,
      "step": 34352
    },
    {
      "epoch": 1.8706581446162014,
      "grad_norm": 0.5499742456332106,
      "learning_rate": 2.185117700378714e-06,
      "loss": 11.8258,
      "step": 34353
    },
    {
      "epoch": 1.8707125986127844,
      "grad_norm": 0.5396749535097277,
      "learning_rate": 2.1832847124368614e-06,
      "loss": 11.8081,
      "step": 34354
    },
    {
      "epoch": 1.8707670526093674,
      "grad_norm": 0.5932369636036566,
      "learning_rate": 2.1814524851341833e-06,
      "loss": 11.8096,
      "step": 34355
    },
    {
      "epoch": 1.8708215066059504,
      "grad_norm": 0.5744945478452651,
      "learning_rate": 2.179621018484945e-06,
      "loss": 11.8209,
      "step": 34356
    },
    {
      "epoch": 1.8708759606025334,
      "grad_norm": 0.5524508983211879,
      "learning_rate": 2.1777903125033694e-06,
      "loss": 11.859,
      "step": 34357
    },
    {
      "epoch": 1.8709304145991164,
      "grad_norm": 0.511186491903641,
      "learning_rate": 2.1759603672037e-06,
      "loss": 11.7553,
      "step": 34358
    },
    {
      "epoch": 1.8709848685956993,
      "grad_norm": 0.597577069022772,
      "learning_rate": 2.1741311826001808e-06,
      "loss": 11.9734,
      "step": 34359
    },
    {
      "epoch": 1.8710393225922826,
      "grad_norm": 0.5804922313953728,
      "learning_rate": 2.172302758707001e-06,
      "loss": 11.6796,
      "step": 34360
    },
    {
      "epoch": 1.8710937765888656,
      "grad_norm": 0.5485987486874512,
      "learning_rate": 2.170475095538427e-06,
      "loss": 11.838,
      "step": 34361
    },
    {
      "epoch": 1.8711482305854485,
      "grad_norm": 0.5347484606120664,
      "learning_rate": 2.1686481931086244e-06,
      "loss": 11.8154,
      "step": 34362
    },
    {
      "epoch": 1.8712026845820315,
      "grad_norm": 0.5123476391533188,
      "learning_rate": 2.166822051431816e-06,
      "loss": 11.7624,
      "step": 34363
    },
    {
      "epoch": 1.8712571385786145,
      "grad_norm": 0.6508918327409489,
      "learning_rate": 2.1649966705222234e-06,
      "loss": 11.745,
      "step": 34364
    },
    {
      "epoch": 1.8713115925751975,
      "grad_norm": 0.5250660221748026,
      "learning_rate": 2.1631720503940134e-06,
      "loss": 11.6672,
      "step": 34365
    },
    {
      "epoch": 1.8713660465717807,
      "grad_norm": 0.6354470342955629,
      "learning_rate": 2.1613481910613963e-06,
      "loss": 11.8038,
      "step": 34366
    },
    {
      "epoch": 1.8714205005683637,
      "grad_norm": 0.48238835692212567,
      "learning_rate": 2.159525092538539e-06,
      "loss": 11.736,
      "step": 34367
    },
    {
      "epoch": 1.8714749545649467,
      "grad_norm": 0.564512669862306,
      "learning_rate": 2.157702754839608e-06,
      "loss": 11.8659,
      "step": 34368
    },
    {
      "epoch": 1.8715294085615297,
      "grad_norm": 0.5635828074828091,
      "learning_rate": 2.1558811779788023e-06,
      "loss": 11.7195,
      "step": 34369
    },
    {
      "epoch": 1.8715838625581127,
      "grad_norm": 0.5534831164136985,
      "learning_rate": 2.154060361970267e-06,
      "loss": 11.7614,
      "step": 34370
    },
    {
      "epoch": 1.8716383165546957,
      "grad_norm": 0.49612570940930945,
      "learning_rate": 2.1522403068281795e-06,
      "loss": 11.8315,
      "step": 34371
    },
    {
      "epoch": 1.8716927705512787,
      "grad_norm": 0.6191222455780548,
      "learning_rate": 2.1504210125666614e-06,
      "loss": 11.8296,
      "step": 34372
    },
    {
      "epoch": 1.8717472245478617,
      "grad_norm": 0.5328092438598759,
      "learning_rate": 2.1486024791998903e-06,
      "loss": 11.6856,
      "step": 34373
    },
    {
      "epoch": 1.8718016785444447,
      "grad_norm": 0.4964358302001134,
      "learning_rate": 2.1467847067419887e-06,
      "loss": 11.7469,
      "step": 34374
    },
    {
      "epoch": 1.8718561325410277,
      "grad_norm": 0.5587324996028472,
      "learning_rate": 2.144967695207101e-06,
      "loss": 11.8047,
      "step": 34375
    },
    {
      "epoch": 1.8719105865376107,
      "grad_norm": 0.5908732823712312,
      "learning_rate": 2.1431514446093594e-06,
      "loss": 11.7955,
      "step": 34376
    },
    {
      "epoch": 1.8719650405341937,
      "grad_norm": 0.5157730303173998,
      "learning_rate": 2.1413359549628862e-06,
      "loss": 11.7474,
      "step": 34377
    },
    {
      "epoch": 1.8720194945307767,
      "grad_norm": 0.5848190070966441,
      "learning_rate": 2.139521226281793e-06,
      "loss": 11.8345,
      "step": 34378
    },
    {
      "epoch": 1.8720739485273596,
      "grad_norm": 0.5816650124556637,
      "learning_rate": 2.1377072585801906e-06,
      "loss": 11.7985,
      "step": 34379
    },
    {
      "epoch": 1.8721284025239426,
      "grad_norm": 0.5434419924222774,
      "learning_rate": 2.1358940518721893e-06,
      "loss": 11.7551,
      "step": 34380
    },
    {
      "epoch": 1.8721828565205256,
      "grad_norm": 0.5277405683949002,
      "learning_rate": 2.1340816061718893e-06,
      "loss": 11.8552,
      "step": 34381
    },
    {
      "epoch": 1.8722373105171086,
      "grad_norm": 0.48254009195127046,
      "learning_rate": 2.1322699214933793e-06,
      "loss": 11.7246,
      "step": 34382
    },
    {
      "epoch": 1.8722917645136916,
      "grad_norm": 0.5822475979806188,
      "learning_rate": 2.1304589978507595e-06,
      "loss": 11.7647,
      "step": 34383
    },
    {
      "epoch": 1.8723462185102748,
      "grad_norm": 0.6217129123806907,
      "learning_rate": 2.1286488352580846e-06,
      "loss": 11.765,
      "step": 34384
    },
    {
      "epoch": 1.8724006725068578,
      "grad_norm": 0.6047053339299858,
      "learning_rate": 2.126839433729466e-06,
      "loss": 11.7751,
      "step": 34385
    },
    {
      "epoch": 1.8724551265034408,
      "grad_norm": 0.600128254369047,
      "learning_rate": 2.125030793278948e-06,
      "loss": 11.8939,
      "step": 34386
    },
    {
      "epoch": 1.8725095805000238,
      "grad_norm": 0.5384829704509522,
      "learning_rate": 2.1232229139206196e-06,
      "loss": 11.7675,
      "step": 34387
    },
    {
      "epoch": 1.8725640344966068,
      "grad_norm": 0.546335750772106,
      "learning_rate": 2.121415795668513e-06,
      "loss": 11.7511,
      "step": 34388
    },
    {
      "epoch": 1.8726184884931898,
      "grad_norm": 0.5903271827455143,
      "learning_rate": 2.119609438536685e-06,
      "loss": 11.7408,
      "step": 34389
    },
    {
      "epoch": 1.872672942489773,
      "grad_norm": 0.5416514137835514,
      "learning_rate": 2.11780384253919e-06,
      "loss": 11.7675,
      "step": 34390
    },
    {
      "epoch": 1.872727396486356,
      "grad_norm": 0.5236622722793727,
      "learning_rate": 2.1159990076900727e-06,
      "loss": 11.8181,
      "step": 34391
    },
    {
      "epoch": 1.872781850482939,
      "grad_norm": 0.5594087362080239,
      "learning_rate": 2.114194934003366e-06,
      "loss": 11.7708,
      "step": 34392
    },
    {
      "epoch": 1.872836304479522,
      "grad_norm": 0.5846820398290147,
      "learning_rate": 2.112391621493093e-06,
      "loss": 11.8291,
      "step": 34393
    },
    {
      "epoch": 1.872890758476105,
      "grad_norm": 0.519282890580764,
      "learning_rate": 2.1105890701732743e-06,
      "loss": 11.7426,
      "step": 34394
    },
    {
      "epoch": 1.872945212472688,
      "grad_norm": 0.5318798358420342,
      "learning_rate": 2.1087872800579333e-06,
      "loss": 11.7557,
      "step": 34395
    },
    {
      "epoch": 1.872999666469271,
      "grad_norm": 0.5705674234138456,
      "learning_rate": 2.1069862511610694e-06,
      "loss": 11.6563,
      "step": 34396
    },
    {
      "epoch": 1.873054120465854,
      "grad_norm": 0.4792590627200801,
      "learning_rate": 2.1051859834967156e-06,
      "loss": 11.6679,
      "step": 34397
    },
    {
      "epoch": 1.873108574462437,
      "grad_norm": 0.6341616288108749,
      "learning_rate": 2.1033864770788503e-06,
      "loss": 11.7423,
      "step": 34398
    },
    {
      "epoch": 1.87316302845902,
      "grad_norm": 0.5527814185344745,
      "learning_rate": 2.1015877319214507e-06,
      "loss": 11.9045,
      "step": 34399
    },
    {
      "epoch": 1.873217482455603,
      "grad_norm": 0.5782207855250743,
      "learning_rate": 2.0997897480385386e-06,
      "loss": 11.7253,
      "step": 34400
    },
    {
      "epoch": 1.873271936452186,
      "grad_norm": 0.5304157200611321,
      "learning_rate": 2.09799252544407e-06,
      "loss": 11.6856,
      "step": 34401
    },
    {
      "epoch": 1.873326390448769,
      "grad_norm": 0.5394937941037234,
      "learning_rate": 2.096196064152034e-06,
      "loss": 11.8111,
      "step": 34402
    },
    {
      "epoch": 1.873380844445352,
      "grad_norm": 0.5954734149265039,
      "learning_rate": 2.0944003641763966e-06,
      "loss": 11.843,
      "step": 34403
    },
    {
      "epoch": 1.873435298441935,
      "grad_norm": 0.551728549063149,
      "learning_rate": 2.0926054255311136e-06,
      "loss": 11.7095,
      "step": 34404
    },
    {
      "epoch": 1.873489752438518,
      "grad_norm": 0.5462890870139695,
      "learning_rate": 2.0908112482301622e-06,
      "loss": 11.7751,
      "step": 34405
    },
    {
      "epoch": 1.873544206435101,
      "grad_norm": 0.5500798009257194,
      "learning_rate": 2.0890178322874653e-06,
      "loss": 11.8971,
      "step": 34406
    },
    {
      "epoch": 1.873598660431684,
      "grad_norm": 0.6280786023691517,
      "learning_rate": 2.0872251777170005e-06,
      "loss": 11.8629,
      "step": 34407
    },
    {
      "epoch": 1.873653114428267,
      "grad_norm": 0.5237086988875659,
      "learning_rate": 2.0854332845327007e-06,
      "loss": 11.7362,
      "step": 34408
    },
    {
      "epoch": 1.87370756842485,
      "grad_norm": 0.5262139772734832,
      "learning_rate": 2.083642152748466e-06,
      "loss": 11.8662,
      "step": 34409
    },
    {
      "epoch": 1.873762022421433,
      "grad_norm": 0.5298680180649677,
      "learning_rate": 2.0818517823782745e-06,
      "loss": 11.7746,
      "step": 34410
    },
    {
      "epoch": 1.873816476418016,
      "grad_norm": 0.5205467443915855,
      "learning_rate": 2.0800621734360035e-06,
      "loss": 11.7789,
      "step": 34411
    },
    {
      "epoch": 1.873870930414599,
      "grad_norm": 0.6486251825344184,
      "learning_rate": 2.078273325935598e-06,
      "loss": 11.7703,
      "step": 34412
    },
    {
      "epoch": 1.8739253844111823,
      "grad_norm": 0.5099860177347235,
      "learning_rate": 2.0764852398909684e-06,
      "loss": 11.7349,
      "step": 34413
    },
    {
      "epoch": 1.8739798384077653,
      "grad_norm": 0.5821970470073644,
      "learning_rate": 2.074697915316004e-06,
      "loss": 11.8652,
      "step": 34414
    },
    {
      "epoch": 1.8740342924043483,
      "grad_norm": 0.5218286181112657,
      "learning_rate": 2.0729113522246155e-06,
      "loss": 11.8774,
      "step": 34415
    },
    {
      "epoch": 1.8740887464009313,
      "grad_norm": 0.5745509749204943,
      "learning_rate": 2.0711255506306814e-06,
      "loss": 11.6911,
      "step": 34416
    },
    {
      "epoch": 1.8741432003975143,
      "grad_norm": 0.6063070917365466,
      "learning_rate": 2.069340510548112e-06,
      "loss": 11.7745,
      "step": 34417
    },
    {
      "epoch": 1.8741976543940972,
      "grad_norm": 0.5818062815628678,
      "learning_rate": 2.067556231990775e-06,
      "loss": 11.7646,
      "step": 34418
    },
    {
      "epoch": 1.8742521083906802,
      "grad_norm": 0.5071365366911584,
      "learning_rate": 2.065772714972525e-06,
      "loss": 11.857,
      "step": 34419
    },
    {
      "epoch": 1.8743065623872632,
      "grad_norm": 0.5530547851562596,
      "learning_rate": 2.063989959507262e-06,
      "loss": 11.7571,
      "step": 34420
    },
    {
      "epoch": 1.8743610163838462,
      "grad_norm": 0.5200060839544695,
      "learning_rate": 2.062207965608831e-06,
      "loss": 11.7833,
      "step": 34421
    },
    {
      "epoch": 1.8744154703804292,
      "grad_norm": 0.5272580332080232,
      "learning_rate": 2.0604267332911096e-06,
      "loss": 11.6509,
      "step": 34422
    },
    {
      "epoch": 1.8744699243770122,
      "grad_norm": 0.567258798481717,
      "learning_rate": 2.0586462625679203e-06,
      "loss": 11.8684,
      "step": 34423
    },
    {
      "epoch": 1.8745243783735952,
      "grad_norm": 0.5232239501305265,
      "learning_rate": 2.0568665534531184e-06,
      "loss": 11.739,
      "step": 34424
    },
    {
      "epoch": 1.8745788323701782,
      "grad_norm": 0.5078560684131614,
      "learning_rate": 2.0550876059605594e-06,
      "loss": 11.7891,
      "step": 34425
    },
    {
      "epoch": 1.8746332863667612,
      "grad_norm": 0.5451956301968227,
      "learning_rate": 2.053309420104055e-06,
      "loss": 11.7283,
      "step": 34426
    },
    {
      "epoch": 1.8746877403633442,
      "grad_norm": 0.5816667572424823,
      "learning_rate": 2.0515319958974487e-06,
      "loss": 11.8666,
      "step": 34427
    },
    {
      "epoch": 1.8747421943599272,
      "grad_norm": 0.5185057030445498,
      "learning_rate": 2.0497553333545637e-06,
      "loss": 11.7493,
      "step": 34428
    },
    {
      "epoch": 1.8747966483565102,
      "grad_norm": 0.5482425338200828,
      "learning_rate": 2.0479794324891887e-06,
      "loss": 11.7361,
      "step": 34429
    },
    {
      "epoch": 1.8748511023530934,
      "grad_norm": 0.5393914245802071,
      "learning_rate": 2.046204293315168e-06,
      "loss": 11.7319,
      "step": 34430
    },
    {
      "epoch": 1.8749055563496764,
      "grad_norm": 0.5342755576887441,
      "learning_rate": 2.04442991584628e-06,
      "loss": 11.7162,
      "step": 34431
    },
    {
      "epoch": 1.8749600103462594,
      "grad_norm": 0.515268639026471,
      "learning_rate": 2.042656300096335e-06,
      "loss": 11.7082,
      "step": 34432
    },
    {
      "epoch": 1.8750144643428424,
      "grad_norm": 0.5224587268132403,
      "learning_rate": 2.0408834460791227e-06,
      "loss": 11.7594,
      "step": 34433
    },
    {
      "epoch": 1.8750689183394254,
      "grad_norm": 0.5892220430621732,
      "learning_rate": 2.0391113538084316e-06,
      "loss": 11.8748,
      "step": 34434
    },
    {
      "epoch": 1.8751233723360083,
      "grad_norm": 0.5619071835590458,
      "learning_rate": 2.037340023298029e-06,
      "loss": 11.742,
      "step": 34435
    },
    {
      "epoch": 1.8751778263325916,
      "grad_norm": 0.5974927532265994,
      "learning_rate": 2.035569454561692e-06,
      "loss": 11.774,
      "step": 34436
    },
    {
      "epoch": 1.8752322803291746,
      "grad_norm": 0.525811668882926,
      "learning_rate": 2.0337996476132214e-06,
      "loss": 11.7742,
      "step": 34437
    },
    {
      "epoch": 1.8752867343257575,
      "grad_norm": 0.5553818211235978,
      "learning_rate": 2.0320306024663395e-06,
      "loss": 11.7991,
      "step": 34438
    },
    {
      "epoch": 1.8753411883223405,
      "grad_norm": 0.5918921431889814,
      "learning_rate": 2.0302623191348126e-06,
      "loss": 11.8323,
      "step": 34439
    },
    {
      "epoch": 1.8753956423189235,
      "grad_norm": 0.5258447972096325,
      "learning_rate": 2.0284947976324077e-06,
      "loss": 11.7244,
      "step": 34440
    },
    {
      "epoch": 1.8754500963155065,
      "grad_norm": 0.532229743224184,
      "learning_rate": 2.026728037972847e-06,
      "loss": 11.7806,
      "step": 34441
    },
    {
      "epoch": 1.8755045503120895,
      "grad_norm": 0.5235079017539482,
      "learning_rate": 2.0249620401698865e-06,
      "loss": 11.7687,
      "step": 34442
    },
    {
      "epoch": 1.8755590043086725,
      "grad_norm": 0.6012943838447397,
      "learning_rate": 2.023196804237237e-06,
      "loss": 11.8524,
      "step": 34443
    },
    {
      "epoch": 1.8756134583052555,
      "grad_norm": 0.5462113512508756,
      "learning_rate": 2.021432330188655e-06,
      "loss": 11.7749,
      "step": 34444
    },
    {
      "epoch": 1.8756679123018385,
      "grad_norm": 0.496202303587155,
      "learning_rate": 2.0196686180378397e-06,
      "loss": 11.6893,
      "step": 34445
    },
    {
      "epoch": 1.8757223662984215,
      "grad_norm": 0.5559307539325787,
      "learning_rate": 2.017905667798514e-06,
      "loss": 11.7268,
      "step": 34446
    },
    {
      "epoch": 1.8757768202950045,
      "grad_norm": 0.513245831471796,
      "learning_rate": 2.0161434794843893e-06,
      "loss": 11.7377,
      "step": 34447
    },
    {
      "epoch": 1.8758312742915875,
      "grad_norm": 0.5683770587521315,
      "learning_rate": 2.014382053109165e-06,
      "loss": 11.793,
      "step": 34448
    },
    {
      "epoch": 1.8758857282881705,
      "grad_norm": 0.5482440403752477,
      "learning_rate": 2.012621388686531e-06,
      "loss": 11.8129,
      "step": 34449
    },
    {
      "epoch": 1.8759401822847535,
      "grad_norm": 0.5094562828130171,
      "learning_rate": 2.0108614862301866e-06,
      "loss": 11.7884,
      "step": 34450
    },
    {
      "epoch": 1.8759946362813364,
      "grad_norm": 0.6093824328480545,
      "learning_rate": 2.0091023457538105e-06,
      "loss": 11.8513,
      "step": 34451
    },
    {
      "epoch": 1.8760490902779194,
      "grad_norm": 0.535910924486717,
      "learning_rate": 2.0073439672711024e-06,
      "loss": 11.878,
      "step": 34452
    },
    {
      "epoch": 1.8761035442745024,
      "grad_norm": 0.5940137508258937,
      "learning_rate": 2.005586350795707e-06,
      "loss": 11.8551,
      "step": 34453
    },
    {
      "epoch": 1.8761579982710856,
      "grad_norm": 0.5250192951520369,
      "learning_rate": 2.003829496341325e-06,
      "loss": 11.7418,
      "step": 34454
    },
    {
      "epoch": 1.8762124522676686,
      "grad_norm": 0.5981081202272566,
      "learning_rate": 2.0020734039215782e-06,
      "loss": 11.7875,
      "step": 34455
    },
    {
      "epoch": 1.8762669062642516,
      "grad_norm": 0.6114501315845203,
      "learning_rate": 2.000318073550156e-06,
      "loss": 11.8459,
      "step": 34456
    },
    {
      "epoch": 1.8763213602608346,
      "grad_norm": 0.4906140205056725,
      "learning_rate": 1.9985635052406915e-06,
      "loss": 11.7518,
      "step": 34457
    },
    {
      "epoch": 1.8763758142574176,
      "grad_norm": 0.5900485469254249,
      "learning_rate": 1.99680969900683e-06,
      "loss": 11.8525,
      "step": 34458
    },
    {
      "epoch": 1.8764302682540008,
      "grad_norm": 0.5923781937237657,
      "learning_rate": 1.995056654862215e-06,
      "loss": 11.7313,
      "step": 34459
    },
    {
      "epoch": 1.8764847222505838,
      "grad_norm": 0.5422296100505702,
      "learning_rate": 1.9933043728204702e-06,
      "loss": 11.7604,
      "step": 34460
    },
    {
      "epoch": 1.8765391762471668,
      "grad_norm": 0.4869438152224834,
      "learning_rate": 1.9915528528952175e-06,
      "loss": 11.6852,
      "step": 34461
    },
    {
      "epoch": 1.8765936302437498,
      "grad_norm": 0.5031597530798516,
      "learning_rate": 1.9898020951001016e-06,
      "loss": 11.7802,
      "step": 34462
    },
    {
      "epoch": 1.8766480842403328,
      "grad_norm": 0.5869418858344352,
      "learning_rate": 1.9880520994487115e-06,
      "loss": 11.7233,
      "step": 34463
    },
    {
      "epoch": 1.8767025382369158,
      "grad_norm": 0.5768040134503095,
      "learning_rate": 1.986302865954681e-06,
      "loss": 11.7805,
      "step": 34464
    },
    {
      "epoch": 1.8767569922334988,
      "grad_norm": 0.5407664945084442,
      "learning_rate": 1.984554394631577e-06,
      "loss": 11.7714,
      "step": 34465
    },
    {
      "epoch": 1.8768114462300818,
      "grad_norm": 0.5181589858084656,
      "learning_rate": 1.9828066854930328e-06,
      "loss": 11.8283,
      "step": 34466
    },
    {
      "epoch": 1.8768659002266648,
      "grad_norm": 0.5918453354931499,
      "learning_rate": 1.981059738552604e-06,
      "loss": 11.7435,
      "step": 34467
    },
    {
      "epoch": 1.8769203542232478,
      "grad_norm": 0.6205678159164849,
      "learning_rate": 1.9793135538239028e-06,
      "loss": 11.8655,
      "step": 34468
    },
    {
      "epoch": 1.8769748082198308,
      "grad_norm": 0.5245735247157259,
      "learning_rate": 1.9775681313204953e-06,
      "loss": 11.8524,
      "step": 34469
    },
    {
      "epoch": 1.8770292622164138,
      "grad_norm": 0.5423945105094565,
      "learning_rate": 1.9758234710559485e-06,
      "loss": 11.7362,
      "step": 34470
    },
    {
      "epoch": 1.8770837162129967,
      "grad_norm": 0.5091349420645608,
      "learning_rate": 1.9740795730438524e-06,
      "loss": 11.7449,
      "step": 34471
    },
    {
      "epoch": 1.8771381702095797,
      "grad_norm": 0.5295070002646127,
      "learning_rate": 1.9723364372977394e-06,
      "loss": 11.6581,
      "step": 34472
    },
    {
      "epoch": 1.8771926242061627,
      "grad_norm": 0.5319138956302755,
      "learning_rate": 1.970594063831177e-06,
      "loss": 11.7964,
      "step": 34473
    },
    {
      "epoch": 1.8772470782027457,
      "grad_norm": 0.564039425531116,
      "learning_rate": 1.9688524526577213e-06,
      "loss": 11.6605,
      "step": 34474
    },
    {
      "epoch": 1.8773015321993287,
      "grad_norm": 0.5564437896006181,
      "learning_rate": 1.9671116037909056e-06,
      "loss": 11.6495,
      "step": 34475
    },
    {
      "epoch": 1.8773559861959117,
      "grad_norm": 0.6194062080551602,
      "learning_rate": 1.9653715172442743e-06,
      "loss": 11.7993,
      "step": 34476
    },
    {
      "epoch": 1.877410440192495,
      "grad_norm": 0.6026581706880969,
      "learning_rate": 1.9636321930313507e-06,
      "loss": 11.8409,
      "step": 34477
    },
    {
      "epoch": 1.877464894189078,
      "grad_norm": 0.5308738460868556,
      "learning_rate": 1.961893631165668e-06,
      "loss": 11.8138,
      "step": 34478
    },
    {
      "epoch": 1.877519348185661,
      "grad_norm": 0.545804917861025,
      "learning_rate": 1.9601558316607482e-06,
      "loss": 11.7917,
      "step": 34479
    },
    {
      "epoch": 1.877573802182244,
      "grad_norm": 0.6841642340268604,
      "learning_rate": 1.9584187945300812e-06,
      "loss": 11.793,
      "step": 34480
    },
    {
      "epoch": 1.877628256178827,
      "grad_norm": 0.5817327235030416,
      "learning_rate": 1.9566825197872007e-06,
      "loss": 11.8931,
      "step": 34481
    },
    {
      "epoch": 1.8776827101754099,
      "grad_norm": 0.5477205931273549,
      "learning_rate": 1.9549470074455957e-06,
      "loss": 11.7111,
      "step": 34482
    },
    {
      "epoch": 1.877737164171993,
      "grad_norm": 0.5515966502652739,
      "learning_rate": 1.9532122575187663e-06,
      "loss": 11.8316,
      "step": 34483
    },
    {
      "epoch": 1.877791618168576,
      "grad_norm": 0.4810358389520921,
      "learning_rate": 1.9514782700202018e-06,
      "loss": 11.7007,
      "step": 34484
    },
    {
      "epoch": 1.877846072165159,
      "grad_norm": 0.5236914559142065,
      "learning_rate": 1.94974504496338e-06,
      "loss": 11.7711,
      "step": 34485
    },
    {
      "epoch": 1.877900526161742,
      "grad_norm": 0.5716203529750217,
      "learning_rate": 1.948012582361791e-06,
      "loss": 11.8149,
      "step": 34486
    },
    {
      "epoch": 1.877954980158325,
      "grad_norm": 0.5472489722211471,
      "learning_rate": 1.94628088222889e-06,
      "loss": 11.862,
      "step": 34487
    },
    {
      "epoch": 1.878009434154908,
      "grad_norm": 0.5891219500076248,
      "learning_rate": 1.9445499445781666e-06,
      "loss": 11.7981,
      "step": 34488
    },
    {
      "epoch": 1.878063888151491,
      "grad_norm": 0.5271053617140028,
      "learning_rate": 1.9428197694230543e-06,
      "loss": 11.7337,
      "step": 34489
    },
    {
      "epoch": 1.878118342148074,
      "grad_norm": 0.5517514029627272,
      "learning_rate": 1.94109035677702e-06,
      "loss": 11.9251,
      "step": 34490
    },
    {
      "epoch": 1.878172796144657,
      "grad_norm": 0.5427870678584611,
      "learning_rate": 1.9393617066535196e-06,
      "loss": 11.676,
      "step": 34491
    },
    {
      "epoch": 1.87822725014124,
      "grad_norm": 0.5426510252194713,
      "learning_rate": 1.937633819065987e-06,
      "loss": 11.7591,
      "step": 34492
    },
    {
      "epoch": 1.878281704137823,
      "grad_norm": 0.5740285832375813,
      "learning_rate": 1.935906694027856e-06,
      "loss": 11.6884,
      "step": 34493
    },
    {
      "epoch": 1.878336158134406,
      "grad_norm": 0.5414019091039056,
      "learning_rate": 1.9341803315525488e-06,
      "loss": 11.6635,
      "step": 34494
    },
    {
      "epoch": 1.878390612130989,
      "grad_norm": 0.5739667806560893,
      "learning_rate": 1.9324547316535104e-06,
      "loss": 11.7823,
      "step": 34495
    },
    {
      "epoch": 1.878445066127572,
      "grad_norm": 0.5395552988286576,
      "learning_rate": 1.9307298943441523e-06,
      "loss": 11.7178,
      "step": 34496
    },
    {
      "epoch": 1.878499520124155,
      "grad_norm": 0.4898450664483993,
      "learning_rate": 1.929005819637886e-06,
      "loss": 11.7795,
      "step": 34497
    },
    {
      "epoch": 1.878553974120738,
      "grad_norm": 0.4883211430185245,
      "learning_rate": 1.9272825075481226e-06,
      "loss": 11.73,
      "step": 34498
    },
    {
      "epoch": 1.878608428117321,
      "grad_norm": 0.5283353580397148,
      "learning_rate": 1.925559958088241e-06,
      "loss": 11.7586,
      "step": 34499
    },
    {
      "epoch": 1.8786628821139042,
      "grad_norm": 0.5234978536808704,
      "learning_rate": 1.9238381712716636e-06,
      "loss": 11.9098,
      "step": 34500
    },
    {
      "epoch": 1.8787173361104872,
      "grad_norm": 0.5749481724090452,
      "learning_rate": 1.9221171471117684e-06,
      "loss": 11.7901,
      "step": 34501
    },
    {
      "epoch": 1.8787717901070702,
      "grad_norm": 0.5375052414148336,
      "learning_rate": 1.9203968856219224e-06,
      "loss": 11.6708,
      "step": 34502
    },
    {
      "epoch": 1.8788262441036532,
      "grad_norm": 0.5686279822901938,
      "learning_rate": 1.918677386815537e-06,
      "loss": 11.9018,
      "step": 34503
    },
    {
      "epoch": 1.8788806981002362,
      "grad_norm": 0.5093394053034656,
      "learning_rate": 1.9169586507059577e-06,
      "loss": 11.8013,
      "step": 34504
    },
    {
      "epoch": 1.8789351520968192,
      "grad_norm": 0.571703094684345,
      "learning_rate": 1.9152406773065513e-06,
      "loss": 11.8528,
      "step": 34505
    },
    {
      "epoch": 1.8789896060934024,
      "grad_norm": 0.5498325117636943,
      "learning_rate": 1.9135234666306844e-06,
      "loss": 11.6962,
      "step": 34506
    },
    {
      "epoch": 1.8790440600899854,
      "grad_norm": 0.5639003883480723,
      "learning_rate": 1.9118070186917137e-06,
      "loss": 11.8434,
      "step": 34507
    },
    {
      "epoch": 1.8790985140865684,
      "grad_norm": 0.5960834375712886,
      "learning_rate": 1.9100913335029833e-06,
      "loss": 11.8433,
      "step": 34508
    },
    {
      "epoch": 1.8791529680831514,
      "grad_norm": 0.5022991779269027,
      "learning_rate": 1.908376411077828e-06,
      "loss": 11.5936,
      "step": 34509
    },
    {
      "epoch": 1.8792074220797343,
      "grad_norm": 0.5216442959501958,
      "learning_rate": 1.9066622514295807e-06,
      "loss": 11.7873,
      "step": 34510
    },
    {
      "epoch": 1.8792618760763173,
      "grad_norm": 0.5401386107197366,
      "learning_rate": 1.9049488545715865e-06,
      "loss": 11.8959,
      "step": 34511
    },
    {
      "epoch": 1.8793163300729003,
      "grad_norm": 0.5106782080787051,
      "learning_rate": 1.9032362205171572e-06,
      "loss": 11.7131,
      "step": 34512
    },
    {
      "epoch": 1.8793707840694833,
      "grad_norm": 0.5643095554772389,
      "learning_rate": 1.9015243492796154e-06,
      "loss": 11.7519,
      "step": 34513
    },
    {
      "epoch": 1.8794252380660663,
      "grad_norm": 0.5136996016727898,
      "learning_rate": 1.8998132408722724e-06,
      "loss": 11.7354,
      "step": 34514
    },
    {
      "epoch": 1.8794796920626493,
      "grad_norm": 0.5778985080824569,
      "learning_rate": 1.898102895308429e-06,
      "loss": 11.7375,
      "step": 34515
    },
    {
      "epoch": 1.8795341460592323,
      "grad_norm": 0.548016986559713,
      "learning_rate": 1.8963933126013856e-06,
      "loss": 11.6967,
      "step": 34516
    },
    {
      "epoch": 1.8795886000558153,
      "grad_norm": 0.5332376768984646,
      "learning_rate": 1.8946844927644425e-06,
      "loss": 11.625,
      "step": 34517
    },
    {
      "epoch": 1.8796430540523983,
      "grad_norm": 0.5273766765887075,
      "learning_rate": 1.8929764358109003e-06,
      "loss": 11.7244,
      "step": 34518
    },
    {
      "epoch": 1.8796975080489813,
      "grad_norm": 0.5529812715269194,
      "learning_rate": 1.8912691417540152e-06,
      "loss": 11.5003,
      "step": 34519
    },
    {
      "epoch": 1.8797519620455643,
      "grad_norm": 0.5449124996821897,
      "learning_rate": 1.8895626106070763e-06,
      "loss": 11.8011,
      "step": 34520
    },
    {
      "epoch": 1.8798064160421473,
      "grad_norm": 0.5784908225647555,
      "learning_rate": 1.88785684238334e-06,
      "loss": 11.8468,
      "step": 34521
    },
    {
      "epoch": 1.8798608700387303,
      "grad_norm": 0.5575582693831567,
      "learning_rate": 1.886151837096084e-06,
      "loss": 11.7628,
      "step": 34522
    },
    {
      "epoch": 1.8799153240353133,
      "grad_norm": 0.5495828736806819,
      "learning_rate": 1.884447594758576e-06,
      "loss": 11.7958,
      "step": 34523
    },
    {
      "epoch": 1.8799697780318965,
      "grad_norm": 0.5294812376046137,
      "learning_rate": 1.8827441153840496e-06,
      "loss": 11.7934,
      "step": 34524
    },
    {
      "epoch": 1.8800242320284795,
      "grad_norm": 0.527139784511984,
      "learning_rate": 1.8810413989857722e-06,
      "loss": 11.7265,
      "step": 34525
    },
    {
      "epoch": 1.8800786860250625,
      "grad_norm": 0.551246936249197,
      "learning_rate": 1.8793394455769552e-06,
      "loss": 11.6954,
      "step": 34526
    },
    {
      "epoch": 1.8801331400216454,
      "grad_norm": 0.5171860086676685,
      "learning_rate": 1.8776382551708548e-06,
      "loss": 11.7661,
      "step": 34527
    },
    {
      "epoch": 1.8801875940182284,
      "grad_norm": 0.5310445627915399,
      "learning_rate": 1.875937827780705e-06,
      "loss": 11.702,
      "step": 34528
    },
    {
      "epoch": 1.8802420480148117,
      "grad_norm": 0.541717433002239,
      "learning_rate": 1.874238163419706e-06,
      "loss": 11.693,
      "step": 34529
    },
    {
      "epoch": 1.8802965020113946,
      "grad_norm": 0.5230157604658559,
      "learning_rate": 1.8725392621010917e-06,
      "loss": 11.9111,
      "step": 34530
    },
    {
      "epoch": 1.8803509560079776,
      "grad_norm": 0.5148423224131015,
      "learning_rate": 1.870841123838063e-06,
      "loss": 11.6381,
      "step": 34531
    },
    {
      "epoch": 1.8804054100045606,
      "grad_norm": 0.5341492950235581,
      "learning_rate": 1.8691437486438313e-06,
      "loss": 11.8082,
      "step": 34532
    },
    {
      "epoch": 1.8804598640011436,
      "grad_norm": 0.575986502869782,
      "learning_rate": 1.867447136531597e-06,
      "loss": 11.7954,
      "step": 34533
    },
    {
      "epoch": 1.8805143179977266,
      "grad_norm": 0.5303804944689541,
      "learning_rate": 1.8657512875145388e-06,
      "loss": 11.7158,
      "step": 34534
    },
    {
      "epoch": 1.8805687719943096,
      "grad_norm": 0.5393704055071213,
      "learning_rate": 1.8640562016058794e-06,
      "loss": 11.7661,
      "step": 34535
    },
    {
      "epoch": 1.8806232259908926,
      "grad_norm": 0.5307814562574126,
      "learning_rate": 1.8623618788187524e-06,
      "loss": 11.7323,
      "step": 34536
    },
    {
      "epoch": 1.8806776799874756,
      "grad_norm": 0.5245617896281497,
      "learning_rate": 1.860668319166381e-06,
      "loss": 11.6104,
      "step": 34537
    },
    {
      "epoch": 1.8807321339840586,
      "grad_norm": 0.5890252640452627,
      "learning_rate": 1.8589755226618987e-06,
      "loss": 11.7918,
      "step": 34538
    },
    {
      "epoch": 1.8807865879806416,
      "grad_norm": 0.639612477749031,
      "learning_rate": 1.857283489318473e-06,
      "loss": 11.7559,
      "step": 34539
    },
    {
      "epoch": 1.8808410419772246,
      "grad_norm": 0.5893184885080097,
      "learning_rate": 1.8555922191492825e-06,
      "loss": 11.7902,
      "step": 34540
    },
    {
      "epoch": 1.8808954959738076,
      "grad_norm": 0.542572726714157,
      "learning_rate": 1.8539017121674495e-06,
      "loss": 11.7784,
      "step": 34541
    },
    {
      "epoch": 1.8809499499703906,
      "grad_norm": 0.5043011800731687,
      "learning_rate": 1.8522119683861528e-06,
      "loss": 11.7943,
      "step": 34542
    },
    {
      "epoch": 1.8810044039669735,
      "grad_norm": 0.5480161379186976,
      "learning_rate": 1.8505229878185038e-06,
      "loss": 11.7191,
      "step": 34543
    },
    {
      "epoch": 1.8810588579635565,
      "grad_norm": 0.5861594521838982,
      "learning_rate": 1.8488347704776477e-06,
      "loss": 11.9703,
      "step": 34544
    },
    {
      "epoch": 1.8811133119601395,
      "grad_norm": 0.513443270616714,
      "learning_rate": 1.8471473163767295e-06,
      "loss": 11.7988,
      "step": 34545
    },
    {
      "epoch": 1.8811677659567225,
      "grad_norm": 0.5734974278795244,
      "learning_rate": 1.8454606255288386e-06,
      "loss": 11.6857,
      "step": 34546
    },
    {
      "epoch": 1.8812222199533057,
      "grad_norm": 0.601132744975943,
      "learning_rate": 1.8437746979471093e-06,
      "loss": 11.8225,
      "step": 34547
    },
    {
      "epoch": 1.8812766739498887,
      "grad_norm": 0.4990396662536802,
      "learning_rate": 1.842089533644653e-06,
      "loss": 11.7413,
      "step": 34548
    },
    {
      "epoch": 1.8813311279464717,
      "grad_norm": 0.5113258685520127,
      "learning_rate": 1.8404051326345596e-06,
      "loss": 11.7624,
      "step": 34549
    },
    {
      "epoch": 1.8813855819430547,
      "grad_norm": 0.5341688777125242,
      "learning_rate": 1.8387214949299514e-06,
      "loss": 11.6345,
      "step": 34550
    },
    {
      "epoch": 1.8814400359396377,
      "grad_norm": 0.531137488453673,
      "learning_rate": 1.837038620543896e-06,
      "loss": 11.7813,
      "step": 34551
    },
    {
      "epoch": 1.8814944899362207,
      "grad_norm": 0.5573666641696239,
      "learning_rate": 1.8353565094894941e-06,
      "loss": 11.8014,
      "step": 34552
    },
    {
      "epoch": 1.881548943932804,
      "grad_norm": 0.5511885283203259,
      "learning_rate": 1.8336751617798132e-06,
      "loss": 11.7881,
      "step": 34553
    },
    {
      "epoch": 1.881603397929387,
      "grad_norm": 0.5616453379529558,
      "learning_rate": 1.8319945774279534e-06,
      "loss": 11.8446,
      "step": 34554
    },
    {
      "epoch": 1.88165785192597,
      "grad_norm": 0.5507030940031911,
      "learning_rate": 1.8303147564469492e-06,
      "loss": 11.7357,
      "step": 34555
    },
    {
      "epoch": 1.881712305922553,
      "grad_norm": 0.49553125219301664,
      "learning_rate": 1.8286356988498898e-06,
      "loss": 11.6769,
      "step": 34556
    },
    {
      "epoch": 1.881766759919136,
      "grad_norm": 0.5741465435718031,
      "learning_rate": 1.8269574046498205e-06,
      "loss": 11.8025,
      "step": 34557
    },
    {
      "epoch": 1.8818212139157189,
      "grad_norm": 0.520782133536226,
      "learning_rate": 1.8252798738597866e-06,
      "loss": 11.7877,
      "step": 34558
    },
    {
      "epoch": 1.8818756679123019,
      "grad_norm": 0.574023474605489,
      "learning_rate": 1.8236031064928548e-06,
      "loss": 11.7481,
      "step": 34559
    },
    {
      "epoch": 1.8819301219088849,
      "grad_norm": 0.5782810539166594,
      "learning_rate": 1.8219271025620489e-06,
      "loss": 11.833,
      "step": 34560
    },
    {
      "epoch": 1.8819845759054679,
      "grad_norm": 0.6411418693401827,
      "learning_rate": 1.820251862080391e-06,
      "loss": 11.714,
      "step": 34561
    },
    {
      "epoch": 1.8820390299020509,
      "grad_norm": 0.5110308301288894,
      "learning_rate": 1.8185773850609267e-06,
      "loss": 11.7653,
      "step": 34562
    },
    {
      "epoch": 1.8820934838986338,
      "grad_norm": 0.5322964441196149,
      "learning_rate": 1.8169036715166677e-06,
      "loss": 11.744,
      "step": 34563
    },
    {
      "epoch": 1.8821479378952168,
      "grad_norm": 0.6191140639186177,
      "learning_rate": 1.8152307214606368e-06,
      "loss": 11.7733,
      "step": 34564
    },
    {
      "epoch": 1.8822023918917998,
      "grad_norm": 0.5484913916318791,
      "learning_rate": 1.8135585349058236e-06,
      "loss": 11.7513,
      "step": 34565
    },
    {
      "epoch": 1.8822568458883828,
      "grad_norm": 0.5840247447273171,
      "learning_rate": 1.8118871118652515e-06,
      "loss": 11.8865,
      "step": 34566
    },
    {
      "epoch": 1.8823112998849658,
      "grad_norm": 0.48549691500511627,
      "learning_rate": 1.8102164523519206e-06,
      "loss": 11.7408,
      "step": 34567
    },
    {
      "epoch": 1.8823657538815488,
      "grad_norm": 0.5255324341211323,
      "learning_rate": 1.8085465563787985e-06,
      "loss": 11.8428,
      "step": 34568
    },
    {
      "epoch": 1.8824202078781318,
      "grad_norm": 0.5819003900039607,
      "learning_rate": 1.8068774239589082e-06,
      "loss": 11.8125,
      "step": 34569
    },
    {
      "epoch": 1.882474661874715,
      "grad_norm": 0.5291091749693664,
      "learning_rate": 1.8052090551051837e-06,
      "loss": 11.8378,
      "step": 34570
    },
    {
      "epoch": 1.882529115871298,
      "grad_norm": 0.5638550577042505,
      "learning_rate": 1.8035414498306258e-06,
      "loss": 11.5874,
      "step": 34571
    },
    {
      "epoch": 1.882583569867881,
      "grad_norm": 0.5451776922628029,
      "learning_rate": 1.8018746081482018e-06,
      "loss": 11.68,
      "step": 34572
    },
    {
      "epoch": 1.882638023864464,
      "grad_norm": 0.6496991930191914,
      "learning_rate": 1.800208530070857e-06,
      "loss": 11.758,
      "step": 34573
    },
    {
      "epoch": 1.882692477861047,
      "grad_norm": 0.5364858740991691,
      "learning_rate": 1.798543215611581e-06,
      "loss": 11.7761,
      "step": 34574
    },
    {
      "epoch": 1.88274693185763,
      "grad_norm": 0.556211449417477,
      "learning_rate": 1.7968786647832747e-06,
      "loss": 11.8209,
      "step": 34575
    },
    {
      "epoch": 1.8828013858542132,
      "grad_norm": 0.5515509456259532,
      "learning_rate": 1.7952148775989275e-06,
      "loss": 11.8054,
      "step": 34576
    },
    {
      "epoch": 1.8828558398507962,
      "grad_norm": 0.6478233205323938,
      "learning_rate": 1.7935518540714514e-06,
      "loss": 11.8756,
      "step": 34577
    },
    {
      "epoch": 1.8829102938473792,
      "grad_norm": 0.5265629244130572,
      "learning_rate": 1.7918895942137804e-06,
      "loss": 11.7566,
      "step": 34578
    },
    {
      "epoch": 1.8829647478439622,
      "grad_norm": 0.5120578431869297,
      "learning_rate": 1.7902280980388596e-06,
      "loss": 11.676,
      "step": 34579
    },
    {
      "epoch": 1.8830192018405452,
      "grad_norm": 0.5100364969390428,
      "learning_rate": 1.7885673655595902e-06,
      "loss": 11.8194,
      "step": 34580
    },
    {
      "epoch": 1.8830736558371282,
      "grad_norm": 0.6390320572686244,
      "learning_rate": 1.7869073967888839e-06,
      "loss": 11.8057,
      "step": 34581
    },
    {
      "epoch": 1.8831281098337112,
      "grad_norm": 0.6243437268660795,
      "learning_rate": 1.7852481917396636e-06,
      "loss": 11.8041,
      "step": 34582
    },
    {
      "epoch": 1.8831825638302941,
      "grad_norm": 0.5538963090657304,
      "learning_rate": 1.7835897504248078e-06,
      "loss": 11.7161,
      "step": 34583
    },
    {
      "epoch": 1.8832370178268771,
      "grad_norm": 0.5899599698293884,
      "learning_rate": 1.7819320728572508e-06,
      "loss": 11.8435,
      "step": 34584
    },
    {
      "epoch": 1.8832914718234601,
      "grad_norm": 0.5630917006394409,
      "learning_rate": 1.780275159049849e-06,
      "loss": 11.7576,
      "step": 34585
    },
    {
      "epoch": 1.8833459258200431,
      "grad_norm": 0.5680457632757285,
      "learning_rate": 1.778619009015503e-06,
      "loss": 11.817,
      "step": 34586
    },
    {
      "epoch": 1.8834003798166261,
      "grad_norm": 0.6171949631287466,
      "learning_rate": 1.7769636227670805e-06,
      "loss": 11.7812,
      "step": 34587
    },
    {
      "epoch": 1.883454833813209,
      "grad_norm": 0.5608462336342018,
      "learning_rate": 1.7753090003174711e-06,
      "loss": 11.6799,
      "step": 34588
    },
    {
      "epoch": 1.883509287809792,
      "grad_norm": 0.614067725318175,
      "learning_rate": 1.773655141679531e-06,
      "loss": 11.876,
      "step": 34589
    },
    {
      "epoch": 1.883563741806375,
      "grad_norm": 0.5843568433323433,
      "learning_rate": 1.772002046866117e-06,
      "loss": 11.8226,
      "step": 34590
    },
    {
      "epoch": 1.883618195802958,
      "grad_norm": 0.5584145875563088,
      "learning_rate": 1.770349715890085e-06,
      "loss": 11.727,
      "step": 34591
    },
    {
      "epoch": 1.883672649799541,
      "grad_norm": 0.553526472848038,
      "learning_rate": 1.7686981487642918e-06,
      "loss": 12.0033,
      "step": 34592
    },
    {
      "epoch": 1.8837271037961243,
      "grad_norm": 0.5529915250229152,
      "learning_rate": 1.76704734550156e-06,
      "loss": 11.7626,
      "step": 34593
    },
    {
      "epoch": 1.8837815577927073,
      "grad_norm": 0.5301679117743063,
      "learning_rate": 1.7653973061147688e-06,
      "loss": 11.8459,
      "step": 34594
    },
    {
      "epoch": 1.8838360117892903,
      "grad_norm": 0.6135710586873316,
      "learning_rate": 1.7637480306166964e-06,
      "loss": 11.8688,
      "step": 34595
    },
    {
      "epoch": 1.8838904657858733,
      "grad_norm": 0.511829590838322,
      "learning_rate": 1.7620995190202105e-06,
      "loss": 11.7596,
      "step": 34596
    },
    {
      "epoch": 1.8839449197824563,
      "grad_norm": 0.507604732791635,
      "learning_rate": 1.7604517713381008e-06,
      "loss": 11.7183,
      "step": 34597
    },
    {
      "epoch": 1.8839993737790393,
      "grad_norm": 0.5305830915378384,
      "learning_rate": 1.7588047875832013e-06,
      "loss": 11.7633,
      "step": 34598
    },
    {
      "epoch": 1.8840538277756225,
      "grad_norm": 0.5432931869801424,
      "learning_rate": 1.7571585677683133e-06,
      "loss": 11.8153,
      "step": 34599
    },
    {
      "epoch": 1.8841082817722055,
      "grad_norm": 0.585687679615243,
      "learning_rate": 1.7555131119062374e-06,
      "loss": 11.7612,
      "step": 34600
    },
    {
      "epoch": 1.8841627357687885,
      "grad_norm": 0.5167908151283527,
      "learning_rate": 1.7538684200097632e-06,
      "loss": 11.7514,
      "step": 34601
    },
    {
      "epoch": 1.8842171897653714,
      "grad_norm": 0.5467397447246971,
      "learning_rate": 1.7522244920916698e-06,
      "loss": 11.8294,
      "step": 34602
    },
    {
      "epoch": 1.8842716437619544,
      "grad_norm": 0.6097082491675548,
      "learning_rate": 1.7505813281647797e-06,
      "loss": 11.8247,
      "step": 34603
    },
    {
      "epoch": 1.8843260977585374,
      "grad_norm": 0.5857637570024662,
      "learning_rate": 1.7489389282418278e-06,
      "loss": 11.7459,
      "step": 34604
    },
    {
      "epoch": 1.8843805517551204,
      "grad_norm": 0.5376543082440299,
      "learning_rate": 1.7472972923356035e-06,
      "loss": 11.845,
      "step": 34605
    },
    {
      "epoch": 1.8844350057517034,
      "grad_norm": 0.5574582469953928,
      "learning_rate": 1.7456564204588854e-06,
      "loss": 11.8851,
      "step": 34606
    },
    {
      "epoch": 1.8844894597482864,
      "grad_norm": 0.5628027342243649,
      "learning_rate": 1.744016312624408e-06,
      "loss": 11.8243,
      "step": 34607
    },
    {
      "epoch": 1.8845439137448694,
      "grad_norm": 0.4962840768947248,
      "learning_rate": 1.7423769688449388e-06,
      "loss": 11.7812,
      "step": 34608
    },
    {
      "epoch": 1.8845983677414524,
      "grad_norm": 0.5508468887795229,
      "learning_rate": 1.740738389133234e-06,
      "loss": 11.8607,
      "step": 34609
    },
    {
      "epoch": 1.8846528217380354,
      "grad_norm": 0.5677877786463065,
      "learning_rate": 1.7391005735020172e-06,
      "loss": 11.6974,
      "step": 34610
    },
    {
      "epoch": 1.8847072757346184,
      "grad_norm": 0.6195100329471449,
      "learning_rate": 1.7374635219640334e-06,
      "loss": 11.756,
      "step": 34611
    },
    {
      "epoch": 1.8847617297312014,
      "grad_norm": 0.5438648655468336,
      "learning_rate": 1.735827234532006e-06,
      "loss": 11.7925,
      "step": 34612
    },
    {
      "epoch": 1.8848161837277844,
      "grad_norm": 0.532194074554113,
      "learning_rate": 1.7341917112186801e-06,
      "loss": 11.7477,
      "step": 34613
    },
    {
      "epoch": 1.8848706377243674,
      "grad_norm": 0.5519866347819824,
      "learning_rate": 1.7325569520367458e-06,
      "loss": 11.6991,
      "step": 34614
    },
    {
      "epoch": 1.8849250917209504,
      "grad_norm": 0.5740721307059174,
      "learning_rate": 1.730922956998926e-06,
      "loss": 11.6997,
      "step": 34615
    },
    {
      "epoch": 1.8849795457175333,
      "grad_norm": 0.5505885728565861,
      "learning_rate": 1.7292897261179442e-06,
      "loss": 11.6975,
      "step": 34616
    },
    {
      "epoch": 1.8850339997141166,
      "grad_norm": 0.5209356291069546,
      "learning_rate": 1.7276572594064677e-06,
      "loss": 11.7307,
      "step": 34617
    },
    {
      "epoch": 1.8850884537106996,
      "grad_norm": 0.5481467800813334,
      "learning_rate": 1.72602555687722e-06,
      "loss": 11.6987,
      "step": 34618
    },
    {
      "epoch": 1.8851429077072825,
      "grad_norm": 0.5774504452202481,
      "learning_rate": 1.7243946185428794e-06,
      "loss": 11.8261,
      "step": 34619
    },
    {
      "epoch": 1.8851973617038655,
      "grad_norm": 0.5343436440407358,
      "learning_rate": 1.722764444416125e-06,
      "loss": 11.7185,
      "step": 34620
    },
    {
      "epoch": 1.8852518157004485,
      "grad_norm": 0.6282483195123607,
      "learning_rate": 1.7211350345096355e-06,
      "loss": 11.5196,
      "step": 34621
    },
    {
      "epoch": 1.8853062696970315,
      "grad_norm": 0.5681866131835598,
      "learning_rate": 1.7195063888360786e-06,
      "loss": 11.767,
      "step": 34622
    },
    {
      "epoch": 1.8853607236936147,
      "grad_norm": 0.5812783220152261,
      "learning_rate": 1.7178785074081216e-06,
      "loss": 11.9382,
      "step": 34623
    },
    {
      "epoch": 1.8854151776901977,
      "grad_norm": 0.526674766439873,
      "learning_rate": 1.7162513902384214e-06,
      "loss": 11.6523,
      "step": 34624
    },
    {
      "epoch": 1.8854696316867807,
      "grad_norm": 0.541344562149101,
      "learning_rate": 1.7146250373396455e-06,
      "loss": 11.8034,
      "step": 34625
    },
    {
      "epoch": 1.8855240856833637,
      "grad_norm": 0.5301244783231273,
      "learning_rate": 1.7129994487244061e-06,
      "loss": 11.6972,
      "step": 34626
    },
    {
      "epoch": 1.8855785396799467,
      "grad_norm": 0.5027119398328848,
      "learning_rate": 1.7113746244053818e-06,
      "loss": 11.7451,
      "step": 34627
    },
    {
      "epoch": 1.8856329936765297,
      "grad_norm": 0.5125000611931756,
      "learning_rate": 1.709750564395185e-06,
      "loss": 11.7177,
      "step": 34628
    },
    {
      "epoch": 1.8856874476731127,
      "grad_norm": 0.5237211434302128,
      "learning_rate": 1.7081272687064609e-06,
      "loss": 11.765,
      "step": 34629
    },
    {
      "epoch": 1.8857419016696957,
      "grad_norm": 0.639306643617262,
      "learning_rate": 1.7065047373518105e-06,
      "loss": 11.6949,
      "step": 34630
    },
    {
      "epoch": 1.8857963556662787,
      "grad_norm": 0.5658080562396036,
      "learning_rate": 1.7048829703438685e-06,
      "loss": 11.7816,
      "step": 34631
    },
    {
      "epoch": 1.8858508096628617,
      "grad_norm": 0.5362130035284358,
      "learning_rate": 1.7032619676952356e-06,
      "loss": 11.815,
      "step": 34632
    },
    {
      "epoch": 1.8859052636594447,
      "grad_norm": 0.5027667432043391,
      "learning_rate": 1.7016417294185349e-06,
      "loss": 11.6848,
      "step": 34633
    },
    {
      "epoch": 1.8859597176560277,
      "grad_norm": 0.6115580945655306,
      "learning_rate": 1.7000222555263346e-06,
      "loss": 11.7466,
      "step": 34634
    },
    {
      "epoch": 1.8860141716526106,
      "grad_norm": 0.5362984113650079,
      "learning_rate": 1.6984035460312687e-06,
      "loss": 11.7809,
      "step": 34635
    },
    {
      "epoch": 1.8860686256491936,
      "grad_norm": 0.5479685299225547,
      "learning_rate": 1.6967856009458826e-06,
      "loss": 11.913,
      "step": 34636
    },
    {
      "epoch": 1.8861230796457766,
      "grad_norm": 0.5556102697829234,
      "learning_rate": 1.6951684202827888e-06,
      "loss": 11.8064,
      "step": 34637
    },
    {
      "epoch": 1.8861775336423596,
      "grad_norm": 0.5221456171550668,
      "learning_rate": 1.6935520040545328e-06,
      "loss": 11.798,
      "step": 34638
    },
    {
      "epoch": 1.8862319876389426,
      "grad_norm": 0.5436265338598268,
      "learning_rate": 1.6919363522737263e-06,
      "loss": 11.7613,
      "step": 34639
    },
    {
      "epoch": 1.8862864416355258,
      "grad_norm": 0.5368084362797071,
      "learning_rate": 1.6903214649529043e-06,
      "loss": 11.8183,
      "step": 34640
    },
    {
      "epoch": 1.8863408956321088,
      "grad_norm": 0.6089297506825833,
      "learning_rate": 1.6887073421046117e-06,
      "loss": 11.7475,
      "step": 34641
    },
    {
      "epoch": 1.8863953496286918,
      "grad_norm": 0.5182800471381385,
      "learning_rate": 1.6870939837414278e-06,
      "loss": 11.7802,
      "step": 34642
    },
    {
      "epoch": 1.8864498036252748,
      "grad_norm": 0.5707836716971906,
      "learning_rate": 1.685481389875887e-06,
      "loss": 11.8407,
      "step": 34643
    },
    {
      "epoch": 1.8865042576218578,
      "grad_norm": 0.5300192048478521,
      "learning_rate": 1.6838695605205346e-06,
      "loss": 11.6704,
      "step": 34644
    },
    {
      "epoch": 1.8865587116184408,
      "grad_norm": 0.5188660417075077,
      "learning_rate": 1.682258495687905e-06,
      "loss": 11.7235,
      "step": 34645
    },
    {
      "epoch": 1.886613165615024,
      "grad_norm": 0.571706838943987,
      "learning_rate": 1.6806481953905106e-06,
      "loss": 11.7495,
      "step": 34646
    },
    {
      "epoch": 1.886667619611607,
      "grad_norm": 0.531707708010136,
      "learning_rate": 1.6790386596408858e-06,
      "loss": 11.6087,
      "step": 34647
    },
    {
      "epoch": 1.88672207360819,
      "grad_norm": 0.5221425359013574,
      "learning_rate": 1.6774298884515427e-06,
      "loss": 11.7859,
      "step": 34648
    },
    {
      "epoch": 1.886776527604773,
      "grad_norm": 0.512961130352283,
      "learning_rate": 1.6758218818350046e-06,
      "loss": 11.7506,
      "step": 34649
    },
    {
      "epoch": 1.886830981601356,
      "grad_norm": 0.5830322974777441,
      "learning_rate": 1.6742146398037617e-06,
      "loss": 11.8445,
      "step": 34650
    },
    {
      "epoch": 1.886885435597939,
      "grad_norm": 0.5815686863134284,
      "learning_rate": 1.6726081623703038e-06,
      "loss": 11.8466,
      "step": 34651
    },
    {
      "epoch": 1.886939889594522,
      "grad_norm": 0.5212978927462005,
      "learning_rate": 1.6710024495471433e-06,
      "loss": 11.7543,
      "step": 34652
    },
    {
      "epoch": 1.886994343591105,
      "grad_norm": 0.5131432598187302,
      "learning_rate": 1.6693975013467478e-06,
      "loss": 11.7428,
      "step": 34653
    },
    {
      "epoch": 1.887048797587688,
      "grad_norm": 0.5679582808671314,
      "learning_rate": 1.6677933177816184e-06,
      "loss": 11.8561,
      "step": 34654
    },
    {
      "epoch": 1.887103251584271,
      "grad_norm": 0.5255067247519414,
      "learning_rate": 1.6661898988642123e-06,
      "loss": 11.7917,
      "step": 34655
    },
    {
      "epoch": 1.887157705580854,
      "grad_norm": 0.563327143505627,
      "learning_rate": 1.6645872446070078e-06,
      "loss": 11.8587,
      "step": 34656
    },
    {
      "epoch": 1.887212159577437,
      "grad_norm": 0.5254916221724062,
      "learning_rate": 1.6629853550224618e-06,
      "loss": 11.7731,
      "step": 34657
    },
    {
      "epoch": 1.88726661357402,
      "grad_norm": 0.569864926068086,
      "learning_rate": 1.661384230123031e-06,
      "loss": 11.6328,
      "step": 34658
    },
    {
      "epoch": 1.887321067570603,
      "grad_norm": 0.5249439267573294,
      "learning_rate": 1.659783869921172e-06,
      "loss": 11.7202,
      "step": 34659
    },
    {
      "epoch": 1.887375521567186,
      "grad_norm": 0.514860100244056,
      "learning_rate": 1.6581842744293307e-06,
      "loss": 11.8043,
      "step": 34660
    },
    {
      "epoch": 1.887429975563769,
      "grad_norm": 0.5258641290207511,
      "learning_rate": 1.6565854436599303e-06,
      "loss": 11.7038,
      "step": 34661
    },
    {
      "epoch": 1.887484429560352,
      "grad_norm": 0.5452153717444679,
      "learning_rate": 1.6549873776254166e-06,
      "loss": 11.8005,
      "step": 34662
    },
    {
      "epoch": 1.887538883556935,
      "grad_norm": 0.5488223450922112,
      "learning_rate": 1.6533900763382125e-06,
      "loss": 11.7307,
      "step": 34663
    },
    {
      "epoch": 1.887593337553518,
      "grad_norm": 0.5598097191485137,
      "learning_rate": 1.651793539810742e-06,
      "loss": 11.7573,
      "step": 34664
    },
    {
      "epoch": 1.887647791550101,
      "grad_norm": 0.5145815345171996,
      "learning_rate": 1.650197768055417e-06,
      "loss": 11.7986,
      "step": 34665
    },
    {
      "epoch": 1.887702245546684,
      "grad_norm": 0.5233328645199831,
      "learning_rate": 1.6486027610846499e-06,
      "loss": 11.7454,
      "step": 34666
    },
    {
      "epoch": 1.887756699543267,
      "grad_norm": 0.5236644100545705,
      "learning_rate": 1.647008518910842e-06,
      "loss": 11.7264,
      "step": 34667
    },
    {
      "epoch": 1.88781115353985,
      "grad_norm": 0.48575464511707533,
      "learning_rate": 1.6454150415463832e-06,
      "loss": 11.695,
      "step": 34668
    },
    {
      "epoch": 1.8878656075364333,
      "grad_norm": 0.5971994981377021,
      "learning_rate": 1.6438223290036747e-06,
      "loss": 11.7124,
      "step": 34669
    },
    {
      "epoch": 1.8879200615330163,
      "grad_norm": 0.5287751875479139,
      "learning_rate": 1.6422303812951068e-06,
      "loss": 11.7466,
      "step": 34670
    },
    {
      "epoch": 1.8879745155295993,
      "grad_norm": 0.5880344382484398,
      "learning_rate": 1.640639198433036e-06,
      "loss": 11.9114,
      "step": 34671
    },
    {
      "epoch": 1.8880289695261823,
      "grad_norm": 0.5139600890663518,
      "learning_rate": 1.6390487804298527e-06,
      "loss": 11.7664,
      "step": 34672
    },
    {
      "epoch": 1.8880834235227653,
      "grad_norm": 0.5631639749303369,
      "learning_rate": 1.6374591272979244e-06,
      "loss": 11.7814,
      "step": 34673
    },
    {
      "epoch": 1.8881378775193483,
      "grad_norm": 0.5110284170270594,
      "learning_rate": 1.6358702390496084e-06,
      "loss": 11.6061,
      "step": 34674
    },
    {
      "epoch": 1.8881923315159312,
      "grad_norm": 0.5624146109225929,
      "learning_rate": 1.634282115697261e-06,
      "loss": 11.8418,
      "step": 34675
    },
    {
      "epoch": 1.8882467855125142,
      "grad_norm": 0.5334597198508444,
      "learning_rate": 1.6326947572532281e-06,
      "loss": 11.7894,
      "step": 34676
    },
    {
      "epoch": 1.8883012395090972,
      "grad_norm": 0.5403257393225684,
      "learning_rate": 1.6311081637298665e-06,
      "loss": 11.7361,
      "step": 34677
    },
    {
      "epoch": 1.8883556935056802,
      "grad_norm": 0.5198175447217575,
      "learning_rate": 1.6295223351394884e-06,
      "loss": 11.7902,
      "step": 34678
    },
    {
      "epoch": 1.8884101475022632,
      "grad_norm": 0.5247410821422873,
      "learning_rate": 1.627937271494462e-06,
      "loss": 11.6793,
      "step": 34679
    },
    {
      "epoch": 1.8884646014988462,
      "grad_norm": 0.5245438830662863,
      "learning_rate": 1.626352972807077e-06,
      "loss": 11.8101,
      "step": 34680
    },
    {
      "epoch": 1.8885190554954292,
      "grad_norm": 0.5197853217966982,
      "learning_rate": 1.6247694390896685e-06,
      "loss": 11.7402,
      "step": 34681
    },
    {
      "epoch": 1.8885735094920122,
      "grad_norm": 0.6099667352408009,
      "learning_rate": 1.6231866703545594e-06,
      "loss": 11.7899,
      "step": 34682
    },
    {
      "epoch": 1.8886279634885952,
      "grad_norm": 0.541271358248567,
      "learning_rate": 1.6216046666140405e-06,
      "loss": 11.7236,
      "step": 34683
    },
    {
      "epoch": 1.8886824174851782,
      "grad_norm": 0.5240606070074448,
      "learning_rate": 1.620023427880435e-06,
      "loss": 11.8339,
      "step": 34684
    },
    {
      "epoch": 1.8887368714817612,
      "grad_norm": 0.5808576552608417,
      "learning_rate": 1.6184429541660106e-06,
      "loss": 11.7861,
      "step": 34685
    },
    {
      "epoch": 1.8887913254783442,
      "grad_norm": 0.4815287009993346,
      "learning_rate": 1.6168632454830802e-06,
      "loss": 11.8493,
      "step": 34686
    },
    {
      "epoch": 1.8888457794749274,
      "grad_norm": 0.5570491946695187,
      "learning_rate": 1.6152843018439111e-06,
      "loss": 11.8726,
      "step": 34687
    },
    {
      "epoch": 1.8889002334715104,
      "grad_norm": 0.5613414560471554,
      "learning_rate": 1.6137061232607942e-06,
      "loss": 11.6397,
      "step": 34688
    },
    {
      "epoch": 1.8889546874680934,
      "grad_norm": 0.518802792322339,
      "learning_rate": 1.6121287097460192e-06,
      "loss": 11.6943,
      "step": 34689
    },
    {
      "epoch": 1.8890091414646764,
      "grad_norm": 0.5352295290433896,
      "learning_rate": 1.6105520613118097e-06,
      "loss": 11.6351,
      "step": 34690
    },
    {
      "epoch": 1.8890635954612593,
      "grad_norm": 0.6134100862756692,
      "learning_rate": 1.6089761779704449e-06,
      "loss": 11.9323,
      "step": 34691
    },
    {
      "epoch": 1.8891180494578426,
      "grad_norm": 0.5595482894017889,
      "learning_rate": 1.6074010597341927e-06,
      "loss": 11.8118,
      "step": 34692
    },
    {
      "epoch": 1.8891725034544256,
      "grad_norm": 0.49573460899014316,
      "learning_rate": 1.6058267066152765e-06,
      "loss": 11.7629,
      "step": 34693
    },
    {
      "epoch": 1.8892269574510085,
      "grad_norm": 0.48920837880105894,
      "learning_rate": 1.6042531186259646e-06,
      "loss": 11.7556,
      "step": 34694
    },
    {
      "epoch": 1.8892814114475915,
      "grad_norm": 0.5987163818653659,
      "learning_rate": 1.6026802957784691e-06,
      "loss": 11.7194,
      "step": 34695
    },
    {
      "epoch": 1.8893358654441745,
      "grad_norm": 0.5411999354481595,
      "learning_rate": 1.6011082380850472e-06,
      "loss": 11.8184,
      "step": 34696
    },
    {
      "epoch": 1.8893903194407575,
      "grad_norm": 0.5333057676812357,
      "learning_rate": 1.5995369455579001e-06,
      "loss": 11.8026,
      "step": 34697
    },
    {
      "epoch": 1.8894447734373405,
      "grad_norm": 0.531896850009571,
      "learning_rate": 1.5979664182092514e-06,
      "loss": 11.8222,
      "step": 34698
    },
    {
      "epoch": 1.8894992274339235,
      "grad_norm": 0.5833299775246741,
      "learning_rate": 1.5963966560513465e-06,
      "loss": 11.8024,
      "step": 34699
    },
    {
      "epoch": 1.8895536814305065,
      "grad_norm": 0.49339265781876485,
      "learning_rate": 1.5948276590963318e-06,
      "loss": 11.7774,
      "step": 34700
    },
    {
      "epoch": 1.8896081354270895,
      "grad_norm": 0.5774944496657055,
      "learning_rate": 1.5932594273564528e-06,
      "loss": 11.8376,
      "step": 34701
    },
    {
      "epoch": 1.8896625894236725,
      "grad_norm": 0.5468619797962504,
      "learning_rate": 1.5916919608438885e-06,
      "loss": 11.5729,
      "step": 34702
    },
    {
      "epoch": 1.8897170434202555,
      "grad_norm": 0.5493901200405292,
      "learning_rate": 1.5901252595708293e-06,
      "loss": 11.8508,
      "step": 34703
    },
    {
      "epoch": 1.8897714974168385,
      "grad_norm": 0.5532054980662606,
      "learning_rate": 1.5885593235494657e-06,
      "loss": 11.6965,
      "step": 34704
    },
    {
      "epoch": 1.8898259514134215,
      "grad_norm": 0.5317276365390414,
      "learning_rate": 1.586994152791954e-06,
      "loss": 11.7395,
      "step": 34705
    },
    {
      "epoch": 1.8898804054100045,
      "grad_norm": 0.549413117040086,
      "learning_rate": 1.5854297473104961e-06,
      "loss": 11.7359,
      "step": 34706
    },
    {
      "epoch": 1.8899348594065875,
      "grad_norm": 0.5398246119148612,
      "learning_rate": 1.5838661071172268e-06,
      "loss": 11.7162,
      "step": 34707
    },
    {
      "epoch": 1.8899893134031704,
      "grad_norm": 0.5557072050995043,
      "learning_rate": 1.5823032322243248e-06,
      "loss": 11.7413,
      "step": 34708
    },
    {
      "epoch": 1.8900437673997534,
      "grad_norm": 0.5064212637841524,
      "learning_rate": 1.580741122643936e-06,
      "loss": 11.7254,
      "step": 34709
    },
    {
      "epoch": 1.8900982213963367,
      "grad_norm": 0.5484464114083785,
      "learning_rate": 1.5791797783882178e-06,
      "loss": 11.7821,
      "step": 34710
    },
    {
      "epoch": 1.8901526753929196,
      "grad_norm": 0.6471395838410182,
      "learning_rate": 1.5776191994693046e-06,
      "loss": 11.8386,
      "step": 34711
    },
    {
      "epoch": 1.8902071293895026,
      "grad_norm": 0.5416635492934879,
      "learning_rate": 1.57605938589932e-06,
      "loss": 11.7639,
      "step": 34712
    },
    {
      "epoch": 1.8902615833860856,
      "grad_norm": 0.5751980565896976,
      "learning_rate": 1.5745003376903987e-06,
      "loss": 11.8457,
      "step": 34713
    },
    {
      "epoch": 1.8903160373826686,
      "grad_norm": 0.4850482014936915,
      "learning_rate": 1.5729420548546758e-06,
      "loss": 11.775,
      "step": 34714
    },
    {
      "epoch": 1.8903704913792516,
      "grad_norm": 0.5402958731500984,
      "learning_rate": 1.5713845374042634e-06,
      "loss": 11.7768,
      "step": 34715
    },
    {
      "epoch": 1.8904249453758348,
      "grad_norm": 0.5487319009951255,
      "learning_rate": 1.5698277853512634e-06,
      "loss": 11.8579,
      "step": 34716
    },
    {
      "epoch": 1.8904793993724178,
      "grad_norm": 0.5120235422922093,
      "learning_rate": 1.5682717987077988e-06,
      "loss": 11.6429,
      "step": 34717
    },
    {
      "epoch": 1.8905338533690008,
      "grad_norm": 0.5177553700338551,
      "learning_rate": 1.5667165774859604e-06,
      "loss": 11.8002,
      "step": 34718
    },
    {
      "epoch": 1.8905883073655838,
      "grad_norm": 0.5220602464741136,
      "learning_rate": 1.5651621216978274e-06,
      "loss": 11.7366,
      "step": 34719
    },
    {
      "epoch": 1.8906427613621668,
      "grad_norm": 0.5365320626779672,
      "learning_rate": 1.5636084313555122e-06,
      "loss": 11.7242,
      "step": 34720
    },
    {
      "epoch": 1.8906972153587498,
      "grad_norm": 0.5375619959670963,
      "learning_rate": 1.5620555064710939e-06,
      "loss": 11.7972,
      "step": 34721
    },
    {
      "epoch": 1.8907516693553328,
      "grad_norm": 0.5818501230587102,
      "learning_rate": 1.56050334705663e-06,
      "loss": 11.6516,
      "step": 34722
    },
    {
      "epoch": 1.8908061233519158,
      "grad_norm": 0.5517358627891079,
      "learning_rate": 1.558951953124199e-06,
      "loss": 11.7736,
      "step": 34723
    },
    {
      "epoch": 1.8908605773484988,
      "grad_norm": 0.5677645282670712,
      "learning_rate": 1.5574013246858587e-06,
      "loss": 11.5162,
      "step": 34724
    },
    {
      "epoch": 1.8909150313450818,
      "grad_norm": 0.5050136781134978,
      "learning_rate": 1.5558514617536878e-06,
      "loss": 11.78,
      "step": 34725
    },
    {
      "epoch": 1.8909694853416648,
      "grad_norm": 0.5766250407978719,
      "learning_rate": 1.5543023643397214e-06,
      "loss": 11.815,
      "step": 34726
    },
    {
      "epoch": 1.8910239393382478,
      "grad_norm": 0.5869011819929751,
      "learning_rate": 1.5527540324560052e-06,
      "loss": 11.8199,
      "step": 34727
    },
    {
      "epoch": 1.8910783933348307,
      "grad_norm": 0.5516178557245982,
      "learning_rate": 1.5512064661145854e-06,
      "loss": 11.8658,
      "step": 34728
    },
    {
      "epoch": 1.8911328473314137,
      "grad_norm": 0.5291889074957623,
      "learning_rate": 1.5496596653274965e-06,
      "loss": 11.7808,
      "step": 34729
    },
    {
      "epoch": 1.8911873013279967,
      "grad_norm": 0.4693384375666813,
      "learning_rate": 1.5481136301067623e-06,
      "loss": 11.7714,
      "step": 34730
    },
    {
      "epoch": 1.8912417553245797,
      "grad_norm": 0.5802511886742016,
      "learning_rate": 1.5465683604644177e-06,
      "loss": 11.8109,
      "step": 34731
    },
    {
      "epoch": 1.8912962093211627,
      "grad_norm": 0.5188283475927622,
      "learning_rate": 1.5450238564124531e-06,
      "loss": 11.7134,
      "step": 34732
    },
    {
      "epoch": 1.891350663317746,
      "grad_norm": 0.5295055212468985,
      "learning_rate": 1.5434801179629034e-06,
      "loss": 11.8336,
      "step": 34733
    },
    {
      "epoch": 1.891405117314329,
      "grad_norm": 0.5805184217072353,
      "learning_rate": 1.5419371451277476e-06,
      "loss": 11.8499,
      "step": 34734
    },
    {
      "epoch": 1.891459571310912,
      "grad_norm": 0.5713144011010932,
      "learning_rate": 1.5403949379190096e-06,
      "loss": 11.7572,
      "step": 34735
    },
    {
      "epoch": 1.891514025307495,
      "grad_norm": 0.5472873777591046,
      "learning_rate": 1.5388534963486801e-06,
      "loss": 11.7151,
      "step": 34736
    },
    {
      "epoch": 1.891568479304078,
      "grad_norm": 0.5995463581218854,
      "learning_rate": 1.5373128204287268e-06,
      "loss": 11.8759,
      "step": 34737
    },
    {
      "epoch": 1.891622933300661,
      "grad_norm": 0.5159348677970845,
      "learning_rate": 1.5357729101711517e-06,
      "loss": 11.6406,
      "step": 34738
    },
    {
      "epoch": 1.891677387297244,
      "grad_norm": 0.5494847554065374,
      "learning_rate": 1.5342337655879112e-06,
      "loss": 11.7725,
      "step": 34739
    },
    {
      "epoch": 1.891731841293827,
      "grad_norm": 0.5773632038676855,
      "learning_rate": 1.5326953866909855e-06,
      "loss": 11.7697,
      "step": 34740
    },
    {
      "epoch": 1.89178629529041,
      "grad_norm": 0.5364709842273305,
      "learning_rate": 1.531157773492331e-06,
      "loss": 11.7066,
      "step": 34741
    },
    {
      "epoch": 1.891840749286993,
      "grad_norm": 0.4894740729461135,
      "learning_rate": 1.529620926003905e-06,
      "loss": 11.6814,
      "step": 34742
    },
    {
      "epoch": 1.891895203283576,
      "grad_norm": 0.5604563994796176,
      "learning_rate": 1.528084844237665e-06,
      "loss": 11.8408,
      "step": 34743
    },
    {
      "epoch": 1.891949657280159,
      "grad_norm": 0.5781736009590769,
      "learning_rate": 1.5265495282055453e-06,
      "loss": 11.8307,
      "step": 34744
    },
    {
      "epoch": 1.892004111276742,
      "grad_norm": 0.542475410913117,
      "learning_rate": 1.5250149779195034e-06,
      "loss": 11.7901,
      "step": 34745
    },
    {
      "epoch": 1.892058565273325,
      "grad_norm": 0.5735364564893207,
      "learning_rate": 1.5234811933914406e-06,
      "loss": 11.6456,
      "step": 34746
    },
    {
      "epoch": 1.892113019269908,
      "grad_norm": 0.5097625218550237,
      "learning_rate": 1.521948174633314e-06,
      "loss": 11.7917,
      "step": 34747
    },
    {
      "epoch": 1.892167473266491,
      "grad_norm": 0.5090108605281053,
      "learning_rate": 1.5204159216570258e-06,
      "loss": 11.698,
      "step": 34748
    },
    {
      "epoch": 1.892221927263074,
      "grad_norm": 0.5908728465726407,
      "learning_rate": 1.518884434474499e-06,
      "loss": 11.8075,
      "step": 34749
    },
    {
      "epoch": 1.892276381259657,
      "grad_norm": 0.5441837765889614,
      "learning_rate": 1.5173537130976578e-06,
      "loss": 11.7498,
      "step": 34750
    },
    {
      "epoch": 1.89233083525624,
      "grad_norm": 0.523598287999606,
      "learning_rate": 1.5158237575383816e-06,
      "loss": 11.7017,
      "step": 34751
    },
    {
      "epoch": 1.892385289252823,
      "grad_norm": 0.5610758206507106,
      "learning_rate": 1.5142945678085719e-06,
      "loss": 11.8497,
      "step": 34752
    },
    {
      "epoch": 1.892439743249406,
      "grad_norm": 0.4921037485056164,
      "learning_rate": 1.512766143920119e-06,
      "loss": 11.7247,
      "step": 34753
    },
    {
      "epoch": 1.892494197245989,
      "grad_norm": 0.5580199380516381,
      "learning_rate": 1.5112384858849137e-06,
      "loss": 11.666,
      "step": 34754
    },
    {
      "epoch": 1.892548651242572,
      "grad_norm": 0.5300540095189908,
      "learning_rate": 1.5097115937148464e-06,
      "loss": 11.8728,
      "step": 34755
    },
    {
      "epoch": 1.892603105239155,
      "grad_norm": 0.5143627162845545,
      "learning_rate": 1.5081854674217632e-06,
      "loss": 11.8203,
      "step": 34756
    },
    {
      "epoch": 1.8926575592357382,
      "grad_norm": 0.5525976886598135,
      "learning_rate": 1.5066601070175657e-06,
      "loss": 11.8271,
      "step": 34757
    },
    {
      "epoch": 1.8927120132323212,
      "grad_norm": 0.5606215905430598,
      "learning_rate": 1.5051355125140775e-06,
      "loss": 11.8279,
      "step": 34758
    },
    {
      "epoch": 1.8927664672289042,
      "grad_norm": 0.5401105593454313,
      "learning_rate": 1.5036116839231785e-06,
      "loss": 11.7004,
      "step": 34759
    },
    {
      "epoch": 1.8928209212254872,
      "grad_norm": 0.5605481407711291,
      "learning_rate": 1.5020886212567254e-06,
      "loss": 11.8379,
      "step": 34760
    },
    {
      "epoch": 1.8928753752220702,
      "grad_norm": 0.4993905054794739,
      "learning_rate": 1.5005663245265423e-06,
      "loss": 11.7567,
      "step": 34761
    },
    {
      "epoch": 1.8929298292186534,
      "grad_norm": 0.5246236421027616,
      "learning_rate": 1.499044793744464e-06,
      "loss": 11.7809,
      "step": 34762
    },
    {
      "epoch": 1.8929842832152364,
      "grad_norm": 0.5593205693022092,
      "learning_rate": 1.497524028922337e-06,
      "loss": 11.7566,
      "step": 34763
    },
    {
      "epoch": 1.8930387372118194,
      "grad_norm": 0.5288054314884846,
      "learning_rate": 1.4960040300719846e-06,
      "loss": 11.7922,
      "step": 34764
    },
    {
      "epoch": 1.8930931912084024,
      "grad_norm": 0.60937038346665,
      "learning_rate": 1.494484797205231e-06,
      "loss": 11.8041,
      "step": 34765
    },
    {
      "epoch": 1.8931476452049854,
      "grad_norm": 0.5091641163587646,
      "learning_rate": 1.492966330333867e-06,
      "loss": 11.5661,
      "step": 34766
    },
    {
      "epoch": 1.8932020992015683,
      "grad_norm": 0.5661740741072653,
      "learning_rate": 1.4914486294697271e-06,
      "loss": 11.8103,
      "step": 34767
    },
    {
      "epoch": 1.8932565531981513,
      "grad_norm": 0.5477024116714037,
      "learning_rate": 1.4899316946246022e-06,
      "loss": 11.7092,
      "step": 34768
    },
    {
      "epoch": 1.8933110071947343,
      "grad_norm": 0.47994673754978207,
      "learning_rate": 1.4884155258102828e-06,
      "loss": 11.6568,
      "step": 34769
    },
    {
      "epoch": 1.8933654611913173,
      "grad_norm": 0.5191042016630331,
      "learning_rate": 1.4869001230385815e-06,
      "loss": 11.7265,
      "step": 34770
    },
    {
      "epoch": 1.8934199151879003,
      "grad_norm": 0.49992275049186446,
      "learning_rate": 1.4853854863212447e-06,
      "loss": 11.7216,
      "step": 34771
    },
    {
      "epoch": 1.8934743691844833,
      "grad_norm": 0.537041236182039,
      "learning_rate": 1.483871615670085e-06,
      "loss": 11.7867,
      "step": 34772
    },
    {
      "epoch": 1.8935288231810663,
      "grad_norm": 0.5737665435883587,
      "learning_rate": 1.4823585110968486e-06,
      "loss": 11.9449,
      "step": 34773
    },
    {
      "epoch": 1.8935832771776493,
      "grad_norm": 0.5464830454808277,
      "learning_rate": 1.480846172613315e-06,
      "loss": 11.752,
      "step": 34774
    },
    {
      "epoch": 1.8936377311742323,
      "grad_norm": 0.5219663064655751,
      "learning_rate": 1.4793346002312524e-06,
      "loss": 11.7249,
      "step": 34775
    },
    {
      "epoch": 1.8936921851708153,
      "grad_norm": 0.5716351177460596,
      "learning_rate": 1.4778237939623962e-06,
      "loss": 11.7522,
      "step": 34776
    },
    {
      "epoch": 1.8937466391673983,
      "grad_norm": 0.526522341466927,
      "learning_rate": 1.4763137538185146e-06,
      "loss": 11.6462,
      "step": 34777
    },
    {
      "epoch": 1.8938010931639813,
      "grad_norm": 0.5344246023625869,
      "learning_rate": 1.4748044798113315e-06,
      "loss": 11.612,
      "step": 34778
    },
    {
      "epoch": 1.8938555471605643,
      "grad_norm": 0.590258437698882,
      "learning_rate": 1.473295971952604e-06,
      "loss": 11.7494,
      "step": 34779
    },
    {
      "epoch": 1.8939100011571475,
      "grad_norm": 0.49805653184473947,
      "learning_rate": 1.4717882302540454e-06,
      "loss": 11.7127,
      "step": 34780
    },
    {
      "epoch": 1.8939644551537305,
      "grad_norm": 0.5563173794132958,
      "learning_rate": 1.4702812547273793e-06,
      "loss": 11.7098,
      "step": 34781
    },
    {
      "epoch": 1.8940189091503135,
      "grad_norm": 0.5431409688264829,
      "learning_rate": 1.4687750453843296e-06,
      "loss": 11.8068,
      "step": 34782
    },
    {
      "epoch": 1.8940733631468964,
      "grad_norm": 0.5453814382357687,
      "learning_rate": 1.4672696022366095e-06,
      "loss": 11.839,
      "step": 34783
    },
    {
      "epoch": 1.8941278171434794,
      "grad_norm": 0.5824055691910709,
      "learning_rate": 1.4657649252959204e-06,
      "loss": 11.8123,
      "step": 34784
    },
    {
      "epoch": 1.8941822711400624,
      "grad_norm": 0.5229939638108031,
      "learning_rate": 1.4642610145739755e-06,
      "loss": 11.6223,
      "step": 34785
    },
    {
      "epoch": 1.8942367251366456,
      "grad_norm": 0.5515915628150705,
      "learning_rate": 1.462757870082454e-06,
      "loss": 11.7954,
      "step": 34786
    },
    {
      "epoch": 1.8942911791332286,
      "grad_norm": 0.5806034845993756,
      "learning_rate": 1.4612554918330579e-06,
      "loss": 11.6721,
      "step": 34787
    },
    {
      "epoch": 1.8943456331298116,
      "grad_norm": 0.5664725552415001,
      "learning_rate": 1.4597538798374554e-06,
      "loss": 11.6593,
      "step": 34788
    },
    {
      "epoch": 1.8944000871263946,
      "grad_norm": 0.5032714377117913,
      "learning_rate": 1.4582530341073487e-06,
      "loss": 11.7285,
      "step": 34789
    },
    {
      "epoch": 1.8944545411229776,
      "grad_norm": 0.5413865182310977,
      "learning_rate": 1.4567529546543833e-06,
      "loss": 11.8393,
      "step": 34790
    },
    {
      "epoch": 1.8945089951195606,
      "grad_norm": 0.5406849285193681,
      "learning_rate": 1.4552536414902285e-06,
      "loss": 11.7651,
      "step": 34791
    },
    {
      "epoch": 1.8945634491161436,
      "grad_norm": 0.5438037667417679,
      "learning_rate": 1.453755094626552e-06,
      "loss": 11.8675,
      "step": 34792
    },
    {
      "epoch": 1.8946179031127266,
      "grad_norm": 0.48452529742500633,
      "learning_rate": 1.4522573140750008e-06,
      "loss": 11.7498,
      "step": 34793
    },
    {
      "epoch": 1.8946723571093096,
      "grad_norm": 0.5504460965053491,
      "learning_rate": 1.4507602998472204e-06,
      "loss": 11.758,
      "step": 34794
    },
    {
      "epoch": 1.8947268111058926,
      "grad_norm": 0.5358206341292236,
      "learning_rate": 1.4492640519548572e-06,
      "loss": 11.8275,
      "step": 34795
    },
    {
      "epoch": 1.8947812651024756,
      "grad_norm": 0.5243397355325661,
      "learning_rate": 1.4477685704095356e-06,
      "loss": 11.8239,
      "step": 34796
    },
    {
      "epoch": 1.8948357190990586,
      "grad_norm": 0.5464946328937796,
      "learning_rate": 1.4462738552229128e-06,
      "loss": 11.7493,
      "step": 34797
    },
    {
      "epoch": 1.8948901730956416,
      "grad_norm": 0.6018134686648586,
      "learning_rate": 1.444779906406568e-06,
      "loss": 11.7378,
      "step": 34798
    },
    {
      "epoch": 1.8949446270922246,
      "grad_norm": 0.550844955802657,
      "learning_rate": 1.4432867239721592e-06,
      "loss": 11.7504,
      "step": 34799
    },
    {
      "epoch": 1.8949990810888075,
      "grad_norm": 0.5681524069504759,
      "learning_rate": 1.4417943079312768e-06,
      "loss": 11.6855,
      "step": 34800
    },
    {
      "epoch": 1.8950535350853905,
      "grad_norm": 0.6080234433877981,
      "learning_rate": 1.4403026582955337e-06,
      "loss": 11.8908,
      "step": 34801
    },
    {
      "epoch": 1.8951079890819735,
      "grad_norm": 0.5256429397977135,
      "learning_rate": 1.4388117750765207e-06,
      "loss": 11.6753,
      "step": 34802
    },
    {
      "epoch": 1.8951624430785567,
      "grad_norm": 0.5545665709539822,
      "learning_rate": 1.4373216582858285e-06,
      "loss": 11.6976,
      "step": 34803
    },
    {
      "epoch": 1.8952168970751397,
      "grad_norm": 0.5014566453021978,
      "learning_rate": 1.43583230793507e-06,
      "loss": 11.7029,
      "step": 34804
    },
    {
      "epoch": 1.8952713510717227,
      "grad_norm": 0.538242420412618,
      "learning_rate": 1.4343437240357916e-06,
      "loss": 11.7059,
      "step": 34805
    },
    {
      "epoch": 1.8953258050683057,
      "grad_norm": 0.5268630452561806,
      "learning_rate": 1.432855906599595e-06,
      "loss": 11.6758,
      "step": 34806
    },
    {
      "epoch": 1.8953802590648887,
      "grad_norm": 0.586093835070148,
      "learning_rate": 1.4313688556380377e-06,
      "loss": 11.7773,
      "step": 34807
    },
    {
      "epoch": 1.8954347130614717,
      "grad_norm": 0.5295547503407131,
      "learning_rate": 1.4298825711626884e-06,
      "loss": 11.683,
      "step": 34808
    },
    {
      "epoch": 1.895489167058055,
      "grad_norm": 0.5564299490582887,
      "learning_rate": 1.4283970531851044e-06,
      "loss": 11.7742,
      "step": 34809
    },
    {
      "epoch": 1.895543621054638,
      "grad_norm": 0.6002470713604036,
      "learning_rate": 1.4269123017168318e-06,
      "loss": 11.8173,
      "step": 34810
    },
    {
      "epoch": 1.895598075051221,
      "grad_norm": 0.6225778379082046,
      "learning_rate": 1.4254283167694172e-06,
      "loss": 12.0001,
      "step": 34811
    },
    {
      "epoch": 1.895652529047804,
      "grad_norm": 0.5628654435386629,
      "learning_rate": 1.4239450983544068e-06,
      "loss": 11.7122,
      "step": 34812
    },
    {
      "epoch": 1.895706983044387,
      "grad_norm": 0.49712718917347337,
      "learning_rate": 1.4224626464833246e-06,
      "loss": 11.7467,
      "step": 34813
    },
    {
      "epoch": 1.8957614370409699,
      "grad_norm": 0.5111721505786188,
      "learning_rate": 1.420980961167717e-06,
      "loss": 11.7691,
      "step": 34814
    },
    {
      "epoch": 1.8958158910375529,
      "grad_norm": 0.5331358144303171,
      "learning_rate": 1.4195000424190751e-06,
      "loss": 11.8331,
      "step": 34815
    },
    {
      "epoch": 1.8958703450341359,
      "grad_norm": 0.6406120784888542,
      "learning_rate": 1.418019890248956e-06,
      "loss": 11.6602,
      "step": 34816
    },
    {
      "epoch": 1.8959247990307189,
      "grad_norm": 0.5092941387868579,
      "learning_rate": 1.4165405046688285e-06,
      "loss": 11.7415,
      "step": 34817
    },
    {
      "epoch": 1.8959792530273019,
      "grad_norm": 0.5475993714739043,
      "learning_rate": 1.4150618856902164e-06,
      "loss": 11.6683,
      "step": 34818
    },
    {
      "epoch": 1.8960337070238849,
      "grad_norm": 0.5204246077710504,
      "learning_rate": 1.4135840333246219e-06,
      "loss": 11.8169,
      "step": 34819
    },
    {
      "epoch": 1.8960881610204678,
      "grad_norm": 0.5832453250973386,
      "learning_rate": 1.4121069475835247e-06,
      "loss": 11.8012,
      "step": 34820
    },
    {
      "epoch": 1.8961426150170508,
      "grad_norm": 0.5604925075904181,
      "learning_rate": 1.4106306284784265e-06,
      "loss": 11.7356,
      "step": 34821
    },
    {
      "epoch": 1.8961970690136338,
      "grad_norm": 0.5131718079393366,
      "learning_rate": 1.409155076020785e-06,
      "loss": 11.8441,
      "step": 34822
    },
    {
      "epoch": 1.8962515230102168,
      "grad_norm": 0.512722172347258,
      "learning_rate": 1.407680290222091e-06,
      "loss": 11.812,
      "step": 34823
    },
    {
      "epoch": 1.8963059770067998,
      "grad_norm": 0.6201351620301891,
      "learning_rate": 1.406206271093824e-06,
      "loss": 11.8832,
      "step": 34824
    },
    {
      "epoch": 1.8963604310033828,
      "grad_norm": 0.5031496210608059,
      "learning_rate": 1.4047330186474085e-06,
      "loss": 11.6435,
      "step": 34825
    },
    {
      "epoch": 1.8964148849999658,
      "grad_norm": 0.5593957168834991,
      "learning_rate": 1.403260532894346e-06,
      "loss": 11.8077,
      "step": 34826
    },
    {
      "epoch": 1.896469338996549,
      "grad_norm": 0.5424184431281923,
      "learning_rate": 1.4017888138460388e-06,
      "loss": 11.7674,
      "step": 34827
    },
    {
      "epoch": 1.896523792993132,
      "grad_norm": 0.5589082158369856,
      "learning_rate": 1.4003178615139777e-06,
      "loss": 11.7461,
      "step": 34828
    },
    {
      "epoch": 1.896578246989715,
      "grad_norm": 0.5458886550404423,
      "learning_rate": 1.3988476759095758e-06,
      "loss": 11.7457,
      "step": 34829
    },
    {
      "epoch": 1.896632700986298,
      "grad_norm": 0.5190695557372617,
      "learning_rate": 1.3973782570442684e-06,
      "loss": 11.7606,
      "step": 34830
    },
    {
      "epoch": 1.896687154982881,
      "grad_norm": 0.5457732202297492,
      "learning_rate": 1.3959096049294795e-06,
      "loss": 11.7472,
      "step": 34831
    },
    {
      "epoch": 1.8967416089794642,
      "grad_norm": 0.5009467775369594,
      "learning_rate": 1.3944417195766335e-06,
      "loss": 11.8235,
      "step": 34832
    },
    {
      "epoch": 1.8967960629760472,
      "grad_norm": 0.5759621152852069,
      "learning_rate": 1.3929746009971433e-06,
      "loss": 11.7632,
      "step": 34833
    },
    {
      "epoch": 1.8968505169726302,
      "grad_norm": 0.5048663985822598,
      "learning_rate": 1.3915082492024334e-06,
      "loss": 11.7819,
      "step": 34834
    },
    {
      "epoch": 1.8969049709692132,
      "grad_norm": 0.49344584842703065,
      "learning_rate": 1.3900426642038721e-06,
      "loss": 11.6221,
      "step": 34835
    },
    {
      "epoch": 1.8969594249657962,
      "grad_norm": 0.5586343453616,
      "learning_rate": 1.388577846012884e-06,
      "loss": 11.8319,
      "step": 34836
    },
    {
      "epoch": 1.8970138789623792,
      "grad_norm": 0.5940360802026438,
      "learning_rate": 1.3871137946408597e-06,
      "loss": 11.9579,
      "step": 34837
    },
    {
      "epoch": 1.8970683329589622,
      "grad_norm": 0.5076628703369054,
      "learning_rate": 1.3856505100991678e-06,
      "loss": 11.7297,
      "step": 34838
    },
    {
      "epoch": 1.8971227869555451,
      "grad_norm": 0.6115903399663384,
      "learning_rate": 1.3841879923991885e-06,
      "loss": 11.8168,
      "step": 34839
    },
    {
      "epoch": 1.8971772409521281,
      "grad_norm": 0.5392599585338174,
      "learning_rate": 1.3827262415523124e-06,
      "loss": 11.814,
      "step": 34840
    },
    {
      "epoch": 1.8972316949487111,
      "grad_norm": 0.5822170653633048,
      "learning_rate": 1.381265257569897e-06,
      "loss": 11.772,
      "step": 34841
    },
    {
      "epoch": 1.8972861489452941,
      "grad_norm": 0.5396554052281479,
      "learning_rate": 1.3798050404633e-06,
      "loss": 11.735,
      "step": 34842
    },
    {
      "epoch": 1.8973406029418771,
      "grad_norm": 0.517078631050961,
      "learning_rate": 1.3783455902438792e-06,
      "loss": 11.692,
      "step": 34843
    },
    {
      "epoch": 1.8973950569384601,
      "grad_norm": 0.5173106123180002,
      "learning_rate": 1.3768869069229695e-06,
      "loss": 11.8019,
      "step": 34844
    },
    {
      "epoch": 1.897449510935043,
      "grad_norm": 0.5681062754518449,
      "learning_rate": 1.3754289905119288e-06,
      "loss": 11.6775,
      "step": 34845
    },
    {
      "epoch": 1.897503964931626,
      "grad_norm": 0.6031698787342817,
      "learning_rate": 1.3739718410221037e-06,
      "loss": 11.831,
      "step": 34846
    },
    {
      "epoch": 1.897558418928209,
      "grad_norm": 0.49968257780963066,
      "learning_rate": 1.372515458464796e-06,
      "loss": 11.8007,
      "step": 34847
    },
    {
      "epoch": 1.897612872924792,
      "grad_norm": 0.5205453754959796,
      "learning_rate": 1.3710598428513633e-06,
      "loss": 11.6712,
      "step": 34848
    },
    {
      "epoch": 1.897667326921375,
      "grad_norm": 0.5691755387365655,
      "learning_rate": 1.3696049941930967e-06,
      "loss": 11.8457,
      "step": 34849
    },
    {
      "epoch": 1.8977217809179583,
      "grad_norm": 0.49082575073352686,
      "learning_rate": 1.3681509125013314e-06,
      "loss": 11.7527,
      "step": 34850
    },
    {
      "epoch": 1.8977762349145413,
      "grad_norm": 0.5544832287565732,
      "learning_rate": 1.3666975977873697e-06,
      "loss": 11.8401,
      "step": 34851
    },
    {
      "epoch": 1.8978306889111243,
      "grad_norm": 0.5811081783586601,
      "learning_rate": 1.3652450500625026e-06,
      "loss": 11.7448,
      "step": 34852
    },
    {
      "epoch": 1.8978851429077073,
      "grad_norm": 0.5831570482655312,
      "learning_rate": 1.3637932693380318e-06,
      "loss": 11.7733,
      "step": 34853
    },
    {
      "epoch": 1.8979395969042903,
      "grad_norm": 0.5353805915594402,
      "learning_rate": 1.3623422556252375e-06,
      "loss": 11.7708,
      "step": 34854
    },
    {
      "epoch": 1.8979940509008733,
      "grad_norm": 0.6718657003779454,
      "learning_rate": 1.3608920089354217e-06,
      "loss": 11.8133,
      "step": 34855
    },
    {
      "epoch": 1.8980485048974565,
      "grad_norm": 0.5680055538659191,
      "learning_rate": 1.3594425292798418e-06,
      "loss": 11.7966,
      "step": 34856
    },
    {
      "epoch": 1.8981029588940395,
      "grad_norm": 0.5177834196991179,
      "learning_rate": 1.3579938166697892e-06,
      "loss": 11.726,
      "step": 34857
    },
    {
      "epoch": 1.8981574128906225,
      "grad_norm": 0.5225753726577712,
      "learning_rate": 1.3565458711165101e-06,
      "loss": 11.7457,
      "step": 34858
    },
    {
      "epoch": 1.8982118668872054,
      "grad_norm": 0.48728829065372314,
      "learning_rate": 1.3550986926312736e-06,
      "loss": 11.7121,
      "step": 34859
    },
    {
      "epoch": 1.8982663208837884,
      "grad_norm": 0.5293433402269976,
      "learning_rate": 1.353652281225337e-06,
      "loss": 11.6661,
      "step": 34860
    },
    {
      "epoch": 1.8983207748803714,
      "grad_norm": 0.6206070880327738,
      "learning_rate": 1.3522066369099473e-06,
      "loss": 11.8236,
      "step": 34861
    },
    {
      "epoch": 1.8983752288769544,
      "grad_norm": 0.5734420945731776,
      "learning_rate": 1.3507617596963396e-06,
      "loss": 11.841,
      "step": 34862
    },
    {
      "epoch": 1.8984296828735374,
      "grad_norm": 0.5792092352979687,
      "learning_rate": 1.3493176495957493e-06,
      "loss": 11.8518,
      "step": 34863
    },
    {
      "epoch": 1.8984841368701204,
      "grad_norm": 0.6182898818515133,
      "learning_rate": 1.3478743066194011e-06,
      "loss": 11.7188,
      "step": 34864
    },
    {
      "epoch": 1.8985385908667034,
      "grad_norm": 0.5467828552647913,
      "learning_rate": 1.3464317307785413e-06,
      "loss": 11.7695,
      "step": 34865
    },
    {
      "epoch": 1.8985930448632864,
      "grad_norm": 0.5113883760743647,
      "learning_rate": 1.344989922084361e-06,
      "loss": 11.7157,
      "step": 34866
    },
    {
      "epoch": 1.8986474988598694,
      "grad_norm": 0.5428993289081379,
      "learning_rate": 1.3435488805480845e-06,
      "loss": 11.8631,
      "step": 34867
    },
    {
      "epoch": 1.8987019528564524,
      "grad_norm": 0.5534172518042819,
      "learning_rate": 1.342108606180925e-06,
      "loss": 11.7682,
      "step": 34868
    },
    {
      "epoch": 1.8987564068530354,
      "grad_norm": 0.5327623657273541,
      "learning_rate": 1.340669098994063e-06,
      "loss": 11.6865,
      "step": 34869
    },
    {
      "epoch": 1.8988108608496184,
      "grad_norm": 0.5025208902523031,
      "learning_rate": 1.339230358998711e-06,
      "loss": 11.8081,
      "step": 34870
    },
    {
      "epoch": 1.8988653148462014,
      "grad_norm": 0.5593433130700441,
      "learning_rate": 1.3377923862060493e-06,
      "loss": 11.7972,
      "step": 34871
    },
    {
      "epoch": 1.8989197688427844,
      "grad_norm": 0.5229594386061016,
      "learning_rate": 1.3363551806272578e-06,
      "loss": 11.7237,
      "step": 34872
    },
    {
      "epoch": 1.8989742228393676,
      "grad_norm": 0.5664251049278367,
      "learning_rate": 1.3349187422735166e-06,
      "loss": 11.746,
      "step": 34873
    },
    {
      "epoch": 1.8990286768359506,
      "grad_norm": 0.5076171833846765,
      "learning_rate": 1.3334830711559832e-06,
      "loss": 11.8043,
      "step": 34874
    },
    {
      "epoch": 1.8990831308325336,
      "grad_norm": 0.611139018203511,
      "learning_rate": 1.3320481672858487e-06,
      "loss": 11.8499,
      "step": 34875
    },
    {
      "epoch": 1.8991375848291165,
      "grad_norm": 0.5403750894731288,
      "learning_rate": 1.3306140306742377e-06,
      "loss": 11.7036,
      "step": 34876
    },
    {
      "epoch": 1.8991920388256995,
      "grad_norm": 0.5165284539906411,
      "learning_rate": 1.32918066133233e-06,
      "loss": 11.7363,
      "step": 34877
    },
    {
      "epoch": 1.8992464928222825,
      "grad_norm": 0.6575863872562808,
      "learning_rate": 1.32774805927125e-06,
      "loss": 11.8829,
      "step": 34878
    },
    {
      "epoch": 1.8993009468188657,
      "grad_norm": 0.5447377032095552,
      "learning_rate": 1.3263162245021553e-06,
      "loss": 11.8546,
      "step": 34879
    },
    {
      "epoch": 1.8993554008154487,
      "grad_norm": 0.5401548915126622,
      "learning_rate": 1.324885157036171e-06,
      "loss": 11.8171,
      "step": 34880
    },
    {
      "epoch": 1.8994098548120317,
      "grad_norm": 0.6040337382301092,
      "learning_rate": 1.3234548568844318e-06,
      "loss": 11.8438,
      "step": 34881
    },
    {
      "epoch": 1.8994643088086147,
      "grad_norm": 0.518418596072161,
      "learning_rate": 1.3220253240580516e-06,
      "loss": 11.7245,
      "step": 34882
    },
    {
      "epoch": 1.8995187628051977,
      "grad_norm": 0.5037391785542648,
      "learning_rate": 1.3205965585681545e-06,
      "loss": 11.7445,
      "step": 34883
    },
    {
      "epoch": 1.8995732168017807,
      "grad_norm": 0.530468799518843,
      "learning_rate": 1.319168560425843e-06,
      "loss": 11.7476,
      "step": 34884
    },
    {
      "epoch": 1.8996276707983637,
      "grad_norm": 0.5375853461548761,
      "learning_rate": 1.3177413296422303e-06,
      "loss": 11.7333,
      "step": 34885
    },
    {
      "epoch": 1.8996821247949467,
      "grad_norm": 0.5772703460003947,
      "learning_rate": 1.3163148662284074e-06,
      "loss": 11.837,
      "step": 34886
    },
    {
      "epoch": 1.8997365787915297,
      "grad_norm": 0.4990338475383423,
      "learning_rate": 1.3148891701954768e-06,
      "loss": 11.7326,
      "step": 34887
    },
    {
      "epoch": 1.8997910327881127,
      "grad_norm": 0.5248471910613208,
      "learning_rate": 1.3134642415545074e-06,
      "loss": 11.8689,
      "step": 34888
    },
    {
      "epoch": 1.8998454867846957,
      "grad_norm": 0.5597959271870933,
      "learning_rate": 1.31204008031659e-06,
      "loss": 11.8377,
      "step": 34889
    },
    {
      "epoch": 1.8998999407812787,
      "grad_norm": 0.5805332350728085,
      "learning_rate": 1.3106166864928048e-06,
      "loss": 11.7712,
      "step": 34890
    },
    {
      "epoch": 1.8999543947778617,
      "grad_norm": 0.5226958549523403,
      "learning_rate": 1.30919406009421e-06,
      "loss": 11.865,
      "step": 34891
    },
    {
      "epoch": 1.9000088487744446,
      "grad_norm": 0.4992119103294987,
      "learning_rate": 1.3077722011318738e-06,
      "loss": 11.5739,
      "step": 34892
    },
    {
      "epoch": 1.9000633027710276,
      "grad_norm": 0.5891212049804477,
      "learning_rate": 1.3063511096168546e-06,
      "loss": 11.8646,
      "step": 34893
    },
    {
      "epoch": 1.9001177567676106,
      "grad_norm": 0.5596632414305603,
      "learning_rate": 1.3049307855601877e-06,
      "loss": 11.864,
      "step": 34894
    },
    {
      "epoch": 1.9001722107641936,
      "grad_norm": 0.5521483647653863,
      "learning_rate": 1.3035112289729422e-06,
      "loss": 11.6744,
      "step": 34895
    },
    {
      "epoch": 1.9002266647607768,
      "grad_norm": 0.5352249635968738,
      "learning_rate": 1.3020924398661426e-06,
      "loss": 11.7585,
      "step": 34896
    },
    {
      "epoch": 1.9002811187573598,
      "grad_norm": 0.7299639279554151,
      "learning_rate": 1.3006744182508245e-06,
      "loss": 11.6792,
      "step": 34897
    },
    {
      "epoch": 1.9003355727539428,
      "grad_norm": 0.5527762706527218,
      "learning_rate": 1.2992571641380124e-06,
      "loss": 11.7747,
      "step": 34898
    },
    {
      "epoch": 1.9003900267505258,
      "grad_norm": 0.5634854671760061,
      "learning_rate": 1.297840677538731e-06,
      "loss": 11.625,
      "step": 34899
    },
    {
      "epoch": 1.9004444807471088,
      "grad_norm": 0.5102684204527475,
      "learning_rate": 1.296424958463993e-06,
      "loss": 11.72,
      "step": 34900
    },
    {
      "epoch": 1.9004989347436918,
      "grad_norm": 0.5235308436576003,
      "learning_rate": 1.2950100069248018e-06,
      "loss": 11.8088,
      "step": 34901
    },
    {
      "epoch": 1.900553388740275,
      "grad_norm": 0.5487990887186326,
      "learning_rate": 1.2935958229321698e-06,
      "loss": 11.6855,
      "step": 34902
    },
    {
      "epoch": 1.900607842736858,
      "grad_norm": 0.5080858442369418,
      "learning_rate": 1.292182406497089e-06,
      "loss": 11.8006,
      "step": 34903
    },
    {
      "epoch": 1.900662296733441,
      "grad_norm": 0.5764407085076949,
      "learning_rate": 1.29076975763055e-06,
      "loss": 11.6335,
      "step": 34904
    },
    {
      "epoch": 1.900716750730024,
      "grad_norm": 0.5435858284155105,
      "learning_rate": 1.2893578763435443e-06,
      "loss": 11.6927,
      "step": 34905
    },
    {
      "epoch": 1.900771204726607,
      "grad_norm": 0.611928884447569,
      "learning_rate": 1.2879467626470298e-06,
      "loss": 11.519,
      "step": 34906
    },
    {
      "epoch": 1.90082565872319,
      "grad_norm": 0.520866001968785,
      "learning_rate": 1.28653641655202e-06,
      "loss": 11.747,
      "step": 34907
    },
    {
      "epoch": 1.900880112719773,
      "grad_norm": 0.5933677414894604,
      "learning_rate": 1.2851268380694393e-06,
      "loss": 11.8656,
      "step": 34908
    },
    {
      "epoch": 1.900934566716356,
      "grad_norm": 0.5794399435556268,
      "learning_rate": 1.2837180272102678e-06,
      "loss": 11.8481,
      "step": 34909
    },
    {
      "epoch": 1.900989020712939,
      "grad_norm": 0.5638927687898467,
      "learning_rate": 1.2823099839854636e-06,
      "loss": 11.7772,
      "step": 34910
    },
    {
      "epoch": 1.901043474709522,
      "grad_norm": 0.5723553513297639,
      "learning_rate": 1.2809027084059733e-06,
      "loss": 11.6843,
      "step": 34911
    },
    {
      "epoch": 1.901097928706105,
      "grad_norm": 0.5956225606954836,
      "learning_rate": 1.2794962004827437e-06,
      "loss": 11.6345,
      "step": 34912
    },
    {
      "epoch": 1.901152382702688,
      "grad_norm": 0.5187793448726408,
      "learning_rate": 1.2780904602266997e-06,
      "loss": 11.7602,
      "step": 34913
    },
    {
      "epoch": 1.901206836699271,
      "grad_norm": 0.6625625031443545,
      "learning_rate": 1.2766854876487766e-06,
      "loss": 11.7441,
      "step": 34914
    },
    {
      "epoch": 1.901261290695854,
      "grad_norm": 0.6301397190904372,
      "learning_rate": 1.2752812827599103e-06,
      "loss": 11.6538,
      "step": 34915
    },
    {
      "epoch": 1.901315744692437,
      "grad_norm": 0.5569390578766883,
      "learning_rate": 1.2738778455710031e-06,
      "loss": 11.7446,
      "step": 34916
    },
    {
      "epoch": 1.90137019868902,
      "grad_norm": 0.5747944201762731,
      "learning_rate": 1.272475176092991e-06,
      "loss": 11.7425,
      "step": 34917
    },
    {
      "epoch": 1.901424652685603,
      "grad_norm": 0.5872191550488564,
      "learning_rate": 1.2710732743367536e-06,
      "loss": 11.7856,
      "step": 34918
    },
    {
      "epoch": 1.901479106682186,
      "grad_norm": 0.5258704340142019,
      "learning_rate": 1.2696721403132273e-06,
      "loss": 11.8165,
      "step": 34919
    },
    {
      "epoch": 1.901533560678769,
      "grad_norm": 0.5250324522894962,
      "learning_rate": 1.2682717740332694e-06,
      "loss": 11.764,
      "step": 34920
    },
    {
      "epoch": 1.901588014675352,
      "grad_norm": 0.5188689434614271,
      "learning_rate": 1.266872175507794e-06,
      "loss": 11.6813,
      "step": 34921
    },
    {
      "epoch": 1.901642468671935,
      "grad_norm": 0.526756402285969,
      "learning_rate": 1.2654733447476807e-06,
      "loss": 11.7247,
      "step": 34922
    },
    {
      "epoch": 1.901696922668518,
      "grad_norm": 0.5317676067744854,
      "learning_rate": 1.2640752817637992e-06,
      "loss": 11.8025,
      "step": 34923
    },
    {
      "epoch": 1.901751376665101,
      "grad_norm": 0.5196343589005462,
      "learning_rate": 1.2626779865670403e-06,
      "loss": 11.7772,
      "step": 34924
    },
    {
      "epoch": 1.901805830661684,
      "grad_norm": 0.5156707167033562,
      "learning_rate": 1.2612814591682399e-06,
      "loss": 11.8267,
      "step": 34925
    },
    {
      "epoch": 1.9018602846582673,
      "grad_norm": 0.6338894852183763,
      "learning_rate": 1.2598856995782782e-06,
      "loss": 11.8186,
      "step": 34926
    },
    {
      "epoch": 1.9019147386548503,
      "grad_norm": 0.5402170246045844,
      "learning_rate": 1.258490707808002e-06,
      "loss": 11.8636,
      "step": 34927
    },
    {
      "epoch": 1.9019691926514333,
      "grad_norm": 0.5225756205200797,
      "learning_rate": 1.2570964838682586e-06,
      "loss": 11.7981,
      "step": 34928
    },
    {
      "epoch": 1.9020236466480163,
      "grad_norm": 0.5036981029513681,
      "learning_rate": 1.2557030277698945e-06,
      "loss": 11.8473,
      "step": 34929
    },
    {
      "epoch": 1.9020781006445993,
      "grad_norm": 0.5302650787095671,
      "learning_rate": 1.2543103395237454e-06,
      "loss": 11.7383,
      "step": 34930
    },
    {
      "epoch": 1.9021325546411822,
      "grad_norm": 0.5640592816036747,
      "learning_rate": 1.2529184191406474e-06,
      "loss": 11.8471,
      "step": 34931
    },
    {
      "epoch": 1.9021870086377652,
      "grad_norm": 0.5283403228744005,
      "learning_rate": 1.2515272666314027e-06,
      "loss": 11.663,
      "step": 34932
    },
    {
      "epoch": 1.9022414626343482,
      "grad_norm": 0.5660956425878506,
      "learning_rate": 1.250136882006847e-06,
      "loss": 11.7854,
      "step": 34933
    },
    {
      "epoch": 1.9022959166309312,
      "grad_norm": 0.5202634141052236,
      "learning_rate": 1.2487472652777832e-06,
      "loss": 11.764,
      "step": 34934
    },
    {
      "epoch": 1.9023503706275142,
      "grad_norm": 0.5103911372556909,
      "learning_rate": 1.2473584164550245e-06,
      "loss": 11.6607,
      "step": 34935
    },
    {
      "epoch": 1.9024048246240972,
      "grad_norm": 0.5184577260363474,
      "learning_rate": 1.2459703355493736e-06,
      "loss": 11.7746,
      "step": 34936
    },
    {
      "epoch": 1.9024592786206802,
      "grad_norm": 0.5519968341127948,
      "learning_rate": 1.2445830225716104e-06,
      "loss": 11.6335,
      "step": 34937
    },
    {
      "epoch": 1.9025137326172632,
      "grad_norm": 0.5758726591964479,
      "learning_rate": 1.2431964775325267e-06,
      "loss": 11.8986,
      "step": 34938
    },
    {
      "epoch": 1.9025681866138462,
      "grad_norm": 0.5901030153822988,
      "learning_rate": 1.2418107004429135e-06,
      "loss": 11.793,
      "step": 34939
    },
    {
      "epoch": 1.9026226406104292,
      "grad_norm": 0.5969603883994249,
      "learning_rate": 1.2404256913135404e-06,
      "loss": 11.8338,
      "step": 34940
    },
    {
      "epoch": 1.9026770946070122,
      "grad_norm": 0.5266099654687076,
      "learning_rate": 1.2390414501551984e-06,
      "loss": 11.7505,
      "step": 34941
    },
    {
      "epoch": 1.9027315486035952,
      "grad_norm": 0.5501564057848704,
      "learning_rate": 1.2376579769786124e-06,
      "loss": 11.7582,
      "step": 34942
    },
    {
      "epoch": 1.9027860026001784,
      "grad_norm": 0.5913755801523414,
      "learning_rate": 1.2362752717945625e-06,
      "loss": 11.7118,
      "step": 34943
    },
    {
      "epoch": 1.9028404565967614,
      "grad_norm": 0.5740167269089241,
      "learning_rate": 1.2348933346137958e-06,
      "loss": 11.7911,
      "step": 34944
    },
    {
      "epoch": 1.9028949105933444,
      "grad_norm": 0.49452260907208945,
      "learning_rate": 1.2335121654470705e-06,
      "loss": 11.6782,
      "step": 34945
    },
    {
      "epoch": 1.9029493645899274,
      "grad_norm": 0.5383000224602161,
      "learning_rate": 1.232131764305111e-06,
      "loss": 11.844,
      "step": 34946
    },
    {
      "epoch": 1.9030038185865104,
      "grad_norm": 0.5316669763904741,
      "learning_rate": 1.2307521311986536e-06,
      "loss": 11.8824,
      "step": 34947
    },
    {
      "epoch": 1.9030582725830933,
      "grad_norm": 0.5807140241198657,
      "learning_rate": 1.229373266138445e-06,
      "loss": 11.7666,
      "step": 34948
    },
    {
      "epoch": 1.9031127265796766,
      "grad_norm": 0.5643874459202864,
      "learning_rate": 1.2279951691351876e-06,
      "loss": 11.8332,
      "step": 34949
    },
    {
      "epoch": 1.9031671805762596,
      "grad_norm": 0.5849060923044125,
      "learning_rate": 1.2266178401995954e-06,
      "loss": 11.7471,
      "step": 34950
    },
    {
      "epoch": 1.9032216345728425,
      "grad_norm": 0.5169059632750649,
      "learning_rate": 1.225241279342404e-06,
      "loss": 11.8156,
      "step": 34951
    },
    {
      "epoch": 1.9032760885694255,
      "grad_norm": 0.5676538317201736,
      "learning_rate": 1.2238654865742938e-06,
      "loss": 11.7099,
      "step": 34952
    },
    {
      "epoch": 1.9033305425660085,
      "grad_norm": 0.555361791346935,
      "learning_rate": 1.2224904619059673e-06,
      "loss": 11.8238,
      "step": 34953
    },
    {
      "epoch": 1.9033849965625915,
      "grad_norm": 0.5340937115298117,
      "learning_rate": 1.2211162053481162e-06,
      "loss": 11.7109,
      "step": 34954
    },
    {
      "epoch": 1.9034394505591745,
      "grad_norm": 0.5149744115786244,
      "learning_rate": 1.2197427169114317e-06,
      "loss": 11.6893,
      "step": 34955
    },
    {
      "epoch": 1.9034939045557575,
      "grad_norm": 0.500862523279399,
      "learning_rate": 1.2183699966066052e-06,
      "loss": 11.765,
      "step": 34956
    },
    {
      "epoch": 1.9035483585523405,
      "grad_norm": 0.5919726300378692,
      "learning_rate": 1.216998044444284e-06,
      "loss": 11.7503,
      "step": 34957
    },
    {
      "epoch": 1.9036028125489235,
      "grad_norm": 0.5216627565954235,
      "learning_rate": 1.2156268604351707e-06,
      "loss": 11.7937,
      "step": 34958
    },
    {
      "epoch": 1.9036572665455065,
      "grad_norm": 0.5350789227643593,
      "learning_rate": 1.2142564445898896e-06,
      "loss": 11.7579,
      "step": 34959
    },
    {
      "epoch": 1.9037117205420895,
      "grad_norm": 0.5672783880458488,
      "learning_rate": 1.2128867969191326e-06,
      "loss": 11.7553,
      "step": 34960
    },
    {
      "epoch": 1.9037661745386725,
      "grad_norm": 0.4998683821936918,
      "learning_rate": 1.2115179174335244e-06,
      "loss": 11.796,
      "step": 34961
    },
    {
      "epoch": 1.9038206285352555,
      "grad_norm": 0.49377638318305933,
      "learning_rate": 1.2101498061437234e-06,
      "loss": 11.7123,
      "step": 34962
    },
    {
      "epoch": 1.9038750825318385,
      "grad_norm": 0.5317072407810388,
      "learning_rate": 1.208782463060365e-06,
      "loss": 11.7181,
      "step": 34963
    },
    {
      "epoch": 1.9039295365284215,
      "grad_norm": 0.6067866019767924,
      "learning_rate": 1.2074158881940744e-06,
      "loss": 11.7319,
      "step": 34964
    },
    {
      "epoch": 1.9039839905250044,
      "grad_norm": 0.5461599204792117,
      "learning_rate": 1.2060500815554877e-06,
      "loss": 11.7649,
      "step": 34965
    },
    {
      "epoch": 1.9040384445215877,
      "grad_norm": 0.5512674428664004,
      "learning_rate": 1.2046850431552293e-06,
      "loss": 11.8201,
      "step": 34966
    },
    {
      "epoch": 1.9040928985181707,
      "grad_norm": 0.551134458254267,
      "learning_rate": 1.203320773003902e-06,
      "loss": 11.7467,
      "step": 34967
    },
    {
      "epoch": 1.9041473525147536,
      "grad_norm": 0.49863207282143507,
      "learning_rate": 1.2019572711121196e-06,
      "loss": 11.659,
      "step": 34968
    },
    {
      "epoch": 1.9042018065113366,
      "grad_norm": 0.5695888542949131,
      "learning_rate": 1.2005945374904958e-06,
      "loss": 11.7296,
      "step": 34969
    },
    {
      "epoch": 1.9042562605079196,
      "grad_norm": 0.5737764133068176,
      "learning_rate": 1.199232572149611e-06,
      "loss": 11.8685,
      "step": 34970
    },
    {
      "epoch": 1.9043107145045026,
      "grad_norm": 0.6623615169926745,
      "learning_rate": 1.197871375100068e-06,
      "loss": 11.8326,
      "step": 34971
    },
    {
      "epoch": 1.9043651685010858,
      "grad_norm": 0.5531808203617824,
      "learning_rate": 1.196510946352436e-06,
      "loss": 11.6211,
      "step": 34972
    },
    {
      "epoch": 1.9044196224976688,
      "grad_norm": 0.5011029737950699,
      "learning_rate": 1.1951512859173064e-06,
      "loss": 11.7735,
      "step": 34973
    },
    {
      "epoch": 1.9044740764942518,
      "grad_norm": 0.5635643958075988,
      "learning_rate": 1.1937923938052598e-06,
      "loss": 11.7972,
      "step": 34974
    },
    {
      "epoch": 1.9045285304908348,
      "grad_norm": 0.4859862049418657,
      "learning_rate": 1.1924342700268431e-06,
      "loss": 11.7243,
      "step": 34975
    },
    {
      "epoch": 1.9045829844874178,
      "grad_norm": 0.5534201210675502,
      "learning_rate": 1.191076914592626e-06,
      "loss": 11.9276,
      "step": 34976
    },
    {
      "epoch": 1.9046374384840008,
      "grad_norm": 0.5011984913829557,
      "learning_rate": 1.1897203275131664e-06,
      "loss": 11.7358,
      "step": 34977
    },
    {
      "epoch": 1.9046918924805838,
      "grad_norm": 0.5675792748840617,
      "learning_rate": 1.1883645087990224e-06,
      "loss": 11.769,
      "step": 34978
    },
    {
      "epoch": 1.9047463464771668,
      "grad_norm": 0.5026309591187752,
      "learning_rate": 1.1870094584607195e-06,
      "loss": 11.5996,
      "step": 34979
    },
    {
      "epoch": 1.9048008004737498,
      "grad_norm": 0.5581656370747227,
      "learning_rate": 1.1856551765088042e-06,
      "loss": 11.77,
      "step": 34980
    },
    {
      "epoch": 1.9048552544703328,
      "grad_norm": 0.5663165010056022,
      "learning_rate": 1.1843016629538128e-06,
      "loss": 11.7731,
      "step": 34981
    },
    {
      "epoch": 1.9049097084669158,
      "grad_norm": 0.5172392838909518,
      "learning_rate": 1.1829489178062481e-06,
      "loss": 11.695,
      "step": 34982
    },
    {
      "epoch": 1.9049641624634988,
      "grad_norm": 0.5497732495772556,
      "learning_rate": 1.181596941076646e-06,
      "loss": 11.8683,
      "step": 34983
    },
    {
      "epoch": 1.9050186164600817,
      "grad_norm": 0.5413920720232699,
      "learning_rate": 1.18024573277552e-06,
      "loss": 11.6641,
      "step": 34984
    },
    {
      "epoch": 1.9050730704566647,
      "grad_norm": 0.547364218464008,
      "learning_rate": 1.1788952929133846e-06,
      "loss": 11.6054,
      "step": 34985
    },
    {
      "epoch": 1.9051275244532477,
      "grad_norm": 0.5487273460843867,
      "learning_rate": 1.1775456215007197e-06,
      "loss": 11.7632,
      "step": 34986
    },
    {
      "epoch": 1.9051819784498307,
      "grad_norm": 0.5150943124030233,
      "learning_rate": 1.1761967185480394e-06,
      "loss": 11.7124,
      "step": 34987
    },
    {
      "epoch": 1.9052364324464137,
      "grad_norm": 0.5506035108844471,
      "learning_rate": 1.174848584065824e-06,
      "loss": 11.584,
      "step": 34988
    },
    {
      "epoch": 1.9052908864429967,
      "grad_norm": 0.5648535428830471,
      "learning_rate": 1.1735012180645543e-06,
      "loss": 11.7641,
      "step": 34989
    },
    {
      "epoch": 1.90534534043958,
      "grad_norm": 0.5546063878312949,
      "learning_rate": 1.1721546205547218e-06,
      "loss": 11.8059,
      "step": 34990
    },
    {
      "epoch": 1.905399794436163,
      "grad_norm": 0.5946641378680784,
      "learning_rate": 1.1708087915467846e-06,
      "loss": 11.8523,
      "step": 34991
    },
    {
      "epoch": 1.905454248432746,
      "grad_norm": 0.5573673365857627,
      "learning_rate": 1.1694637310512125e-06,
      "loss": 11.6701,
      "step": 34992
    },
    {
      "epoch": 1.905508702429329,
      "grad_norm": 0.5404327531472225,
      "learning_rate": 1.1681194390784634e-06,
      "loss": 11.8141,
      "step": 34993
    },
    {
      "epoch": 1.905563156425912,
      "grad_norm": 0.542847902451296,
      "learning_rate": 1.1667759156389847e-06,
      "loss": 11.7001,
      "step": 34994
    },
    {
      "epoch": 1.9056176104224951,
      "grad_norm": 0.5595541408136204,
      "learning_rate": 1.1654331607432457e-06,
      "loss": 11.7201,
      "step": 34995
    },
    {
      "epoch": 1.905672064419078,
      "grad_norm": 0.5288638373784262,
      "learning_rate": 1.1640911744016602e-06,
      "loss": 11.8266,
      "step": 34996
    },
    {
      "epoch": 1.905726518415661,
      "grad_norm": 0.5970397485499701,
      "learning_rate": 1.1627499566246869e-06,
      "loss": 11.7677,
      "step": 34997
    },
    {
      "epoch": 1.905780972412244,
      "grad_norm": 0.5616017549771714,
      "learning_rate": 1.1614095074227282e-06,
      "loss": 11.7861,
      "step": 34998
    },
    {
      "epoch": 1.905835426408827,
      "grad_norm": 0.5635818877416472,
      "learning_rate": 1.1600698268062315e-06,
      "loss": 11.8511,
      "step": 34999
    },
    {
      "epoch": 1.90588988040541,
      "grad_norm": 0.5023035649247424,
      "learning_rate": 1.1587309147856217e-06,
      "loss": 11.7506,
      "step": 35000
    },
    {
      "epoch": 1.905944334401993,
      "grad_norm": 0.613569279577508,
      "learning_rate": 1.1573927713712796e-06,
      "loss": 11.7161,
      "step": 35001
    },
    {
      "epoch": 1.905998788398576,
      "grad_norm": 0.5262782385154369,
      "learning_rate": 1.156055396573641e-06,
      "loss": 11.7928,
      "step": 35002
    },
    {
      "epoch": 1.906053242395159,
      "grad_norm": 0.5671827952736994,
      "learning_rate": 1.1547187904030866e-06,
      "loss": 11.7986,
      "step": 35003
    },
    {
      "epoch": 1.906107696391742,
      "grad_norm": 0.4781231166406721,
      "learning_rate": 1.153382952870008e-06,
      "loss": 11.6468,
      "step": 35004
    },
    {
      "epoch": 1.906162150388325,
      "grad_norm": 0.579281572459171,
      "learning_rate": 1.152047883984808e-06,
      "loss": 11.825,
      "step": 35005
    },
    {
      "epoch": 1.906216604384908,
      "grad_norm": 0.5480966016272341,
      "learning_rate": 1.1507135837578453e-06,
      "loss": 11.8069,
      "step": 35006
    },
    {
      "epoch": 1.906271058381491,
      "grad_norm": 0.5126263720681261,
      "learning_rate": 1.1493800521995334e-06,
      "loss": 11.7038,
      "step": 35007
    },
    {
      "epoch": 1.906325512378074,
      "grad_norm": 0.6769674666132978,
      "learning_rate": 1.1480472893201977e-06,
      "loss": 11.7818,
      "step": 35008
    },
    {
      "epoch": 1.906379966374657,
      "grad_norm": 0.5307353123798643,
      "learning_rate": 1.1467152951302407e-06,
      "loss": 11.7825,
      "step": 35009
    },
    {
      "epoch": 1.90643442037124,
      "grad_norm": 0.5434710656450639,
      "learning_rate": 1.1453840696399986e-06,
      "loss": 11.7149,
      "step": 35010
    },
    {
      "epoch": 1.906488874367823,
      "grad_norm": 0.516724319134081,
      "learning_rate": 1.144053612859819e-06,
      "loss": 11.7374,
      "step": 35011
    },
    {
      "epoch": 1.906543328364406,
      "grad_norm": 0.5074130646145986,
      "learning_rate": 1.142723924800071e-06,
      "loss": 11.7568,
      "step": 35012
    },
    {
      "epoch": 1.9065977823609892,
      "grad_norm": 0.5503949663861676,
      "learning_rate": 1.1413950054710687e-06,
      "loss": 11.849,
      "step": 35013
    },
    {
      "epoch": 1.9066522363575722,
      "grad_norm": 0.5871422964262129,
      "learning_rate": 1.1400668548831595e-06,
      "loss": 11.6125,
      "step": 35014
    },
    {
      "epoch": 1.9067066903541552,
      "grad_norm": 0.5684713531708555,
      "learning_rate": 1.1387394730466793e-06,
      "loss": 11.7334,
      "step": 35015
    },
    {
      "epoch": 1.9067611443507382,
      "grad_norm": 0.5339253585247791,
      "learning_rate": 1.1374128599719314e-06,
      "loss": 11.6277,
      "step": 35016
    },
    {
      "epoch": 1.9068155983473212,
      "grad_norm": 0.6252833041214423,
      "learning_rate": 1.1360870156692405e-06,
      "loss": 11.7459,
      "step": 35017
    },
    {
      "epoch": 1.9068700523439042,
      "grad_norm": 0.5034389970866846,
      "learning_rate": 1.1347619401489206e-06,
      "loss": 11.7586,
      "step": 35018
    },
    {
      "epoch": 1.9069245063404874,
      "grad_norm": 0.47682379643605904,
      "learning_rate": 1.1334376334212638e-06,
      "loss": 11.759,
      "step": 35019
    },
    {
      "epoch": 1.9069789603370704,
      "grad_norm": 0.5630438106014305,
      "learning_rate": 1.1321140954965836e-06,
      "loss": 11.7311,
      "step": 35020
    },
    {
      "epoch": 1.9070334143336534,
      "grad_norm": 0.5712070275988911,
      "learning_rate": 1.13079132638515e-06,
      "loss": 11.7696,
      "step": 35021
    },
    {
      "epoch": 1.9070878683302364,
      "grad_norm": 0.5284279871716235,
      "learning_rate": 1.1294693260972877e-06,
      "loss": 11.7122,
      "step": 35022
    },
    {
      "epoch": 1.9071423223268194,
      "grad_norm": 0.5604633493767001,
      "learning_rate": 1.1281480946432332e-06,
      "loss": 11.8063,
      "step": 35023
    },
    {
      "epoch": 1.9071967763234023,
      "grad_norm": 0.5505765064327027,
      "learning_rate": 1.1268276320332893e-06,
      "loss": 11.7465,
      "step": 35024
    },
    {
      "epoch": 1.9072512303199853,
      "grad_norm": 0.5023365981449948,
      "learning_rate": 1.1255079382777144e-06,
      "loss": 11.7987,
      "step": 35025
    },
    {
      "epoch": 1.9073056843165683,
      "grad_norm": 0.5342114329208129,
      "learning_rate": 1.1241890133867671e-06,
      "loss": 11.7392,
      "step": 35026
    },
    {
      "epoch": 1.9073601383131513,
      "grad_norm": 0.5126656077229695,
      "learning_rate": 1.1228708573707058e-06,
      "loss": 11.819,
      "step": 35027
    },
    {
      "epoch": 1.9074145923097343,
      "grad_norm": 0.5578011430400167,
      "learning_rate": 1.121553470239789e-06,
      "loss": 11.8733,
      "step": 35028
    },
    {
      "epoch": 1.9074690463063173,
      "grad_norm": 0.5135564892159001,
      "learning_rate": 1.1202368520042527e-06,
      "loss": 11.7069,
      "step": 35029
    },
    {
      "epoch": 1.9075235003029003,
      "grad_norm": 0.5355118246894365,
      "learning_rate": 1.1189210026743225e-06,
      "loss": 11.7699,
      "step": 35030
    },
    {
      "epoch": 1.9075779542994833,
      "grad_norm": 0.599579854014161,
      "learning_rate": 1.1176059222602676e-06,
      "loss": 11.919,
      "step": 35031
    },
    {
      "epoch": 1.9076324082960663,
      "grad_norm": 0.5283005049642556,
      "learning_rate": 1.1162916107722798e-06,
      "loss": 11.8298,
      "step": 35032
    },
    {
      "epoch": 1.9076868622926493,
      "grad_norm": 0.5101769098963235,
      "learning_rate": 1.1149780682205957e-06,
      "loss": 11.7398,
      "step": 35033
    },
    {
      "epoch": 1.9077413162892323,
      "grad_norm": 0.5498331710435225,
      "learning_rate": 1.113665294615418e-06,
      "loss": 11.8011,
      "step": 35034
    },
    {
      "epoch": 1.9077957702858153,
      "grad_norm": 0.5832457360829273,
      "learning_rate": 1.112353289966961e-06,
      "loss": 11.6953,
      "step": 35035
    },
    {
      "epoch": 1.9078502242823985,
      "grad_norm": 0.5279775767510355,
      "learning_rate": 1.1110420542854384e-06,
      "loss": 11.8075,
      "step": 35036
    },
    {
      "epoch": 1.9079046782789815,
      "grad_norm": 0.5205927459750868,
      "learning_rate": 1.109731587581031e-06,
      "loss": 11.7359,
      "step": 35037
    },
    {
      "epoch": 1.9079591322755645,
      "grad_norm": 0.5881479585714262,
      "learning_rate": 1.108421889863931e-06,
      "loss": 11.7509,
      "step": 35038
    },
    {
      "epoch": 1.9080135862721475,
      "grad_norm": 0.5403880405130227,
      "learning_rate": 1.1071129611443408e-06,
      "loss": 11.8494,
      "step": 35039
    },
    {
      "epoch": 1.9080680402687304,
      "grad_norm": 0.6171662767645459,
      "learning_rate": 1.1058048014324084e-06,
      "loss": 11.9234,
      "step": 35040
    },
    {
      "epoch": 1.9081224942653134,
      "grad_norm": 0.5373698224508806,
      "learning_rate": 1.1044974107383255e-06,
      "loss": 11.6986,
      "step": 35041
    },
    {
      "epoch": 1.9081769482618967,
      "grad_norm": 0.5334942218426715,
      "learning_rate": 1.1031907890722615e-06,
      "loss": 11.7383,
      "step": 35042
    },
    {
      "epoch": 1.9082314022584796,
      "grad_norm": 0.5215693598715035,
      "learning_rate": 1.101884936444364e-06,
      "loss": 11.5175,
      "step": 35043
    },
    {
      "epoch": 1.9082858562550626,
      "grad_norm": 0.5439630012830671,
      "learning_rate": 1.1005798528648025e-06,
      "loss": 11.6718,
      "step": 35044
    },
    {
      "epoch": 1.9083403102516456,
      "grad_norm": 0.5169738020244817,
      "learning_rate": 1.0992755383437137e-06,
      "loss": 11.692,
      "step": 35045
    },
    {
      "epoch": 1.9083947642482286,
      "grad_norm": 0.5267399179343627,
      "learning_rate": 1.0979719928912446e-06,
      "loss": 11.7187,
      "step": 35046
    },
    {
      "epoch": 1.9084492182448116,
      "grad_norm": 0.5240608640848374,
      "learning_rate": 1.0966692165175318e-06,
      "loss": 11.7991,
      "step": 35047
    },
    {
      "epoch": 1.9085036722413946,
      "grad_norm": 0.5479382086282198,
      "learning_rate": 1.0953672092327006e-06,
      "loss": 11.8021,
      "step": 35048
    },
    {
      "epoch": 1.9085581262379776,
      "grad_norm": 0.5635261883703365,
      "learning_rate": 1.094065971046887e-06,
      "loss": 11.6825,
      "step": 35049
    },
    {
      "epoch": 1.9086125802345606,
      "grad_norm": 0.5543134097400486,
      "learning_rate": 1.0927655019701943e-06,
      "loss": 11.9005,
      "step": 35050
    },
    {
      "epoch": 1.9086670342311436,
      "grad_norm": 0.5595174971811848,
      "learning_rate": 1.0914658020127477e-06,
      "loss": 11.8898,
      "step": 35051
    },
    {
      "epoch": 1.9087214882277266,
      "grad_norm": 0.5394529378112851,
      "learning_rate": 1.0901668711846614e-06,
      "loss": 11.7729,
      "step": 35052
    },
    {
      "epoch": 1.9087759422243096,
      "grad_norm": 0.5580037430524944,
      "learning_rate": 1.0888687094960049e-06,
      "loss": 11.7652,
      "step": 35053
    },
    {
      "epoch": 1.9088303962208926,
      "grad_norm": 0.5193653645595908,
      "learning_rate": 1.0875713169569146e-06,
      "loss": 11.7774,
      "step": 35054
    },
    {
      "epoch": 1.9088848502174756,
      "grad_norm": 0.5520991921498845,
      "learning_rate": 1.0862746935774381e-06,
      "loss": 11.6531,
      "step": 35055
    },
    {
      "epoch": 1.9089393042140586,
      "grad_norm": 0.5447856659242378,
      "learning_rate": 1.0849788393676896e-06,
      "loss": 11.8382,
      "step": 35056
    },
    {
      "epoch": 1.9089937582106415,
      "grad_norm": 0.5596622529204539,
      "learning_rate": 1.0836837543377165e-06,
      "loss": 11.8003,
      "step": 35057
    },
    {
      "epoch": 1.9090482122072245,
      "grad_norm": 0.5716859031427447,
      "learning_rate": 1.082389438497622e-06,
      "loss": 11.739,
      "step": 35058
    },
    {
      "epoch": 1.9091026662038075,
      "grad_norm": 0.5529552089456344,
      "learning_rate": 1.0810958918574533e-06,
      "loss": 11.7993,
      "step": 35059
    },
    {
      "epoch": 1.9091571202003907,
      "grad_norm": 0.590923171759506,
      "learning_rate": 1.0798031144272691e-06,
      "loss": 11.7941,
      "step": 35060
    },
    {
      "epoch": 1.9092115741969737,
      "grad_norm": 0.503203522617532,
      "learning_rate": 1.0785111062171282e-06,
      "loss": 11.6206,
      "step": 35061
    },
    {
      "epoch": 1.9092660281935567,
      "grad_norm": 0.5159708977929509,
      "learning_rate": 1.0772198672370782e-06,
      "loss": 11.775,
      "step": 35062
    },
    {
      "epoch": 1.9093204821901397,
      "grad_norm": 0.5641158346192275,
      "learning_rate": 1.0759293974971441e-06,
      "loss": 11.8559,
      "step": 35063
    },
    {
      "epoch": 1.9093749361867227,
      "grad_norm": 0.5603151273798148,
      "learning_rate": 1.0746396970073846e-06,
      "loss": 11.6271,
      "step": 35064
    },
    {
      "epoch": 1.909429390183306,
      "grad_norm": 0.4953463292263194,
      "learning_rate": 1.073350765777803e-06,
      "loss": 11.814,
      "step": 35065
    },
    {
      "epoch": 1.909483844179889,
      "grad_norm": 0.5693987504248924,
      "learning_rate": 1.0720626038184467e-06,
      "loss": 11.8154,
      "step": 35066
    },
    {
      "epoch": 1.909538298176472,
      "grad_norm": 0.49598276689437376,
      "learning_rate": 1.0707752111393189e-06,
      "loss": 11.6372,
      "step": 35067
    },
    {
      "epoch": 1.909592752173055,
      "grad_norm": 0.5512118148892549,
      "learning_rate": 1.0694885877504334e-06,
      "loss": 11.9031,
      "step": 35068
    },
    {
      "epoch": 1.909647206169638,
      "grad_norm": 0.5783836042091899,
      "learning_rate": 1.0682027336617939e-06,
      "loss": 11.8721,
      "step": 35069
    },
    {
      "epoch": 1.909701660166221,
      "grad_norm": 0.5378272435245747,
      "learning_rate": 1.0669176488834032e-06,
      "loss": 11.5571,
      "step": 35070
    },
    {
      "epoch": 1.9097561141628039,
      "grad_norm": 0.5237454585595565,
      "learning_rate": 1.0656333334252532e-06,
      "loss": 11.6939,
      "step": 35071
    },
    {
      "epoch": 1.9098105681593869,
      "grad_norm": 0.5758661119362838,
      "learning_rate": 1.064349787297325e-06,
      "loss": 11.7597,
      "step": 35072
    },
    {
      "epoch": 1.9098650221559699,
      "grad_norm": 0.5862225418671749,
      "learning_rate": 1.0630670105096108e-06,
      "loss": 11.7829,
      "step": 35073
    },
    {
      "epoch": 1.9099194761525529,
      "grad_norm": 0.6267844827987517,
      "learning_rate": 1.0617850030720688e-06,
      "loss": 11.8347,
      "step": 35074
    },
    {
      "epoch": 1.9099739301491359,
      "grad_norm": 0.5373539663053456,
      "learning_rate": 1.0605037649946915e-06,
      "loss": 11.762,
      "step": 35075
    },
    {
      "epoch": 1.9100283841457188,
      "grad_norm": 0.561820988168871,
      "learning_rate": 1.059223296287415e-06,
      "loss": 11.7992,
      "step": 35076
    },
    {
      "epoch": 1.9100828381423018,
      "grad_norm": 0.5095559453567118,
      "learning_rate": 1.0579435969602203e-06,
      "loss": 11.7944,
      "step": 35077
    },
    {
      "epoch": 1.9101372921388848,
      "grad_norm": 0.4888594334549358,
      "learning_rate": 1.056664667023044e-06,
      "loss": 11.8618,
      "step": 35078
    },
    {
      "epoch": 1.9101917461354678,
      "grad_norm": 0.5353049185426352,
      "learning_rate": 1.0553865064858448e-06,
      "loss": 11.7997,
      "step": 35079
    },
    {
      "epoch": 1.9102462001320508,
      "grad_norm": 0.5223524984841007,
      "learning_rate": 1.0541091153585481e-06,
      "loss": 11.6515,
      "step": 35080
    },
    {
      "epoch": 1.9103006541286338,
      "grad_norm": 0.5663943238723873,
      "learning_rate": 1.0528324936510902e-06,
      "loss": 11.7447,
      "step": 35081
    },
    {
      "epoch": 1.9103551081252168,
      "grad_norm": 0.5433977328800638,
      "learning_rate": 1.0515566413733969e-06,
      "loss": 11.8036,
      "step": 35082
    },
    {
      "epoch": 1.9104095621218,
      "grad_norm": 0.5724687132210393,
      "learning_rate": 1.0502815585354042e-06,
      "loss": 11.7263,
      "step": 35083
    },
    {
      "epoch": 1.910464016118383,
      "grad_norm": 0.5410814429033893,
      "learning_rate": 1.0490072451470046e-06,
      "loss": 11.8019,
      "step": 35084
    },
    {
      "epoch": 1.910518470114966,
      "grad_norm": 0.5795403204573999,
      "learning_rate": 1.0477337012181232e-06,
      "loss": 11.864,
      "step": 35085
    },
    {
      "epoch": 1.910572924111549,
      "grad_norm": 0.5843096261051132,
      "learning_rate": 1.0464609267586522e-06,
      "loss": 11.8717,
      "step": 35086
    },
    {
      "epoch": 1.910627378108132,
      "grad_norm": 0.5547838288076281,
      "learning_rate": 1.045188921778506e-06,
      "loss": 11.7711,
      "step": 35087
    },
    {
      "epoch": 1.910681832104715,
      "grad_norm": 0.5089451102713335,
      "learning_rate": 1.0439176862875654e-06,
      "loss": 11.6835,
      "step": 35088
    },
    {
      "epoch": 1.9107362861012982,
      "grad_norm": 0.6201359920253151,
      "learning_rate": 1.0426472202957004e-06,
      "loss": 11.6396,
      "step": 35089
    },
    {
      "epoch": 1.9107907400978812,
      "grad_norm": 0.5881591957683492,
      "learning_rate": 1.0413775238128253e-06,
      "loss": 11.8939,
      "step": 35090
    },
    {
      "epoch": 1.9108451940944642,
      "grad_norm": 0.5522354787812104,
      "learning_rate": 1.0401085968487766e-06,
      "loss": 11.7936,
      "step": 35091
    },
    {
      "epoch": 1.9108996480910472,
      "grad_norm": 0.5235783192898293,
      "learning_rate": 1.0388404394134577e-06,
      "loss": 11.7094,
      "step": 35092
    },
    {
      "epoch": 1.9109541020876302,
      "grad_norm": 0.5104367125330096,
      "learning_rate": 1.0375730515167047e-06,
      "loss": 11.7649,
      "step": 35093
    },
    {
      "epoch": 1.9110085560842132,
      "grad_norm": 0.6352311273546106,
      "learning_rate": 1.0363064331683769e-06,
      "loss": 11.9029,
      "step": 35094
    },
    {
      "epoch": 1.9110630100807962,
      "grad_norm": 0.5808278235580587,
      "learning_rate": 1.0350405843783216e-06,
      "loss": 11.7317,
      "step": 35095
    },
    {
      "epoch": 1.9111174640773791,
      "grad_norm": 0.5439018501035373,
      "learning_rate": 1.0337755051563868e-06,
      "loss": 11.7704,
      "step": 35096
    },
    {
      "epoch": 1.9111719180739621,
      "grad_norm": 0.5535290140400511,
      "learning_rate": 1.0325111955124201e-06,
      "loss": 11.7334,
      "step": 35097
    },
    {
      "epoch": 1.9112263720705451,
      "grad_norm": 0.5355060497331602,
      "learning_rate": 1.0312476554562356e-06,
      "loss": 11.7896,
      "step": 35098
    },
    {
      "epoch": 1.9112808260671281,
      "grad_norm": 0.5190122938502716,
      "learning_rate": 1.02998488499767e-06,
      "loss": 11.8073,
      "step": 35099
    },
    {
      "epoch": 1.9113352800637111,
      "grad_norm": 0.5204723389919724,
      "learning_rate": 1.028722884146538e-06,
      "loss": 11.7878,
      "step": 35100
    },
    {
      "epoch": 1.911389734060294,
      "grad_norm": 0.5564546198705889,
      "learning_rate": 1.0274616529126536e-06,
      "loss": 11.8006,
      "step": 35101
    },
    {
      "epoch": 1.911444188056877,
      "grad_norm": 0.48794837984224265,
      "learning_rate": 1.0262011913058312e-06,
      "loss": 11.6623,
      "step": 35102
    },
    {
      "epoch": 1.91149864205346,
      "grad_norm": 0.5390515935615477,
      "learning_rate": 1.024941499335863e-06,
      "loss": 11.828,
      "step": 35103
    },
    {
      "epoch": 1.911553096050043,
      "grad_norm": 0.5235486524429563,
      "learning_rate": 1.0236825770125414e-06,
      "loss": 11.7012,
      "step": 35104
    },
    {
      "epoch": 1.911607550046626,
      "grad_norm": 0.6402393695134693,
      "learning_rate": 1.0224244243456693e-06,
      "loss": 11.7515,
      "step": 35105
    },
    {
      "epoch": 1.9116620040432093,
      "grad_norm": 0.515391054436438,
      "learning_rate": 1.0211670413450169e-06,
      "loss": 11.8416,
      "step": 35106
    },
    {
      "epoch": 1.9117164580397923,
      "grad_norm": 0.49968832184637235,
      "learning_rate": 1.0199104280203763e-06,
      "loss": 11.768,
      "step": 35107
    },
    {
      "epoch": 1.9117709120363753,
      "grad_norm": 0.6274243618865167,
      "learning_rate": 1.0186545843815065e-06,
      "loss": 11.7803,
      "step": 35108
    },
    {
      "epoch": 1.9118253660329583,
      "grad_norm": 0.5097495364866304,
      "learning_rate": 1.0173995104381773e-06,
      "loss": 11.779,
      "step": 35109
    },
    {
      "epoch": 1.9118798200295413,
      "grad_norm": 0.5261417959977284,
      "learning_rate": 1.0161452062001587e-06,
      "loss": 11.8723,
      "step": 35110
    },
    {
      "epoch": 1.9119342740261243,
      "grad_norm": 0.5459633910347668,
      "learning_rate": 1.0148916716771761e-06,
      "loss": 11.8145,
      "step": 35111
    },
    {
      "epoch": 1.9119887280227075,
      "grad_norm": 0.5602467867364136,
      "learning_rate": 1.0136389068790108e-06,
      "loss": 11.779,
      "step": 35112
    },
    {
      "epoch": 1.9120431820192905,
      "grad_norm": 0.5941671271494218,
      "learning_rate": 1.0123869118153883e-06,
      "loss": 11.6405,
      "step": 35113
    },
    {
      "epoch": 1.9120976360158735,
      "grad_norm": 0.5483298253064384,
      "learning_rate": 1.0111356864960341e-06,
      "loss": 11.6138,
      "step": 35114
    },
    {
      "epoch": 1.9121520900124565,
      "grad_norm": 0.5154942674498695,
      "learning_rate": 1.0098852309307072e-06,
      "loss": 11.8123,
      "step": 35115
    },
    {
      "epoch": 1.9122065440090394,
      "grad_norm": 0.5525543639258955,
      "learning_rate": 1.0086355451290997e-06,
      "loss": 11.8114,
      "step": 35116
    },
    {
      "epoch": 1.9122609980056224,
      "grad_norm": 0.5227600270568211,
      "learning_rate": 1.007386629100948e-06,
      "loss": 11.7644,
      "step": 35117
    },
    {
      "epoch": 1.9123154520022054,
      "grad_norm": 0.5627995273649774,
      "learning_rate": 1.006138482855956e-06,
      "loss": 11.6594,
      "step": 35118
    },
    {
      "epoch": 1.9123699059987884,
      "grad_norm": 0.5338480996883471,
      "learning_rate": 1.0048911064038268e-06,
      "loss": 11.7901,
      "step": 35119
    },
    {
      "epoch": 1.9124243599953714,
      "grad_norm": 0.5442165476819657,
      "learning_rate": 1.003644499754275e-06,
      "loss": 11.6895,
      "step": 35120
    },
    {
      "epoch": 1.9124788139919544,
      "grad_norm": 0.5476404177077041,
      "learning_rate": 1.0023986629169813e-06,
      "loss": 11.7744,
      "step": 35121
    },
    {
      "epoch": 1.9125332679885374,
      "grad_norm": 0.5437739273626775,
      "learning_rate": 1.0011535959016494e-06,
      "loss": 11.7232,
      "step": 35122
    },
    {
      "epoch": 1.9125877219851204,
      "grad_norm": 0.5599405951560065,
      "learning_rate": 9.999092987179381e-07,
      "loss": 11.7276,
      "step": 35123
    },
    {
      "epoch": 1.9126421759817034,
      "grad_norm": 0.49533066232163964,
      "learning_rate": 9.986657713755288e-07,
      "loss": 11.7487,
      "step": 35124
    },
    {
      "epoch": 1.9126966299782864,
      "grad_norm": 0.5326189323704272,
      "learning_rate": 9.974230138841024e-07,
      "loss": 11.7116,
      "step": 35125
    },
    {
      "epoch": 1.9127510839748694,
      "grad_norm": 0.5294952348997704,
      "learning_rate": 9.961810262533178e-07,
      "loss": 11.8349,
      "step": 35126
    },
    {
      "epoch": 1.9128055379714524,
      "grad_norm": 0.6010079103320684,
      "learning_rate": 9.94939808492834e-07,
      "loss": 11.8098,
      "step": 35127
    },
    {
      "epoch": 1.9128599919680354,
      "grad_norm": 0.5693917939451096,
      "learning_rate": 9.93699360612299e-07,
      "loss": 11.6701,
      "step": 35128
    },
    {
      "epoch": 1.9129144459646186,
      "grad_norm": 0.4904476688547059,
      "learning_rate": 9.924596826213716e-07,
      "loss": 11.6774,
      "step": 35129
    },
    {
      "epoch": 1.9129688999612016,
      "grad_norm": 0.4995587318185474,
      "learning_rate": 9.912207745296665e-07,
      "loss": 11.7444,
      "step": 35130
    },
    {
      "epoch": 1.9130233539577846,
      "grad_norm": 0.5644042562087479,
      "learning_rate": 9.899826363468311e-07,
      "loss": 11.774,
      "step": 35131
    },
    {
      "epoch": 1.9130778079543675,
      "grad_norm": 0.6478245834824435,
      "learning_rate": 9.887452680825138e-07,
      "loss": 11.7276,
      "step": 35132
    },
    {
      "epoch": 1.9131322619509505,
      "grad_norm": 0.5025718304468798,
      "learning_rate": 9.875086697463066e-07,
      "loss": 11.7485,
      "step": 35133
    },
    {
      "epoch": 1.9131867159475335,
      "grad_norm": 0.5132824629887921,
      "learning_rate": 9.86272841347835e-07,
      "loss": 11.5666,
      "step": 35134
    },
    {
      "epoch": 1.9132411699441167,
      "grad_norm": 0.5160447393687133,
      "learning_rate": 9.85037782896714e-07,
      "loss": 11.8409,
      "step": 35135
    },
    {
      "epoch": 1.9132956239406997,
      "grad_norm": 0.5831760641182391,
      "learning_rate": 9.838034944025354e-07,
      "loss": 11.7491,
      "step": 35136
    },
    {
      "epoch": 1.9133500779372827,
      "grad_norm": 0.5000670887401855,
      "learning_rate": 9.82569975874914e-07,
      "loss": 11.6315,
      "step": 35137
    },
    {
      "epoch": 1.9134045319338657,
      "grad_norm": 0.5364252803555681,
      "learning_rate": 9.813372273234311e-07,
      "loss": 11.6952,
      "step": 35138
    },
    {
      "epoch": 1.9134589859304487,
      "grad_norm": 0.5859474620258575,
      "learning_rate": 9.80105248757679e-07,
      "loss": 11.773,
      "step": 35139
    },
    {
      "epoch": 1.9135134399270317,
      "grad_norm": 0.5183197641860716,
      "learning_rate": 9.788740401872277e-07,
      "loss": 11.8488,
      "step": 35140
    },
    {
      "epoch": 1.9135678939236147,
      "grad_norm": 0.5311511630162272,
      "learning_rate": 9.776436016216473e-07,
      "loss": 11.885,
      "step": 35141
    },
    {
      "epoch": 1.9136223479201977,
      "grad_norm": 0.5721275190221004,
      "learning_rate": 9.764139330705412e-07,
      "loss": 11.8198,
      "step": 35142
    },
    {
      "epoch": 1.9136768019167807,
      "grad_norm": 0.5439242256473281,
      "learning_rate": 9.751850345434355e-07,
      "loss": 11.7999,
      "step": 35143
    },
    {
      "epoch": 1.9137312559133637,
      "grad_norm": 0.5025895639682192,
      "learning_rate": 9.739569060498998e-07,
      "loss": 11.6802,
      "step": 35144
    },
    {
      "epoch": 1.9137857099099467,
      "grad_norm": 0.522579173012891,
      "learning_rate": 9.727295475994714e-07,
      "loss": 11.7436,
      "step": 35145
    },
    {
      "epoch": 1.9138401639065297,
      "grad_norm": 0.5540909247206703,
      "learning_rate": 9.715029592017088e-07,
      "loss": 11.8201,
      "step": 35146
    },
    {
      "epoch": 1.9138946179031127,
      "grad_norm": 0.5299964103986897,
      "learning_rate": 9.702771408661604e-07,
      "loss": 11.8015,
      "step": 35147
    },
    {
      "epoch": 1.9139490718996957,
      "grad_norm": 0.5460718503237835,
      "learning_rate": 9.690520926023294e-07,
      "loss": 11.7618,
      "step": 35148
    },
    {
      "epoch": 1.9140035258962786,
      "grad_norm": 0.5642893204770933,
      "learning_rate": 9.678278144197639e-07,
      "loss": 11.7825,
      "step": 35149
    },
    {
      "epoch": 1.9140579798928616,
      "grad_norm": 0.5199466557557606,
      "learning_rate": 9.666043063279673e-07,
      "loss": 11.8994,
      "step": 35150
    },
    {
      "epoch": 1.9141124338894446,
      "grad_norm": 0.5713658464418989,
      "learning_rate": 9.653815683364764e-07,
      "loss": 11.8829,
      "step": 35151
    },
    {
      "epoch": 1.9141668878860276,
      "grad_norm": 0.5318327808591264,
      "learning_rate": 9.641596004547726e-07,
      "loss": 11.7626,
      "step": 35152
    },
    {
      "epoch": 1.9142213418826108,
      "grad_norm": 0.5282506846256024,
      "learning_rate": 9.629384026923816e-07,
      "loss": 11.8018,
      "step": 35153
    },
    {
      "epoch": 1.9142757958791938,
      "grad_norm": 0.5720525814173348,
      "learning_rate": 9.617179750587957e-07,
      "loss": 11.8022,
      "step": 35154
    },
    {
      "epoch": 1.9143302498757768,
      "grad_norm": 0.5258208866729774,
      "learning_rate": 9.604983175634852e-07,
      "loss": 11.9336,
      "step": 35155
    },
    {
      "epoch": 1.9143847038723598,
      "grad_norm": 0.5159547586650071,
      "learning_rate": 9.592794302159646e-07,
      "loss": 11.7323,
      "step": 35156
    },
    {
      "epoch": 1.9144391578689428,
      "grad_norm": 0.5610796327787909,
      "learning_rate": 9.580613130256822e-07,
      "loss": 11.8646,
      "step": 35157
    },
    {
      "epoch": 1.9144936118655258,
      "grad_norm": 0.5767096618913562,
      "learning_rate": 9.5684396600213e-07,
      "loss": 11.8894,
      "step": 35158
    },
    {
      "epoch": 1.914548065862109,
      "grad_norm": 0.5294956672415153,
      "learning_rate": 9.556273891547673e-07,
      "loss": 11.7973,
      "step": 35159
    },
    {
      "epoch": 1.914602519858692,
      "grad_norm": 0.5391260841721528,
      "learning_rate": 9.544115824930532e-07,
      "loss": 11.7829,
      "step": 35160
    },
    {
      "epoch": 1.914656973855275,
      "grad_norm": 0.5405415191523248,
      "learning_rate": 9.53196546026458e-07,
      "loss": 11.7448,
      "step": 35161
    },
    {
      "epoch": 1.914711427851858,
      "grad_norm": 0.5841377470590796,
      "learning_rate": 9.519822797643962e-07,
      "loss": 11.8143,
      "step": 35162
    },
    {
      "epoch": 1.914765881848441,
      "grad_norm": 0.6184059434122927,
      "learning_rate": 9.507687837163492e-07,
      "loss": 11.729,
      "step": 35163
    },
    {
      "epoch": 1.914820335845024,
      "grad_norm": 0.5255925439390693,
      "learning_rate": 9.495560578917318e-07,
      "loss": 11.7779,
      "step": 35164
    },
    {
      "epoch": 1.914874789841607,
      "grad_norm": 0.5253838804623853,
      "learning_rate": 9.483441022999695e-07,
      "loss": 11.7927,
      "step": 35165
    },
    {
      "epoch": 1.91492924383819,
      "grad_norm": 0.5980539158184621,
      "learning_rate": 9.471329169504995e-07,
      "loss": 11.9218,
      "step": 35166
    },
    {
      "epoch": 1.914983697834773,
      "grad_norm": 0.5617716604024158,
      "learning_rate": 9.459225018527251e-07,
      "loss": 11.8383,
      "step": 35167
    },
    {
      "epoch": 1.915038151831356,
      "grad_norm": 0.5623123857783747,
      "learning_rate": 9.447128570160723e-07,
      "loss": 11.6922,
      "step": 35168
    },
    {
      "epoch": 1.915092605827939,
      "grad_norm": 0.5506840989740919,
      "learning_rate": 9.435039824499559e-07,
      "loss": 11.8841,
      "step": 35169
    },
    {
      "epoch": 1.915147059824522,
      "grad_norm": 0.53043677168556,
      "learning_rate": 9.422958781637569e-07,
      "loss": 11.6814,
      "step": 35170
    },
    {
      "epoch": 1.915201513821105,
      "grad_norm": 0.5097510373721862,
      "learning_rate": 9.410885441668793e-07,
      "loss": 11.7102,
      "step": 35171
    },
    {
      "epoch": 1.915255967817688,
      "grad_norm": 0.5199440180916941,
      "learning_rate": 9.398819804687043e-07,
      "loss": 11.7086,
      "step": 35172
    },
    {
      "epoch": 1.915310421814271,
      "grad_norm": 0.5762046373965938,
      "learning_rate": 9.386761870786243e-07,
      "loss": 11.9591,
      "step": 35173
    },
    {
      "epoch": 1.915364875810854,
      "grad_norm": 0.5586414190330928,
      "learning_rate": 9.374711640060096e-07,
      "loss": 11.8247,
      "step": 35174
    },
    {
      "epoch": 1.915419329807437,
      "grad_norm": 0.5381081233376709,
      "learning_rate": 9.362669112602307e-07,
      "loss": 11.708,
      "step": 35175
    },
    {
      "epoch": 1.9154737838040201,
      "grad_norm": 0.559612607472234,
      "learning_rate": 9.350634288506466e-07,
      "loss": 11.7455,
      "step": 35176
    },
    {
      "epoch": 1.915528237800603,
      "grad_norm": 0.5597247638643995,
      "learning_rate": 9.338607167866276e-07,
      "loss": 11.7732,
      "step": 35177
    },
    {
      "epoch": 1.915582691797186,
      "grad_norm": 0.5832729433958163,
      "learning_rate": 9.326587750775329e-07,
      "loss": 11.8304,
      "step": 35178
    },
    {
      "epoch": 1.915637145793769,
      "grad_norm": 0.5401258334400405,
      "learning_rate": 9.314576037326772e-07,
      "loss": 11.7651,
      "step": 35179
    },
    {
      "epoch": 1.915691599790352,
      "grad_norm": 0.546976026951646,
      "learning_rate": 9.302572027614309e-07,
      "loss": 11.8512,
      "step": 35180
    },
    {
      "epoch": 1.915746053786935,
      "grad_norm": 0.5205921963085846,
      "learning_rate": 9.290575721731198e-07,
      "loss": 11.7567,
      "step": 35181
    },
    {
      "epoch": 1.9158005077835183,
      "grad_norm": 0.5490305793584135,
      "learning_rate": 9.278587119770698e-07,
      "loss": 11.7593,
      "step": 35182
    },
    {
      "epoch": 1.9158549617801013,
      "grad_norm": 0.5386055788617824,
      "learning_rate": 9.266606221826069e-07,
      "loss": 11.6951,
      "step": 35183
    },
    {
      "epoch": 1.9159094157766843,
      "grad_norm": 0.4888484154079049,
      "learning_rate": 9.254633027990456e-07,
      "loss": 11.7217,
      "step": 35184
    },
    {
      "epoch": 1.9159638697732673,
      "grad_norm": 0.5561620307709734,
      "learning_rate": 9.242667538356898e-07,
      "loss": 11.8107,
      "step": 35185
    },
    {
      "epoch": 1.9160183237698503,
      "grad_norm": 0.533857419028927,
      "learning_rate": 9.230709753018651e-07,
      "loss": 11.804,
      "step": 35186
    },
    {
      "epoch": 1.9160727777664333,
      "grad_norm": 0.5740980065069025,
      "learning_rate": 9.218759672068422e-07,
      "loss": 11.7448,
      "step": 35187
    },
    {
      "epoch": 1.9161272317630162,
      "grad_norm": 0.5348810160177284,
      "learning_rate": 9.206817295599246e-07,
      "loss": 11.7366,
      "step": 35188
    },
    {
      "epoch": 1.9161816857595992,
      "grad_norm": 0.5639451320436223,
      "learning_rate": 9.194882623704049e-07,
      "loss": 11.7164,
      "step": 35189
    },
    {
      "epoch": 1.9162361397561822,
      "grad_norm": 0.49415545276392187,
      "learning_rate": 9.182955656475645e-07,
      "loss": 11.529,
      "step": 35190
    },
    {
      "epoch": 1.9162905937527652,
      "grad_norm": 0.557829386350709,
      "learning_rate": 9.171036394006737e-07,
      "loss": 11.7855,
      "step": 35191
    },
    {
      "epoch": 1.9163450477493482,
      "grad_norm": 0.5947507546188459,
      "learning_rate": 9.159124836390032e-07,
      "loss": 11.849,
      "step": 35192
    },
    {
      "epoch": 1.9163995017459312,
      "grad_norm": 0.5565434666075539,
      "learning_rate": 9.147220983718119e-07,
      "loss": 11.7487,
      "step": 35193
    },
    {
      "epoch": 1.9164539557425142,
      "grad_norm": 0.5687280597467993,
      "learning_rate": 9.135324836083592e-07,
      "loss": 11.7756,
      "step": 35194
    },
    {
      "epoch": 1.9165084097390972,
      "grad_norm": 0.5369250923409661,
      "learning_rate": 9.123436393578822e-07,
      "loss": 11.7732,
      "step": 35195
    },
    {
      "epoch": 1.9165628637356802,
      "grad_norm": 0.6092539975696989,
      "learning_rate": 9.111555656296511e-07,
      "loss": 11.692,
      "step": 35196
    },
    {
      "epoch": 1.9166173177322632,
      "grad_norm": 0.5167452649292813,
      "learning_rate": 9.09968262432892e-07,
      "loss": 11.7265,
      "step": 35197
    },
    {
      "epoch": 1.9166717717288462,
      "grad_norm": 0.5385597377763062,
      "learning_rate": 9.087817297768308e-07,
      "loss": 11.833,
      "step": 35198
    },
    {
      "epoch": 1.9167262257254294,
      "grad_norm": 0.5416782185874432,
      "learning_rate": 9.075959676707046e-07,
      "loss": 11.6064,
      "step": 35199
    },
    {
      "epoch": 1.9167806797220124,
      "grad_norm": 0.537139385306688,
      "learning_rate": 9.064109761237282e-07,
      "loss": 11.681,
      "step": 35200
    },
    {
      "epoch": 1.9168351337185954,
      "grad_norm": 0.6357180096696138,
      "learning_rate": 9.052267551451165e-07,
      "loss": 11.808,
      "step": 35201
    },
    {
      "epoch": 1.9168895877151784,
      "grad_norm": 0.6055184640562326,
      "learning_rate": 9.040433047440844e-07,
      "loss": 11.788,
      "step": 35202
    },
    {
      "epoch": 1.9169440417117614,
      "grad_norm": 0.5045144401513638,
      "learning_rate": 9.028606249298355e-07,
      "loss": 11.7997,
      "step": 35203
    },
    {
      "epoch": 1.9169984957083444,
      "grad_norm": 0.5459174889517361,
      "learning_rate": 9.016787157115514e-07,
      "loss": 11.8305,
      "step": 35204
    },
    {
      "epoch": 1.9170529497049276,
      "grad_norm": 0.4910436462510843,
      "learning_rate": 9.004975770984358e-07,
      "loss": 11.7572,
      "step": 35205
    },
    {
      "epoch": 1.9171074037015106,
      "grad_norm": 0.5478522802477646,
      "learning_rate": 8.993172090996593e-07,
      "loss": 11.8115,
      "step": 35206
    },
    {
      "epoch": 1.9171618576980936,
      "grad_norm": 0.5559894155851322,
      "learning_rate": 8.981376117244255e-07,
      "loss": 11.8449,
      "step": 35207
    },
    {
      "epoch": 1.9172163116946765,
      "grad_norm": 0.538612690716442,
      "learning_rate": 8.969587849818828e-07,
      "loss": 11.7906,
      "step": 35208
    },
    {
      "epoch": 1.9172707656912595,
      "grad_norm": 0.5084388156015692,
      "learning_rate": 8.957807288812126e-07,
      "loss": 11.7268,
      "step": 35209
    },
    {
      "epoch": 1.9173252196878425,
      "grad_norm": 0.4832681966615627,
      "learning_rate": 8.946034434315742e-07,
      "loss": 11.7395,
      "step": 35210
    },
    {
      "epoch": 1.9173796736844255,
      "grad_norm": 0.5738225621440586,
      "learning_rate": 8.934269286421159e-07,
      "loss": 11.8553,
      "step": 35211
    },
    {
      "epoch": 1.9174341276810085,
      "grad_norm": 0.48711473108013803,
      "learning_rate": 8.922511845219971e-07,
      "loss": 11.7085,
      "step": 35212
    },
    {
      "epoch": 1.9174885816775915,
      "grad_norm": 0.5202528831488631,
      "learning_rate": 8.910762110803439e-07,
      "loss": 11.8102,
      "step": 35213
    },
    {
      "epoch": 1.9175430356741745,
      "grad_norm": 0.519394116891388,
      "learning_rate": 8.899020083263043e-07,
      "loss": 11.7792,
      "step": 35214
    },
    {
      "epoch": 1.9175974896707575,
      "grad_norm": 0.519271590028062,
      "learning_rate": 8.887285762690156e-07,
      "loss": 11.8252,
      "step": 35215
    },
    {
      "epoch": 1.9176519436673405,
      "grad_norm": 0.5455615839571457,
      "learning_rate": 8.875559149175816e-07,
      "loss": 11.8203,
      "step": 35216
    },
    {
      "epoch": 1.9177063976639235,
      "grad_norm": 0.5390491771381011,
      "learning_rate": 8.863840242811394e-07,
      "loss": 11.744,
      "step": 35217
    },
    {
      "epoch": 1.9177608516605065,
      "grad_norm": 0.5617895119488668,
      "learning_rate": 8.85212904368804e-07,
      "loss": 11.7927,
      "step": 35218
    },
    {
      "epoch": 1.9178153056570895,
      "grad_norm": 0.49923998977061224,
      "learning_rate": 8.84042555189657e-07,
      "loss": 11.7869,
      "step": 35219
    },
    {
      "epoch": 1.9178697596536725,
      "grad_norm": 0.5392256851222319,
      "learning_rate": 8.828729767528354e-07,
      "loss": 11.7381,
      "step": 35220
    },
    {
      "epoch": 1.9179242136502554,
      "grad_norm": 0.5580498866571791,
      "learning_rate": 8.817041690673989e-07,
      "loss": 11.7798,
      "step": 35221
    },
    {
      "epoch": 1.9179786676468384,
      "grad_norm": 0.5553461907649563,
      "learning_rate": 8.805361321424732e-07,
      "loss": 11.8,
      "step": 35222
    },
    {
      "epoch": 1.9180331216434217,
      "grad_norm": 0.538107620107467,
      "learning_rate": 8.793688659871069e-07,
      "loss": 11.6728,
      "step": 35223
    },
    {
      "epoch": 1.9180875756400046,
      "grad_norm": 0.5358888643911297,
      "learning_rate": 8.782023706103925e-07,
      "loss": 11.8671,
      "step": 35224
    },
    {
      "epoch": 1.9181420296365876,
      "grad_norm": 0.5473243382040065,
      "learning_rate": 8.770366460214008e-07,
      "loss": 11.7563,
      "step": 35225
    },
    {
      "epoch": 1.9181964836331706,
      "grad_norm": 0.5619020179221044,
      "learning_rate": 8.75871692229191e-07,
      "loss": 11.8151,
      "step": 35226
    },
    {
      "epoch": 1.9182509376297536,
      "grad_norm": 0.646298529253663,
      "learning_rate": 8.747075092428336e-07,
      "loss": 11.7864,
      "step": 35227
    },
    {
      "epoch": 1.9183053916263366,
      "grad_norm": 0.5448105127013941,
      "learning_rate": 8.73544097071366e-07,
      "loss": 11.6923,
      "step": 35228
    },
    {
      "epoch": 1.9183598456229198,
      "grad_norm": 0.5097718353891445,
      "learning_rate": 8.723814557238474e-07,
      "loss": 11.7472,
      "step": 35229
    },
    {
      "epoch": 1.9184142996195028,
      "grad_norm": 0.6254578642322416,
      "learning_rate": 8.712195852093152e-07,
      "loss": 11.7263,
      "step": 35230
    },
    {
      "epoch": 1.9184687536160858,
      "grad_norm": 0.5030137023454851,
      "learning_rate": 8.700584855367955e-07,
      "loss": 11.7914,
      "step": 35231
    },
    {
      "epoch": 1.9185232076126688,
      "grad_norm": 0.524172881676483,
      "learning_rate": 8.688981567153365e-07,
      "loss": 11.7223,
      "step": 35232
    },
    {
      "epoch": 1.9185776616092518,
      "grad_norm": 0.5398621754495531,
      "learning_rate": 8.677385987539532e-07,
      "loss": 11.9072,
      "step": 35233
    },
    {
      "epoch": 1.9186321156058348,
      "grad_norm": 0.5669199821899846,
      "learning_rate": 8.665798116616496e-07,
      "loss": 11.6823,
      "step": 35234
    },
    {
      "epoch": 1.9186865696024178,
      "grad_norm": 0.558354801015339,
      "learning_rate": 8.654217954474408e-07,
      "loss": 11.7699,
      "step": 35235
    },
    {
      "epoch": 1.9187410235990008,
      "grad_norm": 0.5154310645832462,
      "learning_rate": 8.642645501203416e-07,
      "loss": 11.7,
      "step": 35236
    },
    {
      "epoch": 1.9187954775955838,
      "grad_norm": 0.5253994038750188,
      "learning_rate": 8.631080756893562e-07,
      "loss": 11.734,
      "step": 35237
    },
    {
      "epoch": 1.9188499315921668,
      "grad_norm": 0.582087935449611,
      "learning_rate": 8.619523721634548e-07,
      "loss": 11.8067,
      "step": 35238
    },
    {
      "epoch": 1.9189043855887498,
      "grad_norm": 0.5274111672244881,
      "learning_rate": 8.607974395516416e-07,
      "loss": 11.7308,
      "step": 35239
    },
    {
      "epoch": 1.9189588395853328,
      "grad_norm": 0.5722438827303705,
      "learning_rate": 8.596432778628983e-07,
      "loss": 11.6561,
      "step": 35240
    },
    {
      "epoch": 1.9190132935819157,
      "grad_norm": 0.5250881599929415,
      "learning_rate": 8.584898871061841e-07,
      "loss": 11.7011,
      "step": 35241
    },
    {
      "epoch": 1.9190677475784987,
      "grad_norm": 0.6673704896322545,
      "learning_rate": 8.573372672904923e-07,
      "loss": 11.8617,
      "step": 35242
    },
    {
      "epoch": 1.9191222015750817,
      "grad_norm": 0.5550575162151639,
      "learning_rate": 8.561854184247597e-07,
      "loss": 11.6047,
      "step": 35243
    },
    {
      "epoch": 1.9191766555716647,
      "grad_norm": 0.609772576805787,
      "learning_rate": 8.550343405179573e-07,
      "loss": 11.811,
      "step": 35244
    },
    {
      "epoch": 1.9192311095682477,
      "grad_norm": 0.49943831048443377,
      "learning_rate": 8.538840335790443e-07,
      "loss": 11.7641,
      "step": 35245
    },
    {
      "epoch": 1.919285563564831,
      "grad_norm": 0.5877683080457907,
      "learning_rate": 8.527344976169471e-07,
      "loss": 11.7762,
      "step": 35246
    },
    {
      "epoch": 1.919340017561414,
      "grad_norm": 0.6439500745850081,
      "learning_rate": 8.51585732640614e-07,
      "loss": 11.7894,
      "step": 35247
    },
    {
      "epoch": 1.919394471557997,
      "grad_norm": 0.537615234870368,
      "learning_rate": 8.504377386589824e-07,
      "loss": 11.7894,
      "step": 35248
    },
    {
      "epoch": 1.91944892555458,
      "grad_norm": 0.5793091396641953,
      "learning_rate": 8.492905156809671e-07,
      "loss": 11.8368,
      "step": 35249
    },
    {
      "epoch": 1.919503379551163,
      "grad_norm": 0.5705253878050031,
      "learning_rate": 8.481440637154947e-07,
      "loss": 11.8582,
      "step": 35250
    },
    {
      "epoch": 1.919557833547746,
      "grad_norm": 0.4914867197880977,
      "learning_rate": 8.469983827714911e-07,
      "loss": 11.6931,
      "step": 35251
    },
    {
      "epoch": 1.919612287544329,
      "grad_norm": 0.529209964279529,
      "learning_rate": 8.458534728578494e-07,
      "loss": 11.6899,
      "step": 35252
    },
    {
      "epoch": 1.919666741540912,
      "grad_norm": 0.5437293401220309,
      "learning_rate": 8.447093339834844e-07,
      "loss": 11.7877,
      "step": 35253
    },
    {
      "epoch": 1.919721195537495,
      "grad_norm": 0.5633118117687254,
      "learning_rate": 8.435659661572892e-07,
      "loss": 11.7408,
      "step": 35254
    },
    {
      "epoch": 1.919775649534078,
      "grad_norm": 0.5199596714588441,
      "learning_rate": 8.424233693881456e-07,
      "loss": 11.7181,
      "step": 35255
    },
    {
      "epoch": 1.919830103530661,
      "grad_norm": 0.4993925874144359,
      "learning_rate": 8.412815436849575e-07,
      "loss": 11.7928,
      "step": 35256
    },
    {
      "epoch": 1.919884557527244,
      "grad_norm": 0.4955370450074922,
      "learning_rate": 8.401404890565845e-07,
      "loss": 11.5063,
      "step": 35257
    },
    {
      "epoch": 1.919939011523827,
      "grad_norm": 0.5295129961762205,
      "learning_rate": 8.390002055119084e-07,
      "loss": 11.7691,
      "step": 35258
    },
    {
      "epoch": 1.91999346552041,
      "grad_norm": 0.5341947484634186,
      "learning_rate": 8.378606930597999e-07,
      "loss": 11.9021,
      "step": 35259
    },
    {
      "epoch": 1.920047919516993,
      "grad_norm": 0.49459724132531996,
      "learning_rate": 8.367219517091074e-07,
      "loss": 11.6327,
      "step": 35260
    },
    {
      "epoch": 1.920102373513576,
      "grad_norm": 0.552824060456351,
      "learning_rate": 8.355839814687017e-07,
      "loss": 11.7846,
      "step": 35261
    },
    {
      "epoch": 1.920156827510159,
      "grad_norm": 0.5945931968247791,
      "learning_rate": 8.344467823474311e-07,
      "loss": 11.7119,
      "step": 35262
    },
    {
      "epoch": 1.920211281506742,
      "grad_norm": 0.5165917185399999,
      "learning_rate": 8.33310354354122e-07,
      "loss": 11.7467,
      "step": 35263
    },
    {
      "epoch": 1.920265735503325,
      "grad_norm": 0.5398463526646301,
      "learning_rate": 8.321746974976341e-07,
      "loss": 11.689,
      "step": 35264
    },
    {
      "epoch": 1.920320189499908,
      "grad_norm": 0.5122382129269146,
      "learning_rate": 8.310398117867713e-07,
      "loss": 11.686,
      "step": 35265
    },
    {
      "epoch": 1.920374643496491,
      "grad_norm": 0.4963068238640571,
      "learning_rate": 8.299056972303821e-07,
      "loss": 11.769,
      "step": 35266
    },
    {
      "epoch": 1.920429097493074,
      "grad_norm": 0.5939624631643763,
      "learning_rate": 8.287723538372705e-07,
      "loss": 11.841,
      "step": 35267
    },
    {
      "epoch": 1.920483551489657,
      "grad_norm": 0.5601564948005896,
      "learning_rate": 8.276397816162629e-07,
      "loss": 11.6909,
      "step": 35268
    },
    {
      "epoch": 1.9205380054862402,
      "grad_norm": 0.5345917124766969,
      "learning_rate": 8.265079805761522e-07,
      "loss": 11.6399,
      "step": 35269
    },
    {
      "epoch": 1.9205924594828232,
      "grad_norm": 0.5148940207216658,
      "learning_rate": 8.253769507257536e-07,
      "loss": 11.6773,
      "step": 35270
    },
    {
      "epoch": 1.9206469134794062,
      "grad_norm": 0.5404967793073,
      "learning_rate": 8.242466920738601e-07,
      "loss": 11.6103,
      "step": 35271
    },
    {
      "epoch": 1.9207013674759892,
      "grad_norm": 0.5186321777187234,
      "learning_rate": 8.231172046292424e-07,
      "loss": 11.875,
      "step": 35272
    },
    {
      "epoch": 1.9207558214725722,
      "grad_norm": 0.5180517870130527,
      "learning_rate": 8.219884884007045e-07,
      "loss": 11.7603,
      "step": 35273
    },
    {
      "epoch": 1.9208102754691552,
      "grad_norm": 0.504696692025867,
      "learning_rate": 8.208605433970173e-07,
      "loss": 11.7911,
      "step": 35274
    },
    {
      "epoch": 1.9208647294657384,
      "grad_norm": 0.5492580926459601,
      "learning_rate": 8.197333696269515e-07,
      "loss": 11.827,
      "step": 35275
    },
    {
      "epoch": 1.9209191834623214,
      "grad_norm": 0.5867281708173184,
      "learning_rate": 8.186069670992779e-07,
      "loss": 11.7523,
      "step": 35276
    },
    {
      "epoch": 1.9209736374589044,
      "grad_norm": 0.6153437883647792,
      "learning_rate": 8.17481335822734e-07,
      "loss": 11.7761,
      "step": 35277
    },
    {
      "epoch": 1.9210280914554874,
      "grad_norm": 0.5026668114833792,
      "learning_rate": 8.163564758060905e-07,
      "loss": 11.6629,
      "step": 35278
    },
    {
      "epoch": 1.9210825454520704,
      "grad_norm": 0.5320818665546363,
      "learning_rate": 8.152323870581069e-07,
      "loss": 11.7831,
      "step": 35279
    },
    {
      "epoch": 1.9211369994486533,
      "grad_norm": 0.5681963567239882,
      "learning_rate": 8.141090695874876e-07,
      "loss": 11.8499,
      "step": 35280
    },
    {
      "epoch": 1.9211914534452363,
      "grad_norm": 0.4949459466197695,
      "learning_rate": 8.129865234030032e-07,
      "loss": 11.7889,
      "step": 35281
    },
    {
      "epoch": 1.9212459074418193,
      "grad_norm": 0.5349621663826487,
      "learning_rate": 8.11864748513369e-07,
      "loss": 11.7013,
      "step": 35282
    },
    {
      "epoch": 1.9213003614384023,
      "grad_norm": 0.5774407758647135,
      "learning_rate": 8.107437449273114e-07,
      "loss": 11.8837,
      "step": 35283
    },
    {
      "epoch": 1.9213548154349853,
      "grad_norm": 0.5460741780068611,
      "learning_rate": 8.096235126535456e-07,
      "loss": 11.8375,
      "step": 35284
    },
    {
      "epoch": 1.9214092694315683,
      "grad_norm": 0.5667162986954193,
      "learning_rate": 8.085040517007758e-07,
      "loss": 11.7073,
      "step": 35285
    },
    {
      "epoch": 1.9214637234281513,
      "grad_norm": 0.5757809426311653,
      "learning_rate": 8.073853620777061e-07,
      "loss": 11.8311,
      "step": 35286
    },
    {
      "epoch": 1.9215181774247343,
      "grad_norm": 0.5300675981497448,
      "learning_rate": 8.062674437930517e-07,
      "loss": 11.6415,
      "step": 35287
    },
    {
      "epoch": 1.9215726314213173,
      "grad_norm": 0.5504102236991174,
      "learning_rate": 8.051502968554947e-07,
      "loss": 11.8556,
      "step": 35288
    },
    {
      "epoch": 1.9216270854179003,
      "grad_norm": 0.5219588919888506,
      "learning_rate": 8.04033921273728e-07,
      "loss": 11.7289,
      "step": 35289
    },
    {
      "epoch": 1.9216815394144833,
      "grad_norm": 0.532605736327605,
      "learning_rate": 8.029183170564225e-07,
      "loss": 11.9089,
      "step": 35290
    },
    {
      "epoch": 1.9217359934110663,
      "grad_norm": 0.5545374656691011,
      "learning_rate": 8.018034842122601e-07,
      "loss": 11.6906,
      "step": 35291
    },
    {
      "epoch": 1.9217904474076493,
      "grad_norm": 0.5785721568173094,
      "learning_rate": 8.006894227499118e-07,
      "loss": 11.8068,
      "step": 35292
    },
    {
      "epoch": 1.9218449014042325,
      "grad_norm": 0.5855414702112186,
      "learning_rate": 7.995761326780371e-07,
      "loss": 11.8148,
      "step": 35293
    },
    {
      "epoch": 1.9218993554008155,
      "grad_norm": 0.5362638828443235,
      "learning_rate": 7.984636140052959e-07,
      "loss": 11.7047,
      "step": 35294
    },
    {
      "epoch": 1.9219538093973985,
      "grad_norm": 0.5477454407750199,
      "learning_rate": 7.973518667403368e-07,
      "loss": 11.7107,
      "step": 35295
    },
    {
      "epoch": 1.9220082633939815,
      "grad_norm": 0.4952478844872741,
      "learning_rate": 7.962408908918085e-07,
      "loss": 11.7421,
      "step": 35296
    },
    {
      "epoch": 1.9220627173905644,
      "grad_norm": 0.5232201606893285,
      "learning_rate": 7.951306864683372e-07,
      "loss": 11.7441,
      "step": 35297
    },
    {
      "epoch": 1.9221171713871477,
      "grad_norm": 0.5422497469933022,
      "learning_rate": 7.940212534785718e-07,
      "loss": 11.8066,
      "step": 35298
    },
    {
      "epoch": 1.9221716253837307,
      "grad_norm": 0.5773271614851948,
      "learning_rate": 7.929125919311387e-07,
      "loss": 11.8527,
      "step": 35299
    },
    {
      "epoch": 1.9222260793803136,
      "grad_norm": 0.5069161014283836,
      "learning_rate": 7.918047018346419e-07,
      "loss": 11.6907,
      "step": 35300
    },
    {
      "epoch": 1.9222805333768966,
      "grad_norm": 0.5265502382157767,
      "learning_rate": 7.906975831977303e-07,
      "loss": 11.8117,
      "step": 35301
    },
    {
      "epoch": 1.9223349873734796,
      "grad_norm": 0.5979082352160378,
      "learning_rate": 7.895912360289748e-07,
      "loss": 11.7516,
      "step": 35302
    },
    {
      "epoch": 1.9223894413700626,
      "grad_norm": 0.4978611012246709,
      "learning_rate": 7.884856603370017e-07,
      "loss": 11.6019,
      "step": 35303
    },
    {
      "epoch": 1.9224438953666456,
      "grad_norm": 0.5553903313803183,
      "learning_rate": 7.873808561304041e-07,
      "loss": 11.8605,
      "step": 35304
    },
    {
      "epoch": 1.9224983493632286,
      "grad_norm": 0.486835429970468,
      "learning_rate": 7.862768234177643e-07,
      "loss": 11.6685,
      "step": 35305
    },
    {
      "epoch": 1.9225528033598116,
      "grad_norm": 0.5372476764856517,
      "learning_rate": 7.85173562207675e-07,
      "loss": 11.7563,
      "step": 35306
    },
    {
      "epoch": 1.9226072573563946,
      "grad_norm": 0.5329644864882591,
      "learning_rate": 7.840710725087186e-07,
      "loss": 11.6573,
      "step": 35307
    },
    {
      "epoch": 1.9226617113529776,
      "grad_norm": 0.7622171642642938,
      "learning_rate": 7.829693543294658e-07,
      "loss": 11.8296,
      "step": 35308
    },
    {
      "epoch": 1.9227161653495606,
      "grad_norm": 0.6431431136042206,
      "learning_rate": 7.818684076784766e-07,
      "loss": 11.9103,
      "step": 35309
    },
    {
      "epoch": 1.9227706193461436,
      "grad_norm": 0.5997994841951583,
      "learning_rate": 7.807682325643218e-07,
      "loss": 11.7868,
      "step": 35310
    },
    {
      "epoch": 1.9228250733427266,
      "grad_norm": 0.5578947978803397,
      "learning_rate": 7.796688289955389e-07,
      "loss": 11.8253,
      "step": 35311
    },
    {
      "epoch": 1.9228795273393096,
      "grad_norm": 0.5286349018608756,
      "learning_rate": 7.785701969806991e-07,
      "loss": 11.6868,
      "step": 35312
    },
    {
      "epoch": 1.9229339813358926,
      "grad_norm": 0.583396661750172,
      "learning_rate": 7.77472336528351e-07,
      "loss": 11.861,
      "step": 35313
    },
    {
      "epoch": 1.9229884353324755,
      "grad_norm": 0.5394180114020597,
      "learning_rate": 7.763752476469988e-07,
      "loss": 11.696,
      "step": 35314
    },
    {
      "epoch": 1.9230428893290585,
      "grad_norm": 0.6335493071578289,
      "learning_rate": 7.752789303452024e-07,
      "loss": 11.6758,
      "step": 35315
    },
    {
      "epoch": 1.9230973433256418,
      "grad_norm": 0.5197597917635172,
      "learning_rate": 7.741833846314773e-07,
      "loss": 11.8524,
      "step": 35316
    },
    {
      "epoch": 1.9231517973222247,
      "grad_norm": 0.5378542335438624,
      "learning_rate": 7.730886105143387e-07,
      "loss": 11.6057,
      "step": 35317
    },
    {
      "epoch": 1.9232062513188077,
      "grad_norm": 0.5634133311170223,
      "learning_rate": 7.719946080023021e-07,
      "loss": 11.8391,
      "step": 35318
    },
    {
      "epoch": 1.9232607053153907,
      "grad_norm": 0.5735120756466036,
      "learning_rate": 7.70901377103872e-07,
      "loss": 11.8891,
      "step": 35319
    },
    {
      "epoch": 1.9233151593119737,
      "grad_norm": 0.5057175099043302,
      "learning_rate": 7.698089178275635e-07,
      "loss": 11.8658,
      "step": 35320
    },
    {
      "epoch": 1.9233696133085567,
      "grad_norm": 0.5543803388535611,
      "learning_rate": 7.68717230181859e-07,
      "loss": 11.8473,
      "step": 35321
    },
    {
      "epoch": 1.92342406730514,
      "grad_norm": 0.5872870913362307,
      "learning_rate": 7.676263141752404e-07,
      "loss": 11.817,
      "step": 35322
    },
    {
      "epoch": 1.923478521301723,
      "grad_norm": 0.51143281922094,
      "learning_rate": 7.665361698162121e-07,
      "loss": 11.7323,
      "step": 35323
    },
    {
      "epoch": 1.923532975298306,
      "grad_norm": 0.5044437056489144,
      "learning_rate": 7.654467971132451e-07,
      "loss": 11.7122,
      "step": 35324
    },
    {
      "epoch": 1.923587429294889,
      "grad_norm": 0.5450926954154512,
      "learning_rate": 7.643581960747992e-07,
      "loss": 11.6908,
      "step": 35325
    },
    {
      "epoch": 1.923641883291472,
      "grad_norm": 0.5756954108170521,
      "learning_rate": 7.632703667093344e-07,
      "loss": 11.848,
      "step": 35326
    },
    {
      "epoch": 1.923696337288055,
      "grad_norm": 0.4863770442798545,
      "learning_rate": 7.621833090253327e-07,
      "loss": 11.6358,
      "step": 35327
    },
    {
      "epoch": 1.9237507912846379,
      "grad_norm": 0.508350176005348,
      "learning_rate": 7.61097023031232e-07,
      "loss": 11.8138,
      "step": 35328
    },
    {
      "epoch": 1.9238052452812209,
      "grad_norm": 0.5992019878117516,
      "learning_rate": 7.60011508735492e-07,
      "loss": 11.6911,
      "step": 35329
    },
    {
      "epoch": 1.9238596992778039,
      "grad_norm": 0.5505014228180067,
      "learning_rate": 7.589267661465393e-07,
      "loss": 11.8331,
      "step": 35330
    },
    {
      "epoch": 1.9239141532743869,
      "grad_norm": 0.5914248529883434,
      "learning_rate": 7.578427952728117e-07,
      "loss": 11.6986,
      "step": 35331
    },
    {
      "epoch": 1.9239686072709699,
      "grad_norm": 0.5849041648920296,
      "learning_rate": 7.567595961227469e-07,
      "loss": 11.8056,
      "step": 35332
    },
    {
      "epoch": 1.9240230612675528,
      "grad_norm": 0.5272009266257041,
      "learning_rate": 7.556771687047492e-07,
      "loss": 11.5029,
      "step": 35333
    },
    {
      "epoch": 1.9240775152641358,
      "grad_norm": 0.5604447882902809,
      "learning_rate": 7.545955130272675e-07,
      "loss": 11.8475,
      "step": 35334
    },
    {
      "epoch": 1.9241319692607188,
      "grad_norm": 0.5504769635070138,
      "learning_rate": 7.535146290986839e-07,
      "loss": 11.8832,
      "step": 35335
    },
    {
      "epoch": 1.9241864232573018,
      "grad_norm": 0.5029963049925109,
      "learning_rate": 7.524345169274139e-07,
      "loss": 11.7606,
      "step": 35336
    },
    {
      "epoch": 1.9242408772538848,
      "grad_norm": 0.5312068726749273,
      "learning_rate": 7.513551765218618e-07,
      "loss": 11.8833,
      "step": 35337
    },
    {
      "epoch": 1.9242953312504678,
      "grad_norm": 0.5125757415662441,
      "learning_rate": 7.5027660789041e-07,
      "loss": 11.7914,
      "step": 35338
    },
    {
      "epoch": 1.924349785247051,
      "grad_norm": 0.5649880153604923,
      "learning_rate": 7.491988110414517e-07,
      "loss": 11.6746,
      "step": 35339
    },
    {
      "epoch": 1.924404239243634,
      "grad_norm": 0.5012386348472996,
      "learning_rate": 7.4812178598338e-07,
      "loss": 11.795,
      "step": 35340
    },
    {
      "epoch": 1.924458693240217,
      "grad_norm": 0.5576508571761037,
      "learning_rate": 7.47045532724544e-07,
      "loss": 11.6963,
      "step": 35341
    },
    {
      "epoch": 1.9245131472368,
      "grad_norm": 0.5388868489754646,
      "learning_rate": 7.459700512733369e-07,
      "loss": 11.8101,
      "step": 35342
    },
    {
      "epoch": 1.924567601233383,
      "grad_norm": 0.5524929396746211,
      "learning_rate": 7.448953416380966e-07,
      "loss": 11.755,
      "step": 35343
    },
    {
      "epoch": 1.924622055229966,
      "grad_norm": 0.5281468662307155,
      "learning_rate": 7.438214038272162e-07,
      "loss": 11.6712,
      "step": 35344
    },
    {
      "epoch": 1.9246765092265492,
      "grad_norm": 0.5019786033804621,
      "learning_rate": 7.427482378490114e-07,
      "loss": 11.7936,
      "step": 35345
    },
    {
      "epoch": 1.9247309632231322,
      "grad_norm": 0.4985147264600398,
      "learning_rate": 7.416758437118421e-07,
      "loss": 11.7349,
      "step": 35346
    },
    {
      "epoch": 1.9247854172197152,
      "grad_norm": 0.6123380396788846,
      "learning_rate": 7.406042214240572e-07,
      "loss": 11.7973,
      "step": 35347
    },
    {
      "epoch": 1.9248398712162982,
      "grad_norm": 0.5479284479493136,
      "learning_rate": 7.395333709939722e-07,
      "loss": 11.7835,
      "step": 35348
    },
    {
      "epoch": 1.9248943252128812,
      "grad_norm": 0.611817874657679,
      "learning_rate": 7.384632924299139e-07,
      "loss": 11.841,
      "step": 35349
    },
    {
      "epoch": 1.9249487792094642,
      "grad_norm": 0.5113991007685988,
      "learning_rate": 7.373939857402201e-07,
      "loss": 11.7846,
      "step": 35350
    },
    {
      "epoch": 1.9250032332060472,
      "grad_norm": 0.852600629121004,
      "learning_rate": 7.363254509331952e-07,
      "loss": 11.6187,
      "step": 35351
    },
    {
      "epoch": 1.9250576872026302,
      "grad_norm": 0.5040558355011653,
      "learning_rate": 7.352576880171547e-07,
      "loss": 11.821,
      "step": 35352
    },
    {
      "epoch": 1.9251121411992131,
      "grad_norm": 0.5355425617423119,
      "learning_rate": 7.34190697000392e-07,
      "loss": 11.7523,
      "step": 35353
    },
    {
      "epoch": 1.9251665951957961,
      "grad_norm": 0.49322277301141726,
      "learning_rate": 7.331244778912116e-07,
      "loss": 11.8026,
      "step": 35354
    },
    {
      "epoch": 1.9252210491923791,
      "grad_norm": 0.5022982131103606,
      "learning_rate": 7.32059030697907e-07,
      "loss": 11.843,
      "step": 35355
    },
    {
      "epoch": 1.9252755031889621,
      "grad_norm": 0.5167383581109334,
      "learning_rate": 7.30994355428749e-07,
      "loss": 11.7494,
      "step": 35356
    },
    {
      "epoch": 1.9253299571855451,
      "grad_norm": 0.532425071896843,
      "learning_rate": 7.299304520920313e-07,
      "loss": 11.8206,
      "step": 35357
    },
    {
      "epoch": 1.925384411182128,
      "grad_norm": 0.5259448588169953,
      "learning_rate": 7.288673206960139e-07,
      "loss": 11.7795,
      "step": 35358
    },
    {
      "epoch": 1.925438865178711,
      "grad_norm": 0.5248042446606963,
      "learning_rate": 7.278049612489901e-07,
      "loss": 11.8046,
      "step": 35359
    },
    {
      "epoch": 1.925493319175294,
      "grad_norm": 0.5425324117292638,
      "learning_rate": 7.267433737591867e-07,
      "loss": 11.8307,
      "step": 35360
    },
    {
      "epoch": 1.925547773171877,
      "grad_norm": 0.5062761099637456,
      "learning_rate": 7.256825582348748e-07,
      "loss": 11.7454,
      "step": 35361
    },
    {
      "epoch": 1.92560222716846,
      "grad_norm": 0.5352392168668737,
      "learning_rate": 7.246225146843144e-07,
      "loss": 11.7295,
      "step": 35362
    },
    {
      "epoch": 1.9256566811650433,
      "grad_norm": 0.5184501313269085,
      "learning_rate": 7.235632431157325e-07,
      "loss": 11.7991,
      "step": 35363
    },
    {
      "epoch": 1.9257111351616263,
      "grad_norm": 0.5216924404581125,
      "learning_rate": 7.225047435373666e-07,
      "loss": 11.8121,
      "step": 35364
    },
    {
      "epoch": 1.9257655891582093,
      "grad_norm": 0.5641134648379601,
      "learning_rate": 7.21447015957466e-07,
      "loss": 11.8982,
      "step": 35365
    },
    {
      "epoch": 1.9258200431547923,
      "grad_norm": 0.5949317025428467,
      "learning_rate": 7.20390060384224e-07,
      "loss": 11.7294,
      "step": 35366
    },
    {
      "epoch": 1.9258744971513753,
      "grad_norm": 0.5402392675001839,
      "learning_rate": 7.193338768259005e-07,
      "loss": 11.7455,
      "step": 35367
    },
    {
      "epoch": 1.9259289511479585,
      "grad_norm": 0.6061302741210017,
      "learning_rate": 7.182784652906671e-07,
      "loss": 11.7955,
      "step": 35368
    },
    {
      "epoch": 1.9259834051445415,
      "grad_norm": 0.5470773797960535,
      "learning_rate": 7.172238257867614e-07,
      "loss": 11.7651,
      "step": 35369
    },
    {
      "epoch": 1.9260378591411245,
      "grad_norm": 0.5421452448098459,
      "learning_rate": 7.161699583223658e-07,
      "loss": 11.7071,
      "step": 35370
    },
    {
      "epoch": 1.9260923131377075,
      "grad_norm": 0.5261602671605133,
      "learning_rate": 7.151168629056737e-07,
      "loss": 11.8025,
      "step": 35371
    },
    {
      "epoch": 1.9261467671342904,
      "grad_norm": 0.5599441292931336,
      "learning_rate": 7.140645395449008e-07,
      "loss": 11.7575,
      "step": 35372
    },
    {
      "epoch": 1.9262012211308734,
      "grad_norm": 0.5600867048068385,
      "learning_rate": 7.130129882482073e-07,
      "loss": 11.6645,
      "step": 35373
    },
    {
      "epoch": 1.9262556751274564,
      "grad_norm": 0.5715260293873371,
      "learning_rate": 7.119622090237754e-07,
      "loss": 11.8297,
      "step": 35374
    },
    {
      "epoch": 1.9263101291240394,
      "grad_norm": 0.5468786873557178,
      "learning_rate": 7.109122018797764e-07,
      "loss": 11.8405,
      "step": 35375
    },
    {
      "epoch": 1.9263645831206224,
      "grad_norm": 0.5285094818170534,
      "learning_rate": 7.098629668243706e-07,
      "loss": 11.7535,
      "step": 35376
    },
    {
      "epoch": 1.9264190371172054,
      "grad_norm": 0.5607105169284818,
      "learning_rate": 7.088145038657179e-07,
      "loss": 11.6754,
      "step": 35377
    },
    {
      "epoch": 1.9264734911137884,
      "grad_norm": 0.49940159785988447,
      "learning_rate": 7.077668130119786e-07,
      "loss": 11.7179,
      "step": 35378
    },
    {
      "epoch": 1.9265279451103714,
      "grad_norm": 0.4907213726895333,
      "learning_rate": 7.067198942713016e-07,
      "loss": 11.7589,
      "step": 35379
    },
    {
      "epoch": 1.9265823991069544,
      "grad_norm": 0.5659091368713514,
      "learning_rate": 7.056737476518138e-07,
      "loss": 11.9466,
      "step": 35380
    },
    {
      "epoch": 1.9266368531035374,
      "grad_norm": 0.5643104997612685,
      "learning_rate": 7.046283731616643e-07,
      "loss": 11.901,
      "step": 35381
    },
    {
      "epoch": 1.9266913071001204,
      "grad_norm": 0.5137125062711562,
      "learning_rate": 7.035837708089799e-07,
      "loss": 11.8287,
      "step": 35382
    },
    {
      "epoch": 1.9267457610967034,
      "grad_norm": 0.5954182365678513,
      "learning_rate": 7.025399406018652e-07,
      "loss": 11.7802,
      "step": 35383
    },
    {
      "epoch": 1.9268002150932864,
      "grad_norm": 0.5230969445519821,
      "learning_rate": 7.014968825484691e-07,
      "loss": 11.7085,
      "step": 35384
    },
    {
      "epoch": 1.9268546690898694,
      "grad_norm": 0.5325847700152697,
      "learning_rate": 7.004545966568854e-07,
      "loss": 11.671,
      "step": 35385
    },
    {
      "epoch": 1.9269091230864526,
      "grad_norm": 0.5498119132753048,
      "learning_rate": 6.994130829352074e-07,
      "loss": 11.873,
      "step": 35386
    },
    {
      "epoch": 1.9269635770830356,
      "grad_norm": 0.5351749950341316,
      "learning_rate": 6.98372341391551e-07,
      "loss": 11.7139,
      "step": 35387
    },
    {
      "epoch": 1.9270180310796186,
      "grad_norm": 0.5443566007650343,
      "learning_rate": 6.973323720339986e-07,
      "loss": 11.8042,
      "step": 35388
    },
    {
      "epoch": 1.9270724850762015,
      "grad_norm": 0.5390561097638134,
      "learning_rate": 6.962931748706547e-07,
      "loss": 11.7709,
      "step": 35389
    },
    {
      "epoch": 1.9271269390727845,
      "grad_norm": 0.565712985744551,
      "learning_rate": 6.952547499095797e-07,
      "loss": 11.8785,
      "step": 35390
    },
    {
      "epoch": 1.9271813930693675,
      "grad_norm": 0.5805432028399905,
      "learning_rate": 6.942170971588558e-07,
      "loss": 11.8645,
      "step": 35391
    },
    {
      "epoch": 1.9272358470659507,
      "grad_norm": 0.625499406559887,
      "learning_rate": 6.931802166265544e-07,
      "loss": 11.7966,
      "step": 35392
    },
    {
      "epoch": 1.9272903010625337,
      "grad_norm": 0.517132198749326,
      "learning_rate": 6.92144108320747e-07,
      "loss": 11.7365,
      "step": 35393
    },
    {
      "epoch": 1.9273447550591167,
      "grad_norm": 0.5861462659480324,
      "learning_rate": 6.911087722494824e-07,
      "loss": 11.8478,
      "step": 35394
    },
    {
      "epoch": 1.9273992090556997,
      "grad_norm": 0.5619082696620521,
      "learning_rate": 6.90074208420799e-07,
      "loss": 11.7398,
      "step": 35395
    },
    {
      "epoch": 1.9274536630522827,
      "grad_norm": 0.5148110059500736,
      "learning_rate": 6.890404168427567e-07,
      "loss": 11.7284,
      "step": 35396
    },
    {
      "epoch": 1.9275081170488657,
      "grad_norm": 0.5645150423670221,
      "learning_rate": 6.880073975233825e-07,
      "loss": 11.9115,
      "step": 35397
    },
    {
      "epoch": 1.9275625710454487,
      "grad_norm": 0.6195674293480592,
      "learning_rate": 6.869751504707255e-07,
      "loss": 11.7153,
      "step": 35398
    },
    {
      "epoch": 1.9276170250420317,
      "grad_norm": 0.5095060571853789,
      "learning_rate": 6.859436756928017e-07,
      "loss": 11.7286,
      "step": 35399
    },
    {
      "epoch": 1.9276714790386147,
      "grad_norm": 0.5364047275544772,
      "learning_rate": 6.849129731976378e-07,
      "loss": 11.7814,
      "step": 35400
    },
    {
      "epoch": 1.9277259330351977,
      "grad_norm": 0.5484462847732213,
      "learning_rate": 6.838830429932386e-07,
      "loss": 11.6552,
      "step": 35401
    },
    {
      "epoch": 1.9277803870317807,
      "grad_norm": 0.5992665760189813,
      "learning_rate": 6.828538850876309e-07,
      "loss": 11.8318,
      "step": 35402
    },
    {
      "epoch": 1.9278348410283637,
      "grad_norm": 0.5201795017438869,
      "learning_rate": 6.818254994887973e-07,
      "loss": 11.8069,
      "step": 35403
    },
    {
      "epoch": 1.9278892950249467,
      "grad_norm": 0.5738206542609812,
      "learning_rate": 6.807978862047537e-07,
      "loss": 11.8386,
      "step": 35404
    },
    {
      "epoch": 1.9279437490215297,
      "grad_norm": 0.535493682482242,
      "learning_rate": 6.797710452434714e-07,
      "loss": 11.6852,
      "step": 35405
    },
    {
      "epoch": 1.9279982030181126,
      "grad_norm": 0.5544308372407974,
      "learning_rate": 6.78744976612955e-07,
      "loss": 11.8048,
      "step": 35406
    },
    {
      "epoch": 1.9280526570146956,
      "grad_norm": 0.6455312421116323,
      "learning_rate": 6.777196803211649e-07,
      "loss": 11.8757,
      "step": 35407
    },
    {
      "epoch": 1.9281071110112786,
      "grad_norm": 0.49445511996845926,
      "learning_rate": 6.766951563760948e-07,
      "loss": 11.6895,
      "step": 35408
    },
    {
      "epoch": 1.9281615650078618,
      "grad_norm": 0.5431249779727285,
      "learning_rate": 6.756714047856938e-07,
      "loss": 11.7307,
      "step": 35409
    },
    {
      "epoch": 1.9282160190044448,
      "grad_norm": 0.5803088746264368,
      "learning_rate": 6.746484255579222e-07,
      "loss": 11.7811,
      "step": 35410
    },
    {
      "epoch": 1.9282704730010278,
      "grad_norm": 0.514407079938101,
      "learning_rate": 6.736262187007513e-07,
      "loss": 11.5352,
      "step": 35411
    },
    {
      "epoch": 1.9283249269976108,
      "grad_norm": 0.5249767434297249,
      "learning_rate": 6.726047842221084e-07,
      "loss": 11.8586,
      "step": 35412
    },
    {
      "epoch": 1.9283793809941938,
      "grad_norm": 0.5934939878917341,
      "learning_rate": 6.715841221299534e-07,
      "loss": 11.6992,
      "step": 35413
    },
    {
      "epoch": 1.9284338349907768,
      "grad_norm": 0.5314712080866695,
      "learning_rate": 6.705642324322248e-07,
      "loss": 11.7601,
      "step": 35414
    },
    {
      "epoch": 1.92848828898736,
      "grad_norm": 0.5375466080781637,
      "learning_rate": 6.695451151368381e-07,
      "loss": 11.5534,
      "step": 35415
    },
    {
      "epoch": 1.928542742983943,
      "grad_norm": 0.5091149042748288,
      "learning_rate": 6.685267702517317e-07,
      "loss": 11.7133,
      "step": 35416
    },
    {
      "epoch": 1.928597196980526,
      "grad_norm": 0.5299152011881831,
      "learning_rate": 6.675091977848102e-07,
      "loss": 11.7254,
      "step": 35417
    },
    {
      "epoch": 1.928651650977109,
      "grad_norm": 0.5082856489996755,
      "learning_rate": 6.664923977440119e-07,
      "loss": 11.7531,
      "step": 35418
    },
    {
      "epoch": 1.928706104973692,
      "grad_norm": 0.5583785568851579,
      "learning_rate": 6.65476370137208e-07,
      "loss": 11.7428,
      "step": 35419
    },
    {
      "epoch": 1.928760558970275,
      "grad_norm": 0.5235554945693908,
      "learning_rate": 6.644611149723257e-07,
      "loss": 11.6274,
      "step": 35420
    },
    {
      "epoch": 1.928815012966858,
      "grad_norm": 0.5536106244421286,
      "learning_rate": 6.634466322572586e-07,
      "loss": 11.756,
      "step": 35421
    },
    {
      "epoch": 1.928869466963441,
      "grad_norm": 0.598404157875678,
      "learning_rate": 6.624329219998781e-07,
      "loss": 11.7432,
      "step": 35422
    },
    {
      "epoch": 1.928923920960024,
      "grad_norm": 0.5319733815319337,
      "learning_rate": 6.614199842080893e-07,
      "loss": 11.7326,
      "step": 35423
    },
    {
      "epoch": 1.928978374956607,
      "grad_norm": 0.5583179663698711,
      "learning_rate": 6.604078188897523e-07,
      "loss": 11.8577,
      "step": 35424
    },
    {
      "epoch": 1.92903282895319,
      "grad_norm": 0.6470860211071525,
      "learning_rate": 6.593964260527385e-07,
      "loss": 11.6878,
      "step": 35425
    },
    {
      "epoch": 1.929087282949773,
      "grad_norm": 0.5941266616046202,
      "learning_rate": 6.583858057049308e-07,
      "loss": 11.9415,
      "step": 35426
    },
    {
      "epoch": 1.929141736946356,
      "grad_norm": 0.5077404659532531,
      "learning_rate": 6.57375957854156e-07,
      "loss": 11.7247,
      "step": 35427
    },
    {
      "epoch": 1.929196190942939,
      "grad_norm": 0.5371235103430241,
      "learning_rate": 6.56366882508297e-07,
      "loss": 11.697,
      "step": 35428
    },
    {
      "epoch": 1.929250644939522,
      "grad_norm": 0.5335731010933777,
      "learning_rate": 6.553585796751916e-07,
      "loss": 11.8056,
      "step": 35429
    },
    {
      "epoch": 1.929305098936105,
      "grad_norm": 0.5326915148899944,
      "learning_rate": 6.543510493626781e-07,
      "loss": 11.7094,
      "step": 35430
    },
    {
      "epoch": 1.929359552932688,
      "grad_norm": 0.5458947629846782,
      "learning_rate": 6.533442915785836e-07,
      "loss": 11.866,
      "step": 35431
    },
    {
      "epoch": 1.9294140069292711,
      "grad_norm": 0.7991401788652261,
      "learning_rate": 6.523383063307465e-07,
      "loss": 11.6802,
      "step": 35432
    },
    {
      "epoch": 1.9294684609258541,
      "grad_norm": 0.55400984203346,
      "learning_rate": 6.513330936269824e-07,
      "loss": 11.7957,
      "step": 35433
    },
    {
      "epoch": 1.929522914922437,
      "grad_norm": 0.5486984418403,
      "learning_rate": 6.503286534751185e-07,
      "loss": 11.8621,
      "step": 35434
    },
    {
      "epoch": 1.92957736891902,
      "grad_norm": 0.5650074778730099,
      "learning_rate": 6.493249858829597e-07,
      "loss": 11.9049,
      "step": 35435
    },
    {
      "epoch": 1.929631822915603,
      "grad_norm": 0.5348936191789725,
      "learning_rate": 6.483220908583221e-07,
      "loss": 11.7854,
      "step": 35436
    },
    {
      "epoch": 1.929686276912186,
      "grad_norm": 0.5484853027327932,
      "learning_rate": 6.47319968408977e-07,
      "loss": 11.7551,
      "step": 35437
    },
    {
      "epoch": 1.9297407309087693,
      "grad_norm": 0.596676607543935,
      "learning_rate": 6.463186185427405e-07,
      "loss": 11.9053,
      "step": 35438
    },
    {
      "epoch": 1.9297951849053523,
      "grad_norm": 0.5244260156174037,
      "learning_rate": 6.453180412673843e-07,
      "loss": 11.8333,
      "step": 35439
    },
    {
      "epoch": 1.9298496389019353,
      "grad_norm": 0.6116199511475648,
      "learning_rate": 6.44318236590713e-07,
      "loss": 11.8368,
      "step": 35440
    },
    {
      "epoch": 1.9299040928985183,
      "grad_norm": 0.5096518970216495,
      "learning_rate": 6.433192045204762e-07,
      "loss": 11.8159,
      "step": 35441
    },
    {
      "epoch": 1.9299585468951013,
      "grad_norm": 0.508541784617971,
      "learning_rate": 6.423209450644452e-07,
      "loss": 11.7323,
      "step": 35442
    },
    {
      "epoch": 1.9300130008916843,
      "grad_norm": 0.5555457044378989,
      "learning_rate": 6.413234582303918e-07,
      "loss": 11.8258,
      "step": 35443
    },
    {
      "epoch": 1.9300674548882673,
      "grad_norm": 0.5365538766671122,
      "learning_rate": 6.403267440260763e-07,
      "loss": 11.7692,
      "step": 35444
    },
    {
      "epoch": 1.9301219088848502,
      "grad_norm": 0.5190823342156515,
      "learning_rate": 6.39330802459237e-07,
      "loss": 11.6126,
      "step": 35445
    },
    {
      "epoch": 1.9301763628814332,
      "grad_norm": 0.5493325550371453,
      "learning_rate": 6.383356335376234e-07,
      "loss": 11.8537,
      "step": 35446
    },
    {
      "epoch": 1.9302308168780162,
      "grad_norm": 0.568077630585568,
      "learning_rate": 6.373412372689735e-07,
      "loss": 11.887,
      "step": 35447
    },
    {
      "epoch": 1.9302852708745992,
      "grad_norm": 0.5642403482181683,
      "learning_rate": 6.363476136610369e-07,
      "loss": 11.815,
      "step": 35448
    },
    {
      "epoch": 1.9303397248711822,
      "grad_norm": 0.5813725648126096,
      "learning_rate": 6.353547627215073e-07,
      "loss": 11.8681,
      "step": 35449
    },
    {
      "epoch": 1.9303941788677652,
      "grad_norm": 0.5534224653106757,
      "learning_rate": 6.343626844581229e-07,
      "loss": 11.7836,
      "step": 35450
    },
    {
      "epoch": 1.9304486328643482,
      "grad_norm": 0.5909850161885183,
      "learning_rate": 6.333713788785999e-07,
      "loss": 11.7532,
      "step": 35451
    },
    {
      "epoch": 1.9305030868609312,
      "grad_norm": 0.5617021735881333,
      "learning_rate": 6.323808459906544e-07,
      "loss": 11.8077,
      "step": 35452
    },
    {
      "epoch": 1.9305575408575142,
      "grad_norm": 0.5088197444062463,
      "learning_rate": 6.31391085801969e-07,
      "loss": 11.7859,
      "step": 35453
    },
    {
      "epoch": 1.9306119948540972,
      "grad_norm": 0.5667951383496671,
      "learning_rate": 6.304020983202486e-07,
      "loss": 11.7958,
      "step": 35454
    },
    {
      "epoch": 1.9306664488506802,
      "grad_norm": 0.5631110617421782,
      "learning_rate": 6.294138835531982e-07,
      "loss": 11.7575,
      "step": 35455
    },
    {
      "epoch": 1.9307209028472634,
      "grad_norm": 0.4771713502211208,
      "learning_rate": 6.284264415084895e-07,
      "loss": 11.6977,
      "step": 35456
    },
    {
      "epoch": 1.9307753568438464,
      "grad_norm": 0.6243482940799704,
      "learning_rate": 6.274397721937941e-07,
      "loss": 11.8624,
      "step": 35457
    },
    {
      "epoch": 1.9308298108404294,
      "grad_norm": 0.5242654353878011,
      "learning_rate": 6.264538756167837e-07,
      "loss": 11.8154,
      "step": 35458
    },
    {
      "epoch": 1.9308842648370124,
      "grad_norm": 0.5354684200038378,
      "learning_rate": 6.25468751785141e-07,
      "loss": 11.8179,
      "step": 35459
    },
    {
      "epoch": 1.9309387188335954,
      "grad_norm": 0.5434240838919755,
      "learning_rate": 6.244844007065265e-07,
      "loss": 11.8165,
      "step": 35460
    },
    {
      "epoch": 1.9309931728301784,
      "grad_norm": 0.5358392178486855,
      "learning_rate": 6.235008223885785e-07,
      "loss": 11.8687,
      "step": 35461
    },
    {
      "epoch": 1.9310476268267616,
      "grad_norm": 0.5295326054658354,
      "learning_rate": 6.225180168389578e-07,
      "loss": 11.6926,
      "step": 35462
    },
    {
      "epoch": 1.9311020808233446,
      "grad_norm": 0.5109574104897429,
      "learning_rate": 6.215359840652912e-07,
      "loss": 11.5464,
      "step": 35463
    },
    {
      "epoch": 1.9311565348199276,
      "grad_norm": 0.5126465925223631,
      "learning_rate": 6.205547240752396e-07,
      "loss": 11.9344,
      "step": 35464
    },
    {
      "epoch": 1.9312109888165105,
      "grad_norm": 0.562043269587614,
      "learning_rate": 6.195742368764191e-07,
      "loss": 11.8007,
      "step": 35465
    },
    {
      "epoch": 1.9312654428130935,
      "grad_norm": 0.516122612809263,
      "learning_rate": 6.185945224764456e-07,
      "loss": 11.8307,
      "step": 35466
    },
    {
      "epoch": 1.9313198968096765,
      "grad_norm": 0.5150862949424221,
      "learning_rate": 6.176155808829575e-07,
      "loss": 11.7832,
      "step": 35467
    },
    {
      "epoch": 1.9313743508062595,
      "grad_norm": 0.5312854379425083,
      "learning_rate": 6.16637412103549e-07,
      "loss": 11.5806,
      "step": 35468
    },
    {
      "epoch": 1.9314288048028425,
      "grad_norm": 0.5217652543439314,
      "learning_rate": 6.156600161458359e-07,
      "loss": 11.6881,
      "step": 35469
    },
    {
      "epoch": 1.9314832587994255,
      "grad_norm": 0.5782885527472418,
      "learning_rate": 6.146833930174234e-07,
      "loss": 11.8026,
      "step": 35470
    },
    {
      "epoch": 1.9315377127960085,
      "grad_norm": 0.5043587481644304,
      "learning_rate": 6.137075427258943e-07,
      "loss": 11.7683,
      "step": 35471
    },
    {
      "epoch": 1.9315921667925915,
      "grad_norm": 0.5348028428426579,
      "learning_rate": 6.127324652788424e-07,
      "loss": 11.6308,
      "step": 35472
    },
    {
      "epoch": 1.9316466207891745,
      "grad_norm": 0.6041838035066207,
      "learning_rate": 6.117581606838507e-07,
      "loss": 11.7065,
      "step": 35473
    },
    {
      "epoch": 1.9317010747857575,
      "grad_norm": 0.544692518559051,
      "learning_rate": 6.10784628948502e-07,
      "loss": 11.7577,
      "step": 35474
    },
    {
      "epoch": 1.9317555287823405,
      "grad_norm": 0.5529879675505308,
      "learning_rate": 6.098118700803568e-07,
      "loss": 11.6863,
      "step": 35475
    },
    {
      "epoch": 1.9318099827789235,
      "grad_norm": 0.5296035195851783,
      "learning_rate": 6.088398840869758e-07,
      "loss": 11.8179,
      "step": 35476
    },
    {
      "epoch": 1.9318644367755065,
      "grad_norm": 0.7115354645674317,
      "learning_rate": 6.078686709759307e-07,
      "loss": 11.6767,
      "step": 35477
    },
    {
      "epoch": 1.9319188907720894,
      "grad_norm": 0.5158612277346752,
      "learning_rate": 6.068982307547599e-07,
      "loss": 11.8101,
      "step": 35478
    },
    {
      "epoch": 1.9319733447686727,
      "grad_norm": 0.550608074542295,
      "learning_rate": 6.059285634310241e-07,
      "loss": 11.7904,
      "step": 35479
    },
    {
      "epoch": 1.9320277987652557,
      "grad_norm": 0.5920005190755275,
      "learning_rate": 6.049596690122506e-07,
      "loss": 11.7993,
      "step": 35480
    },
    {
      "epoch": 1.9320822527618386,
      "grad_norm": 0.6001616860296559,
      "learning_rate": 6.039915475059777e-07,
      "loss": 11.7818,
      "step": 35481
    },
    {
      "epoch": 1.9321367067584216,
      "grad_norm": 0.5419926409527603,
      "learning_rate": 6.030241989197438e-07,
      "loss": 11.8246,
      "step": 35482
    },
    {
      "epoch": 1.9321911607550046,
      "grad_norm": 0.516740599127782,
      "learning_rate": 6.020576232610542e-07,
      "loss": 11.812,
      "step": 35483
    },
    {
      "epoch": 1.9322456147515876,
      "grad_norm": 0.5638874430702876,
      "learning_rate": 6.010918205374361e-07,
      "loss": 11.7153,
      "step": 35484
    },
    {
      "epoch": 1.9323000687481708,
      "grad_norm": 0.5943364813106105,
      "learning_rate": 6.001267907564057e-07,
      "loss": 11.6287,
      "step": 35485
    },
    {
      "epoch": 1.9323545227447538,
      "grad_norm": 0.5856328089669531,
      "learning_rate": 5.991625339254458e-07,
      "loss": 11.7378,
      "step": 35486
    },
    {
      "epoch": 1.9324089767413368,
      "grad_norm": 0.5527893443743865,
      "learning_rate": 5.981990500520729e-07,
      "loss": 11.8416,
      "step": 35487
    },
    {
      "epoch": 1.9324634307379198,
      "grad_norm": 0.5339137887451216,
      "learning_rate": 5.972363391437696e-07,
      "loss": 11.8466,
      "step": 35488
    },
    {
      "epoch": 1.9325178847345028,
      "grad_norm": 0.6167140816443636,
      "learning_rate": 5.962744012080413e-07,
      "loss": 11.7049,
      "step": 35489
    },
    {
      "epoch": 1.9325723387310858,
      "grad_norm": 0.5891799387008039,
      "learning_rate": 5.953132362523372e-07,
      "loss": 11.6747,
      "step": 35490
    },
    {
      "epoch": 1.9326267927276688,
      "grad_norm": 0.5361965009771246,
      "learning_rate": 5.943528442841517e-07,
      "loss": 11.7759,
      "step": 35491
    },
    {
      "epoch": 1.9326812467242518,
      "grad_norm": 0.5822303092732038,
      "learning_rate": 5.933932253109454e-07,
      "loss": 11.8003,
      "step": 35492
    },
    {
      "epoch": 1.9327357007208348,
      "grad_norm": 0.5725928998089853,
      "learning_rate": 5.924343793401898e-07,
      "loss": 11.9443,
      "step": 35493
    },
    {
      "epoch": 1.9327901547174178,
      "grad_norm": 0.5570129172893064,
      "learning_rate": 5.914763063793349e-07,
      "loss": 11.7837,
      "step": 35494
    },
    {
      "epoch": 1.9328446087140008,
      "grad_norm": 0.49833430172502596,
      "learning_rate": 5.905190064358301e-07,
      "loss": 11.7331,
      "step": 35495
    },
    {
      "epoch": 1.9328990627105838,
      "grad_norm": 0.48798094462713315,
      "learning_rate": 5.89562479517125e-07,
      "loss": 11.7916,
      "step": 35496
    },
    {
      "epoch": 1.9329535167071668,
      "grad_norm": 0.5102213066152852,
      "learning_rate": 5.886067256306472e-07,
      "loss": 11.8337,
      "step": 35497
    },
    {
      "epoch": 1.9330079707037497,
      "grad_norm": 0.5262706629664062,
      "learning_rate": 5.876517447838347e-07,
      "loss": 11.7168,
      "step": 35498
    },
    {
      "epoch": 1.9330624247003327,
      "grad_norm": 0.5274663062535434,
      "learning_rate": 5.866975369841155e-07,
      "loss": 11.7886,
      "step": 35499
    },
    {
      "epoch": 1.9331168786969157,
      "grad_norm": 0.5142626794514161,
      "learning_rate": 5.857441022389054e-07,
      "loss": 11.7051,
      "step": 35500
    },
    {
      "epoch": 1.9331713326934987,
      "grad_norm": 0.5462579250009353,
      "learning_rate": 5.84791440555621e-07,
      "loss": 11.8119,
      "step": 35501
    },
    {
      "epoch": 1.933225786690082,
      "grad_norm": 0.5515835747959236,
      "learning_rate": 5.838395519416784e-07,
      "loss": 11.8155,
      "step": 35502
    },
    {
      "epoch": 1.933280240686665,
      "grad_norm": 0.5028460130939895,
      "learning_rate": 5.828884364044607e-07,
      "loss": 11.8183,
      "step": 35503
    },
    {
      "epoch": 1.933334694683248,
      "grad_norm": 0.5402563121058426,
      "learning_rate": 5.819380939513841e-07,
      "loss": 11.8034,
      "step": 35504
    },
    {
      "epoch": 1.933389148679831,
      "grad_norm": 0.504151490597803,
      "learning_rate": 5.809885245898206e-07,
      "loss": 11.7673,
      "step": 35505
    },
    {
      "epoch": 1.933443602676414,
      "grad_norm": 0.5367612822078753,
      "learning_rate": 5.800397283271752e-07,
      "loss": 11.8184,
      "step": 35506
    },
    {
      "epoch": 1.933498056672997,
      "grad_norm": 0.6087798429679145,
      "learning_rate": 5.790917051707978e-07,
      "loss": 11.7996,
      "step": 35507
    },
    {
      "epoch": 1.9335525106695801,
      "grad_norm": 0.5058173259283866,
      "learning_rate": 5.781444551280823e-07,
      "loss": 11.7636,
      "step": 35508
    },
    {
      "epoch": 1.933606964666163,
      "grad_norm": 0.5104911424120063,
      "learning_rate": 5.771979782063897e-07,
      "loss": 11.7917,
      "step": 35509
    },
    {
      "epoch": 1.933661418662746,
      "grad_norm": 0.5269449614424476,
      "learning_rate": 5.762522744130805e-07,
      "loss": 11.8466,
      "step": 35510
    },
    {
      "epoch": 1.933715872659329,
      "grad_norm": 0.6238794507286992,
      "learning_rate": 5.753073437555046e-07,
      "loss": 11.8674,
      "step": 35511
    },
    {
      "epoch": 1.933770326655912,
      "grad_norm": 0.5058354461907706,
      "learning_rate": 5.743631862410115e-07,
      "loss": 11.7186,
      "step": 35512
    },
    {
      "epoch": 1.933824780652495,
      "grad_norm": 0.6166115157126538,
      "learning_rate": 5.73419801876951e-07,
      "loss": 11.8859,
      "step": 35513
    },
    {
      "epoch": 1.933879234649078,
      "grad_norm": 0.5258356103363334,
      "learning_rate": 5.724771906706505e-07,
      "loss": 11.7401,
      "step": 35514
    },
    {
      "epoch": 1.933933688645661,
      "grad_norm": 0.5825308617439735,
      "learning_rate": 5.715353526294376e-07,
      "loss": 11.8494,
      "step": 35515
    },
    {
      "epoch": 1.933988142642244,
      "grad_norm": 0.5293163193035932,
      "learning_rate": 5.705942877606396e-07,
      "loss": 11.7866,
      "step": 35516
    },
    {
      "epoch": 1.934042596638827,
      "grad_norm": 0.5160769179071342,
      "learning_rate": 5.69653996071573e-07,
      "loss": 11.5809,
      "step": 35517
    },
    {
      "epoch": 1.93409705063541,
      "grad_norm": 0.49040082299889726,
      "learning_rate": 5.687144775695541e-07,
      "loss": 11.7248,
      "step": 35518
    },
    {
      "epoch": 1.934151504631993,
      "grad_norm": 0.5418900609016448,
      "learning_rate": 5.677757322618881e-07,
      "loss": 11.6884,
      "step": 35519
    },
    {
      "epoch": 1.934205958628576,
      "grad_norm": 0.4995072799842643,
      "learning_rate": 5.668377601558694e-07,
      "loss": 11.7995,
      "step": 35520
    },
    {
      "epoch": 1.934260412625159,
      "grad_norm": 0.5595781752915366,
      "learning_rate": 5.659005612587919e-07,
      "loss": 11.9312,
      "step": 35521
    },
    {
      "epoch": 1.934314866621742,
      "grad_norm": 0.5210870773252599,
      "learning_rate": 5.649641355779501e-07,
      "loss": 11.6057,
      "step": 35522
    },
    {
      "epoch": 1.934369320618325,
      "grad_norm": 0.5302416990401119,
      "learning_rate": 5.640284831206266e-07,
      "loss": 11.8015,
      "step": 35523
    },
    {
      "epoch": 1.934423774614908,
      "grad_norm": 0.5602177390948307,
      "learning_rate": 5.630936038940826e-07,
      "loss": 11.7848,
      "step": 35524
    },
    {
      "epoch": 1.934478228611491,
      "grad_norm": 0.674124727988565,
      "learning_rate": 5.621594979056122e-07,
      "loss": 11.8302,
      "step": 35525
    },
    {
      "epoch": 1.9345326826080742,
      "grad_norm": 0.525792940730806,
      "learning_rate": 5.612261651624651e-07,
      "loss": 11.7443,
      "step": 35526
    },
    {
      "epoch": 1.9345871366046572,
      "grad_norm": 0.542087456639,
      "learning_rate": 5.60293605671891e-07,
      "loss": 11.7502,
      "step": 35527
    },
    {
      "epoch": 1.9346415906012402,
      "grad_norm": 0.6032764142708296,
      "learning_rate": 5.593618194411509e-07,
      "loss": 11.8292,
      "step": 35528
    },
    {
      "epoch": 1.9346960445978232,
      "grad_norm": 0.5052367945536791,
      "learning_rate": 5.584308064774834e-07,
      "loss": 11.6832,
      "step": 35529
    },
    {
      "epoch": 1.9347504985944062,
      "grad_norm": 0.5790057680640368,
      "learning_rate": 5.575005667881383e-07,
      "loss": 11.8919,
      "step": 35530
    },
    {
      "epoch": 1.9348049525909894,
      "grad_norm": 0.5269509280830994,
      "learning_rate": 5.565711003803542e-07,
      "loss": 11.7955,
      "step": 35531
    },
    {
      "epoch": 1.9348594065875724,
      "grad_norm": 0.5238391545030523,
      "learning_rate": 5.556424072613365e-07,
      "loss": 11.7945,
      "step": 35532
    },
    {
      "epoch": 1.9349138605841554,
      "grad_norm": 0.5270434170481811,
      "learning_rate": 5.547144874383348e-07,
      "loss": 11.8329,
      "step": 35533
    },
    {
      "epoch": 1.9349683145807384,
      "grad_norm": 0.5729170709511119,
      "learning_rate": 5.537873409185434e-07,
      "loss": 11.6325,
      "step": 35534
    },
    {
      "epoch": 1.9350227685773214,
      "grad_norm": 0.6404820091796056,
      "learning_rate": 5.52860967709179e-07,
      "loss": 11.8465,
      "step": 35535
    },
    {
      "epoch": 1.9350772225739044,
      "grad_norm": 0.572087389665054,
      "learning_rate": 5.519353678174465e-07,
      "loss": 11.5826,
      "step": 35536
    },
    {
      "epoch": 1.9351316765704873,
      "grad_norm": 0.5094313491671534,
      "learning_rate": 5.510105412505406e-07,
      "loss": 11.7929,
      "step": 35537
    },
    {
      "epoch": 1.9351861305670703,
      "grad_norm": 0.5397780977723677,
      "learning_rate": 5.500864880156553e-07,
      "loss": 11.6658,
      "step": 35538
    },
    {
      "epoch": 1.9352405845636533,
      "grad_norm": 0.5748822295368378,
      "learning_rate": 5.491632081199626e-07,
      "loss": 11.6088,
      "step": 35539
    },
    {
      "epoch": 1.9352950385602363,
      "grad_norm": 0.6642626763132676,
      "learning_rate": 5.482407015706681e-07,
      "loss": 11.9099,
      "step": 35540
    },
    {
      "epoch": 1.9353494925568193,
      "grad_norm": 0.49085089642363156,
      "learning_rate": 5.473189683749325e-07,
      "loss": 11.7104,
      "step": 35541
    },
    {
      "epoch": 1.9354039465534023,
      "grad_norm": 0.5444184984950573,
      "learning_rate": 5.463980085399167e-07,
      "loss": 11.6211,
      "step": 35542
    },
    {
      "epoch": 1.9354584005499853,
      "grad_norm": 0.561893596251195,
      "learning_rate": 5.45477822072793e-07,
      "loss": 11.7222,
      "step": 35543
    },
    {
      "epoch": 1.9355128545465683,
      "grad_norm": 0.6352864392320118,
      "learning_rate": 5.44558408980711e-07,
      "loss": 11.7787,
      "step": 35544
    },
    {
      "epoch": 1.9355673085431513,
      "grad_norm": 0.6272580156377827,
      "learning_rate": 5.436397692708206e-07,
      "loss": 11.8069,
      "step": 35545
    },
    {
      "epoch": 1.9356217625397343,
      "grad_norm": 0.6382483803998975,
      "learning_rate": 5.427219029502717e-07,
      "loss": 11.8302,
      "step": 35546
    },
    {
      "epoch": 1.9356762165363173,
      "grad_norm": 0.5368684556254641,
      "learning_rate": 5.418048100261808e-07,
      "loss": 11.8036,
      "step": 35547
    },
    {
      "epoch": 1.9357306705329003,
      "grad_norm": 0.5487146134663398,
      "learning_rate": 5.408884905057088e-07,
      "loss": 11.8738,
      "step": 35548
    },
    {
      "epoch": 1.9357851245294835,
      "grad_norm": 0.554506082062712,
      "learning_rate": 5.399729443959611e-07,
      "loss": 11.8091,
      "step": 35549
    },
    {
      "epoch": 1.9358395785260665,
      "grad_norm": 0.5123244370176998,
      "learning_rate": 5.390581717040766e-07,
      "loss": 11.8091,
      "step": 35550
    },
    {
      "epoch": 1.9358940325226495,
      "grad_norm": 0.5316891974945974,
      "learning_rate": 5.381441724371384e-07,
      "loss": 11.7525,
      "step": 35551
    },
    {
      "epoch": 1.9359484865192325,
      "grad_norm": 0.5424051800823211,
      "learning_rate": 5.37230946602274e-07,
      "loss": 11.7651,
      "step": 35552
    },
    {
      "epoch": 1.9360029405158155,
      "grad_norm": 0.5074854984584203,
      "learning_rate": 5.363184942065891e-07,
      "loss": 11.785,
      "step": 35553
    },
    {
      "epoch": 1.9360573945123984,
      "grad_norm": 0.5125622814230949,
      "learning_rate": 5.354068152571668e-07,
      "loss": 11.773,
      "step": 35554
    },
    {
      "epoch": 1.9361118485089817,
      "grad_norm": 0.5326360895208188,
      "learning_rate": 5.344959097611014e-07,
      "loss": 11.7576,
      "step": 35555
    },
    {
      "epoch": 1.9361663025055647,
      "grad_norm": 0.5538480999750912,
      "learning_rate": 5.33585777725476e-07,
      "loss": 11.7114,
      "step": 35556
    },
    {
      "epoch": 1.9362207565021476,
      "grad_norm": 0.5686752853133247,
      "learning_rate": 5.326764191573741e-07,
      "loss": 11.7576,
      "step": 35557
    },
    {
      "epoch": 1.9362752104987306,
      "grad_norm": 0.5361564625880706,
      "learning_rate": 5.317678340638566e-07,
      "loss": 11.7045,
      "step": 35558
    },
    {
      "epoch": 1.9363296644953136,
      "grad_norm": 0.5772105290727838,
      "learning_rate": 5.308600224519844e-07,
      "loss": 11.7416,
      "step": 35559
    },
    {
      "epoch": 1.9363841184918966,
      "grad_norm": 0.554722701537153,
      "learning_rate": 5.299529843288409e-07,
      "loss": 11.7756,
      "step": 35560
    },
    {
      "epoch": 1.9364385724884796,
      "grad_norm": 0.5684014070116883,
      "learning_rate": 5.290467197014537e-07,
      "loss": 11.8636,
      "step": 35561
    },
    {
      "epoch": 1.9364930264850626,
      "grad_norm": 0.594703750715069,
      "learning_rate": 5.281412285768839e-07,
      "loss": 11.6712,
      "step": 35562
    },
    {
      "epoch": 1.9365474804816456,
      "grad_norm": 0.5671834112957976,
      "learning_rate": 5.272365109621702e-07,
      "loss": 11.7228,
      "step": 35563
    },
    {
      "epoch": 1.9366019344782286,
      "grad_norm": 0.533732680674147,
      "learning_rate": 5.263325668643404e-07,
      "loss": 11.7501,
      "step": 35564
    },
    {
      "epoch": 1.9366563884748116,
      "grad_norm": 0.5474627287160215,
      "learning_rate": 5.254293962904444e-07,
      "loss": 11.8011,
      "step": 35565
    },
    {
      "epoch": 1.9367108424713946,
      "grad_norm": 0.537744675436712,
      "learning_rate": 5.245269992474766e-07,
      "loss": 11.7322,
      "step": 35566
    },
    {
      "epoch": 1.9367652964679776,
      "grad_norm": 0.6856842289761352,
      "learning_rate": 5.236253757424758e-07,
      "loss": 11.8531,
      "step": 35567
    },
    {
      "epoch": 1.9368197504645606,
      "grad_norm": 0.528508075490259,
      "learning_rate": 5.227245257824475e-07,
      "loss": 11.8795,
      "step": 35568
    },
    {
      "epoch": 1.9368742044611436,
      "grad_norm": 0.5458446312322265,
      "learning_rate": 5.218244493743862e-07,
      "loss": 11.7738,
      "step": 35569
    },
    {
      "epoch": 1.9369286584577265,
      "grad_norm": 0.5817752403466625,
      "learning_rate": 5.209251465253196e-07,
      "loss": 11.8707,
      "step": 35570
    },
    {
      "epoch": 1.9369831124543095,
      "grad_norm": 0.5757848563191925,
      "learning_rate": 5.200266172422085e-07,
      "loss": 11.8887,
      "step": 35571
    },
    {
      "epoch": 1.9370375664508928,
      "grad_norm": 0.5366360202542961,
      "learning_rate": 5.191288615320478e-07,
      "loss": 11.7073,
      "step": 35572
    },
    {
      "epoch": 1.9370920204474757,
      "grad_norm": 0.5009043441019304,
      "learning_rate": 5.182318794018315e-07,
      "loss": 11.6977,
      "step": 35573
    },
    {
      "epoch": 1.9371464744440587,
      "grad_norm": 0.5453093343904455,
      "learning_rate": 5.173356708585208e-07,
      "loss": 11.8582,
      "step": 35574
    },
    {
      "epoch": 1.9372009284406417,
      "grad_norm": 0.5530277793269945,
      "learning_rate": 5.164402359090992e-07,
      "loss": 11.8872,
      "step": 35575
    },
    {
      "epoch": 1.9372553824372247,
      "grad_norm": 0.5171475565767923,
      "learning_rate": 5.155455745605276e-07,
      "loss": 11.764,
      "step": 35576
    },
    {
      "epoch": 1.9373098364338077,
      "grad_norm": 0.5567248051119154,
      "learning_rate": 5.146516868197448e-07,
      "loss": 11.7506,
      "step": 35577
    },
    {
      "epoch": 1.937364290430391,
      "grad_norm": 0.5453075674968821,
      "learning_rate": 5.137585726937233e-07,
      "loss": 11.7504,
      "step": 35578
    },
    {
      "epoch": 1.937418744426974,
      "grad_norm": 0.5437802702069786,
      "learning_rate": 5.128662321893906e-07,
      "loss": 11.7658,
      "step": 35579
    },
    {
      "epoch": 1.937473198423557,
      "grad_norm": 0.5910353216614329,
      "learning_rate": 5.11974665313697e-07,
      "loss": 11.8461,
      "step": 35580
    },
    {
      "epoch": 1.93752765242014,
      "grad_norm": 0.5412227303444469,
      "learning_rate": 5.1108387207357e-07,
      "loss": 11.7498,
      "step": 35581
    },
    {
      "epoch": 1.937582106416723,
      "grad_norm": 0.5293252489419342,
      "learning_rate": 5.101938524759486e-07,
      "loss": 11.7091,
      "step": 35582
    },
    {
      "epoch": 1.937636560413306,
      "grad_norm": 0.5686496400691846,
      "learning_rate": 5.093046065277385e-07,
      "loss": 11.8361,
      "step": 35583
    },
    {
      "epoch": 1.937691014409889,
      "grad_norm": 0.6288173121877694,
      "learning_rate": 5.084161342358562e-07,
      "loss": 11.7634,
      "step": 35584
    },
    {
      "epoch": 1.9377454684064719,
      "grad_norm": 0.5355093804957634,
      "learning_rate": 5.075284356072185e-07,
      "loss": 11.798,
      "step": 35585
    },
    {
      "epoch": 1.9377999224030549,
      "grad_norm": 0.5980710019276938,
      "learning_rate": 5.06641510648731e-07,
      "loss": 11.7094,
      "step": 35586
    },
    {
      "epoch": 1.9378543763996379,
      "grad_norm": 0.5638489993134836,
      "learning_rate": 5.05755359367277e-07,
      "loss": 11.6973,
      "step": 35587
    },
    {
      "epoch": 1.9379088303962209,
      "grad_norm": 0.5412255550250967,
      "learning_rate": 5.048699817697511e-07,
      "loss": 11.8987,
      "step": 35588
    },
    {
      "epoch": 1.9379632843928039,
      "grad_norm": 0.53510416101725,
      "learning_rate": 5.039853778630477e-07,
      "loss": 11.7264,
      "step": 35589
    },
    {
      "epoch": 1.9380177383893868,
      "grad_norm": 0.5512286243043325,
      "learning_rate": 5.03101547654039e-07,
      "loss": 11.6982,
      "step": 35590
    },
    {
      "epoch": 1.9380721923859698,
      "grad_norm": 0.5852851572606463,
      "learning_rate": 5.022184911495864e-07,
      "loss": 11.6774,
      "step": 35591
    },
    {
      "epoch": 1.9381266463825528,
      "grad_norm": 0.5660109327879069,
      "learning_rate": 5.013362083565843e-07,
      "loss": 11.816,
      "step": 35592
    },
    {
      "epoch": 1.9381811003791358,
      "grad_norm": 0.5389142622214665,
      "learning_rate": 5.004546992818715e-07,
      "loss": 11.8273,
      "step": 35593
    },
    {
      "epoch": 1.9382355543757188,
      "grad_norm": 0.5466833615710256,
      "learning_rate": 4.995739639323094e-07,
      "loss": 11.7037,
      "step": 35594
    },
    {
      "epoch": 1.9382900083723018,
      "grad_norm": 0.5101710844844005,
      "learning_rate": 4.98694002314748e-07,
      "loss": 11.7182,
      "step": 35595
    },
    {
      "epoch": 1.938344462368885,
      "grad_norm": 0.5598999360532281,
      "learning_rate": 4.978148144360262e-07,
      "loss": 11.8324,
      "step": 35596
    },
    {
      "epoch": 1.938398916365468,
      "grad_norm": 0.5141143040876567,
      "learning_rate": 4.969364003029941e-07,
      "loss": 11.7442,
      "step": 35597
    },
    {
      "epoch": 1.938453370362051,
      "grad_norm": 0.5282292840073158,
      "learning_rate": 4.960587599224575e-07,
      "loss": 11.8029,
      "step": 35598
    },
    {
      "epoch": 1.938507824358634,
      "grad_norm": 0.5559970232317345,
      "learning_rate": 4.951818933012553e-07,
      "loss": 11.6902,
      "step": 35599
    },
    {
      "epoch": 1.938562278355217,
      "grad_norm": 0.5585131601873601,
      "learning_rate": 4.943058004462042e-07,
      "loss": 11.8164,
      "step": 35600
    },
    {
      "epoch": 1.9386167323518002,
      "grad_norm": 0.5361690102528642,
      "learning_rate": 4.934304813641211e-07,
      "loss": 11.7702,
      "step": 35601
    },
    {
      "epoch": 1.9386711863483832,
      "grad_norm": 0.5611180396196582,
      "learning_rate": 4.925559360618226e-07,
      "loss": 11.6667,
      "step": 35602
    },
    {
      "epoch": 1.9387256403449662,
      "grad_norm": 0.6096788073552318,
      "learning_rate": 4.916821645460812e-07,
      "loss": 11.8067,
      "step": 35603
    },
    {
      "epoch": 1.9387800943415492,
      "grad_norm": 0.606285916864132,
      "learning_rate": 4.908091668237136e-07,
      "loss": 11.7109,
      "step": 35604
    },
    {
      "epoch": 1.9388345483381322,
      "grad_norm": 0.5485993629123017,
      "learning_rate": 4.899369429014922e-07,
      "loss": 11.8023,
      "step": 35605
    },
    {
      "epoch": 1.9388890023347152,
      "grad_norm": 0.5112756488971831,
      "learning_rate": 4.890654927862226e-07,
      "loss": 11.8408,
      "step": 35606
    },
    {
      "epoch": 1.9389434563312982,
      "grad_norm": 0.5109369905202668,
      "learning_rate": 4.881948164846661e-07,
      "loss": 11.8175,
      "step": 35607
    },
    {
      "epoch": 1.9389979103278812,
      "grad_norm": 0.5256283769630259,
      "learning_rate": 4.87324914003584e-07,
      "loss": 11.7667,
      "step": 35608
    },
    {
      "epoch": 1.9390523643244642,
      "grad_norm": 0.5116411016231788,
      "learning_rate": 4.864557853497597e-07,
      "loss": 11.681,
      "step": 35609
    },
    {
      "epoch": 1.9391068183210471,
      "grad_norm": 0.5106795253498314,
      "learning_rate": 4.855874305299435e-07,
      "loss": 11.7391,
      "step": 35610
    },
    {
      "epoch": 1.9391612723176301,
      "grad_norm": 0.5515760517609177,
      "learning_rate": 4.847198495508853e-07,
      "loss": 11.8701,
      "step": 35611
    },
    {
      "epoch": 1.9392157263142131,
      "grad_norm": 0.5935175150195036,
      "learning_rate": 4.838530424193355e-07,
      "loss": 11.7416,
      "step": 35612
    },
    {
      "epoch": 1.9392701803107961,
      "grad_norm": 0.5456600415296303,
      "learning_rate": 4.829870091420219e-07,
      "loss": 11.7238,
      "step": 35613
    },
    {
      "epoch": 1.9393246343073791,
      "grad_norm": 0.5820267786212897,
      "learning_rate": 4.821217497257058e-07,
      "loss": 11.8029,
      "step": 35614
    },
    {
      "epoch": 1.939379088303962,
      "grad_norm": 0.5972214761178529,
      "learning_rate": 4.812572641770929e-07,
      "loss": 11.8013,
      "step": 35615
    },
    {
      "epoch": 1.939433542300545,
      "grad_norm": 0.5705273882659945,
      "learning_rate": 4.803935525029224e-07,
      "loss": 11.7862,
      "step": 35616
    },
    {
      "epoch": 1.939487996297128,
      "grad_norm": 0.6152412570596103,
      "learning_rate": 4.795306147098999e-07,
      "loss": 11.8496,
      "step": 35617
    },
    {
      "epoch": 1.939542450293711,
      "grad_norm": 0.6127777368977805,
      "learning_rate": 4.786684508047201e-07,
      "loss": 11.8153,
      "step": 35618
    },
    {
      "epoch": 1.9395969042902943,
      "grad_norm": 0.5332224816342236,
      "learning_rate": 4.778070607941221e-07,
      "loss": 11.8921,
      "step": 35619
    },
    {
      "epoch": 1.9396513582868773,
      "grad_norm": 0.560218249044104,
      "learning_rate": 4.769464446847782e-07,
      "loss": 11.7938,
      "step": 35620
    },
    {
      "epoch": 1.9397058122834603,
      "grad_norm": 0.5371161705150863,
      "learning_rate": 4.760866024833943e-07,
      "loss": 11.795,
      "step": 35621
    },
    {
      "epoch": 1.9397602662800433,
      "grad_norm": 0.5914869001523443,
      "learning_rate": 4.752275341966428e-07,
      "loss": 11.7295,
      "step": 35622
    },
    {
      "epoch": 1.9398147202766263,
      "grad_norm": 0.5783624654681653,
      "learning_rate": 4.7436923983120717e-07,
      "loss": 11.9032,
      "step": 35623
    },
    {
      "epoch": 1.9398691742732093,
      "grad_norm": 0.5788289930007964,
      "learning_rate": 4.735117193937821e-07,
      "loss": 11.8802,
      "step": 35624
    },
    {
      "epoch": 1.9399236282697925,
      "grad_norm": 0.5369306486402379,
      "learning_rate": 4.726549728910179e-07,
      "loss": 11.8194,
      "step": 35625
    },
    {
      "epoch": 1.9399780822663755,
      "grad_norm": 0.5421521257155484,
      "learning_rate": 4.717990003295758e-07,
      "loss": 11.8647,
      "step": 35626
    },
    {
      "epoch": 1.9400325362629585,
      "grad_norm": 0.5615799146355253,
      "learning_rate": 4.709438017161172e-07,
      "loss": 11.8358,
      "step": 35627
    },
    {
      "epoch": 1.9400869902595415,
      "grad_norm": 0.5786992811693126,
      "learning_rate": 4.700893770572812e-07,
      "loss": 11.7863,
      "step": 35628
    },
    {
      "epoch": 1.9401414442561244,
      "grad_norm": 0.5857360801939799,
      "learning_rate": 4.6923572635974023e-07,
      "loss": 11.7307,
      "step": 35629
    },
    {
      "epoch": 1.9401958982527074,
      "grad_norm": 0.5265721571586228,
      "learning_rate": 4.6838284963010016e-07,
      "loss": 11.7108,
      "step": 35630
    },
    {
      "epoch": 1.9402503522492904,
      "grad_norm": 0.5645426037271813,
      "learning_rate": 4.675307468750112e-07,
      "loss": 11.7081,
      "step": 35631
    },
    {
      "epoch": 1.9403048062458734,
      "grad_norm": 0.5340102986397449,
      "learning_rate": 4.6667941810109026e-07,
      "loss": 11.704,
      "step": 35632
    },
    {
      "epoch": 1.9403592602424564,
      "grad_norm": 0.5791237343761372,
      "learning_rate": 4.6582886331496543e-07,
      "loss": 11.7368,
      "step": 35633
    },
    {
      "epoch": 1.9404137142390394,
      "grad_norm": 0.5726002459202131,
      "learning_rate": 4.649790825232425e-07,
      "loss": 11.8715,
      "step": 35634
    },
    {
      "epoch": 1.9404681682356224,
      "grad_norm": 0.5064762737642653,
      "learning_rate": 4.641300757325273e-07,
      "loss": 11.8122,
      "step": 35635
    },
    {
      "epoch": 1.9405226222322054,
      "grad_norm": 0.5245745992275329,
      "learning_rate": 4.632818429494479e-07,
      "loss": 11.7848,
      "step": 35636
    },
    {
      "epoch": 1.9405770762287884,
      "grad_norm": 0.5583594530276754,
      "learning_rate": 4.6243438418057674e-07,
      "loss": 11.8176,
      "step": 35637
    },
    {
      "epoch": 1.9406315302253714,
      "grad_norm": 0.5467704679271195,
      "learning_rate": 4.6158769943249747e-07,
      "loss": 11.828,
      "step": 35638
    },
    {
      "epoch": 1.9406859842219544,
      "grad_norm": 0.6377707657926502,
      "learning_rate": 4.6074178871181595e-07,
      "loss": 11.9018,
      "step": 35639
    },
    {
      "epoch": 1.9407404382185374,
      "grad_norm": 0.5554824544815146,
      "learning_rate": 4.598966520250936e-07,
      "loss": 11.6895,
      "step": 35640
    },
    {
      "epoch": 1.9407948922151204,
      "grad_norm": 0.5584007560908515,
      "learning_rate": 4.590522893789029e-07,
      "loss": 11.8051,
      "step": 35641
    },
    {
      "epoch": 1.9408493462117036,
      "grad_norm": 0.5427642025318676,
      "learning_rate": 4.5820870077982747e-07,
      "loss": 11.7008,
      "step": 35642
    },
    {
      "epoch": 1.9409038002082866,
      "grad_norm": 0.5466973752396468,
      "learning_rate": 4.5736588623440655e-07,
      "loss": 11.8237,
      "step": 35643
    },
    {
      "epoch": 1.9409582542048696,
      "grad_norm": 0.53261653244681,
      "learning_rate": 4.565238457492016e-07,
      "loss": 11.847,
      "step": 35644
    },
    {
      "epoch": 1.9410127082014526,
      "grad_norm": 0.5893298575811762,
      "learning_rate": 4.5568257933075175e-07,
      "loss": 11.8655,
      "step": 35645
    },
    {
      "epoch": 1.9410671621980355,
      "grad_norm": 0.529436593129781,
      "learning_rate": 4.5484208698562957e-07,
      "loss": 11.7339,
      "step": 35646
    },
    {
      "epoch": 1.9411216161946185,
      "grad_norm": 0.4989824159338248,
      "learning_rate": 4.5400236872032987e-07,
      "loss": 11.7142,
      "step": 35647
    },
    {
      "epoch": 1.9411760701912018,
      "grad_norm": 0.5639858010167177,
      "learning_rate": 4.5316342454141403e-07,
      "loss": 11.7917,
      "step": 35648
    },
    {
      "epoch": 1.9412305241877847,
      "grad_norm": 0.5489765442045642,
      "learning_rate": 4.5232525445538796e-07,
      "loss": 11.6759,
      "step": 35649
    },
    {
      "epoch": 1.9412849781843677,
      "grad_norm": 0.552480917792173,
      "learning_rate": 4.514878584687687e-07,
      "loss": 11.5201,
      "step": 35650
    },
    {
      "epoch": 1.9413394321809507,
      "grad_norm": 0.5364813325782707,
      "learning_rate": 4.506512365880844e-07,
      "loss": 11.6857,
      "step": 35651
    },
    {
      "epoch": 1.9413938861775337,
      "grad_norm": 0.522938519881032,
      "learning_rate": 4.498153888198298e-07,
      "loss": 11.7345,
      "step": 35652
    },
    {
      "epoch": 1.9414483401741167,
      "grad_norm": 0.5362017031750769,
      "learning_rate": 4.4898031517049965e-07,
      "loss": 11.8485,
      "step": 35653
    },
    {
      "epoch": 1.9415027941706997,
      "grad_norm": 0.5341845330736563,
      "learning_rate": 4.4814601564659996e-07,
      "loss": 11.7259,
      "step": 35654
    },
    {
      "epoch": 1.9415572481672827,
      "grad_norm": 0.5555783654732127,
      "learning_rate": 4.473124902546033e-07,
      "loss": 11.8794,
      "step": 35655
    },
    {
      "epoch": 1.9416117021638657,
      "grad_norm": 0.5586152100740498,
      "learning_rate": 4.464797390010045e-07,
      "loss": 11.6914,
      "step": 35656
    },
    {
      "epoch": 1.9416661561604487,
      "grad_norm": 0.5447660271687738,
      "learning_rate": 4.456477618922761e-07,
      "loss": 11.7789,
      "step": 35657
    },
    {
      "epoch": 1.9417206101570317,
      "grad_norm": 0.622647630281201,
      "learning_rate": 4.448165589348796e-07,
      "loss": 11.7603,
      "step": 35658
    },
    {
      "epoch": 1.9417750641536147,
      "grad_norm": 0.5431923100601692,
      "learning_rate": 4.4398613013528766e-07,
      "loss": 11.792,
      "step": 35659
    },
    {
      "epoch": 1.9418295181501977,
      "grad_norm": 0.5790240480772548,
      "learning_rate": 4.4315647549996176e-07,
      "loss": 11.9162,
      "step": 35660
    },
    {
      "epoch": 1.9418839721467807,
      "grad_norm": 0.5117777821406759,
      "learning_rate": 4.4232759503534115e-07,
      "loss": 11.7656,
      "step": 35661
    },
    {
      "epoch": 1.9419384261433636,
      "grad_norm": 0.5479628306654931,
      "learning_rate": 4.414994887478763e-07,
      "loss": 11.622,
      "step": 35662
    },
    {
      "epoch": 1.9419928801399466,
      "grad_norm": 0.6233033102570642,
      "learning_rate": 4.4067215664400643e-07,
      "loss": 11.9512,
      "step": 35663
    },
    {
      "epoch": 1.9420473341365296,
      "grad_norm": 0.5564504797952488,
      "learning_rate": 4.3984559873017086e-07,
      "loss": 11.673,
      "step": 35664
    },
    {
      "epoch": 1.9421017881331128,
      "grad_norm": 0.5448089265380534,
      "learning_rate": 4.3901981501278664e-07,
      "loss": 11.9291,
      "step": 35665
    },
    {
      "epoch": 1.9421562421296958,
      "grad_norm": 0.48524318686328877,
      "learning_rate": 4.3819480549828205e-07,
      "loss": 11.723,
      "step": 35666
    },
    {
      "epoch": 1.9422106961262788,
      "grad_norm": 0.6561161077775467,
      "learning_rate": 4.3737057019307413e-07,
      "loss": 11.7131,
      "step": 35667
    },
    {
      "epoch": 1.9422651501228618,
      "grad_norm": 0.652101449136463,
      "learning_rate": 4.3654710910356886e-07,
      "loss": 11.9336,
      "step": 35668
    },
    {
      "epoch": 1.9423196041194448,
      "grad_norm": 0.4927048110208113,
      "learning_rate": 4.3572442223617225e-07,
      "loss": 11.6496,
      "step": 35669
    },
    {
      "epoch": 1.9423740581160278,
      "grad_norm": 0.5433086032537734,
      "learning_rate": 4.349025095972792e-07,
      "loss": 11.7397,
      "step": 35670
    },
    {
      "epoch": 1.942428512112611,
      "grad_norm": 0.5282950663203201,
      "learning_rate": 4.340813711932734e-07,
      "loss": 11.8879,
      "step": 35671
    },
    {
      "epoch": 1.942482966109194,
      "grad_norm": 0.47735063570741465,
      "learning_rate": 4.3326100703054983e-07,
      "loss": 11.827,
      "step": 35672
    },
    {
      "epoch": 1.942537420105777,
      "grad_norm": 0.5427272413004708,
      "learning_rate": 4.3244141711549223e-07,
      "loss": 11.7919,
      "step": 35673
    },
    {
      "epoch": 1.94259187410236,
      "grad_norm": 0.5197529372982094,
      "learning_rate": 4.316226014544622e-07,
      "loss": 11.7244,
      "step": 35674
    },
    {
      "epoch": 1.942646328098943,
      "grad_norm": 0.5386760180800727,
      "learning_rate": 4.3080456005383243e-07,
      "loss": 11.7665,
      "step": 35675
    },
    {
      "epoch": 1.942700782095526,
      "grad_norm": 0.6072065150359209,
      "learning_rate": 4.2998729291997553e-07,
      "loss": 11.8564,
      "step": 35676
    },
    {
      "epoch": 1.942755236092109,
      "grad_norm": 0.5637670787877125,
      "learning_rate": 4.2917080005921985e-07,
      "loss": 11.7804,
      "step": 35677
    },
    {
      "epoch": 1.942809690088692,
      "grad_norm": 0.565105528113536,
      "learning_rate": 4.28355081477938e-07,
      "loss": 11.8302,
      "step": 35678
    },
    {
      "epoch": 1.942864144085275,
      "grad_norm": 0.5648751695002734,
      "learning_rate": 4.2754013718245836e-07,
      "loss": 11.6565,
      "step": 35679
    },
    {
      "epoch": 1.942918598081858,
      "grad_norm": 0.5920240674745916,
      "learning_rate": 4.267259671791424e-07,
      "loss": 11.7315,
      "step": 35680
    },
    {
      "epoch": 1.942973052078441,
      "grad_norm": 0.5077946424213308,
      "learning_rate": 4.259125714742851e-07,
      "loss": 11.78,
      "step": 35681
    },
    {
      "epoch": 1.943027506075024,
      "grad_norm": 0.545072177508665,
      "learning_rate": 4.25099950074237e-07,
      "loss": 11.7516,
      "step": 35682
    },
    {
      "epoch": 1.943081960071607,
      "grad_norm": 0.5208080262270302,
      "learning_rate": 4.242881029853041e-07,
      "loss": 11.7015,
      "step": 35683
    },
    {
      "epoch": 1.94313641406819,
      "grad_norm": 0.48932973275562786,
      "learning_rate": 4.234770302138147e-07,
      "loss": 11.6227,
      "step": 35684
    },
    {
      "epoch": 1.943190868064773,
      "grad_norm": 0.5030720565581431,
      "learning_rate": 4.2266673176606376e-07,
      "loss": 11.9005,
      "step": 35685
    },
    {
      "epoch": 1.943245322061356,
      "grad_norm": 0.5441757426810855,
      "learning_rate": 4.218572076483573e-07,
      "loss": 11.8475,
      "step": 35686
    },
    {
      "epoch": 1.943299776057939,
      "grad_norm": 0.5935279381739246,
      "learning_rate": 4.210484578669904e-07,
      "loss": 11.826,
      "step": 35687
    },
    {
      "epoch": 1.943354230054522,
      "grad_norm": 0.5280797930627416,
      "learning_rate": 4.202404824282469e-07,
      "loss": 11.7909,
      "step": 35688
    },
    {
      "epoch": 1.9434086840511051,
      "grad_norm": 0.5701949166003447,
      "learning_rate": 4.1943328133841056e-07,
      "loss": 11.782,
      "step": 35689
    },
    {
      "epoch": 1.943463138047688,
      "grad_norm": 0.6112172870600644,
      "learning_rate": 4.1862685460376525e-07,
      "loss": 11.8887,
      "step": 35690
    },
    {
      "epoch": 1.943517592044271,
      "grad_norm": 0.5271576083084188,
      "learning_rate": 4.1782120223057273e-07,
      "loss": 11.697,
      "step": 35691
    },
    {
      "epoch": 1.943572046040854,
      "grad_norm": 0.676886655921369,
      "learning_rate": 4.1701632422510575e-07,
      "loss": 11.8064,
      "step": 35692
    },
    {
      "epoch": 1.943626500037437,
      "grad_norm": 0.58317774283948,
      "learning_rate": 4.1621222059361477e-07,
      "loss": 11.7945,
      "step": 35693
    },
    {
      "epoch": 1.94368095403402,
      "grad_norm": 0.5382340106794387,
      "learning_rate": 4.154088913423615e-07,
      "loss": 11.6911,
      "step": 35694
    },
    {
      "epoch": 1.9437354080306033,
      "grad_norm": 0.5199886284674525,
      "learning_rate": 4.146063364775854e-07,
      "loss": 11.7783,
      "step": 35695
    },
    {
      "epoch": 1.9437898620271863,
      "grad_norm": 0.5757448211057639,
      "learning_rate": 4.13804556005537e-07,
      "loss": 11.8136,
      "step": 35696
    },
    {
      "epoch": 1.9438443160237693,
      "grad_norm": 0.6443045661948109,
      "learning_rate": 4.1300354993244475e-07,
      "loss": 11.8579,
      "step": 35697
    },
    {
      "epoch": 1.9438987700203523,
      "grad_norm": 0.5467342450522417,
      "learning_rate": 4.122033182645368e-07,
      "loss": 11.7118,
      "step": 35698
    },
    {
      "epoch": 1.9439532240169353,
      "grad_norm": 0.5542789289943424,
      "learning_rate": 4.1140386100803063e-07,
      "loss": 11.8412,
      "step": 35699
    },
    {
      "epoch": 1.9440076780135183,
      "grad_norm": 0.580831701424614,
      "learning_rate": 4.1060517816916553e-07,
      "loss": 11.8274,
      "step": 35700
    },
    {
      "epoch": 1.9440621320101013,
      "grad_norm": 0.583477141632455,
      "learning_rate": 4.098072697541144e-07,
      "loss": 11.8083,
      "step": 35701
    },
    {
      "epoch": 1.9441165860066842,
      "grad_norm": 0.561768365617742,
      "learning_rate": 4.090101357691167e-07,
      "loss": 11.7626,
      "step": 35702
    },
    {
      "epoch": 1.9441710400032672,
      "grad_norm": 0.5796074077575896,
      "learning_rate": 4.082137762203564e-07,
      "loss": 11.7128,
      "step": 35703
    },
    {
      "epoch": 1.9442254939998502,
      "grad_norm": 0.5634412532462876,
      "learning_rate": 4.0741819111402846e-07,
      "loss": 11.777,
      "step": 35704
    },
    {
      "epoch": 1.9442799479964332,
      "grad_norm": 0.5646150578537383,
      "learning_rate": 4.0662338045630576e-07,
      "loss": 11.7639,
      "step": 35705
    },
    {
      "epoch": 1.9443344019930162,
      "grad_norm": 0.5284997169497452,
      "learning_rate": 4.058293442533945e-07,
      "loss": 11.7069,
      "step": 35706
    },
    {
      "epoch": 1.9443888559895992,
      "grad_norm": 0.5379691627878892,
      "learning_rate": 4.050360825114563e-07,
      "loss": 11.8555,
      "step": 35707
    },
    {
      "epoch": 1.9444433099861822,
      "grad_norm": 0.5539307429360871,
      "learning_rate": 4.04243595236653e-07,
      "loss": 11.7842,
      "step": 35708
    },
    {
      "epoch": 1.9444977639827652,
      "grad_norm": 0.5455320142949022,
      "learning_rate": 4.0345188243515743e-07,
      "loss": 11.7417,
      "step": 35709
    },
    {
      "epoch": 1.9445522179793482,
      "grad_norm": 0.5941572163906921,
      "learning_rate": 4.0266094411312016e-07,
      "loss": 11.7816,
      "step": 35710
    },
    {
      "epoch": 1.9446066719759312,
      "grad_norm": 0.5993469973740057,
      "learning_rate": 4.018707802766919e-07,
      "loss": 11.7944,
      "step": 35711
    },
    {
      "epoch": 1.9446611259725144,
      "grad_norm": 0.5623196863215588,
      "learning_rate": 4.0108139093202323e-07,
      "loss": 11.7284,
      "step": 35712
    },
    {
      "epoch": 1.9447155799690974,
      "grad_norm": 0.5040781196717365,
      "learning_rate": 4.0029277608524265e-07,
      "loss": 11.7071,
      "step": 35713
    },
    {
      "epoch": 1.9447700339656804,
      "grad_norm": 0.524003300164236,
      "learning_rate": 3.995049357425007e-07,
      "loss": 11.7005,
      "step": 35714
    },
    {
      "epoch": 1.9448244879622634,
      "grad_norm": 0.4866673847118549,
      "learning_rate": 3.987178699098926e-07,
      "loss": 11.7628,
      "step": 35715
    },
    {
      "epoch": 1.9448789419588464,
      "grad_norm": 0.5821477240631966,
      "learning_rate": 3.9793157859358e-07,
      "loss": 11.8356,
      "step": 35716
    },
    {
      "epoch": 1.9449333959554294,
      "grad_norm": 0.510507799479634,
      "learning_rate": 3.971460617996359e-07,
      "loss": 11.767,
      "step": 35717
    },
    {
      "epoch": 1.9449878499520126,
      "grad_norm": 0.6280518385953502,
      "learning_rate": 3.9636131953419975e-07,
      "loss": 11.6978,
      "step": 35718
    },
    {
      "epoch": 1.9450423039485956,
      "grad_norm": 0.5285910951440296,
      "learning_rate": 3.9557735180335567e-07,
      "loss": 11.8543,
      "step": 35719
    },
    {
      "epoch": 1.9450967579451786,
      "grad_norm": 0.5458123047948893,
      "learning_rate": 3.9479415861320977e-07,
      "loss": 11.7192,
      "step": 35720
    },
    {
      "epoch": 1.9451512119417615,
      "grad_norm": 0.5328846918990849,
      "learning_rate": 3.9401173996983507e-07,
      "loss": 11.8453,
      "step": 35721
    },
    {
      "epoch": 1.9452056659383445,
      "grad_norm": 0.5329099872425184,
      "learning_rate": 3.932300958793489e-07,
      "loss": 11.7942,
      "step": 35722
    },
    {
      "epoch": 1.9452601199349275,
      "grad_norm": 0.5598684641929673,
      "learning_rate": 3.924492263477908e-07,
      "loss": 11.7646,
      "step": 35723
    },
    {
      "epoch": 1.9453145739315105,
      "grad_norm": 0.559382593285798,
      "learning_rate": 3.9166913138126704e-07,
      "loss": 11.7958,
      "step": 35724
    },
    {
      "epoch": 1.9453690279280935,
      "grad_norm": 0.5545658316275914,
      "learning_rate": 3.908898109858172e-07,
      "loss": 11.7752,
      "step": 35725
    },
    {
      "epoch": 1.9454234819246765,
      "grad_norm": 0.5180100870720471,
      "learning_rate": 3.901112651675143e-07,
      "loss": 11.6259,
      "step": 35726
    },
    {
      "epoch": 1.9454779359212595,
      "grad_norm": 0.545965859639784,
      "learning_rate": 3.8933349393242003e-07,
      "loss": 11.8456,
      "step": 35727
    },
    {
      "epoch": 1.9455323899178425,
      "grad_norm": 0.6642197509422173,
      "learning_rate": 3.88556497286563e-07,
      "loss": 11.721,
      "step": 35728
    },
    {
      "epoch": 1.9455868439144255,
      "grad_norm": 0.4712969524763708,
      "learning_rate": 3.877802752360049e-07,
      "loss": 11.6598,
      "step": 35729
    },
    {
      "epoch": 1.9456412979110085,
      "grad_norm": 0.5494635343391067,
      "learning_rate": 3.8700482778676327e-07,
      "loss": 11.8528,
      "step": 35730
    },
    {
      "epoch": 1.9456957519075915,
      "grad_norm": 0.5428316624175658,
      "learning_rate": 3.8623015494488877e-07,
      "loss": 11.7118,
      "step": 35731
    },
    {
      "epoch": 1.9457502059041745,
      "grad_norm": 0.4863273638061356,
      "learning_rate": 3.854562567163766e-07,
      "loss": 11.6769,
      "step": 35732
    },
    {
      "epoch": 1.9458046599007575,
      "grad_norm": 0.5746758191981314,
      "learning_rate": 3.8468313310727753e-07,
      "loss": 11.8249,
      "step": 35733
    },
    {
      "epoch": 1.9458591138973405,
      "grad_norm": 0.5685567238136227,
      "learning_rate": 3.8391078412357563e-07,
      "loss": 11.7821,
      "step": 35734
    },
    {
      "epoch": 1.9459135678939237,
      "grad_norm": 0.5237326655910691,
      "learning_rate": 3.831392097712994e-07,
      "loss": 11.7451,
      "step": 35735
    },
    {
      "epoch": 1.9459680218905067,
      "grad_norm": 0.6046829238409155,
      "learning_rate": 3.823684100564329e-07,
      "loss": 11.8291,
      "step": 35736
    },
    {
      "epoch": 1.9460224758870897,
      "grad_norm": 0.533224507214216,
      "learning_rate": 3.8159838498498247e-07,
      "loss": 11.7393,
      "step": 35737
    },
    {
      "epoch": 1.9460769298836726,
      "grad_norm": 0.5436070399414449,
      "learning_rate": 3.8082913456292114e-07,
      "loss": 11.8217,
      "step": 35738
    },
    {
      "epoch": 1.9461313838802556,
      "grad_norm": 0.5453353709124602,
      "learning_rate": 3.800606587962441e-07,
      "loss": 11.8112,
      "step": 35739
    },
    {
      "epoch": 1.9461858378768386,
      "grad_norm": 0.5812766822010963,
      "learning_rate": 3.792929576909132e-07,
      "loss": 11.8766,
      "step": 35740
    },
    {
      "epoch": 1.9462402918734218,
      "grad_norm": 0.5537300880662903,
      "learning_rate": 3.7852603125291265e-07,
      "loss": 11.7217,
      "step": 35741
    },
    {
      "epoch": 1.9462947458700048,
      "grad_norm": 0.5276328697901749,
      "learning_rate": 3.7775987948819316e-07,
      "loss": 11.8891,
      "step": 35742
    },
    {
      "epoch": 1.9463491998665878,
      "grad_norm": 0.5260705314244255,
      "learning_rate": 3.769945024027277e-07,
      "loss": 11.6815,
      "step": 35743
    },
    {
      "epoch": 1.9464036538631708,
      "grad_norm": 0.5785938476405321,
      "learning_rate": 3.7622990000245606e-07,
      "loss": 11.7479,
      "step": 35744
    },
    {
      "epoch": 1.9464581078597538,
      "grad_norm": 0.5896680240690657,
      "learning_rate": 3.75466072293329e-07,
      "loss": 11.7817,
      "step": 35745
    },
    {
      "epoch": 1.9465125618563368,
      "grad_norm": 0.6498796511110028,
      "learning_rate": 3.747030192812862e-07,
      "loss": 11.7938,
      "step": 35746
    },
    {
      "epoch": 1.9465670158529198,
      "grad_norm": 0.7237580619834845,
      "learning_rate": 3.7394074097226725e-07,
      "loss": 11.8408,
      "step": 35747
    },
    {
      "epoch": 1.9466214698495028,
      "grad_norm": 0.5563646385313362,
      "learning_rate": 3.7317923737217876e-07,
      "loss": 11.7168,
      "step": 35748
    },
    {
      "epoch": 1.9466759238460858,
      "grad_norm": 0.5168367031565501,
      "learning_rate": 3.7241850848696023e-07,
      "loss": 11.7155,
      "step": 35749
    },
    {
      "epoch": 1.9467303778426688,
      "grad_norm": 0.5130486735454791,
      "learning_rate": 3.7165855432252926e-07,
      "loss": 11.7459,
      "step": 35750
    },
    {
      "epoch": 1.9467848318392518,
      "grad_norm": 0.6244948898279601,
      "learning_rate": 3.708993748847811e-07,
      "loss": 11.7626,
      "step": 35751
    },
    {
      "epoch": 1.9468392858358348,
      "grad_norm": 0.5530923103096124,
      "learning_rate": 3.701409701796332e-07,
      "loss": 11.7108,
      "step": 35752
    },
    {
      "epoch": 1.9468937398324178,
      "grad_norm": 0.5661141389051748,
      "learning_rate": 3.6938334021296985e-07,
      "loss": 11.8574,
      "step": 35753
    },
    {
      "epoch": 1.9469481938290008,
      "grad_norm": 0.5182482736419168,
      "learning_rate": 3.686264849906973e-07,
      "loss": 11.5661,
      "step": 35754
    },
    {
      "epoch": 1.9470026478255837,
      "grad_norm": 0.5639635402331181,
      "learning_rate": 3.678704045186776e-07,
      "loss": 11.5872,
      "step": 35755
    },
    {
      "epoch": 1.9470571018221667,
      "grad_norm": 0.5171289199709851,
      "learning_rate": 3.6711509880282823e-07,
      "loss": 11.7741,
      "step": 35756
    },
    {
      "epoch": 1.9471115558187497,
      "grad_norm": 0.516906244992106,
      "learning_rate": 3.6636056784898896e-07,
      "loss": 11.8151,
      "step": 35757
    },
    {
      "epoch": 1.9471660098153327,
      "grad_norm": 0.5059858006377905,
      "learning_rate": 3.656068116630329e-07,
      "loss": 11.5865,
      "step": 35758
    },
    {
      "epoch": 1.947220463811916,
      "grad_norm": 0.4941656244664837,
      "learning_rate": 3.6485383025084416e-07,
      "loss": 11.7932,
      "step": 35759
    },
    {
      "epoch": 1.947274917808499,
      "grad_norm": 0.6434683748293571,
      "learning_rate": 3.641016236182404e-07,
      "loss": 11.7971,
      "step": 35760
    },
    {
      "epoch": 1.947329371805082,
      "grad_norm": 0.5282323122304959,
      "learning_rate": 3.6335019177110575e-07,
      "loss": 11.7219,
      "step": 35761
    },
    {
      "epoch": 1.947383825801665,
      "grad_norm": 0.502077551476034,
      "learning_rate": 3.6259953471526885e-07,
      "loss": 11.7572,
      "step": 35762
    },
    {
      "epoch": 1.947438279798248,
      "grad_norm": 0.499740545919771,
      "learning_rate": 3.618496524565584e-07,
      "loss": 11.7284,
      "step": 35763
    },
    {
      "epoch": 1.947492733794831,
      "grad_norm": 0.5412275970791999,
      "learning_rate": 3.6110054500081425e-07,
      "loss": 11.8111,
      "step": 35764
    },
    {
      "epoch": 1.9475471877914141,
      "grad_norm": 0.6620992968835064,
      "learning_rate": 3.6035221235387607e-07,
      "loss": 11.8467,
      "step": 35765
    },
    {
      "epoch": 1.947601641787997,
      "grad_norm": 0.5463680273203441,
      "learning_rate": 3.5960465452152815e-07,
      "loss": 11.8238,
      "step": 35766
    },
    {
      "epoch": 1.94765609578458,
      "grad_norm": 0.6230451822896992,
      "learning_rate": 3.588578715096214e-07,
      "loss": 11.7563,
      "step": 35767
    },
    {
      "epoch": 1.947710549781163,
      "grad_norm": 0.5500990160108249,
      "learning_rate": 3.5811186332393997e-07,
      "loss": 11.7255,
      "step": 35768
    },
    {
      "epoch": 1.947765003777746,
      "grad_norm": 0.5274759632381156,
      "learning_rate": 3.5736662997029046e-07,
      "loss": 11.7417,
      "step": 35769
    },
    {
      "epoch": 1.947819457774329,
      "grad_norm": 0.6134745912046852,
      "learning_rate": 3.566221714544682e-07,
      "loss": 11.8862,
      "step": 35770
    },
    {
      "epoch": 1.947873911770912,
      "grad_norm": 0.5942901860692718,
      "learning_rate": 3.558784877822574e-07,
      "loss": 11.7984,
      "step": 35771
    },
    {
      "epoch": 1.947928365767495,
      "grad_norm": 0.5324894676895175,
      "learning_rate": 3.551355789594535e-07,
      "loss": 11.5921,
      "step": 35772
    },
    {
      "epoch": 1.947982819764078,
      "grad_norm": 0.5208748411718979,
      "learning_rate": 3.543934449918185e-07,
      "loss": 11.7375,
      "step": 35773
    },
    {
      "epoch": 1.948037273760661,
      "grad_norm": 0.5884086872750396,
      "learning_rate": 3.5365208588513666e-07,
      "loss": 11.8309,
      "step": 35774
    },
    {
      "epoch": 1.948091727757244,
      "grad_norm": 0.5344587420224608,
      "learning_rate": 3.5291150164517006e-07,
      "loss": 11.7614,
      "step": 35775
    },
    {
      "epoch": 1.948146181753827,
      "grad_norm": 0.5807156630632645,
      "learning_rate": 3.5217169227765857e-07,
      "loss": 11.8448,
      "step": 35776
    },
    {
      "epoch": 1.94820063575041,
      "grad_norm": 0.6509098455259031,
      "learning_rate": 3.514326577883864e-07,
      "loss": 11.9187,
      "step": 35777
    },
    {
      "epoch": 1.948255089746993,
      "grad_norm": 0.6640034731038533,
      "learning_rate": 3.5069439818308235e-07,
      "loss": 11.8736,
      "step": 35778
    },
    {
      "epoch": 1.948309543743576,
      "grad_norm": 0.5197435171253312,
      "learning_rate": 3.499569134674863e-07,
      "loss": 11.7151,
      "step": 35779
    },
    {
      "epoch": 1.948363997740159,
      "grad_norm": 0.5382578311627094,
      "learning_rate": 3.49220203647338e-07,
      "loss": 11.711,
      "step": 35780
    },
    {
      "epoch": 1.948418451736742,
      "grad_norm": 0.5033336945565409,
      "learning_rate": 3.4848426872836626e-07,
      "loss": 11.8021,
      "step": 35781
    },
    {
      "epoch": 1.9484729057333252,
      "grad_norm": 0.5469208153599247,
      "learning_rate": 3.477491087162887e-07,
      "loss": 11.668,
      "step": 35782
    },
    {
      "epoch": 1.9485273597299082,
      "grad_norm": 0.5779186555582223,
      "learning_rate": 3.4701472361682307e-07,
      "loss": 11.726,
      "step": 35783
    },
    {
      "epoch": 1.9485818137264912,
      "grad_norm": 0.567181339386298,
      "learning_rate": 3.462811134356869e-07,
      "loss": 11.8276,
      "step": 35784
    },
    {
      "epoch": 1.9486362677230742,
      "grad_norm": 0.5817138107091715,
      "learning_rate": 3.455482781785868e-07,
      "loss": 11.7262,
      "step": 35785
    },
    {
      "epoch": 1.9486907217196572,
      "grad_norm": 0.5178025546351565,
      "learning_rate": 3.448162178512071e-07,
      "loss": 11.7248,
      "step": 35786
    },
    {
      "epoch": 1.9487451757162402,
      "grad_norm": 0.6111691085854609,
      "learning_rate": 3.440849324592543e-07,
      "loss": 11.7132,
      "step": 35787
    },
    {
      "epoch": 1.9487996297128234,
      "grad_norm": 0.4830768142712869,
      "learning_rate": 3.4335442200840173e-07,
      "loss": 11.6678,
      "step": 35788
    },
    {
      "epoch": 1.9488540837094064,
      "grad_norm": 0.5205172880075745,
      "learning_rate": 3.4262468650434474e-07,
      "loss": 11.7703,
      "step": 35789
    },
    {
      "epoch": 1.9489085377059894,
      "grad_norm": 0.5361165745380371,
      "learning_rate": 3.4189572595274553e-07,
      "loss": 11.7288,
      "step": 35790
    },
    {
      "epoch": 1.9489629917025724,
      "grad_norm": 0.5943033370828161,
      "learning_rate": 3.4116754035928845e-07,
      "loss": 11.7242,
      "step": 35791
    },
    {
      "epoch": 1.9490174456991554,
      "grad_norm": 0.5666927184201768,
      "learning_rate": 3.4044012972961336e-07,
      "loss": 11.771,
      "step": 35792
    },
    {
      "epoch": 1.9490718996957384,
      "grad_norm": 0.5402614924496174,
      "learning_rate": 3.397134940693936e-07,
      "loss": 11.8572,
      "step": 35793
    },
    {
      "epoch": 1.9491263536923213,
      "grad_norm": 0.5913378435730663,
      "learning_rate": 3.38987633384269e-07,
      "loss": 11.766,
      "step": 35794
    },
    {
      "epoch": 1.9491808076889043,
      "grad_norm": 0.5183274891548123,
      "learning_rate": 3.3826254767990174e-07,
      "loss": 11.8697,
      "step": 35795
    },
    {
      "epoch": 1.9492352616854873,
      "grad_norm": 0.5579197662683869,
      "learning_rate": 3.3753823696190956e-07,
      "loss": 11.7581,
      "step": 35796
    },
    {
      "epoch": 1.9492897156820703,
      "grad_norm": 0.541985234741951,
      "learning_rate": 3.368147012359324e-07,
      "loss": 11.7521,
      "step": 35797
    },
    {
      "epoch": 1.9493441696786533,
      "grad_norm": 0.565821080166775,
      "learning_rate": 3.360919405075991e-07,
      "loss": 11.7639,
      "step": 35798
    },
    {
      "epoch": 1.9493986236752363,
      "grad_norm": 0.590903133530086,
      "learning_rate": 3.353699547825273e-07,
      "loss": 11.6631,
      "step": 35799
    },
    {
      "epoch": 1.9494530776718193,
      "grad_norm": 0.5516842409895033,
      "learning_rate": 3.3464874406634597e-07,
      "loss": 11.8536,
      "step": 35800
    },
    {
      "epoch": 1.9495075316684023,
      "grad_norm": 0.5342780122454649,
      "learning_rate": 3.339283083646283e-07,
      "loss": 11.7583,
      "step": 35801
    },
    {
      "epoch": 1.9495619856649853,
      "grad_norm": 0.4950279828894907,
      "learning_rate": 3.332086476830143e-07,
      "loss": 11.6153,
      "step": 35802
    },
    {
      "epoch": 1.9496164396615683,
      "grad_norm": 0.5697677897162219,
      "learning_rate": 3.324897620270773e-07,
      "loss": 11.7594,
      "step": 35803
    },
    {
      "epoch": 1.9496708936581513,
      "grad_norm": 0.5429610345061743,
      "learning_rate": 3.317716514024127e-07,
      "loss": 11.7725,
      "step": 35804
    },
    {
      "epoch": 1.9497253476547345,
      "grad_norm": 0.5510586216933023,
      "learning_rate": 3.3105431581461623e-07,
      "loss": 11.9017,
      "step": 35805
    },
    {
      "epoch": 1.9497798016513175,
      "grad_norm": 0.5190511466084771,
      "learning_rate": 3.3033775526923883e-07,
      "loss": 11.739,
      "step": 35806
    },
    {
      "epoch": 1.9498342556479005,
      "grad_norm": 0.5364381969782219,
      "learning_rate": 3.296219697718872e-07,
      "loss": 11.8013,
      "step": 35807
    },
    {
      "epoch": 1.9498887096444835,
      "grad_norm": 0.5947973922439259,
      "learning_rate": 3.289069593281013e-07,
      "loss": 11.719,
      "step": 35808
    },
    {
      "epoch": 1.9499431636410665,
      "grad_norm": 0.5521077995128394,
      "learning_rate": 3.2819272394344346e-07,
      "loss": 11.7002,
      "step": 35809
    },
    {
      "epoch": 1.9499976176376494,
      "grad_norm": 0.5764012215332128,
      "learning_rate": 3.274792636234869e-07,
      "loss": 11.8677,
      "step": 35810
    },
    {
      "epoch": 1.9500520716342327,
      "grad_norm": 0.5578297352939146,
      "learning_rate": 3.267665783737606e-07,
      "loss": 11.9374,
      "step": 35811
    },
    {
      "epoch": 1.9501065256308157,
      "grad_norm": 0.5322309769436364,
      "learning_rate": 3.260546681998156e-07,
      "loss": 11.7566,
      "step": 35812
    },
    {
      "epoch": 1.9501609796273986,
      "grad_norm": 0.5433238958891174,
      "learning_rate": 3.253435331071808e-07,
      "loss": 11.7498,
      "step": 35813
    },
    {
      "epoch": 1.9502154336239816,
      "grad_norm": 0.48925847983033494,
      "learning_rate": 3.2463317310138517e-07,
      "loss": 11.7488,
      "step": 35814
    },
    {
      "epoch": 1.9502698876205646,
      "grad_norm": 0.5589126774225989,
      "learning_rate": 3.2392358818796873e-07,
      "loss": 11.8701,
      "step": 35815
    },
    {
      "epoch": 1.9503243416171476,
      "grad_norm": 0.5709214017534532,
      "learning_rate": 3.2321477837242706e-07,
      "loss": 11.7667,
      "step": 35816
    },
    {
      "epoch": 1.9503787956137306,
      "grad_norm": 0.5940625147770527,
      "learning_rate": 3.225067436603002e-07,
      "loss": 11.7525,
      "step": 35817
    },
    {
      "epoch": 1.9504332496103136,
      "grad_norm": 0.4917298206686435,
      "learning_rate": 3.217994840570615e-07,
      "loss": 11.7953,
      "step": 35818
    },
    {
      "epoch": 1.9504877036068966,
      "grad_norm": 0.5999062352893816,
      "learning_rate": 3.210929995682288e-07,
      "loss": 11.7939,
      "step": 35819
    },
    {
      "epoch": 1.9505421576034796,
      "grad_norm": 0.4887759734218869,
      "learning_rate": 3.203872901992977e-07,
      "loss": 11.7573,
      "step": 35820
    },
    {
      "epoch": 1.9505966116000626,
      "grad_norm": 0.5424081437122078,
      "learning_rate": 3.1968235595574156e-07,
      "loss": 11.7486,
      "step": 35821
    },
    {
      "epoch": 1.9506510655966456,
      "grad_norm": 0.5817441787724815,
      "learning_rate": 3.1897819684305606e-07,
      "loss": 11.8727,
      "step": 35822
    },
    {
      "epoch": 1.9507055195932286,
      "grad_norm": 0.5180513380536177,
      "learning_rate": 3.1827481286671454e-07,
      "loss": 11.7364,
      "step": 35823
    },
    {
      "epoch": 1.9507599735898116,
      "grad_norm": 0.49946414602183087,
      "learning_rate": 3.1757220403219043e-07,
      "loss": 11.8093,
      "step": 35824
    },
    {
      "epoch": 1.9508144275863946,
      "grad_norm": 0.5077580998679988,
      "learning_rate": 3.168703703449349e-07,
      "loss": 11.7523,
      "step": 35825
    },
    {
      "epoch": 1.9508688815829776,
      "grad_norm": 0.5229095469697806,
      "learning_rate": 3.161693118104103e-07,
      "loss": 11.7871,
      "step": 35826
    },
    {
      "epoch": 1.9509233355795605,
      "grad_norm": 0.5624545584192372,
      "learning_rate": 3.154690284340789e-07,
      "loss": 11.8014,
      "step": 35827
    },
    {
      "epoch": 1.9509777895761435,
      "grad_norm": 0.6411604832075658,
      "learning_rate": 3.1476952022136964e-07,
      "loss": 11.8469,
      "step": 35828
    },
    {
      "epoch": 1.9510322435727268,
      "grad_norm": 0.5604763411659524,
      "learning_rate": 3.1407078717773377e-07,
      "loss": 11.7794,
      "step": 35829
    },
    {
      "epoch": 1.9510866975693097,
      "grad_norm": 0.518215737932677,
      "learning_rate": 3.1337282930860025e-07,
      "loss": 11.7237,
      "step": 35830
    },
    {
      "epoch": 1.9511411515658927,
      "grad_norm": 0.5534885203746889,
      "learning_rate": 3.1267564661938697e-07,
      "loss": 11.7628,
      "step": 35831
    },
    {
      "epoch": 1.9511956055624757,
      "grad_norm": 0.6007370207747913,
      "learning_rate": 3.119792391155341e-07,
      "loss": 11.8422,
      "step": 35832
    },
    {
      "epoch": 1.9512500595590587,
      "grad_norm": 0.5412034443034113,
      "learning_rate": 3.112836068024483e-07,
      "loss": 11.8291,
      "step": 35833
    },
    {
      "epoch": 1.951304513555642,
      "grad_norm": 0.5023854244170816,
      "learning_rate": 3.1058874968554754e-07,
      "loss": 11.675,
      "step": 35834
    },
    {
      "epoch": 1.951358967552225,
      "grad_norm": 0.5998728340734277,
      "learning_rate": 3.0989466777021634e-07,
      "loss": 11.8802,
      "step": 35835
    },
    {
      "epoch": 1.951413421548808,
      "grad_norm": 0.5233710975280607,
      "learning_rate": 3.0920136106186157e-07,
      "loss": 11.7182,
      "step": 35836
    },
    {
      "epoch": 1.951467875545391,
      "grad_norm": 0.5886865846789391,
      "learning_rate": 3.085088295658789e-07,
      "loss": 11.8063,
      "step": 35837
    },
    {
      "epoch": 1.951522329541974,
      "grad_norm": 0.5820195424925961,
      "learning_rate": 3.0781707328765285e-07,
      "loss": 11.7188,
      "step": 35838
    },
    {
      "epoch": 1.951576783538557,
      "grad_norm": 0.5645678801383038,
      "learning_rate": 3.0712609223255696e-07,
      "loss": 11.7586,
      "step": 35839
    },
    {
      "epoch": 1.95163123753514,
      "grad_norm": 0.6150839369493637,
      "learning_rate": 3.0643588640596467e-07,
      "loss": 11.8566,
      "step": 35840
    },
    {
      "epoch": 1.9516856915317229,
      "grad_norm": 0.5641827692573567,
      "learning_rate": 3.0574645581323834e-07,
      "loss": 11.6754,
      "step": 35841
    },
    {
      "epoch": 1.9517401455283059,
      "grad_norm": 0.629641290932581,
      "learning_rate": 3.050578004597626e-07,
      "loss": 11.6847,
      "step": 35842
    },
    {
      "epoch": 1.9517945995248889,
      "grad_norm": 0.5240048791950138,
      "learning_rate": 3.0436992035086656e-07,
      "loss": 11.8017,
      "step": 35843
    },
    {
      "epoch": 1.9518490535214719,
      "grad_norm": 0.5207401473257164,
      "learning_rate": 3.0368281549191245e-07,
      "loss": 11.6369,
      "step": 35844
    },
    {
      "epoch": 1.9519035075180549,
      "grad_norm": 0.5680741577814936,
      "learning_rate": 3.0299648588822946e-07,
      "loss": 11.8314,
      "step": 35845
    },
    {
      "epoch": 1.9519579615146379,
      "grad_norm": 0.5415864232828316,
      "learning_rate": 3.0231093154518e-07,
      "loss": 11.8047,
      "step": 35846
    },
    {
      "epoch": 1.9520124155112208,
      "grad_norm": 0.5483960674646408,
      "learning_rate": 3.016261524680708e-07,
      "loss": 11.8152,
      "step": 35847
    },
    {
      "epoch": 1.9520668695078038,
      "grad_norm": 0.5239235695206559,
      "learning_rate": 3.009421486622421e-07,
      "loss": 11.8147,
      "step": 35848
    },
    {
      "epoch": 1.9521213235043868,
      "grad_norm": 0.6451641908611242,
      "learning_rate": 3.002589201330008e-07,
      "loss": 11.765,
      "step": 35849
    },
    {
      "epoch": 1.9521757775009698,
      "grad_norm": 0.5243196034409832,
      "learning_rate": 2.995764668856649e-07,
      "loss": 11.8682,
      "step": 35850
    },
    {
      "epoch": 1.9522302314975528,
      "grad_norm": 0.562819873362613,
      "learning_rate": 2.988947889255522e-07,
      "loss": 11.8393,
      "step": 35851
    },
    {
      "epoch": 1.952284685494136,
      "grad_norm": 0.5234802610463309,
      "learning_rate": 2.9821388625794756e-07,
      "loss": 11.7525,
      "step": 35852
    },
    {
      "epoch": 1.952339139490719,
      "grad_norm": 0.5223475941794641,
      "learning_rate": 2.975337588881466e-07,
      "loss": 11.7858,
      "step": 35853
    },
    {
      "epoch": 1.952393593487302,
      "grad_norm": 0.5188269331431005,
      "learning_rate": 2.9685440682144515e-07,
      "loss": 11.8606,
      "step": 35854
    },
    {
      "epoch": 1.952448047483885,
      "grad_norm": 0.5103479857297905,
      "learning_rate": 2.961758300631279e-07,
      "loss": 11.6701,
      "step": 35855
    },
    {
      "epoch": 1.952502501480468,
      "grad_norm": 0.5454324703986294,
      "learning_rate": 2.954980286184683e-07,
      "loss": 11.8446,
      "step": 35856
    },
    {
      "epoch": 1.952556955477051,
      "grad_norm": 0.5400136085861927,
      "learning_rate": 2.9482100249274005e-07,
      "loss": 11.7596,
      "step": 35857
    },
    {
      "epoch": 1.9526114094736342,
      "grad_norm": 0.5282519667995272,
      "learning_rate": 2.941447516911944e-07,
      "loss": 11.8004,
      "step": 35858
    },
    {
      "epoch": 1.9526658634702172,
      "grad_norm": 0.5545893747726753,
      "learning_rate": 2.9346927621910494e-07,
      "loss": 11.553,
      "step": 35859
    },
    {
      "epoch": 1.9527203174668002,
      "grad_norm": 0.5537199705247947,
      "learning_rate": 2.927945760817119e-07,
      "loss": 11.7516,
      "step": 35860
    },
    {
      "epoch": 1.9527747714633832,
      "grad_norm": 0.5631832176629185,
      "learning_rate": 2.921206512842778e-07,
      "loss": 11.8641,
      "step": 35861
    },
    {
      "epoch": 1.9528292254599662,
      "grad_norm": 0.5360749020353921,
      "learning_rate": 2.9144750183203175e-07,
      "loss": 11.624,
      "step": 35862
    },
    {
      "epoch": 1.9528836794565492,
      "grad_norm": 0.5334187994037874,
      "learning_rate": 2.90775127730214e-07,
      "loss": 11.9112,
      "step": 35863
    },
    {
      "epoch": 1.9529381334531322,
      "grad_norm": 0.5445344224613425,
      "learning_rate": 2.901035289840537e-07,
      "loss": 11.8131,
      "step": 35864
    },
    {
      "epoch": 1.9529925874497152,
      "grad_norm": 0.5487306338296999,
      "learning_rate": 2.894327055987578e-07,
      "loss": 11.6989,
      "step": 35865
    },
    {
      "epoch": 1.9530470414462981,
      "grad_norm": 0.4904048312942676,
      "learning_rate": 2.887626575795666e-07,
      "loss": 11.6938,
      "step": 35866
    },
    {
      "epoch": 1.9531014954428811,
      "grad_norm": 0.5223437638123665,
      "learning_rate": 2.880933849316647e-07,
      "loss": 11.7829,
      "step": 35867
    },
    {
      "epoch": 1.9531559494394641,
      "grad_norm": 0.5653797954884884,
      "learning_rate": 2.874248876602814e-07,
      "loss": 11.6763,
      "step": 35868
    },
    {
      "epoch": 1.9532104034360471,
      "grad_norm": 0.5670962305740367,
      "learning_rate": 2.867571657706014e-07,
      "loss": 11.6804,
      "step": 35869
    },
    {
      "epoch": 1.9532648574326301,
      "grad_norm": 0.5361234131465358,
      "learning_rate": 2.8609021926782053e-07,
      "loss": 11.8379,
      "step": 35870
    },
    {
      "epoch": 1.9533193114292131,
      "grad_norm": 0.5745048326946705,
      "learning_rate": 2.8542404815712354e-07,
      "loss": 11.722,
      "step": 35871
    },
    {
      "epoch": 1.953373765425796,
      "grad_norm": 0.5558435101449316,
      "learning_rate": 2.847586524436951e-07,
      "loss": 11.8554,
      "step": 35872
    },
    {
      "epoch": 1.953428219422379,
      "grad_norm": 0.5234960261363741,
      "learning_rate": 2.8409403213269794e-07,
      "loss": 11.692,
      "step": 35873
    },
    {
      "epoch": 1.953482673418962,
      "grad_norm": 0.5513330073098779,
      "learning_rate": 2.8343018722930546e-07,
      "loss": 11.875,
      "step": 35874
    },
    {
      "epoch": 1.9535371274155453,
      "grad_norm": 0.5461841410829567,
      "learning_rate": 2.8276711773869146e-07,
      "loss": 11.871,
      "step": 35875
    },
    {
      "epoch": 1.9535915814121283,
      "grad_norm": 0.5587509588084834,
      "learning_rate": 2.821048236659962e-07,
      "loss": 11.7609,
      "step": 35876
    },
    {
      "epoch": 1.9536460354087113,
      "grad_norm": 0.5564115773370264,
      "learning_rate": 2.814433050163823e-07,
      "loss": 11.8441,
      "step": 35877
    },
    {
      "epoch": 1.9537004894052943,
      "grad_norm": 0.5415376775011018,
      "learning_rate": 2.8078256179498995e-07,
      "loss": 11.7623,
      "step": 35878
    },
    {
      "epoch": 1.9537549434018773,
      "grad_norm": 0.5500602009273343,
      "learning_rate": 2.801225940069485e-07,
      "loss": 11.8713,
      "step": 35879
    },
    {
      "epoch": 1.9538093973984603,
      "grad_norm": 0.6269348301275441,
      "learning_rate": 2.7946340165739825e-07,
      "loss": 11.8026,
      "step": 35880
    },
    {
      "epoch": 1.9538638513950435,
      "grad_norm": 0.5750371656321503,
      "learning_rate": 2.788049847514684e-07,
      "loss": 11.8651,
      "step": 35881
    },
    {
      "epoch": 1.9539183053916265,
      "grad_norm": 0.5823462549100343,
      "learning_rate": 2.7814734329426606e-07,
      "loss": 11.7562,
      "step": 35882
    },
    {
      "epoch": 1.9539727593882095,
      "grad_norm": 0.6173978633885092,
      "learning_rate": 2.7749047729092036e-07,
      "loss": 11.7826,
      "step": 35883
    },
    {
      "epoch": 1.9540272133847925,
      "grad_norm": 0.5880719517248111,
      "learning_rate": 2.768343867465273e-07,
      "loss": 11.8653,
      "step": 35884
    },
    {
      "epoch": 1.9540816673813755,
      "grad_norm": 0.5758611817619522,
      "learning_rate": 2.7617907166619384e-07,
      "loss": 11.722,
      "step": 35885
    },
    {
      "epoch": 1.9541361213779584,
      "grad_norm": 0.5832879258558195,
      "learning_rate": 2.7552453205501595e-07,
      "loss": 11.7369,
      "step": 35886
    },
    {
      "epoch": 1.9541905753745414,
      "grad_norm": 0.5287119368124059,
      "learning_rate": 2.7487076791808954e-07,
      "loss": 11.8529,
      "step": 35887
    },
    {
      "epoch": 1.9542450293711244,
      "grad_norm": 0.5157165752512857,
      "learning_rate": 2.7421777926048833e-07,
      "loss": 11.8183,
      "step": 35888
    },
    {
      "epoch": 1.9542994833677074,
      "grad_norm": 0.4865903755511943,
      "learning_rate": 2.735655660872971e-07,
      "loss": 11.7977,
      "step": 35889
    },
    {
      "epoch": 1.9543539373642904,
      "grad_norm": 0.540523073617327,
      "learning_rate": 2.7291412840357855e-07,
      "loss": 11.7901,
      "step": 35890
    },
    {
      "epoch": 1.9544083913608734,
      "grad_norm": 0.5487497903139256,
      "learning_rate": 2.7226346621440633e-07,
      "loss": 11.6646,
      "step": 35891
    },
    {
      "epoch": 1.9544628453574564,
      "grad_norm": 0.5891033440618993,
      "learning_rate": 2.7161357952483204e-07,
      "loss": 11.7166,
      "step": 35892
    },
    {
      "epoch": 1.9545172993540394,
      "grad_norm": 0.5863427885907998,
      "learning_rate": 2.709644683399182e-07,
      "loss": 11.8414,
      "step": 35893
    },
    {
      "epoch": 1.9545717533506224,
      "grad_norm": 0.5819569928162447,
      "learning_rate": 2.7031613266471636e-07,
      "loss": 11.7669,
      "step": 35894
    },
    {
      "epoch": 1.9546262073472054,
      "grad_norm": 0.5619965174082716,
      "learning_rate": 2.696685725042558e-07,
      "loss": 11.7991,
      "step": 35895
    },
    {
      "epoch": 1.9546806613437884,
      "grad_norm": 0.563162459752338,
      "learning_rate": 2.690217878635659e-07,
      "loss": 11.8986,
      "step": 35896
    },
    {
      "epoch": 1.9547351153403714,
      "grad_norm": 0.5085379508253828,
      "learning_rate": 2.6837577874769817e-07,
      "loss": 11.7301,
      "step": 35897
    },
    {
      "epoch": 1.9547895693369544,
      "grad_norm": 0.5520974546880524,
      "learning_rate": 2.6773054516167074e-07,
      "loss": 11.7723,
      "step": 35898
    },
    {
      "epoch": 1.9548440233335376,
      "grad_norm": 0.5518164002357201,
      "learning_rate": 2.670860871104908e-07,
      "loss": 11.7063,
      "step": 35899
    },
    {
      "epoch": 1.9548984773301206,
      "grad_norm": 0.6009453695692543,
      "learning_rate": 2.664424045991765e-07,
      "loss": 11.789,
      "step": 35900
    },
    {
      "epoch": 1.9549529313267036,
      "grad_norm": 0.5692161719869828,
      "learning_rate": 2.657994976327238e-07,
      "loss": 11.7154,
      "step": 35901
    },
    {
      "epoch": 1.9550073853232866,
      "grad_norm": 0.6287841057724544,
      "learning_rate": 2.6515736621615106e-07,
      "loss": 11.7546,
      "step": 35902
    },
    {
      "epoch": 1.9550618393198695,
      "grad_norm": 0.5583070790923353,
      "learning_rate": 2.6451601035443196e-07,
      "loss": 11.8544,
      "step": 35903
    },
    {
      "epoch": 1.9551162933164528,
      "grad_norm": 0.507130766576949,
      "learning_rate": 2.638754300525625e-07,
      "loss": 11.6692,
      "step": 35904
    },
    {
      "epoch": 1.9551707473130358,
      "grad_norm": 0.6557667819767078,
      "learning_rate": 2.632356253155277e-07,
      "loss": 11.7887,
      "step": 35905
    },
    {
      "epoch": 1.9552252013096187,
      "grad_norm": 0.5321084664675033,
      "learning_rate": 2.6259659614829013e-07,
      "loss": 11.6028,
      "step": 35906
    },
    {
      "epoch": 1.9552796553062017,
      "grad_norm": 0.5084910653859216,
      "learning_rate": 2.6195834255583474e-07,
      "loss": 11.7913,
      "step": 35907
    },
    {
      "epoch": 1.9553341093027847,
      "grad_norm": 0.5389367268452145,
      "learning_rate": 2.613208645431242e-07,
      "loss": 11.8112,
      "step": 35908
    },
    {
      "epoch": 1.9553885632993677,
      "grad_norm": 0.5369992545014218,
      "learning_rate": 2.6068416211509906e-07,
      "loss": 11.8744,
      "step": 35909
    },
    {
      "epoch": 1.9554430172959507,
      "grad_norm": 0.5785829195847547,
      "learning_rate": 2.6004823527672196e-07,
      "loss": 11.79,
      "step": 35910
    },
    {
      "epoch": 1.9554974712925337,
      "grad_norm": 0.6076617351802577,
      "learning_rate": 2.594130840329334e-07,
      "loss": 11.7503,
      "step": 35911
    },
    {
      "epoch": 1.9555519252891167,
      "grad_norm": 0.5304492212541593,
      "learning_rate": 2.587787083886739e-07,
      "loss": 11.7637,
      "step": 35912
    },
    {
      "epoch": 1.9556063792856997,
      "grad_norm": 0.5155447374843087,
      "learning_rate": 2.5814510834888393e-07,
      "loss": 11.7147,
      "step": 35913
    },
    {
      "epoch": 1.9556608332822827,
      "grad_norm": 0.5760552732945751,
      "learning_rate": 2.575122839184818e-07,
      "loss": 11.8351,
      "step": 35914
    },
    {
      "epoch": 1.9557152872788657,
      "grad_norm": 0.6113632142587743,
      "learning_rate": 2.568802351023858e-07,
      "loss": 11.8025,
      "step": 35915
    },
    {
      "epoch": 1.9557697412754487,
      "grad_norm": 0.5589133512186288,
      "learning_rate": 2.5624896190552526e-07,
      "loss": 11.79,
      "step": 35916
    },
    {
      "epoch": 1.9558241952720317,
      "grad_norm": 0.5391823032797487,
      "learning_rate": 2.5561846433279633e-07,
      "loss": 11.7124,
      "step": 35917
    },
    {
      "epoch": 1.9558786492686147,
      "grad_norm": 0.5836982543128165,
      "learning_rate": 2.549887423891062e-07,
      "loss": 11.6308,
      "step": 35918
    },
    {
      "epoch": 1.9559331032651976,
      "grad_norm": 0.6049656136525037,
      "learning_rate": 2.543597960793509e-07,
      "loss": 11.846,
      "step": 35919
    },
    {
      "epoch": 1.9559875572617806,
      "grad_norm": 0.5440007353911324,
      "learning_rate": 2.5373162540841547e-07,
      "loss": 11.6707,
      "step": 35920
    },
    {
      "epoch": 1.9560420112583636,
      "grad_norm": 0.5871422008088131,
      "learning_rate": 2.531042303811959e-07,
      "loss": 11.6263,
      "step": 35921
    },
    {
      "epoch": 1.9560964652549468,
      "grad_norm": 0.5726808866474816,
      "learning_rate": 2.5247761100256615e-07,
      "loss": 11.7663,
      "step": 35922
    },
    {
      "epoch": 1.9561509192515298,
      "grad_norm": 0.5255182273916505,
      "learning_rate": 2.51851767277389e-07,
      "loss": 11.695,
      "step": 35923
    },
    {
      "epoch": 1.9562053732481128,
      "grad_norm": 0.7316910928448936,
      "learning_rate": 2.512266992105494e-07,
      "loss": 11.8428,
      "step": 35924
    },
    {
      "epoch": 1.9562598272446958,
      "grad_norm": 0.5466159138712406,
      "learning_rate": 2.5060240680689907e-07,
      "loss": 11.7252,
      "step": 35925
    },
    {
      "epoch": 1.9563142812412788,
      "grad_norm": 0.5120301825262785,
      "learning_rate": 2.499788900712896e-07,
      "loss": 11.8882,
      "step": 35926
    },
    {
      "epoch": 1.9563687352378618,
      "grad_norm": 0.5531173590274187,
      "learning_rate": 2.493561490085727e-07,
      "loss": 11.7761,
      "step": 35927
    },
    {
      "epoch": 1.956423189234445,
      "grad_norm": 0.6035921506793719,
      "learning_rate": 2.48734183623589e-07,
      "loss": 11.7222,
      "step": 35928
    },
    {
      "epoch": 1.956477643231028,
      "grad_norm": 0.5275334665423085,
      "learning_rate": 2.4811299392117906e-07,
      "loss": 11.8672,
      "step": 35929
    },
    {
      "epoch": 1.956532097227611,
      "grad_norm": 0.5079229798744881,
      "learning_rate": 2.4749257990617224e-07,
      "loss": 11.7034,
      "step": 35930
    },
    {
      "epoch": 1.956586551224194,
      "grad_norm": 0.5632949809556068,
      "learning_rate": 2.46872941583387e-07,
      "loss": 11.7346,
      "step": 35931
    },
    {
      "epoch": 1.956641005220777,
      "grad_norm": 0.538298404356519,
      "learning_rate": 2.4625407895765285e-07,
      "loss": 11.7376,
      "step": 35932
    },
    {
      "epoch": 1.95669545921736,
      "grad_norm": 0.5876919351619007,
      "learning_rate": 2.4563599203376584e-07,
      "loss": 11.6859,
      "step": 35933
    },
    {
      "epoch": 1.956749913213943,
      "grad_norm": 0.556354689695551,
      "learning_rate": 2.450186808165555e-07,
      "loss": 11.7755,
      "step": 35934
    },
    {
      "epoch": 1.956804367210526,
      "grad_norm": 0.5260272536635366,
      "learning_rate": 2.4440214531079586e-07,
      "loss": 11.7413,
      "step": 35935
    },
    {
      "epoch": 1.956858821207109,
      "grad_norm": 0.5286884270683888,
      "learning_rate": 2.4378638552129404e-07,
      "loss": 11.8724,
      "step": 35936
    },
    {
      "epoch": 1.956913275203692,
      "grad_norm": 0.5494134903867526,
      "learning_rate": 2.4317140145284634e-07,
      "loss": 11.7719,
      "step": 35937
    },
    {
      "epoch": 1.956967729200275,
      "grad_norm": 0.5925043697881837,
      "learning_rate": 2.425571931102266e-07,
      "loss": 11.6398,
      "step": 35938
    },
    {
      "epoch": 1.957022183196858,
      "grad_norm": 0.5943765468098076,
      "learning_rate": 2.419437604982089e-07,
      "loss": 11.9939,
      "step": 35939
    },
    {
      "epoch": 1.957076637193441,
      "grad_norm": 0.5535055704856124,
      "learning_rate": 2.4133110362156705e-07,
      "loss": 11.8611,
      "step": 35940
    },
    {
      "epoch": 1.957131091190024,
      "grad_norm": 0.5040609159798455,
      "learning_rate": 2.4071922248506404e-07,
      "loss": 11.6314,
      "step": 35941
    },
    {
      "epoch": 1.957185545186607,
      "grad_norm": 0.5475128848344688,
      "learning_rate": 2.401081170934627e-07,
      "loss": 11.7481,
      "step": 35942
    },
    {
      "epoch": 1.95723999918319,
      "grad_norm": 0.4946371589393999,
      "learning_rate": 2.394977874515036e-07,
      "loss": 11.7724,
      "step": 35943
    },
    {
      "epoch": 1.957294453179773,
      "grad_norm": 0.5377785856912255,
      "learning_rate": 2.3888823356393864e-07,
      "loss": 11.7273,
      "step": 35944
    },
    {
      "epoch": 1.9573489071763561,
      "grad_norm": 0.5307371097819616,
      "learning_rate": 2.3827945543551943e-07,
      "loss": 11.5581,
      "step": 35945
    },
    {
      "epoch": 1.9574033611729391,
      "grad_norm": 0.5651747604016063,
      "learning_rate": 2.376714530709534e-07,
      "loss": 11.8479,
      "step": 35946
    },
    {
      "epoch": 1.957457815169522,
      "grad_norm": 0.5283998511994507,
      "learning_rate": 2.3706422647499226e-07,
      "loss": 11.7311,
      "step": 35947
    },
    {
      "epoch": 1.957512269166105,
      "grad_norm": 0.5407110930783279,
      "learning_rate": 2.3645777565235448e-07,
      "loss": 11.639,
      "step": 35948
    },
    {
      "epoch": 1.957566723162688,
      "grad_norm": 0.5639319181617664,
      "learning_rate": 2.3585210060774742e-07,
      "loss": 11.7106,
      "step": 35949
    },
    {
      "epoch": 1.957621177159271,
      "grad_norm": 0.505142217294847,
      "learning_rate": 2.3524720134587842e-07,
      "loss": 11.6646,
      "step": 35950
    },
    {
      "epoch": 1.9576756311558543,
      "grad_norm": 0.5746314781996184,
      "learning_rate": 2.3464307787146588e-07,
      "loss": 11.8535,
      "step": 35951
    },
    {
      "epoch": 1.9577300851524373,
      "grad_norm": 0.5174578720733319,
      "learning_rate": 2.34039730189195e-07,
      "loss": 11.7564,
      "step": 35952
    },
    {
      "epoch": 1.9577845391490203,
      "grad_norm": 0.5251376682898412,
      "learning_rate": 2.33437158303762e-07,
      "loss": 11.7435,
      "step": 35953
    },
    {
      "epoch": 1.9578389931456033,
      "grad_norm": 0.5353396839108149,
      "learning_rate": 2.3283536221986312e-07,
      "loss": 11.7581,
      "step": 35954
    },
    {
      "epoch": 1.9578934471421863,
      "grad_norm": 0.5340541763313547,
      "learning_rate": 2.322343419421502e-07,
      "loss": 11.7559,
      "step": 35955
    },
    {
      "epoch": 1.9579479011387693,
      "grad_norm": 0.5204118864493482,
      "learning_rate": 2.316340974753306e-07,
      "loss": 11.8139,
      "step": 35956
    },
    {
      "epoch": 1.9580023551353523,
      "grad_norm": 0.5972537787744839,
      "learning_rate": 2.3103462882404503e-07,
      "loss": 11.7654,
      "step": 35957
    },
    {
      "epoch": 1.9580568091319352,
      "grad_norm": 0.505470952708302,
      "learning_rate": 2.3043593599296753e-07,
      "loss": 11.7382,
      "step": 35958
    },
    {
      "epoch": 1.9581112631285182,
      "grad_norm": 0.517484750902238,
      "learning_rate": 2.2983801898674996e-07,
      "loss": 11.7658,
      "step": 35959
    },
    {
      "epoch": 1.9581657171251012,
      "grad_norm": 0.6277265276756603,
      "learning_rate": 2.2924087781004412e-07,
      "loss": 11.8407,
      "step": 35960
    },
    {
      "epoch": 1.9582201711216842,
      "grad_norm": 0.5175984169503505,
      "learning_rate": 2.2864451246749074e-07,
      "loss": 11.6994,
      "step": 35961
    },
    {
      "epoch": 1.9582746251182672,
      "grad_norm": 0.5292819777409135,
      "learning_rate": 2.280489229637417e-07,
      "loss": 11.7636,
      "step": 35962
    },
    {
      "epoch": 1.9583290791148502,
      "grad_norm": 0.5432526409607965,
      "learning_rate": 2.274541093034044e-07,
      "loss": 11.8115,
      "step": 35963
    },
    {
      "epoch": 1.9583835331114332,
      "grad_norm": 0.5721332638844125,
      "learning_rate": 2.2686007149111955e-07,
      "loss": 11.7229,
      "step": 35964
    },
    {
      "epoch": 1.9584379871080162,
      "grad_norm": 0.5306818464954678,
      "learning_rate": 2.2626680953149459e-07,
      "loss": 11.6406,
      "step": 35965
    },
    {
      "epoch": 1.9584924411045992,
      "grad_norm": 0.5303782181200513,
      "learning_rate": 2.2567432342915916e-07,
      "loss": 11.8844,
      "step": 35966
    },
    {
      "epoch": 1.9585468951011822,
      "grad_norm": 0.5247701926028948,
      "learning_rate": 2.250826131887096e-07,
      "loss": 11.8682,
      "step": 35967
    },
    {
      "epoch": 1.9586013490977654,
      "grad_norm": 0.4854950431216873,
      "learning_rate": 2.244916788147533e-07,
      "loss": 11.7764,
      "step": 35968
    },
    {
      "epoch": 1.9586558030943484,
      "grad_norm": 0.5879777289575042,
      "learning_rate": 2.239015203118866e-07,
      "loss": 11.8634,
      "step": 35969
    },
    {
      "epoch": 1.9587102570909314,
      "grad_norm": 0.5213985472647493,
      "learning_rate": 2.2331213768468363e-07,
      "loss": 11.8124,
      "step": 35970
    },
    {
      "epoch": 1.9587647110875144,
      "grad_norm": 0.5562151313233744,
      "learning_rate": 2.2272353093774067e-07,
      "loss": 11.8171,
      "step": 35971
    },
    {
      "epoch": 1.9588191650840974,
      "grad_norm": 0.5000176824145144,
      "learning_rate": 2.2213570007563188e-07,
      "loss": 11.7175,
      "step": 35972
    },
    {
      "epoch": 1.9588736190806804,
      "grad_norm": 0.5177771222958579,
      "learning_rate": 2.215486451029314e-07,
      "loss": 11.821,
      "step": 35973
    },
    {
      "epoch": 1.9589280730772636,
      "grad_norm": 0.5053136477219986,
      "learning_rate": 2.2096236602420216e-07,
      "loss": 11.8311,
      "step": 35974
    },
    {
      "epoch": 1.9589825270738466,
      "grad_norm": 0.5547501316099366,
      "learning_rate": 2.2037686284399617e-07,
      "loss": 11.744,
      "step": 35975
    },
    {
      "epoch": 1.9590369810704296,
      "grad_norm": 0.5583075796416789,
      "learning_rate": 2.197921355668875e-07,
      "loss": 11.7493,
      "step": 35976
    },
    {
      "epoch": 1.9590914350670126,
      "grad_norm": 0.4865546407423326,
      "learning_rate": 2.1920818419739475e-07,
      "loss": 11.8093,
      "step": 35977
    },
    {
      "epoch": 1.9591458890635955,
      "grad_norm": 0.5159941282380012,
      "learning_rate": 2.1862500874008097e-07,
      "loss": 11.845,
      "step": 35978
    },
    {
      "epoch": 1.9592003430601785,
      "grad_norm": 0.5738069207992845,
      "learning_rate": 2.1804260919946472e-07,
      "loss": 11.7668,
      "step": 35979
    },
    {
      "epoch": 1.9592547970567615,
      "grad_norm": 0.544092997987775,
      "learning_rate": 2.1746098558008688e-07,
      "loss": 11.8062,
      "step": 35980
    },
    {
      "epoch": 1.9593092510533445,
      "grad_norm": 0.5430659754606114,
      "learning_rate": 2.16880137886466e-07,
      "loss": 11.7779,
      "step": 35981
    },
    {
      "epoch": 1.9593637050499275,
      "grad_norm": 0.5533043904573204,
      "learning_rate": 2.163000661231096e-07,
      "loss": 11.7627,
      "step": 35982
    },
    {
      "epoch": 1.9594181590465105,
      "grad_norm": 0.5491360911113741,
      "learning_rate": 2.157207702945474e-07,
      "loss": 11.755,
      "step": 35983
    },
    {
      "epoch": 1.9594726130430935,
      "grad_norm": 0.5939356269830557,
      "learning_rate": 2.151422504052758e-07,
      "loss": 11.8318,
      "step": 35984
    },
    {
      "epoch": 1.9595270670396765,
      "grad_norm": 0.5346205594886096,
      "learning_rate": 2.145645064597801e-07,
      "loss": 11.7513,
      "step": 35985
    },
    {
      "epoch": 1.9595815210362595,
      "grad_norm": 0.5220698824521048,
      "learning_rate": 2.139875384625789e-07,
      "loss": 11.7206,
      "step": 35986
    },
    {
      "epoch": 1.9596359750328425,
      "grad_norm": 0.5531759666057698,
      "learning_rate": 2.1341134641813532e-07,
      "loss": 11.7065,
      "step": 35987
    },
    {
      "epoch": 1.9596904290294255,
      "grad_norm": 0.5283570502275963,
      "learning_rate": 2.1283593033094572e-07,
      "loss": 11.8188,
      "step": 35988
    },
    {
      "epoch": 1.9597448830260085,
      "grad_norm": 0.5472719714085682,
      "learning_rate": 2.1226129020547324e-07,
      "loss": 11.6957,
      "step": 35989
    },
    {
      "epoch": 1.9597993370225915,
      "grad_norm": 0.5066518263393189,
      "learning_rate": 2.1168742604619206e-07,
      "loss": 11.7311,
      "step": 35990
    },
    {
      "epoch": 1.9598537910191745,
      "grad_norm": 0.5613462019930039,
      "learning_rate": 2.111143378575653e-07,
      "loss": 11.7848,
      "step": 35991
    },
    {
      "epoch": 1.9599082450157577,
      "grad_norm": 0.5879546238432012,
      "learning_rate": 2.105420256440449e-07,
      "loss": 11.8254,
      "step": 35992
    },
    {
      "epoch": 1.9599626990123407,
      "grad_norm": 0.5143005623587931,
      "learning_rate": 2.09970489410094e-07,
      "loss": 11.7918,
      "step": 35993
    },
    {
      "epoch": 1.9600171530089237,
      "grad_norm": 0.5088615201152062,
      "learning_rate": 2.0939972916014238e-07,
      "loss": 11.7431,
      "step": 35994
    },
    {
      "epoch": 1.9600716070055066,
      "grad_norm": 0.6261982315383645,
      "learning_rate": 2.0882974489863093e-07,
      "loss": 11.7243,
      "step": 35995
    },
    {
      "epoch": 1.9601260610020896,
      "grad_norm": 0.5667986517146926,
      "learning_rate": 2.0826053662998946e-07,
      "loss": 11.7451,
      "step": 35996
    },
    {
      "epoch": 1.9601805149986726,
      "grad_norm": 0.5579407408108145,
      "learning_rate": 2.0769210435865883e-07,
      "loss": 11.6502,
      "step": 35997
    },
    {
      "epoch": 1.9602349689952558,
      "grad_norm": 0.5572480299646217,
      "learning_rate": 2.0712444808904662e-07,
      "loss": 11.6107,
      "step": 35998
    },
    {
      "epoch": 1.9602894229918388,
      "grad_norm": 0.498678713069211,
      "learning_rate": 2.0655756782557157e-07,
      "loss": 11.8045,
      "step": 35999
    },
    {
      "epoch": 1.9603438769884218,
      "grad_norm": 0.521042250840242,
      "learning_rate": 2.059914635726301e-07,
      "loss": 11.7027,
      "step": 36000
    },
    {
      "epoch": 1.9603983309850048,
      "grad_norm": 0.5331566584946978,
      "learning_rate": 2.0542613533465204e-07,
      "loss": 11.7363,
      "step": 36001
    },
    {
      "epoch": 1.9604527849815878,
      "grad_norm": 0.5443088404153574,
      "learning_rate": 2.0486158311600057e-07,
      "loss": 11.7402,
      "step": 36002
    },
    {
      "epoch": 1.9605072389781708,
      "grad_norm": 0.5525041906124101,
      "learning_rate": 2.0429780692108325e-07,
      "loss": 11.7649,
      "step": 36003
    },
    {
      "epoch": 1.9605616929747538,
      "grad_norm": 0.6303177546887656,
      "learning_rate": 2.0373480675428546e-07,
      "loss": 11.7865,
      "step": 36004
    },
    {
      "epoch": 1.9606161469713368,
      "grad_norm": 0.5915968681274381,
      "learning_rate": 2.031725826199815e-07,
      "loss": 11.8998,
      "step": 36005
    },
    {
      "epoch": 1.9606706009679198,
      "grad_norm": 0.5011611063775976,
      "learning_rate": 2.0261113452254566e-07,
      "loss": 11.7123,
      "step": 36006
    },
    {
      "epoch": 1.9607250549645028,
      "grad_norm": 0.5570439052407872,
      "learning_rate": 2.0205046246634108e-07,
      "loss": 11.7016,
      "step": 36007
    },
    {
      "epoch": 1.9607795089610858,
      "grad_norm": 0.512534127629289,
      "learning_rate": 2.0149056645573094e-07,
      "loss": 11.788,
      "step": 36008
    },
    {
      "epoch": 1.9608339629576688,
      "grad_norm": 0.5184480782611407,
      "learning_rate": 2.0093144649505623e-07,
      "loss": 11.7903,
      "step": 36009
    },
    {
      "epoch": 1.9608884169542518,
      "grad_norm": 0.5857843005853354,
      "learning_rate": 2.0037310258868013e-07,
      "loss": 11.8534,
      "step": 36010
    },
    {
      "epoch": 1.9609428709508347,
      "grad_norm": 0.5111709976498129,
      "learning_rate": 1.998155347409436e-07,
      "loss": 11.5414,
      "step": 36011
    },
    {
      "epoch": 1.9609973249474177,
      "grad_norm": 0.5292095211027054,
      "learning_rate": 1.9925874295617652e-07,
      "loss": 11.732,
      "step": 36012
    },
    {
      "epoch": 1.9610517789440007,
      "grad_norm": 0.5856080790092227,
      "learning_rate": 1.9870272723870875e-07,
      "loss": 11.8398,
      "step": 36013
    },
    {
      "epoch": 1.9611062329405837,
      "grad_norm": 0.5033977054195551,
      "learning_rate": 1.9814748759285906e-07,
      "loss": 11.7954,
      "step": 36014
    },
    {
      "epoch": 1.961160686937167,
      "grad_norm": 0.577376962998737,
      "learning_rate": 1.9759302402295731e-07,
      "loss": 11.8011,
      "step": 36015
    },
    {
      "epoch": 1.96121514093375,
      "grad_norm": 0.5397605666328251,
      "learning_rate": 1.9703933653331118e-07,
      "loss": 11.757,
      "step": 36016
    },
    {
      "epoch": 1.961269594930333,
      "grad_norm": 0.5541654365163368,
      "learning_rate": 1.9648642512821724e-07,
      "loss": 11.7338,
      "step": 36017
    },
    {
      "epoch": 1.961324048926916,
      "grad_norm": 0.7217196531735636,
      "learning_rate": 1.9593428981198315e-07,
      "loss": 11.7926,
      "step": 36018
    },
    {
      "epoch": 1.961378502923499,
      "grad_norm": 0.519998290054018,
      "learning_rate": 1.9538293058890543e-07,
      "loss": 11.6498,
      "step": 36019
    },
    {
      "epoch": 1.961432956920082,
      "grad_norm": 0.5272176504867117,
      "learning_rate": 1.948323474632696e-07,
      "loss": 11.7032,
      "step": 36020
    },
    {
      "epoch": 1.9614874109166651,
      "grad_norm": 0.5099707324800663,
      "learning_rate": 1.9428254043935003e-07,
      "loss": 11.451,
      "step": 36021
    },
    {
      "epoch": 1.9615418649132481,
      "grad_norm": 0.5179851898636314,
      "learning_rate": 1.93733509521421e-07,
      "loss": 11.6959,
      "step": 36022
    },
    {
      "epoch": 1.961596318909831,
      "grad_norm": 0.5372911964255624,
      "learning_rate": 1.9318525471376802e-07,
      "loss": 11.8403,
      "step": 36023
    },
    {
      "epoch": 1.961650772906414,
      "grad_norm": 0.761199762720354,
      "learning_rate": 1.9263777602064327e-07,
      "loss": 11.727,
      "step": 36024
    },
    {
      "epoch": 1.961705226902997,
      "grad_norm": 0.4977141123013079,
      "learning_rate": 1.9209107344629885e-07,
      "loss": 11.7505,
      "step": 36025
    },
    {
      "epoch": 1.96175968089958,
      "grad_norm": 0.549922404983853,
      "learning_rate": 1.9154514699499805e-07,
      "loss": 11.7971,
      "step": 36026
    },
    {
      "epoch": 1.961814134896163,
      "grad_norm": 0.5653266048395228,
      "learning_rate": 1.909999966709819e-07,
      "loss": 11.8807,
      "step": 36027
    },
    {
      "epoch": 1.961868588892746,
      "grad_norm": 0.5776493133370724,
      "learning_rate": 1.9045562247848036e-07,
      "loss": 11.7873,
      "step": 36028
    },
    {
      "epoch": 1.961923042889329,
      "grad_norm": 0.5485644438477592,
      "learning_rate": 1.899120244217345e-07,
      "loss": 11.7357,
      "step": 36029
    },
    {
      "epoch": 1.961977496885912,
      "grad_norm": 0.5381699981451984,
      "learning_rate": 1.8936920250497425e-07,
      "loss": 11.6875,
      "step": 36030
    },
    {
      "epoch": 1.962031950882495,
      "grad_norm": 0.5204981639255041,
      "learning_rate": 1.8882715673241847e-07,
      "loss": 11.7527,
      "step": 36031
    },
    {
      "epoch": 1.962086404879078,
      "grad_norm": 0.4941649129436532,
      "learning_rate": 1.8828588710827488e-07,
      "loss": 11.7216,
      "step": 36032
    },
    {
      "epoch": 1.962140858875661,
      "grad_norm": 0.5787425359730036,
      "learning_rate": 1.8774539363676235e-07,
      "loss": 11.8042,
      "step": 36033
    },
    {
      "epoch": 1.962195312872244,
      "grad_norm": 0.5298592707443699,
      "learning_rate": 1.8720567632207753e-07,
      "loss": 11.7328,
      "step": 36034
    },
    {
      "epoch": 1.962249766868827,
      "grad_norm": 0.6048682113144075,
      "learning_rate": 1.8666673516842814e-07,
      "loss": 11.7806,
      "step": 36035
    },
    {
      "epoch": 1.96230422086541,
      "grad_norm": 0.5203259207628885,
      "learning_rate": 1.8612857017998864e-07,
      "loss": 11.7484,
      "step": 36036
    },
    {
      "epoch": 1.962358674861993,
      "grad_norm": 0.5203654124111285,
      "learning_rate": 1.8559118136095566e-07,
      "loss": 11.7989,
      "step": 36037
    },
    {
      "epoch": 1.9624131288585762,
      "grad_norm": 0.5574154978845504,
      "learning_rate": 1.8505456871550364e-07,
      "loss": 11.803,
      "step": 36038
    },
    {
      "epoch": 1.9624675828551592,
      "grad_norm": 0.5187730591400143,
      "learning_rate": 1.8451873224780703e-07,
      "loss": 11.7569,
      "step": 36039
    },
    {
      "epoch": 1.9625220368517422,
      "grad_norm": 0.5397345979656596,
      "learning_rate": 1.8398367196202914e-07,
      "loss": 11.5374,
      "step": 36040
    },
    {
      "epoch": 1.9625764908483252,
      "grad_norm": 0.5249696001010945,
      "learning_rate": 1.8344938786233334e-07,
      "loss": 11.8053,
      "step": 36041
    },
    {
      "epoch": 1.9626309448449082,
      "grad_norm": 0.6110236430723038,
      "learning_rate": 1.8291587995288295e-07,
      "loss": 11.8182,
      "step": 36042
    },
    {
      "epoch": 1.9626853988414912,
      "grad_norm": 0.5413015579929863,
      "learning_rate": 1.82383148237808e-07,
      "loss": 11.7702,
      "step": 36043
    },
    {
      "epoch": 1.9627398528380744,
      "grad_norm": 0.583288513988857,
      "learning_rate": 1.8185119272126072e-07,
      "loss": 11.714,
      "step": 36044
    },
    {
      "epoch": 1.9627943068346574,
      "grad_norm": 0.5334439277205439,
      "learning_rate": 1.8132001340739334e-07,
      "loss": 11.8311,
      "step": 36045
    },
    {
      "epoch": 1.9628487608312404,
      "grad_norm": 0.6313076412553729,
      "learning_rate": 1.8078961030030262e-07,
      "loss": 11.9019,
      "step": 36046
    },
    {
      "epoch": 1.9629032148278234,
      "grad_norm": 0.5802374704330805,
      "learning_rate": 1.8025998340414075e-07,
      "loss": 11.7941,
      "step": 36047
    },
    {
      "epoch": 1.9629576688244064,
      "grad_norm": 0.5530229621499704,
      "learning_rate": 1.7973113272301556e-07,
      "loss": 11.7996,
      "step": 36048
    },
    {
      "epoch": 1.9630121228209894,
      "grad_norm": 0.611200710106464,
      "learning_rate": 1.79203058261046e-07,
      "loss": 11.7493,
      "step": 36049
    },
    {
      "epoch": 1.9630665768175724,
      "grad_norm": 0.5538728300789155,
      "learning_rate": 1.7867576002232878e-07,
      "loss": 11.8603,
      "step": 36050
    },
    {
      "epoch": 1.9631210308141553,
      "grad_norm": 0.5902849805325706,
      "learning_rate": 1.781492380109606e-07,
      "loss": 11.7395,
      "step": 36051
    },
    {
      "epoch": 1.9631754848107383,
      "grad_norm": 0.5172885994447218,
      "learning_rate": 1.7762349223106046e-07,
      "loss": 11.7958,
      "step": 36052
    },
    {
      "epoch": 1.9632299388073213,
      "grad_norm": 0.5611427830043566,
      "learning_rate": 1.770985226866806e-07,
      "loss": 11.7454,
      "step": 36053
    },
    {
      "epoch": 1.9632843928039043,
      "grad_norm": 0.5455502001926145,
      "learning_rate": 1.7657432938193997e-07,
      "loss": 11.8378,
      "step": 36054
    },
    {
      "epoch": 1.9633388468004873,
      "grad_norm": 0.5278946672983367,
      "learning_rate": 1.7605091232089088e-07,
      "loss": 11.6538,
      "step": 36055
    },
    {
      "epoch": 1.9633933007970703,
      "grad_norm": 0.6774585307507369,
      "learning_rate": 1.7552827150760788e-07,
      "loss": 11.9573,
      "step": 36056
    },
    {
      "epoch": 1.9634477547936533,
      "grad_norm": 0.5560376521290947,
      "learning_rate": 1.7500640694616544e-07,
      "loss": 11.8007,
      "step": 36057
    },
    {
      "epoch": 1.9635022087902363,
      "grad_norm": 0.5959644389597665,
      "learning_rate": 1.744853186406048e-07,
      "loss": 11.8268,
      "step": 36058
    },
    {
      "epoch": 1.9635566627868193,
      "grad_norm": 0.5390154979311883,
      "learning_rate": 1.7396500659498938e-07,
      "loss": 11.7693,
      "step": 36059
    },
    {
      "epoch": 1.9636111167834023,
      "grad_norm": 0.5274492519851942,
      "learning_rate": 1.7344547081337147e-07,
      "loss": 11.7415,
      "step": 36060
    },
    {
      "epoch": 1.9636655707799853,
      "grad_norm": 0.602824620451956,
      "learning_rate": 1.7292671129977013e-07,
      "loss": 11.7833,
      "step": 36061
    },
    {
      "epoch": 1.9637200247765685,
      "grad_norm": 0.5294160814401556,
      "learning_rate": 1.724087280582376e-07,
      "loss": 11.6379,
      "step": 36062
    },
    {
      "epoch": 1.9637744787731515,
      "grad_norm": 0.49065103439773283,
      "learning_rate": 1.7189152109280405e-07,
      "loss": 11.8398,
      "step": 36063
    },
    {
      "epoch": 1.9638289327697345,
      "grad_norm": 0.5378153877668171,
      "learning_rate": 1.7137509040746625e-07,
      "loss": 11.861,
      "step": 36064
    },
    {
      "epoch": 1.9638833867663175,
      "grad_norm": 0.520991498039108,
      "learning_rate": 1.7085943600626542e-07,
      "loss": 11.827,
      "step": 36065
    },
    {
      "epoch": 1.9639378407629005,
      "grad_norm": 0.5675381880773707,
      "learning_rate": 1.7034455789320947e-07,
      "loss": 11.8059,
      "step": 36066
    },
    {
      "epoch": 1.9639922947594837,
      "grad_norm": 0.48008876926079747,
      "learning_rate": 1.698304560722841e-07,
      "loss": 11.5584,
      "step": 36067
    },
    {
      "epoch": 1.9640467487560667,
      "grad_norm": 0.542396205604003,
      "learning_rate": 1.693171305475083e-07,
      "loss": 11.7639,
      "step": 36068
    },
    {
      "epoch": 1.9641012027526497,
      "grad_norm": 0.51188379760167,
      "learning_rate": 1.688045813228567e-07,
      "loss": 11.8636,
      "step": 36069
    },
    {
      "epoch": 1.9641556567492326,
      "grad_norm": 0.5727338721667895,
      "learning_rate": 1.682928084023261e-07,
      "loss": 11.7454,
      "step": 36070
    },
    {
      "epoch": 1.9642101107458156,
      "grad_norm": 0.5429382756620393,
      "learning_rate": 1.6778181178989106e-07,
      "loss": 11.7864,
      "step": 36071
    },
    {
      "epoch": 1.9642645647423986,
      "grad_norm": 0.6027145212717409,
      "learning_rate": 1.672715914895262e-07,
      "loss": 11.7084,
      "step": 36072
    },
    {
      "epoch": 1.9643190187389816,
      "grad_norm": 0.5824037743355904,
      "learning_rate": 1.6676214750520613e-07,
      "loss": 11.7379,
      "step": 36073
    },
    {
      "epoch": 1.9643734727355646,
      "grad_norm": 0.5365915612304775,
      "learning_rate": 1.6625347984088324e-07,
      "loss": 11.7134,
      "step": 36074
    },
    {
      "epoch": 1.9644279267321476,
      "grad_norm": 0.5111232224399088,
      "learning_rate": 1.657455885005099e-07,
      "loss": 11.6599,
      "step": 36075
    },
    {
      "epoch": 1.9644823807287306,
      "grad_norm": 0.5047051101211683,
      "learning_rate": 1.6523847348803856e-07,
      "loss": 11.8254,
      "step": 36076
    },
    {
      "epoch": 1.9645368347253136,
      "grad_norm": 0.5362279851786779,
      "learning_rate": 1.6473213480742156e-07,
      "loss": 11.7533,
      "step": 36077
    },
    {
      "epoch": 1.9645912887218966,
      "grad_norm": 0.6139211616223916,
      "learning_rate": 1.642265724625891e-07,
      "loss": 11.8627,
      "step": 36078
    },
    {
      "epoch": 1.9646457427184796,
      "grad_norm": 0.5461223989996234,
      "learning_rate": 1.6372178645747138e-07,
      "loss": 11.736,
      "step": 36079
    },
    {
      "epoch": 1.9647001967150626,
      "grad_norm": 0.5359178466362011,
      "learning_rate": 1.6321777679599858e-07,
      "loss": 11.8026,
      "step": 36080
    },
    {
      "epoch": 1.9647546507116456,
      "grad_norm": 0.560500157611614,
      "learning_rate": 1.627145434820898e-07,
      "loss": 11.7971,
      "step": 36081
    },
    {
      "epoch": 1.9648091047082286,
      "grad_norm": 0.516300969854972,
      "learning_rate": 1.6221208651964192e-07,
      "loss": 11.6479,
      "step": 36082
    },
    {
      "epoch": 1.9648635587048116,
      "grad_norm": 0.5579620078889198,
      "learning_rate": 1.6171040591258512e-07,
      "loss": 11.8091,
      "step": 36083
    },
    {
      "epoch": 1.9649180127013945,
      "grad_norm": 0.6173431760203307,
      "learning_rate": 1.612095016648163e-07,
      "loss": 11.7821,
      "step": 36084
    },
    {
      "epoch": 1.9649724666979778,
      "grad_norm": 1.0244614891651473,
      "learning_rate": 1.607093737802101e-07,
      "loss": 11.8587,
      "step": 36085
    },
    {
      "epoch": 1.9650269206945608,
      "grad_norm": 0.537696805089532,
      "learning_rate": 1.6021002226268567e-07,
      "loss": 11.7809,
      "step": 36086
    },
    {
      "epoch": 1.9650813746911437,
      "grad_norm": 0.5401388137255373,
      "learning_rate": 1.5971144711610654e-07,
      "loss": 11.7671,
      "step": 36087
    },
    {
      "epoch": 1.9651358286877267,
      "grad_norm": 0.49885934768199663,
      "learning_rate": 1.5921364834435847e-07,
      "loss": 11.7048,
      "step": 36088
    },
    {
      "epoch": 1.9651902826843097,
      "grad_norm": 0.5756231592654099,
      "learning_rate": 1.587166259513051e-07,
      "loss": 11.8952,
      "step": 36089
    },
    {
      "epoch": 1.9652447366808927,
      "grad_norm": 0.5280664174641819,
      "learning_rate": 1.5822037994080997e-07,
      "loss": 11.6791,
      "step": 36090
    },
    {
      "epoch": 1.965299190677476,
      "grad_norm": 0.5578361953831942,
      "learning_rate": 1.5772491031673665e-07,
      "loss": 11.7343,
      "step": 36091
    },
    {
      "epoch": 1.965353644674059,
      "grad_norm": 0.5191451354626264,
      "learning_rate": 1.572302170829376e-07,
      "loss": 11.7964,
      "step": 36092
    },
    {
      "epoch": 1.965408098670642,
      "grad_norm": 0.5843496016765073,
      "learning_rate": 1.5673630024326536e-07,
      "loss": 11.7899,
      "step": 36093
    },
    {
      "epoch": 1.965462552667225,
      "grad_norm": 0.5260539209834371,
      "learning_rate": 1.5624315980155014e-07,
      "loss": 11.6003,
      "step": 36094
    },
    {
      "epoch": 1.965517006663808,
      "grad_norm": 0.5505038763418566,
      "learning_rate": 1.5575079576164443e-07,
      "loss": 11.6972,
      "step": 36095
    },
    {
      "epoch": 1.965571460660391,
      "grad_norm": 0.5464706381830436,
      "learning_rate": 1.552592081273563e-07,
      "loss": 11.871,
      "step": 36096
    },
    {
      "epoch": 1.965625914656974,
      "grad_norm": 0.6060484170851027,
      "learning_rate": 1.5476839690250488e-07,
      "loss": 11.9239,
      "step": 36097
    },
    {
      "epoch": 1.9656803686535569,
      "grad_norm": 0.5348211016419099,
      "learning_rate": 1.542783620909316e-07,
      "loss": 11.7832,
      "step": 36098
    },
    {
      "epoch": 1.9657348226501399,
      "grad_norm": 0.527604278349885,
      "learning_rate": 1.537891036964334e-07,
      "loss": 11.8067,
      "step": 36099
    },
    {
      "epoch": 1.9657892766467229,
      "grad_norm": 0.545937571217918,
      "learning_rate": 1.5330062172281835e-07,
      "loss": 11.8276,
      "step": 36100
    },
    {
      "epoch": 1.9658437306433059,
      "grad_norm": 0.6159916248778422,
      "learning_rate": 1.5281291617387227e-07,
      "loss": 11.9024,
      "step": 36101
    },
    {
      "epoch": 1.9658981846398889,
      "grad_norm": 0.5393725227542722,
      "learning_rate": 1.5232598705340327e-07,
      "loss": 11.8396,
      "step": 36102
    },
    {
      "epoch": 1.9659526386364718,
      "grad_norm": 0.5850637917033171,
      "learning_rate": 1.5183983436519722e-07,
      "loss": 11.6428,
      "step": 36103
    },
    {
      "epoch": 1.9660070926330548,
      "grad_norm": 0.6049639957187175,
      "learning_rate": 1.5135445811301774e-07,
      "loss": 11.8286,
      "step": 36104
    },
    {
      "epoch": 1.9660615466296378,
      "grad_norm": 0.5391677320089677,
      "learning_rate": 1.508698583006507e-07,
      "loss": 11.6355,
      "step": 36105
    },
    {
      "epoch": 1.9661160006262208,
      "grad_norm": 0.5963141921636924,
      "learning_rate": 1.5038603493187087e-07,
      "loss": 11.8,
      "step": 36106
    },
    {
      "epoch": 1.9661704546228038,
      "grad_norm": 0.5607816036452667,
      "learning_rate": 1.4990298801043079e-07,
      "loss": 11.78,
      "step": 36107
    },
    {
      "epoch": 1.966224908619387,
      "grad_norm": 0.5467884372063281,
      "learning_rate": 1.4942071754009413e-07,
      "loss": 11.8496,
      "step": 36108
    },
    {
      "epoch": 1.96627936261597,
      "grad_norm": 0.5237521257677624,
      "learning_rate": 1.4893922352460233e-07,
      "loss": 11.8206,
      "step": 36109
    },
    {
      "epoch": 1.966333816612553,
      "grad_norm": 0.5031099102344975,
      "learning_rate": 1.4845850596770793e-07,
      "loss": 11.7422,
      "step": 36110
    },
    {
      "epoch": 1.966388270609136,
      "grad_norm": 0.5607691559665281,
      "learning_rate": 1.4797856487314132e-07,
      "loss": 11.8437,
      "step": 36111
    },
    {
      "epoch": 1.966442724605719,
      "grad_norm": 0.6055633202044425,
      "learning_rate": 1.4749940024464393e-07,
      "loss": 11.9146,
      "step": 36112
    },
    {
      "epoch": 1.966497178602302,
      "grad_norm": 0.5361187480751732,
      "learning_rate": 1.47021012085935e-07,
      "loss": 11.7409,
      "step": 36113
    },
    {
      "epoch": 1.9665516325988852,
      "grad_norm": 0.653688839131671,
      "learning_rate": 1.465434004007338e-07,
      "loss": 11.9156,
      "step": 36114
    },
    {
      "epoch": 1.9666060865954682,
      "grad_norm": 0.5149642296727405,
      "learning_rate": 1.4606656519275952e-07,
      "loss": 11.789,
      "step": 36115
    },
    {
      "epoch": 1.9666605405920512,
      "grad_norm": 0.5676475229880306,
      "learning_rate": 1.4559050646570925e-07,
      "loss": 11.8223,
      "step": 36116
    },
    {
      "epoch": 1.9667149945886342,
      "grad_norm": 0.5252849505699192,
      "learning_rate": 1.4511522422330227e-07,
      "loss": 11.7996,
      "step": 36117
    },
    {
      "epoch": 1.9667694485852172,
      "grad_norm": 0.5877817757785959,
      "learning_rate": 1.4464071846921333e-07,
      "loss": 11.7568,
      "step": 36118
    },
    {
      "epoch": 1.9668239025818002,
      "grad_norm": 0.5882393227807133,
      "learning_rate": 1.4416698920716177e-07,
      "loss": 11.8876,
      "step": 36119
    },
    {
      "epoch": 1.9668783565783832,
      "grad_norm": 0.5704054098512796,
      "learning_rate": 1.4369403644080014e-07,
      "loss": 11.7115,
      "step": 36120
    },
    {
      "epoch": 1.9669328105749662,
      "grad_norm": 0.5289613004063723,
      "learning_rate": 1.4322186017382554e-07,
      "loss": 11.9047,
      "step": 36121
    },
    {
      "epoch": 1.9669872645715492,
      "grad_norm": 0.5453688217064268,
      "learning_rate": 1.427504604098906e-07,
      "loss": 11.7507,
      "step": 36122
    },
    {
      "epoch": 1.9670417185681321,
      "grad_norm": 0.5198981806085811,
      "learning_rate": 1.4227983715269232e-07,
      "loss": 11.6551,
      "step": 36123
    },
    {
      "epoch": 1.9670961725647151,
      "grad_norm": 0.5253699755300153,
      "learning_rate": 1.4180999040586117e-07,
      "loss": 11.7519,
      "step": 36124
    },
    {
      "epoch": 1.9671506265612981,
      "grad_norm": 0.5393785358843216,
      "learning_rate": 1.413409201730609e-07,
      "loss": 11.859,
      "step": 36125
    },
    {
      "epoch": 1.9672050805578811,
      "grad_norm": 0.5558675984047506,
      "learning_rate": 1.408726264579441e-07,
      "loss": 11.7202,
      "step": 36126
    },
    {
      "epoch": 1.9672595345544641,
      "grad_norm": 0.5586216581087697,
      "learning_rate": 1.404051092641523e-07,
      "loss": 11.89,
      "step": 36127
    },
    {
      "epoch": 1.967313988551047,
      "grad_norm": 0.537080041467235,
      "learning_rate": 1.3993836859531596e-07,
      "loss": 11.5992,
      "step": 36128
    },
    {
      "epoch": 1.96736844254763,
      "grad_norm": 0.5485479140489351,
      "learning_rate": 1.3947240445505437e-07,
      "loss": 11.8157,
      "step": 36129
    },
    {
      "epoch": 1.967422896544213,
      "grad_norm": 0.5432455194118807,
      "learning_rate": 1.3900721684702023e-07,
      "loss": 11.6892,
      "step": 36130
    },
    {
      "epoch": 1.967477350540796,
      "grad_norm": 0.5711861910186576,
      "learning_rate": 1.385428057747995e-07,
      "loss": 11.8944,
      "step": 36131
    },
    {
      "epoch": 1.9675318045373793,
      "grad_norm": 0.5397806512059342,
      "learning_rate": 1.3807917124201154e-07,
      "loss": 11.7456,
      "step": 36132
    },
    {
      "epoch": 1.9675862585339623,
      "grad_norm": 0.5562135585852154,
      "learning_rate": 1.3761631325227564e-07,
      "loss": 11.8171,
      "step": 36133
    },
    {
      "epoch": 1.9676407125305453,
      "grad_norm": 0.5412331934578432,
      "learning_rate": 1.3715423180917786e-07,
      "loss": 11.8402,
      "step": 36134
    },
    {
      "epoch": 1.9676951665271283,
      "grad_norm": 0.4967746914567025,
      "learning_rate": 1.366929269163264e-07,
      "loss": 11.7487,
      "step": 36135
    },
    {
      "epoch": 1.9677496205237113,
      "grad_norm": 0.5395412386790123,
      "learning_rate": 1.362323985772851e-07,
      "loss": 11.7525,
      "step": 36136
    },
    {
      "epoch": 1.9678040745202945,
      "grad_norm": 0.6026132408976158,
      "learning_rate": 1.3577264679566214e-07,
      "loss": 11.8774,
      "step": 36137
    },
    {
      "epoch": 1.9678585285168775,
      "grad_norm": 0.5580241116103313,
      "learning_rate": 1.353136715749992e-07,
      "loss": 11.8646,
      "step": 36138
    },
    {
      "epoch": 1.9679129825134605,
      "grad_norm": 0.5345519822307682,
      "learning_rate": 1.3485547291890442e-07,
      "loss": 11.7214,
      "step": 36139
    },
    {
      "epoch": 1.9679674365100435,
      "grad_norm": 0.5376399763605038,
      "learning_rate": 1.3439805083090838e-07,
      "loss": 11.6297,
      "step": 36140
    },
    {
      "epoch": 1.9680218905066265,
      "grad_norm": 0.5098290078463686,
      "learning_rate": 1.3394140531458598e-07,
      "loss": 11.8686,
      "step": 36141
    },
    {
      "epoch": 1.9680763445032095,
      "grad_norm": 0.49553484030839434,
      "learning_rate": 1.3348553637347882e-07,
      "loss": 11.746,
      "step": 36142
    },
    {
      "epoch": 1.9681307984997924,
      "grad_norm": 0.5133708628717928,
      "learning_rate": 1.3303044401112852e-07,
      "loss": 11.6988,
      "step": 36143
    },
    {
      "epoch": 1.9681852524963754,
      "grad_norm": 0.6003084040624407,
      "learning_rate": 1.325761282310878e-07,
      "loss": 11.9923,
      "step": 36144
    },
    {
      "epoch": 1.9682397064929584,
      "grad_norm": 0.517819959884357,
      "learning_rate": 1.3212258903688714e-07,
      "loss": 11.729,
      "step": 36145
    },
    {
      "epoch": 1.9682941604895414,
      "grad_norm": 0.5401976231040443,
      "learning_rate": 1.3166982643204595e-07,
      "loss": 11.6976,
      "step": 36146
    },
    {
      "epoch": 1.9683486144861244,
      "grad_norm": 0.5557863424856615,
      "learning_rate": 1.3121784042009478e-07,
      "loss": 11.786,
      "step": 36147
    },
    {
      "epoch": 1.9684030684827074,
      "grad_norm": 0.5617659525947073,
      "learning_rate": 1.3076663100453079e-07,
      "loss": 11.8253,
      "step": 36148
    },
    {
      "epoch": 1.9684575224792904,
      "grad_norm": 0.5604156723002983,
      "learning_rate": 1.303161981888734e-07,
      "loss": 11.6716,
      "step": 36149
    },
    {
      "epoch": 1.9685119764758734,
      "grad_norm": 0.5239089373389401,
      "learning_rate": 1.2986654197664205e-07,
      "loss": 11.8228,
      "step": 36150
    },
    {
      "epoch": 1.9685664304724564,
      "grad_norm": 0.5449828924114282,
      "learning_rate": 1.294176623713006e-07,
      "loss": 11.6734,
      "step": 36151
    },
    {
      "epoch": 1.9686208844690394,
      "grad_norm": 0.5389916291346643,
      "learning_rate": 1.2896955937635734e-07,
      "loss": 11.7565,
      "step": 36152
    },
    {
      "epoch": 1.9686753384656224,
      "grad_norm": 0.5836005375121872,
      "learning_rate": 1.2852223299529843e-07,
      "loss": 11.9432,
      "step": 36153
    },
    {
      "epoch": 1.9687297924622054,
      "grad_norm": 0.5841893821371371,
      "learning_rate": 1.280756832315988e-07,
      "loss": 11.8333,
      "step": 36154
    },
    {
      "epoch": 1.9687842464587886,
      "grad_norm": 0.5275668060950642,
      "learning_rate": 1.2762991008873348e-07,
      "loss": 11.7275,
      "step": 36155
    },
    {
      "epoch": 1.9688387004553716,
      "grad_norm": 0.5217971901470458,
      "learning_rate": 1.2718491357016637e-07,
      "loss": 11.8409,
      "step": 36156
    },
    {
      "epoch": 1.9688931544519546,
      "grad_norm": 0.5498514013741188,
      "learning_rate": 1.2674069367935026e-07,
      "loss": 11.8383,
      "step": 36157
    },
    {
      "epoch": 1.9689476084485376,
      "grad_norm": 0.5214066287859432,
      "learning_rate": 1.2629725041976014e-07,
      "loss": 11.7107,
      "step": 36158
    },
    {
      "epoch": 1.9690020624451205,
      "grad_norm": 0.5597912773438624,
      "learning_rate": 1.2585458379481552e-07,
      "loss": 11.7649,
      "step": 36159
    },
    {
      "epoch": 1.9690565164417035,
      "grad_norm": 0.5673417262170881,
      "learning_rate": 1.2541269380799136e-07,
      "loss": 11.7681,
      "step": 36160
    },
    {
      "epoch": 1.9691109704382868,
      "grad_norm": 0.5287815407454016,
      "learning_rate": 1.249715804626961e-07,
      "loss": 11.7399,
      "step": 36161
    },
    {
      "epoch": 1.9691654244348697,
      "grad_norm": 0.5069560083107458,
      "learning_rate": 1.2453124376237134e-07,
      "loss": 11.7218,
      "step": 36162
    },
    {
      "epoch": 1.9692198784314527,
      "grad_norm": 0.518075331035977,
      "learning_rate": 1.2409168371044776e-07,
      "loss": 11.6591,
      "step": 36163
    },
    {
      "epoch": 1.9692743324280357,
      "grad_norm": 0.5758893774532786,
      "learning_rate": 1.2365290031032263e-07,
      "loss": 11.7872,
      "step": 36164
    },
    {
      "epoch": 1.9693287864246187,
      "grad_norm": 0.5899421541870227,
      "learning_rate": 1.2321489356543758e-07,
      "loss": 11.7815,
      "step": 36165
    },
    {
      "epoch": 1.9693832404212017,
      "grad_norm": 0.5995609830507865,
      "learning_rate": 1.2277766347917886e-07,
      "loss": 11.9585,
      "step": 36166
    },
    {
      "epoch": 1.9694376944177847,
      "grad_norm": 0.5431532295982318,
      "learning_rate": 1.223412100549437e-07,
      "loss": 11.7666,
      "step": 36167
    },
    {
      "epoch": 1.9694921484143677,
      "grad_norm": 0.5715447701570818,
      "learning_rate": 1.219055332961405e-07,
      "loss": 11.8167,
      "step": 36168
    },
    {
      "epoch": 1.9695466024109507,
      "grad_norm": 0.5283972325969607,
      "learning_rate": 1.2147063320614438e-07,
      "loss": 11.7003,
      "step": 36169
    },
    {
      "epoch": 1.9696010564075337,
      "grad_norm": 0.6585150801629824,
      "learning_rate": 1.2103650978834147e-07,
      "loss": 11.8573,
      "step": 36170
    },
    {
      "epoch": 1.9696555104041167,
      "grad_norm": 0.565393904228365,
      "learning_rate": 1.2060316304610686e-07,
      "loss": 11.7446,
      "step": 36171
    },
    {
      "epoch": 1.9697099644006997,
      "grad_norm": 0.5742858779716876,
      "learning_rate": 1.2017059298281564e-07,
      "loss": 11.7469,
      "step": 36172
    },
    {
      "epoch": 1.9697644183972827,
      "grad_norm": 0.5166168818491959,
      "learning_rate": 1.1973879960183176e-07,
      "loss": 11.7861,
      "step": 36173
    },
    {
      "epoch": 1.9698188723938657,
      "grad_norm": 0.5302467696516732,
      "learning_rate": 1.1930778290650814e-07,
      "loss": 11.7151,
      "step": 36174
    },
    {
      "epoch": 1.9698733263904487,
      "grad_norm": 0.5700285546687027,
      "learning_rate": 1.188775429001865e-07,
      "loss": 11.7277,
      "step": 36175
    },
    {
      "epoch": 1.9699277803870316,
      "grad_norm": 0.5419987626982037,
      "learning_rate": 1.1844807958623083e-07,
      "loss": 11.7204,
      "step": 36176
    },
    {
      "epoch": 1.9699822343836146,
      "grad_norm": 0.4873817336037878,
      "learning_rate": 1.1801939296797182e-07,
      "loss": 11.6233,
      "step": 36177
    },
    {
      "epoch": 1.9700366883801979,
      "grad_norm": 0.49949415826014854,
      "learning_rate": 1.1759148304875122e-07,
      "loss": 11.6539,
      "step": 36178
    },
    {
      "epoch": 1.9700911423767808,
      "grad_norm": 0.5398713427126732,
      "learning_rate": 1.1716434983187751e-07,
      "loss": 11.7904,
      "step": 36179
    },
    {
      "epoch": 1.9701455963733638,
      "grad_norm": 0.5627717589213662,
      "learning_rate": 1.1673799332069247e-07,
      "loss": 11.7619,
      "step": 36180
    },
    {
      "epoch": 1.9702000503699468,
      "grad_norm": 0.5212479404199305,
      "learning_rate": 1.1631241351850453e-07,
      "loss": 11.7151,
      "step": 36181
    },
    {
      "epoch": 1.9702545043665298,
      "grad_norm": 0.5318582358039198,
      "learning_rate": 1.1588761042862218e-07,
      "loss": 11.7379,
      "step": 36182
    },
    {
      "epoch": 1.9703089583631128,
      "grad_norm": 0.5851594414369534,
      "learning_rate": 1.1546358405434277e-07,
      "loss": 11.77,
      "step": 36183
    },
    {
      "epoch": 1.970363412359696,
      "grad_norm": 0.5343863641122449,
      "learning_rate": 1.1504033439896367e-07,
      "loss": 11.7686,
      "step": 36184
    },
    {
      "epoch": 1.970417866356279,
      "grad_norm": 0.5995527303681142,
      "learning_rate": 1.1461786146579334e-07,
      "loss": 11.7558,
      "step": 36185
    },
    {
      "epoch": 1.970472320352862,
      "grad_norm": 0.5154038835467099,
      "learning_rate": 1.1419616525809584e-07,
      "loss": 11.6241,
      "step": 36186
    },
    {
      "epoch": 1.970526774349445,
      "grad_norm": 0.5522059941507242,
      "learning_rate": 1.1377524577916854e-07,
      "loss": 11.7807,
      "step": 36187
    },
    {
      "epoch": 1.970581228346028,
      "grad_norm": 0.577810220383659,
      "learning_rate": 1.1335510303226437e-07,
      "loss": 11.767,
      "step": 36188
    },
    {
      "epoch": 1.970635682342611,
      "grad_norm": 0.5724667765123155,
      "learning_rate": 1.1293573702068072e-07,
      "loss": 11.6692,
      "step": 36189
    },
    {
      "epoch": 1.970690136339194,
      "grad_norm": 0.6336511842378724,
      "learning_rate": 1.1251714774764833e-07,
      "loss": 11.7558,
      "step": 36190
    },
    {
      "epoch": 1.970744590335777,
      "grad_norm": 0.5545459044883255,
      "learning_rate": 1.1209933521643124e-07,
      "loss": 11.7489,
      "step": 36191
    },
    {
      "epoch": 1.97079904433236,
      "grad_norm": 0.5551963448600635,
      "learning_rate": 1.1168229943028241e-07,
      "loss": 11.7147,
      "step": 36192
    },
    {
      "epoch": 1.970853498328943,
      "grad_norm": 0.4912865617046046,
      "learning_rate": 1.112660403924437e-07,
      "loss": 11.7612,
      "step": 36193
    },
    {
      "epoch": 1.970907952325526,
      "grad_norm": 0.5470037838420644,
      "learning_rate": 1.1085055810615696e-07,
      "loss": 11.6291,
      "step": 36194
    },
    {
      "epoch": 1.970962406322109,
      "grad_norm": 0.5738744663975652,
      "learning_rate": 1.1043585257464184e-07,
      "loss": 11.8226,
      "step": 36195
    },
    {
      "epoch": 1.971016860318692,
      "grad_norm": 0.5450275334591849,
      "learning_rate": 1.1002192380112908e-07,
      "loss": 11.8505,
      "step": 36196
    },
    {
      "epoch": 1.971071314315275,
      "grad_norm": 0.553970158224054,
      "learning_rate": 1.0960877178883833e-07,
      "loss": 11.7523,
      "step": 36197
    },
    {
      "epoch": 1.971125768311858,
      "grad_norm": 0.5110423403723059,
      "learning_rate": 1.0919639654097813e-07,
      "loss": 11.7543,
      "step": 36198
    },
    {
      "epoch": 1.971180222308441,
      "grad_norm": 0.5611192989680622,
      "learning_rate": 1.0878479806076813e-07,
      "loss": 11.8191,
      "step": 36199
    },
    {
      "epoch": 1.971234676305024,
      "grad_norm": 0.5475503858666368,
      "learning_rate": 1.0837397635139468e-07,
      "loss": 11.7071,
      "step": 36200
    },
    {
      "epoch": 1.9712891303016071,
      "grad_norm": 0.5519847935598601,
      "learning_rate": 1.0796393141605521e-07,
      "loss": 11.6902,
      "step": 36201
    },
    {
      "epoch": 1.9713435842981901,
      "grad_norm": 0.5856589134744672,
      "learning_rate": 1.0755466325793606e-07,
      "loss": 11.6871,
      "step": 36202
    },
    {
      "epoch": 1.9713980382947731,
      "grad_norm": 0.518004116995887,
      "learning_rate": 1.0714617188022357e-07,
      "loss": 11.6014,
      "step": 36203
    },
    {
      "epoch": 1.971452492291356,
      "grad_norm": 0.6271548455564989,
      "learning_rate": 1.067384572861041e-07,
      "loss": 11.8362,
      "step": 36204
    },
    {
      "epoch": 1.971506946287939,
      "grad_norm": 0.594355934820692,
      "learning_rate": 1.0633151947873065e-07,
      "loss": 11.708,
      "step": 36205
    },
    {
      "epoch": 1.971561400284522,
      "grad_norm": 0.530694315035782,
      "learning_rate": 1.0592535846127849e-07,
      "loss": 11.7427,
      "step": 36206
    },
    {
      "epoch": 1.9716158542811053,
      "grad_norm": 0.5497407668420948,
      "learning_rate": 1.0551997423690063e-07,
      "loss": 11.7952,
      "step": 36207
    },
    {
      "epoch": 1.9716703082776883,
      "grad_norm": 0.5237390509891918,
      "learning_rate": 1.0511536680876122e-07,
      "loss": 11.7361,
      "step": 36208
    },
    {
      "epoch": 1.9717247622742713,
      "grad_norm": 0.5521301418158289,
      "learning_rate": 1.0471153617999108e-07,
      "loss": 11.7663,
      "step": 36209
    },
    {
      "epoch": 1.9717792162708543,
      "grad_norm": 0.6364183485173556,
      "learning_rate": 1.0430848235373214e-07,
      "loss": 11.9224,
      "step": 36210
    },
    {
      "epoch": 1.9718336702674373,
      "grad_norm": 0.5466973097287856,
      "learning_rate": 1.0390620533312634e-07,
      "loss": 11.8913,
      "step": 36211
    },
    {
      "epoch": 1.9718881242640203,
      "grad_norm": 0.5324499588525479,
      "learning_rate": 1.035047051212934e-07,
      "loss": 11.8186,
      "step": 36212
    },
    {
      "epoch": 1.9719425782606033,
      "grad_norm": 0.5269022730122065,
      "learning_rate": 1.0310398172136415e-07,
      "loss": 11.7582,
      "step": 36213
    },
    {
      "epoch": 1.9719970322571863,
      "grad_norm": 0.7373916070420619,
      "learning_rate": 1.0270403513645832e-07,
      "loss": 11.7461,
      "step": 36214
    },
    {
      "epoch": 1.9720514862537692,
      "grad_norm": 0.519722595597763,
      "learning_rate": 1.0230486536967344e-07,
      "loss": 11.7973,
      "step": 36215
    },
    {
      "epoch": 1.9721059402503522,
      "grad_norm": 0.5892860532731992,
      "learning_rate": 1.0190647242411811e-07,
      "loss": 11.6721,
      "step": 36216
    },
    {
      "epoch": 1.9721603942469352,
      "grad_norm": 0.5406902008203212,
      "learning_rate": 1.0150885630288987e-07,
      "loss": 11.6886,
      "step": 36217
    },
    {
      "epoch": 1.9722148482435182,
      "grad_norm": 0.557050987792401,
      "learning_rate": 1.0111201700908624e-07,
      "loss": 11.8016,
      "step": 36218
    },
    {
      "epoch": 1.9722693022401012,
      "grad_norm": 0.5353839650173511,
      "learning_rate": 1.0071595454578253e-07,
      "loss": 11.8251,
      "step": 36219
    },
    {
      "epoch": 1.9723237562366842,
      "grad_norm": 0.5437374207443233,
      "learning_rate": 1.0032066891606517e-07,
      "loss": 11.8201,
      "step": 36220
    },
    {
      "epoch": 1.9723782102332672,
      "grad_norm": 0.5638026184592565,
      "learning_rate": 9.992616012300949e-08,
      "loss": 11.9145,
      "step": 36221
    },
    {
      "epoch": 1.9724326642298502,
      "grad_norm": 0.5387061459151387,
      "learning_rate": 9.953242816967968e-08,
      "loss": 11.7386,
      "step": 36222
    },
    {
      "epoch": 1.9724871182264332,
      "grad_norm": 0.5329417947855982,
      "learning_rate": 9.913947305913995e-08,
      "loss": 11.7285,
      "step": 36223
    },
    {
      "epoch": 1.9725415722230162,
      "grad_norm": 0.5437957574204663,
      "learning_rate": 9.874729479444345e-08,
      "loss": 11.7789,
      "step": 36224
    },
    {
      "epoch": 1.9725960262195994,
      "grad_norm": 0.5532036869654136,
      "learning_rate": 9.835589337864326e-08,
      "loss": 11.8299,
      "step": 36225
    },
    {
      "epoch": 1.9726504802161824,
      "grad_norm": 0.5942121282855267,
      "learning_rate": 9.796526881478141e-08,
      "loss": 11.8215,
      "step": 36226
    },
    {
      "epoch": 1.9727049342127654,
      "grad_norm": 0.5400951194990719,
      "learning_rate": 9.75754211058999e-08,
      "loss": 11.8526,
      "step": 36227
    },
    {
      "epoch": 1.9727593882093484,
      "grad_norm": 0.5148944871312742,
      "learning_rate": 9.718635025501855e-08,
      "loss": 11.8274,
      "step": 36228
    },
    {
      "epoch": 1.9728138422059314,
      "grad_norm": 0.5032515065594267,
      "learning_rate": 9.679805626517934e-08,
      "loss": 11.6671,
      "step": 36229
    },
    {
      "epoch": 1.9728682962025144,
      "grad_norm": 0.5470610301875037,
      "learning_rate": 9.641053913937992e-08,
      "loss": 11.6036,
      "step": 36230
    },
    {
      "epoch": 1.9729227501990976,
      "grad_norm": 0.53765309824049,
      "learning_rate": 9.602379888065116e-08,
      "loss": 11.8478,
      "step": 36231
    },
    {
      "epoch": 1.9729772041956806,
      "grad_norm": 0.5587061458410771,
      "learning_rate": 9.563783549200178e-08,
      "loss": 11.7793,
      "step": 36232
    },
    {
      "epoch": 1.9730316581922636,
      "grad_norm": 0.5366956153541028,
      "learning_rate": 9.525264897641828e-08,
      "loss": 11.6919,
      "step": 36233
    },
    {
      "epoch": 1.9730861121888466,
      "grad_norm": 0.511307339449537,
      "learning_rate": 9.486823933689826e-08,
      "loss": 11.7315,
      "step": 36234
    },
    {
      "epoch": 1.9731405661854295,
      "grad_norm": 0.5064735221613555,
      "learning_rate": 9.448460657645042e-08,
      "loss": 11.7467,
      "step": 36235
    },
    {
      "epoch": 1.9731950201820125,
      "grad_norm": 0.5419981151780828,
      "learning_rate": 9.410175069803906e-08,
      "loss": 11.7226,
      "step": 36236
    },
    {
      "epoch": 1.9732494741785955,
      "grad_norm": 0.6664414139562269,
      "learning_rate": 9.371967170463958e-08,
      "loss": 11.7607,
      "step": 36237
    },
    {
      "epoch": 1.9733039281751785,
      "grad_norm": 0.5885293274527131,
      "learning_rate": 9.333836959923847e-08,
      "loss": 11.8146,
      "step": 36238
    },
    {
      "epoch": 1.9733583821717615,
      "grad_norm": 0.5063013234728269,
      "learning_rate": 9.295784438478894e-08,
      "loss": 11.7359,
      "step": 36239
    },
    {
      "epoch": 1.9734128361683445,
      "grad_norm": 0.5694645873013854,
      "learning_rate": 9.257809606426638e-08,
      "loss": 11.8281,
      "step": 36240
    },
    {
      "epoch": 1.9734672901649275,
      "grad_norm": 0.5171677678543232,
      "learning_rate": 9.219912464060176e-08,
      "loss": 11.6704,
      "step": 36241
    },
    {
      "epoch": 1.9735217441615105,
      "grad_norm": 0.559913902747464,
      "learning_rate": 9.182093011674831e-08,
      "loss": 11.6763,
      "step": 36242
    },
    {
      "epoch": 1.9735761981580935,
      "grad_norm": 0.5667602243596593,
      "learning_rate": 9.144351249564809e-08,
      "loss": 11.7827,
      "step": 36243
    },
    {
      "epoch": 1.9736306521546765,
      "grad_norm": 0.556631289938428,
      "learning_rate": 9.106687178024321e-08,
      "loss": 11.7439,
      "step": 36244
    },
    {
      "epoch": 1.9736851061512595,
      "grad_norm": 0.5730097167908303,
      "learning_rate": 9.069100797345353e-08,
      "loss": 11.6917,
      "step": 36245
    },
    {
      "epoch": 1.9737395601478425,
      "grad_norm": 0.542573366011134,
      "learning_rate": 9.031592107821008e-08,
      "loss": 11.8264,
      "step": 36246
    },
    {
      "epoch": 1.9737940141444255,
      "grad_norm": 0.5295022511165232,
      "learning_rate": 8.99416110974216e-08,
      "loss": 11.5568,
      "step": 36247
    },
    {
      "epoch": 1.9738484681410087,
      "grad_norm": 0.6218076862655832,
      "learning_rate": 8.956807803399691e-08,
      "loss": 11.8278,
      "step": 36248
    },
    {
      "epoch": 1.9739029221375917,
      "grad_norm": 0.5000786730024737,
      "learning_rate": 8.919532189085589e-08,
      "loss": 11.7464,
      "step": 36249
    },
    {
      "epoch": 1.9739573761341747,
      "grad_norm": 0.5658771570438295,
      "learning_rate": 8.88233426708851e-08,
      "loss": 11.8297,
      "step": 36250
    },
    {
      "epoch": 1.9740118301307576,
      "grad_norm": 0.5070732070418988,
      "learning_rate": 8.845214037697113e-08,
      "loss": 11.7376,
      "step": 36251
    },
    {
      "epoch": 1.9740662841273406,
      "grad_norm": 0.5577908444636813,
      "learning_rate": 8.808171501201167e-08,
      "loss": 11.85,
      "step": 36252
    },
    {
      "epoch": 1.9741207381239236,
      "grad_norm": 0.48970841671500775,
      "learning_rate": 8.77120665788933e-08,
      "loss": 11.7533,
      "step": 36253
    },
    {
      "epoch": 1.9741751921205068,
      "grad_norm": 0.5301730065638747,
      "learning_rate": 8.734319508046929e-08,
      "loss": 11.6283,
      "step": 36254
    },
    {
      "epoch": 1.9742296461170898,
      "grad_norm": 0.5573255084230732,
      "learning_rate": 8.697510051962621e-08,
      "loss": 11.7985,
      "step": 36255
    },
    {
      "epoch": 1.9742841001136728,
      "grad_norm": 0.5134263743120279,
      "learning_rate": 8.660778289921734e-08,
      "loss": 11.7688,
      "step": 36256
    },
    {
      "epoch": 1.9743385541102558,
      "grad_norm": 0.5559404718408812,
      "learning_rate": 8.624124222210705e-08,
      "loss": 11.7263,
      "step": 36257
    },
    {
      "epoch": 1.9743930081068388,
      "grad_norm": 0.5465372986633922,
      "learning_rate": 8.587547849112642e-08,
      "loss": 11.7493,
      "step": 36258
    },
    {
      "epoch": 1.9744474621034218,
      "grad_norm": 0.6107175971590492,
      "learning_rate": 8.551049170915094e-08,
      "loss": 11.8658,
      "step": 36259
    },
    {
      "epoch": 1.9745019161000048,
      "grad_norm": 0.5091342860021256,
      "learning_rate": 8.514628187898944e-08,
      "loss": 11.6125,
      "step": 36260
    },
    {
      "epoch": 1.9745563700965878,
      "grad_norm": 0.48977344812386725,
      "learning_rate": 8.47828490034952e-08,
      "loss": 11.7845,
      "step": 36261
    },
    {
      "epoch": 1.9746108240931708,
      "grad_norm": 0.5106483975923595,
      "learning_rate": 8.442019308547711e-08,
      "loss": 11.7983,
      "step": 36262
    },
    {
      "epoch": 1.9746652780897538,
      "grad_norm": 0.5654591778208484,
      "learning_rate": 8.405831412776621e-08,
      "loss": 11.8042,
      "step": 36263
    },
    {
      "epoch": 1.9747197320863368,
      "grad_norm": 0.5390632403146243,
      "learning_rate": 8.369721213318248e-08,
      "loss": 11.6458,
      "step": 36264
    },
    {
      "epoch": 1.9747741860829198,
      "grad_norm": 0.6122913704702639,
      "learning_rate": 8.333688710451259e-08,
      "loss": 11.7465,
      "step": 36265
    },
    {
      "epoch": 1.9748286400795028,
      "grad_norm": 0.5812981482406246,
      "learning_rate": 8.29773390445765e-08,
      "loss": 11.8522,
      "step": 36266
    },
    {
      "epoch": 1.9748830940760858,
      "grad_norm": 0.5626605446982994,
      "learning_rate": 8.261856795617195e-08,
      "loss": 11.5391,
      "step": 36267
    },
    {
      "epoch": 1.9749375480726687,
      "grad_norm": 0.5645370918586674,
      "learning_rate": 8.226057384206343e-08,
      "loss": 11.7902,
      "step": 36268
    },
    {
      "epoch": 1.9749920020692517,
      "grad_norm": 0.5304535637631037,
      "learning_rate": 8.190335670507088e-08,
      "loss": 11.7177,
      "step": 36269
    },
    {
      "epoch": 1.9750464560658347,
      "grad_norm": 0.5090654073949913,
      "learning_rate": 8.154691654794766e-08,
      "loss": 11.7826,
      "step": 36270
    },
    {
      "epoch": 1.975100910062418,
      "grad_norm": 0.5589764265176818,
      "learning_rate": 8.119125337346934e-08,
      "loss": 11.8097,
      "step": 36271
    },
    {
      "epoch": 1.975155364059001,
      "grad_norm": 0.4915100717076199,
      "learning_rate": 8.083636718441145e-08,
      "loss": 11.8505,
      "step": 36272
    },
    {
      "epoch": 1.975209818055584,
      "grad_norm": 0.555226556038492,
      "learning_rate": 8.048225798351628e-08,
      "loss": 11.7101,
      "step": 36273
    },
    {
      "epoch": 1.975264272052167,
      "grad_norm": 0.5304210386952574,
      "learning_rate": 8.012892577354824e-08,
      "loss": 11.8985,
      "step": 36274
    },
    {
      "epoch": 1.97531872604875,
      "grad_norm": 0.5224139660861945,
      "learning_rate": 7.977637055726073e-08,
      "loss": 11.6573,
      "step": 36275
    },
    {
      "epoch": 1.975373180045333,
      "grad_norm": 0.5965846458124298,
      "learning_rate": 7.942459233738486e-08,
      "loss": 11.838,
      "step": 36276
    },
    {
      "epoch": 1.9754276340419161,
      "grad_norm": 0.5522995528529425,
      "learning_rate": 7.90735911166629e-08,
      "loss": 11.7547,
      "step": 36277
    },
    {
      "epoch": 1.9754820880384991,
      "grad_norm": 0.574676307176989,
      "learning_rate": 7.87233668978149e-08,
      "loss": 11.627,
      "step": 36278
    },
    {
      "epoch": 1.975536542035082,
      "grad_norm": 0.5173689839251362,
      "learning_rate": 7.83739196835831e-08,
      "loss": 11.8449,
      "step": 36279
    },
    {
      "epoch": 1.975590996031665,
      "grad_norm": 0.5135224400747184,
      "learning_rate": 7.802524947666534e-08,
      "loss": 11.6768,
      "step": 36280
    },
    {
      "epoch": 1.975645450028248,
      "grad_norm": 0.5251815678512176,
      "learning_rate": 7.767735627978168e-08,
      "loss": 11.7705,
      "step": 36281
    },
    {
      "epoch": 1.975699904024831,
      "grad_norm": 0.586207223170747,
      "learning_rate": 7.733024009562994e-08,
      "loss": 11.8765,
      "step": 36282
    },
    {
      "epoch": 1.975754358021414,
      "grad_norm": 0.5574241248148462,
      "learning_rate": 7.698390092691909e-08,
      "loss": 11.8348,
      "step": 36283
    },
    {
      "epoch": 1.975808812017997,
      "grad_norm": 0.7273289480528171,
      "learning_rate": 7.663833877634696e-08,
      "loss": 11.933,
      "step": 36284
    },
    {
      "epoch": 1.97586326601458,
      "grad_norm": 0.5668389575448375,
      "learning_rate": 7.629355364657809e-08,
      "loss": 11.8707,
      "step": 36285
    },
    {
      "epoch": 1.975917720011163,
      "grad_norm": 0.50882618487503,
      "learning_rate": 7.59495455403103e-08,
      "loss": 11.7991,
      "step": 36286
    },
    {
      "epoch": 1.975972174007746,
      "grad_norm": 0.5920375125195722,
      "learning_rate": 7.560631446023036e-08,
      "loss": 11.8265,
      "step": 36287
    },
    {
      "epoch": 1.976026628004329,
      "grad_norm": 0.5267960479256812,
      "learning_rate": 7.52638604089806e-08,
      "loss": 11.7645,
      "step": 36288
    },
    {
      "epoch": 1.976081082000912,
      "grad_norm": 0.5596102602300225,
      "learning_rate": 7.492218338923662e-08,
      "loss": 11.7543,
      "step": 36289
    },
    {
      "epoch": 1.976135535997495,
      "grad_norm": 0.5709911412573924,
      "learning_rate": 7.458128340366299e-08,
      "loss": 11.7865,
      "step": 36290
    },
    {
      "epoch": 1.976189989994078,
      "grad_norm": 0.5582750815536331,
      "learning_rate": 7.424116045489094e-08,
      "loss": 11.8644,
      "step": 36291
    },
    {
      "epoch": 1.976244443990661,
      "grad_norm": 0.5237505920322937,
      "learning_rate": 7.390181454558498e-08,
      "loss": 11.7387,
      "step": 36292
    },
    {
      "epoch": 1.976298897987244,
      "grad_norm": 0.5097076329400535,
      "learning_rate": 7.356324567837636e-08,
      "loss": 11.7185,
      "step": 36293
    },
    {
      "epoch": 1.976353351983827,
      "grad_norm": 0.5429137780618095,
      "learning_rate": 7.32254538558963e-08,
      "loss": 11.8103,
      "step": 36294
    },
    {
      "epoch": 1.9764078059804102,
      "grad_norm": 0.5841830201353465,
      "learning_rate": 7.288843908076492e-08,
      "loss": 11.7246,
      "step": 36295
    },
    {
      "epoch": 1.9764622599769932,
      "grad_norm": 0.5782744501294761,
      "learning_rate": 7.255220135562457e-08,
      "loss": 11.8672,
      "step": 36296
    },
    {
      "epoch": 1.9765167139735762,
      "grad_norm": 0.5426315506050473,
      "learning_rate": 7.221674068306206e-08,
      "loss": 11.7543,
      "step": 36297
    },
    {
      "epoch": 1.9765711679701592,
      "grad_norm": 0.5466101557634807,
      "learning_rate": 7.188205706570861e-08,
      "loss": 11.5791,
      "step": 36298
    },
    {
      "epoch": 1.9766256219667422,
      "grad_norm": 0.5375236649410577,
      "learning_rate": 7.154815050616215e-08,
      "loss": 11.7321,
      "step": 36299
    },
    {
      "epoch": 1.9766800759633252,
      "grad_norm": 0.4964365523418336,
      "learning_rate": 7.121502100700949e-08,
      "loss": 11.6377,
      "step": 36300
    },
    {
      "epoch": 1.9767345299599084,
      "grad_norm": 0.6305503821042118,
      "learning_rate": 7.088266857084858e-08,
      "loss": 11.8237,
      "step": 36301
    },
    {
      "epoch": 1.9767889839564914,
      "grad_norm": 0.5358256091857745,
      "learning_rate": 7.055109320025511e-08,
      "loss": 11.7467,
      "step": 36302
    },
    {
      "epoch": 1.9768434379530744,
      "grad_norm": 0.5497933257512864,
      "learning_rate": 7.0220294897827e-08,
      "loss": 11.8914,
      "step": 36303
    },
    {
      "epoch": 1.9768978919496574,
      "grad_norm": 0.5984924652247515,
      "learning_rate": 6.989027366611778e-08,
      "loss": 11.7903,
      "step": 36304
    },
    {
      "epoch": 1.9769523459462404,
      "grad_norm": 0.550647246077146,
      "learning_rate": 6.956102950770315e-08,
      "loss": 11.8169,
      "step": 36305
    },
    {
      "epoch": 1.9770067999428234,
      "grad_norm": 0.5377030961924383,
      "learning_rate": 6.923256242513664e-08,
      "loss": 11.7392,
      "step": 36306
    },
    {
      "epoch": 1.9770612539394063,
      "grad_norm": 0.5272190434435017,
      "learning_rate": 6.890487242097177e-08,
      "loss": 11.7926,
      "step": 36307
    },
    {
      "epoch": 1.9771157079359893,
      "grad_norm": 0.5148450472477232,
      "learning_rate": 6.857795949776203e-08,
      "loss": 11.7307,
      "step": 36308
    },
    {
      "epoch": 1.9771701619325723,
      "grad_norm": 0.5522228817044328,
      "learning_rate": 6.825182365806093e-08,
      "loss": 11.8029,
      "step": 36309
    },
    {
      "epoch": 1.9772246159291553,
      "grad_norm": 0.5048907866642403,
      "learning_rate": 6.792646490437759e-08,
      "loss": 11.5917,
      "step": 36310
    },
    {
      "epoch": 1.9772790699257383,
      "grad_norm": 0.5817331170603239,
      "learning_rate": 6.760188323926552e-08,
      "loss": 11.7732,
      "step": 36311
    },
    {
      "epoch": 1.9773335239223213,
      "grad_norm": 0.5569403418667223,
      "learning_rate": 6.727807866523384e-08,
      "loss": 11.8178,
      "step": 36312
    },
    {
      "epoch": 1.9773879779189043,
      "grad_norm": 0.5254249993695699,
      "learning_rate": 6.695505118481382e-08,
      "loss": 11.7582,
      "step": 36313
    },
    {
      "epoch": 1.9774424319154873,
      "grad_norm": 0.5173435106520328,
      "learning_rate": 6.66328008005035e-08,
      "loss": 11.7492,
      "step": 36314
    },
    {
      "epoch": 1.9774968859120703,
      "grad_norm": 0.531210626335476,
      "learning_rate": 6.631132751482305e-08,
      "loss": 11.8797,
      "step": 36315
    },
    {
      "epoch": 1.9775513399086533,
      "grad_norm": 0.5094864697529042,
      "learning_rate": 6.59906313302594e-08,
      "loss": 11.7976,
      "step": 36316
    },
    {
      "epoch": 1.9776057939052363,
      "grad_norm": 0.5615927347246377,
      "learning_rate": 6.567071224931054e-08,
      "loss": 11.7936,
      "step": 36317
    },
    {
      "epoch": 1.9776602479018195,
      "grad_norm": 0.5895509348544351,
      "learning_rate": 6.535157027446337e-08,
      "loss": 11.6687,
      "step": 36318
    },
    {
      "epoch": 1.9777147018984025,
      "grad_norm": 0.5208164665987108,
      "learning_rate": 6.503320540820479e-08,
      "loss": 11.8263,
      "step": 36319
    },
    {
      "epoch": 1.9777691558949855,
      "grad_norm": 0.5077054033679513,
      "learning_rate": 6.47156176530106e-08,
      "loss": 11.704,
      "step": 36320
    },
    {
      "epoch": 1.9778236098915685,
      "grad_norm": 0.5650257572135685,
      "learning_rate": 6.439880701134548e-08,
      "loss": 11.8266,
      "step": 36321
    },
    {
      "epoch": 1.9778780638881515,
      "grad_norm": 0.5194485004209728,
      "learning_rate": 6.408277348567415e-08,
      "loss": 11.6489,
      "step": 36322
    },
    {
      "epoch": 1.9779325178847345,
      "grad_norm": 0.5101688774812396,
      "learning_rate": 6.37675170784502e-08,
      "loss": 11.753,
      "step": 36323
    },
    {
      "epoch": 1.9779869718813177,
      "grad_norm": 0.49336019726026487,
      "learning_rate": 6.34530377921272e-08,
      "loss": 11.6734,
      "step": 36324
    },
    {
      "epoch": 1.9780414258779007,
      "grad_norm": 0.5976201422843452,
      "learning_rate": 6.313933562915874e-08,
      "loss": 11.8844,
      "step": 36325
    },
    {
      "epoch": 1.9780958798744837,
      "grad_norm": 0.5111855462812013,
      "learning_rate": 6.282641059197625e-08,
      "loss": 11.7917,
      "step": 36326
    },
    {
      "epoch": 1.9781503338710666,
      "grad_norm": 0.5422945022187791,
      "learning_rate": 6.251426268301108e-08,
      "loss": 11.7145,
      "step": 36327
    },
    {
      "epoch": 1.9782047878676496,
      "grad_norm": 0.510816362739602,
      "learning_rate": 6.220289190470575e-08,
      "loss": 11.8235,
      "step": 36328
    },
    {
      "epoch": 1.9782592418642326,
      "grad_norm": 0.5620730410934067,
      "learning_rate": 6.18922982594583e-08,
      "loss": 11.8108,
      "step": 36329
    },
    {
      "epoch": 1.9783136958608156,
      "grad_norm": 0.5609138016276113,
      "learning_rate": 6.158248174970017e-08,
      "loss": 11.7199,
      "step": 36330
    },
    {
      "epoch": 1.9783681498573986,
      "grad_norm": 0.5202221186895,
      "learning_rate": 6.12734423778405e-08,
      "loss": 11.7697,
      "step": 36331
    },
    {
      "epoch": 1.9784226038539816,
      "grad_norm": 0.572701285802484,
      "learning_rate": 6.096518014627738e-08,
      "loss": 11.9467,
      "step": 36332
    },
    {
      "epoch": 1.9784770578505646,
      "grad_norm": 0.5556647868062884,
      "learning_rate": 6.065769505740892e-08,
      "loss": 11.754,
      "step": 36333
    },
    {
      "epoch": 1.9785315118471476,
      "grad_norm": 0.5417592795573949,
      "learning_rate": 6.035098711362209e-08,
      "loss": 11.7906,
      "step": 36334
    },
    {
      "epoch": 1.9785859658437306,
      "grad_norm": 0.5681967901406085,
      "learning_rate": 6.004505631730383e-08,
      "loss": 11.6744,
      "step": 36335
    },
    {
      "epoch": 1.9786404198403136,
      "grad_norm": 0.5218432259138275,
      "learning_rate": 5.973990267084118e-08,
      "loss": 11.8088,
      "step": 36336
    },
    {
      "epoch": 1.9786948738368966,
      "grad_norm": 0.509235367162828,
      "learning_rate": 5.9435526176598863e-08,
      "loss": 11.7769,
      "step": 36337
    },
    {
      "epoch": 1.9787493278334796,
      "grad_norm": 0.5982462132848184,
      "learning_rate": 5.913192683694169e-08,
      "loss": 11.9151,
      "step": 36338
    },
    {
      "epoch": 1.9788037818300626,
      "grad_norm": 0.6370102271443775,
      "learning_rate": 5.882910465424551e-08,
      "loss": 11.8847,
      "step": 36339
    },
    {
      "epoch": 1.9788582358266456,
      "grad_norm": 0.5820695328139325,
      "learning_rate": 5.852705963084182e-08,
      "loss": 11.7506,
      "step": 36340
    },
    {
      "epoch": 1.9789126898232288,
      "grad_norm": 0.5480070882482397,
      "learning_rate": 5.8225791769106476e-08,
      "loss": 11.7188,
      "step": 36341
    },
    {
      "epoch": 1.9789671438198118,
      "grad_norm": 0.5234889592020289,
      "learning_rate": 5.7925301071348746e-08,
      "loss": 11.8682,
      "step": 36342
    },
    {
      "epoch": 1.9790215978163948,
      "grad_norm": 0.5489082558769252,
      "learning_rate": 5.7625587539944514e-08,
      "loss": 11.7167,
      "step": 36343
    },
    {
      "epoch": 1.9790760518129777,
      "grad_norm": 0.5211445365558279,
      "learning_rate": 5.732665117719194e-08,
      "loss": 11.866,
      "step": 36344
    },
    {
      "epoch": 1.9791305058095607,
      "grad_norm": 0.5313983787771561,
      "learning_rate": 5.702849198542248e-08,
      "loss": 11.6887,
      "step": 36345
    },
    {
      "epoch": 1.9791849598061437,
      "grad_norm": 0.5935234732911259,
      "learning_rate": 5.673110996696762e-08,
      "loss": 11.9406,
      "step": 36346
    },
    {
      "epoch": 1.979239413802727,
      "grad_norm": 0.5406266078476016,
      "learning_rate": 5.6434505124136616e-08,
      "loss": 11.7714,
      "step": 36347
    },
    {
      "epoch": 1.97929386779931,
      "grad_norm": 0.577116098170172,
      "learning_rate": 5.613867745922763e-08,
      "loss": 11.7744,
      "step": 36348
    },
    {
      "epoch": 1.979348321795893,
      "grad_norm": 0.5639332310482835,
      "learning_rate": 5.584362697453882e-08,
      "loss": 11.8665,
      "step": 36349
    },
    {
      "epoch": 1.979402775792476,
      "grad_norm": 0.586943512552007,
      "learning_rate": 5.554935367237946e-08,
      "loss": 11.7265,
      "step": 36350
    },
    {
      "epoch": 1.979457229789059,
      "grad_norm": 0.5458017642790904,
      "learning_rate": 5.525585755502549e-08,
      "loss": 11.8465,
      "step": 36351
    },
    {
      "epoch": 1.979511683785642,
      "grad_norm": 0.6195057890363825,
      "learning_rate": 5.496313862476399e-08,
      "loss": 11.7799,
      "step": 36352
    },
    {
      "epoch": 1.979566137782225,
      "grad_norm": 0.5576611249620743,
      "learning_rate": 5.46711968838598e-08,
      "loss": 11.7103,
      "step": 36353
    },
    {
      "epoch": 1.979620591778808,
      "grad_norm": 0.5828726139288403,
      "learning_rate": 5.438003233459998e-08,
      "loss": 11.8346,
      "step": 36354
    },
    {
      "epoch": 1.9796750457753909,
      "grad_norm": 0.5360299264781121,
      "learning_rate": 5.40896449792494e-08,
      "loss": 11.8725,
      "step": 36355
    },
    {
      "epoch": 1.9797294997719739,
      "grad_norm": 0.5294166679762473,
      "learning_rate": 5.3800034820050696e-08,
      "loss": 11.7766,
      "step": 36356
    },
    {
      "epoch": 1.9797839537685569,
      "grad_norm": 0.5134185955352489,
      "learning_rate": 5.3511201859268725e-08,
      "loss": 11.6572,
      "step": 36357
    },
    {
      "epoch": 1.9798384077651399,
      "grad_norm": 0.6311800638950488,
      "learning_rate": 5.3223146099135035e-08,
      "loss": 11.9058,
      "step": 36358
    },
    {
      "epoch": 1.9798928617617229,
      "grad_norm": 0.5834818119326419,
      "learning_rate": 5.2935867541914487e-08,
      "loss": 11.7748,
      "step": 36359
    },
    {
      "epoch": 1.9799473157583058,
      "grad_norm": 0.5675728564670772,
      "learning_rate": 5.264936618981642e-08,
      "loss": 11.692,
      "step": 36360
    },
    {
      "epoch": 1.9800017697548888,
      "grad_norm": 0.5399396559574882,
      "learning_rate": 5.236364204507238e-08,
      "loss": 11.7302,
      "step": 36361
    },
    {
      "epoch": 1.9800562237514718,
      "grad_norm": 0.5459817402648429,
      "learning_rate": 5.207869510992502e-08,
      "loss": 11.7319,
      "step": 36362
    },
    {
      "epoch": 1.9801106777480548,
      "grad_norm": 0.5013970198304216,
      "learning_rate": 5.179452538656149e-08,
      "loss": 11.8196,
      "step": 36363
    },
    {
      "epoch": 1.9801651317446378,
      "grad_norm": 0.5595841064832009,
      "learning_rate": 5.151113287721332e-08,
      "loss": 11.8034,
      "step": 36364
    },
    {
      "epoch": 1.980219585741221,
      "grad_norm": 0.5352657501973164,
      "learning_rate": 5.122851758406766e-08,
      "loss": 11.787,
      "step": 36365
    },
    {
      "epoch": 1.980274039737804,
      "grad_norm": 0.5478585939635453,
      "learning_rate": 5.094667950933385e-08,
      "loss": 11.7097,
      "step": 36366
    },
    {
      "epoch": 1.980328493734387,
      "grad_norm": 0.5369698837798761,
      "learning_rate": 5.0665618655210136e-08,
      "loss": 11.7658,
      "step": 36367
    },
    {
      "epoch": 1.98038294773097,
      "grad_norm": 0.5646834408333486,
      "learning_rate": 5.038533502386145e-08,
      "loss": 11.8147,
      "step": 36368
    },
    {
      "epoch": 1.980437401727553,
      "grad_norm": 0.5748268256684115,
      "learning_rate": 5.0105828617474925e-08,
      "loss": 11.7697,
      "step": 36369
    },
    {
      "epoch": 1.9804918557241362,
      "grad_norm": 0.5403407442121676,
      "learning_rate": 4.982709943823771e-08,
      "loss": 11.7193,
      "step": 36370
    },
    {
      "epoch": 1.9805463097207192,
      "grad_norm": 0.558221577878052,
      "learning_rate": 4.9549147488303635e-08,
      "loss": 11.7359,
      "step": 36371
    },
    {
      "epoch": 1.9806007637173022,
      "grad_norm": 0.513380065672353,
      "learning_rate": 4.927197276982653e-08,
      "loss": 11.7409,
      "step": 36372
    },
    {
      "epoch": 1.9806552177138852,
      "grad_norm": 0.5309627229713108,
      "learning_rate": 4.899557528498244e-08,
      "loss": 11.7988,
      "step": 36373
    },
    {
      "epoch": 1.9807096717104682,
      "grad_norm": 0.5099223751420987,
      "learning_rate": 4.871995503591409e-08,
      "loss": 11.7858,
      "step": 36374
    },
    {
      "epoch": 1.9807641257070512,
      "grad_norm": 0.5365808849213447,
      "learning_rate": 4.8445112024753106e-08,
      "loss": 11.7678,
      "step": 36375
    },
    {
      "epoch": 1.9808185797036342,
      "grad_norm": 0.5243160670745014,
      "learning_rate": 4.817104625364221e-08,
      "loss": 11.8326,
      "step": 36376
    },
    {
      "epoch": 1.9808730337002172,
      "grad_norm": 0.5720339580511433,
      "learning_rate": 4.789775772472416e-08,
      "loss": 11.835,
      "step": 36377
    },
    {
      "epoch": 1.9809274876968002,
      "grad_norm": 0.5196927828287613,
      "learning_rate": 4.762524644010835e-08,
      "loss": 11.7265,
      "step": 36378
    },
    {
      "epoch": 1.9809819416933832,
      "grad_norm": 0.564518442792885,
      "learning_rate": 4.735351240192642e-08,
      "loss": 11.863,
      "step": 36379
    },
    {
      "epoch": 1.9810363956899661,
      "grad_norm": 0.5165776029013883,
      "learning_rate": 4.7082555612287804e-08,
      "loss": 11.7244,
      "step": 36380
    },
    {
      "epoch": 1.9810908496865491,
      "grad_norm": 0.5488738136409199,
      "learning_rate": 4.6812376073290806e-08,
      "loss": 11.6977,
      "step": 36381
    },
    {
      "epoch": 1.9811453036831321,
      "grad_norm": 0.5675803755394957,
      "learning_rate": 4.6542973787044866e-08,
      "loss": 11.7042,
      "step": 36382
    },
    {
      "epoch": 1.9811997576797151,
      "grad_norm": 0.5550562381550627,
      "learning_rate": 4.6274348755637185e-08,
      "loss": 11.9232,
      "step": 36383
    },
    {
      "epoch": 1.9812542116762981,
      "grad_norm": 0.5136672127163762,
      "learning_rate": 4.60065009811661e-08,
      "loss": 11.7147,
      "step": 36384
    },
    {
      "epoch": 1.981308665672881,
      "grad_norm": 0.5152296456485161,
      "learning_rate": 4.5739430465718826e-08,
      "loss": 11.805,
      "step": 36385
    },
    {
      "epoch": 1.981363119669464,
      "grad_norm": 0.5471769060936504,
      "learning_rate": 4.547313721136037e-08,
      "loss": 11.7434,
      "step": 36386
    },
    {
      "epoch": 1.981417573666047,
      "grad_norm": 0.5797772295236849,
      "learning_rate": 4.520762122015576e-08,
      "loss": 11.7883,
      "step": 36387
    },
    {
      "epoch": 1.9814720276626303,
      "grad_norm": 0.5224920669972644,
      "learning_rate": 4.4942882494192204e-08,
      "loss": 11.8305,
      "step": 36388
    },
    {
      "epoch": 1.9815264816592133,
      "grad_norm": 0.596610670032578,
      "learning_rate": 4.467892103550142e-08,
      "loss": 11.7957,
      "step": 36389
    },
    {
      "epoch": 1.9815809356557963,
      "grad_norm": 0.49203453727486535,
      "learning_rate": 4.441573684615952e-08,
      "loss": 11.7003,
      "step": 36390
    },
    {
      "epoch": 1.9816353896523793,
      "grad_norm": 0.5130185467092806,
      "learning_rate": 4.415332992820931e-08,
      "loss": 11.7963,
      "step": 36391
    },
    {
      "epoch": 1.9816898436489623,
      "grad_norm": 0.4771372129257163,
      "learning_rate": 4.38917002836714e-08,
      "loss": 11.6706,
      "step": 36392
    },
    {
      "epoch": 1.9817442976455453,
      "grad_norm": 0.519489722446061,
      "learning_rate": 4.36308479145997e-08,
      "loss": 11.7753,
      "step": 36393
    },
    {
      "epoch": 1.9817987516421285,
      "grad_norm": 0.5289121310084038,
      "learning_rate": 4.3370772823014824e-08,
      "loss": 11.7648,
      "step": 36394
    },
    {
      "epoch": 1.9818532056387115,
      "grad_norm": 0.5192001536176016,
      "learning_rate": 4.3111475010948474e-08,
      "loss": 11.7289,
      "step": 36395
    },
    {
      "epoch": 1.9819076596352945,
      "grad_norm": 0.5346676956395546,
      "learning_rate": 4.285295448041016e-08,
      "loss": 11.7567,
      "step": 36396
    },
    {
      "epoch": 1.9819621136318775,
      "grad_norm": 0.5346781785191166,
      "learning_rate": 4.259521123339827e-08,
      "loss": 11.8591,
      "step": 36397
    },
    {
      "epoch": 1.9820165676284605,
      "grad_norm": 0.5311257454039943,
      "learning_rate": 4.233824527194452e-08,
      "loss": 11.807,
      "step": 36398
    },
    {
      "epoch": 1.9820710216250434,
      "grad_norm": 0.5162823434128065,
      "learning_rate": 4.208205659802511e-08,
      "loss": 11.6991,
      "step": 36399
    },
    {
      "epoch": 1.9821254756216264,
      "grad_norm": 0.5467079572874044,
      "learning_rate": 4.182664521362734e-08,
      "loss": 11.8395,
      "step": 36400
    },
    {
      "epoch": 1.9821799296182094,
      "grad_norm": 0.5242538245582777,
      "learning_rate": 4.15720111207607e-08,
      "loss": 11.7732,
      "step": 36401
    },
    {
      "epoch": 1.9822343836147924,
      "grad_norm": 0.5335046877696261,
      "learning_rate": 4.13181543213903e-08,
      "loss": 11.8929,
      "step": 36402
    },
    {
      "epoch": 1.9822888376113754,
      "grad_norm": 0.6177807711391469,
      "learning_rate": 4.106507481749233e-08,
      "loss": 11.9371,
      "step": 36403
    },
    {
      "epoch": 1.9823432916079584,
      "grad_norm": 0.5412095866919604,
      "learning_rate": 4.0812772611042993e-08,
      "loss": 11.8959,
      "step": 36404
    },
    {
      "epoch": 1.9823977456045414,
      "grad_norm": 0.6220907738212006,
      "learning_rate": 4.0561247703985174e-08,
      "loss": 11.9239,
      "step": 36405
    },
    {
      "epoch": 1.9824521996011244,
      "grad_norm": 0.5125961095662214,
      "learning_rate": 4.0310500098295066e-08,
      "loss": 11.7893,
      "step": 36406
    },
    {
      "epoch": 1.9825066535977074,
      "grad_norm": 0.5520342049921835,
      "learning_rate": 4.0060529795904466e-08,
      "loss": 11.7016,
      "step": 36407
    },
    {
      "epoch": 1.9825611075942904,
      "grad_norm": 0.5570034617747267,
      "learning_rate": 3.9811336798778466e-08,
      "loss": 11.7271,
      "step": 36408
    },
    {
      "epoch": 1.9826155615908734,
      "grad_norm": 0.5632854091028757,
      "learning_rate": 3.9562921108837745e-08,
      "loss": 11.8128,
      "step": 36409
    },
    {
      "epoch": 1.9826700155874564,
      "grad_norm": 0.5297255578447356,
      "learning_rate": 3.93152827280141e-08,
      "loss": 11.7802,
      "step": 36410
    },
    {
      "epoch": 1.9827244695840396,
      "grad_norm": 0.5445943732363671,
      "learning_rate": 3.906842165823932e-08,
      "loss": 11.9233,
      "step": 36411
    },
    {
      "epoch": 1.9827789235806226,
      "grad_norm": 0.5533317778112701,
      "learning_rate": 3.882233790143408e-08,
      "loss": 11.7697,
      "step": 36412
    },
    {
      "epoch": 1.9828333775772056,
      "grad_norm": 0.5434511729045822,
      "learning_rate": 3.8577031459519075e-08,
      "loss": 11.9149,
      "step": 36413
    },
    {
      "epoch": 1.9828878315737886,
      "grad_norm": 0.5692900136423962,
      "learning_rate": 3.8332502334381684e-08,
      "loss": 11.7982,
      "step": 36414
    },
    {
      "epoch": 1.9829422855703716,
      "grad_norm": 0.531192699353065,
      "learning_rate": 3.808875052793148e-08,
      "loss": 11.7621,
      "step": 36415
    },
    {
      "epoch": 1.9829967395669545,
      "grad_norm": 0.6144391606959894,
      "learning_rate": 3.7845776042078064e-08,
      "loss": 11.7797,
      "step": 36416
    },
    {
      "epoch": 1.9830511935635378,
      "grad_norm": 0.5129640212616524,
      "learning_rate": 3.7603578878686597e-08,
      "loss": 11.7144,
      "step": 36417
    },
    {
      "epoch": 1.9831056475601208,
      "grad_norm": 0.5452520192810797,
      "learning_rate": 3.736215903966667e-08,
      "loss": 11.7853,
      "step": 36418
    },
    {
      "epoch": 1.9831601015567037,
      "grad_norm": 0.539465194466536,
      "learning_rate": 3.7121516526872345e-08,
      "loss": 11.7376,
      "step": 36419
    },
    {
      "epoch": 1.9832145555532867,
      "grad_norm": 0.5754108188862822,
      "learning_rate": 3.688165134219102e-08,
      "loss": 11.8138,
      "step": 36420
    },
    {
      "epoch": 1.9832690095498697,
      "grad_norm": 0.8407767216307679,
      "learning_rate": 3.664256348747674e-08,
      "loss": 11.8776,
      "step": 36421
    },
    {
      "epoch": 1.9833234635464527,
      "grad_norm": 0.5060928623516017,
      "learning_rate": 3.640425296459471e-08,
      "loss": 11.7418,
      "step": 36422
    },
    {
      "epoch": 1.9833779175430357,
      "grad_norm": 0.5452587918783603,
      "learning_rate": 3.616671977539898e-08,
      "loss": 11.8999,
      "step": 36423
    },
    {
      "epoch": 1.9834323715396187,
      "grad_norm": 0.47998421597627455,
      "learning_rate": 3.5929963921732534e-08,
      "loss": 11.7164,
      "step": 36424
    },
    {
      "epoch": 1.9834868255362017,
      "grad_norm": 0.5355787736840985,
      "learning_rate": 3.569398540544944e-08,
      "loss": 11.7603,
      "step": 36425
    },
    {
      "epoch": 1.9835412795327847,
      "grad_norm": 0.5291167248010584,
      "learning_rate": 3.545878422835935e-08,
      "loss": 11.7304,
      "step": 36426
    },
    {
      "epoch": 1.9835957335293677,
      "grad_norm": 0.5715189458160215,
      "learning_rate": 3.522436039231636e-08,
      "loss": 11.8027,
      "step": 36427
    },
    {
      "epoch": 1.9836501875259507,
      "grad_norm": 0.5517468892746402,
      "learning_rate": 3.499071389911901e-08,
      "loss": 11.8163,
      "step": 36428
    },
    {
      "epoch": 1.9837046415225337,
      "grad_norm": 0.49261450619098257,
      "learning_rate": 3.475784475061028e-08,
      "loss": 11.7235,
      "step": 36429
    },
    {
      "epoch": 1.9837590955191167,
      "grad_norm": 0.6400201614758865,
      "learning_rate": 3.452575294858873e-08,
      "loss": 11.8131,
      "step": 36430
    },
    {
      "epoch": 1.9838135495156997,
      "grad_norm": 0.5098108880897478,
      "learning_rate": 3.429443849485292e-08,
      "loss": 11.7697,
      "step": 36431
    },
    {
      "epoch": 1.9838680035122827,
      "grad_norm": 0.5896519442328669,
      "learning_rate": 3.4063901391212514e-08,
      "loss": 11.7852,
      "step": 36432
    },
    {
      "epoch": 1.9839224575088656,
      "grad_norm": 0.5735672501499843,
      "learning_rate": 3.3834141639454976e-08,
      "loss": 11.8876,
      "step": 36433
    },
    {
      "epoch": 1.9839769115054486,
      "grad_norm": 0.5641724388866606,
      "learning_rate": 3.360515924136776e-08,
      "loss": 11.7415,
      "step": 36434
    },
    {
      "epoch": 1.9840313655020319,
      "grad_norm": 0.529777008549507,
      "learning_rate": 3.337695419872722e-08,
      "loss": 11.716,
      "step": 36435
    },
    {
      "epoch": 1.9840858194986148,
      "grad_norm": 0.5489142191689512,
      "learning_rate": 3.314952651330971e-08,
      "loss": 11.7486,
      "step": 36436
    },
    {
      "epoch": 1.9841402734951978,
      "grad_norm": 0.5623179765079697,
      "learning_rate": 3.29228761868916e-08,
      "loss": 11.7828,
      "step": 36437
    },
    {
      "epoch": 1.9841947274917808,
      "grad_norm": 0.535042210829423,
      "learning_rate": 3.269700322122704e-08,
      "loss": 11.7817,
      "step": 36438
    },
    {
      "epoch": 1.9842491814883638,
      "grad_norm": 0.5466478260566233,
      "learning_rate": 3.247190761808128e-08,
      "loss": 11.7459,
      "step": 36439
    },
    {
      "epoch": 1.984303635484947,
      "grad_norm": 0.527203248150186,
      "learning_rate": 3.2247589379197364e-08,
      "loss": 11.6313,
      "step": 36440
    },
    {
      "epoch": 1.98435808948153,
      "grad_norm": 0.5582220042491767,
      "learning_rate": 3.202404850631835e-08,
      "loss": 11.8305,
      "step": 36441
    },
    {
      "epoch": 1.984412543478113,
      "grad_norm": 0.506057199041528,
      "learning_rate": 3.180128500117619e-08,
      "loss": 11.7391,
      "step": 36442
    },
    {
      "epoch": 1.984466997474696,
      "grad_norm": 0.5030049299910746,
      "learning_rate": 3.1579298865525017e-08,
      "loss": 11.8705,
      "step": 36443
    },
    {
      "epoch": 1.984521451471279,
      "grad_norm": 0.5600930442138133,
      "learning_rate": 3.13580901010746e-08,
      "loss": 11.7956,
      "step": 36444
    },
    {
      "epoch": 1.984575905467862,
      "grad_norm": 0.542997673197189,
      "learning_rate": 3.113765870954577e-08,
      "loss": 11.7903,
      "step": 36445
    },
    {
      "epoch": 1.984630359464445,
      "grad_norm": 0.5415007233739847,
      "learning_rate": 3.091800469264827e-08,
      "loss": 11.7567,
      "step": 36446
    },
    {
      "epoch": 1.984684813461028,
      "grad_norm": 0.6795185988670197,
      "learning_rate": 3.0699128052114056e-08,
      "loss": 11.9329,
      "step": 36447
    },
    {
      "epoch": 1.984739267457611,
      "grad_norm": 0.5924422573874469,
      "learning_rate": 3.048102878961956e-08,
      "loss": 11.8243,
      "step": 36448
    },
    {
      "epoch": 1.984793721454194,
      "grad_norm": 0.6158281607939818,
      "learning_rate": 3.026370690686342e-08,
      "loss": 11.8351,
      "step": 36449
    },
    {
      "epoch": 1.984848175450777,
      "grad_norm": 0.7523505407724791,
      "learning_rate": 3.0047162405544284e-08,
      "loss": 11.673,
      "step": 36450
    },
    {
      "epoch": 1.98490262944736,
      "grad_norm": 0.5584705033824234,
      "learning_rate": 2.983139528734968e-08,
      "loss": 11.6747,
      "step": 36451
    },
    {
      "epoch": 1.984957083443943,
      "grad_norm": 0.5710800115087628,
      "learning_rate": 2.9616405553944958e-08,
      "loss": 11.7849,
      "step": 36452
    },
    {
      "epoch": 1.985011537440526,
      "grad_norm": 0.5356196072968289,
      "learning_rate": 2.9402193207017648e-08,
      "loss": 11.625,
      "step": 36453
    },
    {
      "epoch": 1.985065991437109,
      "grad_norm": 0.47829264998905807,
      "learning_rate": 2.918875824821088e-08,
      "loss": 11.7912,
      "step": 36454
    },
    {
      "epoch": 1.985120445433692,
      "grad_norm": 0.5136787925091855,
      "learning_rate": 2.8976100679212192e-08,
      "loss": 11.7793,
      "step": 36455
    },
    {
      "epoch": 1.985174899430275,
      "grad_norm": 0.5177346165727557,
      "learning_rate": 2.8764220501642514e-08,
      "loss": 11.7661,
      "step": 36456
    },
    {
      "epoch": 1.985229353426858,
      "grad_norm": 0.5417904388014813,
      "learning_rate": 2.855311771717828e-08,
      "loss": 11.8325,
      "step": 36457
    },
    {
      "epoch": 1.9852838074234411,
      "grad_norm": 0.5065989398736146,
      "learning_rate": 2.8342792327451517e-08,
      "loss": 11.7289,
      "step": 36458
    },
    {
      "epoch": 1.9853382614200241,
      "grad_norm": 0.5506387088913992,
      "learning_rate": 2.8133244334094255e-08,
      "loss": 11.7723,
      "step": 36459
    },
    {
      "epoch": 1.9853927154166071,
      "grad_norm": 0.5698689998099923,
      "learning_rate": 2.7924473738738522e-08,
      "loss": 11.7432,
      "step": 36460
    },
    {
      "epoch": 1.98544716941319,
      "grad_norm": 0.5959147601358578,
      "learning_rate": 2.771648054300524e-08,
      "loss": 11.7884,
      "step": 36461
    },
    {
      "epoch": 1.985501623409773,
      "grad_norm": 0.5890098564455121,
      "learning_rate": 2.750926474851534e-08,
      "loss": 11.9101,
      "step": 36462
    },
    {
      "epoch": 1.985556077406356,
      "grad_norm": 0.5885155735101597,
      "learning_rate": 2.7302826356878642e-08,
      "loss": 11.8218,
      "step": 36463
    },
    {
      "epoch": 1.9856105314029393,
      "grad_norm": 0.5177049109306143,
      "learning_rate": 2.7097165369704967e-08,
      "loss": 11.5807,
      "step": 36464
    },
    {
      "epoch": 1.9856649853995223,
      "grad_norm": 0.5207191332899337,
      "learning_rate": 2.689228178858194e-08,
      "loss": 11.7763,
      "step": 36465
    },
    {
      "epoch": 1.9857194393961053,
      "grad_norm": 0.5097256203044499,
      "learning_rate": 2.6688175615119383e-08,
      "loss": 11.8482,
      "step": 36466
    },
    {
      "epoch": 1.9857738933926883,
      "grad_norm": 0.5937435099745028,
      "learning_rate": 2.6484846850882705e-08,
      "loss": 11.8284,
      "step": 36467
    },
    {
      "epoch": 1.9858283473892713,
      "grad_norm": 0.5430212541291283,
      "learning_rate": 2.6282295497470634e-08,
      "loss": 11.6684,
      "step": 36468
    },
    {
      "epoch": 1.9858828013858543,
      "grad_norm": 0.46856519532394764,
      "learning_rate": 2.608052155645968e-08,
      "loss": 11.619,
      "step": 36469
    },
    {
      "epoch": 1.9859372553824373,
      "grad_norm": 0.5376089571158441,
      "learning_rate": 2.587952502940416e-08,
      "loss": 11.7194,
      "step": 36470
    },
    {
      "epoch": 1.9859917093790203,
      "grad_norm": 0.5370737918031686,
      "learning_rate": 2.5679305917880592e-08,
      "loss": 11.7159,
      "step": 36471
    },
    {
      "epoch": 1.9860461633756032,
      "grad_norm": 0.534456341970588,
      "learning_rate": 2.547986422343218e-08,
      "loss": 11.721,
      "step": 36472
    },
    {
      "epoch": 1.9861006173721862,
      "grad_norm": 0.5693198990854652,
      "learning_rate": 2.5281199947624347e-08,
      "loss": 11.8178,
      "step": 36473
    },
    {
      "epoch": 1.9861550713687692,
      "grad_norm": 0.5395660060236315,
      "learning_rate": 2.5083313091989192e-08,
      "loss": 11.9495,
      "step": 36474
    },
    {
      "epoch": 1.9862095253653522,
      "grad_norm": 0.5594244618991824,
      "learning_rate": 2.488620365808103e-08,
      "loss": 11.746,
      "step": 36475
    },
    {
      "epoch": 1.9862639793619352,
      "grad_norm": 0.5300685926005979,
      "learning_rate": 2.4689871647420872e-08,
      "loss": 11.7432,
      "step": 36476
    },
    {
      "epoch": 1.9863184333585182,
      "grad_norm": 0.5428098946373274,
      "learning_rate": 2.4494317061540817e-08,
      "loss": 11.7523,
      "step": 36477
    },
    {
      "epoch": 1.9863728873551012,
      "grad_norm": 0.5526840885415635,
      "learning_rate": 2.4299539901950773e-08,
      "loss": 11.6715,
      "step": 36478
    },
    {
      "epoch": 1.9864273413516842,
      "grad_norm": 0.5000964365460533,
      "learning_rate": 2.410554017018285e-08,
      "loss": 11.7401,
      "step": 36479
    },
    {
      "epoch": 1.9864817953482672,
      "grad_norm": 0.5147461783983301,
      "learning_rate": 2.3912317867724742e-08,
      "loss": 11.7364,
      "step": 36480
    },
    {
      "epoch": 1.9865362493448504,
      "grad_norm": 0.8427748965577434,
      "learning_rate": 2.3719872996097457e-08,
      "loss": 11.7933,
      "step": 36481
    },
    {
      "epoch": 1.9865907033414334,
      "grad_norm": 0.4886377342918437,
      "learning_rate": 2.3528205556788697e-08,
      "loss": 11.8437,
      "step": 36482
    },
    {
      "epoch": 1.9866451573380164,
      "grad_norm": 0.5983192921560333,
      "learning_rate": 2.3337315551297257e-08,
      "loss": 11.9141,
      "step": 36483
    },
    {
      "epoch": 1.9866996113345994,
      "grad_norm": 0.553652626616461,
      "learning_rate": 2.3147202981099738e-08,
      "loss": 11.8146,
      "step": 36484
    },
    {
      "epoch": 1.9867540653311824,
      "grad_norm": 0.6418428452775194,
      "learning_rate": 2.2957867847661628e-08,
      "loss": 11.8464,
      "step": 36485
    },
    {
      "epoch": 1.9868085193277654,
      "grad_norm": 0.5455281673631264,
      "learning_rate": 2.2769310152481736e-08,
      "loss": 11.82,
      "step": 36486
    },
    {
      "epoch": 1.9868629733243486,
      "grad_norm": 0.7700380742973841,
      "learning_rate": 2.2581529897014454e-08,
      "loss": 11.8567,
      "step": 36487
    },
    {
      "epoch": 1.9869174273209316,
      "grad_norm": 0.508514097771357,
      "learning_rate": 2.2394527082714168e-08,
      "loss": 11.7757,
      "step": 36488
    },
    {
      "epoch": 1.9869718813175146,
      "grad_norm": 0.5886567007444015,
      "learning_rate": 2.2208301711046376e-08,
      "loss": 11.8596,
      "step": 36489
    },
    {
      "epoch": 1.9870263353140976,
      "grad_norm": 0.5549673988248771,
      "learning_rate": 2.2022853783443263e-08,
      "loss": 11.7638,
      "step": 36490
    },
    {
      "epoch": 1.9870807893106806,
      "grad_norm": 0.5587925331336404,
      "learning_rate": 2.183818330137033e-08,
      "loss": 11.8355,
      "step": 36491
    },
    {
      "epoch": 1.9871352433072635,
      "grad_norm": 0.5328003337131507,
      "learning_rate": 2.1654290266237553e-08,
      "loss": 11.6843,
      "step": 36492
    },
    {
      "epoch": 1.9871896973038465,
      "grad_norm": 0.5352286581011919,
      "learning_rate": 2.147117467949933e-08,
      "loss": 11.8615,
      "step": 36493
    },
    {
      "epoch": 1.9872441513004295,
      "grad_norm": 0.521618915135555,
      "learning_rate": 2.1288836542554537e-08,
      "loss": 11.6261,
      "step": 36494
    },
    {
      "epoch": 1.9872986052970125,
      "grad_norm": 0.5312015028180979,
      "learning_rate": 2.1107275856846466e-08,
      "loss": 11.6559,
      "step": 36495
    },
    {
      "epoch": 1.9873530592935955,
      "grad_norm": 0.5332096331879077,
      "learning_rate": 2.0926492623762894e-08,
      "loss": 11.8785,
      "step": 36496
    },
    {
      "epoch": 1.9874075132901785,
      "grad_norm": 0.5424721875130121,
      "learning_rate": 2.074648684472491e-08,
      "loss": 11.6787,
      "step": 36497
    },
    {
      "epoch": 1.9874619672867615,
      "grad_norm": 0.5406570202646456,
      "learning_rate": 2.0567258521131394e-08,
      "loss": 11.6913,
      "step": 36498
    },
    {
      "epoch": 1.9875164212833445,
      "grad_norm": 0.5796295827097839,
      "learning_rate": 2.0388807654370125e-08,
      "loss": 11.8214,
      "step": 36499
    },
    {
      "epoch": 1.9875708752799275,
      "grad_norm": 0.5368021958895797,
      "learning_rate": 2.021113424583998e-08,
      "loss": 11.8478,
      "step": 36500
    },
    {
      "epoch": 1.9876253292765105,
      "grad_norm": 0.48074724813872993,
      "learning_rate": 2.0034238296906537e-08,
      "loss": 11.5588,
      "step": 36501
    },
    {
      "epoch": 1.9876797832730935,
      "grad_norm": 0.5065848752096026,
      "learning_rate": 1.9858119808957575e-08,
      "loss": 11.7973,
      "step": 36502
    },
    {
      "epoch": 1.9877342372696765,
      "grad_norm": 0.5759036275316106,
      "learning_rate": 1.968277878335867e-08,
      "loss": 11.7654,
      "step": 36503
    },
    {
      "epoch": 1.9877886912662597,
      "grad_norm": 0.5591605320109206,
      "learning_rate": 1.950821522147539e-08,
      "loss": 11.6315,
      "step": 36504
    },
    {
      "epoch": 1.9878431452628427,
      "grad_norm": 0.5598232048830637,
      "learning_rate": 1.933442912465111e-08,
      "loss": 11.8587,
      "step": 36505
    },
    {
      "epoch": 1.9878975992594257,
      "grad_norm": 0.5734144479691698,
      "learning_rate": 1.9161420494262505e-08,
      "loss": 11.6655,
      "step": 36506
    },
    {
      "epoch": 1.9879520532560087,
      "grad_norm": 0.5181470637516574,
      "learning_rate": 1.8989189331641843e-08,
      "loss": 11.8563,
      "step": 36507
    },
    {
      "epoch": 1.9880065072525916,
      "grad_norm": 0.5678972236587222,
      "learning_rate": 1.881773563812139e-08,
      "loss": 11.7367,
      "step": 36508
    },
    {
      "epoch": 1.9880609612491746,
      "grad_norm": 0.5275689525463271,
      "learning_rate": 1.864705941503342e-08,
      "loss": 11.5711,
      "step": 36509
    },
    {
      "epoch": 1.9881154152457579,
      "grad_norm": 0.5226776011583704,
      "learning_rate": 1.8477160663721295e-08,
      "loss": 11.8611,
      "step": 36510
    },
    {
      "epoch": 1.9881698692423408,
      "grad_norm": 0.5907563497459758,
      "learning_rate": 1.830803938550618e-08,
      "loss": 11.9167,
      "step": 36511
    },
    {
      "epoch": 1.9882243232389238,
      "grad_norm": 0.5558311979413025,
      "learning_rate": 1.8139695581687043e-08,
      "loss": 11.8081,
      "step": 36512
    },
    {
      "epoch": 1.9882787772355068,
      "grad_norm": 0.5408375610092016,
      "learning_rate": 1.7972129253573942e-08,
      "loss": 11.8025,
      "step": 36513
    },
    {
      "epoch": 1.9883332312320898,
      "grad_norm": 0.6487321049336314,
      "learning_rate": 1.7805340402488046e-08,
      "loss": 11.9208,
      "step": 36514
    },
    {
      "epoch": 1.9883876852286728,
      "grad_norm": 0.5331738608573782,
      "learning_rate": 1.763932902970611e-08,
      "loss": 11.8855,
      "step": 36515
    },
    {
      "epoch": 1.9884421392252558,
      "grad_norm": 0.49844084490852403,
      "learning_rate": 1.7474095136538193e-08,
      "loss": 11.746,
      "step": 36516
    },
    {
      "epoch": 1.9884965932218388,
      "grad_norm": 0.5258561191422564,
      "learning_rate": 1.7309638724249954e-08,
      "loss": 11.7594,
      "step": 36517
    },
    {
      "epoch": 1.9885510472184218,
      "grad_norm": 0.477293214613775,
      "learning_rate": 1.714595979412925e-08,
      "loss": 11.6868,
      "step": 36518
    },
    {
      "epoch": 1.9886055012150048,
      "grad_norm": 0.5407193248464377,
      "learning_rate": 1.698305834745284e-08,
      "loss": 11.7135,
      "step": 36519
    },
    {
      "epoch": 1.9886599552115878,
      "grad_norm": 0.5378190096474488,
      "learning_rate": 1.6820934385475272e-08,
      "loss": 11.7109,
      "step": 36520
    },
    {
      "epoch": 1.9887144092081708,
      "grad_norm": 0.5317356370743387,
      "learning_rate": 1.6659587909473307e-08,
      "loss": 11.742,
      "step": 36521
    },
    {
      "epoch": 1.9887688632047538,
      "grad_norm": 0.5760190090144073,
      "learning_rate": 1.649901892069039e-08,
      "loss": 11.8167,
      "step": 36522
    },
    {
      "epoch": 1.9888233172013368,
      "grad_norm": 0.5291254998874673,
      "learning_rate": 1.6339227420381076e-08,
      "loss": 11.7414,
      "step": 36523
    },
    {
      "epoch": 1.9888777711979198,
      "grad_norm": 0.5248665097299513,
      "learning_rate": 1.618021340977771e-08,
      "loss": 11.7631,
      "step": 36524
    },
    {
      "epoch": 1.9889322251945027,
      "grad_norm": 0.5435504748659484,
      "learning_rate": 1.602197689013485e-08,
      "loss": 11.6174,
      "step": 36525
    },
    {
      "epoch": 1.9889866791910857,
      "grad_norm": 0.5212935552363418,
      "learning_rate": 1.5864517862662632e-08,
      "loss": 11.8869,
      "step": 36526
    },
    {
      "epoch": 1.9890411331876687,
      "grad_norm": 0.5796327661169799,
      "learning_rate": 1.5707836328604508e-08,
      "loss": 11.7896,
      "step": 36527
    },
    {
      "epoch": 1.989095587184252,
      "grad_norm": 0.535665979283029,
      "learning_rate": 1.5551932289170623e-08,
      "loss": 11.73,
      "step": 36528
    },
    {
      "epoch": 1.989150041180835,
      "grad_norm": 0.5853900510831671,
      "learning_rate": 1.5396805745560018e-08,
      "loss": 11.657,
      "step": 36529
    },
    {
      "epoch": 1.989204495177418,
      "grad_norm": 1.653872634805348,
      "learning_rate": 1.5242456699005037e-08,
      "loss": 11.9419,
      "step": 36530
    },
    {
      "epoch": 1.989258949174001,
      "grad_norm": 0.5540657531119669,
      "learning_rate": 1.508888515068252e-08,
      "loss": 11.7679,
      "step": 36531
    },
    {
      "epoch": 1.989313403170584,
      "grad_norm": 0.5260239814293589,
      "learning_rate": 1.4936091101802608e-08,
      "loss": 11.8596,
      "step": 36532
    },
    {
      "epoch": 1.989367857167167,
      "grad_norm": 0.5370210294461955,
      "learning_rate": 1.4784074553542138e-08,
      "loss": 11.7527,
      "step": 36533
    },
    {
      "epoch": 1.9894223111637501,
      "grad_norm": 0.5940283262279022,
      "learning_rate": 1.4632835507089049e-08,
      "loss": 11.9127,
      "step": 36534
    },
    {
      "epoch": 1.9894767651603331,
      "grad_norm": 0.5591647115083277,
      "learning_rate": 1.4482373963620177e-08,
      "loss": 11.7366,
      "step": 36535
    },
    {
      "epoch": 1.989531219156916,
      "grad_norm": 0.5998355448929859,
      "learning_rate": 1.4332689924301257e-08,
      "loss": 11.8799,
      "step": 36536
    },
    {
      "epoch": 1.989585673153499,
      "grad_norm": 0.5143044566579497,
      "learning_rate": 1.4183783390298022e-08,
      "loss": 11.778,
      "step": 36537
    },
    {
      "epoch": 1.989640127150082,
      "grad_norm": 0.5242737544104311,
      "learning_rate": 1.4035654362776207e-08,
      "loss": 11.6855,
      "step": 36538
    },
    {
      "epoch": 1.989694581146665,
      "grad_norm": 0.5185734968571442,
      "learning_rate": 1.3888302842868239e-08,
      "loss": 11.6778,
      "step": 36539
    },
    {
      "epoch": 1.989749035143248,
      "grad_norm": 0.5775111635310636,
      "learning_rate": 1.3741728831750955e-08,
      "loss": 11.7708,
      "step": 36540
    },
    {
      "epoch": 1.989803489139831,
      "grad_norm": 0.533025138209696,
      "learning_rate": 1.3595932330534578e-08,
      "loss": 11.7419,
      "step": 36541
    },
    {
      "epoch": 1.989857943136414,
      "grad_norm": 0.5913819883562282,
      "learning_rate": 1.3450913340362637e-08,
      "loss": 11.8257,
      "step": 36542
    },
    {
      "epoch": 1.989912397132997,
      "grad_norm": 0.5666547350860178,
      "learning_rate": 1.330667186236756e-08,
      "loss": 11.7689,
      "step": 36543
    },
    {
      "epoch": 1.98996685112958,
      "grad_norm": 0.5617870311779647,
      "learning_rate": 1.3163207897659569e-08,
      "loss": 11.7593,
      "step": 36544
    },
    {
      "epoch": 1.990021305126163,
      "grad_norm": 0.5278297550209919,
      "learning_rate": 1.3020521447371093e-08,
      "loss": 11.761,
      "step": 36545
    },
    {
      "epoch": 1.990075759122746,
      "grad_norm": 0.55048504306361,
      "learning_rate": 1.2878612512601251e-08,
      "loss": 11.7731,
      "step": 36546
    },
    {
      "epoch": 1.990130213119329,
      "grad_norm": 0.5374731388214943,
      "learning_rate": 1.2737481094460269e-08,
      "loss": 11.7788,
      "step": 36547
    },
    {
      "epoch": 1.990184667115912,
      "grad_norm": 0.577594614368669,
      "learning_rate": 1.2597127194036163e-08,
      "loss": 11.8198,
      "step": 36548
    },
    {
      "epoch": 1.990239121112495,
      "grad_norm": 0.5727313297973341,
      "learning_rate": 1.2457550812416951e-08,
      "loss": 11.7288,
      "step": 36549
    },
    {
      "epoch": 1.990293575109078,
      "grad_norm": 0.5253241306983374,
      "learning_rate": 1.2318751950701757e-08,
      "loss": 11.7712,
      "step": 36550
    },
    {
      "epoch": 1.9903480291056612,
      "grad_norm": 0.5600070923392805,
      "learning_rate": 1.2180730609967495e-08,
      "loss": 11.6915,
      "step": 36551
    },
    {
      "epoch": 1.9904024831022442,
      "grad_norm": 0.5692639945535776,
      "learning_rate": 1.2043486791279978e-08,
      "loss": 11.8965,
      "step": 36552
    },
    {
      "epoch": 1.9904569370988272,
      "grad_norm": 0.5878580842199417,
      "learning_rate": 1.1907020495716126e-08,
      "loss": 11.6622,
      "step": 36553
    },
    {
      "epoch": 1.9905113910954102,
      "grad_norm": 0.5272999207693095,
      "learning_rate": 1.1771331724319545e-08,
      "loss": 11.6701,
      "step": 36554
    },
    {
      "epoch": 1.9905658450919932,
      "grad_norm": 0.5069394035366999,
      "learning_rate": 1.1636420478167154e-08,
      "loss": 11.6037,
      "step": 36555
    },
    {
      "epoch": 1.9906202990885762,
      "grad_norm": 0.5307850074672694,
      "learning_rate": 1.1502286758291459e-08,
      "loss": 11.7663,
      "step": 36556
    },
    {
      "epoch": 1.9906747530851594,
      "grad_norm": 0.5553898205062658,
      "learning_rate": 1.136893056574717e-08,
      "loss": 11.8379,
      "step": 36557
    },
    {
      "epoch": 1.9907292070817424,
      "grad_norm": 0.5542884498644787,
      "learning_rate": 1.12363519015668e-08,
      "loss": 11.769,
      "step": 36558
    },
    {
      "epoch": 1.9907836610783254,
      "grad_norm": 0.5387905649061595,
      "learning_rate": 1.110455076678285e-08,
      "loss": 11.7206,
      "step": 36559
    },
    {
      "epoch": 1.9908381150749084,
      "grad_norm": 0.5412154444315345,
      "learning_rate": 1.0973527162405628e-08,
      "loss": 11.8113,
      "step": 36560
    },
    {
      "epoch": 1.9908925690714914,
      "grad_norm": 0.5987705883005071,
      "learning_rate": 1.0843281089478741e-08,
      "loss": 11.843,
      "step": 36561
    },
    {
      "epoch": 1.9909470230680744,
      "grad_norm": 0.6542224568495725,
      "learning_rate": 1.071381254900139e-08,
      "loss": 11.8524,
      "step": 36562
    },
    {
      "epoch": 1.9910014770646574,
      "grad_norm": 0.5760613911943637,
      "learning_rate": 1.0585121541972776e-08,
      "loss": 11.6326,
      "step": 36563
    },
    {
      "epoch": 1.9910559310612403,
      "grad_norm": 0.5359531398178513,
      "learning_rate": 1.0457208069414304e-08,
      "loss": 11.8028,
      "step": 36564
    },
    {
      "epoch": 1.9911103850578233,
      "grad_norm": 0.471814764239132,
      "learning_rate": 1.033007213230297e-08,
      "loss": 11.7115,
      "step": 36565
    },
    {
      "epoch": 1.9911648390544063,
      "grad_norm": 0.5769235638220788,
      "learning_rate": 1.0203713731626873e-08,
      "loss": 11.8379,
      "step": 36566
    },
    {
      "epoch": 1.9912192930509893,
      "grad_norm": 0.5238240480289093,
      "learning_rate": 1.0078132868385215e-08,
      "loss": 11.8243,
      "step": 36567
    },
    {
      "epoch": 1.9912737470475723,
      "grad_norm": 0.5364322950304945,
      "learning_rate": 9.953329543543888e-09,
      "loss": 11.6682,
      "step": 36568
    },
    {
      "epoch": 1.9913282010441553,
      "grad_norm": 0.593512180407724,
      "learning_rate": 9.829303758068786e-09,
      "loss": 11.8686,
      "step": 36569
    },
    {
      "epoch": 1.9913826550407383,
      "grad_norm": 0.5605312415944048,
      "learning_rate": 9.706055512925804e-09,
      "loss": 11.8327,
      "step": 36570
    },
    {
      "epoch": 1.9914371090373213,
      "grad_norm": 0.5514882556214906,
      "learning_rate": 9.583584809080837e-09,
      "loss": 11.7296,
      "step": 36571
    },
    {
      "epoch": 1.9914915630339043,
      "grad_norm": 0.5429878775553492,
      "learning_rate": 9.461891647488675e-09,
      "loss": 11.7423,
      "step": 36572
    },
    {
      "epoch": 1.9915460170304873,
      "grad_norm": 0.5691855765732323,
      "learning_rate": 9.340976029081905e-09,
      "loss": 11.7875,
      "step": 36573
    },
    {
      "epoch": 1.9916004710270705,
      "grad_norm": 0.5521564499331408,
      "learning_rate": 9.22083795481532e-09,
      "loss": 11.8163,
      "step": 36574
    },
    {
      "epoch": 1.9916549250236535,
      "grad_norm": 0.5701652563157238,
      "learning_rate": 9.101477425610405e-09,
      "loss": 11.8349,
      "step": 36575
    },
    {
      "epoch": 1.9917093790202365,
      "grad_norm": 0.5855597507007764,
      "learning_rate": 8.982894442399747e-09,
      "loss": 11.807,
      "step": 36576
    },
    {
      "epoch": 1.9917638330168195,
      "grad_norm": 0.5525490524446289,
      "learning_rate": 8.865089006104832e-09,
      "loss": 11.7881,
      "step": 36577
    },
    {
      "epoch": 1.9918182870134025,
      "grad_norm": 0.5457276094545386,
      "learning_rate": 8.748061117647144e-09,
      "loss": 11.7254,
      "step": 36578
    },
    {
      "epoch": 1.9918727410099855,
      "grad_norm": 0.6052376554592634,
      "learning_rate": 8.631810777937067e-09,
      "loss": 11.7223,
      "step": 36579
    },
    {
      "epoch": 1.9919271950065687,
      "grad_norm": 0.5890097010004732,
      "learning_rate": 8.516337987862777e-09,
      "loss": 11.8369,
      "step": 36580
    },
    {
      "epoch": 1.9919816490031517,
      "grad_norm": 0.5788997216648428,
      "learning_rate": 8.401642748345762e-09,
      "loss": 11.7819,
      "step": 36581
    },
    {
      "epoch": 1.9920361029997347,
      "grad_norm": 0.6038637351962896,
      "learning_rate": 8.287725060263097e-09,
      "loss": 11.5915,
      "step": 36582
    },
    {
      "epoch": 1.9920905569963177,
      "grad_norm": 0.5987582867117255,
      "learning_rate": 8.17458492450296e-09,
      "loss": 11.79,
      "step": 36583
    },
    {
      "epoch": 1.9921450109929006,
      "grad_norm": 0.5258476968420109,
      "learning_rate": 8.062222341942427e-09,
      "loss": 11.7373,
      "step": 36584
    },
    {
      "epoch": 1.9921994649894836,
      "grad_norm": 0.5378773063919398,
      "learning_rate": 7.950637313458576e-09,
      "loss": 11.83,
      "step": 36585
    },
    {
      "epoch": 1.9922539189860666,
      "grad_norm": 0.5388884010314146,
      "learning_rate": 7.839829839928481e-09,
      "loss": 11.6727,
      "step": 36586
    },
    {
      "epoch": 1.9923083729826496,
      "grad_norm": 0.52512221499079,
      "learning_rate": 7.729799922195912e-09,
      "loss": 11.7655,
      "step": 36587
    },
    {
      "epoch": 1.9923628269792326,
      "grad_norm": 0.548364727159414,
      "learning_rate": 7.620547561126845e-09,
      "loss": 11.6767,
      "step": 36588
    },
    {
      "epoch": 1.9924172809758156,
      "grad_norm": 0.5262306886882376,
      "learning_rate": 7.51207275757615e-09,
      "loss": 11.7849,
      "step": 36589
    },
    {
      "epoch": 1.9924717349723986,
      "grad_norm": 0.5296609115846772,
      "learning_rate": 7.404375512365391e-09,
      "loss": 11.727,
      "step": 36590
    },
    {
      "epoch": 1.9925261889689816,
      "grad_norm": 0.5197247939044222,
      "learning_rate": 7.297455826360544e-09,
      "loss": 11.74,
      "step": 36591
    },
    {
      "epoch": 1.9925806429655646,
      "grad_norm": 0.5061356816214125,
      "learning_rate": 7.191313700372071e-09,
      "loss": 11.5471,
      "step": 36592
    },
    {
      "epoch": 1.9926350969621476,
      "grad_norm": 0.6189454476855245,
      "learning_rate": 7.0859491352437415e-09,
      "loss": 11.8413,
      "step": 36593
    },
    {
      "epoch": 1.9926895509587306,
      "grad_norm": 0.5116906400228126,
      "learning_rate": 6.981362131774916e-09,
      "loss": 11.6078,
      "step": 36594
    },
    {
      "epoch": 1.9927440049553136,
      "grad_norm": 0.5443973249715622,
      "learning_rate": 6.8775526907871594e-09,
      "loss": 11.802,
      "step": 36595
    },
    {
      "epoch": 1.9927984589518966,
      "grad_norm": 0.5226089155217758,
      "learning_rate": 6.7745208130909345e-09,
      "loss": 11.8135,
      "step": 36596
    },
    {
      "epoch": 1.9928529129484795,
      "grad_norm": 0.5249070542445562,
      "learning_rate": 6.6722664994744995e-09,
      "loss": 11.753,
      "step": 36597
    },
    {
      "epoch": 1.9929073669450628,
      "grad_norm": 0.5731575805681722,
      "learning_rate": 6.570789750759421e-09,
      "loss": 11.7456,
      "step": 36598
    },
    {
      "epoch": 1.9929618209416458,
      "grad_norm": 0.6046180240218758,
      "learning_rate": 6.470090567711751e-09,
      "loss": 11.8848,
      "step": 36599
    },
    {
      "epoch": 1.9930162749382287,
      "grad_norm": 0.5238906110334236,
      "learning_rate": 6.370168951119748e-09,
      "loss": 11.6064,
      "step": 36600
    },
    {
      "epoch": 1.9930707289348117,
      "grad_norm": 0.5442536443886653,
      "learning_rate": 6.27102490176057e-09,
      "loss": 11.8224,
      "step": 36601
    },
    {
      "epoch": 1.9931251829313947,
      "grad_norm": 0.6052913011545425,
      "learning_rate": 6.17265842041137e-09,
      "loss": 11.8681,
      "step": 36602
    },
    {
      "epoch": 1.993179636927978,
      "grad_norm": 0.5277548338212886,
      "learning_rate": 6.075069507838205e-09,
      "loss": 11.7395,
      "step": 36603
    },
    {
      "epoch": 1.993234090924561,
      "grad_norm": 0.6245102734589224,
      "learning_rate": 5.978258164784922e-09,
      "loss": 11.8652,
      "step": 36604
    },
    {
      "epoch": 1.993288544921144,
      "grad_norm": 0.5429962968186917,
      "learning_rate": 5.882224392017577e-09,
      "loss": 11.5717,
      "step": 36605
    },
    {
      "epoch": 1.993342998917727,
      "grad_norm": 0.5307810892273391,
      "learning_rate": 5.786968190280018e-09,
      "loss": 11.7404,
      "step": 36606
    },
    {
      "epoch": 1.99339745291431,
      "grad_norm": 0.515158080338061,
      "learning_rate": 5.692489560304992e-09,
      "loss": 11.6824,
      "step": 36607
    },
    {
      "epoch": 1.993451906910893,
      "grad_norm": 0.5087382019976419,
      "learning_rate": 5.598788502847452e-09,
      "loss": 11.787,
      "step": 36608
    },
    {
      "epoch": 1.993506360907476,
      "grad_norm": 0.5465389543262851,
      "learning_rate": 5.505865018606837e-09,
      "loss": 11.7854,
      "step": 36609
    },
    {
      "epoch": 1.993560814904059,
      "grad_norm": 0.5229986948256299,
      "learning_rate": 5.4137191083381e-09,
      "loss": 11.747,
      "step": 36610
    },
    {
      "epoch": 1.993615268900642,
      "grad_norm": 0.5524090689207177,
      "learning_rate": 5.322350772729579e-09,
      "loss": 11.8494,
      "step": 36611
    },
    {
      "epoch": 1.9936697228972249,
      "grad_norm": 0.6229786307791355,
      "learning_rate": 5.231760012502917e-09,
      "loss": 11.9247,
      "step": 36612
    },
    {
      "epoch": 1.9937241768938079,
      "grad_norm": 0.5276054388819608,
      "learning_rate": 5.14194682836866e-09,
      "loss": 11.7887,
      "step": 36613
    },
    {
      "epoch": 1.9937786308903909,
      "grad_norm": 0.5685137900197232,
      "learning_rate": 5.052911221015144e-09,
      "loss": 11.843,
      "step": 36614
    },
    {
      "epoch": 1.9938330848869739,
      "grad_norm": 0.6003547945268849,
      "learning_rate": 4.96465319114181e-09,
      "loss": 11.8157,
      "step": 36615
    },
    {
      "epoch": 1.9938875388835569,
      "grad_norm": 0.503537665410977,
      "learning_rate": 4.877172739436997e-09,
      "loss": 11.7783,
      "step": 36616
    },
    {
      "epoch": 1.9939419928801398,
      "grad_norm": 0.5524766885056016,
      "learning_rate": 4.7904698665779405e-09,
      "loss": 11.8226,
      "step": 36617
    },
    {
      "epoch": 1.9939964468767228,
      "grad_norm": 0.5839317510664969,
      "learning_rate": 4.704544573230774e-09,
      "loss": 11.7921,
      "step": 36618
    },
    {
      "epoch": 1.9940509008733058,
      "grad_norm": 0.4937412077674242,
      "learning_rate": 4.619396860072734e-09,
      "loss": 11.7857,
      "step": 36619
    },
    {
      "epoch": 1.9941053548698888,
      "grad_norm": 0.619428771642005,
      "learning_rate": 4.535026727758851e-09,
      "loss": 11.7916,
      "step": 36620
    },
    {
      "epoch": 1.994159808866472,
      "grad_norm": 0.5241669488118829,
      "learning_rate": 4.45143417695526e-09,
      "loss": 11.7554,
      "step": 36621
    },
    {
      "epoch": 1.994214262863055,
      "grad_norm": 0.6134667912022922,
      "learning_rate": 4.3686192083058904e-09,
      "loss": 11.9199,
      "step": 36622
    },
    {
      "epoch": 1.994268716859638,
      "grad_norm": 0.5666420771301303,
      "learning_rate": 4.286581822454672e-09,
      "loss": 11.688,
      "step": 36623
    },
    {
      "epoch": 1.994323170856221,
      "grad_norm": 0.6716040216061332,
      "learning_rate": 4.20532202003443e-09,
      "loss": 11.901,
      "step": 36624
    },
    {
      "epoch": 1.994377624852804,
      "grad_norm": 0.5038869191413636,
      "learning_rate": 4.124839801689095e-09,
      "loss": 11.7773,
      "step": 36625
    },
    {
      "epoch": 1.994432078849387,
      "grad_norm": 0.5304847927368819,
      "learning_rate": 4.045135168040392e-09,
      "loss": 11.7209,
      "step": 36626
    },
    {
      "epoch": 1.9944865328459702,
      "grad_norm": 0.5834841566192898,
      "learning_rate": 3.966208119698944e-09,
      "loss": 11.7493,
      "step": 36627
    },
    {
      "epoch": 1.9945409868425532,
      "grad_norm": 0.5297166483552079,
      "learning_rate": 3.888058657297577e-09,
      "loss": 11.8569,
      "step": 36628
    },
    {
      "epoch": 1.9945954408391362,
      "grad_norm": 0.5238660667981947,
      "learning_rate": 3.8106867814136086e-09,
      "loss": 11.8502,
      "step": 36629
    },
    {
      "epoch": 1.9946498948357192,
      "grad_norm": 0.5234309470475411,
      "learning_rate": 3.734092492679864e-09,
      "loss": 11.6249,
      "step": 36630
    },
    {
      "epoch": 1.9947043488323022,
      "grad_norm": 0.5139694921264127,
      "learning_rate": 3.658275791684762e-09,
      "loss": 11.7672,
      "step": 36631
    },
    {
      "epoch": 1.9947588028288852,
      "grad_norm": 0.5264909707605255,
      "learning_rate": 3.583236679005619e-09,
      "loss": 11.7835,
      "step": 36632
    },
    {
      "epoch": 1.9948132568254682,
      "grad_norm": 0.5242808653399074,
      "learning_rate": 3.5089751552308537e-09,
      "loss": 11.6735,
      "step": 36633
    },
    {
      "epoch": 1.9948677108220512,
      "grad_norm": 0.5194141757764744,
      "learning_rate": 3.435491220937781e-09,
      "loss": 11.7064,
      "step": 36634
    },
    {
      "epoch": 1.9949221648186342,
      "grad_norm": 0.5211938629199598,
      "learning_rate": 3.362784876703717e-09,
      "loss": 11.7095,
      "step": 36635
    },
    {
      "epoch": 1.9949766188152172,
      "grad_norm": 0.6333757755520796,
      "learning_rate": 3.290856123083774e-09,
      "loss": 11.7457,
      "step": 36636
    },
    {
      "epoch": 1.9950310728118001,
      "grad_norm": 0.5435958077205914,
      "learning_rate": 3.219704960655268e-09,
      "loss": 11.7806,
      "step": 36637
    },
    {
      "epoch": 1.9950855268083831,
      "grad_norm": 0.5210460749138116,
      "learning_rate": 3.1493313899511046e-09,
      "loss": 11.7653,
      "step": 36638
    },
    {
      "epoch": 1.9951399808049661,
      "grad_norm": 0.5117830193675624,
      "learning_rate": 3.0797354115263964e-09,
      "loss": 11.8055,
      "step": 36639
    },
    {
      "epoch": 1.9951944348015491,
      "grad_norm": 0.514988069970701,
      "learning_rate": 3.0109170259251528e-09,
      "loss": 11.6839,
      "step": 36640
    },
    {
      "epoch": 1.9952488887981321,
      "grad_norm": 0.5542102234589764,
      "learning_rate": 2.942876233680281e-09,
      "loss": 11.7812,
      "step": 36641
    },
    {
      "epoch": 1.995303342794715,
      "grad_norm": 0.516034683978556,
      "learning_rate": 2.8756130353246867e-09,
      "loss": 11.8199,
      "step": 36642
    },
    {
      "epoch": 1.995357796791298,
      "grad_norm": 0.5155532397239551,
      "learning_rate": 2.8091274313801762e-09,
      "loss": 11.7568,
      "step": 36643
    },
    {
      "epoch": 1.9954122507878813,
      "grad_norm": 0.6329309551328337,
      "learning_rate": 2.743419422357452e-09,
      "loss": 11.6283,
      "step": 36644
    },
    {
      "epoch": 1.9954667047844643,
      "grad_norm": 0.4963986603524539,
      "learning_rate": 2.6784890087672154e-09,
      "loss": 11.7111,
      "step": 36645
    },
    {
      "epoch": 1.9955211587810473,
      "grad_norm": 0.6076339871650054,
      "learning_rate": 2.61433619112017e-09,
      "loss": 11.8352,
      "step": 36646
    },
    {
      "epoch": 1.9955756127776303,
      "grad_norm": 0.5234783332036234,
      "learning_rate": 2.5509609699159163e-09,
      "loss": 11.7935,
      "step": 36647
    },
    {
      "epoch": 1.9956300667742133,
      "grad_norm": 0.5392005307077706,
      "learning_rate": 2.4883633456429524e-09,
      "loss": 11.7426,
      "step": 36648
    },
    {
      "epoch": 1.9956845207707963,
      "grad_norm": 0.5887162033689871,
      "learning_rate": 2.4265433187897757e-09,
      "loss": 11.8386,
      "step": 36649
    },
    {
      "epoch": 1.9957389747673795,
      "grad_norm": 0.5396301072348566,
      "learning_rate": 2.365500889844885e-09,
      "loss": 11.7071,
      "step": 36650
    },
    {
      "epoch": 1.9957934287639625,
      "grad_norm": 0.575394282904709,
      "learning_rate": 2.3052360592745738e-09,
      "loss": 11.7565,
      "step": 36651
    },
    {
      "epoch": 1.9958478827605455,
      "grad_norm": 0.770807824973402,
      "learning_rate": 2.245748827534033e-09,
      "loss": 11.8931,
      "step": 36652
    },
    {
      "epoch": 1.9959023367571285,
      "grad_norm": 0.5378501184092315,
      "learning_rate": 2.1870391951117618e-09,
      "loss": 11.7475,
      "step": 36653
    },
    {
      "epoch": 1.9959567907537115,
      "grad_norm": 0.5328546146445313,
      "learning_rate": 2.129107162451849e-09,
      "loss": 11.7567,
      "step": 36654
    },
    {
      "epoch": 1.9960112447502945,
      "grad_norm": 0.5642278048320595,
      "learning_rate": 2.0719527300094853e-09,
      "loss": 11.7792,
      "step": 36655
    },
    {
      "epoch": 1.9960656987468774,
      "grad_norm": 0.5407460073021433,
      "learning_rate": 2.0155758982176588e-09,
      "loss": 11.8449,
      "step": 36656
    },
    {
      "epoch": 1.9961201527434604,
      "grad_norm": 0.5918623888172896,
      "learning_rate": 1.959976667520458e-09,
      "loss": 11.8475,
      "step": 36657
    },
    {
      "epoch": 1.9961746067400434,
      "grad_norm": 0.5208077918237171,
      "learning_rate": 1.905155038361972e-09,
      "loss": 11.715,
      "step": 36658
    },
    {
      "epoch": 1.9962290607366264,
      "grad_norm": 0.5547570054299232,
      "learning_rate": 1.8511110111529839e-09,
      "loss": 11.7199,
      "step": 36659
    },
    {
      "epoch": 1.9962835147332094,
      "grad_norm": 0.5406713324055593,
      "learning_rate": 1.7978445863153782e-09,
      "loss": 11.8062,
      "step": 36660
    },
    {
      "epoch": 1.9963379687297924,
      "grad_norm": 0.5336696006867239,
      "learning_rate": 1.7453557642710393e-09,
      "loss": 11.8141,
      "step": 36661
    },
    {
      "epoch": 1.9963924227263754,
      "grad_norm": 0.5206807405477759,
      "learning_rate": 1.6936445454307504e-09,
      "loss": 11.7819,
      "step": 36662
    },
    {
      "epoch": 1.9964468767229584,
      "grad_norm": 0.5767948374382369,
      "learning_rate": 1.642710930183089e-09,
      "loss": 11.772,
      "step": 36663
    },
    {
      "epoch": 1.9965013307195414,
      "grad_norm": 0.585163110915863,
      "learning_rate": 1.5925549189388377e-09,
      "loss": 11.7973,
      "step": 36664
    },
    {
      "epoch": 1.9965557847161244,
      "grad_norm": 0.5016347834194185,
      "learning_rate": 1.5431765120754727e-09,
      "loss": 11.7421,
      "step": 36665
    },
    {
      "epoch": 1.9966102387127074,
      "grad_norm": 0.517762633363895,
      "learning_rate": 1.4945757099815716e-09,
      "loss": 11.7985,
      "step": 36666
    },
    {
      "epoch": 1.9966646927092904,
      "grad_norm": 0.5241380060054428,
      "learning_rate": 1.4467525130346105e-09,
      "loss": 11.6881,
      "step": 36667
    },
    {
      "epoch": 1.9967191467058736,
      "grad_norm": 0.5181787526820097,
      "learning_rate": 1.399706921612065e-09,
      "loss": 11.7816,
      "step": 36668
    },
    {
      "epoch": 1.9967736007024566,
      "grad_norm": 0.5505127884394181,
      "learning_rate": 1.3534389360692068e-09,
      "loss": 11.6572,
      "step": 36669
    },
    {
      "epoch": 1.9968280546990396,
      "grad_norm": 0.5693830525480125,
      "learning_rate": 1.3079485567724093e-09,
      "loss": 11.8402,
      "step": 36670
    },
    {
      "epoch": 1.9968825086956226,
      "grad_norm": 0.5860973761223902,
      "learning_rate": 1.2632357840769438e-09,
      "loss": 11.8876,
      "step": 36671
    },
    {
      "epoch": 1.9969369626922056,
      "grad_norm": 0.5704348010003918,
      "learning_rate": 1.2193006183269794e-09,
      "loss": 11.8041,
      "step": 36672
    },
    {
      "epoch": 1.9969914166887888,
      "grad_norm": 0.6279119288513697,
      "learning_rate": 1.1761430598666856e-09,
      "loss": 11.8818,
      "step": 36673
    },
    {
      "epoch": 1.9970458706853718,
      "grad_norm": 0.5198706796872268,
      "learning_rate": 1.133763109029129e-09,
      "loss": 11.7613,
      "step": 36674
    },
    {
      "epoch": 1.9971003246819548,
      "grad_norm": 0.6034201104231257,
      "learning_rate": 1.0921607661473765e-09,
      "loss": 11.8964,
      "step": 36675
    },
    {
      "epoch": 1.9971547786785377,
      "grad_norm": 0.49949102841187076,
      "learning_rate": 1.051336031543393e-09,
      "loss": 11.7939,
      "step": 36676
    },
    {
      "epoch": 1.9972092326751207,
      "grad_norm": 0.5363153137472171,
      "learning_rate": 1.0112889055391428e-09,
      "loss": 11.7587,
      "step": 36677
    },
    {
      "epoch": 1.9972636866717037,
      "grad_norm": 0.5810910374213107,
      "learning_rate": 9.720193884343865e-10,
      "loss": 11.724,
      "step": 36678
    },
    {
      "epoch": 1.9973181406682867,
      "grad_norm": 0.5210939134204752,
      "learning_rate": 9.335274805399863e-10,
      "loss": 11.6994,
      "step": 36679
    },
    {
      "epoch": 1.9973725946648697,
      "grad_norm": 0.5381576891492463,
      "learning_rate": 8.958131821557025e-10,
      "loss": 11.8286,
      "step": 36680
    },
    {
      "epoch": 1.9974270486614527,
      "grad_norm": 0.530220337335392,
      "learning_rate": 8.588764935812954e-10,
      "loss": 11.755,
      "step": 36681
    },
    {
      "epoch": 1.9974815026580357,
      "grad_norm": 0.5728679182487828,
      "learning_rate": 8.227174150943206e-10,
      "loss": 11.787,
      "step": 36682
    },
    {
      "epoch": 1.9975359566546187,
      "grad_norm": 0.5205716047367589,
      "learning_rate": 7.87335946972334e-10,
      "loss": 11.8618,
      "step": 36683
    },
    {
      "epoch": 1.9975904106512017,
      "grad_norm": 0.5241118339580726,
      "learning_rate": 7.527320895150958e-10,
      "loss": 11.5183,
      "step": 36684
    },
    {
      "epoch": 1.9976448646477847,
      "grad_norm": 0.5088502198791645,
      "learning_rate": 7.189058429557527e-10,
      "loss": 11.7576,
      "step": 36685
    },
    {
      "epoch": 1.9976993186443677,
      "grad_norm": 0.5470162434631054,
      "learning_rate": 6.85857207594065e-10,
      "loss": 11.6734,
      "step": 36686
    },
    {
      "epoch": 1.9977537726409507,
      "grad_norm": 0.5110775201370584,
      "learning_rate": 6.535861836520773e-10,
      "loss": 11.7879,
      "step": 36687
    },
    {
      "epoch": 1.9978082266375337,
      "grad_norm": 0.5608037502661866,
      "learning_rate": 6.220927714073455e-10,
      "loss": 11.6387,
      "step": 36688
    },
    {
      "epoch": 1.9978626806341166,
      "grad_norm": 0.5188198153542338,
      "learning_rate": 5.913769710930161e-10,
      "loss": 11.8053,
      "step": 36689
    },
    {
      "epoch": 1.9979171346306996,
      "grad_norm": 0.5757468699154831,
      "learning_rate": 5.614387829533385e-10,
      "loss": 11.568,
      "step": 36690
    },
    {
      "epoch": 1.9979715886272829,
      "grad_norm": 0.5732020279908595,
      "learning_rate": 5.322782072103571e-10,
      "loss": 11.7973,
      "step": 36691
    },
    {
      "epoch": 1.9980260426238658,
      "grad_norm": 0.5661612494643344,
      "learning_rate": 5.038952441083211e-10,
      "loss": 11.8387,
      "step": 36692
    },
    {
      "epoch": 1.9980804966204488,
      "grad_norm": 0.5364384218893937,
      "learning_rate": 4.762898938470706e-10,
      "loss": 11.7906,
      "step": 36693
    },
    {
      "epoch": 1.9981349506170318,
      "grad_norm": 0.6917035468321034,
      "learning_rate": 4.494621566597523e-10,
      "loss": 11.8456,
      "step": 36694
    },
    {
      "epoch": 1.9981894046136148,
      "grad_norm": 0.5582074763770993,
      "learning_rate": 4.2341203274620653e-10,
      "loss": 11.7563,
      "step": 36695
    },
    {
      "epoch": 1.9982438586101978,
      "grad_norm": 0.5400802746154967,
      "learning_rate": 3.981395223062734e-10,
      "loss": 11.8059,
      "step": 36696
    },
    {
      "epoch": 1.998298312606781,
      "grad_norm": 0.6784869398514226,
      "learning_rate": 3.736446255508952e-10,
      "loss": 11.8526,
      "step": 36697
    },
    {
      "epoch": 1.998352766603364,
      "grad_norm": 0.5680346951485059,
      "learning_rate": 3.499273426466054e-10,
      "loss": 11.8861,
      "step": 36698
    },
    {
      "epoch": 1.998407220599947,
      "grad_norm": 0.5160730365803441,
      "learning_rate": 3.2698767379324426e-10,
      "loss": 11.7039,
      "step": 36699
    },
    {
      "epoch": 1.99846167459653,
      "grad_norm": 0.5123216668387516,
      "learning_rate": 3.0482561917954954e-10,
      "loss": 11.7522,
      "step": 36700
    },
    {
      "epoch": 1.998516128593113,
      "grad_norm": 0.5254371195560049,
      "learning_rate": 2.8344117894985036e-10,
      "loss": 11.8134,
      "step": 36701
    },
    {
      "epoch": 1.998570582589696,
      "grad_norm": 0.5854105432065376,
      "learning_rate": 2.6283435329288453e-10,
      "loss": 11.7867,
      "step": 36702
    },
    {
      "epoch": 1.998625036586279,
      "grad_norm": 0.5862551505034992,
      "learning_rate": 2.4300514236408336e-10,
      "loss": 11.9344,
      "step": 36703
    },
    {
      "epoch": 1.998679490582862,
      "grad_norm": 0.5140052659654867,
      "learning_rate": 2.23953546318878e-10,
      "loss": 11.7432,
      "step": 36704
    },
    {
      "epoch": 1.998733944579445,
      "grad_norm": 0.5604743133364141,
      "learning_rate": 2.0567956529049527e-10,
      "loss": 11.7821,
      "step": 36705
    },
    {
      "epoch": 1.998788398576028,
      "grad_norm": 0.5240572928921524,
      "learning_rate": 1.881831994454686e-10,
      "loss": 11.6996,
      "step": 36706
    },
    {
      "epoch": 1.998842852572611,
      "grad_norm": 0.5405593267654587,
      "learning_rate": 1.7146444889482026e-10,
      "loss": 11.711,
      "step": 36707
    },
    {
      "epoch": 1.998897306569194,
      "grad_norm": 0.5587512939355626,
      "learning_rate": 1.5552331379398155e-10,
      "loss": 11.8544,
      "step": 36708
    },
    {
      "epoch": 1.998951760565777,
      "grad_norm": 0.5475112401914279,
      "learning_rate": 1.403597942428725e-10,
      "loss": 11.7343,
      "step": 36709
    },
    {
      "epoch": 1.99900621456236,
      "grad_norm": 0.5490761635001964,
      "learning_rate": 1.2597389037471984e-10,
      "loss": 11.8575,
      "step": 36710
    },
    {
      "epoch": 1.999060668558943,
      "grad_norm": 0.5125875521480342,
      "learning_rate": 1.1236560230054593e-10,
      "loss": 11.6728,
      "step": 36711
    },
    {
      "epoch": 1.999115122555526,
      "grad_norm": 0.5796859431852404,
      "learning_rate": 9.953493010916859e-11,
      "loss": 11.8984,
      "step": 36712
    },
    {
      "epoch": 1.999169576552109,
      "grad_norm": 0.5733528789181468,
      "learning_rate": 8.748187392271234e-11,
      "loss": 11.8825,
      "step": 36713
    },
    {
      "epoch": 1.9992240305486921,
      "grad_norm": 0.5596695605625579,
      "learning_rate": 7.620643382999503e-11,
      "loss": 11.7445,
      "step": 36714
    },
    {
      "epoch": 1.9992784845452751,
      "grad_norm": 0.519350111406422,
      "learning_rate": 6.570860990873229e-11,
      "loss": 11.7687,
      "step": 36715
    },
    {
      "epoch": 1.9993329385418581,
      "grad_norm": 0.5701236866048116,
      "learning_rate": 5.598840224774193e-11,
      "loss": 11.8148,
      "step": 36716
    },
    {
      "epoch": 1.999387392538441,
      "grad_norm": 0.554621853887413,
      "learning_rate": 4.7045810924739586e-11,
      "loss": 11.7872,
      "step": 36717
    },
    {
      "epoch": 1.999441846535024,
      "grad_norm": 0.531288056186138,
      "learning_rate": 3.888083600633863e-11,
      "loss": 11.7555,
      "step": 36718
    },
    {
      "epoch": 1.999496300531607,
      "grad_norm": 0.541802893876769,
      "learning_rate": 3.149347754805021e-11,
      "loss": 11.7749,
      "step": 36719
    },
    {
      "epoch": 1.9995507545281903,
      "grad_norm": 0.6008596151002624,
      "learning_rate": 2.4883735627589944e-11,
      "loss": 11.77,
      "step": 36720
    },
    {
      "epoch": 1.9996052085247733,
      "grad_norm": 0.4876281226770037,
      "learning_rate": 1.905161026716229e-11,
      "loss": 11.7265,
      "step": 36721
    },
    {
      "epoch": 1.9996596625213563,
      "grad_norm": 0.5839723800419259,
      "learning_rate": 1.399710154448286e-11,
      "loss": 11.7743,
      "step": 36722
    },
    {
      "epoch": 1.9997141165179393,
      "grad_norm": 0.570160454135997,
      "learning_rate": 9.720209470653885e-12,
      "loss": 11.7272,
      "step": 36723
    },
    {
      "epoch": 1.9997685705145223,
      "grad_norm": 0.5308431239774418,
      "learning_rate": 6.220934101186515e-12,
      "loss": 11.7632,
      "step": 36724
    },
    {
      "epoch": 1.9998230245111053,
      "grad_norm": 0.5191686533476448,
      "learning_rate": 3.4992754471829813e-12,
      "loss": 11.7697,
      "step": 36725
    },
    {
      "epoch": 1.9998774785076883,
      "grad_norm": 0.525138338749835,
      "learning_rate": 1.5552335419499743e-12,
      "loss": 11.7664,
      "step": 36726
    },
    {
      "epoch": 1.9999319325042713,
      "grad_norm": 0.5327611039604181,
      "learning_rate": 3.8880838548749357e-13,
      "loss": 11.6781,
      "step": 36727
    },
    {
      "epoch": 1.9999863865008543,
      "grad_norm": 0.5408051043597044,
      "learning_rate": 0.0,
      "loss": 11.8148,
      "step": 36728
    },
    {
      "epoch": 1.9999863865008543,
      "step": 36728,
      "total_flos": 4582719071862784.0,
      "train_loss": 6.056929863945536,
      "train_runtime": 133830.3994,
      "train_samples_per_second": 35.128,
      "train_steps_per_second": 0.274
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 36728,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 250,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 4582719071862784.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}